Topics in Monetary Economics: The International Monetary...

ECON 425

Topics in Monetary Economics:

The International Monetary System

from the Gold Standard to COVID-19

Fabio Ghironi

University of Washington

Spring 2020



Flattening Curves

Fabio Ghironi

University of Washington,

CEPR, and NBER

Introduction

• We are living a crisis that will be talked about in history books, with a pandemic that has

already killed thousands and sent many more to hospitals, many recovering only after long

stays in intensive care units (ICUs).

• Late reaction to the looming danger by many governments has resulted in having to impose

draconian measures to contain the virus and preserve the viability of strained health care

systems.

• This has required halting most economic activity, with costs that already huge and will grow

larger before the situation improves.

• Economic policymakers—central banks and governments—in the United States and

elsewhere have been taking actions to sustain the economy through the crisis.

• Time will tell whether the combinations of public health and economic policy actions

implemented by many countries are successful at managing the crisis and what combined

outcome they contribute to producing for world health and the global economy.

1

Introduction, Continued

• The crisis and the economic policy responses will be a central topic of this course.

• In particular, we will discuss the choices made by today’s monetary and fiscal policymakers

considering them through the lenses of past experiences in the international monetary

system from the Gold Standard to the present day.

• Before we delve into that, it is useful to introduce a simple conceptual framework for analysis.

• We do that by referring to a short article by Pierre-Olivier Gourinchas of U.C. Berkeley—

“Flattening the Pandemic and Recession Curves”—that was published just recently in a

March 2020 VoxEU.org e-book.

• The following slides mostly reproduce the article, with the occasional addition of a few

thoughts of mine.

2

Flattening Curves

• As Gourinchas notes, Figure 1 from his article summarizes how public health experts

approach the crisis.

• In the short run, the capacity of any country’s health system is finite (number of ICU and

other hospital beds, number of ventilators, number of skilled health professionals).

• This limits the number of patients that can be properly treated at any point in time, creating

an upper bound on this number represented by the flat line.

• If not contained, given the transmission rate of the virus, the pandemic would quickly

overwhelm any health system, leaving many patients with quickly deteriorating pulmonary

conditions without access to treatment.

• The fatality rate would surge and casualties would skyrocket to numbers that no sane person

would ever want to see.

– A 2% case fatality rate for overwhelmed health systems (most likely an optimistic number)

and 50% of the world population infected would result in 76 million dead.

• This scenario corresponds to the red curve in the figure.

3

1

“Flattening the Pandemic and Recession Curves” Pierre-Olivier Gourinchas1 Professor of Economics, UC Berkeley, visiting Princeton University March 13, 2020 We are facing a joint health and economic crisis of unprecedented proportions in recent history. I want to start by acknowledging that containment of the pandemic is the utmost priority. Figure 1 summarizes how public health experts approach the problem.

Figure 1: Flattening the Pandemic Curve

In the short run, the capacity of any country’s health system is finite (capacity of Intensive Care Units, number of hospital beds, number of skilled health professionals, ventilators….). This puts an upper bound on the number of patients that can be properly treated, at any given point in time and is represented by the flat line in the Figure. Unchecked, and given what we know of the transmission rate of the coronavirus, the pandemic would quickly overwhelm any health system, leaving many infected patients with deteriorating pulmonary conditions without any treatment. The fatality rate would surge. The threat is almost beyond comprehension. With a 2% case fatality rate baseline for overwhelmed health systems, and 50% of the world population infected, 1% of the world population -76 million people- would die. This scenario corresponds to the red curve in Figure 1. The part of the curve

1 [email protected]. Many thanks to Mark Aguiar, Olivier Blanchard, Jacques Delpla, Oleg Itskhoki and Jonathan Parker for useful comments.

Flattening Curves, Continued

• The part of the curve above the capacity of the health care system faces a sharply higher

mortality risk (shaded red area).

• Public health policy actions taken by many governments aim to “flatten the curve” by

reducing the transmission rate of the virus.

• This is accomplished by imposing drastic social distancing measures and promoting health

practices.

• Doing so spreads the pandemic over time, allowing more people to receive proper health

treatment, ultimately leading to a lower fatality rate.

• This is the blue curve in the figure.

4


• These policies, where implemented, have delivered very strong results.

• Countries that adopted drastic containment measures—such as Singapore and Taiwan—or

the Chinese regions outside Hubei have seen the number of cases grow at a markedly

slower rate.

• Clearly and unambiguously, this is the right short-run public health policy.

• Drastic “lockdown” becomes desirable and inevitable especially if delayed reaction by

authorities and failure to quell localized spikes in infection imply that the virus has spread

enough that hospital capacity is already under threat and there are not the time and

resources to implement large-scale test-and-trace approaches such as South Korea’s and

Singapore’s.

5


• Let us assume that authorities do the right thing for public health, so we are on the flattened

(blue) infection curve.

• What are the macroeconomic implications?

• As Gourinchas observes, in the short run, flattening the infection curve inevitably steepens

the macroeconomic recession curve.

• Social distancing has required closing schools, universities, most non-essential businesses,

and asking most of the working-age population to stay at home.

• Those able to work from home are a small fraction of the overall labor force, and disruption

to work and family routines affects even their productivity negatively.

• Essentially, the health policy response that is necessary to save millions of lives plunges the

economy into a sudden stop, with dramatic declines of production and trade.

6


• In a perfect world, once public health authorities give the all-clear signal, the switch on the

economy would be flipped back “on,” and the economic engine would immediately begin

humming again.

• But even in that “ideal” scenario world, the damage to the economy would be large.

• To see this, assume that, relative to a baseline, containment measures reduce economic

activity by 50% for one month and 25% for another month, after which the economy returns

to the baseline.

• Such a sharp but short-lived decline in activity would not seem unreasonable if we thought

that all that is happening to economic resources is that a majority of the labor force is

currently shuttered at home.

• But even that scenario would still imply a decline in annual GDP growth of 6.5% relative to

the previous year.

• Extend the 25% shutdown for just another month, and the decline in annual GDP growth

(relative to the previous year) would be almost 10%!

• For comparison, the decline in GDP growth in the U.S. during the 2008-09 Great Recession

was around 4.5%.

7


• Why are the numbers so much larger now?

• The short answer, writes Gourinchas, is that even at the peak of the financial crisis, when

the U.S. economy was shedding jobs at the rate of 800 thousand workers per month, the

vast majority of people were still employed and working.

– The unemployment rate in the U.S. peaked at ‘just’ 10%.

• By contrast, the coronavirus is creating a situation where, for a brief amount of time, 50% or

more of people may not be able to work.

• The impact on economic activity is comparatively that much larger.

• And this is true in the “ideal,” optimistic scenario—the world in which the we can just flip the

switch back on.

8


• We do not live in such world.

• In reality, we can anticipate a much slower process, with permanent loss of economic

activity.

• Improperly managed, the economic cost of the crisis could be much larger and longer

lasting.

• The reason is that modern economy is a complex web of interconnected parties: employees,

firms, suppliers, consumers, banks and financial intermediaries.

• Everyone is someone else’s employee, customer, lender, etc.

• A sudden stop like the one described above can easily trigger a cascading chain of events,

fueled by individually rational but collectively catastrophic decisions.

9


• Much like people, for reasons that are individually sensible, may ignore self-isolation

instructions, resulting in a steepening of the infection curve, so economic actors can make

individual decisions that amplify and precipitate a much larger economic downturn.

• Self-isolated customers may have fewer opportunities to spend.

• Yet, faced with uncertainties about future economic prospects, a common impulse may be

to cut down spending even further.

• This, in turn, would make it harder for firms to earn income—even out of current inventories.

• Faced with declining demand for their products (the collapse in demand is likely to be near

total in some sectors), firms will want to cut costs, shedding workers to avoid a complete

collapse.

• Banks, with a worsening portfolio of non-performing loans, will naturally want to cut lending,

further darkening the prospects of the non-financial sector.

• Suppliers will ask to be paid, etc.

10


• Panic or loss of confidence adds another layer.

• The result would be cascading business failures, with an associated surge in layoffs and a

build-up in financial fragilities.

• In other words, as Gourinchas puts it, a real danger is that the virus mutates and infect our

economic system even as we manage to root it out of our bodies.

• In both cases, infection and recession, externalities are key: actions that are individually

sensible, but collectively harmful.

• The upshot is that the economy too faces a “flatten the curve” problem.

11


• Without proper macroeconomic support, the impact of the downturn could be represented

by the red curve in Figure 2.

• It plots output lost during a sharp downturn, amplified by the economic decisions of millions

of economic agents trying to protect themselves by cutting spending, shelving investment,

cutting down credit, and generally hunkering down.

• For the economy, it is isolation that creates the negative externalities.

• The blue shaded area represents the economic downturn if we could prevent any additional

“economic infection,” i.e., limit the loss of economic activity to the lost production during the

public health containment period.

• This is likely to be a large negative number.

• But the red line—and additional red shaded area—represents the additional loss of

economic activity once the economy itself gets ‘infected’ and the various negative feedback

loops and amplification mechanisms described above kick-in.

12

4

Economists will recognize that the problem in both cases -infection and recession- are externalities: actions that are individually perfectly sensible, but collectively harmful. The upshot is that the economy too faces a ‘flatten the curve’ problem. Without proper macroeconomic support, the impact of the downturn could be represented by the red curve in Figure 2. It plots output lost during a sharp an intense downturn, amplified by the economic decisions of millions of economic agents trying to protect themselves by cutting spending, shelving investment, cutting down credit and generally hunkering down. Notice the irony: for the economy, it is isolation that has negative externalities!

Figure 2: Flattening the Recession Curve

In this figure, the blue shaded area represents the economic downturn, if we could prevent any additional ‘economic infection’, i.e. limit the loss of economic activity to the lost production during the public health containment period. As already discussed, this is likely to be a large negative number. The red line -and additional red shaded area- represents the additional loss of economic activity once the economy itself gets ‘infected’ and the various negative feedback loops and amplification mechanisms described above kick-in. Not by coincidence, the measures that help solve the health crisis can make the economic crisis worse -at least in the short run- and vice versa: stricter health policy forces a larger economic shutdown, a larger blue-shaded area. Yet, while this may look like a trade-off, it is not really one: unemployment versus lost lives, there is not much to debate -at least at the infection and fatality rates that we are witnessing. Moreover, even if no containment measures were implemented, a recession would occur anyway, fueled by the precautionary and/or panic behavior of households and firms faced with the uncertainty of dealing with a pandemic with an inadequate public health response.


• The measures that help solve the health crisis can make the economic crisis worse, at least

in the short run, and vice versa:

– Stricter health policy forces a larger economic shutdown, a larger blue-shaded area.

• Yet, while this may look like a tradeoff, it is not really one: unemployment versus lost lives,

not much to debate.

• Moreover, even if no containment measures were implemented, a recession would occur

anyway, fueled by the precautionary and/or panic behavior of households and firms faced

with the uncertainty of dealing with a pandemic with an inadequate public health response.

• In the longer run—an intertemporal dimension that Gourinchas does not really address—the

economic costs of not containing the pandemic and facing the resulting devastation of

society would most likely dwarf those of an even very large short-to-medium-run recession.

13


• Fortunately, economic policy can act decisively to prevent these the “economic infections”

Gourinchas talks about.

• The basic objective here is also to flatten the curve and limit the economic damage to what

is inevitable given that output is not produced when the labor force is largely quarantined.

• Modern economic system (again, when properly managed) contain a number of fail-safes

that are designed to prevent or limit catastrophic collapses of this type, that have been

designed also building on lessons from past experiences.

• Central banks can provide emergency liquidity to the financial sector.

• Fiscal stabilizers (the decline in government fiscal revenues and the increase in transfers)

help lessen the blow of downturns on household and firms’ bottom lines.

• Governments can deploy discretionary targeted fiscal measures or broader programs to

support economic activity.

14


• These measures help flatten the economic curve, i.e., limit the economic loss.

• This is what central banks and governments around the world have been trying to

accomplish with their recent actions (and, most likely, those that will come).

• It is, however, important to keep in mind what economic policy can and cannot do.

• The objective is not and cannot be to eliminate the recession altogether.

• The recession will be there, it will be massive, but, hopefully, short-lived.

• The priority is to short-circuit all the negative feedback loops and channels of contagion that

otherwise amplify this negative shock.

15


• Unchecked, the recession threatens to destroy the complex network of economic linkages

that allows the economy to operate and that would take a long time to repair.

• From this perspective, the priority should be:

1. To ensure that workers can remain employed (and collect their paycheck) even if

quarantined or forced to stay home to look after dependents.

– Temporary layoff assistance is a key component. Without it, it is even unclear whether

public health advisories can be followed. Households need to be able to make basic

payments (rent, utilities, mortgages, insurance).

2. To ensure that firms can weather the storm without going into bankruptcy, with easier

borrowing terms, possibly temporarily waving tax or payroll payments, suspending loan

payments, or providing direct financial assistance where needed.

– This especially for small- and medium-size enterprises whose financial resources are

more limited. Larger U.S. corporations, for instance, have already demonstrated the

ability to operate successfully under Chapter 11 bankruptcy proceedings, if needed.

3. To support the financial system as non-performing loans will mount, so as to ensure the

crisis does not morph into a financial crisis of the type the world experienced in 2007-08.

16


• These measures will dampen the amplification loops and greatly reduce the economic

downturn.

• The timing is important.

• Economic measures are most acutely needed while the economy is in shutdown mode.

• Stimulus packages after the health crisis is over are only needed if we do not manage to

act decisively right now to avoid a catastrophic collapse and could potentially be much more

costly.

17


• Moreover, strict health policies can dampen dramatically the spread of the disease but they

leave a larger fraction of the population unexposed to it.

• This implies that containment measures might need to be imposed for a longer time or

re-tightened, otherwise the pandemic might flare up again.

• How long could production remain shut down before the size of the recession becomes

catastrophically large?

• The simple calculations above indicated that one month at 50% and 2 months at 25% would

already cost 10% of annual output.

• Another two months at 75% production would cost another 5% of annual output.

• This indicates that the right strategy is dynamic.

18


• In terms of Figure 1, a key goal of fiscal policy should be to expand the capacity of the health

system.

• Raising the horizontal line in that figure allows both to treat more patients, but also to relax

the containment measures.

• This directly benefits the economy, without degrading the public health response.

• Another key goal should be to put in place the structure needed to implement South Korea’s

strategy of large scale testing and tracing of infected cases and their contacts.

• This will make it possible to respond to localized spikes in infections that may happen in the

future more effectively, without having to resort to draconian lockdowns.

19


• Inevitably, these measures will have a heavy fiscal cost.

• Gourinchas consider that governments may need to provide income support on a scale

roughly comparable to the output lost.

• If total output loss is of the order of 10% of annual GDP, this would mean deploying a

comparable level of fiscal resources, even before accounting for other health expenditures.

• By comparison, while the markets cheered the U.K. announcement of a $39bn stimulus

package, this represented “only” 1.5% of UK output.

20


• Should we worry about this?

• Put differently, are we constrained in what we can do on the macroeconomic side, much like

hospital beds, ventilators, and the number of health professionals constrain the capacity of

the health system?

• Gourinchas argues that we are not, with some exceptions.

• Borrowing rates for governments are at historical lows, even more so since the beginning of

the crisis.

• At these borrowing rates, even a 10% increase in debt-GDP only increases annual interest

costs by 0.1% of GDP.

• If now is not the time to borrow to support an economy on the verge of collapse, when is a

good time?

• Most advanced economies should be able to face such a one-off increase in public debt.

• The U.S. Congress just recently approved a $2tn economic package.

21


• Eurozone countries like Italy, with elevated pubic debt levels, are more constrained.

• The answer is clear: Strong signals should be sent that the Eurozone will support these

efforts.

• This could take many forms.

• The European Commission has already temporarily waived budgetary rules, allowing

countries like Italy to run larger deficits.

• But given Italy’s elevated debt levels, and the fragility of the current financial environment,

one needs to make sure there is a backstop.

22


• Substantial support could come from the European Central Bank (ECB) via its Outright

Monetary Transactions (OMT) program, an instrument introduced by Mario Draghi’s ECB at

the height of the Eurozone sovereign debt crisis in 2012.

• But OMT comes with a program: Countries that apply for OMT support must subject

themselves to adjustment requirements which are entirely superfluous in this case and

create a stigma effect that makes countries very reluctant to use this tool.

• The crisis created by the coronavirus is a common shock, for all European Union (EU)

countries.

• Basic economic theory implies that there should be a common answer.

23


• The best response, then, would be to issue a Eurozone bond with two specific purposes: to

finance necessary health expenditures and to prevent economic dislocation in the affected

countries.

• Several Eurozone members are opposed to the creation of a Eurobond.

• One of the main objections is so-called moral hazard, i.e., the potential risk that countries

would apply insufficient fiscal discipline in the future, in the hope that other countries foot

the bill.

• This objection should fade when we think of the challenges created by the coronavirus.

• The virus cares little about incentives, or borders for that matter.

• A Eurozone “Coronavirus” bond would send a strong signal that European countries stand

together when confronted by a common shock.

• It will be more powerful than monetary policy in restoring economic confidence and enabling

the health authorities in all the affected European countries to fight the real battle.

24


• Should a targeted Eurobond issuance prove a bridge too far for European policymakers,

there alternative suggested by Gourinchas is a coordinated jumbo sovereign debt

issuance—between 10% to 20% of GDP—coordinated with an expansion of Quantitative

Easing by the ECB, which would also provide much needed fiscal space.

• A third solution, proposed by Lorenzo Bini Smaghi (a former member of the ECB’s Executive

Board) in a March 28, 2020 VoxEU.org article would be to have all Eurozone countries

collectively applying for OMT European Stability Mechanism (ESM) support with minimal

conditionality.

• This would unlock the large amount of resources that the ESM and ECB could mobilize

for governments without generating stigma effects for any specific country and with the

advantage of sidestepping the parliamentary approval steps that would be needed to

introduce the Coronavirus bond.

25


• Regardless of specifics, the bottom line is that bold policy initiatives are needed to contain

the looming recession.

• The right combination starts with public health policy in the driver seat to limit human

contagion.

– Containing the pandemic is the utmost priority.

– Failure to do that would imply costs (including economic costs) far larger than even the

large costs that are being incurred now.

• Fiscal, financial, and monetary policies should be designed to accompany the shock to our

economic system and limit economic contagion.

• As Gourinchas concludes, this is not the time to be cautious.

26



Macroeconomic Policy in the Time of Coronavirus

Fabio Ghironi


CEPR, and NBER

Introduction

• Gourinchas’ article that we talked about gave us a simple conceptual framework to think

about the exercise that policymakers around the world are engaging in now: flattening the

pandemic curve and also trying to flatten the recession curve.

• Recession-curve flattening is the responsibility of macroeconomic policy—mostly monetary

and fiscal policy.

• Luca Fornaro (CREI, Barcelona, Spain) and Martin Wolf (University of Vienna, Austria)

present a simple analytical framework to think about the economic side of the crisis and

the role of policy in their 2020 CEPR Discussion Paper on “COVID-19 Coronavirus and

Macroeconomic Policy.”

• The framework builds on work on “Stagnation Traps” that Fornaro did with Gianluca Benigno

(Federal Reserve Bank of New York) and that was published in the Review of Economic

Studies in 2018.

• These slides mostly reproduce the Fornaro-Wolf paper, with the addition of some

background material and comments.

1

The Fornaro-Wolf Model of the COVID-19 Economic Crisis

The New Keynesian Framework

• The starting point of Fornaro and Wolf’s model is a simplified version of the so-called New

Keynesian macroeconomic model, which started being developed in the 1980s.

• Assume that variables below are measured in logs.

– For instance, yt is the log of GDP in period t.

– In many versions of the model, variables actually measure log-deviations from trend

levels: yt ≡ log Yt − log Y trendt , where Yt is GDP and Y trend

t is its trend level.

– In the case of interest or inflation rates, logs of gross rates are taken.

• The basic New Keynesian model boils down to a three-equation log-linear system.

– Log-linear means linear in logs.

• The log-linear system can be derived from a fully microfounded non-linear model that

specifies optimization problems for households and firms using a technique known as

log-linearization.

– You would learn the foundations of the New Keynesian model and the log-linearization

technique if you took ECON 401 with me.

2

The Fornaro-Wolf Model of the COVID-19 Economic Crisis, Continued

The Intertemporal IS

• There is an equation that describes how output today depends on the ex ante real interest

rate and on expected future output:

yt = −σ [it − Et (πt+1)] + Et (yt+1) + ut, (1)

where σ > 0 is a parameter that measures the responsiveness of today’s output to the

real interest rate, it is the nominal interest rate, πt+1 is inflation next period, Et (.) is the

expectation of the variable inside the parentheses based on information available at time t,

and ut is an exogenous demand shock.

• This equation is an intertemporal IS equation.

– Today’s GDP contracts if the real interest rate increases; today’s GDP rises if it is

expected to rise tomorrow.

– ut is an exogenous demand shock.

• This equation describes the demand-side of the economy.

3


The New Keynesian Phillips Curve

• There is an equation that describes how today’s inflation depends on today’s output and on

expected future inflation:

πt = λyt + βEt (πt+1) + zt, (2)

where λ > 0 and β > 0 are parameters, and zt is an exogenous shock.

– Current inflation rises if GDP rises and if inflation is expected to rise tomorrow.

• This equation is known as New Keynesian Phillips Curve and it describes the supply-side of

the model.

4


Monetary Policy

• Finally, there is an equation that describes monetary policy in terms of what is known

as a Taylor-type rule for interest rate setting, from the 1993 article by John Taylor in the

Carnegie-Rochester Conference Series on Public Policy that started the literature on the

Taylor rule.

• For instance:

it = απt + γyt + xt, (3)

where α > 0 and γ > 0 are policy response parameters, and xt is an exogenous policy

shock.

• Equations (1)-(3) are a system of three dynamic equations for the three endogenous

variables yt, πt, and it as functions of the exogenous shocks ut, zt, and xt.

• Usually, it is assumed that these shocks follow so-called first–order autoregressive

processes, in which the level of the shock today depends on its level last period and on an

innovation in the current period.

• If you took ECON 401 with me, you would also learn how to solve systems like (1)-(3) to

simulate, for instance, what happens to the economy when there is one or another shock.

5


• The model in Fornaro and Wolf’s paper is a version of the New Keynesian framework.

• It assumes that output is a function of labor (lt) and of exogenous productivity (at):

yt = at + lt (4)

• There is a maximum level of employment lt which is assumed to be efficient, i.e., it is the

level that maximizes welfare in the economy.

– A microeconomic foundation of this assumption is in Benigno and Fornaro’s (2018)

article.

– It is based on assuming that agents experience no disutility from working and can supply

up to lt units of labor. Involuntary unemployment can arise as a consequence of sticky

wages.

• When lt = lt, the economy operates at full employment and output is equal to potential,

while when lt < lt there is some involuntary unemployment and output is below potential.

• The growth rate of productivity is defined by gt ≡ at − at−1.

6


• Fornaro and Wolf note that output and employment in their model are determined by

aggregate demand (AD), as in the Keynesian tradition, although, as we shall see, the

AD and AS (aggregate supply) blocks of their framework mix both demand and supply

ingredients.

• As in equation (1) of the New Keynesian system above, AD depends on expectations of

future output yt+1 and on the real interest rate, denoted rt.

– Fornaro and Wolf omit the expectation operator since they do not develop a model with

stochastic exogenous shocks as we would in general.

• First, a lower real interest rate (denoted with rt) boosts AD, for instance by encouraging

expenditure financed by borrowing.

• Second, current demand is increasing in expectations of future output.

– Consumers are more willing to spend in the present if they anticipate a higher future

income.

7


• This leads Fornaro and Wolf to specify the intertemporal IS of their model as:

yt = −rt + yt+1. (5)

• The only difference with equation (1) is the assumption σ = 1 that Fornaro and Wolf implicitly

make, and the omission of the exogenous shock ut.

• The real interest rate is related to the nominal interest rate and expected inflation by the

equation known as Fisher equation, which we used implicitly in equation (1):

rt = it − πt+1. (6)

• A simplifying assumption in the first part of Fornaro and Wolf’s analysis is that inflation is

fixed and such that πt = πt+1 = π.

• This implies that, by controlling the nominal interest rate, the central bank is de facto setting

the real interest rate.

– As Fornaro and Wolf point out, all it takes for the result we shall analyze is some rigidity

in prices or wages.

– The assumption of constant inflation is a limiting case in which prices are fully rigid.

8


• Since there is no concern about inflation (it is assumed constant), Fornaro and Wolf assume

that the central bank sets the nominal interest rate according to the simple rule:

it = ı + φ(lt − lt

)(7)

where ı and φ are positive constants.

• Under this rule, the central bank aims at stabilizing output at potential by stabilizing

employment at its efficient level.

• This is accomplished by lowering the interest rate when employment falls below lt.

9


• Substituting (7) into (6), and solving for rt gives:

rt = ı + φ(lt − lt

)− π, (8)

where we used the assumption of constant inflation.

• Next substitute the time-t production function and its t + 1 version into the IS equation (5) to

obtain:

at + lt = −rt + at+1 + lt+1.

• Recall the definition of gt ≡ at − at−1. Hence, this equation becomes:

lt = −rt + gt+1 + lt+1.

10


• Finally, substituting (8) into this equation yields:

lt = −ı− φ(lt − l

)+ π + gt+1 + lt+1,

which can be rearranged by collecting the terms in lt as:

lt (1 + φ) = −ı + π + φlt + gt+1 + lt+1. (9)

• This equation describes the AD side of Fornaro and Wolf’s New Keynesian model, combining

the intertemporal IS and the policy specification, and taking also into account the element of

supply represented by the production function.

• Note that in standard New Keynesian models in which policy is modeled as a Taylor-type

rule, one does not need the specify an LM equation to fully describe the AD side of the

economy.

• It is possible to determine all variables of interest fully without any reference to an LM

condition.

11

The AD Effect of the Coronavirus

• How do Fornaro and Wolf model the AD effect of the pandemic?

• A first thing to observe is that the pandemic surely reduced the welfare-maximizing level of

employment.

• The reason is that many occupations require in-person social interactions, which facilitate

the spread of the virus.

• Limiting employment, by imposing an economic lockdown, is thus desirable to mitigate the

impact of the virus on public health.

• These considerations can be captured through a fall in lt.

• But this is not what Fornaro and Wolf do. In fact, they abstract from this effect of the crisis,

and they assume that efficient employment is constant and such that lt = l in every period.

12

The AD Effect of the Coronavirus, Continued

• A second channel through which the pandemic impacts the economy is that the virus

and lockdown are likely to generate a reduction in future productive capacity—by making

firms scrap investment plans, companies go bankrupt, and by destroying employment

relationships.

• All these effects produce a long-lasting supply disruption, which might very well extend far

beyond the end of the epidemic.

• In the model, this can be captured by a persistent drop in labor productivity growth.

• This is what Fornaro and Wolf focus on.

• Notice that they refer to this as an AD effect of the pandemic, but it really comes from the

supply ingredient of the AD block of their model—the production function.

13


• Analytically, it is convenient to focus on the limiting case in which the fall in productivity

growth caused by the coronavirus is permanent, even if the results can be generalized to

the case of a persistent, but not permanent, drop in productivity growth.

• Studying what happens if the rate of labor productivity growth falls permanently is equivalent

to studying the effects of a permanently lower level of a constant gt.

• Since gt is constant at a level g, all other variables are constant over time, and the AD

equation (9) becomes:

φ(l − l

)= −ı + π + g. (10)

where l is the constant level of employment implied by the AD equation with constant lt and

constant gt.

14


• Figure 1 in Fornaro and Wolf’s paper plots the AD schedule in a diagram with l on the

horizontal axis and g on the vertical axis.

• The schedule is upward sloping, because lower productivity growth is associated with

expectations of lower future income, and thus with weaker aggregate demand.

• Lower aggregate demand, in turn, depresses output and employment.

• As shown in the figure, for given g, this equation determines employment l.

15

(a) (b)

Figure 1: Impact of coronavirus on aggregate demand and employment.

result is that the supply disruption is persistent, so as to induce agents to revise downward their

expectations of future income.3

What does it take to restore full employment? The central bank needs to inject further mone-

tary stimulus, i.e. it needs to lower i. Graphically, this corresponds to a rightward shift of the AD

curve. If the monetary stimulus is strong enough, full employment is restored, as illustrated by

the right panel of Figure 1. This simple model thus lends support to the idea that central banks

might need to respond to the Covid-19 outbreak by easing monetary policy.

In reality, however, restoring full employment through monetary stimulus might not be that

easy. First, social-distancing is impairing households’ ability to spend. A reduction in interest rates

might thus have a much weaker impact on demand, compared to normal times.4 Second, interest

rates are currently very low. This reduces central banks’ ability to cut policy rates, because of the

effective lower bound constraint. We will go back to this point later on.

Let us now spend a few words on inflation. Suppose that the prices set by firms are increasing

in the marginal cost of production. Higher wages, therefore, push up prices by increasing marginal

costs. Higher labor productivity, instead, lowers prices by reducing the marginal cost of production.

We can then write price inflation πt as

π = πw − g, (4)

where πw denotes nominal wage inflation. Let us also assume the existence of a wage Phillips curve

πw = ξ(l − l), where ξ > 0 so that wage inflation is positively related to employment. Inflation is

3This effect is well known from the literature on news shocks (e.g., Lorenzoni, 2009).4To capture this effect, we can replace equation (2) with

yt = −σrt + yt+1,

where σ > 0 determines the sensitivity of aggregate demand to changes in the interest rate. Social distancing wouldthen lead to a reduction in σ.

4


• Now imagine that we start from an equilibrium characterized by full employment (l = l) and

suppose that the coronavirus epidemic causes a (previously unexpected) fall in g to g′ < g.

• The outcome is illustrated by the left panel of Figure 1:

– The fall in productivity growth translates into lower aggregate demand.

– The central bank reacts by cutting the policy rate, but not enough to prevent unemploy-

ment from arising.

– The result is a drop in employment below its efficient level (l′ < l).

• In this simple model, therefore, the negative supply shock triggered by the coronavirus

generates a fall in AD and involuntary unemployment.

• The crucial assumption behind this result is that the supply disruption is persistent, so as to

induce agents to revise downward their expectations of future income.

16


• What does it take to restore full employment?

• The central bank needs to inject further monetary stimulus above and beyond what is

dictated by the endogenous part of the policy rule, i.e., it needs to lower ı.

• Graphically, this corresponds to a rightward shift of the AD curve.

• If the monetary stimulus is strong enough, full employment is restored, as illustrated by the

right panel of Figure 1.

• Thus, the simple model lends support to the idea that central bank had to respond to the

COVID-19 outbreak by easing monetary policy.

• In reality, however, restoring full employment through monetary stimulus might not be that

easy.

17


• First, social distancing is impairing households’ ability to spend.

• Therefore, a reduction in interest rates might have a much weaker impact on demand

compared to normal times.

• As Fornaro and Wolf note, this can be captured by replacing their equation (5) with the more

general specification:

yt = −σrt + yt+1,

where 0 < σ < 1.

• The effect of social distancing on the ability of monetary policy to affect output would then

be captured by the lower value of σ, which has fallen below 1.

18


• Moreover, interest rates are currently very low, reducing the ability of central banks to cut

policy rates.

– The Fed’s recent actions brought the federal funds target rate back to essentially zero,

which is viewed as the lower bound for the Fed’s interest rate.

– Some central banks—like the ECB—had policy rates already in negative territory before

the current crisis.

• However, before studying this problem explicitly, Fornaro and Wolf shift their focus to inflation

determination, relaxing the assumption of fixed inflation.

19


• Suppose that prices set by firms are increasing in marginal costs.

• The production function

Yt = AtLt

(the level counterpart to the log equation (4)) implies that producing a unit of output requires

1/At units of labor.

• Hence, the marginal cost of production is given by the ration of the wage to productivity,

Wt/At, where Wt is the nominal wage.

• Higher wages, therefore, push up prices by increasing marginal costs.

• Higher labor productivity, instead, lowers prices by reducing the marginal cost of production.

20


• Assuming that prices are flexible (and markups of price over marginal cost) are therefore

constant, we can then write (log gross) price inflation as:

π = πw − g, (11)

where πw denotes nominal wage inflation.

• Now assume the existence of a wage Phillips curve

πw = ξ(l − l

), (12)

where ξ > 0, so that wage inflation is positively related to employment.

– This amounts to an assumption of wage stickiness.

• Equations (11) and (12) together imply that inflation is determined by:

π = ξ(l − l

)− g. (13)

• Will COVID-19 lead to higher or lower inflation?

21


• It depends!

• Lower productivity growth tends to push inflation up.

– This is the classic notion that negative supply shocks are inflationary.

• But lower employment pushes wage inflation down.

– This effect points toward lower price inflation.

• The relative strength of these two effects depends on the slope of the wage Phillips curve

(ξ).

• It is hard, a priori, to say whether the coronavirus outbreak will lead to higher or lower

inflation.

• The Fornaro-Wolf model suggests that central banks may face a tradeoff between stabilizing

employment at its efficient level and inflation.

22

The Supply-Demand Doom Loop

• So far, the analysis took the rate of labor productivity growth as an exogenous variable.

• In reality, firms can increase their labor productivity by investing to increase their capital

stock, or to develop innovations that improve the quality of their products.

• It is reasonable to assume that firms’ investment decisions depend on AD.

• First, when demand is strong the return from investment tends to be high.

• Weak AD, consequently, depresses firms’ incentives to invest.

23

The Supply-Demand Doom Loop, Continued

• Moreover, due to financial market imperfections (such as asymmetric information problems),

many firms must rely on internal funds to finance investment.

• Weak aggregate demand reduces firms’ operating profits and erodes their net worth, forcing

financially-constrained firms to scrap investment plans.

• These effects give rise to a positive relationship between investment—and so labor

productivity growth—and aggregate demand.

– There are also other channels through which a spell of weak AD can produce a drop in

future potential output.

– For instance, weak AD generates layoffs.

– Due to search-and-matching frictions in the labor market, it might not be easy to restore

these matches quickly once demand recovers, with negative effects on productivity.

• These effects are fully microfounded in Fornaro’s Review of Economic Studies article with

Benigno.

24


• In their analysis of the COVID-19 crisis, Fornaro and Wolf simply assume that productivity

growth evolves according to

g = χl + g, (14)

where χ and g are two positive constants.

• The term l captures the endogenous component of productivity growth.

• The rationale behind this term is that higher AD, which is associated with higher employment,

leads to higher investment and faster productivity growth.

• g instead, captures all the factors that can affect productivity independently of demand—such

as the spread of the coronavirus and the associated lockdown.

25


• The relation between g and l in equation (14) is plotted in Figure 2 as the GG schedule,

which Fornaro and Wolf view as summarizing the supply side of their simple model.

– Figure 2 also plots the AD schedule.

• For the reasons mentioned above, the GG schedule is upward sloping.

• The equilibrium is thus determined by the intersection of two upward sloping curves.

• This signals the presence of amplification effects.

26

Figure 2: The supply-demand doom loop.

Let’s now go through the macroeconomic impact of a negative supply shock triggered by the

coronavirus spread, which we capture by a fall in g. As shown in Figure 2, the fall in g makes the

GG curve shift toward the right. If monetary policy holds i constant, the new equilibrium features

lower productivity growth and lower employment.

What is interesting, is that now a supply-demand doom loop takes place. As before, the initial

negative supply shock depresses aggregate demand. But now lower demand induces firms to cut

back on their investment, which generates an endogenous drop in productivity growth and future

potential output. Lower productivity growth, in turn, induces a further cut in demand, which

again lowers investment and growth. This vicious spiral, or supply-demand doom loop, amplifies

the impact of the initial supply shock on employment and labor productivity growth.

Now monetary interventions aiming at sustaining demand have a multiplier effect - because they

reverse the supply-demand doom loop. Suppose that the central bank eases monetary policy, by

lowering i. This intervention increases aggregate demand. Moreover, higher demand induces firms

to increase investment. In turn, this sustains consumers’ expectations of future income, leading to

a further rise in demand, and so on. Under this scenario, a monetary expansion has a particularly

large impact on employment and productivity, because it counteracts the supply-demand doom

loop.

Unfortunately, as we argued above, central banks might be able to impart only a limited amount

of monetary stimulus to the economy. But the supply-demand doom loop can be reversed also

through appropriate fiscal policy interventions. Imagine that governments can implement policies

to sustain investment, so that now the GG equation becomes

g = χl + g + s, (GG)

where s captures government policies aiming at increasing investment. A higher s, for instance,

can be interpreted as a rise in subsidies to firms’ investment, an increase in public investment,

public credit provision to financially-constrained firms or even subsidies to prevent the breakup

of workers-firms matches. All these policies, in fact, lead to higher aggregate investment - and

6


• Consider the macroeconomic impact of a negative supply shock triggered by the coronavirus

spread, captured by a fall in g.

• As shown in Figure 2, the fall in g makes the GG schedule shift toward the right.

• If monetary policy holds ı constant, the new equilibrium features lower productivity growth

and lower employment.

• But now a supply-demand doom loop takes place:

– As before, the initial negative supply shock depresses AD.

– But now lower demand induces firms to cut back on their investment, which generates

an endogenous drop in productivity growth and future potential output.

– Lower productivity growth, in turn, induces a further cut in demand, which again lowers

investment and growth.

• This vicious spiral, or supply-demand doom loop, amplifies the impact of the initial supply

shock on employment and labor productivity growth.

27


• Now monetary interventions aiming at sustaining demand have a multiplier effect because

they reverse the supply-demand doom loop.

• Suppose that the central bank eases monetary policy, by lowering ı.

• This intervention increases aggregate demand.

• Moreover, higher demand induces firms to increase investment.

• In turn, this sustains consumers’ expectations of future income, leading to a further rise in

demand, and so on.

28


• Under this scenario, a monetary expansion has a particularly large impact on employment

and productivity, because it counteracts the supply-demand doom loop.

• Unfortunately, as noted above, the amount of stimulus to the economy that central banks

can impart by lowering interest rates may be limited.

• But the supply-demand doom loop can be reversed also through appropriate fiscal policy

interventions.

29


• Imagine that governments can implement policies to sustain investment, so that now the

GG equation becomes:

g = χl + g + s,

where s captures government policies aiming at increasing investment.

– A higher s, for instance, can be interpreted as a rise in subsidies to firms’ investment, an

increase in public investment, public credit provision to financially-constrained firms or

even subsidies to prevent the breakup of worker-firm matches.

– All these policies, in fact, lead to higher aggregate investment—and therefore higher

labor productivity growth—for given aggregate demand.

• Graphically, a rise in s generates an upward shift of the GG curve, leading to higher

productivity growth and employment.

30


• These fiscal interventions, which—in Fornaro and Wolf’s interpretation—act on the supply

side of the economy, also affect aggregate demand.

• The reason is that higher investment boosts expectations of future growth and income,

leading agents to increase spending in the present.

• In turn, higher aggregate demand leads to a further rise in investment and productivity

growth, etc.

• The bottom line is that fiscal interventions supporting investment reverse the supply-demand

doom loop, and so they trigger a positive multiplier effect on economic activity.

31


• A comment at this point:

– I am not entirely convinced by Fornaro and Wolf’s choice of labels for demand-side and

supply-side forces in their model.

– By virtue of how it is constructed and developed, each side actually contains a mixture of

demand and supply ingredients.

· Think about the use of the production function in the AD block of the model.

– If you think of national accounting, investment is a component of AD. A policy that boosts

investment is therefore hardly a supply-side policy.

– Demand and supply forces interact within each block of the model—and then across

blocks once they are put together to determine the overall equilibrium.

– The language that Fornaro and Wolf use is a convenient way to simplify an underlying

web of effects.

32

Animal Spirits and Stagnation Traps

• At this point, let us return to the issue of the effective lower bound on central bank interest

rate setting that we briefly mentioned above.

• Suppose there is a lower bound il below which the central bank cannot push the interest

rate.

• If we were building the model from its non-linear foundations, we would impose this

assumption on a non-linear rule for interest rate setting (rule (7) follows from taking logs of a

multiplicative rule) and we would observe that log-linearization is a bad technique to use to

think about the lower bound on interest rate setting.

• Fornaro and Wolf sidestep this problem by imposing the lower bound directly on the

log-linear rule, which still allows them to make a key point.

33

Animal Spirits and Stagnation Traps, Continued

• This implies that the interest rate rule becomes:

i = maxı + φ

(l − l

), il. (15)

– The policy rate is constrained to being the maximum between ı + φ(l − l

)and il.

– Circumstances may be such that the central bank would like to push the interest rate

below il (ı + φ(l − l

)< il), but it cannot.

– When this happens, the interest rate is stuck at il.

• It turns out that this creates a major problem.

34


• If demand is weak enough that the interest rate hits the lower bound, the economy

experiences a liquidity trap.

• The AD equation now becomes

maxı + φ

(l − l

), il

= π + g. (16)

• As shown in the left panel of Figure 3, the AD equation now exhibits a kink, and it becomes

horizontal for values of l low enough to trigger a liquidity trap.

• As before, imagine that the coronavirus outbreak induces a downward shift of the GG curve,

from GG to GG’.

• As drawn in the figure, there are now two intersections between the AD and the GG’ curve.

• This means that two equilibria are possible.

35

(a) (b)

Figure 3: Stagnation traps and fiscal policy.

cutting investment, which negatively affects productivity growth. Initial pessimistic expectations

of weak growth thus become self-fulfilling. Importantly, this self-fulfilling feedback loop can take

place only if the fundamentals of the economy are sufficiently weak (notice that the equilibrium is

unique before the coronavirus causes a drop in g). The coronavirus epidemic, therefore, can open

the door to expectation-driven stagnation traps precisely by weakening the growth fundamentals

of the economy.

Which policy interventions can prevent a stagnation trap from taking place? There is little

that conventional monetary policy can do, since the policy rate is constrained by the zero lower

bound. Luckily, fiscal policy - and in particular policies that sustain investment - can be of help.

Suppose that the government reacts to the coronavirus outbreak by increasing s. As illustrated

by the right panel of Figure 3, this policy induces an upward shift of the GG curve, from GG′ to

GG′′. If this shift is large enough, the stagnation trap equilibrium disappears. In economic terms,

this means that only a sufficiently aggressive fiscal intervention can rule out stagnation traps. A

timid intervention, in fact, will not do the job (think about a small upward shift of the GG curve).

Taking stock, this coronavirus outbreak might cause a persistent supply disruption, which might

last far longer than the epidemic itself (and the associated economic lockdown). We show that, in

this case, the spread of the virus might cause a demand-driven slump, give rise to a supply-demand

doom loop, and open the door to stagnation traps induced by pessimistic animal spirits. Monetary

policy is likely to be insufficient in mitigating the slump induced by the coronavirus shock. Instead,

aggressive fiscal policy interventions to support investment - and more broadly future productivity

capacity - can play a key role in sustaining employment and growth, by reversing the supply-

demand doom loop. This is especially true if governments will need to jumpstart their economies

out of stagnation traps driven by pessimistic animal spirits.6

6Of course, financing a large fiscal stimulus package represents a difficult challenge for governments. While we donot address this issue here, our analysis suggests that fiscal interventions to stimulate investment are likely to triggerpositive multiplier effects on economic activity. Taking into account these effects is important to design optimalfiscal packages.

8


• The first equilibrium, corresponding to the point (l′, g′), is the one we analyzed in the case

without the lower bound on interest rate setting.

• The second equilibrium, corresponding to the point (l”, g”), is new and is the equilibrium in

which the economy is stuck in a liquidity trap (i = il)

• In this equilibrium, both growth and employment are depressed (l” < l′ and g” < g′).

• This second equilibrium is an example of the stagnation traps that Fornaro studied with

Benigno.

36


• Notice that nothing “fundamental” determines which equilibrium prevails.

• In fact, agents can coordinate their expectations on either of the two equilibria.

• Therefore, pessimistic animal spirits (a concept that Keynes emphasized) can push the

economy into a stagnation trap.

• Now the coronavirus shock not only triggers a supply-demand doom loop, it also places

the economy in a danger zone in which animal spirits and agents’ expectations can affect

employment and productivity growth.

37


• To see how this can happen, imagine that agents become pessimistic about future growth.

• Due to the lower bound on the interest rate, the central bank cannot counteract the

associated drop in demand.

• As a result, employment and economic activity drop.

• Firms react by cutting investment, which negatively affects productivity growth.

• Initial pessimistic expectations of weak growth thus become self-fulfilling.

• Importantly, this self-fulfilling feedback loop can take place only if the fundamentals of the

economy are sufficiently weak (notice that the equilibrium is unique before the coronavirus

causes a drop in g).

• The coronavirus epidemic, therefore, can open the door to expectation-driven stagnation

traps precisely by weakening the growth fundamentals of the economy.

38


• Which policy interventions can prevent a stagnation trap from taking place?

• There is little that conventional monetary policy can do, since the policy rate is constrained

by the lower bound.

• Luckily, fiscal policy—and in particular policies that sustain investment in the Benigno-

Fornaro-Wolf framework—can be of help.

39


• Suppose that the government reacts to the coronavirus outbreak by increasing s.

• As illustrated by the right panel of Figure 3, this policy induces an upward shift of the GG

curve, from GG’ to GG".

• If this shift is large enough, the stagnation trap equilibrium disappears.

• In economic terms,this means that only a sufficiently aggressive fiscal intervention can rule

out stagnation traps.

• A timid intervention, in fact, will not do the job (think about a small upward shift of the GG

curve).

40

Fornaro-Wolf Conclusion

• The coronavirus outbreak might cause a persistent supply disruption, which might last far

longer than the pandemic itself (and the associated economic lockdown).

• Fornaro and Wolf’s analysis shows that, in this case, the spread of the virus might cause

a demand-driven slump, give rise to a supply-demand doom loop, and open the door to

stagnation traps induced by pessimistic animal spirits.

• Monetary policy is likely to be insufficient in mitigating the slump induced by the coronavirus

shock.

• Instead, aggressive fiscal policy interventions—in this framework, to support investment and,

more broadly, future productivity capacity—can play a key role in sustaining employment

and growth, by reversing the supply-demand doom loop.

• This is especially true if governments will need to jump-start their economies out of

stagnation traps driven by pessimistic animal spirits.

41

Fornaro-Wolf Conclusion, Continued

• What if the supply shock is short-lived?

• In this case, agents’ expectations about future growth will not be greatly affected, and the

impact on aggregate demand will be small.

• Unfortunately, the pessimistic outcome in which the supply disruption caused by the virus is

going to be severe and protracted is more likely.

– At the time of writing, Fornaro and Wolf said they cannot rule it out. I hope I will be wrong,

but I am increasingly pessimistic.

• If this possibility materializes, Fornaro and Wolf’s simple model suggests that drastic policy

interventions—both monetary and fiscal—will be needed to prevent this negative supply

shock from severely affecting employment and productivity.

• As Gourinchas put it, this is not the time to be cautious.

42

Organized “Throttling Down” and U.S. Economic Performance

• St. Louis Fed President James Bullard has a very interesting perspective on the conduct

of macroeconomic policy in the current circumstances, which he described in a St. Louis

Fed blog on “Expected US Macroeconomic Performance during the Pandemic Adjustment

Period,” which I mostly reproduce below.

• Bullard observes that these actions and policies that have been put in place to combat the

pandemic have had the effect of engineering a controlled, partial, and temporary shutdown

of parts of the U.S. economy.

• This organized “throttling down” radically changes the way we need to think about and

gauge the health of the economy in the near term.

• The U.S. economy will, by design, behave very differently than what is conventionally

assumed in ordinary times—so differently, in fact, that, in Bullard’s view, ordinary business

cycle analysis becomes ineffective and ceases to make sense.

• In the current environment, the goals of macroeconomic policy must be very different, in

some ways the opposite of what we would normally try to accomplish.

43

Bullard’s Proposal: A National Pandemic Adjustment Period

• Bullard’s recommendation would be for the President and Congress to declare a “National

Pandemic Adjustment Period” (NPAP), providing a natural focal point for the expectations of

policymakers and Americans at large concerning what is happening.

• The NPAP would initially extend from now until the end of the second quarter of 2020, and

would be flexible enough to be shortened or extended as necessary depending on how the

virus progresses.

• Special policies would be in effect for the duration of the NPAP, and the dates that these

special policies would expire could be tied to the end date of the NPAP.

44

The Goals of the NPAP

• The NPAP would have three goals:

1. Greatly Reducing Economic Activity.

• The first goal during the NPAP is to intentionally reduce (reduce!) economic activity in order

to meet public health objectives.

• Production is to be carried out only if (1) the good or service is deemed “essential,” or (2)

the good or service can be produced in a way that does not risk transmission of the virus.

• If production is reduced in this way, this will be considered success during the NPAP.

• Bullard’s rough initial estimate of the level of U.S. real GDP (and hence national income)

that meets this public health objective is up to 50% of normal production.

• In other words, the U.S. needs to throttle back the economy to produce at only half its

normal pace.

45

The Goals of the NPAP, Continued

• Important: It would be inappropriate to characterize that outcome as a recession because it

is undertaken intentionally to meet public health objectives.

• In particular, it is inappropriate to argue for “economic stimulus” intending to ramp up

production or create new demand in this situation, as that would work at cross-purposes

with the goal of reducing the level of economic activity in order to meet public health

objectives.

• A better concept is that we should strive to “keep everybody whole” during the NPAP.

46


• A normal quarter of production of goods and services in the U.S. recently, in very round

numbers, is about $5 trillion.

• Producing only half would mean that national income is cut to about $2.5 trillion during the

second quarter of 2020 when the NPAP is in effect.

• This is a quarter-over-quarter drop of 50%, well outside historical experience in the U.S.

• This outcome is expected and temporary and simply reflects the large investment in public

health that will be made in the U.S.

• This change in magnitude is something to be expected and to prepare for, reinforcing the

point that standard business cycle tracking serves little useful purpose in the near term.

• Data during the NPAP will be coming from a special situation.

47


2. Keeping Households and Firms Whole

• The second goal of policymakers is to prevent destruction of livelihoods and firms during the

NPAP.

• This planned, organized partial shutdown will clearly have very uneven effects across

households and firms during the NPAP.

• Some types of businesses are closed down completely, while other types continue to

operate.

• On the household income side, the goal is to keep households whole.

48


• The U.S. already has government income maintenance programs, popularly known as

unemployment insurance (UI).

• These programs should be used extensively and the label on them should change to

“pandemic insurance” (PI) during the NPAP to more appropriately reflect what is happening.

• Heavy use of this facility by individuals—to the extent that it helps to maintain laid-off

workers’ income—should be used as a metric of policy success during the second quarter.

• Heavy use would mean that the government is making the proper transfers to those who

have been disrupted by the health objectives of the country.

• To help accomplish this, benefit replacement rates could be increased substantially from the

current average rate in the U.S. of about 45% to a value close to or equal to 100%.

49


• Moreover, every state has a well-established UI system with rules already in place.

• Stress will be placed on these systems as the number of claims made in the upcoming

weeks may be unprecedented; nonetheless, Bullard views this facility as much better than

the alternative of trying to set up a new system on the fly.

• His initial estimate of the level of pandemic insurance that may be appropriate during the

NPAP period is 30%.

• That is, up to 30% of the workforce could be using this program as part of an optimal policy

response to the pandemic.

50


• The intentional, partial reduction in production means capital will also be unemployed during

the NPAP.

• Conceptually, factories will shut down for a period of time and then reopen once the

pandemic has passed with the capital intact.

• National policy, therefore, needs to make the owners of capital whole during this period.

• Most proposals in this area under consideration in Congress provide loans to businesses,

large and small, so they can start up again after the NPAP.

51


3. Paying for the Pandemic Response

• The third goal is to pay for the pandemic response.

• This is the part of Bullard’s proposal that is least fleshed out.

• He observes that ,if national income falls by 50% during the NPAP, households will not be

able to maintain their normal lifestyles, and consumption is likely to be much lower than

normal.

• Most of this reduction will come as a result of the health objectives themselves—many

avenues for ordinary consumption will simply be closed, and in addition people are being

asked to remain in their homes, resulting in a joint drop of consumption and GDP as an

effect of “hunkering down.”

• Bullard then makes a comment on the fact that the federal government is borrowing

substantially, but most of this is oriented toward maintaining market functioning and

extending loans to businesses.

52


• This is not much information on Bullard’s real thoughts on how the pandemic response

should be paid for.

• At some level, this is not surprising, given Bullard’s position as regional Fed president and,

therefore, “natural” caution in talking about fiscal policy matters.

• My own view?

– This is a time for government borrowing, taking advantage of the very low interest rates

and relying on long-maturity instruments.

– I would not worry about inflation now and I would most definitely defer any tax increases

that may be needed to deal with the longer-term sustainability of accumulated debt until

the economy is back to healthy normal in a sustained fashion.

53

Summarizing Bullard’s View

• Just as incoming macroeconomic data should be interpreted in light of the unprecedented

nature of the public health policy response to COVID-19, so too should the macroeconomic

policies be understood and conducted.

• The phrase “stimulus” may not be entirely appropriate now:

– Many people may not want to fly out of caution or be able to dine out because of legal

decree.

– The goal of macroeconomic policy, at this stage, is not to “stimulate” them to do these

things.

– Rather, at this stage, macroeconomic policy could be better described as maintenance

and support, more a matter of insurance than stimulus.

– For example, enhanced unemployment benefits help maintain the income of workers

temporarily laid off because of a change in demand in the sector where they had been

employed.

54

A Medical Analogy

• I agree with Bullard’s interpretation, which is in many ways consistent with Gourinchas’ view

of the “double curve flattening” exercise.

• We can also think of all this in terms of a medical analogy:

– The U.S. economy and society (and the economies and societies of countries around the

world) are like a patient undergoing a massive surgery that requires the functioning of a

vital organ (the economy—think of it as the heart) to be temporarily suspended.

– We need economic policy to be the extra-corporeal blood circulation machine and the

team of experts and assistants who support the surgeon performing the (public health

operation) and help maximize the probability of successful recovery once the patient

wakes up from anesthesia.

• The Fornaro-Wolf model gives us an initial formal framework for thinking about the

macroeconomic impact of the crisis and the channels through which policies may operate.

• Throughout the quarter, we will discuss how historical events in the functioning of the

international monetary system contributed to shaping the choices of policymakers today.

55



Determination and Intertemporal Sustainability of Net Foreign Debt

Fabio Ghironi


CEPR, and NBER

Introduction

• In the words of Eichengreen’s (2019) Globalizing Capital :

– “The international monetary system is the glue that binds national economies together.

– It lends order and stability to foreign-exchange markets, encourages the elimination of

balance-of-payments problems, and provides access to international credits in the event

of shocks.

– Nations find it difficult to efficiently exploit the gains from trade and foreign lending in the

absence of an adequately functioning international monetary mechanism.

– [...] it is impossible to understand the operation of the international economy without also

understanding its monetary system.”

1


• Understanding the global economic crisis that we are facing and understanding its impact

across countries requires understanding the functioning of the international monetary

system.

• This was shaped by events and economic policy choices that happened over history at least

since the advent of the Gold Standard, with consequences that reverberate to the present

day and inform decisions that policymakers are taking today.

• In order to understand how the international monetary system evolved and functions, we

begin by introducing some key concepts.

2

National Income Accounting and The Balance of Payments

• Consider the definition of GDP from the national income accounts: GDP (Y ) is the sum of

consumption (C), investment (I), government spending (G), and exports (X), minus imports

(IM ):

Yt ≡ Ct + It +Gt +Xt − IMt. (1)

– The subscript t denotes “in period t.” Period t can be a month, a quarter, or a year (or a

different time interval), depending on the time frequency we are focusing on.

– The symbol ≡ denotes the fact that this equation is an identity.

3

National Income Accounting and The Balance of Payments, Continued

Yt ≡ Ct + It +Gt +Xt − IMt.

• Important: Does the fact that we subtract imports mean that imports reduce GDP?

• No! We subtract imports because imports are used in (are part of) consumption, investment,

and government spending.

– In fact, trade in intermediate goods and the phenomenon of so-called global value chains

imply that imports are used also as input in exports.

• Subtracting imports is thus necessary to avoid counting as domestic production (GDP) stuff

that is produced in other countries.

• The difference between exports and imports is known as the trade balance (TB); it is also

referred to as net exports (NX):

TBt ≡ Xt − IMt ≡ NXt. (2)

• Hence, we can equivalently write equation (1) as:

Yt ≡ Ct + It +Gt + TBt. (3)

4


• We can rewrite equation (1) (or (3)) as:

Xt − IMt = Yt − (Ct + It +Gt) . (4)

– The trade balance equals the difference between GDP and the sum of consumption,

investment, and government spending:

– If we produce more than we use for C, I, and G, we run a trade surplus: Xt > IMt or

TBt > 0;

– If we produce less than we use for C, I, and G, we run a trade deficit: Xt < IMt or

TBt < 0.

5


• Now think about the resources the country has during period t:

– The country begins the period with some assets and some liabilities toward other

countries.

· Examples of assets can be bonds issued by foreign governments or stocks issued

by foreign firms; an example of liability is bonds issued by our government held by

foreign agents.

– We are going to use the letter B to denote the country’s net foreign assets other than

central bank reserves of foreign currency:

· The difference between the assets (other than central bank reserves) and liabilities of

the country vis-a-vis the rest of the world.

· Note: For simplicity, I assume that reserves of foreign currency used to implement

regimes of limited exchange rate flexibility are held by the central bank. In reality,

they may be held by a different institution depending on the historical period and

institutional framework of the country we are looking at.

6


• Bt denotes the country’s net foreign assets other than reserves at the beginning of period t.

– Bt > 0 when assets are larger than liabilities, otherwise Bt < 0.

• We denote the central bank’s reserves at the beginning of period t with RESt.

7


• We are going to make a huge simplifying assumption and assume that all assets (including

reserves) and liabilities generate the same net return rt during period t:

– Alternatively, think of rt as a synthetic return measure that aggregates the different

returns across different types of assets and liabilities included in Bt.

– If Bt > 0, rtBt is the income on the country’s net foreign assets (other than reserves)

during period t;

– If Bt < 0, rtBt is the interest burden on the country’s net foreign debt during period t;

– rtRESt ≥ 0 is the income generated by the central bank’s reserves (usually held in the

form of interest-bearing, currency-denominated assets rather than cash).

· In the case of the Gold Standard, reserves also took the form of gold, which would not

generate interest income.

• Finally, the country’s resources in period t of course include the GDP Yt that the country

produces during the period.

8


• Putting things together, the country’s resources in period t are:

Bt + rtBt +RESt + rtRESt + Yt. (5)

• The country can use these resources for consumption, investment, government spending,

and to make asset transactions (purchases of assets, sales of liabilities, reserve accu-

mulation) that determine the levels of B and RES with which the country will begin next

period.

• In other words, it has to be:

Ct + It +Gt +Bt+1 +RESt+1 = (1 + rt)(Bt +RESt) + Yt. (6)

• Now, we can rearrange this equation as:

Bt+1 +RESt+1 = (1 + rt)(Bt +RESt) + Yt − (Ct + It +Gt). (7)

• Or, using equation (3),

Bt+1 +RESt+1 = (1 + rt)(Bt +RESt) + TBt. (8)

• This is the law of motion for the countries total net foreign assets (including reserves): Total

net foreign assets next period are determined by this period’s trade balance plus the (gross)

income on the total net foreign assets with which the country began the current period.

9


• Equation (8) can also be rewritten as:

Bt+1 +RESt+1 − (Bt +RESt) = rt(Bt +RESt) + TBt. (9)

– The change in the country’s total net foreign asset position between the current period

and the next one is equal to the sum of the income balance (the income on the starting

net asset position) and the trade balance during the current period.

– The sum of the trade balance and the income balance is known as the current account

(CA):

CAt ≡ rt(Bt +RESt) + TBt– The current account determines the change in the country’s overall net foreign asset

position.

10


• If we take a step back and we use again the definition of the trade balance (2) and equation

(4), it follows that:

CAt ≡ rt(Bt +RESt) + TBt

≡ rt(Bt +RESt) +Xt − IMt

= rt(Bt +RESt) + Yt − (Ct + It +Gt). (10)

– In this expression, rt(Bt + RESt) + Yt − (Ct + Gt) measures the country’s total savings

(S): The country’s income (its GDP plus the net income from its asset holdings) minus its

consumption (private consumption and government consumption):

St ≡ rt(Bt +RESt) + Yt − (Ct +Gt).

– It follows that the current account is also equal to the difference between the country’s

savings and investment:

CAt = St − It. (11)

– When a country runs a CA deficit (or surplus), it is because the country’s savings are

lower (or higher) than the country’s investment.

11


Flexible Exchange Rates

• Now suppose that the country we are looking at operates a flexible exchange rate regime

(like the U.S. since the early 1970s).

• In this case, RESt+1 = RESt = 0, and equation (9) becomes:

Bt+1 −Bt = rtBt + TBt. (12)

• For a country that operates a flexible exchange regime, the current account (CAt ≡rtBt+TBt here) has to be equal to the negative of the capital account (KAt ≡ −(Bt+1−Bt)).

12


• Why is Bt+1 −Bt the negative of the capital account?

• Suppose that the country is running a current account surplus: CAt ≡ rtBt + TBt > 0.

• This means that the country’s net foreign assets are increasing: Bt+1 −Bt > 0.

• In other words, the country is buying foreign assets: domestic capital is flowing to foreign

countries (a capital outflow, or a capital account deficit, KAt < 0) as domestic agents buy

foreign assets (liabilities for foreigners), thus lending to foreign countries.

– Note: A capital outflow also happens when foreigners sell their holdings of the assets of

the country we are considering.

13


• If we run a CA deficit (like the U.S. has been doing for quite some time), we have a capital

inflow (or a KA surplus):

– Foreigners lend to us (finance our deficit) by buying U.S. assets (bonds, stocks—which

become foreign liabilities for us).

– The U.S. net foreign asset position worsens (net foreign debt becomes larger).

• Under flexible exchange rates, current account and capital account add up to zero:

0 = KAt + CAt. (13)

– Current account surpluses (or deficits) are mirrored by capital account deficits (or

surpluses) or capital outflows (or inflows).

14


Fixed or Managed Exchange Rates

• If the country runs a fixed exchange rate regime (or another regime of limited exchange rate

flexibility that requires the use of reserves), the balance of payment equation (9) implies:

RESt+1 −RESt = −(Bt+1 −Bt) + rt(Bt +RESt) + TBt, (14)

or:

RESt+1 −RESt = KAt + CAt. (15)

• Under fixed exchange rates, the sum of capital account and current account determines the

change in the country’s foreign currency reserves.

• In this environment, it is possible for countries to face the simultaneous combination of

a current account deficit (CAt < 0, typically the result of TBt < 0) and a capital outflow

(KAt < 0).

• This is the combination of circumstances that is usually associated with most rapid loss of

central bank reserves.

– KAt + CAt < 0 implies RESt+1 < RESt: The central bank’s reserves are declining.

15


• When a country runs out of reserves, it must let go of the commitment to pegging the

exchange rate.

• Why? Because the central bank uses reserves to keep the exchange rate stable.

• Suppose there is pressure on the currency to depreciate—as it usually happens in the

presence of trade deficit and capital outflow

• The central bank counteracts the pressure by soaking (buying) its own currency in exchange

for reserves.

– This is another way of looking at the fact that KAt + CAt < 0 implies RESt+1 < RESt.

• But a central bank’s reserves of foreign currency (specifically, of the currency against which

the central bank is trying to keep the exchange rate stable) are not infinite.

• Sufficiently large reserve loss (say, as a consequence of a sufficiently large speculative

attack—the sale of domestic-currency-denominated assets by speculators) will cause the

peg to collapse.

16


• We will use this framework many times during the quarter to discuss the historical

experiences of countries at different times in history and under different exchange rate

arrangements.

• When we do, we will sometimes refer to the concept of intertemporal sustainability of a

country’s net foreign debt.

• This is a very important concept, central to understanding crises and the concerns of

policymakers in many countries in the current situation.

17

From the Balance of Payments Equation to Intertemporal Sustainability

• Recall the balance of payments (BOP) equation we just discussed:

RESt+1 −RESt = − (Bt+1 −Bt) + rtRESt + rtBt + TBt. (16)

– The change in central bank reserves is equal to the sum of the capital account

(KAt ≡ − (Bt+1 −Bt)) and the current account (CAt ≡ rtRESt + rtBt + TBt, where TBtis the trade balance).

– TBt = Yt − (Ct + It − Gt): The trade balance is the difference between domestic

output and absorption (given by the sum of consumption, investment, and government

spending).

– Bt+1 − Bt is the change in net foreign assets other than reserves (or the change in net

foreign debt if Bt is negative).

18

From the Balance of Payments Equation to Intertemporal Sustainability,

Continued

RESt+1 −RESt = − (Bt+1 −Bt) + rtRESt + rtBt + TBt.

• Rearrange equation (16) as:

RESt+1 +Bt+1 = RESt +Bt + rtRESt + rtBt + TBt,

or:

RESt+1 +Bt+1 = (1 + rt) (RESt +Bt) + TBt. (17)

• Define the country’s total net foreign assets as At ≡ RESt +Bt.

• Then, equation (17) becomes:

At+1 = (1 + rt)At + TBt. (18)

19


Continued

• Now, to simplify the analysis that follows, assume that the interest rate is constant in all

periods: rt = r ∀t.

• Therefore, equation (18) becomes:

At+1 = (1 + r)At + TBt. (19)

• Notice that the BOP equation (16), or its version (19) given our assumptions, must hold in

very period t, t + 1, t + 2, and so on.

• Thus, we can scroll equation (19) forward by one period to obtain:

At+2 = (1 + r)At+1 + TBt+1. (20)

• Solve this equation for At+1:

At+1 =At+21 + r

− TBt+11 + r

. (21)

20


Continued

At+1 =At+21 + r

− TBt+11 + r

.

• And substituting this expression for At+1 into (19), we have:

At+21 + r

− TBt+11 + r

= (1 + r)At + TBt,

or:

At+21 + r

= (1 + r)At + TBt +TBt+11 + r

. (22)

21


Continued

• Now consider equation (20) scrolled forward by one period:

At+3 = (1 + r)At+2 + TBt+2.

• Hence,

At+2 =At+31 + r

− TBt+21 + r

.

• And substituting this expression for At+2 into (22), we have:(At+31+r −

TBt+21+r

)1 + r

= (1 + r)At + TBt +TBt+11 + r

,

or:

At+3

(1 + r)2= (1 + r)At + TBt +

TBt+11 + r

+TBt+2

(1 + r)2. (23)

22


Continued

• If we use the same type of substitution repeatedly, we get:

At+T

(1 + r)T−1= (1 + r)At + TBt +

TBt+11 + r

+TBt+2

(1 + r)2+ ... +

TBt+T−1

(1 + r)T−1, (24)

where T is a large number.

• We can use the summation symbol∑

to write equation (24) more compactly as:

At+T

(1 + r)T−1= (1 + r)At +

t+T−1∑s=t

TBs

(1 + r)s−t. (25)

– To see how this works, observe that is you set s = t in TBs(1+r)s−t

, you get TBt; if you set

s = t + 1 in TBs(1+r)s−t

, you getTBt+11+r ; if you set s = t + 2 in TBs

(1+r)s−t, you get

TBt+2(1+r)2

; and so on.

23


Continued

• Now let us “unpack” reserves and other net foreign assets by recalling that At ≡ RESt +Bt.

• Hence, we can rewrite equation (25) as:

RESt+T

(1 + r)T−1+

Bt+T

(1 + r)T−1= (1 + r) (RESt +Bt) +

t+T−1∑s=t

TBs

(1 + r)s−t. (26)

• Let T →∞ and consider the limit:

limT→∞

RESt+T

(1 + r)T−1+ limT→∞

Bt+T

(1 + r)T−1= (1 + r) (RESt +Bt) +

∞∑s=t

TBs

(1 + r)s−t. (27)

• Realistically, we can assume that the central bank’s reserves are always finite.

• It follows that

limT→∞

RESt+T

(1 + r)T−1= 0,

because (1 + r)T−1 →∞ if T →∞.

24


Continued

• Therefore, equation (27) becomes:

limT→∞

Bt+T

(1 + r)T−1= (1 + r) (RESt +Bt) +

∞∑s=t

TBs

(1 + r)s−t. (28)

• The long-run behavior of the country’s net foreign asset (or debt) position (other than

reserves) is determined by the sum of its initial total net foreign asset position, plus interest

on it, plus the present discounted value (PDV) of the country’s trade balances from the

current period to the infinite future.

• Equation (28) proves a very important result about the intertemporal sustainability of a

country’s net foreign asset (or debt) position (other than reserves).

25

Intertemporal Sustainability

• For example, consider the case in which the country runs a net foreign debt (B is negative).

• Net foreign debt not exploding to∞ faster than the interest rate, i.e.,

limT→∞

Bt+T

(1 + r)T−1= 0, (29)

is equivalent to the following condition being satisfied:

0 = (1 + r) (RESt +Bt) +

∞∑s=t

TBs

(1 + r)s−t. (30)

• Put differently, suppose Bt < 0 and the country runs a trade deficit today and for some time

in the future.

• To avoid its net foreign debt eventually exploding to ∞, it better be that the country runs

surpluses later in the future, so that the intertemporal sustainability condition (30) is satisfied.

• Note that the larger is RESt, the longer the country can run deficits, but eventually surpluses

are needed for (30) to be satisfied.

26

Intertemporal Sustainability, Continued

• So,

RESt+1 −RESt = − (Bt+1 −Bt) + rtRESt + rtBt + TBttells us how the country’s reserve position changes between t and t + 1.

•

0 = (1 + r) (RESt +Bt) +

∞∑s=t

TBs

(1 + r)s−t

is the condition that ensures that the country’s net foreign asset (or debt) position is

intertemporally sustainable.

• Both equations must be satisfied to avoid crises.

27

Intertemporal Sustainability, Continued

• With flexible exchange rates, these two equations become:

Bt+1 −Bt = rtBt + TBt

and

0 = (1 + r)Bt +

∞∑s=t

TBs

(1 + r)s−t.

28

Intertemporal Sustainability and Growth

• The analysis above abstracts from growth of the economy. In reality, once GDP growth is

accounted for, what matters for sustainability is the debt/GDP ratio, the trade balance/GDP

ratio, and the difference between interest rate and growth rate of GDP.

• The faster is economic growth, the easier it is to sustain a debt position intertemporally.

• We understand this point by relying on material from Maurice Obstfeld and Kenneth Rogoff’s

Foundations of International Macroeconomics (MIT Press, 1996).

29

Intertemporal Sustainability and Growth, Continued

• Rewrite the law of motion for total net foreign assets At ≡ Bt +RESt as:

At+1 = (1 + r)At + TBt. (31)

• Assume that GDP grows at rate g in every period, so that:

Yt+1 = (1 + g)Yt. (32)

• Now divide both sides of (31) by Yt+1 to obtain:

At+1Yt+1

= (1 + r)AtYt+1

+TBtYt+1

. (33)

30


At+1Yt+1

= (1 + r)AtYt+1

+TBtYt+1

• Note that this equation can be rewritten as:

At+1Yt+1

= (1 + r)AtYt

YtYt+1

+TBtYt

YtYt+1

(34)

because all we did was to multiply and divide by Yt both terms on the right-hand side of

equation (33).

• But our assumption about growth, equation (32), implies that

YtYt+1

=1

1 + g.

• Hence, equation (34) becomes:

At+1Yt+1

=

(1 + r

1 + g

)AtYt

+

(1

1 + g

)TBtYt. (35)

31


• Now define the ratio of total net foreign assets over GDP as at ≡ At/Yt and the ratio of the

trade balance to GDP as tbt ≡ TBt/Yt.

• It follows that:

at+1 =

(1 + r

1 + g

)at +

tbt1 + g

. (36)

• This equation is important to understand the role of growth.

32


• Suppose that the government wants to stabilize the country’s overall net foreign asset

position as percentage of GDP at a constant level a

at+1 = at = a.

• To keep the ratio of total net foreign assets to GDP constant it would be the case that At has

to grow at the same rate as Yt, i.e., that it is:

At+1 = (1 + g)At.

• Equation (36) implies that, in order to accomplish the goal of constant total net foreign

assets (as percentage of GDP) at the level a, also the trade balance (as percentage of GDP)

must be constant, at the level tb implied by:

a

(1− 1 + r

1 + g

)=

tb

1 + g,

or:

tb = − (r − g) a. (37)

33


tb = − (r − g) a

• Another way to state this equation is that, for at to be constant at the level a, it has to be:

TBtYt

= − (r − g)AtYt.

• This equation states that, if GDP is growing at rate g, and we assume r > g, in order to keep

net foreign debt constant as a ratio to GDP (assuming that Bt < 0 and RESt is not so large

to turn At positive), the country only needs to run surpluses that allow it to pay out only the

excess of the interest rate over the growth rate.

• Thus, the faster GDP grows (the higher g), the easier it is for the country to stabilize its

foreign debt for given interest rate r.

34


• The ratio

− (r − g)AtYt

measures the burden imposed on the economy by foreign debt.

• The higher this burden, the higher the probability that debt is unsustainable (i.e., the country

will be unable or unwilling to repay).

35


• A similar argument applies to government debt and the government’s deficit or surplus, and

this is the reason why Italian government bonds came under pressure during the euro area

crisis that began in 2010.

• Italian governments accumulated a very large stock of debt, now above 130 percent of GDP,

starting in the 1980s.

• Italy’s growth performance has been abysmal since the late 1990s.

• In an environment of very low growth, larger government surpluses are necessary to

stabilize government debt.

• Put differently, even relatively small deficits can cause the debt/GDP ratio to rise rapidly.

• In 2010 markets lost faith in the ability of Italy’s governments to sustain the large accumulated

stock of debt in an environment of very low growth.

36


• Note that debt sustainability (be it net foreign debt or government debt sustainability) is of

course easier to accomplish when interest rates are very low.

• The analysis above implies that if r < g it would be possible to keep total net foreign debt

constant even if the economy were running a constant trade deficit as percentage of GDP:

– The required trade balance/GDP ratio

− (r − g)AtYt

would be negative when g > r.

• This may lead some to think that debt is a free lunch when interest rates are very low and

g > r.

37


• But we need to be very careful here:

1. It still has to be the case that growth is sufficiently high.

– In reality, the growth rate is an endogenous object, affected by policy choices.

– It is difficult to establish clear thresholds, but excessive reliance on debt could have

negative consequences for growth.

2. The interest rate r is also an endogenous (and time-varying) object that is affected by

policy decisions.

– Excessive reliance on debt could eventually push r into dangerous territory for debt

sustainability.

38


• The current circumstances of the global economy, combined with an environment of very

low interest rates also thanks to the actions of central banks, make it such that large

debt-financed fiscal expansion is globally desirable.

• But it will be important for this debt financing not to become a dangerous addiction once the

crisis is over.

39



What If We Were Still on the Gold Standard?

Fabio Ghironi


CEPR, and NBER

Introduction

• We discussed some perspectives on the ongoing global crisis and the framework to

understand key policymaker concerns about dynamics and sustainability of debt positions.

• Our next step is to start exploring lessons from history for today’s policymakers.

• For instance: In the recent past, members of the U.S. Congress argued that the U.S. should

return to the Gold Standard (GS).

• But the Fed and the U.S. government be in a better position to fight the current crisis if that

decision had been taken and implemented?

• The answer is going to be a resounding no, but getting there requires us to understand how

the GS worked and the lessons that were learned between its advent as the predominant

international monetary regime in the 1870s and its eventual collapse at the beginning of the

1930s.

1


• These slides/notes draw on Eichengreen and Flandreau (1997, edited volume), a 1991

Quarterly Journal of Economics article by Paul Krugman, my own interpretation, and

occasional additional references.

• You would learn many of the same insights (but not all) by reading Eichengreen’s (2019)

Globalizing Capital.

• I start by focusing on how the GS functioned.

• Appendix A of these slides gives you information on how the GS came to become the

predominant international monetary regime in the 1870s: a combination of chance, path

dependence, network externalities, and geopolitical considerations.

2

How Did the Gold Standard Function?

• The GS did not operate in the same way everywhere; it took different forms across

countries—for instance, in the extent of monetary gold circulation or in how much of the

country’s reserves backing its banknotes were held in gold versus bonds denominated in

convertible currencies.

– If I am certain that a country’s currency is convertible into gold, I may choose to hold part

of my reserves in bonds denominated in that currency because they generate interest

income; gold does not.

• We begin the discussion of this question by talking about price stability.

3

Price-Level Dynamics under the Gold Standard,

• John Stuart Mill described in 1865 the mechanism by which the GS was supposed to

preserve price stability.

• He emphasized the link between money supplies and gold stocks and the tendency for the

flow of newly mined gold to respond to changes in the price level.

• Think about the simplest version of a world money market equilibrium condition:

Mt

Pt= Yt,

where Mt is the world quantity of nominal money, Pt is the world price level, and Yt is world

GDP.

– According to this equation, known as the quantity theory of money, money demand (equal

to money supply in equilibrium) is determined by the nominal amount of transactions:

Mt = PtYt.

4

Price-Level Dynamics under the Gold Standard, Continued

• Now suppose that the world economy is expanding (GDP is growing), but the supply of

money is fixed, pinned down by the world stock of gold: Mt = M .


M

Pt= Yt.

• And you immediately see that as money demand for money rises, downward pressure is

placed on prices (“too little money chasing too many goods”):

– If Yt rises, Pt has to fall to preserve money market equilibrium.

5


• Now consider the relative price of gold, denoted ρGOLDt . This is given by:

ρGOLDt ≡ pGOLDt

Pt.

– pGOLDt is the price of gold: the number of units of currency it takes to buy an ounce of

gold.

– Pt tells you how many units of currency it takes to buy a unit of the GDP goods bundle.

– Hence:

ρGOLDt ≡ pGOLDt

Pt=

Units of currencyGold

Units of currencyGoods

=Goods

Gold.

• ρGOLDt tells you how much gold is worth in terms of goods: how many units of good it takes

to trade for an ounce of gold.

6


ρGOLDt ≡ pGOLDt

Pt=

Goods

Gold.

• Let us assume that the price of gold is successfully pegged by the “world central bank (or

government)” at the conversion rate (pGOLD) at which the authority is committed to convert

the currency for an ounce of gold.

– When talking about convertibility under the GS, we refer to the commitment to convert

the currency for gold at the rate determined by the relevant authority.

• Given the conversion ratio pGOLDt = pGOLD, if the price level Pt is falling, it follows that the

relative price ρGOLDt is rising:

– An ounce of gold is worth more in terms of goods.

• This increase in the real, relative value of gold will generate a stronger incentive for mining,

eliciting additional supply of the metal.

• In turn, the larger gold stock will imply an increase in money supply (Mt will rise to M ′ > M ),

limiting the downward pressure on the price level.

7


• Even Mill admitted that time was required for the mining response to occur.

• For instance, it took more than 20 years of deflation (1873-1896) before gold discoveries

in the Klondike and elsewhere reversed the deflation trend and inaugurated a period of

inflation.

• The long deflation, hurting borrowers whose liabilities increased in real value, is the main

reason why, for instance, farmers in the U.S. (who relied on mortgage financing whose

burden became heavier with deflation) strongly opposed the GS.

• Moreover, several economists emphasized the randomness of gold finds, arguing that

chance discoveries dominated relative prices as a determinant of gold production.

• Alfred Marshall, for instance, was dubious that long-run price stability would be provided by

a system dependent on the “hazards of mining.”

8


• Scholars studied how the GS did in terms of price stability in comparison to later monetary

regimes.

• Michael Bordo (1981) found more moderate inflation under the GS than after World War II

(WWII) for Britain and the U.S., but the relatively low average inflation during the GS is partly

the consequence of two decades of deflation.

• In two other studies (1981 and 1993), Bordo compared the volatility of inflation under the

GS to that in the post-WWII period for a number of countries, finding no conclusive evidence

in favor of the GS.

• Others trying to evaluate comparative price predictability across the GS and other regimes

reached mixed conclusions.

• Robert Barsky (1987), however, found that inflation persistence was lower under the GS

than after WWII, a conclusion supported also by results in Bordo’s work with Finn Kydland

(1995).

9

Exchange Rates and International Adjustment under the Gold Standard

• Absent clear-cut superiority of the GS in terms of price dynamics performance, the

outstanding feature of the GS is the extent of exchange rate (ER) stability.

• Crises did happen and forced countries to suspend convertibility of their currencies into gold

at various points.

• Yet, it is striking that none of the major powers—for instance, in North America and Western

Europe—was forced to depart from the GS for any extended time.

• ERs between the core countries of the system were impressively stable.

• Although there were financial and commodity price shocks, balance of payment (BOP)

imbalances were absorbed without destabilizing ERs or otherwise undermining the stability

of the GS.

10

Exchange Rates and International Adjustment under the Gold Standard,

Continued

• This ease of adjustment should perhaps not be credited to the GS itself.

• It could have been nothing more than a reflection of underlying economic conditions.

• In this view, causality ran not from stable ERs to the rapid growth of trade and incomes.

• Rather, ER stability and smooth BOP adjustment were the consequence of industrialization,

which stimulated incomes and trade.

• Nevertheless, generations of scholars have insisted that the GS played a more active role in

the observed ease of adjustment.

• How did this happen?

11

Exchange Rates and International Adjustment under the Gold Standard,

Continued

• Three main approaches to understanding the functioning of international adjustment under

the GS can be distinguished:

1. One emphasizes the efficiency and automaticity of the adjustment mechanism.

2. A second one focuses on the disciplining effect of the GS on fiscal and monetary policies.

3. A third one highlights the combination of policy credibility and flexibility that the GS allowed.

• All three approaches are important to understand features of the system.

12

The Efficiency View of the Gold Standard

• The traditional explanation for ease of BOP adjustment and ER stability under the GS

focuses on the efficiency and automaticity of the adjustment mechanism.

David Hume’s (1752) Price-Specie Flow Model

• Consider a world economy with 2 countries (Home and Foreign) and 2 commodities: a

homogeneous consumption good and gold.

• Suppose there is a “shock” in the Home country—say, a one-time increase in its stock of

gold.

• At initial prices (the prices that prevailed before the shock), there will be excess supply of

gold and excess demand of the consumption good in Home.

• Hence, prices must adjust to restore equilibrium.

13

The Efficiency View of the Gold Standard, Continued

• The benchmark price-specie flow model assumes that transactions occur initially among

domestic residents.

• As residents of Home simultaneously attempt to sell gold and buy consumption, the price of

consumption relative to gold rises.

• The price of consumption in terms of gold is now higher in Home than in Foreign.

• Home agents have an incentive to buy the consumption good from abroad, where it is

relatively cheap.

• Foreigners, for their part, have an incentive to obtain gold from Home, where its price (in

terms of consumption) is low.

• Consumption is shipped from foreign to home and gold is shipped from home to foreign.

14


• Absent changes in production, foreigners must reduce their consumption spending to make

a surplus of consumption available for exports.

• Home residents, meanwhile, increase their consumption spending, absorbing imports.

• Home runs a trade deficit whose corollary is loss of gold; Foreign runs a trade surplus

mirrored by an increase in its gold holdings.

15


• In terms of the formal framework we introduced in the previous set of slides, the Home BOP

equation is:

RESt+1 −RESt = TBt.

– There is no asset in this simple model (hence, Bt+1 = Bt = 0) and gold yields no interest

(hence, rtRESt = 0).

– I am assuming that residents of each country hand gold over to the respective central

bank or government in exchange for banknotes of equivalent value, so flows of gold

between countries as agents trade are equivalent to changes in the authority’s holdings

of gold reserves backing the currency.

16


RESt+1 −RESt = TBt.

• As a consequence of the shock, TBt < 0 and RESt+1 < RESt.

• In Foreign (denoted with a star), a similar equation holds, and it is TB∗t > 0 and

RES∗t+1 > RES∗t .

• Note that in a world of two countries, TBt = −TB∗t :

– A country’s trade surplus must coincide with the other country’s deficit.

• Therefore, RESt+1 −RESt = − (RES∗t+1 −RES∗t ).

– Home’s reserve loss must coincide with Foreign’s gain.

17


• Important: Adjustments in the markets for consumption and gold are two sides of the same

coin.

• Another way of seeing this is by observing that in a world with only two commodities, there

exists only one relative price.

• So, a rise in one commodity price is the same as a fall in the other.

• When one market clears, so must the other, as a result of Walras’ Law.

18


• How long does the adjustment to the shock take?

• Home will run a trade deficit and lose gold to Foreign until the prices of the consumption

good and gold (or the relative price of gold in terms of goods) are equalized in the two

countries.

• At that point, the trade balance returns to zero (assuming that was its level when the shock

happened) and each country’s gold holdings stay constant.

19


• Hume intended his model as analytical device rather than description of reality.

• He was aware, for instance, that transactions do not occur first within and only then between

nations.

• In addition, arbitrage in both gold and consumer goods markets implies that prices will

deviate from the law of one price only to the extent that there are tariff barriers and transport

and insurance costs (McCloskey and Zecher, 1982).

– The law of one price (LOP) states that the price of a given good (or of gold) in two

different locations (such as countries) has to be the same once it is expressed in the

same units.

– If it does not hold, there is an incentive for arbitrage.

– Hence, the force of arbitrage implies that LOP deviations will be observed only to the

extent that arbitrage involves costs and/or markets are segmented.

• But departures of same-item prices from cross-country equalization are central to the theory

we explained.

20


• Studies early in the 20th century documented that relative prices of goods across countries

(the terms of trade) rarely behaved as predicted by the price-specie flow model.

• This—and the fact that the benchmark model did not feature any role for capital flows—

induced subsequent analyses to stress the role of interest rate differentials and capital flows

in the elimination of BOP imbalances.

21


Price-Specie and Capital Flows

• Britain’s Cunliffe Committee (1919), recommending that the GS be restored as soon as

possible after World War I (WWI), described its operation in terms of a price-specie flow

model extended to allow for capital flows and interest rate differentials.

• Consider the same thought experiment as before (a one-time increase in Home’s stock

of gold) under the simplifying assumption that transactions take place first among Home

residents.

• The initial excess supply of gold now has its counterpart in excess demand for both

consumption and financial assets.

• The domestic prices of consumption and assets increase, and the domestic price of gold

falls.

• Home residents have an incentive to obtain consumption and assets from Foreign, while

Foreigners have an incentive to obtain gold from Home.

22


• The value of Home’s net exports of gold equals the value of its net imports of consumption

and financial assets.

• It is no longer the trade deficit that is equal to the outflow of gold—it is the BOP deficit (trade

deficit plus capital outflow) that is equal to the transfer of gold.

• In terms of the formal framework we introduced, we now have that the BOP equation:

RESt+1 −RESt = KAt + CAt

is at work. Or,

RESt+1 −RESt = − (Bt+1 −Bt) + rtBt + rtRESt + TBt,

where I assumed again that gold is handed over to the central bank (or government), and

I also assumed that this can use some of its gold to buy Foreign-currency-denominated,

interest-bearing assets.

23


RESt+1 −RESt = − (Bt+1 −Bt) + rtBt + rtRESt + TBt.

• Note that now it has to be TBt = −TB∗t and Bt = −B∗t in each period (i.e., also

Bt+1 = −B∗t+1):

– Home’s trade deficit mirror’s Foreign’s surplus, and Home’s net foreign assets (or debt)

must mirror Foreign’s net foreign debt (or assets).

– Home and Foreign reserves do not add up to zero; they add up to the total stock of world

gold, which increased as a consequence of the shock.

– But it still has to be the case that RESt+1 −RESt = − (RES∗t+1 −RES∗t ).

24


• It is indifferent whether we describe the adjustment process in terms of price or interest rate

effects, since saying that financial asset prices are higher at home than abroad following the

increase in the stock of gold is the same as saying that interest rates are lower at home than

abroad.

– Remember that the interest rate on Home bonds between t and t + 1 (denoted with it) is

such that

1 + it =1

P bt

,

where P bt is the domestic currency price of the bonds.

25


• Capital flows from the country where interest rates are low (asset prices are high) to where

they are high (asset prices are low), until asset prices and interest rates are equalized

internationally.

• Goods flow from the country where prices are low (Foreign) to the country where they are

high (Home), until good prices are equalized.

• Adjustment to different types of asymmetric shocks across countries (different from the

Home gold discovery we focused on) would work similarly:

– The shock would result in initial price and interest rate discrepancies that would be

absorbed over time as a result of international flows of goods and capital.

26


• We already mentioned that there was not evidence of persistent deviations from price

equalization for goods (larger and slow-to-revert deviations than those resulting from

transport costs, etcetera).

• Early discussions posited that capital flowed gradually in the GS adjustment process,

eliminating interest rate differentials only slowly.

• But later studies (Dick and Floyd, 1992; Calomiris and Hubbard, 1996) pointed out that

observed interest rate differentials were negligible, and quick responses of capital flows

were dominant in adjustment.

• So, while the forces of the price-specie flow model must have been relevant to the functioning

of the GS, they cannot really explain it.

• We must turn somewhere else to continue building our understanding.

• We do this by considering a view of the system that emphasizes discipline or so-called

“rules of the game.”

27

The Discipline View of the Gold Standard

• Economists have long argued that commitment to gold convertibility of the currency served

as a check on inflationary finance.

• According to this view, the key to international adjustment and ER stability was that sound

public finances were an integral part of the GS ideology.

• Governments ran balanced budgets, which insulated central banks from pressure to

purchase government bond and inject currency into the economy.

• This was consistent with the commitment to convertibility, since persistent budget deficits,

leading to monetization and inflation, would have exhausted the central bank’s reserves and

forced the abandonment of convertibility.

28

The Discipline View of the Gold Standard, Continued

• Adherence to the GS became a signal of financial probity, with lenders charging lower

interest rates to countries that adhered to the system.

– Even if convertibility was a signal of probity, it did not prevent countries from running

deficits:

· Some countries (Portugal, Argentina) moved off gold as a result of deficit spending.

• Even for core countries, the GS as a commitment device for fiscal policy is best interpreted

as a contingent rule (Bordo and Kydland, 1995), allowing for deficit financing of government

spending and suspensions of convertibility in exceptional circumstances (such as wars).

29


• Most importantly, the discipline view can also be applied to monetary policy.

• Scholars who highlight this point out that, whatever the behavior of fiscal policy, central

banks saw themselves as responsible for doing what was necessary to defend their

reserves.

• Thus, they adjusted policy interest rates to maintain external balance.

• If an outflow of gold reserves was beginning or about to happen, the central bank responded

immediately by increasing the discount rate.

– The discount rate is the rate at which the central bank lends to other banks.

– These must pledge collateral to obtain liquidity from the central bank, and the central

bank discounts the valuation of this collateral.

– The rate at which it does so is the discount rate.

– So, the higher the discount rate, the harder it is for banks to obtain liquidity from the

central bank.

30


• A higher discount rate put upward pressure on interest rates, reduced the availability of

credit, and discouraged interest-sensitive spending.

• As spending fell, the trade balance improved.

• As financial conditions tightened, capital flowed in from abroad (attracted by higher returns).

31


• In 1925, John Maynard Keynes called the practice of central banks to adjust policy to

prevent reserve loss playing by the GS “rules of the game.”

• Behavior by these rules of the game would help explain why gold flows of the size predicted

by the price-specie flow mechanism were rarely observed.

• A country experiencing a BOP deficit had to adjust by cutting its spending.

• In the price-specie flow model, the mechanism prompting this was a gold outflow, which

would eventually force a reduction of money supply and tighten financial conditions.

• But a central bank playing by the rules could produce the same effect without waiting for any

significant gold movement.

• At the first sign of gold losses, it would raise the discount rate, producing the required

financial tightening, without need for significant gold flows.

32


• In terms of the tools we introduced, recall again the BOP equation and our discussion of the

price-specie flow model after a shock that caused the prices of Home assets and goods to

rise:

RESt+1 −RESt = − (Bt+1 −Bt) + rtBt + rtRESt + TBt

= − (Bt+1 −Bt) + rtBt + rtRESt + Yt − (Ct + It + Gt) .

• When capital outflow (Bt+1 − Bt > 0) and trade deficit (TBt < 0) threaten to reduce reserve

holdings, the central bank contracts monetary policy.

• The higher discount rate causes Home interest rates to rise (Home asset prices to fall)

reducing the incentive to buy assets issued by Foreign, choking the capital outflow Bt+1−Bt.

• Higher Home interest rates also reduce domestic consumption and investment, choking

appetite for imports and the trade deficit.

33


• Therefore, behavior according to the rules of the game requires the Home central bank to

impose a contraction on the economy whenever capital outflow and trade deficit threaten

reserves.

• If the Foreign central bank wants to act cooperatively, it can help by lowering its own discount

rate and expanding its monetary policy.

– Home capital outflow and trade deficit are the result of Home interest rates having fallen

below Foreign and goods prices having risen above, i.e., the imbalances are the result of

cross-country differences in asset and good prices.

– These are narrowed (or eliminated) by a Home contraction, but also by a Foreign

expansion, which lowers Foreign interest rates and expands the Foreign economy.

• In a cooperative environment, the imbalances will be eliminated with both central banks

doing part of the effort.

• As we shall see, cooperation (or lack of cooperation) between authorities of different

countries can have very important consequences.

34


• Did the GS really function as described by the discipline/rules-of-the-game story?

• The problem with this story is that evidence of behavior by the rules was week.

• In 1959, Arthur Bloomfield found that pre-WWI central banks violated the rules in the

majority of years and countries he considered.

• Rather than draining liquidity from the market when their reserves declined (and augmenting

it when they rose), they frequently did the opposite.

• So, we need to take another step to complete our understanding of how the GS worked.

35

The Modern Synthesis: The Gold Standard as a Credible Target Zone

• Bloomfield’s study identified an apparent paradox:

– If central banks were committed to defending the gold parity, how could they at the same

time often violate the rules of the game?

• The answer is that the GS implied not pegged ERs, but rather a narrow band within which

ERs could move.

• Even this narrow band endowed authorities with significant monetary autonomy.

• In the terminology of the 1990s ERs literature, the GS was not a system of currency pegs.

• It was a set of target zones.

• To understand this, we must explain two of the GS’s technical features: the gold points and

gold devices.

36

The Modern Synthesis: The Gold Standard as a Credible Target Zone, Cont.

The Gold Points

• The 19th century gold market was highly developed, but it operated subject to transaction

costs.

• These meant that ERs could fluctuate within narrow margins, known as the gold export and

import points, without offering gold traders arbitrage profits.

• To see this, consider first a world without transaction costs.

• The Bank of England stood ready to convert a pound sterling into an ounce of (11/12 fine)

gold on demand.

• The U.S. Treasury committed to paying out an ounce of gold of equal purity for $4.86.

• Thus, the ER was locked at $4.86 per sterling: the level implied by the ratio of the dollar/gold

parity to pound sterling/gold parity:

dollargold

pound sterlinggold

=dollar

pound sterling.

37


• If the dollar strengthened even slightly (say, so it took $4.85 to buy a pound sterling), there

would be an incentive to

– buy gold from the Bank of England for 1 sterling,

– ship it to the U.S.,

– sell it to the U.S. Treasury for $4.86,

– and convert those dollars into sterling on the foreign exchange market, obtaining more

than 1 sterling (1.0021 sterlings, to be precise).

• This would increase the demand for sterling and reduce that for dollars until equilibrium was

restored, which would happen instantaneously (and without any ER deviation from 4.86) in

a world without transaction costs.

38


• In practice, arbitragers had to pay to ship gold across the Atlantic, to insure it while in

transit, and to borrow the funds needed for its purchase (or forego interest earnings on bank

deposits in the meantime).

• These expenses amounted to approximately 2.5 percent of the value of gold transactions in

the 1840s.

• They dropped to approximately .5 percent before WWI, due to technological progress in

shipping and other markets.

• The dollar/sterling ER could move away from the parity implied by the ratio of the U.S. and

British gold conversion rates by at most this amount before there was an incentive to engage

in arbitrage.

– Note that arbitrage would act as a stabilizing force on the exchange rate, preventing it

from moving further and pushing it back toward the central parity, if the authority whose

gold was being bought did not run out of it.

39


• The bounds of the band around parity within which the ER could fluctuate without triggering

arbitrage were known as the gold export and import points.

• With the ER measured as dollar/sterling, the upper bound would be the U.S. gold export

point and the lower bound would be the U.S. gold import point:

– The dollar would have to depreciate above the gold export point (or appreciate below the

gold import point) for arbitrage and gold flow to happen.

40


Gold Devices

• In 1959, Oskar Morgenstern used estimates constructed by Ernest Seyd, a bullion dealer

who had authored a study on international financial markets in 1868, to argue that exchange

rates between major financial centers had repeatedly strayed outside the bands defined by

the gold points.

• This led some scholars to conclude that the gold market was inefficient, since arbitrage

failed to prevent violations of the gold points.

• But Lawrence Officer, in a series of studies between the 1980s and 1990s, explained away

these anomalies by challenging Morgenstern’s assumption that central banks always sold

gold at the Mint price.

41


• The Bank of England, for instance, although obligated to buy gold at a fixed sterling price

and redeem its notes with gold coins with the same mint price, could also dispense worn

coins that were worth significantly less than their mint price.

• By paying out such coins, the Bank could effectively manipulate the gold points.

• Officer pointed out that ERs never violated the gold points adjusted for these variations.

• The practice of paying out worn coin was an example of the so-called gold devices.

• In addition to paying out coin whose metallic content was eroded by wear, tear, and clipping,

central banks of other countries (such as France and Belgium) on “limping Gold Standards”

could also redeem their notes in depreciated silver.

• Thus, ERs vis-à-vis these countries were potentially not constrained to the same narrow

bands as those of countries on full GSs.

42


Discretionary Monetary Policy and Gold Convertibility

• The key for the compatibility of monetary autonomy (and discretion) with gold convertibility

lay in the gold points and gold devices, which gave central banks room for maneuver.

• Governments could pursue monetary policies that caused the ER to depreciate so long as

this did not violate the gold points and threaten the central bank with reserve losses.

• Indeed, monetary autonomy could be quite significant.

• In 1930, Keynes provided an analysis of the tradeoff between the width of the gold points

band and the extent of monetary autonomy which anticipated Paul Krugman’s 1991 analysis

of target zones in the Quarterly Journal of Economics and Lars Svensson’s 1994 use of the

Krugman model in a Journal of Monetary Economics article to study the tradeoff between

ER stability and monetary independence.

• The Keynes-Svensson analysis showed that even a narrow band endowed authorities with

significant room to maneuver independently and buffer shocks.

• By providing this degree of flexibility, gold points and gold devices contributed importantly to

the GS’s success.43


• Consider the example of a monetary policy expansion that triggered an ER depreciation—

say, because the central bank wanted to support an otherwise weak economy.

• The expansion in domestic credit would produce the lower interest rate desired by the

central bank only if the ER was expected to recover subsequently.

• Suppose that this was so—because of the credibility of the commitment to convertibility.

• Say the ER depreciated toward the gold export point (where the currency was cheap enough

that it was profitable to convert it into gold, export the latter, and use it to purchase foreign

currency).

44


• As soon as the central bank was threatened with reserve loss, investors would anticipate

that the central bank would quickly act in contractionary fashion to keep the ER from straying

outside the band:

– Capital would flow in from abroad in anticipation of the profits that would accrue to

investors in domestic assets once the central bank took steps to strengthen the ER.

• Because the commitment to protecting gold convertibility of the currency (and hence

reserves) was unquestioned, capital flowed in quickly and in significant quantities.

• Interest rates tended to fall, and the depreciation of the exchange rate was limited, keeping

it well inside the gold export point.

45


• This smoothing effect of investors’ behavior on ER dynamics within the boundaries of

credible bands, based on the investors’ expectations of future ER movements, is the same

central mechanism in the analysis of target zones that Krugman and Svensson (and others)

wrote with the resurgence of interest in ER bands generated by the European Monetary

System in the 1980s.

• Under the GS, central banks could deviate from the “rules of the game” because their

commitment to the maintenance of gold convertibility was credible.

• Although it was possible to find repeated “violations of the rules” over periods as short as a

year, central banks behaved according to the rules of the game over longer intervals.

• Put differently, central banks had the ability to breach the rules in the short run, because

there was no question about their obeying them in the long run.

• Knowing that central banks would ultimately take whatever steps were needed to defend

convertibility, investors would shift capital toward weak-currency countries, financing their

deficits even when their central banks temporarily violated the rules of the game.

46


• We are going to model the effects of credibility and expectations on ER dynamics in a target

zone using the analysis in Paul Krugman (1991, Quarterly Journal of Economics).

• Suppose we measure all variables in logs.

• Several models of the ER yield an exchange rate equation of the form:

ε (t) = mD (t) + vD (t) + γE [dε (t) /dt] , (1)

where:

– ε (t) is the ER between the home and foreign currencies (units of home currency per unit

of foreign – thus, an increase in ε (t) indicates depreciation),

– mD (t) is the difference between home and foreign money supplies (mD (t) ≡ m (t) −m∗ (t)),

– vD (t) is an exogenous, relative shock term (vD (t) ≡ v (t)− v∗ (t)),

– γ is a parameter such that γ > 0,

– E is the expectation operator,

– and dε (t) /dt is the change in the ER between time t and the next instant.

47


Continuous vs. Discrete Time

• We assume a continuous-time modeling approach.

• Hence, the continuous-time notation convention ε (t) instead of εt, and the expected change

in the ER measured by E [dε (t) /dt].

– In a discrete-time model, we would write the current ER as εt, the change in the ER

between this period and the next as εt+1 − εt, and the expectation of this change in the

current period as Et (εt+1 − εt).

– Continuous time implies that the distance between two consecutive points in time

becomes arbitrarily small.

– Hence, we write the change between now and the next instant as the differential (d) of

the ER (dε (t)) over an arbitrarily small interval of time (dt).

– The expectation of the change in the ER between now and the next instant is then written

as E [dε (t) /dt].

48


The Intuition for Equation (1)

• Equation (1) states that there are two fundamental determinants of the ER:

– relative money supply mD (t) and the exogenous shock vD (t).

· The ER depreciates (ε (t) increases – it takes more units of the home currency to buy

one unit of foreign currency) if mD (t) increases (i.e., if home monetary policy is more

expansionary than foreign).

· The ER depreciates also if there is an increase in vD (t) (for instance, if there is an

increase in the velocity at which home households dispose of home currency relative

to foreign households).

• In addition, the ER depreciates if it is expected to depreciate, as agents who expect the

currency to lose value immediately start selling it.

49


The Effects of a Credible Target Zone

• For simplicity, we assume that foreign monetary policy and exogenous shocks are fixed and

normalized to 1, so that, in logs, m∗ (t) = v∗ (t) = 0, and equation (1) reduces to:

ε (t) = m (t) + v (t) + γE [dε (t) /dt] . (2)

• To continue simplifying, we also assume that m is shifted by the central bank only to maintain

a perfectly credible target zone:

– The central bank is prepared to reduce m in order to prevent ε from exceeding some

maximum value ε, and to increase m in order to prevent ε from falling below some

minimum value ε.

– As long as ε remains within these boundaries (the gold points of the GS), money supply

remains unchanged.

• Without loss of generality, we can normalize units to center the target zone around zero, so

that ε = −ε.

50


• For simplicity, we also assume that the exogenous shock term v (t) follows a continuous-time

random walk:

dv (t) = σdz (t) ,

where σ > 0 is a parameter and the exogenous z (t) is such that E [dz (t)] = 0, implying

E [dv (t)] = 0.

– We are assuming that the current v (t) is the best predictor of the value of v (t) in the next

instant based on the available information.

· Upward or downward changes in v (t) are equally probable.

• This assumption allows us to focus more transparently on the ER dynamics implied by the

presence of a target zone.

51


• Suppose we plot the ER as a function of v (t) in a diagram.

• The broken lines in Figure 1 indicate the boundaries of the target zone, constraining the ER

to be between ε and ε.

• We consider the behavior of the ER starting from an initial money supply m = 0.

• A naive view of the ER would be as follows:

– Since money supply is locally held constant and v (t) follows a random walk, there should

be no predictable change in the ER, i.e., E [dε (t) /dt] = 0.

– Thus, the ER might simply be expected to equal m (t) + v (t) inside the band, i.e., to

behave like a freely floating ER inside the target zone.

– If successive realizations of the shock v push ε to the edge of the band, then m will be

adjusted to prevent ε from drifting any further.

52

672 QUARTERLYJOURNAL OF ECONOMICS

so simple a structure, yet it will yield some surprising insights about the functioning of a target zone.

Before proceeding to algebraic analysis, it is useful to start with an intuitive approach to the effects of a target zone on exchange rate behavior. Figure I plots the exchange rate against v ; the target zone is indicated by the broken lines that define a band that bounds the exchange rate between -5 and I.We consider the behavior of the exchange rate when starting with some initial money supply, say m = 0.

Now a naive view would run as follows: since m is locally held constant and since v follows a random walk, there should be no predictable change in the exchange rate- E[dsldt]= 0. Thus, the exchange rate might simply be expected to equal m + v inside the band, i.e., to behave like a freely floating rate inside the target zone. If successive shocks to v push the exchange rate to the edge of the band, then the money supply will be adjusted to prevent s from drifting any further; thus, this naive view would suppose a relationship between v and s that looks like the heavy line in Figure I.

Why is this not right? Suppose that it were the correct description of exchange rate behavior, and consider the situation at an exchange rate that is just inside the band, say at point 2. Starting at point 2, if v falls a little, the exchange rate would retreat


• This naive view would suppose a relationship between ε and v that looks like the solid,

45-degree line in Figure 1.

– Starting from m = 0, and since m is not changed until the ER reaches the boundaries of

the band, the naive view would suggest ε (t) = v (t) inside the band, and ε (t) equal to the

maximum or minimum level otherwise.

• But this is not right.

53


• In fact, the behavior of private agents in the expectation of monetary policy actions at the

limits of the band will affect the behavior of the ER inside the band.

• To see this, suppose that the naive view is correct, and consider the situation at an ER level

that is just inside the band, say at point 2.

• Starting at point 2, if v falls a little, the ER would retreat down the 45-degree line to a point

like 1.

• If v rises a little, however, the ER will not rise by an equal amount, because the central bank

will act to defend the target zone.

• So, the ER will move to a point like 3.

• But this says that when we are near the top of the band, a fall in v will reduce ε more than a

rise in v will increase ε.

54


• Since we assumed that v follows a random walk (with equally likely upward or downward

random movements), this implies that the expected rate of change of ε is negative:

E [dε (t) /dt] < 0.

• Because expected depreciation enters the ER equation (2), this will affect the ER itself:

– The ER will be “dragged” down from point 2 to a somewhat lower point.

· As private agents buy home currency in the expectation of central bank action at the

boundary of the bank, this results in immediate appreciation relative to point 2.

• The same must be true (in opposite direction) at the bottom of the band.

• In effect, the relationship between v and ε must be bent as it approaches the edges of the

target zone.

55


• We cannot stop here, however.

• Once ER behavior near the edges of the zone lies off the 45-degree line, this will affect ER

expectations further inside the zone as well.

• Repeated revisions of ER expectations will lead to a relationship between v and ε that looks

like the S-shaped curve in Figure 2, below the 45-degree line in the upper half of the target

zone and above it in the lower half.

– Such repeated revisions will happen as agents trade currencies to remove any

discontinuity in the path of the ER inside the band.

· If any discontinuity remained, it would imply an unexploited opportunity for profits from

arbitrage across currencies.

56

TARGET ZONES AND EXCHANGE RATE DYNAMICS 673

down the 45-degree line, to a point like 1.If v rises a little, however, the exchange rate will not rise by an equal amount, because the monetary authority will act to defend the target zone. So the exchange rate will move to a point like 3.

But this says that when we are near the top of the band, a fall in v will reduce s more than a rise in v will increase s. Since v is assumed to follow a random walk, the expected rate of change of s is negative. Because expected depreciation enters the basic exchange rate equation (I), this will affect the exchange rate itself: the exchange rate would be "dragged" down from 2 to a somewhat lower point. The same must be true at the bottom of the band. In effect, the relationship between v and s must be bent as it approaches the edges of the target zone.

We cannot stop here, however; once exchange rate behavior near the edges of the zone lies off the 45-degree line, this will affect exchange rate expectations further inside the zone as well. I t seems intuitively obvious that repeated revisions of exchange rate expectations will lead to a relationship between v and s that looks like the S-shaped curve in Figure 11, below the 45-degree line in the upper half of the target zone and above it in the lower half.

There are two important points to make about the S-curve in Figure 11. First is the relationship between geometry and behavior.


• As we have observed, the target zone (the credible commitment of the central bank not to

let the ER stray beyond the gold points) has a stabilizing effect on the ER.

• With constant money supply and no target zone, the ER would simply move up and down

the 45-degree line.

• The S-curve in Figure 2, however, is flatter than the 45-degree line – that is, exogenous

shocks have a smaller effect on the ER, and thus the ER itself displays less variation than

under a free float without target zone.

• Note that this reduction in variation occurs even while the ER is inside the band, and thus

no current effort is being made to stabilize it:

– ER dynamics are stabilized by the behavior of private agents anticipating the actions

of the central bank at the limits of the target zone, exactly as in the target zone

characterization of the GS regime.

– In this regime, ERs fluctuated between the gold points, but the credibility of the

commitment to convertibility and the expectation that central banks would act accordingly

stabilized their dynamics between these boundaries.

57


• You may be wondering why a central bank would want to expand monetary policy to prevent

the ER from strengthening past the gold import point:

– After all, if that happened, the central bank would be accumulating gold.

• In our model, we simply follow Krugman and assume that the central bank behaves

symmetrically on both sides of the band.

• In practice, we shall see that asymmetry of incentives did indeed matter, and it played a

crucial role in the eventual collapse of the GS at the beginning of the 1930s.

• For our current purposes, we may observe that sterilization (the practice of central banks

to adjust monetary base in order not to let money supply rise with gold inflows) may not be

successful over long horizons.

• More importantly, a central bank’s gold import point is its partner’s gold export point.

• So, even if the domestic central bank does nothing as its currency is strengthening, it will

be the foreign one that must contract its own policy in order to prevent gold loss, thereby

ensuring the stabilizing effect of the target zone on the left side of the band.

– Of course, this is an example of non-cooperative behavior by the domestic central bank,

with consequences that we shall study.

58


• Bordo and Kydland (1995, Explorations in Economic History ) synthesize the discipline and

discretion views of the GS, by characterizing it as a contingent rule.

• Capital flows were stabilizing, keeping ERs within the gold points band limits, so long as the

commitment to defend currency convertibility was credible.

59


• Bordo and Kydland argue that the same was true even when governments were forced to

suspend convertibility temporarily and allow the ER to violate the band, so long as markets

were convinced that they remained committed to resuming convertibility at the previous rate

once the exceptional disturbance motivating the temporary suspension had passed.

• So long as the markets were convinced of the government’s commitment to gold convertibil-

ity, they would bid up the currency in anticipation of the eventual restoration of convertibility

at the traditional parity, stabilizing the ER.

• Of course, this mechanism only operated under the conditions that the contingency in

response to which suspension took place was exceptional and independently verifiable (war,

natural disaster) and so long as authorities remained committed to restoring convertibility as

soon as that contingency passed.

• If these conditions were met, the GS worked as a contingent rule, to be obeyed in normal

times, but which could be disregarded in crises without damaging the authorities’ credibility.

60

The Roots of Credibility

• The unquestioned credibility of the commitment to gold convertibility ultimately underpinned

the success of the GS system in a world of high capital mobility—which is generally

associated with difficulty of operating regimes that limit ER flexibility.

• Since everything depended on the credibility of the commitment to the maintenance (or

restoration) of convertibility, a key to understanding how the GS worked is to identify the

roots of that credibility.

• In the late 19th century, this rested on economics, diplomacy, politics, and ideology.

• That the decades between 1870 and 1913 were a period of international peace (relative

to the preceding and following years) minimized shocks to government budgets and

international payments.

• International peace also manifested itself in the form of cooperation among central banks in

the management of crisis episodes, with central banks providing each other with resources

in times of need.

61

The Roots of Credibility, Continued

• The absence of a clearly articulated theory linking central bank policy to the state of the

economy limited the pressure on monetary authorities to pursue output and employment

goals.

• The same was true of the absence of a Keynesian theory linking budget deficits to aggregate

demand and employment.

• Thus, the ideology of hard money extended to the maintenance of budget balance, limiting

the scope for fiscal policy to destabilize external accounts.

• In the countries at the core of the system, ruling parliamentary elites held much of their

wealth in government bonds and thus favored price stability.

• Those who valued other policy targets (for instance, workers concerned with unemployment)

rarely had the right to vote.

• For all these reasons, political support for the GS was pervasive.

62


• Admittedly, there were always skeptics, and ruling elites could not always afford to ignore

potential political threats from populist and socialist forces.

• In the U.S., farmers hurt by falling agricultural prices lobbied for the restoration of free silver

coinage.

• Their populist movement came close to pushing the country off the GS in 1896.

• Still, in none of the core countries did anti-GS agitation succeed.

• When authorities in those countries had to choose between interest rate increases to keep

the GS from collapsing and interest rate reductions to stimulate the economy, they never

hesitated to opt for the former.

63


• The situation was different outside the North European and North American core of the

system.

• In Latin America, for example, credibility was less complete.

• Inflationist agitation among farmers (as in the U.S.), combined with unstable politics, shaky

finances, and external shocks to drive governments to depreciated paper.

• But the contrast with Europe highlights what was unique and distinctive about the GS:

– At its core, the commitment to gold convertibility was paramount.

• As we shall see, this constellation of circumstances was unique.

• Neither it nor the commitment to gold would long survive the beginning of the Great

Depression.

64

The Consequences of World War I

• The unquestioned credibility of the commitment to convertibility was the ultimate underpin-

ning of the GS.

• But WWI undermined the mechanism.

• War led to government deficit spending, export controls, and inflation incompatible with a

fixed domestic currency price of gold as governments resorted to printing money to finance

their spending.

• Virtually every country but Britain and the U.S. suspended the convertibility of currency into

gold and placed barriers in the way of international gold shipments.

– Even the British authorities used “red tape” and moral suasion to discourage gold

exports.

65

The Consequences of World War I, Continued

• Budget deficits persisted beyond the conclusion of the war as the financial needs of

reconstruction replaced the conflict as cause of public spending.

• With the return to peacetime conditions, it was no longer feasible for governments to control

prices directly and fully restrict foreign exchange transactions.

• Rates of inflation and currency depreciation accelerated, and even Britain was forced to

suspend convertibility once it became clear that further discount rate increases to defend

the gold points band would have sent the country into a tailspin.

66

Returning to Gold

Britain

• It was only in 1925 that Britain stabilized its financial conditions sufficiently to restore

convertibility.

• Importantly, then Chancellor of the Exchequer Winston Churchill decided to do so at the

prewar sterling per gold rate.

– The belief was that restoring the prewar conversion rate would be crucial for credibility,

based on the pre-war record.

• This turned out to be a major mistake as it resulted in an overvalued exchange rate for the

pound sterling vis-a-vis British trade partners.

67

Returning to Gold, Continued

• Consider the following argument.

• The real exchange rate is a key measure of international competitiveness.

• It measures how many units of the domestic consumption bundle it takes to trade for a unit

of the foreign consumption bundle.

• Letting Pt denote the pound sterling price of British consumption, P ∗t the—say—dollar price

of U.S. consumption, and εt the exchange rate between the pound sterling and the dollar

(pound sterlings per dollar), the real exchange rate between Britain and the U.S. is:

εtP∗t

Pt=

£$

$U.S. consumption

£British consumption

=British consumption

U.S. consumption.

• When the real exchange moves down (i.e., it appreciates), it takes less units of British

consumption to trade for one unit of U.S. consumption.

• That means British consumption has become relatively more expensive, and British products

have lost competitiveness.

68


• Now suppose that the U.S. price level is constant, but the British price level rises because

of pent-up inflationary pressure that is released once price controls are removed at the end

of the war.

• If the parity between the pound sterling and gold is not adjusted upward accordingly

(devaluing the pound sterling relative to gold), the nominal exchange rate εt will not adjust

for the increase in Pt, and the real exchange rate will appreciate, causing Britain to lose

competitiveness.

• That is precisely what happened: By setting the gold parity at the pre-war level, Churchill

delivered an overvalued pound sterling relative to where it should have been to preserve

British trade competitiveness.

69


Other Countries

• The members of the Commonwealth and most smaller countries of continental Europe

quickly restored gold convertibility of the currency once Britain did.

• But France, Belgium, Italy, and Germany found it impossible to turn back the clock.

• Given the length of time that inflation had run out of control, restoring the original gold

parities without immediately incurring massive deficits due to loss of competitiveness would

have required reducing price levels by more than half (often significantly more), which was

impossible economically and politically in most of these countries.

70


• France settled for six months of deflation before stabilizing at a devalued parity (more francs

per ounce of gold than at the original parity) at the end of 1926.

• In Germany, hyperinflation so debased the currency that full monetary reform was required.

• The Weimar Republic established a new currency – the Reichsmark – in 1924 before

restoring gold convertibility.

– The ER between the Reichsmark (RM) and the Papiermark (PM) was set to 1RM =

1012PM ( = 1 trillion PM).

· The name PM was applied to the German currency after suspension of gold

convertibility of the mark in 1914.

· To stabilize the transition from PM to RM, the PM was initially replaced by the

Rentenmark, an interim currency backed by the Deutsche Rentenbank, which owned

industrial and real estate assets.

– The RM was then put back on the GS at the prewar conversion rate.

71

The Collapse of the Gold Standard

• Superficially, the normal state of GS affairs had been restored by the end of 1926, with the

major countries back on gold.

• The system’s structure had not changed in obvious ways, except for the withdrawal of gold

coin from circulation.

– In most countries, residents had been required to turn it over to the authorities to help

finance the war, and authorities were not inclined to turn it back.

• But subsequent experience was very different from the GS of prewar years:

– BOP adjustment was anything but smooth.

– Discount rate increases threatened recession and failed to attract gold inflows.

• The GS as an international system survived for barely five years.

• Starting in 1931 one country after another, beginning with the UK itself, was forced to

abandon convertibility and allow its currency to float downward (i.e., depreciate).

• By 1936 the transition to floating was complete, and the international GS was no more.

72

The Causes of the Collapse

• Multiple explanations have been proposed for this outcome.

• One strand of thought emphasizes that the interwar system was more fragile and susceptible

to shocks than its prewar predecessor.

• Central banks now held a substantial share of their reserves in the form of interest-bearing

foreign exchange.

• For example, the bonds of governments whose currencies were themselves convertible into

gold, mainly sterling and the dollar.

• This was the so-called Gold-Exchange Standard:

– Countries backed their currencies not only with gold but also with holdings of bonds in

currencies that were (supposedly) committed to gold convertibility to be able to expand

their own money supplies.

73

The Causes of the Collapse, Continued

• But any question about the stability of the sterling or the dollar could lead to a massive

liquidation of foreign reserves, as central banks scrambled to replace them with gold before

suffering capital losses due to pound sterling or dollar devaluation.

• By their actions, central banks might thus cause the reserve base of the system to implode

and deflationary pressure to ramify internationally.

• In addition, the prewar pattern of trade relations had been shattered by the creation of new

national borders in Central and Eastern Europe (due mainly to the breakup of the Austro-

Hungarian empire) and by the proliferation of tariffs on imports, on which governments now

relied to raise revenues.

74


• International flows of financial and physical capital did not fit together as neatly as before

1913.

• Before WWI, Britain, the major exporter of financial capital, was also a major exporter of

capital goods and capital flowed in a countercyclical, stabilizing manner.

• This stabilized the payments of the country at the center of the international system:

– When London lent abroad, the receiving countries used the funds to purchase machinery,

equipment, and ships from British industries, automatically balancing Britain’s external

accounts.

• I explain this mechanism in detail in Appendix B.

75


• After WWI, however, Britain was no longer the preeminent lender, the role having been

assumed by the U.S.

• The borrowers no longer relied on one country for their imports of machinery and capital

equipment, as members of the Commonwealth had relied on Britain before 1913.

• Thus, merchandise exports no longer financed foreign lending to the same extent.

76


• Yet, none of the previous facts is enough to explain the collapse of the GS.

• The practice of holding foreign exchange reserves was widespread before 1913, especially

in countries other than England, Germany, France, and the U.S.

• Paper currency and bank deposits accounted for 90 percent of money supply (and gold coin

for 10 percent) already before WWI.

• Countries other than Britain had relied heavily on import tariffs for revenue and other

purposes.

• While London accounted for perhaps half of all overseas lending between 1870 and 1913,

Paris, Berlin, Amsterdam, and, toward the end of the period, New York had also played

significant roles.

• And the linkage from exports of financial capital to exports of physical capital, however

important for Britain, was hardly evident in these other countries.

77


• A second set of explanations of the poor performance of the interwar GS emphasizes the

decline of the international leader, or “hegemon,” Britain, and the reluctant acceptance of

this mantle by the U.S.

• Keynes referred to Britain and the Bank of England as the “conductor of the international

orchestra.”

• According to this view, before 1913, when credit conditions were lax and the Bank of

England raised its discount rate, the impact on international financial markets was so

profound that other central banks had no choice but to follow suit.

• This follow-the-leader behavior brought about a de facto harmonization of policies worldwide.

78


• In Keynes’ metaphor, the Bank of England used its discount rate as the conductor’s baton.

• With British trade and finance accounting for such large shares of the respective world

totals, what was good for Britain was good for the international system.

• The decline in Britain’s commercial and financial leverage over the course of WWI and

through the 1920s, coupled with the reluctance of an isolationist U.S. to accept the

“maestro’s baton,” undermined the harmonization of policies that had prevailed before the

war.

• Central banks and governments no longer played harmoniously, and the climax of this

cacophony was the collapse of the GS in 1931.

79


• In an influential 1973 book on “The World in Depression, 1929-1939,” Charles Kindleberger

generalized this interpretation of the destabilizing effects of the absence of a hegemon.

• Before WWI, he argued, Britain had provided an open market for the goods of countries in

distress.

• British lending fluctuated countercyclically, rising in periods of global slowdown, falling in

booms.

• Thus, Britain acted as international lender of last resort, providing capital to foreign central

banks and governments when the stability of the GS was threatened.

• Between the wars, instead, U.S. lending was procyclical (see Appendix C), and the U.S.

market was heavily protected.

• Moreover, the U.S. failed to acknowledge the need for an international lender of last resort.

• When financial crises happened in 1931, the failure of the U.S. to act as international lender

of last resort doomed the system.

80


• It is no surprise that this interpretation was advanced by one of the leading historians of

international monetary affairs after WWII, a period of overwhelming political dominance by

the U.S.

• But it is actually unclear that the stabilizing influence of Britain had been so dominant prior

to 1913.

• Often, it was Britain itself that was subject to international financial strains.

81


• In 1890, the Bank of England was faced with the insolvency of a major British merchant

bank, Baring Brothers, which had extended bad loans to the government of Argentina.

• In that episode, Britain was the international borrower of last resort (obtaining support from

the Bank of France and the Russian State Bank, among others) not the international lender.

• Often the Bank of England had to alter its discount rate to conform to international financial

conditions rather than leading them.

– This was in order to render its discount rate “effective” – that is, to keep it sufficiently in

touch with market rates to sustain a reasonable volume of business.

– Since the Bank of England had both an Issue Department (responsible for backing

the note circulation with gold) and a Banking Department that undertook commercial

activities and market intervention, too large deviations of its discount rate from the market

could jeopardize the Banking Department’s business, reducing the effectiveness of the

discount rate as the instrument of policy.

82


• As these examples illustrate, international cooperation had also contributed importantly to

the success of the prewar GS.

• Cooperation among the central banks of the leading countries, while episodic, was critical in

times of crisis.

• But it took place only under very specific circumstances, as emphasized by Marc Flandreau

in a 1997 Review of Economic History article.

• Flandreau points out that the extension of cooperation was motivated by selfish interests at

least as much as by any recognition of the existence of common goals.

• While apparent between 1890 and 1910 (“the heydays of the international GS”), cooperation

was much less extensive before and after.

83


• During the interwar period, given the difficult financial situation bequeathed by WWI, the

need for international cooperation was greater than before.

• But the obstacles were formidable.

• Disputes over war debts and reparations, and more generally the fact that the shadow of

war was never far removed, spoiled the climate for cooperation.

• The Bank for International Settlements (BIS), the logical vehicle for last-resort lending,

was disabled by having been created to manage the transfer of German war reparation

payments.

84


• The ideological underpinnings of the prewar GS no longer carried the same force.

• Proto-Keynesian ideas supporting monetary and fiscal policy expansion as tools to

systematically bolster the economy surfaced in a growing number of places.

• Policymakers in different countries interpreted economic weakness in different ways and

prescribed different policy responses, rendering concerted international action all but

impossible.

• Increasingly influential special interest groups emphasized conflicts between the policies

needed to advance cooperation with other countries and those required to address domestic

problems.

85


• With respect to this last point, perhaps the single most important difference between the

prewar GS and the interwar system was the credibility of the commitment to gold.

• Before WWI, there was no question that convertibility was the preeminent goal of policy.

• The 1920s saw a growing consciousness of the problem of unemployment:

– Governments and trade unions gathered and published statistics on its prevalence, many

for the first time.

• Keynes and others articulated theories of the connection between central bank policy and

the economy.

• Interest groups lobbied for output- and employment-friendly policies.

• Parliamentary labor parties gave voice to those concerned with unemployment.

• The extension of the political franchise made it costly for governments to neglect their views.

86


• For all these reasons, when the ER depreciated to the gold export point and a rise in interest

rates was required, it was no longer clear that the central bank had political backing.

• This might render it reluctant to raise interest rates in periods of weakness of the economy.

• Capital, rather than flowing in in anticipation of the ER’s recovery that would follow policy

contraction, might flow out, in anticipation of devaluation, making the central bank’s task

even harder.

– Put differently, conversion rates and gold points based on past record as the foundation

of credibility (such as Churchill’s choice for the pound sterling) might in fact not be

credible in the eyes of forward looking speculators anticipating the policy tradeoffs facing

authorities.

• The conflict between internal and external balance was heightened, and with it the dilemma

facing the monetary authorities.

• Between 1925 and 1931, markets repeatedly tested the extent of central banks’ commitment

to defending their gold reserves.

• In the newly politicized conditions of the interwar period, monetary authorities failed the test,

and this failure ultimately led to the collapse of the interwar GS.87

Appendix A: The Emergence of the International Gold Standard

• The GS is frequently portrayed as the normal state of international monetary affairs prior to

1913.

• In fact, it prevailed on a global scale only from 1880 to 1914.

• Prior to that, currencies were generally based on silver instead of gold or on a combination

of the two metals.

• Britain was the main exception, having been on a full legal GS since 1821 and on a de facto

GS since the eighteenth century.

• The GS became the international norm as result of a combination of accidental events, path

dependence, network externalities, and geopolitical factors not directly tied to economic

considerations.

88

Appendix A: The Emergence of the International Gold Standard, Continued

Accidental Events

• In 1717, Sir Isaac Newton (of physics fame) was the Master of the Mint in Britain.

• In this position, he made a crucial mistake and set too high a silver price for the gold guinea.

– A gold guinea could be exchanged for more ounces of silver at the British Mint than on

the international market in which bullion dealers were trading the metals.

• Overvaluation of gold created an opportunity for arbitrage:

– Suppose you were a bullion dealer in Amsterdam, the market price of gold was X ounces

of silver, and the silver price of the gold guinea set by Newton was X + 1 ounces of silver.

– You could use X ounces of silver to buy an ounce of gold in Amsterdam, ship it to

London, have it coined into a guinea at the British Mint, and ask to convert it into X + 1

ounces of silver, making a completely safe profit.

89


• With the Mint price of silver lower (and the price of gold higher) than its international

market price, any newly minted British silver coins were quickly driven from circulation,as

arbitrageurs bought them at the Mint in exchange for gold, and sold them for a higher price

on the international market.

• Britain received an inflow of gold, but was eventually stripped of silver.

• This was an example of Gresham’s Law: The bad money chases away the good one.

• In this case, gold was the bad money (because it was overvalued at the Mint relative to

its market value), and it chased away the good money (silver) through the operation of

arbitrage.

90


Path Dependence

• In 1819, after the Napoleonic Wars led to inflation and temporary suspension of convertibility

of Bank of England’s notes, Parliament turned the status quo of a gold-backed currency into

law by passing a law that required the Bank of England to make its notes redeemable in

gold at the 1821 market price.

• Newton’s mistake put Britain on a de facto GS; Parliament accepted this outcome and

turned into British law.

91


Network Externalities

• Britain was the center of a large trading bloc—the British Empire—and had trading relations

with many other countries.

• Once Britain settled on a monetary standard, it became convenient for its satellites and

many partners to adopt the same standard to facilitate trade in goods and capital.

• But it took more than this for the GS finally to become the predominant regime in the 1870s.

92


• All the way to the 1870s, the U.S., France, and many other countries operated bimetallic

standards.

• Their Mints stood ready to transform specified quantities of gold or silver into coins of

comparable value.

• The operation of bimetallic regimes implicitly kept the relative market price of the two metals

close to the Mint ratio.

• How do we understand this?

93


• Think again about what happened to Britain: As a consequence of Newton’s pricing mistake,

the British Mint was absorbing gold and releasing silver.

• Notice: The operation of the arbitrage required an expansion of the demand for gold on

the market (so the gold could then be shipped to Britain to be coined into guineas) and

an increase in the supply of silver (released by the British Mint and then used on the

international market to buy gold to continue the arbitrage).

• But this increse in the market demand for gold and supply of silver will cause the market

price of gold to rise and that of silver to fall, closing the gap relative to the Mint ratio that

made the arbitrage profitable to begin with.

– Moreover, expectations will contribute to the closing of the relative price gap, as traders

anticipating that the market price of gold is about to rise will demand more of it now in

expectation of its increase in value.

• If this had happened before the British Mint ran out of silver, the British bimetallic system

would have survived, and its operation would actually have driven the market price close to

the Mint ratio.

94


United States

• In the U.S., the Coinage Act of 1792 established a Mint ratio of 15 ounces of silver to an

ounce of gold, approximating the market price.

• But increases in Mexican and South American silver production soon caused the market

value of silver to fall, so it took 15.5 ounces of silver to trade for one ounce of gold.

• With gold undervalued at the Mint, silver was brought in to be coined, and gold was shipped

abroad where its price was higher (the opposite of the effect of Newton’s undervaluation of

silver in Britain).

• As a result, through the first half of the 1830s, the U.S. was effectively on a silver standard.

• The Coinage Act of 1834 raised the Mint ratio to 16 to 1 in an attempt to restore gold coins

circulation.

• Gold discoveries then depressed the price of gold, causing silver to be exported and gold to

be coined.

• The U.S. was effectively placed on the GS until convertibility was suspended with the

outbreak of the Civil War.

95


France

• While France’s experience was similar in many respects, its larger internal circulation of

both gold and silver insulated it from disturbances to the availability of the two metals.

• Initially, both gold and silver circulated in France because the 15.5 to 1 French Mint ratio

was closer to the market price.

• Suppose, however, that the relative market price of gold rose to 16 ounces of silver per

ounce of gold.

• An arbitrageur could then import 15.5 ounces of silver into France and have it coined at the

Mint.

• He/she could exchange that silver coin for one containing an ounce of gold, export that gold,

and trade it for 16 ounces of silver on the world market, recouping the initial investment and

obtaining an extra half ounce of silver.

96


• As long as the market ratio stayed sufficiently above the Mint ratio, the incentive for arbitrage

remained.

• Arbitrageurs would import silver and export gold until all the gold coins in the country had

been exported.

• Once again, Gresham’s Law at work: the money that is losing value (silver in this case)

driving out the good one (gold).

97


• Only if the Mint and market ratios remained sufficiently close would both gold and silver

circulate.

• This simultaneous circulation was not threatened by small deviations between Mint and

market ratios:

– Arbitrage was not a frictionless activity that could be performed at zero cost:

· Governments charged a fee—known as brassage—to coin bullion.

· In addition, arbitrage took time—the price discrepancy motivating it might disappear

before the transaction was complete—and there were costs of shipping and insurance.

– Therefore, the difference between market and Mint ratios had to be larger than the total

of these costs for arbitrage to be profitable.

– Put differently, costs created a “band of inaction” around the Mint ratios within which

there was no incentive for arbitrage.

· As we shall see, these bands of inaction played a very important role in the functioning

of the GS.

98


• By absorbing one metal and releasing the other, the operation of France’s bimetallic system

reduced the supply of the former available to the rest of the world and increased that of the

latter, sustaining the circulation there of both metals.

• As we noted above, market participants factored this into their expectations.

– If the value of silver fell to the point where arbitrage was about to become profitable,

traders, realizing that the bimetallic system was about to absorb silver and release gold,

bought silver in anticipation.

• The bottom of the band around the Mint ratio thus provided a floor at which the relative price

of the abundant metal was supported.

99


• However, this stabilizing influence was effective only in the face of limited changes in metal

supplies.

• Large movements could strip bimetallic countries of the metal that was underpriced at the

Mint—as it happened to Britain after Newton’s mistake.

• With no more of that metal to release, their monetary systems no longer provided a floor at

which its price was supported.

100


Geopolitical Factors

• Fear of inflation due to gold discoveries deterred many nations from adopting the GS until

the 1870s.

• In 1871, Germany initiated the process that tipped the balance in favor of gold.

• The indemnity received in 1871-73 as victor in the Franco-Prussian War provided the

resources needed to carry out a currency reform after a temporary suspension of

convertibility during the war years.

• Germany established a gold-based currency unit, the mark, and used the indemnity to

purchase about half of the gold needed for circulation.

• The Germans sought to complete the process by selling their silver on world markets, taking

advantage of bimetallic France’s commitment to purchase it (and sell gold).

• Determined not to aid its longtime German rival, France responded by limiting silver coinage.

101


• Without France to operate silver-gold conversion on the needed scale, the bimetallic

bloc was shattered, the market value of gold skyrocketed, and the GS emerged as the

predominant regime as the more countries adopted it, the more attractive it became for

others.

• By the early 1900s, most of the world was on the GS.

– The main exceptions were China (which remained on silver), portions of Latin America

with silver and bimetallic standards, and bimetallic Persia.

102

Appendix B: Countercyclical and Stabilizing Capital Flows

• As Eichengreen’s (2019) Globalizing Capital discusses, capital flows often acted in

countercyclical and stabilizing fashion during the pre-WWI Gold Standard, and they were

tightly connected to goods trade; this changed in the interwar period, when capital flows

became procyclical and destabilizing.

• To understand the stabilizing nature of capital flows in the pre-war period and their

connection to goods trade, consider the effect of a negative shock that causes a GDP

decline in the center country of the system, Britain, and focus on the British money market

equilibrium condition (the LM relation):

Mt

Pt= YtL(it). (3)

– Real money demand (nominal money, Mt, over the price level, Pt) is determined by the

amount of transactions in the economy, or GDP (Yt), and by the interest rate, it.

· We know that there are multiple interest rates in each country, but we abstract from

this for simplicity and assume that there is just one British interest rate (and similarly

in other countries), maneuvered by the central bank.

– The liquidity preference function L(it) is decreasing in the interest rate:

· The higher it, the higher the opportunity cost of holding money instead of interest-

bearing assets; hence, money demand declines.103

Appendix B: Countercyclical and Stabilizing Capital Flows, Continued

• The shock causes British GDP Yt to contract.

• Everything else given, contraction of the British economy implies a contractionary shock

to its partners at the periphery of the system, because British demand for their exports

declines.

• Using a star to denote a representative partner economy (let us call it Foreign):

Y ∗t = C∗t + I∗t + G∗t + X∗t − IM∗t . (4)

– Foreign GDP (Y ∗t ) is equal to the sum of Foreign consumption (C∗t ), investment (I∗t ),

government spending (G∗t ), and trade balance (exports, X∗t , minus imports, IM∗t ).

• Reduced appetite for imports in Britain means that Foreign exports decline, imparting a

contractionary impulse to Foreign GDP.

104


• But now suppose that British money supply (which has to be equal to demand in equilibrium)

is tied to the Bank of England’s holdings of gold and does not change in the short run:

Mt = M .

• And suppose that also the British price level does not respond immediately to the shock,

and it remains constant in the short run: Pt = P .

– Strong assumptions, but we make them to simplify the analysis.

• Hence, equation (3) becomes:

M

P= YtL(it). (5)

• It follows that, in order to preserve money market equilibrium (the equality between the left-

and right-hand sides of the equation), the British interest rate has to fall.

– Remember that L(it) is a decreasing function of it; therefore, a lower interest rate causes

L(it) to rise, compensating for lower Yt.

105


• Assuming that interest rates in Britain and Foreign were equal before the shock to British

GDP happened, it follows that the British interest rate has fallen below Foreign’s.

• As a consequence, capital will flow from Britain to Foreign in search of higher yield.

• In the pre-WWI period, countries at the periphery of the Gold Standard—like Foreign in our

example—would use inflow of capital to finance investment expansion.

• Hence, the inflow of capital from Britain would be associated with an increase in I∗t .

• In equation (4), this would tend to counteract the negative effect of the decline in exports to

Britain (X∗t ), thereby stabilizing (Y ∗t ) against the contractionary effect of the British recession.

106


• Moreover, consider now the balance-of-payments equation for Britain:

RESt+1 −RESt = − (Bt+1 −Bt) + rtRESt + rtBt + TBt. (6)

• All else given, a capital outflow (Bt+1 − Bt > 0) would be bad news for the British balance

of payments and would be associated with a loss of reserves that could, if sustained,

jeopardize the convertibility of the pound sterling into gold.

107


• But now consider the pattern of trade in the pre-WWI period: Britain was the primary supplier

of capital goods to countries at the periphery.

• Expansion in investment at the periphery would be associated with rising demand for British

exports of capital goods.

• Hence, an increase in I∗t would lead to an improvement in the British trade balance:

– Assuming that it started from 0, TBt would become positive.

• This would counteract the effect of the capital outflow and stabilize the British balance of

payments.

• Moreover, rising exports would contribute to the recovery of British GDP from the initial

contractionary shock.

108


• This is the sense in which capital flows were countercyclical and stabilizing during the

pre-WWI period:

– Capital would flow from center to periphery in a manner that would counteract the

adverse effects at the periphery of shocks in the center, and the tight connection between

capital flows and the pattern of trade would in turn act as a stabilizing force back at the

center of the system.

109


• Another sense in which capital flows acted in stabilizing fashion was associated with the

functioning of the target zone system:

– Credibility of the commitment to intervene at the boundaries of the band ensured that if

a currency was approaching the gold export point, capital would flow in in anticipation of

the policy intervention and the implied strengthening of the currency.

– This force tended to push the currency toward the center of the band even before the

central bank intervened, facilitating its task (implying that a smaller increase in the

domestic interest rate would be necessary to keep the currency safely inside the band).

110

Appendix C: Procyclical and Destabilizing Capital Flows

• The mechanism broke during the interwar period, when capital flows became procyclical

and destabilizing.

• An example of this was the effect of the interaction between the Bank of England and the

Federal Reserve.

• When Britain returned to Gold Standard convertibility in the aftermath of WWI, it did so at

an overvalued exchange rate parity.

• Then Chancellor of the Exchequer Winston Churchill opted to return to the Gold Standard at

the pre-WWI parity between sterling and gold, on the assumption that this parity had been

successful before the war and was therefore credible.

• But the inflationary effects of war finance once wartime price controls were removed eroded

implied that the pre-war parity resulted in an overvalued real exchange rate.

111

Appendix C: Procyclical and Destabilizing Capital Flows, Continued

• The real exchange rate—let us denote it with Qt—is the relative price of home consumption

relative to foreign (or the ratio of the foreign price level to home when both are expressed in

home currency):

Qt ≡εtP

∗t

Pt=

home currencyforeign currency

foreign currency

foreign consumption

home currencyhome consumption

=home consumption

foreign consumption.

– When the real exchange rate becomes lower (appreciates), it means that home

consumption has become relatively more valuable (expensive) relative to foreign.

– This is associated with loss of competitiveness of home goods in international markets.

– Now suppose we abstract from target zone dynamics and that the Gold Standard

conversion parities just imply a fixed exchange rate: εt = ε.

– Suppose also that the foreign price level is constant: P ∗t = P ∗.

– It follows that home inflation (or British inflation, taking Britain as home) would imply real

appreciation of the pound sterling, and loss of competitiveness.

112


• As a consequence of Churchill’s choice of a parity that proved overvalued, the Bank of

England’s reserves came under pressure.

• The Federal Reserve tried to help Britain by keeping U.S. interest rates low, trying to tilt

capital flows in favor of Britain and engineering appetite for British exports.

• By the late 1920s, low U.S. interest rates had triggered a run-up in the U.S. stock market

that the Fed viewed as a bubble it was necessary to prick.

• The monetary policy contraction that followed is widely taken as the initial trigger of the

Great Depression.

• It triggered a contraction of GDP in the economy that was establishing itself as the new

economic center of the system (the U.S.), which propagated to its partners in the form of

reduced U.S. demand of their exports.

113


• Moreover, this reduced demand came at a time of high interest rates in the U.S. (from the

policy contraction), implying that the periphery got hit with a double shock: loss of demand

for its product and capital flowing away from the periphery and toward the higher U.S. yields.

• In this sense, capital flows became procyclical and destabilizing: Capital flowed in a way

that no longer lent stability to the system; instead, it amplified the adverse effects on the

periphery of shocks at the center.

114


• What was the crucial reason for this change?

• It was the change in the nature of monetary policy, which went from passive in the pre-WWI

period (think about our discussion above: the Bank of England adjusted the British interest

rate passively just to preserve money market equilibrium) to active in the interwar period

(when policy became driven by considerations other than protection of convertibility at all

costs—in our example, the Fed acts on the U.S. interest rate first to help Britain and then to

prick a perceived asset bubble).

– We know from our discussion of the Gold Standard as a system of target zones that

pre-WWI monetary policymaking was not always purely passive—that the target zones

created room for flexibility.

– But the passive-behavior example that we used captures key forces that were at work in

the functioning of the system.

• When push came to shove, central banks would act with convertibility as the overriding

priority, and this would lead to stabilizing capital flows that, in connection with trade patterns,

would stabilize the system.

115


• In the interwar period, central bank priorities changed.

• This eroded the credibility that they would do “whatever it takes” at the boundaries of the

band to protect the convertibility of the currency.

• When the currency got close to the gold export points, investors could no longer trust that

the central bank would contract policy with one-hundred-percent probability.

• This led them to “test the commitment” of the central bank, selling the currency for fear of its

continued loss of value instead of buying it in anticipation of strengthening.

• As a result, the task of a central bank close to the gold export point became harder, as a

stronger policy contraction than otherwise would be needed to defend the currency.

• This changed the policy tradeoff facing the central bank, as it made the defense of

convertibility more costly in terms of negative consequences for the economy (at a time also

of rising power of labor movements).

• For many central banks, this implied letting go of convertibility and abandoning the Gold

Standard.

116



Lessons from the Great Depression and

the Global Financial Crisis of 2007-08

Fabio Ghironi


CEPR, and NBER

Introduction

• What we learned about the functioning of the Gold Standard (GS) suggested an important

reason why the U.S. and other countries would be badly worse off if we were facing the

current crisis under a GS arrangement:

– No monetary or fiscal authority would want to subordinate its policy choices to external

balance considerations as tightly as imposed by even the most flexible interpretations of

the GS under the current circumstances.

• However, nothing makes the point more clearly than studying the dynamics of the Great

Depression (GD) and, more recently, of the so-called Global Financial Crisis (GFC) of

2007-08 and the Great Recession (GR) that followed it.

• We begin by focusing on the GD.

1

The Macroeconomics of the Great Depression

• When Ben Bernanke wrote his 1995, Journal of Money, Credit, and Banking article on "The

Macroeconomics of the Great Depression," many macroeconomists shared the consensus

that “To understand the Great Depression is the Holy Grail of macroeconomics.”

• The status of the GD as the Grail of Macroeconomics may have been supplanted by the

GFC, the GR, and the decade that followed, but we are probably living now the phenomena

that will be at the top of this sad podium for a long time.

• Nevertheless, understanding the GD helps us understand GFC and GR, and the lessons

from those two events give us insights into the current situation, although we should always

keep in mind that a better understanding will probably come only after the “fog of war” has

cleared.

• The next slides mostly reproduce material from Bernanke’s article.

2

The Macroeconomics of the Great Depression

• Traditional research on the GD focused on shocks to the domestic U.S. economy as primary

cause of both the American and world depressions.

• But no account of the GD would be complete without an explanation of the worldwide nature

of the event, and of the channels through which deflationary forces spread among countries.

• As Bernanke points out, by effectively expanding the data set from one observation (the

U.S.) to many, the shift to a comparative perspective improves our ability to identify the

forces responsible for the world depression.

• Bernanke reviews the causes of the GD and the channels through which it propagated by

distinguishing factors affecting aggregate demand (AD) and aggregate supply (AS).

3

Aggregate Demand: The Gold Standard and World Money Supplies

• We expect prices and output to move in the same direction following demand shocks,

whereas they move in opposite directions following supply shocks.

• Hence, the fact that changes in output and the price level exhibited a strong positive

correlation in almost every country during the GD suggests an important role for AD shocks.

• For many years, the principal debate about the causes of the GD in the U.S. was over the

importance of monetary factors:

– Money supply, output, and prices all fell precipitously in the contraction and rose rapidly

in the recovery.

• But the difficulty was in establishing the causal links among these variables.

4

Aggregate Demand: The Gold Standard and World Money Supplies, Continued

• Milton Friedman and Anna Schwartz, in a very influential 1963 book, argued that the main

lines of causation ran from monetary contraction (the result of monetary policy mistakes)

and continuing crisis in the banking system to declining prices and output.

• Opposing Friedman and Schwartz, Peter Temin contended in a 1976 book that much of the

monetary contraction in fact reflected a passive response of money to output; and that the

main sources of the GD lay on the real side of the economy (for example, an autonomous

drop in consumption in 1930).

• The research on the GS conducted since the 1980s, comparing the experiences of different

countries and taking a global perspective, makes it possible to assert with considerable

confidence that monetary factors played an important causal role, both in the worldwide

decline in prices and output and in their eventual recovery.

5


• Two well-documented observations support this conclusion:

– First, analysis of the operation of the interwar GS has shown that much of the worldwide

monetary contraction of the early 1930s was not a passive response to declining output,

but instead the largely unintended result of an interaction of poorly designed institutions,

shortsighted policymaking, and unfavorable political and economic preconditions.

· Hence the correlation of money and price declines with output declines that was

observed in almost every country is most reasonably interpreted as reflecting primarily

the influence of money on the real economy, rather than vice versa.

6


• – Second, for reasons that were largely historical, political, and philosophical rather than

purely economic, some governments responded to the crises of the early 1930s by

quickly abandoning the GS, while others chose to remain on gold despite adverse

conditions.

· As also Eichengreen also pointed out in his work, countries that left gold were able to

reflate their money supplies and price levels, and did so after some delay; countries

remaining on gold were forced into further deflation.

· To an overwhelming degree, the evidence shows that countries that left the GS

recovered from the GD more quickly than countries that remained on gold.

· Indeed, no country exhibited significant economic recovery while remaining on the

GS.

· The strong dependence of the rate of recovery on the choice of ER regime is further,

powerful evidence for the importance of monetary factors.

7

The Sources of Monetary Contraction

• What most countries experiencing monetary contraction had in common was adherence to

the international GS.

• How did this induce widespread monetary contraction?

• Consider a simple identity that relates the money stock (say, M1) of a country on the GS to

its reserves of monetary gold:

M1 =

(M1

BASE

)•(BASE

RES

)•(RES

GOLD

)• PGOLD •QGOLD, (1)

where:

– M1 = M1 money supply (money and notes in circulation plus commercial bank deposits),

– BASE = monetary base (money and notes in circulation plus reserves of commercial

banks),

– RES = international reserves of the central bank (foreign assets plus gold reserves),

valued in domestic currency,

– GOLD = gold reserves of the central bank, valued in domestic currency = PGOLD •QGOLD,

– PGOLD = the official domestic-currency price of gold, and

– QGOLD = the physical quantity of gold reserves.

8

The Sources of Monetary Contraction, Continued

M1 =

(M1

BASE

)•(BASE

RES

)•(RES

GOLD

)• PGOLD •QGOLD,

• Equation (1) makes the familiar points that, under the GS, a country’s money supply is

affected both by its physical quantity of gold reserves (QGOLD) and the price at which its

central bank stands ready to buy and sell gold (PGOLD).

– Ceteris paribus, an inflow of gold (QGOLD ↑) or a devaluation (PGOLD ↑) ⇒ money

supply ↑.

9


M1 =

(M1

BASE

)•(BASE

RES

)•(RES

GOLD


• However, equation (1) also indicates three additional determinants of money supply under

the GS:

– (1) The “money multiplier,” M1/BASE.

· In fractional-reserve banking systems, the total money supply (including bank

deposits) is larger than the monetary base.

· You should recall from macro theory that the money multiplier is a decreasing function

of the currency-deposit ratio chosen by the public and the reserve-deposit ratio

chosen by commercial banks.

· At the beginning of the 1930s, M1/BASE was relatively low (not much above one)

in countries in which banking was less developed, or in which people retained a

preference for currency in transactions.

· In contrast, in the U.S., this ratio was close to four in 1929.

10


M1 =

(M1

BASE

)•(BASE

RES

)•(RES

GOLD


• – (2) The inverse of the gold backing ratio, BASE/RES.

· Because central banks were allowed to hold domestic assets as well as international

reserves, the ratio BASE/RES – the inverse of the gold backing ratio (also called the

coverage ratio) – exceeded one.

· Statutory requirements usually set a minimum backing ratio (such as the Federal

Reserve’s 40 percent requirement), implying a maximum value for BASE/RES (for

example, 2.5 in the United States).

· However, there was typically no statutory minimum for BASE/RES, an important

asymmetry.

· In particular, sterilization of gold inflows by surplus countries (a reduction in monetary

base in response to gold inflow) reduced average values of BASE/RES.

11


M1 =

(M1

BASE

)•(BASE

RES

)•(RES

GOLD


• – (3) The ratio of international reserves to gold, RES/GOLD.

· Under the gold-exchange standard of the interwar period, foreign exchange convertible

into gold could be counted as international reserves, on a one-to-one basis with gold.

· Hence, except for a few “reserve currency” countries, the ratio RES/GOLD also

usually exceeded one.

• Because the ratio of money to monetary base, the ratio of base to reserves, and the ratio

of reserves to monetary gold were all typically greater than one, the money supplies of GS

countries were often large multiples of the value of gold reserves.

12


• Total stocks of monetary gold continued to grow through the 1930s.

• Hence, the observed sharp declines in money supplies must be attributed entirely to

contractions in the average money-gold ratio.

• Why did the world money-gold ratio decline?

• In the early part of the GD period, prior to 1931, the policies of some major central banks

played an important role.

• Federal Reserve policy turned contractionary in 1928, in an attempt to curb stock market

speculation.

13


• BASE/RES fell between 1928 and 1930, and the U.S. monetary base fell about 6 percent

between June 1928 and June 1930, despite a more-than-10 percent increase in U.S. gold

reserves during the same period.

• This flow of gold into the U.S., like a similarly large inflow into France, drained the reserves

of other GS countries and forced them into parallel tight-money policies.

• However, in 1931 and subsequently, the large declines in the money-gold ratio that occurred

around the world did not reflect anyone’s consciously chosen policy.

• The proximate causes of these declines were the waves of banking panics and ER crises

that followed the failure of Kreditanstalt, the largest bank in Austria, in May 1931, which

Eichengreen discusses in Globalizing Capital (2019).

14


• These developments affected each of the components of the money-gold ratio:

– First, by leading to rises in aggregate currency-deposit and bank reserve-deposit ratios,

banking panics typically led to sharp declines in the money multiplier, M1/BASE.

– Second, ER crises and the associated fears of devaluation led central banks to substitute

gold for foreign exchange reserves.

– This flight from foreign exchange reserves reduced the ratio of total reserves to gold,

RES/GOLD.

– Finally, in the wake of these crises, central banks attempted to increase gold reserves

and coverage ratios as security against future attacks on their currencies; in many

countries, the resulting “scramble for gold” induced continuing declines in the ratio

BASE/RES.

15


• A particularly destabilizing aspect of this process was the tendency of fears about the

soundness of banks and expectations of ER devaluation to reinforce each other.

• An element that the two types of crises had in common was the so-called “hot money,”

short-term deposits held by foreigners in domestic banks.

• On one hand, expectations of devaluation induced outflows of the hot-money deposits (as

well as flight by domestic depositors), which threatened to trigger general bank runs.

• On the other hand, a fall in confidence in a domestic banking system (arising, for example,

from the failure of a major bank) often led to a flight of short-term capital from the country,

draining international reserves and threatening convertibility.

• Other than abandoning the parity altogether, central banks could do little in the face of

combined banking and ER crises, as the former seemed to demand easy money policies

while the latter required monetary tightening.

• The sharp declines in the money-gold ratio during the early 1930s suggest that under the

GS as it operated during this period, there appeared to be multiple potential equilibrium

values of the money supply.

16


• Broadly speaking, when financial investors and other members of the public were “optimistic,”

believing that the banking system would remain stable and gold parities would be defended,

the money-gold ratio and hence the money stock itself remained “high.”

– Confidence in the banks allowed the ratio of inside money to base to remain high, while

confidence in the ER made central banks willing to hold foreign exchange reserves and

to keep relatively low coverage ratios.

• In contrast, when investors and the general public became “pessimistic,” anticipating bank

runs and devaluation, these expectations were to some degree self-confirming and resulted

in “low” values of the money-gold ratio and the money stock.

• As Bernanke points out, an interpretation of the monetary collapse of the interwar period

as a jump from one expectational equilibrium to another one fits neatly with Eichengreen’s

comparison of the classical (1870-1913) and interwar GS periods.

17


• In the classical period, high levels of central bank credibility and international cooperation

generated stabilizing expectations:

– Speculators’ activities tended to reverse rather than exacerbate movements of currency

values away from official exchange rates.

· Recall our analysis of the GS as a credible target zone system.

• In contrast, in the interwar period central banks’ credibility was significantly reduced by the

lack of effective international cooperation (the result of lingering animosities and the lack

of effective leadership) and by changing domestic political equilibria – notably, the growing

power of the labor movement, which reduced the perceived likelihood that the exchange

rate would be defended at the cost of higher unemployment.

• Banking conditions also changed significantly between the earlier and later periods, as war,

reconstruction, and the financial and economic problems of the 1920s left the banks of

many countries in a much weaker financial condition, and thus more crisis-prone.

• For these reasons, destabilizing expectations and a resulting low-level equilibrium for the

money supply seemed much more likely in the interwar environment.

18


• Table 1 presents evidence on equation (1) for six countries: three members of the Gold

Bloc, who remained on the GS until relatively late (Belgium, France, and Poland) and three

countries that abandoned

gold earlier (Sweden, UK, and U.S.).

1. Strong correspondence between GS membership and falling M1 money supplies.

2. Sharp declines in Ml/BASE and RES/GOLD, reflecting (respectively) the banking crises

and ER crises (both of which peaked in 1931).

3. Tendency of gold-surplus countries to sterilize (that is, BASE/RES ↓ in countries

experiencing GOLD ↑).

19

TABLE 1

DETERMINANTS SUPPLY 1929-1936OF THE MONEY IN SIX COUNTRIES, -

FRANCE (devalued October 1936)

MI M IIBASE BASEIRES RESIGOLD PGOLD QGOLD

POLAND (imposed exchange control April 1936, devalued October 1936)

M I MIIBASE BASEIRES RESIGOLD PGOLD QGOLD -

1929 2284 1.339 1.390 1.750 5.92 118.3 1930 2212 1.328 1.709 1.735 5.92 94.9 1931 1945 1.267 1.888 1.355 5.92 101.3 1932 1773 1.275 2.177 1.273 5.92 84.7 1933 1802 1.280 2.496 1.185 5.92 80.3 1934 1861 1.301 2.693 1.056 5.92 84.9 1935 1897 1.277 3.155 1.061 5.92 74.9 1936 2059 1.340 3.634 1.076 5.92 66.3

BELGIUM (devalued March 1935)

M I MIIBASE BASEIRES RESIGOLD PGOLD QGOLD

UNITED KINGDOM (suspended gold standard September 1931)

M I M IIBASE BASEIRES RESIGOLD PGOLD QGOLD

1929 1328 1.560 5.825 1 .0 0.1366 1069.8 1930 1361 1.618 5.699 1 .0 0.1366 1080.8 1931 1229 1.579 6.452 1 .O 0.1366 883.8 1932 1362 1.667 6.823 1 .0 0.1366 877.2 1933 1408 1.680 4.395 1 .0 0.1366 1396.4

SWEDEN (suspended gold standard September 193 1)

M I M I IBASE BASEIRES RESIGOLD PGOLD OGOLD

1929 988 1.498 1.280 2.082 2.48 98.8 1930 1030 1.508 1.082 2.618 2.48 97.2 1931 1021 1.522 2.631 1.238 2.48 83.1 1932 1004 1.373 1.740 2.039 2.48 83.1 1933 1085 1.106 1.202 2.205 2.48 149.2 1934 1205 1.21 1 1.101 2.575 2.48 141.5 1935 1353 1.268 1.029 2.542 2.48 164.5 1936 1557 l.211 1.032 2.355 2.48 213.3

(continued)

10 : MONEY, CREDIT, AND BANKING

TABLE 1 (Continued)

UNITED STATES (suspended gold standard March 1933)

M I M I ,BASE BASEIRES RESIGOLD PGOLD OGOLD

1929 26434 3.788 1.746 1 .O 0.6646 6014.0 1930 24922 3.498 1.655 I .0 0.6646 6478.9 1931 21894 2.831 1.854 1 .0 0.6646 6278.8 1932 20341 2.534 1.900 1 .O 0.6646 6358.6 1933 19759 2.380 2.057 1 .0 0.6646 6072.7 1934 22774 2.396 1.154 1 .0 1.1253 7320.9 1935 27032 2.335 1.144 1 .0 1.1253 8997.8 1936 30852 2.327 1.178 1 .O 1.1253 10004.7

NOTES The table llluslrates the idenllly. equallon ( I ) . for six counlrles Where poss~ble, values are end-of-year. Data sources are glven In the Appendix.

Definitions are as follows. M I = Money and notes in c~rculat~on plus commercial bank depos~ts; in local currency (m~lllons). BASE = Money and notes in circulation plus commerc~al bank reserves; in local currency RES = Intemat~onal reserves (gold plus fore~gn assets); valued in local currency GOLD = Gold reserves; valued in local currency at the official gold price = PGOLD X QGOLD PGOLD = Official gold price (units of local currency per gram), for countries not on the gold standard, a legal fiction rather than a market pnce. QGOLD = Phys~cal quantlty of gold reserbes, In metrlc tons.

all experienced significant gold inflows starting in 1933. This seemingly perverse result reflected the greater confidence of speculators in already depreciated currencies, relative to the clearly overvalued currencies of the Gold Bloc. This flow of gold away from some important Gold Bloc countries was the final nail in the gold standard's coffin.

1.2 . The Macroeconomic Implications of the Choice of Exchange-rate Regime We have seen that countries adhering to the international gold standard suffered

largely unintended and unanticipated declines in their inside money stocks in the late 1920s and early 1930s. These declines in inside money stocks, particularly in 1931 and later, were naturally influenced by macroeconomic conditions; but they were hardly continuous, passive responses to changes in output. Instead, money supplies evolved discontinuously in response to financial and exchange-rate crises, crises whose roots in turn lay primarily in the political and economic conditions of the 1920s and in the institutional structure as rebuilt after the war. Thus, to a first approximation, it seems reasonable to characterize these monetary shocks as exogenous with respect to contemporaneous output, suggesting a significant causal role for monetary forces in the world depression.

However, even stronger evidence for the role of nominal factors in the Depression is provided by a comparison of the experiences of countries that continued to adhere to the gold standard with those that did not. Although, as has been mentioned, the great majority of countries had returned to gold by the late 1920s, there was considerable variation in the strength of national allegiances to gold during the 1930s: Many countries left gold following the crises of 193 1, notably the "sterling bloc" (the United Kingdom and its trading partners). Other countries held out a few years more before capitulating (for example, the United States in 1933, Italy in 1934). Finally, the diehard Gold Bloc nations, led by France, remained on gold until the


• Another interesting phenomenon shown in Table 1 is the tendency of countries devaluing or

leaving the GS to attract gold away from countries still on the GS.

• The UK, Sweden, and the U.S. all experienced significant gold inflows starting in 1933.

• This reflected the greater confidence of speculators in already depreciated currencies,

relative to the clearly overvalued currencies of the remaining Gold Bloc.

• This flow of gold away from some important Gold Bloc countries was the final nail in the

GS’s coffin.

20

The Macroeconomic Implications of the Exchange Rate Regime

• Bottom line: Countries adhering to the international GS suffered largely unintended and

unanticipated declines in their money stocks in the late 1920s and early 1930s.

• These declines in money stocks, particularly in 1931 and later, were naturally influenced by

macroeconomic conditions; but they were hardly continuous, passive responses to changes

in output.

• Instead, money supplies evolved discontinuously in response to financial and ER crises,

whose roots in turn lay primarily in the political and economic conditions of the 1920s and in

the institutional structure as rebuilt after World War I.

• Thus, to a first approximation, it seems reasonable to characterize these monetary shocks

as exogenous with respect to contemporaneous output, suggesting a significant causal role

for monetary forces in the world depression.

• However, even stronger evidence for the role of nominal factors in the GD is provided by a

comparison of the experiences of countries that continued to adhere to the GS with those

that did not.

21

The Macroeconomic Implications of the Exchange Rate Regime, Continued

• Although the great majority of countries had returned to gold by the late 1920s, there was

considerable variation in the strength of national allegiances to gold during the 1930s:

• Many countries left gold following the crises of 1931, notably the “pound sterling bloc” (the

UK and its trading partners).

• Other countries held out a few years more before capitulating (for example, the U.S. in 1933,

Italy in 1934).

• Finally, the diehard Gold Bloc nations, led by France, remained on gold until the final

collapse of the system in late 1936.

• Because countries leaving gold effectively removed the external constraint on monetary

reflation, to the extent that they took advantage of this freedom we should observe these

countries enjoying earlier and stronger recoveries than the countries remaining on the GS.

22


• Ehsan Choudhri and Levis Kochin (1980, Journal of Money, Credit, and Banking) first

documented that a clear divergence between the two groups of countries did occur.

• Eichengreen and Jeffrey Sachs (1985, Journal of Economic History ) found that by 1935

countries that had left gold relatively early had largely recovered from the GD, while the

Gold Bloc countries remained at low levels of output and employment.

• Other studies confirmed these findings.

23


• Now, in practice the decision about whether to leave the GS was endogenous to a degree,

and so we must be concerned with the possibility that some underlying factor accounted for

both the choice of ER regime and the subsequent differences in economic performance.

• But this is very unlikely for two general reasons:

– First, for most countries the decision to remain on or leave the GS was strongly influenced

by internal and external political factors and by prevailing economic and philosophical

beliefs.

· Indeed, economic conditions in 1929 and 1930 were on average quite similar in

those countries that were to leave gold in 1931 and those that would not; thus it is

difficult to view this choice as being simply a reflection of cross-sectional differences

in macroeconomic performance.

– Second, and perhaps even more compelling, is that any bias created by endogeneity of

the decision to leave gold would appear to go the wrong way, as it were, to explain the

facts:

· The presumption is that economically weaker countries, or those suffering the deepest

depressions, would be the first to devalue or abandon gold.

· Yet the evidence is that countries leaving gold recovered substantially more rapidly

and vigorously than those who did not.

· Hence, any correction for endogeneity bias in the choice of ER regime should tend to

strengthen the association of economic expansion and the abandonment of gold.24

AVERAGEBEHAVIOROF SELECTEDMACROVARIABLESFOR COUNTRIESON AND OFF THE GOLD STANDARD, 1930- 1936

1930 1931 1932 1933 1934 1935 1936

1. Manufacturing production (log-change) Average -.066 - . 116 -.090 ,076 ,100 .074 ,072 ON -.I17 -.I73 ,068 .025 - ,001 OFF -.I13 -.057 ,078 ,120 ,008

2. Wholesale prices (log-change) Average -.I16 -.I22 -.045 -.017 ,018 .024 ,048 ON -.I40 -.I33 -.065 -.037 -.038 OFF -.084 -.011 -.002 ,033 ,036

3. M1 money supply (log-change) Average ,016 -.088 -.068 -.006 ,019 .027 ,074 ON -.094 -.088 -.045 -.013 -.067 OFF -.076 -.060 ,007 ,028 .046

4. M1-currency ratio (log-change) Average ,030 -.I29 -.006 -.024 -.002 - ,011 -.011 ON -.I42 -.052 -.009 -.016 - .037 OFF - . lo2 ,014 -.030 ,002 - ,006

5. Nominal wages (log-change) Average .004 -.030 -.053 -.030 -.002 -.001 ,031 ON -.027 -.070 -.033 -.031 - ,022 OFF -.039 -.045 -.029 ,007 ,004

6. Real wages (log-change) Average .I22 ,094 ,007 -.009 -.023 - ,022 -.018 ON ,110 ,064 .032 ,005 ,016 OFF ,059 -.020 -.025 -.032 -.031

7. Employment (log-change) Average -.066 -.I17 -.074 ,050 ,096 ,064 ,068 ON -.I13 -.I37 ,006 ,028 -.016 OFF -.I27 -.047 ,065 ,113 .083

8. Nominal interest rate (percentage points) Average 5.31 5.43 5.29 4.37 3.97 3.89 3.79 ON 5.22 4.20 3.69 3.26 4.05 OFF 5.90 5.68 4.56 4.13 3.86

9. Ex-post real interest rate (percentage points) Average 16.89 9.39 6.51 2.78 1.11 -1.19 -8.93 ON 10.38 9.41 6.94 3.35 -4.92 OFF 7.16 5.47 1.64 0.61 -0.62

10. Relative price of exports (log-change) Average -.033 -.011 -.047 ,076 .084 - ,067 ,039 ON .003 -.019 ,134 ,140 -.I12 OFF -.040 -.058 ,058 ,070 - ,058

11. Real exports (log-change) Average -.073 -.I79 -.222 .014 ,056 ,021 .072 ON -.I93 -.292 -.008 .015 - ,024 OFF -.I46 -.I92 ,021 ,067 ,030

12. Real imports (log-change) Average - .07 1 -.211 -.264 ,004 ,038 ,020 ,049 ON -.I59 -.250 -.006 -.067 -.012 OFF -.315 -.271 ,008 ,070 ,027

13. Real share prices (log-change) Average - . lo7 -.I86 -.214 ,133 ,060 ,091 ,115 ON -.I81 -.219 ,139 -.028 ,062 OFF -.I98 -.211 ,130 ,092 .098

NOTES:For each variable and year, the table presents the overall average value of the vanable, and the average for counlnes on and off the gold standard In that year (see Bernanke and lames 1991). As most countnes were on the gold standard in 1930 and off the gold standard in 1936, disaggregated data for those years are not presented. Data are annual and for up to twenty-six counuies, depending on data availability (see the Appendix). Real wages, real share pnces, and the ex post real rate of Interest are computed using the wholesale price index If a country 1s on the gold standard for a fraction f of a particular year, the values of ~ t s variables for the whole year are counted with the gold standard countries with weight f and wlth non-gold-standard countries with welght Iffor that year. The proponion of country-months "on gold" in each year are as follows: 0.676 (1931), 0.282 (1932), 0.237 (1933), 0.205 (1934), 0.164 (1935).

TABLE 3

DUMMIES,1931-1935

Dependent vanable ONGOLD PANIC Adjusted R2

Manufacturing ( la) - .0704 0.601 production (4.04)

(lb) - ,0496 - .0926 0.634 (2.80) 13.50)

Wholesale prices

Money supply (MI)

M1-currency ratio

Nominal (5a) - ,0204 0.196 wages (2.62)

(5b) -.0145 - .0262 0.219 (1.78) (2.16)

Real wages

Employment

Nominal interest rate

Ex-post real (9a) 2.70 0.264 interest rate (2.07)

(9b) 2.16 2.39 0.266 (1.56) (1.16)

Relative price (10a) ,0464 0.198 of exports (1.70)

(lob) .0288 ,0783 0.213 (1 .OO) (1.83)

Real exports ( l l a ) - ,0745 0.323

Real imports (1 2a) - .OOOO 0.416 (0.00)

(1 2b) ,0232 - ,1036 0.435 (0.75) (2.25)

Real share (13a) - ,0299 0.354 prices

NOTES:Entnes are estimated coefficients from regressions of the dependent variables agalnst durnm~es for adherence to the gold standard (ONGOLD) and for the presence of a banking panlc (PANIC). Absolute values of r-statlstlcs are In parentheses. Dependent variables are measured In log-changes, except for the nomlnal and ex post real Interest rates. whlch are In percentage polnts (levels) Data are annual, 1931 to 1935 ~nclus~ve, (see the Append~x) Each regresslon mcludes aand for up to twenty-six countries, dependmg on data avallab~l~ty complete set of year dummies ONGOLD and PANIC are measured as the number of months durlng the year In whlch the country was on gold or experiencing a banking panlc (see text), d~vided by twelve


• The evidence presented by Bernanke is consistent with the conclusions that:

1. Monetary contraction was an important source of the GD in all countries.

2. Subsequent to 1931 or 1932, there was a sharp divergence between countries which

remained on the GS and those that left it.

3. This divergence arose because countries leaving the GS had greater freedom to initiate

expansionary monetary policies.

25


• These conclusions reiterate points made by Eichengreen in a 1992 Economic History

Review article:

– The stage for U.S. recession was set by U.S. monetary policy contraction.

– Interacting with existing imbalances in the pattern of international settlements, this

caused an even more contractionary shift in policy abroad, as gold flowed out of most

GS countries.

– U.S. exports weakened, and the Wall Street crash of 1929 led consumers to defer

spending on expensive items (consumer durables).

– This magnified the cyclical sensitivity of the durables sector to the downturn.

– Once the open market operations executed by the Federal Reserve Bank of New York in

the wake of the 1929 Wall Street crash were unwound, a further move toward contraction

in U.S. monetary policy reinforced deflationary tendencies.

– And a decline in the flexibility of U.S. labor markets in the 1920s limited the economy’s

ability to adjust.

– The shock(s) from the U.S. propagated to other countries via the GS constraint.

– Countries that abandoned the GS were better positioned to engage in expansionary

policies and recover more quickly.

26

Aggregate Supply

• The discussion above highlights monetary factors as crucial in the GD, but macroeconomists

then face a challenge:

– Macroeconomic theory predicts that money should be neutral in the long run and

changes in monetary conditions should not have persistent real effects.

– How did the monetary policy changes mentioned above have such long lasting effects in

the GD years?

• This is an AS-side puzzle: Why did the process of adjustment to nominal shocks appear to

take so long in interwar economies?

• Bernanke discusses two leading explanations for this in his article: induced financial crises

and sticky nominal wages.

27

Aggregate Supply, Continued

• The financial crises argument builds on work by Bernanke himself and other authors that

expands on Irving Fisher’s (1933, Econometrica) debt-deflation argument.

• Fisher envisioned a dynamic process in which falling asset and commodity prices created

pressure on nominal debtors, forcing them into distress sales of assets, which in turn led to

further price declines and financial difficulties.

• His diagnosis led him to urge President Roosevelt to subordinate ER considerations to the

need for reflation, advice that (ultimately) FDR followed.

• Fisher’s idea was less influential in academic circles, though, because of the counterargu-

ment that debt-deflation represented no more than a redistribution from one group (debtors)

to another (creditors).

• Absent implausibly large differences in marginal spending propensities among the groups, it

was suggested, pure redistributions should have no significant macroeconomic effects.

28


• However, the debt-deflation idea was revived in work by Bernanke and other authors who

focused on the consequences of imperfect information and agency costs in capital markets.

• According to the agency approach, the structure of balance sheets provides an important

mechanism for aligning the incentives of the borrower (the agent) and the lender (the

principal).

• One central feature of the balance sheet is the borrower’s net worth, defined to be the

borrower’s own (“internal”) funds plus the collateral value of his illiquid assets.

• Many simple principal-agent models imply that a decline in the borrower’s net worth

increases the deadweight agency costs of lending, and thus the net cost of financing the

borrower’s proposed investments.

• Intuitively, if a borrower can contribute relatively little to his or her own project and hence

must rely primarily on external finance, then the borrower’s incentives to take actions that

are not in the lender’s interest may be relatively high.

• The result is both deadweight losses (for example, inefficiently high risk-taking or low effort)

and the necessity of costly information provision and monitoring.

• If the borrower’s net worth falls below a threshold level, he or she may not be able to obtain

funds at all.

29


• From the agency perspective, a debt-deflation that unexpectedly redistributes wealth away

from borrowers is not a macroeconomically neutral event:

– To the extent that potential borrowers have unique or lower-cost access to particular

investment projects or spending opportunities, the loss of borrower net worth effectively

cuts off these opportunities from the economy.

– Thus, for example, a financially distressed firm may not be able to obtain working capital

necessary to expand production, or to fund a project that would be viable under better

financial conditions.

– Similarly, a household whose current nominal income has fallen relative to its debts may

be barred from purchasing a new home, even though purchase is justified by its expected

income.

• By inducing financial distress in borrower firms and households, debt-deflation can have

real effects on the economy.

30


• If the extent of debt-deflation is sufficiently severe, debt-deflation can also threaten the

health of banks and other financial intermediaries.

• Banks typically have both nominal assets and nominal liabilities and so over a certain range

are hedged against deflation.

• However, as the distress of banks’ borrowers increases, the banks’ nominal claims are

replaced by claims on real assets (for example, collateral).

• From that point, deflation squeezes the banks as well, and banking crises may follow.

• Bernanke’s article discusses the significant role of these dynamics in the GD.

• It also reviews how slow adjustment in nominal wages was important for these dynamics.

31


• Nominal stickiness prevented downward wage adjustments that could have eased the

situation of employment, with high wages prevailing especially in the Gold Bloc countries.

• An open question, however, is what motivated such long lasting inertia in nominal wages.

• Given the costs of this inertia, it is not straightforward to reconcile it with rational economic

behavior.

• One interesting possibility explored by Bernanke is that the nonindexation of financial

contracts, and the associated debt deflation, might in some way have been a source of the

slow adjustment of wages and other prices.

• Such a link would most likely arise for political reasons: As deflation proceeded, both

the growing threat of financial crisis and the complaints of debtors increased pressure on

governments to intervene in the economy in ways that inhibited adjustment.

• Examples of such interventions can be found for France, one of the countries where the GD

had the longest lasting effects.

• Even in the U.S., Eichengreen highlights changes in the characteristics of the labor market

that reduced its flexibility in the post-WWI period and contributed to the Depression.

32

A Model of Monetary Policy Interactions in the Interwar Gold Standard

• Our discussion of the GS and the GD highlighted at various points the role of cooperation

versus central bank competition in the accumulation of gold reserves, and its potential for

imparting a deflationary bias to monetary policy under the GS.

• We now study a simple model of policy interactions under the GS that captures some key

features of our discussions.

• The model was developed by Eichengreen in a 1984 article in Explorations in Economic

History.

• It is a stylized model whose purpose is not to capture all the details of the functioning of the

GS.

• Its purpose is to provide a simple macroeconomic framework which highlights how the

actions of a foreign central bank can affect a country’s internal and external position and the

incentives these repercussion effects provide the home country’s central bank to respond to

foreign initiatives.

• The next slides mostly reproduce material from Eichengreen’s paper.

33

A Model of Monetary Policy Interactions in the Interwar Gold Standard,

Continued

• The model has two essential ingredients:

1. First, it assumes that each central bank has more targets than instruments, forcing it to

confront a tradeoff between its various objectives.

2. Second, it assumes that each domestic target variable is affected by the actions of the

foreign central bank.

• The latter is the assumption of interdependence.

Monetary Policy Instruments

• Central banks in the model affect the size of the money multiplier through changes in their

discount rates.

• Central banks can also alter the composition of the monetary base through open market

operations and changes in fiduciary circulation, but this instrument is held inactive in the

analysis for reasons we will discuss.

34


Continued

• The idea that changes in central bank discount rates affect the relationship between gold

reserves and money supply captures the fact that, in the interwar period, the link between

gold reserves and broadly defined monetary aggregates was loose in many countries.

– Central banks frequently held gold in excess of that required to back notes in circulation,

enabling them to intervene in financial markets with purchases or sales of bonds and

bills and to alter the monetary base without any accompanying change in reserves.

– Only the need to maintain confidence in the convertibility of the currency placed limits on

such discretionary actions.

– Similarly, commercial banks, even if free of statutory reserve requirements, had

the normal incentive to hold precautionary reserves to guard against unanticipated

withdrawals.

– The size of such precautionary reserves was determined in part by the cost of feasible

alternatives, including rediscounting at the central bank.

– Therefore, the central bank discount rate influenced broadly defined monetary aggregates

through its impact on the money multiplier.

35


Continued

• The model assumes that each central bank has target levels of gold reserves and prices,

and sets policy optimally in an effort to accomplish these targets.

• Although the evidence suggests that central banks followed rules of thumb when setting

discount rates, the assumption of optimizing behavior is a useful simplifying device.

• The assumption that each bank had a target level of reserves follows from the observation

that, while a central bank could feel more secure in its ability to defend the convertibility of

the currency with a larger gold reserve on hand, it was less profitable to hold barren metal

than interest bearing financial assets.

• If central banks balanced the costs of extra gold against the benefits, this would have led

them to consider a target stock of gold.

36


Continued

• The idea that central banks had a target level for prices also is a simplifying assumption.

• Frequently, it is argued that central banks were concerned ultimately with the domestic

currency price of gold, and that they desired only to prevent such fluctuations in prices and

economic activity as might threaten convertibility.

• In this view, the price level should not be treated as an independent goal but merely an

intermediate target whose achievement was helpful for reaching the ultimate objective of

maintaining convertibility.

• At the same time, there was pressure throughout the interwar period for central banks to

respond actively to internal conditions.

37


Continued

• Such pressure is evident, for example, in the proposals adopted by the Financial Commission

at the Genoa Conference in May 1922.

• These proposals, largely drafted by Ralph Hawtrey of the British Treasury and presented to

the Conference by the British delegation headed by the Chancellor of the Exchequer, Sir

Robert Horne, resolved among other things that central banks should pursue credit policies

“not only with a view to maintaining currencies at par with one another, but also with a view

to preventing undue fluctuations in the purchasing power of gold.”

• Price stabilization was seen by the British as “the remedy for the unemployment evil.”

38


Continued

• Critics of central banking reiterated this point throughout the 1920s.

• Keynes for one had argued before Britain’s return to gold that restoration of the prewar

pound sterling parity would provoke a disastrous deflation of wages and prices, necessary

to restore competitiveness with an overvalued parity.

• By 1930, he was again arguing that a change in Bank of England policy was required to

stabilize British prices.

• Treasury officials such as Sir Otto Niemeyer warned that high discount rates would be a

serious inconvenience to British industry.

• Central bankers may have held a different view of the role for monetary policy, but it is

difficult to dispute that pressures such as these would have caused them to prefer stable

prices to either inflation or deflation and relatively smooth output growth to more violent

internal fluctuations.

• In the model below, prices and production move together, so the target of a stable price level

can be thought of as shorthand for stable prices, output, and employment.

39

The Structure of the Model

• For simplicity, the model assumes that the world consists of two countries, home and

foreign.

• Each country has a money supply M , equal to the product of the monetary base and the

money multiplier µ.

• The base is made up of domestic credit C and the central bank’s gold reserves G.

• C can be positive or negative:

– It is negative if the central banks hold excess gold reserves and positive if there is a

fiduciary issue outstanding.

40

The Structure of the Model, Continued

• The money multiplier depends negatively on the central bank’s discount rate R.

– A rise in the discount rate increases the cost of rediscounting at the central bank,

inducing the banking sector to hold a larger ratio of precautionary reserves to deposits.

• Using asterisks to indicate foreign variables,

M = µ (R) (C + G) , µR < 0, (2)

M ∗ = µ∗ (R∗) (C∗ + G∗) , µ∗R∗ < 0, (3)

where µR (µ∗R∗) denotes the derivative of µ (µ∗) with respect to R (R∗).

41


• The world stock of gold is denoted by G, and

G = G + G∗. (4)

• We assume that demand for real money balances is proportional to output:

M/P = φY, φ > 0, (5)

M ∗/P ∗ = φ∗Y ∗, φ∗ > 0. (6)

– Agents hold money in proportion to the amount of their transactions, which is determined

by income and the price level.

– For simplicity, we abstract from the effect of interest rates on money demand.

42


• Output (= income) in each country is an increasing function of the domestic price level:

Y = Y (P ), 0 ≤ YP ≤ ∞, (7)

Y ∗ = Y ∗(P ∗), 0 ≤ Y ∗P ∗ ≤ ∞, (8)

where YP (Y ∗P ∗) denotes the derivative of Y (Y ∗) with respect to P (P ∗).

– Think of these as the supply curves of an aggregation of profit-maximizing firms.

– Then the level of production will depend positively on output prices and negatively on

wages and other costs.

– Rather than introducing costs explicitly, simply note that the classical, flexible-price full

employment model with vertical AS in the Y -P space (YP = Y ∗P ∗ = 0) and the Keynesian

scenario with horizontal AS (YP = Y ∗P ∗ =∞) can be treated as special cases.

43


• The model is closed by assuming purchasing power parity (PPP):

P = εP ∗, (9)

where ε is the ER (units of home currency per unit of foreign).

– PPP states that output has the same price in the two countries when prices are evaluated

in the same currency.

– It assumes that agents in the two countries have equal preferences and trade is

frictionless, so that arbitrage removes any difference in prices across countries (when

expressed in the same currency).

– PPP is a strong assumption, which is often violated, but it is useful to simplify the model

and focus on the issues of interest to us.

• We abstract from target zone dynamics of the ER and assume that the ER is fixed at ε = 1

for simplicity, implying P = P ∗.

44


• Rearrange (5) asM = PφY and substitute into (3) (thus imposing money market equilibrium)

to obtain:

PφY = µ (R) (C + G) .

• Next, substitute (7) into this equation to obtain:

PφY (P ) = µ (R) (C + G) . (10)

• Similarly in the foreign country:

Pφ∗Y ∗(P ) = µ∗ (R∗)(C∗ + G−G

), (11)

where we also used (4) and PPP (9) with ε = 1.

• Equations (10) and (11) are two equations in two endogenous variables (G and P ).

– We take the amount of credit in the two countries (C and C∗) as exogenous.

• Equations (10) and (11) state that the price level in each country must adjust to keep

equilibrium in the money market (money demand = money supply) and the gold stock must

adjust to insure the money supply’s equilibrium international distribution.

45

The Tradeoff Between Target Variables

• To understand the tradeoff between target variables facing central banks, we differentiate

equations (10) and (11) holding exogenous variables and the foreign discount rate fixed.

– Our purpose is to understand the effect of changes in the home central bank’s discount

rate on the variables the central bank cares about – gold reserves and the price level –

for given foreign discount rate and amounts of credit.

– In performing the differentiation required for this purpose, we use d to denote differential

(so that dx is the differential of variable x, for any variable x), and we use the following

general results:

· df (x) = fxdx: The differential of the function f (x) of a variable x is equal to the first

derivative of f (x) with respect to x (denoted by fx) times the differential of x (this tells

us by how much an increment dx in x causes f (x) to change);

· d (xf (x)) = (dx) f (x) + xdf (x) = (dx) f (x) + xfxdx: This is simply an application of

the rule for differentiation of a product: The differential of the product of x times f (x)

is equal to the differential of x times f (x) plus x times the differential of f (x).

46

The Tradeoff Between Target Variables, Continued

• Differentiating (10) and using the general results above yields:

dPφY (P ) + PφdY (P ) = dµ (R) (C + G) + µ (R) d (C + G) ,

or, using φY (P ) = M/P from the money demand and output equations, and recalling that

we are holding C fixed,

dPM

P+ PφYPdP = µRdR (C + G) + µ (R) dG,

• Finally, this equation can be rewritten as:

dP

(M

P+ PφYP

)= µRdR (C + G) + µ (R) dG. (12)

47


• Similarly, differentiating (11) yields:

dPφ∗Y ∗ (P ) + Pφ∗dY ∗ (P ) = dµ∗ (R∗)(C∗ + G−G

)+ µ∗ (R∗) d

(C + G−G

),

or:

dPM ∗

P+ Pφ∗Y ∗P ∗dP = µ∗R∗dR

∗ (C∗ + G−G)− µ∗ (R∗) dG,

or:

dP

(M ∗

P+ Pφ∗Y ∗P ∗

)= −µ∗ (R∗) dG, (13)

because we are holding the foreign discount rate fixed (i.e., dR∗ = 0).

– Note: We used Y ∗P = Y ∗P ∗ (because P = P ∗).

48


• From (13),

dP = − µ∗ (R∗)M∗

P + Pφ∗Y ∗P ∗dG. (14)

• Substitute this into (12):

− µ∗ (R∗)M∗

P + Pφ∗Y ∗P ∗dG

(M

P+ PφYP

)= µRdR (C + G) + µ (R) dG.

• Move all the terms involving dG to the left-hand side and collect:

−dG[

µ∗ (R∗)M∗

P + Pφ∗Y ∗P ∗

(M

P+ PφYP

)+ µ (R)

]= µRdR (C + G) ,

or:

dG

[µ∗ (R∗)

(MP + PφYP

)+ µ (R)

(M∗

P + Pφ∗Y ∗P ∗)

M∗

P + Pφ∗Y ∗P ∗

]= −µRdR (C + G) ,

where we imposed the common denominator in the left-hand side and we changed the

signs of both sides of the equation.

49


• Define

Ω ≡ µ∗ (R∗)

(M

P+ PφYP

)+ µ (R)

(M ∗

P+ Pφ∗Y ∗P ∗

).

Note that Ω > 0 (since YP and Y ∗P ∗ are both positive).

• Then we have:

dG = −µR (C + G)

(M∗

P + Pφ∗Y ∗P ∗)

ΩdR. (15)

• And, substituting (15) into (14) yields:

dP =µ∗ (R∗)µR (C + G)

ΩdR. (16)

50


• Equations (15) and (16) can be rearranged as:

dG

dR= −

µR (C + G)(M∗

P + Pφ∗Y ∗P ∗)

Ω, (17)

dP

dR=

µ∗ (R∗)µR (C + G)

Ω. (18)

• These two equations give us two important pieces of information:

– Equation (17) gives us the differential of G with respect to R.

· It tells us by how much the stock of gold held by the home central bank changes if the

home central bank marginally increases its discount rate.

· Since Ω > 0, Y ∗P ∗ ≥ 0, and µR < 0, it follows that dG/dR > 0:

· The stock of gold held by the home central bank increases if the central bank

increases its discount rate.

51


• – Equation (18) gives us the differential of P with respect to R.

· It tells us by how much the home price level changes (and, by virtue of PPP and ε = 1,

by how much the foreign price level changes) if the home central bank marginally

increases its discount rate.

· Since Ω > 0 and µR < 0, it follows that dP/dR < 0:

· The (home and foreign) price level decreases if the home central bank increases

its discount rate.

• Intuitively: An increase in the domestic discount rate decreases the domestic money

multiplier, putting downward pressure on the price level and, by reducing domestic money

supply relative to money demand, attracting gold to the home country to restore money

market equilibrium.

– Note: The decrease in the price level reduces money demand by reducing the price of

good purchases and, in general, income (since YP and Y ∗P ∗ are positive).

– Thus, even if the gold inflow pushes money supply back up, overall, money demand and

supply (the left- and right-hand sides of equation (10), respectively) settle at a lower level

than before the change in R.

52


• Thus, there is a tradeoff between central bank objectives:

– When the central bank contracts monetary policy, it improves its holdings of gold, but it

also causes deflation (and a reduction in output, as long as YP is positive).

• Similar results hold for the foreign central bank.

– You should verify that as an exercise: replace G with G − G∗ in equations (10) and (11)

and differentiate holding R fixed instead of R∗.

– You should be able to verify that dG∗/dR∗ > 0 and dP/dR∗ < 0 (see also below on this).

53

Interdependence: The External Effects of Policy

• If we differentiate equations (10) and (11) holding R fixed but letting R∗ vary, we can find

the effects of changes in the foreign central bank’s discount rate on the home central bank’s

target variables for given home discount rate.

• Proceeding similarly to what we did before, we have:

dG

dR∗=

µ∗R∗(C∗ + G−G

) (MP + φPYP

)Ω

< 0, (19)

dP

dR∗=

µ∗R∗µ (R)(C∗ + G−G

)Ω

< 0, (20)

where the inequalities follow from YP ≥ 0 and µ∗R∗ < 0.

– You should verify these results as an exercise.

• Intuitively: An increase in the foreign discount rate reduces the foreign money multiplier and

the foreign money supply, reducing the common world price level and attracting gold from

the home country.

• Again, analogous results hold for the effects of changes in R on the foreign central banks

target variables (and you should try to verify it as an exercise).

54

The Central Banks’ Objective Functions

• The model assumes that the central banks have target levels of gold reserves and prices,

and each central bank maximizes an objective function that depends on its holdings of gold

and the domestic price level.

• The home central bank maximizes the function:

W = W (G,P ) , (21)

and the foreign central bank maximizes the function:

W ∗ = W ∗ (G∗, P ∗) . (22)

• We assume that these functions are “well behaved”:

– They are continuous in both variables and twice differentiable (i.e., we can calculate first

and second derivatives with respect to each argument).

– They reach a unique maximum when gold reserves and prices are at their target levels.

55

The Central Banks’ Objective Functions, Continued

• An example of objective function that has these properties is:

W = −[(G−GT

)2+ α

(P − P T

)2], (23)

where GT and P T are the target levels of G and P , respectively, and α > 0 measures the

relative weight of gold reserves versus the price level in the central banks objective function.

• This objective function shows that the central bank is made worse off by deviations (in either

direction) of gold reserves and price level from target.

• Ideally, the central bank would like to have G = GT and P = P T , ensuring W = 0.

• Any deviation of gold reserves and price from target implies W < 0.

• The next figure shows you two plots of the function for example numerical values of GT

(= 500), GT (= 500), and α (= 1/2, i.e., the central bank attaches a larger weight to gold

reserves than the price level in its objective function).

56

The Central Banks’ Objective Functions, Continued

• As you could see, the objective function looks like a dome, or a mountain, with its maximum

at G = GT and P = P T .

• Now imagine cutting horizontal slices through this dome, and eliminating the W axis from

the diagram.

• The contours we are left with in Figure 1 represent combinations of G and P that result in

the same value of W .

– Think of the map of a mountain where you have contour lines that give you combinations

of longitude and latitude that result in the same altitude on the mountain.

• These contours are called indifference curves.

• The central bank is indifferent across points on a given contour curve because they result in

the same level of W .

• Contours closer to the center represent higher levels of W , i.e., the central bank is better off

when it can position itself on a contour closer to the center point – the bliss point at which W

is maximized (W = 0 in the numerical example I gave you).

57

CENTRAL BANK COOPERATION 73

where Q = V(M*/P + P$*Y&) + V*(M/P -t P$Yp) > Q. A rise in the domestic discount rate decreases the domestic money multiplier, putting downward pressure on the price level and, by reducing domestic money supply relative to money demand, attracting gold to the home country.

The second element we desire is that target variables in the home country are affected by the actions of the foreign central bank. Agam from (3’)

dG -= dR*

V$(C* + t? - G)(M/P + $f’Yp) < o

cl

dP -= dR*

vg*v(c* + 77 - G) < Q (7)

i-l

An increase in the foreign discount rate reduces the foreign money multiplier and the foreign money supply, attracting gold from the home country and depressing the world price level. Analogous results hold for the foreign country.

Recall that each central bank has a target level of gold reserves and a target price level. Each bank maximizes an objective function of the form

F = F(G, P) F” = F*(G*, P*).

From (6), we know that, given R *, the domestic central bank can vary pi to attain different combinations of G and P. In Fig. 1, the frontier of feasible combinations is labeled AA. We represent the central bank’s objective function F by a set of well-behaved indifference curves, on the assumption that the central bank is indifferent between a price level and gold reserves slightly above or below optimal levels. The optimal setting for R is the one which achieves a G-P combination tangent to an indifference curve, at the point labeled E in Fig. 1.

RGuRE 1

Central Bank Behavior

• Absent any constraint, the central bank would reach the bliss point.

• But the central bank has only one instrument (the discount rate R) and two targets (G and

P ), and there is a tradeoff between the two targets.

• Equations (10) and (11) define the set of G and P combinations that the home central bank

can reach by manipulating its discount rate, given the foreign central bank’s discount rate.

• For simplicity, let us assume that the money multiplier and output functions are such that

the frontier of the set of G and P combinations that are feasible for the home central bank to

reach is a straight line.

• Assume that this is the line AA in Figure 1 for an initial foreign discount rate R∗.

• How does the home central bank choose its discount rate?

58

Central Bank Behavior, Continued

• The home central bank would never choose a discount rate that positions it inside the AA

frontier.

• This would place the central bank on an indifference curve that is further away from the bliss

point than it is feasible for the home central bank to accomplish.

• The central bank will set R to generate the G-P combination that places the central bank on

the indifference curve closest to the bliss point.

• This happens at point E in Figure 1, where the frontier of the feasible G-P set is tangent to

the highest possible indifference curve (“altitude contour”) that the central bank can reach.

59


• Now consider the consequences of an increase in the foreign discount rate R∗.

• We verified above that, for given R, this causes both P and G to decrease.

• In our diagram, this means that the boundary of the feasible P -G set for the home central

bank shifts inward, to A’A’.

• If it does not adjust R, the home central bank is forced to accept lower prices, smaller gold

reserves, or a combination of the two.

• As drawn, it moves to point D, tangent to a less desirable indifference curve where both

prices and reserves have fallen.

60


• The same experiments can be conducted for the foreign central bank, but the analysis

becomes interesting once we combine the two banks’ problems and consider their

interaction.

• This can be done by transposing the indifference curves from P -G to R-R∗ space, as in

Figure 2.

• Each P -G combination (i.e., each point) in Figure 1 corresponds to a unique R-R∗

combination (i.e., a unique point in Figure 2).

• Algebraically, we can use equations (10) and (11) to solve for P and G in terms of R and R∗.

• By using the central bank’s objective function, we can then map the ranking of P -G

combinations into a ranking of R-R∗ combinations.

61

76 BARRY EICHENGRJZEN

Consider now a rise in R*. This shifts the AA locus inward to A’A’. The home country’s central bank is forced to accept lower prices, smaller gold reserves, or a combination of the two. As drawn, it moves to a point such as D tangent to a less desirable indifference curve where both prices and reserves have fallen.

The same experiments can be conducted for the foreign central bank. The analysis becomes interesting once we combine the two banks’ problems and consider their interaction. This can be done by transposing the indifference curves to R-R* space, as in Fig. 2. We read off from Fig. 1 the home country’s rankings of different combinations of the two discount rates. Thus, point E in Fig. 2 at the center of the home country’s (solid) indifference curves corresponds to point E in Fig. 1. Similarly, the dashed indifference curves above and to the left of the solid curves represent the preferences of the foreign central bank. The fact that the foreign central bank’s indifference curves lie above and to the left of the home central bank’s indifference curves reflects the assumption that each bank prefers to hold a relatively large share of the world’s gold stock; this is accomplished in our model when each bank’s discount rate is high relative to that of its rival.

The downward sloping pp locus depicts combinations of R and R* for which a price level P obtains. Along nonintersecting curves below and to the left of pp, prices are higher, while above and to the right, prices are lower. As the figure is drawn, the two central banks share a common rank ordering over prices. This is an assumption of convenience which allows us to speak of “their” most preferred price level and to interpret central banks’ strategic interaction in terms of a competitive scramble for gold.

The B and B* curves in Fig. 2 are the reaction functions of the two central banks. The B curve represents the locus of utility-maximizing discount rates for the home central bank, given the foreign discount rate. It is the locus of points where the tangent to a home indifference curve

FIGURE 2


• Point E in Figure 2 at the center of the home country’s (solid) indifference curves corresponds

to home’s bliss point in Figure 1.

• Similarly, the dashed indifference curves above and to the left of the solid curves represent

the preferences of the foreign central bank over R-R∗ combinations.

• We assume that each bank would like to hold a relatively large share of the world’s gold

stock.

• Thus, the foreign central bank’s indifference curves lie above and to the left of the home

central bank’s indifference curves, because the goal of holding a share of total world gold

larger than 1/2 is accomplished in the model when each bank’s discount rate is high relative

to that of its rival.

62


• The downward sloping P P line depicts combinations of R and R∗ for which the price level

P T obtains.

– Again, we assume for simplicity that functional forms are such that the locus P P is linear.

• Along parallel lines below and to the left of P P , prices are higher (since discount rates are

low relative to what is required to generate P = P T ), while above and to the right, prices are

lower.

• As the figure is drawn, we are assuming that the two central banks share a common price

target and rank ordering over prices.

• This is a convenient assumption that allows us to speak of a common most preferred price

level and to interpret central banks’ strategic interaction in terms of a competitive scramble

for gold.

63


• The B and B∗ lines in Figure 2 represent the reaction functions of the two central banks.

• Our discussion of Figure 1 helps us explain the home central bank’s choice of R for any

given R∗.

• Holding R fixed, an increase in R∗ forces the home central bank to accept lower prices and

reserves in Figure 1, moving it to point D, further away from the bliss point than the initial

position E.

• But the home central bank can respond by increasing its discount rate or lowering it.

– Increasing R causes P to decrease further, but G increases.

– Decreasing R causes P to increase, but G decreases further.

64


• Under the assumption that central banks attach a larger weight to gold reserves than the

price level in their objective functions, the home central bank will respond to higher R∗ by

increasing its own discount rate, resulting in the positively sloped reaction function B in

Figure 2.

– Since the central bank cares more about its gold reserves than the price level, it is willing

to trade further deviation of P from target against stemming the loss of reserves.

• Similarly, the foreign central bank will respond to higher R by increasing R∗, leading to the

positively sloped reaction function B∗.

• We assume that the home central bank’s reaction function is steeper than the foreign central

bank’s reaction function to ensure stability of the dynamics we discuss below.

65

Strategies and Solution

• We can now solve for the equilibrium values of R and R∗ under different assumptions about

the strategies of central banks.

• First, we assume that each bank chooses its optimal strategy taking the other bank’s

discount rate as given.

– This is the model’s non-cooperative, Cournot solution, or the Nash equilibrium of the

game between the central banks.

• Next, we assume that one bank continues to employ this strategy but the other anticipates

its rival’s reaction.

– This yields the Stackelberg leader-follower solution.

• Finally, we assume that the two central banks coordinate their actions to achieve what we

call a Pareto efficient outcome.

– This is the cooperative solution.

• By comparing the Stackelberg and cooperative solutions with the Nash equilibrium, we can

analyze the implications of central bank leadership and cooperation.

66

Strategies and Solution, Continued

• Consider Figure 2.

• Starting from a point such as K, if each central bank takes the discount rate set by its foreign

counterpart as given and adjusts its own discount rate accordingly, successive discount rate

increases lead to the Nash equilibrium N at the intersection of B and B∗.

– Note: Here you can see how our assumption on the relative steepness of B and B∗ is

important for stability:

· There would be no convergence if B∗ were steeper than B.

• The desire of the rival banks to possess incompatibly large shares of the world’s gold stock

causes both banks to raise their discount rates above the level consistent with price level

P T .

• In this framework, international competition for scarce gold reserves imparts a deflationary

bias to world prices and, for values of YP , and Y ∗P ∗ greater than zero, results in a lower level

of output in each country than along the P P curve.

67


• To compare this outcome with the leadership solution, assume that the home country acts

as a Stackelberg leader, setting its discount rate on the basis of accurate anticipations of

foreign reactions, while the foreign central bank acts as a Stackelberg follower, taking the

home country’s discount rate as given.

• The home country’s central bank sets its discount rate so that its foreign rival’s response

leads it to point M, where a home indifference curve is tangent to the foreign reaction

function.

• The home country’s central bank recognizes that if it increases its discount rate, the foreign

central bank will respond in kind.

• Compared with the Nash equilibrium, the home country’s central bank holds less gold, but

both countries benefit from a higher price level and a higher output.

• While the home country is forced to maintain slimmer gold reserves, leadership has less of

a deflationary bias than Nash behavior by both banks.

• For well-behaved indifference curves like those in Figure 2, both countries prefer M to N.

68


• However, neither the Nash equilibrium nor the leadership solution lies on the Pareto efficient

frontier – the locus of the points such that neither of the two players can be made better off

without making the other worse off.

• That frontier – also known as the contract curve – is the locus of points of tangency between

the two sets of indifference curves.

• In Figure 2, the contract curve is drawn as the heavy line connecting E with E∗.

• Starting from points away from this curve, it is always possible to make one of the players

better off without making the other worse off by moving to a point on the curve.

• The contract curve may lie on either side of (or cross) the P P line depending on the precise

form of the two countries’ objective functions.

• The cooperative solution will be a point on the contract curve.

• Precisely what point is reached (and thus what distribution of the world gold stock) will

depend on the bargaining power of the two central banks.

• But, for well-behaved indifference curves such as those in Figure 2, the cooperative solution

will have less of a deflationary bias than either the Nash or Stackelberg equilibria.

69


• Thus, the strategic interaction of the two central banks imparts a deflationary bias to the

world economy.

• Leadership has advantages over other noncooperative strategies:

– A country which takes its rival’s reaction into account, if willing to function with slimmer

gold reserves, can induce a reduction in both discount rates, raising prices and

stimulating production in both countries.

• Cooperation has advantages over leadership:

– Assuming that the two central banks have similar preferences over price levels, through

cooperation both discount rates can be lowered, and the deflationary bias can be further

reduced.

70

Eichengreen’s Model and Policy Interactions under the Interwar Gold Standard

• The model illustrates and explains the deflationary effects of a non-cooperative “scramble”

for gold across central banks, culminating in monetary conditions that contributed to the GD.

• We can use the model to analyze specific examples of international monetary relations in

the 1920s that concretely illustrate the roles of central bank leadership and cooperation – or

lack thereof.

• In his 1984 article, Eichengreen focuses on Anglo-French relations in the summers of 1927

and 1928.

• France followed Britain’s return to the GS after WWI by slightly more than a year.

• Unlike the British, who returned to the GS at the prewar sterling parity, the French stabilized

the franc at 20% of its prewar gold parity in 1926.

71

Eichengreen’s Model and Policy Interactions under the Interwar Gold Standard,

Continued

• – This stabilization was de facto rather than de jure.

– Although the Bank of France intervened to prevent ER fluctuations, the GS was not fully

restored until 1928.

– In the interim, there remained suspicions that the Bank of France would attempt an

eventual return to the prewar parity.

– Indeed there was sentiment favoring such a course within the Bank itself.

– At the same time, there were compelling arguments favoring the maintenance of the

devalued parity.

– Prices had lagged behind the ER over the period of depreciation before 1926, leaving

France with a competitive advantage in international markets (low prices and a devalued

ER).

– Assuming that prices would also lag behind an appreciating ER, efforts to restore the

prewar gold price would undermine French exporters’ competitive position and entail

extensive output and employment losses.

– Ultimately, these arguments prevailed, and no attempt was made to restore the prewar

parity.

– Rather, the Bank of France intervened as necessary to prevent appreciation.

72


Continued

• France had experienced heavy capital inflows since July 1926, comprised of both repatriated

French funds and foreign capital.

• Through the first 5 months of 1927, France continued to attract foreign balances.

• To prevent the currency from appreciating the Bank of France was forced to intervene in the

foreign exchange market with substantial sales of francs.

• Between November 1926 and May 1927, its foreign exchange holdings increased from a

mere $50 million to an impressive $770 million.

• To a considerable extent these foreign funds had come from across the English Channel.

• These were not the only pressures felt by the Bank of England.

73


Continued

• In 1926 the General Strike in conjunction with currency stabilization in still other countries

had rendered London a less attractive repository for short-term capital.

• New York was maintaining a high discount rate to curb stock market speculation, yet

investment on the New York stock exchange was continuing to drain funds from London.

• The Bank of France’s request in May 1927 that the Bank of England earmark 3 million

weekly in gold to be exchanged for its sterling reserves brought these matters to a head.

• The standard response to such external pressures would have been a rise in the Bank of

England’s discount rate.

• However, Montagu Norman, the Bank of England’s interwar Governor, felt constrained by

the slow recovery of British industry.

• From the British point of view, the preferred alternative to a rise in the London rate was a

reduction in the Paris rate.

74


Continued

• Through cooperation, discount rates in both countries could be lowered without undermining

ER stability, giving a boost to industrial production.

• In terms of Figure 2 in Eichengreen’s article, the alternatives can be thought of as moving

from an initial point of disequilibrium L toward the contract curve (in the expansionary

direction) rather than forcing additional deflation by moving to the Nash solution.

• In an attempt to achieve a cooperative solution, Norman traveled to Paris, where he

conferred with Emile Moreau, Governor of the Bank of France.

• According to an account subsequently published by Moreau, Norman complained that he

could not raise the Bank of England’s discount rate without “provoking a riot,” given the

depressed condition of British industry.

• According to Moreau, Norman argued that Paris had the power to force upon London a rate

increase, but Norman suggested to Moreau the alternative of a reduction in the Paris rate.

• The French, suspicious always of what Moreau referred as “the imperialism of the Bank of

England,” maintained that the responsibility for corrective measures rested with the deficit

country.75


Continued

• Ultimately, the efforts to achieve a cooperative solution for discount rate adjustment failed.

• In the end, neither discount rate was adjusted, but the two countries adopted rather more

indirect responses.

• The Bank of England intervened in the London market to drain off domestic credit, a move

with much the same effect as a rise in discount rate but with less symbolic content.

• In consequence, bankers deposits at the Bank of England declined by more than 2.5% over

the next 6 months.

• The Bank of France did take certain steps to ease the pressure on its British counterpart:

– It agreed to take gold from the Federal Reserve rather than the Bank of England, to leave

gold acquired through sales of sterling earmarked at the Bank of England, to continue

to hold high quality sterling assets among its reserves, and to alter its foreign exchange

support prices in such a way as to discriminate against sterling purchases and in favor of

other currencies.

76


Continued

• We can see in this episode both the advantages and limits of cooperation.

• Given the existence of unemployment and the danger of further deflation, there was an

argument for a discount rate reduction by Paris rather than an increase by London.

• In the end, the policies which were adopted resembled a combination of the two: tighter

credit conditions in Britain and more accommodating actions by the Bank of France.

• It is difficult to say whether these measures would have sufficed in the longer run.

• Indeed, their actions suggest that the central bankers themselves thought these measures

insufficient.

• Little more than a month after Norman and Moreau’s Paris meeting, several central bankers

(including Norman and a representative for Moreau) convened on Long Island to discuss

further prospects for coordination of national monetary policies.

• Little is known of the substance of their discussions, but in the wake of the meeting came a

reduction in the New York discount rate.77


Continued

• The Fed may have had a variety of reasons for this initiative.

• While the desire to counter recessionary pressures at home and facilitate seasonal crop

sales may have been among its motives, it also may have been the Fed’s intention to help

relieve the pressure on London.

• Thus, the crisis was surmounted.

• These experiences in attempting to, negotiate a cooperative solution may have conditioned

Norman and Moreau’s responses to the exchange market pressures of the following

summer.

78


Continued

• Over the first half of 1928 the Bank of England had drawn comfort from steadily increasing

reserves.

• British gold holdings rose from £153 million in January to £171 million at the beginning of

July.

• At that point the Federal Reserve’s discount rate, which had been increased repeatedly

to moderate the New York stock market boom, reached 5%, and the Bank of England’s

position began to deteriorate.

• Initially, Norman let gold go, a total of £7 million over the course of the summer.

• There was also the temptation to raise the discount rate.

• However, the Bank was conscious of the danger that a rise in its discount rate would induce

an upward adjustment by the European central banks, intensifying deflationary pressures

but resulting in no improvement in London’s position.

79


Continued

• The French as well as the British wished to augment their gold reserves, so any rate

increase was likely to be matched by Paris.

• France was running a substantial balance of payments surplus, and all the additional foreign

exchange acquired by the Bank of France from June 1928 was routinely converted into gold.

• In fact, the Bank of France may have been doing more than this.

• Since August 1927 it had been selling sterling on the spot market and repurchasing it

forward (i.e., repurchasing it for a certain future date for a price agreed upon today).

• The value of these forward contracts reached a peak at approximately £120 million in June

of 1928.

• If the Bank of France treated the sterling it repurchased as these forward contracts matured

as increments to its holdings, making them eligible for conversion into gold, the pressure it

placed on London would have been all the more intense.

80


Continued

• The Bank of England, increasingly alarmed by its reserve losses, notified foreign central

banks of its desire to avoid raising its discount rate out of concern for the state of industry.

• Again it attempted to negotiate a cooperative solution in which foreign central banks would

lower their discount rates to relieve the pressure on London, and again its efforts failed.

• The alternatives for the British were to initiate the necessary rate adjustments and ignore

the Continental reaction, or to base their strategy on the likely response of the Bank of

France and others.

• In terms of Figure 2, the choice was between competitive discount rate increases leading to

further deflation until reaching the Nash solution at point N, and a strategy which took the

likely French reaction into account, leading to the leadership solution at point M.

• Ultimately, the Bank of England resisted the impulse to increase its discount rate, once

again draining funds from the market and manipulating the price of gold.

• It would seem that the difficulties of achieving a cooperative solution in 1927 encouraged

the Bank of England to adopt a leadership strategy in the summer of 1928.

81

Summary

• The model of international policy interdependence that we studied is inspired by comments

by John Maynard Keynes and Montagu Norman that the interwar system can be understood

as a competitive struggle for gold.

• The model shows that if two central banks play a non-cooperative game in which they both

seek to augment their gold reserves, they will tend to raise their discount rates significantly

above the level consistent with price stability, depressing incomes at home and abroad.

• The model illustrates the advantages that monetary policy cooperation would have had

under the interwar GS.

• While central bank policy was but one factor at work in the world economy in the 1920s,

the model captures features of observed dynamics when applied to a period marked by

historically high discount rates, conflicts over the distribution of gold, and steady deflation

culminating in the GD.

• Before moving to our next topic, it is worth discussing some aspects of the model.

82

The Number of Policy Instruments versus Targets

• A central assumption of the model is that central banks have less instruments than targets.

• This assumption is crucial for there to exist a tradeoff between targets:

– If the number of instruments were equal to the number of targets, there would be no

tradeoff, as the central bank could assign each instrument to a target.

• Interestingly, the model does potentially feature a second instrument of policy for central

banks in addition to the discount rate: credit (C for home, C∗ for foreign in the model), i.e.,

the extent to which there is a fiduciary issue.

• We took this as exogenously fixed, but in practice it reflects central bank decisions.

83

The Number of Policy Instruments versus Targets, Continued

• In fact, the discussion of observed policies above does recognize that central banks used

the credit instrument as an alternative to discount rate changes.

• The key to preserve policy tradeoffs in the model is to require that both instruments not be

used together.

• Alternatively, if we wanted to allow for contemporaneous use of both instruments by central

banks in the model, we could preserve the existence of tradeoffs by expanding the range of

targets for central banks.

– The Dutch economist Jan Tinbergen, first winner of the Nobel Prize in economics, was

first to formalize the importance of number of policy objectives versus instruments when

studying the conduct of economic policy.

84

The Sustainability of Cooperation

• When we study policy cooperation in models such as the one we explored, we always face

the issue of what makes the cooperative equilibrium sustainable.

• The cooperative equilibrium, being a point on the contract curve in Figure 2, is outside the

best response function for at least one central bank and, in general, for both.

• This means that, if there is no mechanism that credibly binds both central banks to sticking

to the cooperative policies, the cooperative equilibrium will generally unravel, as each central

bank has an incentive to “cheat” on the central bank that is playing the cooperative action.

– If the foreign central bank plays the cooperative action associated with a certain point on

the contract curve that differs from the home central bank’s bliss point, the home central

bank can be made better off by cheating and playing the non-cooperative discount rate

action implied by its own best response function B.

• In the end, this shared incentive to cheat will result in both central banks finding themselves

in the inferior, non-cooperative equilibrium N in Figure 2.

• What can prevent this from happening?

85

The Sustainability of Cooperation, Continued

• A first answer is credible commitment mechanisms that force central banks to sticking to

international cooperation agreements.

– International institutions at the heart of the post-World War II Bretton Woods regime were

designed with the goal (among others) of facilitating and supporting cooperation.

• A second answer is the fact that the game between the central banks is not a one-shot,

one-time-only interaction.

• It is a repeated game that takes place many times over time.

• In such an environment, if central banks care sufficiently about the future, they may refrain

from “cheating” because they would recognize that the future costs of such cheating

(implied by the retaliation by the other central bank and the reversion to the non-cooperative

outcome) more than offset the short-run benefits from cheating on the cooperating partner.

86

Objective Functions

• Another issue that is worth mentioning briefly before concluding our discussion of

cooperation versus non-cooperation in the interwar GS is whether objective functions such

as those we posited for the central banks in our model should be taken as appropriate

representations of policy objectives.

• If we think that policymakers should care about the welfare of agents in the economy, it

is not clear that objective functions in terms of gold reserve and price level targets are an

accurate representation of, say, household welfare.

• Now, we know that the objective functions we posited were empirically plausible representa-

tions of central bank objectives under the interwar GS for reasons we discussed.

• We may argue that central banks at that time were not necessarily only concerned about

the welfare of agents in the economy.

• And we may argue that the same applies to policymakers in later times.

• Yet, when evaluating policy from a normative perspective (i.e., trying to address the question

of what a policymaker should do), we would like to do that in terms of a criterion that does

have a clear connection to agents’ welfare.

87

Objective Functions, Continued

• In modern macroeconomics, agents’ welfare is measured by the utility function that we

assume agents maximize with their choices of, say, consumption and leisure versus labor

effort.

• In modern macro frameworks, we evaluate policies in terms of whether or not they are

consistent with maximizing such measures of welfare.

• Much literature has been written over the years (some of which we will study later on) that

still posits objective functions for policymakers that are in terms of, say, deviations of inflation

and employment (or output) from targets rather than directly connected to household utility

from consumption and leisure versus labor effort (or other variables that may matter for an

agent’s utility).

• Michael Woodford of Columbia University and then other scholars have proven conditions

subject to which such objective functions can (in different scenarios) be taken as rigorous

approximations to household welfare as a function of consumption and leisure.

88

Objective Functions, Continued

• In the models we will see, like in the one we saw, there is no such foundation for the

policymakers’ objectives, and we simply take them as plausible empirical representations of

the policymakers’ preference functions.

• However, now you are aware of the question whether or not such policy objective functions

should be the metric for policy evaluation, of the fact that there are conditions subject to

which these objective functions are indeed a rigorous approximation to household welfare,

and of the fact that, in general, when these conditions are not satisfied, we can still think of

these objective functions as shortcuts to capture observed behavior by policymakers.

89

Lessons from the Great Depression and the Global Financial Crisis of 2007-08

• The dynamics of the GD and the different experiences of countries that left the GS quickly

or stuck with it longer explain why it would have been disastrous for the U.S. and other

countries if they still had been on the GS in 2007-08, and even more disastrous now.

• In fact, it was the lessons learned from studying the GD that informed Bernanke’s

policymaking during the GFC, and that ensured that we had a GR but not another GD.

• Not being on the GS allowed the Fed and other key central banks the flexibility to respond

aggressively and in coordinated fashion.

• The U.S. government implemented stimulus, others did too, though governments of some

key economies implemented less fiscal expansion than would have been desirable for a

variety of reasons, and, according to many commentators, even U.S. fiscal stimulus was

reversed too quickly.

• This contributed to a weak recovery from the GR that left permanent scars on the economy.

• It is also because of these lessons that we are seeing central bank cooperation and efforts

to implement as much monetary and fiscal expansion as possible today.

90



The Bretton Woods System and the Transition

to Floating Exchange Rates

Fabio Ghironi


CEPR, and NBER

Introduction

• Chapter 4 of Barry Eichengreen’s (2019) Globalizing Capital gives you a historical overview

of the Bretton Woods (BW) system that governed international monetary interactions from

the end of World War II to the beginning of the 1970s.

• These slides reproduce material from Michael Bordo’s chapter in a volume on A Retrospec-

tive on the Bretton Woods System that Bordo and Eichengreen edited in 1993, adding my

own explanations of key conceptual issues.

1

The Origins of the Bretton Woods System

• The planning that led to the creation of the BW system was predicated on the belief that the

mistakes of the interwar period were to be avoided.

• These mistakes, as they were perceived, included:

1. Wildly fluctuating ERs after WWI and the collapse of the interwar GS (the Gold Exchange

Standard – GES);

2. The international transmission of deflation and the resort to beggar-thy-neighbor devalua-

tions;

3. Trade and exchange restrictions and bilateralism.

• The goal was the negotiation of an international monetary constitution based on stable ERs,

national full employment policies, and cooperation.

• Three issues dominated the perception of the interwar experience: the flaws of the GES,

the case against floating ERs, and bilateralism.

2

The Problems of the Gold Exchange Standard

• There were three major problems in the functioning of the GES, the first two of which

received much attention from the architects of the BW system:

1. The adjustment problem,

2. the liquidity problem,

3. and the confidence problem.

3

The Problems of the Gold Exchange Standard, Continued

The Adjustment Problem

• The adjustment problem involved two issues:

– First, asymmetric adjustment between deficit and surplus countries, which contributed to

a deflationary bias,

– and, second, the failure by all countries to follow the “rules of the game.”

• The U.S. and France (with undervalued gold parities) together absorbed 53% of the world’s

monetary gold reserves by 1924.

• As surplus countries, they sterilized gold inflows to avoid the inflationary consequences of

monetary expansion.

– In terms of Ben Bernanke’s money supply equation that we studied, surplus countries

responded to gold inflows by reducing their money base in order to prevent money supply

from rising.

• At the same time, the steady drain of gold from Britain (with an overvalued parity) and other

deficit countries ultimately caused them to contract their money supplies.

4


• Sterilization by the surplus countries only worsened the gold drain from the deficit countries,

as sterilization prevented the price changes that would improve the competitive position of

deficit countries.

• Virtually every central bank followed a policy of offsetting changes in international reserves

by changes in domestic credit, rather than letting money supply adjust in a way consistent

with restoring external balance, hence breaking the “rules of the game.”

• This slowed the adjustment of relative national prices and incomes required to restore

balance of payments equilibrium under the GS, where a deficit (surplus) country should

have allowed its money supply to contract (increase) along with reserves in order to

induce price and income decreases (increases) that would contribute to restoring external

equilibrium.

5


The Liquidity Problem

• The interwar liquidity problem involved gold supplies that—at the prevailing set of gold

parities—were inadequate to serve as backing of national currencies and also finance the

growth of world output and trade.

• To economize on gold, countries other than the center ones used their holdings of the key

currencies as international reserves.

• But, as these currency holdings rose relative to the gold holdings of center countries,

peripheral countries had an incentive to reduce their currency holdings for fear of a

convertibility crisis.

6


The Confidence Problem

• The interwar confidence problem involved the transition from London to New York as primary

financial and reserve center for international currency transactions, and a shift between the

key currencies and gold.

• Shifts of foreign currency balances from now weak to strong reserve centers weakened the

system because they increased the likelihood of confidence crisis in the weak center.

• In turn, if such a crisis materialized, it would put pressure also on the strong center.

• A shift between the key currencies and gold occurred when foreign holders of key currency

balances, fearing the reserve centers’ inability to convert their outstanding liabilities into

gold at the fixed parity, staged a run on the “bank,” precipitating interruption of convertibility,

as happened to Britain in 1931.

7

Exchange Rate Volatility and Competitive Devaluations

• Estonian international economist Ragnar Nurkse published a very influential book on

International Currency Experience in 1944 in which he based the case against floating ERs

on the interwar experience.

• Nurkse argued that freely floating ERs inevitably led to destabilizing speculation, mostly

based on the French experience of 1922-1926.

• His argument was that French franc depreciation led speculators to anticipate further

depreciation, ensuring the outcome by virtue of self-fulfilling speculative capital flight.

• At the same time, the depreciation would worsen (rather than improve) the trade balance

because importers, anticipating further depreciation and higher prices, would import more in

the current period, and exporters, expecting the domestic currency price of exports to rise,

would hold back sales to get a better price later.

8

Exchange Rate Volatility and Competitive Devaluations, Continued

• Furthermore, according to Nurkse, the 1931 devaluation of the pound sterling led to a spate

of competitive devaluations involving considerable excess movement of ERs because of the

destabilizing nature of speculative capital flows.

• The common belief held by many economists and policymakers was that devaluations were

intended to “beggar thy neighbor”:

– Expand the domestic economy at the expense of its trading partners, whose competitive

position was hurt by the domestic devaluation.

• This belief was widely held even if Nurkse himself pointed out that overall world trade

increased in several cases in which devaluations were accompanied by monetary policy

expansion (so that devaluations did not simply redistribute a given amount of trade in favor

of the devaluing countries).

• Nurkse—and the architects of the BW system—believed that the competitive devaluations

and speculative capital flows of the 1930s inevitably led to the observed outcome of

a pervasive system of exchange controls (limiting both flows of goods and capital), to

protectionist trade policy responses, and to the fragmentation of the international economy

into a series of discriminating bilateral agreements.

9


• On the basis of his interpretation of the interwar experience, Nurkse actually envisioned a

world monetary system much along the lines of how BW was designed to be.

• To avoid the problems of floating ERs, he would have fixed ERs, but, in the event of structural

disequilibria in the BOP, changes in parity would be allowed rather than exchange controls

on current account transactions (i.e., transactions involving goods and services).

• In the case of short-run payments disequilibria, international reserves would serve as a

buffer.

• ER parities should be set by an international agreement rather than on a piecemeal basis

by each country.

10


• Monetary and fiscal policy would be coordinated between countries to ensure full

employment.

• Speculative capital flows would be avoided by pervasive capital controls.

• Finally, the deflationary bias of the interwar GS would be avoided by imposing discriminatory

exchange controls against “scarce” currencies.

– If a country ran persistent surpluses, its currency could become “scarce” as deficit

countries needed it to purchase disproportionately large imports from that country.

– International reserves flowed from the deficit countries to the surplus country, and

sterilization by the latter could result in deflationary pressure (as it did under the interwar

GS).

– To encourage “scarce” currency countries to let their money supplies grow with inflow

of reserves, Nurkse envisioned the possibility of discriminatory controls against such

scarce currencies.

11

Was the Perception of the Interwar Experience Correct?

• Many aspects of the perception of the flaws of the interwar GS have been challenged.

• Milton Friedman in a famous 1953 paper on “The Case for Flexible Exchange Rates” argued

that every case of destabilizing speculation that Nurkse documented involved a prospective

change in government policies that would otherwise have changed the ER anyway—that

the market just facilitated the movement to the new rate.

• Barry Eichengreen and Jeffrey Sachs in a 1985 article in the Journal of Economic History

presented strong evidence against the prevalence of beggar-thy-neighbor devaluations in

the 1930s.

• They showed that most devaluations were accompanied by expansionary monetary policy

that resulted in expansion of the overall GDP and trade “pie” rather than just a change in the

way a given-size pie was sliced.

• Put differently, Eichengreen and Sachs showed that devaluation with monetary expansion

was not a zero-sum game.

• Moreover, they showed that even the devaluations that were accompanied by monetary

policy expansion not did not significantly reduce income in other countries.

12

The Case for Flexible Exchange Rates

• To understand Milton Friedman’s case for flexible ERs, consider a world of two countries,

Home and Foreign.

• Each country is specialized in production of its own good: Home produces good h and

Foreign produces good f .

• Consumers in each country want to consume both the domestic good and the good

produced by the other country, and we assume that Home consumers and Foreign

consumers have identical preferences over the two goods.

13

The Case for Flexible Exchange Rates, Continued

• In particular, we assume that the Home and Foreign consumption baskets (C and C∗) are

given by, respectively:

C =c(h)ac(f )1−a

aa (1− a)1−aand C∗ =

c∗(h)ac∗(f )1−a

aa (1− a)1−a, (1)

where a star denotes the Foreign country, c(h) (c∗(h)) is Home (Foreign) consumption of the

Home good, c(f ) (c∗(f )) is Home (Foreign) consumption of the Foreign good, and 0 ≤ a ≤ 1.

– In each country, the consumption basket combines the domestic good and the imported

one in Cobb-Douglas fashion, with the same weights a and 1 − a attached to the Home

and Foreign goods, respectively.

– The larger is a, the more the Home good is important as a share of both Home and

Foreign consumption.

– The constant aa (1− a)1−a at the denominator is just a constant that simplifies algebra

below.

14


• Let p(h) and p(f ) be the prices of the two goods in Home (therefore, these prices are in

Home currency), and p∗(h) and p∗(f ) be the prices of the two goods in Foreign (therefore,

these prices are in Foreign currency).

• It is possible to show that consumption preferences as in (1) imply the following expressions

for the CPIs of the two countries (P and P ∗):

P = p(h)ap(f )1−a and P ∗ = p∗(h)ap∗(f )1−a. (2)

– If we did not have the constant aa (1− a)1−a at the denominators of the expressions in

(1), it would show up at the denominators of the expressions in (2).

• Now suppose that Home consumers have decided how much to consume of the overall

bundle C. How would they allocate their consumption between the Home good and the

Foreign good?

15


• It is also possible to show that their demand of the two goods would be determined by the

following expressions:

c(h) = a

(p(h)

P

)−1C and c(f ) = (1− a)

(p(f )

P

)−1C. (3)

– Home consumers would demand more of each good the more they want to consume of

the overall basked—the higher is C).

– They would demand more of the Home (Foreign) good the more the Home (Foreign)

good is important in the basket—the higher is a (respectively, 1− a).

– They would demand less of each good the higher its price relative to the overall price of

the basket.

• The behavior of Foreign consumers would be similar, and their demands of the two goods

would obey:

c∗(h) = a

(p∗(h)

P ∗

)−1C∗ and c∗(f ) = (1− a)

(p∗(f )

P ∗

)−1C∗. (4)

16


• Now consider the relative price p (h) /P that determines the Home consumer’s demand of

the Home good. Using the expression of the Home CPI:

p(h)

P=

p(h)

p(h)ap(f )1−a=

(p(h)

p(f )

)1−a. (5)

• Similarly, the relative price that matters for the Home consumer’s demand of the Foreign

good is such that:

p(f )

P=

p(f )

p(h)ap(f )1−a=

(p(f )

p(h)

)a=

(p(h)

p(f )

)−a. (6)

• Similar equations hold in Foreign:

p∗(h)

P ∗=

(p∗(h)

p∗(f )

)1−aand

p∗(f )

P ∗=

(p∗(h)

p∗(f )

)−a. (7)

17


p(h)

P=

(p(h)

p(f )

)1−a,

p(f )

P=

(p(h)

p(f )

)−a,

p∗(h)

P ∗=

(p∗(h)

p∗(f )

)1−aand

p∗(f )

P ∗=

(p∗(h)

p∗(f )

)−a.

• Equations (5)-(7) (and the demand equations (3)-(4)) tell us that the allocation of demand in

each country crucially depends on the relative price: the price of the Home good relative to

the Foreign good: p (h) /p(f ) in Home and p∗ (h) /p∗(f ) in Foreign.

• In each country, that price ratio is the ratio between how many currency units it takes to buy

a unit of the Home good and how many currency units it takes to buy a unit of the Foreign

good.

• Hence, the price ratio tell us how many units of the Foreign good it takes to exchange for

the Home good.

• When this increases, the Home good is becoming more expensive (it is worth more units of

the Foreign goods), and demand shifts away from the Home good and toward the Foreign

good.

• By how much? It depends on the size of the share parameter a.

18


• Now assume that trade is frictionless (there are no trade barriers) and prices are flexible.

• This implies that arbitrage will ensure that the Law of One Price (LOP) holds: The price of a

given good is equal in the two countries when expressed in a the same currency:

p(f ) = εp∗(f )...and p∗(h) =p(h)

ε,

where ε is the exchange rate (units of Home currency per unit of the foreign currency).

• Using the LOP, we have

p(h)

p(f )=

p(h)

εp∗(f ).

• And

p∗(h)

p∗(f )=

p(h)ε

p∗(f )=

p(h)

εp∗(f ).

• Because of the LOP, the allocation of demand between the two goods is determined by

exactly the same relative price in both countries: The price of the good Home produces

(in Home currency) relative to the price of the good Home imports (in Foreign currency,

converted into Home currency by the exchange rate).

• This relative price is also known as Home’s terms of trade (TOT).

19


• Now consider the type of thought exercise we did when we studied Hume’s price-specie

flow model: Suppose there is an asymmetric shock between the two countries—say, a new

technology is introduced in Home that implies an increase in the supply of the Home good.

• We know that, at pre-shock prices, we have an excess supply of the Home good and an

excess demand of the Foreign good.

• For markets to clear, demand (at Home and in Foreign) has to shift toward the Home good

and away from the Foreign good.

• For this to happen, the ratio p (h) / (εp∗(f )) must fall, because this will cause both price ratios

p(h)/p(f ) and p∗(h)/p∗(f ) that determine the allocation of demand in Home and Foreign to

fall, and it will cause consumers to shift their demand toward the good whose supply has

become more abundant.

• This happens immediately and without problem if all prices are flexible and respond

immediately to economic shocks.

20


• But now suppose that prices are not flexible.

• Suppose that Home producers had set the price of the Home good for the Home market in

Home currency before the realization of the shock.

• Denote this pre-set price with p(h).

• Suppose that Home producers let the price of the Home good in the Foreign country be

determined by the LOP, so that

p∗(h) =p(h)

ε.

• Note that this implies that p(h) does not move immediately in response to the shock, but

p∗(h) can still move if ε does.

21


• Similarly, suppose that Foreign producers had set the price of the Foreign good for the

Foreign market in Foreign currency before the realization of the shock.

• Denote this pre-set price with p∗(f ).

• Suppose that Home producers let the price of the Home good in the Foreign country be

determined by the LOP, so that

p(f ) = εp∗(f ).

• Note that this implies that p∗(f ) does not move immediately in response to the shock, but

p(f ) can still move if ε does.

22


• Consider the ratio p (h) / (εp∗(f )) that we know to be crucial for the allocation of consumer

demand in response to the shocks.

• Under the assumptions we are making, this ratio becomes:

p(h)

εp∗(f ).

• The prices p(h) and p∗(f ) cannot adjust immediately to the shock.

• In order for our world economy to adjust efficiently to the shock, it has to be the case that ε

is allowed to move, in particular, that the Home currency is allowed to depreciate (ε rises).

• If the exchange rate cannot move, then the adjustment of the Home and Foreign economies

to the shock is hampered until the prices can finally adjust, delaying the process of

international adjustment.

• This mechanism is at the heart of Friedman’s case for flexible exchange rates:

– We need exchange rates to move to bring about the efficient adjustment of international

relative prices (TOTs) to asymmetric shocks in an environment in which prices are sticky

and therefore they cannot move to deliver the desired expenditure switching effects.

23


• Friedman’s argument in favor of flexible exchange rates is crucially important and it is still

the one that you are going to hear most frequently mentioned in literature and media.

• Whenever you hear someone say (or read someone write) that countries must have the

ability to let their exchange rates move to respond to asymmetric shocks, you can bet on the

fact that the argument has Friedman at its core.

• In the mid-to-late 1990s, Maurice Obstfeld (U.C. Berkeley) and Kenneth Rogoff (Harvard

University) argued that European countries were making a bad mistake by adopting the

euro precisely on the basis of Friedman’s argument.

• The Euro Area crisis that began in 2010 re-heated reheated that discussion, with many

suggesting that countries like Greece and Italy should leave the euro and regain the ability

to rely on the ER as a shock absorber.

• We will talk later on about reasons why, even taking Friedman into account, countries may

still want to give up ER flexibility today, but first let us mention a reason why Friedman’s

argument may not have the general validity that many automatically attach to it.

24


• Recall something important: In building Friedman’s case for ER flexibility, we used the LOP

assumption:

– No trade barrier and producers that were setting prices only in their own currency, letting

the price in their export market be determined by the LOP.

• With respect to price setting, we made an assumption that is known among international

macroeconomists as producer currency pricing (PCP).

• Now suppose that producers do not engage in PCP, but they actually set prices in the

currency of the destination market:

– Home producers set a Home currency price for the Home market and a Foreign currency

price for the Foreign market, Foreign producers set a Foreign currency price for the

Foreign market and a Home currency price for the Home market.

– In other words, producers now engage in what is known as local (or consumer) currency

pricing (LCP): They set prices in the currency of consumers, not letting passively any

price be determined by the LOP.

25


• It is possible to verify that the LOP would still hold in a world of flexible prices.

• But now consider what happens if prices were pre-set relative to the shock that increases

the supply of the Home good in our example above.

• The prices of the Home good in the Home market (p(h)) and in the Foreign market (p∗(h))

were both pre-set by Home producers at levels p(h) and p∗(h) before the shock happened.

• The prices of the Foreign good in the Foreign market (p∗(f )) and in the Home market (p(f ))

were both pre-set by Foreign producers at levels p∗(f ) and p(f ) before the shock happened.

• The LOP no longer holds:

– Because both p(h) and p∗(h) are pre-set, movements of ε can no longer deliver

p∗(h) = p(h)/ε, and because both p∗(f ) and p(f ) are pre-set, it can no longer be the case

that p(f ) = εp∗(f ).

– Note: For deviations from the LOP to happen, it must be the case that the firms we are

talking about have the ability to segment markets (for instance, through country-specific

distribution networks). We are implicitly assuming that this is what happens in this

scenario.

26


• The crucial relative price ratios

p(h)

p(f )and

p∗(h)

p∗(f )

that ultimately matter for the allocation of consumer demand between the two goods in the

two economies are completely pre-determined and insulated from ER variation.

• In fact, both CPIs are completely pre-determined and cannot respond to the shock on

impact, because:

P = p(h)ap(f )1−a and P ∗ = p∗(h)ap∗(f )1−a.

• In this world, the ER has no expenditure switching role to play in the short-run adjustment to

shocks.

• This is the world that was studied by Caroline Betts (USC) and Michael Devereux (UBC) in

a paper published in the Journal of International Economics in 2000.

27


• Devereux and Charles Engel (now at the University of Wisconsin, then at the UW)

documented in their work in the late 1990s that LCP was a pervasive phenomenon in

European countries and they studied the implications of LCP for the optimal conduct of

monetary policy.

• They showed that a fixed exchange rate is optimal under LCP (for instance, see their 2003

Review of Economic Studies article).

• The reason is that exchange rate movements cannot deliver the expenditure switching at

the core of Friedman’s idea but instead introduce income volatility (it is still the case that, for

instance, a Home producer’s revenue generated by selling its good to Foreign consumers,

is subject to the effect of ER volatility once converted back into Home currency).

• In this environment, a fixed ER is the best monetary policy.

28


• Obstfeld and Rogoff on one side and Devereux and Engel on the other were the leaders of

two camps that generated very interesting discussions and research between the mid-to-late

1990s and the mid 2000s.

• Naturally, scholars developed a variety of models in which one or the other assumption

about price setting did not deliver the expected result and/or one or the other mechanism

was downplayed.

• The discussion on the importance of the role of the currency of trade invoicing and price

setting are still ongoing today.

• See for instance the work on the so-called dominant currency paradigm (DCP) that the

current IMF Chief Economist Gita Gopinath has been doing with some coauthors (including

Pierre-Olivier Gourinchas).

29


• Under DCP firms in the Home and Foreign countries choose to invoice their trade and to set

traded-good prices in the “dominant” currency of a third country: the dollar, which became

the predominant international currency because of how the BW system worked.

– Gopinath’s work was actually preceded by work on the choice of the currency of invoicing

and price setting that was done in the mid-to-late 2000s by Devereux and Engel with

Peter Storgaard (Danmarks Nationalbank), and by Linda Goldberg (NY Fed) and Cedric

Tille (now at the University of Geneva, then at the NY Fed).

• Discussions on these topics are still far from settled, but what we covered gives you a good

idea of the importance of certain assumptions, the key arguments they supported, and

ultimately how they contributed to thought processes by both researchers and policymakers.

30

The Perception of the Interwar Experience and the Design of the Bretton

Woods System

• In hindsight, the breakdown of the interwar GS owed more to inappropriate policies by

France and the U.S. than to the perceived structural flaws of the system, although it is the

case that adherence to gold convertibility during the GD was a key determinant of worldwide

deflation and depression.

• But the perception of the flaws of the interwar GS was crucial in the design of the BW

system.

• The planning and bargaining that took place both in the U.S. and Britain during WWII

and leading to the BW conference of 1944 reflected both a common vision of the future

international system and different national concerns.

31

Common Beliefs, Different Expectations, Different Priorities

• The architects on both sides of the Atlantic wanted a system that would avoid the defects of

the interwar period and promote world peace.

• Common beliefs included a multilateral payment system, stable ERs, and full employment.

• However, key differences in the positions of power and economic importance of the two

allies dictated quite different national concerns.

• The U.S. was about to emerge from WWII as the strongest and richest power in the world.

– It expected to be a creditor nation, running BOP surpluses.

• Britain’s resources and power were greatly weakened by the war.

– It expected to be a debtor nation, running BOP deficits.

32

Common Beliefs, Different Expectations, Different Priorities, Continued

• These expectations were reflected in the priorities of the two key delegations at the Bretton

Woods conference: the British delegation, led by John Maynard Keynes, and the U.S.

delegation, led by Harry Dexter White.

• The Roosevelt administration attached great importance to the elimination of discriminatory

trade and exchange controls by the British, especially a system of imperial preference

negotiated at Ottawa in 1931, and of bilateral agreements.

• Restoration of a multilateral payments system based on convertible currencies was

paramount.

• This system should allow free, multilateral trade in goods and services without restricting

the ability of agents to convert across currencies for the purposes of such trade.

• The U.S. administration viewed this as central to a system that would support long-term

international peace by ensuring the most beneficial allocation of resources and productive

activities and, most importantly, a system in which American companies (which had not

faced the devastation that European and Asian competitors had faced) would flourish.

33

Common Beliefs, Different Expectations, Different Priorities, Continued

• On their side, the British were most concerned that the U.S. not repeat the deflationary

policies of the 1930s.

• They wanted the freedom to pursue domestic full employment policies without concern over

the state of the BOP.

• They also wanted the preservation of the British Commonwealth, assistance in wartime

reconstruction, and relief in the repayment of outstanding debt accumulated during the war.

34

Negotiations and Plans

• Wartime negotiations leading to the Atlantic Charter (August 14, 1941) and the Mutual Aid

Agreement (February 23, 1942) led to a compromise between the two nations, setting the

planning for BW.

• The British accepted a multilateral payment system in which the pound sterling would

become freely convertible with other currencies in exchange for a U.S. commitment to

maintain full employment, the preservation of the Commonwealth, generous terms on

lend-lease (the special arrangement for lending from the U.S. to Britain during the war), and

U.S. assistance in the postwar recovery.

• Against this background of negotiations, the British Treasury team of planners led by

Keynes and the American team led by White drafted two competing plans for the postwar

international monetary system.

• Final versions of the plans, known as the Keynes and White plans, were published in 1943.

• Following intense negotiation, a compromise between the two plans led to the Joint

Statement by Experts on the Establishment of an International Monetary Fund (1944).

• The Joint Statement served as the working draft at the BW conference and led directly to

the Articles of Agreement of the International Monetary Fund (1944).

35

The Keynes Plan

• The Keynes plan was designed to encourage the expansion of world trade and activity by

the generous provision of international liquidity.

• It was also designed to shield the domestic economy from foreign disturbances via the

provision of a buffer stock of international reserves.

• Recall a country’s intertemporal net foreign debt sustainability constraint:

0 = (1 + r)(Bt + RESt) +

∞∑s=t

(1

1 + r

)s−tTBs.

• Think of period t as the beginning of the BW system.

• Keynes led the delegation of a country that would be starting life under BW in debt

(Bt < 0), deficit (TBt < 0), and that expected to remain in deficit for several years:

TBt+1 < 0, TBt+2 < 0, and so on.

• In essence, Keynes wanted the system to be endowed with large liquidity such that the

country could access the equivalent of large RESt > 0, which would make it possible

to sustain several years of deficit before being forced to adjust policies in contractionary

fashion to stabilize the BOP.

36

The Keynes Plan, Continued

• Operationally, the heart of the plan was the establishment of a supranational central bank,

the International Clearing Union (ICU), that would issue a new international currency to be

called bancor.

• The nominal value of the bancor was to be fixed in terms of gold.

• Every national currency would then set its par value in terms of bancor.

– Gold could be paid into the ICU and would serve as reserve, but bancor could not be

redeemed in gold.

– Gold could also be used in settlements between members.

• The central bank of each member nation would keep accounts with the ICU to settle

balances between other members at par in bancor.

37


• Surplus nations would maintain credit balances earning interest.

• Deficit nations could settle their balances by obtaining overdrafts that would bear interest

and that would be transferred to the credit of the surplus countries.

• Each country would be assigned a quota to determine the limit on resources it could obtain.

• The plan would provide generous liquidity—between $25 and $30 billion.

38


• The Keynes plan also imposed extensive regulations governing the balances of both debtors

and creditors.

• Debtors faced increasingly stiff penalties as they used up successive portions of their

quotas, the penalties including having to devalue and the imposition of capital controls.

• Creditors were to take measures including expansionary domestic credit, appreciation of

their currencies, cutting tariffs, and extending international development loans.

• Creditors would bear more of the burden of adjustment because there was no limit on the

amount of bancor liquidity they would have to accept (liquidity that could not be used for

other purposes), whereas debtors were limited by their quotas.

• A final important aspect of the Keynes plan was the provision of permanent capital controls

to prevent destabilizing speculation against the fixed ER parities.

39

The White Plan

• The White plan put more emphasis on ER stability and less on the generous provision of

international liquidity.

• The key institution was the creation of a United Nations Stabilization Fund.

• Each member would contribute to the fund a quota consisting of gold and its own currency.

• The total subscription was to be $5 billion.

• A deficit country would draw resources from the Fund by selling its currency for that of

another member, rather than running an overdraft with the ICU as in the Keynes plan.

• Consequently, the balances of its currency at the Fund would increase, while those of the

member whose currency had been drawn would decline.

40

The White Plan, Continued

• Each member would declare a par value for its currency in terms of unitas, an international

unit of account worth ten gold U.S. dollars.

• Each member was obliged to maintain that par value, except in the event of a fundamental

disequilibrium in the BOP, when it could be altered.

• In that case, if the proposed change was less than 10%, it could be made after consultation

with the Fund.

• If larger, it needed approval by three-quarters of the members.

41

The White Plan, Continued

• As with the Keynes plan, various penalties were imposed on debtor countries when their

borrowings exceeded their quotas.

• These included the Fund’s suggestion of appropriate domestic policies to facilitate

adjustment.

• Less pressure was placed on creditor countries.

• However, a scarce currency clause (if a surplus country’s currency became “scarce” for

other member countries) allowed the Fund to recommend rationing its use by exchange

controls.

42

The Compromise

• In the months that followed the publication of the two plans, an initial compromise was

reached in the Joint Statement, and further compromise was made by the time of the

Articles of Agreement.

• In the compromise, the British gave up the ICU, bancor, the overdraft system, and the

generous provision of liquidity, settling for $8 billion in the Joint Statement, $8.8 billion at

BW.

• The British gained greater national policy autonomy for members, who were allowed

discretion over changes in the ER (with Fund approval required to correct fundamental

payments disequilibria—though precisely what constituted a fundamental disequilibrium

was never defined clearly, and this turned out to be a major problem).

• The British also gained the authorization of capital controls in the agreement.

43

The Compromise, Continued

• The Fund’s discretion to intervene in domestic policymaking to facilitate adjustment was

eliminated, as was explicit conditionality on credit drawings.

• The scarce currency clause was viewed as a solution to British concerns over the

deflationary potential of the U.S. running a large surplus.

• Finally, British requests for postwar assistance were dropped from the negotiations, although

they were met in part by a transition clause allowing exchange controls on current account

transactions for a number of years, by the exclusion of outstanding sterling liability balances

from the convertibility requirements, and by the Anglo-American Loan of $3.75 billion on

December 6, 1945.

44

The Compromise, Continued

• A number of special conditions may have contributed to the successful outcome of the BW

negotiations:

1. The U.S. and Britain (with the aid of Canada) were able to work out an agreement that was

a compromise between their two national interests without involving other countries that

were either belligerents or under occupation.

2. As the strongest economic power, the U.S. was able to dominate the terms of the

agreement.

3. A strong sense of idealism prevailed in both countries—that they had an obligation to

create a system that would promote lasting peace.

4. The negotiators on both sides of the Atlantic shared a common set of perceptions of the

problems of the interwar period, had common views on the importance of full employment

and a liberal multilateral payment system, and greatly respected John Maynard Keynes.

45

The Articles of Agreement

• The Articles of Agreement incorporated elements of both the Keynes and the White plans,

although, in the end, U.S. concerns predominated.

– At the same time as the Articles of Agreement for the International Monetary Fund (IMF)

were signed, the International Bank for Reconstruction and Development (the World

Bank) was established.

– The Charter of the International Trade Organization (ITO) was signed in 1947, but never

ratified.

– It was succeeded by the General Agreement for Tariffs and Trade (GATT), originally

negotiated in Geneva in 1947 as an interim institution until the ITO came into force.

46

The Articles of Agreement, Continued

• The stated objectives of the IMF were:

– to promote international monetary cooperation,

– to facilitate the maintenance of full employment and rapid growth,

– to maintain stable ERs and avoid competitive devaluations,

– to provide a multilateral payments system and eliminate exchange restrictions,

– to provide resources to meet BOP disequilibria without resort to drastic measures,

– and to shorten the duration and lessen the degree of payments disequilibria.

• The main points of the Articles were the creation of the par value system, multilateral

payments, the use of the Fund’s resources, the Fund’s powers, and its organization.

47

The Par Value System

• Article IV defined the numeraire of the international monetary system as either gold or the

U.S. dollar of the weight and fineness on July 1, 1944.

• All members were urged to declare a par value and maintain the ER rate within a 1% margin

on either side of parity.

• Parities could be changed in the event of a fundamental disequilibrium in the balance of

payments at the decision of the members, after consultation with the Fund.

• However, the Fund would not disapprove the change if it was less than 10%, and, if it was

more than 10%, the Fund would decide within seventy-two hours.

• Unauthorized ER changes could make members ineligible to use the Fund’s resources and,

if they were to persist, could lead to a member’s expulsion.

• A uniform change in the parity value of all currencies (in terms of gold) required a majority

of the total voting power and also had to be approved by every member with 10% or more

of the total quota.

48

Multilateral Payments

• Members were supposed to make their currencies convertible across each other for

current account transactions (trade in goods and services, Article VIII), but capital controls

(controls preventing residents from converting currencies for purposes of capital account

transactions) were permitted (Article VI.3).

• Members were also to avoid discriminatory currency arrangements.

• The convertibility requirement applied to foreign balances that would be accumulated for

current account transactions but exempted previously accumulated balances (presumably,

to protect Britain from the effect of conversion of pound sterling liabilities accumulated by

the end of World War II).

49

Multilateral Payments, Continued

• Importantly, countries could avoid declaring their currencies convertible immediately upon

joining the system by invoking Article XIV, which allowed a three-year transition period after

establishment of the IMF.

• During this transition period, existing exchange control restricting the ability to convert

currencies for current account transactions could be maintained.

• After the three-year grace period, the Fund was to report on the state of convertibility of the

currency in question.

• After five years and every year thereafter, under Article XIV, the members had to justify their

position to the Fund in individual consultation.

50

Multilateral Payments, Continued

To Make Sure You Understand: Convertibility in the Gold Standard Versus the Bretton

Woods System

• – Under the classical GS, convertibility referred to the ability of a private individual freely to

convert a unit of any national currency into gold at the official fixed price.

– A suspension of convertibility meant that the ER and a national currency became flexible,

but the individual could still freely transact in either asset.

– By the late 1930s, convertibility referred to the ability of a private individual freely to make

and receive payments in international transactions in terms of the currency of another

country.

– Under BW, convertibility meant the freedom for individuals to engage in current account

transactions (trade in goods and services) without being subject to exchange controls.

– This was sometimes referred to as market convertibility to distinguish it from official

convertibility, whereby the central bank of each country must be freely willing to buy and

sell foreign exchange (primarily dollars) to keep its exchange rate within 1% margins

around the fixed parity and the U.S. must be freely willing to buy and sell gold to keep the

dollar price of gold within 1% margins around the fixed parity of $35.00 per ounce.

51

The Fund’s Resources

• As under the White plan, members could obtain resources from the IMF to help finance

short- or medium-term disequilibria of their BOPs.

• The total Fund, contributed by members’ quotas (25% in gold, 75% in currencies), was set

at $8.8 billion.

• It could be raised every five years if the majority of members wanted to do so.

• The Fund set a number of conditions on the use of its resources by deficit countries to

prevent it from accumulating deficit-country currencies and depleting its holdings of the

currencies of surplus countries.

– Members could draw on their quotas without condition.

– But once they borrowed beyond their quotas, increasingly more exacting conditions were

required.

52

The Fund’s Resources, Continued

• The Fund also established requirements and conditions for repayment of loans, including

having the right to decide the currency in which the repayment would be made.

• In the case of countries prone to running large surpluses, the scarce currency clause (Article

VII) could come into play.

• If the Fund’s holdings of a currency were insufficient to satisfy the demand for it by other

members, it could declare that currency scarce and then urge members to ration its use by

discriminatory exchange controls.

53

The Powers of the Fund

• The Fund had considerably less discretionary power over the domestic policies of its

members than either Keynes or White wanted, but it still had the power to strongly influence

the international monetary system.

• It had the authority to approve or disapprove of changes in parity, discriminatory practices,

and the conditionality that was implicit in members’ access to borrowing in excess of their

quotas.

• It could declare currencies scarce, declare members ineligible to use its resources (as it did

against France in 1948), and expel members.

• It also had considerable power as the premier international monetary organization in

consulting and cooperating with national and other international monetary authorities.

54

The Fund’s Organization

• The Fund was to be governed by a board of governors appointed by the members.

• The board would make the major policy decisions, such as approving parity changes.

• Operations of the Fund were to be directed by executive directors, appointed by the

members, and a managing director, selected by the executive directors.

• Major changes, such as a uniform change in the par value of all currencies relative to gold

or the Second Amendment creating the Special Drawing Right (SDR—see below), would

require a majority vote by the members.

• The number of votes in turn was tied to the size of each member’s quota, which was

determined by its economic size.

55

How Was the Bretton Woods System Supposed to Work?

• The architects never spelled out exactly how the system was supposed to work.

• However, subsequent research has suggested a number of salient features.

• John Williamson (Peterson Institute for International Economics) described the BW system

in a 1985 American Economic Review (Papers and Proceedings) article as a comprehensive

set of rules for assigning macroeconomic policies to objectives:

– Using ER parity adjustment when needed to maintain medium-run external balance,

– using monetary and fiscal policy to achieve short-run internal balance (internal policy

objectives),

– and using international reserves to provide a buffer stock to allow short-run departures

from external balance.

56

How Was the Bretton Woods System Supposed to Work? Continued

A Constellation with Gold and the Dollar at Its Center

• Currencies were treated as equal in the Articles of Agreement.

• This meant that, in theory, each country was required to maintain its par value by intervening

in the currency of every other country (buying and selling the currency of every other

country).

• In practice, because the U.S. was the only country that pegged its currency in terms of gold

(bought and sold gold in exchange of its currency), all other countries would fix their parities

in terms of dollars and intervene to keep their ERs within 1% of parity with the dollar.

57

How Was the Bretton Woods System Supposed to Work? Continued

International Reserves, Medium-Term Disequilibria, and Fundamental Disequilibria

• Countries would use their international reserves or draw resources from the IMF to finance

payments deficits.

• In the case of surpluses, countries would temporarily build up foreign currency reserves or

use accumulated foreign currency to repurchase their currencies from the Fund.

• In the event of medium-term disequilibria, countries would use monetary and fiscal policy to

alter aggregate demand and restore equilibrium.

• In the event of a fundamental disequilibrium, presumably reflecting either some structural

shock or sustained inflation, a member was supposed to adjust its relevant exchange rate

parities by an amount sufficient to restore external equilibrium.

• Capital controls were required to prevent destabilizing speculation from forcing members to

alter their parities prematurely or unintentionally.

58

Reality Turned Out Different

• Notwithstanding its careful design and the intentions of its architects, the system that began

operations after the BW conference and the establishment of the IMF ended up functioning

differently in many major respects from what the architects expected and intended.

• Contrary to expectations, the positions of the U.S. as a surplus country and those of key

partners, such as Western Europe and Japan, as deficit countries were quickly reversed.

• Multilateral convertibility was established only in 1958, after a first attempt in the late 1940s

resulted in a large devaluation of the pound sterling, a wave of devaluations by other

countries, and the reintroduction of restrictions to convertibility.

• The Marshall Plan implied that European countries ultimately had access to an amount of

international liquidity similar to that envisioned by the Keynes plan rather than the smaller

White-plan-based resources of the IMF.

• Rising dollar balances held by other countries created fear of a convertibility crisis—since

the dollar was ultimately supposed to be convertible into gold.

59

Reality Turned Out Different, Continued

• Ultimately, U.S. macroeconomic policymaking that was inconsistent with the responsibility of

the system’s center country to deliver price stability for the entire system, and disagreement

with other key countries over policy priorities, led to the collapse of the system at the

beginning of the 1970s, with President Nixon closing the “gold window” on August 15, 1971.

• The following slides focus on the key conceptual issues to understand the events

summarized in the previous bullets.

• Additional historical details are in the Appendix.

60

Rapid Role Reversals

• The two figures that follow (taken from Bordo’s chapter) illustrate the rapid role reversals

between surplus and deficit positions that happened shortly after

• After the end of World War II, Europe ran a massive current account deficit, reflecting the

demand for essential imports and the reduced capacity of export industries.

• The deficit of the Organization for European Economic Cooperation (OEEC) countries,

aggravated by the exceptionally bad winter of 1946-47, reached a 1947 high of nearly $7

billion.

• This, combined with other countries’ deficits, matched the U.S. current account surplus,

since, as the only major industrial country operating at full capacity, the U.S. supplied the

needed imports.

• Deficit countries were extremely concerned by the risk of shortage of dollars with which to

settle their deficits.

• Yet, as the figures show, the problem was gone by the mid-1950s.

61

Rapid Role Reversals, Continued

• Several factors contributed to this:

1. the Marshall Plan,

2. the European Payments Union (EPU),

3. the 1949 devaluation of the pound sterling and the wave of devaluations it triggered.

62

The Marshall Plan

• The Marshall Plan funneled approximately $13 billion in aid to Western Europe between

1948 and 1952.

• U.S. aid was to pay for essential imports and provide international reserves.

• Each recipient government committed to providing matching funds in local currency to be

used for investment in the productive capacity of industry, agriculture, and infrastructure.

• By 1952, in part thanks to the Marshall Plan, the OEEC countries had achieved a 39%

increase in industrial production, a doubling of exports, an increase in imports by one-third,

and a current account surplus.

63

The Marshall Plan, Continued

• According to Barry Eichengreen, the Marshall Plan permanently raised the growth rate of the

recipients, but not through investment and government spending in productive infrastructure.

• It did so by improving productivity and investor confidence.

• In turn, this was achieved by reducing the political instability associated with the “war of

attrition” between workers and property owners and by filling the role usually played by

foreign private investment.

64


Italy

• The example of Italy illustrates the importance and deep motives of the Marshall Plan.

• Italy became a republic shortly after World War II, with a referendum held on June 2, 1946.

• In the years that immediately followed, the key political conflict was between the Christian

Democratic Party and the Communist Party.

• The Communist Party came close to parliamentary majority, but never accomplished it.

• A communist-governed Italy would have been a terrible blow to U.S. strategic interests in

Europe and the Mediterranean, as the country would have obviously shifted toward the

Soviet bloc.

• The Marshall Plan contributed to a successful enough economic performance in those

crucial years that the majority of the Italian electorate chose to reward the more conservative

Christian Democrats, and Italy never fell under communist control.

65


• U.S. geopolitical interests and strategy—the goal of facing the Warsaw Pact with a strong

Western Europe committed to the U.S. alliance—were thus key to the function of the

Marshall Plan.

• This, in turn, gave Italy (and other European countries) the international liquidity that Keynes

would have liked European countries to have based on his plan for the BW system.

• And this helped change the functioning of the BW system relative to what was expected to

happen when the system was designed.

66

The EPU and the Return to Convertibility

• It took twelve years from the declaration of official par values by thirty-two nations in

December 1946 to the achievement of convertibility for current account transactions by the

major industrial countries as specified by the BW Articles.

• Western European countries tried several schemes to facilitate the payments process before

establishing the EPU.

• The EPU, established on September 19, 1950 by the OEEC countries, functioned essentially

like a commercial bank clearinghouse.

• At the end of each month, each member would clear its net debit or credit position against

all other members with the EPU (the Bank for International Settlements, BIS, acting as its

agents).

• The unit of account for these clearings was the U.S. dollar.

• The EPU was extremely successful in reducing the volume of payments transactions (by

providing in-house multilateral clearing), facilitating trade between the member countries,

and in providing the background for the gradual liberalization of payments that led to the

eventual restoration of multilateral convertibility in 1958.

67

The 1949 Devaluation of the Pound Sterling

• Finally, the 1949 devaluation of the pound sterling helped stabilize the international system

by reducing both European deficits and the U.S. surplus (and also strengthen the UK

economy).

• As soon as it was established, the IMF put pressure on member countries to declare par

values and currency convertibility as soon as possible.

• The logic was that even if countries declared overvalued parities, the rules of the BW system

would later allow them to adjust those parities as needed anyway.

• The UK and other countries succumbed to the pressure and made an initial attempt to

declare convertibility in 1947.

• But the same mistake as Churchill’s after World War I was repeated, with the choice of

pre-World War II parities for several currencies.

68

The 1949 Devaluation of the Pound Sterling, Continued

• Although the UK BOP and current account deficits (in dollars and gold) were shrinking in

1948 ad 1949, a U.S. recession reduced the demand for British goods in the second quarter

of 1949.

• This, coupled with the growing belief that the pound sterling would be devalued, triggered a

speculative attack on the currency.

• On September 18, 1949, twenty-four hours after the IMF was informed, the pound sterling

was devalued by 30.5% to $2.80 per sterling.

• Shortly thereafter, twenty-three countries devalued their parity by, in most cases, similar

magnitudes.

69

The 1949 Devaluation of the Pound Sterling, Continued

• The devaluations of 1949 were important for the BW system for two reasons.

• First, along with the Marshall Plan aid, they helped more the European countries from

current account deficit to surplus, a movement important to the eventual restoration of

convertibility.

• Second, they revealed a basic weakness of the adjustable peg BW arrangement:

– The one-way option of speculation against parity.

• By allowing parity changes only in the event of fundamental disequilibrium, the BW system

encouraged monetary authorities to delay parity adjustment until they were sure it was

necessary.

• By that time, speculators would also be sure, and they would take a position from which

they could not lose:

– If the currency was devalued, they won;

– If it was not, they lost only the interest (if any) on the speculative funds.

· Interest on speculative funds is paid when the speculator, anticipating a currency’s

devaluation, borrows funds in that currency to sell it in exchange for another currency.

· If the first currency does not devalue, the interest to be paid on the initial borrowing is

a net loss for the speculator.70

The Emergence of the Dollar as Key Currency

• During the period between 1947 and 1958, the dollar emerged as the key currency of the

international monetary system.

• The sterling was the dominant reserve currency at the beginning of the period, but it was

eclipsed by the dollar by the end of the 1950s.

• Because of the size of the U.S. economy, its importance in world trade, and its open

and deep capital markets, the dollar emerged in the 1950s as the dominant international

currency for private transactions.

• It was used as a unit of account in invoicing imports and exports, as a medium of exchange

in serving as a vehicle currency for interbank transactions, and as a store of value for private

claims.

71

The Emergence of the Dollar as Key Currency, Continued

• Simultaneously, the dollar emerged as the official international money.

• This stemmed from its use as a unit of account to define the parities of member countries in

the IMF.

• The dollar was also used as the primary intervention currency:

– Members maintained their fixed parities by buying and selling dollars.

• Finally, because of its role as a unit of account and a medium of exchange and its growing

private acceptance, it became the dominant international store of value to be used as

reserve.

72

The Emergence of the Dollar as Key Currency, Continued

• The growing private and official demand for dollars in the global economy was supplied

through private and official long-term capital outflows in excess of current account surpluses

(when there were current account surpluses), which produced a series of official settlements

BOP deficits beginning in 1950.

• By 1958-59, the U.S. BOP deficit became a source of policy concern.

73

Summary: The Bretton Woods System in 1958

• The convertible BW system that began at the end of 1958 differed in a number of ways from

the system intended by its architects.

• These included:

– the dominance of the U.S. in the international monetary order (soon to be challenged by

a reemerging continental Europe),

– reduced prestige of the IMF (see the Appendix),

– the decline of the sterling and the rise of the dollar as key currency,

– a shift from the adjustable peg system toward a de facto fixed ER regime,

– and, finally, growing capital mobility.

• Despite the prevalence of capital controls in most countries, private long-term capital

mobility increased considerably in the 1950s, and speculative short-term capital movements

emerged as a powerful force.

• While the outlook for the BW system never looked brighter than in December 1958,

emerging signs of weakness were soon to be revealed.

74

The Heyday of Bretton Woods: 1959-1967

• After the introduction of convertibility, the full-fledged BW system was in operation:

– The U.S. dollar was pegged to gold.

– Other currencies were pegged to dollar with ± 1 percent margins of fluctuations around

the parity.

– This implied cross-currency pegs with ± 2 percent margins of fluctuation.

• The system was very different from its original design.

1. Instead of a system of equal currencies, it evolved into a modified gold exchange standard:

a gold dollar standard, prone to the same problems as the interwar GS.

2. Instead of a system of adjustable pegs, it evolved into a system of de facto fixed ERs as the

experience of the pound sterling and other currencies in 1949 quickly convinced monetary

authorities that orderly parity realignments were impossible.

75

The Three Problems

• As the system evolved into a variant of the interwar GS, the same crucial problems emerged:

1. The adjustment problem:

– Deficit countries (concerned with unemployment) were unwilling to contract (as required

by BOP adjustment).

– Key surplus countries (Germany) were unwilling to expand and let prices rise.

– ER adjustment was only the last resort.

2. The liquidity problem:

– Provision of liquidity helps delay using other policy tools for adjustment in the short

term.

– But it can make things worse in the long term.

– The perceived problem was that there was not enough liquidity in the system to finance

world growth: not enough gold, not enough IMF resources, and supply of dollars tied to

U.S. BOP deficit.

3. The confidence problem:

– As dollar holdings outside the U.S. increased, confidence in the ability to convert them

into gold at the fixed parity declined.

– The threat of a convertibility crisis loomed progressively larger.

76

Intertwined Problems

• The three problems were connected:

• More reserves (liquidity) in the system mean less burden of adjustment on deficit countries,

but more burden on surplus countries.

– No matter the initial distribution, reserves will be eventually be redistributed to surplus

countries as they are used to settle deficits.

• Conversely, deficit countries disproportionately bear the burden of deficient liquidity.

• On the other hand, the more adequate the adjustment mechanism, the less need for liquidity

there is.

77

Intertwined Problems, Continued

• Finally, as liquidity provided by the center country increases, confidence declines.

• For the U.S., adjustment and confidence were tightly intertwined, because the main concern

over the U.S. BOP deficit was the threat to U.S. gold reserves.

• But as the U.S. deficit provided liquidity to the system, adjustment by the U.S. meant liquidity

shortage for the system.

• Note that these problems would not arise under a pure gold coin standard, a pure dollar

standard (with no connection to gold), or a floating ER regime.

78

The Adjustment Problem

• The U.S. BOP deficit (caused mainly by long-term capital outflows) was perceived as a

problem by U.S. policymakers for its effects on confidence in dollar convertibility and the

dollar’s role in providing liquidity.

• Europeans viewed the U.S. deficit as problem for different reasons:

– As reserve currency country, the U.S. did not have to adjust policies to target the BOP.

– As long as everyone was willing to continue soaking dollars, BOP deficits did not force

the Fed to contract monetary policy.

– The Fed routinely sterilized dollar outflows preventing them from translating into U.S.

monetary contraction.

79

The Adjustment Problem, Continued

• – This practice was resented: Germans viewed the U.S. as exporting inflation; the French

resented U.S. financial dominance and the seigniorage the U.S. earned on its liabilities.

– Acting on this perception, the French started converting dollars into gold in 1965.

– The French proposed solution to the problem was to double the dollar price of gold

(devaluing the dollar relative to gold and matching the decline in the real price of gold

since 1934).

– Capital gains on revaluation of world’s gold reserves would have been enough to retire

outstanding dollar (and sterling) balances.

– Once the U.S. returned to BOP equilibrium, the system could resume working as a

classical GS.

80

The Adjustment Problem, Continued

• Some economists argued that U.S. deficits were not an issue:

– According to this view, deficits were demand-determined because other countries

voluntarily held dollars for the value of dollar service flow.

• U.S. authorities actually tried four types of measures to address the deficit:

– Capital controls (for instance, Interest Equalization Tax in 1963) to discourage capital

exports.

– Trade balance measures to discourage official spending abroad, encourage exports, and

discourage imports.

– Monetary-fiscal mix during the Kennedy and Johnson administrations: Expansionary

fiscal policy to address to 1960-1961 recession and Operation Twist (raising short-term

rates to “twist” the yield curve (encouraging capital inflow) while keeping low long-term

rates to support growth.

– Gold conversion policies to prevent/discourage foreign authorities from converting dollars

into gold.

81

Possible Solutions to The Adjustment Problem

• U.S. authorities, the IMF, and the OECD proposed increasing world liquidity by providing

an alternative reserve asset, ameliorating the need for the U.S. to run deficits to provide

liquidity for the system.

– Negotiations in 1964-1967 resulted in the creation of the Special Drawing Rights.

• Unilateral devaluation of the dollar relative to gold—the solution preferred by France—was

rejected by Congress for the likelihood of it being quickly followed by other countries and the

consequences on U.S. credibility and confidence that this would not happen again.

• Multilateral parity adjustment was rejected by France as mostly a temporary solution.

• Floating ERs were receiving increasing support in academia, but were still strongly opposed

by the IMF.

82

The Limited Success of U.S. Policies

• U.S. policies had little success, as they could not eliminate the inherent problem of the gold

dollar standard:

– Gold production was just not sufficient to finance world growth.

– Eventually, the world economy would be bound for deflation if there was no alternative

source of liquidity (the dollar).

– Introducing an alternative reserve asset would not fix the problem as long as this asset

was ultimately convertible into gold.

83

The Liquidity Problem

Triffin’s Dilemma

• Robert Triffin (Yale University) published an extremely influential book on Gold and the

Dollar Crisis in 1960, pointing out the insufficiency of international liquidity (reserves) under

the BW system.

• Gold production was not sufficient to generate enough monetary gold to finance the growth

of world GDP and trade (as a low relative price of gold had reduced production incentives

and stimulated private demand).

• A large gap opened between GDP and trade growth and the growth of gold reserves in

1958.

• The gold shortfall would have to be made up by dollars from the U.S. BOP deficits.

• But, running continuous deficits, U.S. monetary gold reserves would decline both absolutely

and relative to dollar liabilities until an eventual convertibility crisis.

• Triffin’s view was that U.S. authorities would eventually close the deficit before the onset of

the crisis, but this would cause massive dollar shortage and international deflation.

84

Solutions

• Triffin’s proposed solution was to go back to Keynes’ ICU, converting all existing reserves

into an international currency and having the IMF serve as the world’s central bank to

provide generous liquidity.

• In the years after Triffin’s book, three types of solutions were proposed and implemented:

1. Expanding the IMF’s resources.

– Member quotas were increased in 1960 and 1966.

– The General Agreement to Borrow (GAB) in 1961 provided the IMF a further $6 billion

credit line.

2. Creating new resources outside the IMF.

– Swap arrangements and other agreements among central banks provided liquidity, but

did not solve the root of the issue.

3. Creating a new reserve asset: the Special Drawing Rights (SDR), on which you find some

details in the Appendix.

• In the end, none of the solutions prevented the eventual collapse of the system.

85

The Confidence Problem

• Ultimately, the key problem of the convertible BW period was a confidence crisis for the

dollar.

• As Triffin noted, the system was dynamically unstable as long as the growth of monetary

gold was insufficient to finance world growth and prevent the U.S. monetary gold stock from

declining relative to U.S. dollar liabilities.

• Pressure on U.S. gold reserves would mount and, eventually, a confidence crisis would lead

to collapse of the system (much like in 1931) absent drastic policy changes in the U.S.

• An international lender of last resort (as advocated by Charles Kindleberger in 1973) could

delay the crisis, but it would eventually became ineffective so long as a gold shortage

persisted.

86

On the Way to Crisis

• In the 1960s, U.S. monetary policy became increasingly expansionary to finance the

Vietnam War, and U.S. competitiveness deteriorated.

• In 1968, the U.S. removed the 25% required gold backing against Fed notes.

• Gold was thus demonetized and the links between gold production and other market

sources of gold and reserves were removed.

• The U.S. exercised extensive moral suasion to discourage dollar reserve conversion into

gold.

• These steps pushed the system in the direction of a de facto dollar standard, but the threat

of dollar convertibility crisis still persisted as the dollar was still formally tied to gold.

87

On the Way to Crisis, Continued

• Several solutions to the gold dollar standard confidence problem were proposed:

1. Introduction of a new reserve asset, substituting SDRs (or something similar) for dollars

and gold.

– But this would not solve the problem if the asset was convertible into gold and the gold

market situation remained unchanged.

· This is why the use of SDRs was restricted to financing BOP deficits.

88

On the Way to Crisis, Continued

1. Doubling the official price of gold would alleviate adjustment and liquidity problems (and

would encourage gold production and reduce private demand).

– But what would prevent such changes in the price of gold from happening again in the

future?

– Market participants would anticipate future changes and reduce their dollar holdings

permanently (as it happened to the sterling after 1949 and the additional devaluation of

1967).

– Besides, the problem would only be postponed: Eventually, gold scarcity would surface

again.

2. A variant of solution 2 was a unilateral devaluation of the dollar, which was rejected by

Congress for the reasons above.

3. Officially move to a pure dollar standard.

89

The n-1 Country Problem

• An official dollar standard would solve the so-called n− 1 BOPs problem studied by Robert

Mundell in his 1968 book on International Economics.

• In a world of n countries and n currencies, there are only n − 1 independent bilateral ERs

and BOPs.

• As the nth currency, the dollar allowed the rest of the world to peg their ERs independently

and target their BOPs.

• According to this argument, the U.S. had to follow a passive BOP policy, i.e., a policy of

benign neglect.

• But the major constraint on the U.S. was that its monetary policy would have to stabilize the

price of traded goods in order to deliver price stability to the entire system.

• The world price level would then be anchored by commodity arbitrage and monetary policy

adjustment to follow the U.S. lead.

• And growth in the rest of the world would be financed by dollars supplied by the U.S. deficit.

90

The n-1 Country Problem, Continued

• Gold convertibility, however, posed a problem:

– As long as the U.S. was committed to gold convertibility, its ability elastically to supply the

dollars demanded by the rest of the world would be curtailed by the threat of a confidence

crisis.

• The solution would then be to demonetize gold.

• But there were two problems with this solution:

1. The Europeans were unwilling to go along with dollar hegemony.

2. Without gold convertibility, there was no commitment mechanism to constrain the U.S. to

follow a stable monetary policy.

• As it turned out, the dollar standard that emerged de facto in 1968 collapsed precisely for

these reasons.

91

The Collapse of the Bretton Woods System

• The collapse of the system was triggered by an acceleration in world inflation, in large part

the consequence of an earlier acceleration of inflation in the U.S.

• This increase in inflation in the U.S. and the rest of the world was closely related to an

increase in money growth and, crucially, money growth relative to real output growth.

• In the early 1960s, expansionary monetary policy reflected a growing preference of

policymakers for full employment over price stability.

• In the 1960s, money supply grew to help finance budget deficits associated with both the

Vietnam war and social programs, with changes in money base closely correlated with the

government’s budget deficit.

92

The Collapse of the Bretton Woods System, Continued

• How did U.S. money growth and inflation spread to the rest of the world?

• U.S. money growth was the primary determinant of world money growth because of the

asymmetrical relationship between the U.S. and the rest of the world.

• The international role of the dollar and the position of the U.S. as the center of the system,

implied that the other countries had to “import” U.S. monetary policy.

• The U.S. could sterilize reserve flows, but the rest of the world could not.

• As dollar reserves in Germany, Japan, and other countries grew in the late 1960s and at the

beginning of the 1970s, sterilization became increasingly difficult for them.

• To keep the ER fixed, these countries had to let domestic money supplies grow and inflation

rise.

• The only alternative to importing U.S. inflation was to float, the route that all countries

eventually took in 1973.

93


• Moreover, Bela Balassa (Johns Hopkins University) pointed out another source of pressure

on the system in an immensely influential Journal of Political Economy article in 1964.

• As traded-sector productivity grew faster (relative to non-traded-sector productivity) in

Germany, Japan, and other surplus countries, this caused the relative prices of non-traded

goods to rise.

• In turn, this put upward pressure on these countries CPIs.

• Expansionary U.S. monetary and fiscal policy in the late 1960s exacerbated the misalign-

ment by further causing the dollar to be overvalued (and currencies such as the D-mark

undervalued) relative to the level that would have ensured BOP equilibrium.

• A revaluation by the surplus countries would have helped, but it would have only delayed

the collapse of Bretton Woods absent drastic changes in U.S. monetary and fiscal policies.

• In addition, Germany and other surplus countries resisted revaluation on the grounds that it

would harm the competitive position of their export industries.

94


• On May 5, 1971, Germany allowed the D-mark to float.

• Similar action by Austria, Belgium, the Netherlands, and Switzerland followed.

• In early August, France and Britain announced the intent to convert their dollar holdings into

gold.

• On August 15, 1971 President Nixon closed the gold window.

• The system of “adjustable” pegs was abandoned for good 19 months later.

95

Importing the nth Country’s Monetary Policy under Fixed Exchange Rates

• We can easily understand why fixed ERs imply importing the monetary policy set by the

center country (the nth country) by referring to the equation that determines the exchange

rate in Paul Krugman’s model of target zone dynamics that we studied:

ε(t) = m (t)−m∗ (t) + v(t)− v∗(t) + γE (dε(t)/dt) ,

where ε(t) is the exchange rate (units of Home currency per unit of Foreign), m (t) is Home

money supply, m∗ (t) is Foreign money supply, v(t) and v∗(t) money velocity shocks in the

two countries, E(dε(t)/dt) is the expectation of the change in the exchange rate between

the current instant and the next, and γ > 0.

• To simplify, assume v(t) = v∗(t) = 0.

• Hence:

ε(t) = m (t)−m∗ (t) + γE (dε(t)/dt) .

96

Importing the nth Country’s Monetary Policy under Fixed Exchange Rates

• Suppose that Home is the country that pegs the exchange rate and Foreign is the center

country.

• All variables measure log-deviations from trend levels.

• If the exchange rate is (credibly) fixed, it follows that ε(t) = E (dε(t)/dt) = 0.

• This implies immediately that it has to be m (t) = m∗ (t):

– Home monetary policy must shadow Foreign.

97

Importing the nth Country’s Monetary Policy under Fixed Exchange Rates,

Continued

• Another way to reach the same conclusion allows us to relate this point to the role of capital

mobility and uses the concept of uncovered interest rate parity (UIP)—which, as we shall

see, is actually part of the relations that lead to obtaining the Krugman ER equation.

• Suppose that agents can freely invest in Home or Foreign currency-denominated bonds.

– We know that there were barriers to capital mobility during the BW years, but they had

become less and less successful over the years. Let us simplify and assume that there

are no barriers.

• Home bonds issued at time t pay the nominal interest rate ıt at time t + 1, Foreign bonds

issued at time t pay the nominal interest rate ı∗t at time t + 1.

• Both these interest rates are measured as log-deviations (or percentage deviations) of the

underlying gross interest rates (respectively, 1 + it and 1 + i∗t ) from their long-run levels.

98


Continued

• Absence of unexploited arbitrage opportunities between the two bonds required the following

condition to hold:

ıt − ı∗t = Et (εt+1)− εt,where Et is the expectation operator conditional on time-t information and the exchange

rate εt is measured as percentage (or log) deviation from its long-run level.

• This is the UIP condition.

• It states that for agents to be indifferent between the two bonds, any interest rate differential

must be compensated by the expectation of exchange rate depreciation.

• The Appendix shows you how to obtain this condition from a rigorous portfolio optimization

problem.

99


Continued

• Suppose again that it is the Home country that pegs its currency to Foreign.

• If the exchange rate is credibly fixed, it follows that εt = Et (εt+1) = 0.

• Therefore, it has to be that the Home interest rate (determined by Home monetary policy) is

equal to the Foreign interest rate (determined by Foreign monetary policy): ıt = ı∗t .

• Free capital mobility and a fixed exchange rate imply giving up monetary independence.

100


Continued

• This is a manifestation of the so-called trilemma of international finance: It is impossible

these three things at the same time: free capital mobility, a fixed exchange rate, monetary

policy independence.

• Authorities must necessarily choose two out of three of these things.

• The term trilemma was coined by Maurice Obstfeld (U.C. Berkeley), Jay Shambaugh

(George Washington University), and Alan Taylor (U.C. Davis) in the early 2000s, but

Tommaso Padoa-Schioppa (one of the “founding fathers” of the euro and a member of the

European Central Bank’s first Executive Board) had already referred to the same concept

as the “impossible trinity” in the 1980s.

• In an influential 2013 paper, Helene Rey (London Business School) showed that flexible ERs

are not sufficient to grant monetary policy independence, as many central banks’ interest

rates end up tracking the Federal Reserve’s even if the exchange rates are in principle

flexible.

• There is much evidence of deviations from UIP and explaining them has generated its

own research literature, but we are going to use UIP at various points in the course as a

simplified tool that helps us understand important forces, in this case, the fact that a fixed

exchange rate (combined with capital mobility) implies importing the monetary policy of the

center country.101


Continued

• Now ask yourselves: What interest rate policy should the Home central bank commit to

and announce if it wanted to deliver a fixed exchange rate, i.e., if it wanted to deliver

εt = Et (εt+1) = 0?

• You may be tempted to answer ıt = ı∗t .

• This is a mistake that many, including many international macroeconomists, make.

• But if you impose ıt = ı∗t .in the UIP condition, all you get is that the expected change in the

ER is zero: Et (εt+1)− εt = 0.

• Any unexpected change can happen: Put differently, the ER remains completely indetermi-

nate.

• Or, another way to say this is that εt = Et (εt+1) = 0 (a fixed exchange rate) implies ıt = ı∗t ,

but ıt = ı∗t does not imply εt = Et (εt+1) = 0 (a fixed exchange rate).

102


Continued

• In a 2007 article in the Journal of Economic Dynamics and Control, Gianluca Benigno (NY

Fed), Pierpaolo Benigno (LUISS University, Rome), and I showed that committing to and

announcing the policy ıt = ı∗t + τ εt with τ > 0, i.e., committing to shadowing the center

country’s policy and to increasing the interest rate whenever the currency depreciates,

or lowering it whenever the currency appreciates—combined with a commitment to never

letting the currency become worthless—would deliver a fixed exchange rate.

• It would be εt = Et (εt+1) = 0 and, in equilibrium, we would observe ıt = ı∗t .

• The proof of this result is in the Appendix of these slides.

103

Importing Inflation

• The last few slides showed you that pegging the ER implies importing the monetary policy

of the country to whose currency one is pegging.

• Now suppose we model German price level dynamics and the effects of pegging the D-mark

to the dollar in an environment of expansionary Federal Reserve policy.

• Suppose the German consumption basket consists of domestic and imported (U.S.) tradable

goods and domestic non-tradables:

C =CηTC

1−ηNT

ηη (1− η)1−η, (8)

where CT is consumption of the tradable bundle, CNT is consumption of non-tradables, and

0 ≤ η ≤ 1.

• Suppose the tradable bundle combines German goods and U.S. goods as follows:

CT =CT (g)aCT (us)1−a

aa (1− a)1−a, (9)

where CT (g) denotes German consumption of German tradables, CT (us) is German imports

of U.S. goods, and 0 ≤ a ≤ 1.

104

Importing Inflation, Continued

• We know from our discussion of Friedman’s case for flexible exchange rates that equation

(8) implies that the German CPI (P ) has the form:

P = P ηTP

1−ηNT ,

where PT is the price of the bundle of traded goods and PNT is the price of non-tradables.

• Equation (9) implies that the price of the traded bundle is such that:

PT = PT (g)aPT (us)1−a,

where PT (g) is the price of domestic tradables and PT (us) is the price of U.S. exports in

Germany (therefore, it is a D-mark price).

• Assume that there are no impediments to trade between the U.S. and Germany and that

PCP results in the LOP:

PT (us) = εP $T (us),

where ε is the exchange rate (D-marks per dollar) and P $T (us) is the price of U.S. goods in

the U.S.

• We know there are reasons why the LOP may not hold, but let us assume it to simplify the

argument.

105

Importing Inflation, Continued

• Now, if the exchange rate is fixed, U.S. monetary expansion that puts upward pressure on

P $T (us) automatically translates into upward pressure on PT (us), which causes PT to rise.

• Moreover, the fact that Germany shadows U.S. monetary policy when the exchange rate is

fixed implies that Germany’s own monetary expansion pushes PT (g) up, further increasing

PT .

• Finally, faster productivity growth (i.e., more rapidly declining unit production costs) in

tradables than non-tradables as documented by Balassa implies that PNT is rising relative

to PT .

• All these forces combine to deliver faster CPI inflation in Germany, something that German

policymakers were hardly willing to tolerate.

106

Why Did the Bretton Woods System Collapse?

• In the end, BW collapsed for three basic reasons:

1. Two major flaws undermined the system:

a. One was the gold exchange standard, which placed the U.S. under threat of a

convertibility crisis.

· In reaction, the U.S. pursued policies that ended up making adjustment more difficult.

b. The other flaw was the adjustable peg system.

· Because the costs of discrete parity changes were deemed so high in the presence

of growing capital mobility, the system evolved into a reluctant fixed ER system

without any effective adjustment mechanism.

107

Why Did the Bretton Woods System Collapse? Continued

1. U.S. monetary policy was inappropriate for a key currency.

– By inflating, the U.S. followed an inappropriate policy for a center currency country after

1965.

– Although the acceleration of inflation was low by the standards of the 1970s, when

superimposed on the cumulation of low inflation since World War II, it was sufficient to

cause high market demand for gold (because its relative price had fallen) threatening

dollar convertibility.

– Once the regime had evolved into a de facto dollar standard, the obligation of the U.S.

was to maintain price stability.

– Instead, it conducted an inflationary policy that ultimately destroyed the system.

108

Why Did the Bretton Woods System Collapse? Continued

1. Surplus countries were increasingly unwilling to adjust.

– The major industrial countries were unwilling to absorb ever increasing dollar balances

and revalue their currencies.

– In turn, this reflected basic differences in the underlying inflation rates that countries

were willing to accept.

– The growing gap between the sovereign interests of the U.S. and other major industrial

countries in part reflected the relative decline in U.S. power as these countries

completed their post-World War II recovery.

– At the same time as U.S. power declined relative to continental Europe and Japan,

coordination groups such as the G10 lost effectiveness, and no other focal points of

power emerged.

– Thus, the stage was set for a decentralized global monetary system in which the major

currencies floated, with exchange rates determined by market forces.

109

Appendix A. Finding the CPI Expression

• We pointed out in the main text that if the consumption basket (C) has the form:

C =c(h)ac(f )1−a

aa (1− a)1−a,

then the CPI (P ) of the economy is:

P = p(h)ap(f )1−a,

where p(h) is the price of a unit of c(h), p(f ) is the price of a unit of c(f ), and 0 ≤ a ≤ 1.

• How do we obtain this expression for P?

• P is defined as the minimum amount of spending that is needed to purchase one unit of C.

110

Appendix A. Finding the CPI Expression, Continued

• In other words,

P = minc(h),c(f)

p(h)c(h) + p(f )c(f ) subject to...C = 1.

• Given the expression of C, this problem can be rewritten as:

P = minc(h),c(f)

p(h)c(h) + p(f )c(f ) subject to...c(h)ac(f )1−a

aa (1− a)1−a= 1.

• The Lagrangian for this minimization problem is:

L = p(h)c(h) + p(f )c(f ) + µ

[1− c(h)ac(f )1−a

aa (1− a)1−a

].

111


• Taking the derivatives of L with respect to c(h) and c(f ) and setting them equal to zero

yields the first-order conditions (FOCs) for the minimization problem:

∂L

∂c(h)= p(h)− µac(h)a−1c(f )1−a

aa (1− a)1−a= p(h)− µ aC

c(h)= p(h)− µ a

c(h)= 0,

∂L

∂c(f )= p(f )− µ(1− a)c(h)ac(f )1−a−1

aa (1− a)1−a= p(f )− µ(1− a)C

c(f )= p(f )− µ1− a

c(f )= 0,

where we used the expression of C and then the fact that C has to be equal to 1 in each of

these FOCs.

• These FOCs imply:

c(h) = µa

p(h),

and:

c(f ) = µ1− ap(f )

,

respectively.

112


• Now substitute these expressions for c(h) and c(f ) in the constraint

c(h)ac(f )1−a

aa (1− a)1−a= 1.


µa aa

p(h)aµ1−a (1−a)1−a

p(f)1−a

aa (1− a)1−a= 1,

which implies:

µ = p(h)ap(f )1−a.

113


• Next, recall that P is defined as the minimum level of spending such that C = 1.

• This implies that the expression for P follows from substituting the expressions for c(h) and

c(f ) implied by the FOCs in p(h)c(h) + p(f )c(f ), or:

p(h)µa

p(h)+ p(f )µ

(1− a)

p(f ).

• Immediate algebra shows that this expression reduces to µ.

• But we proved above that µ = p(h)ap(f )1−a.

• Hence, P = p(h)ap(f )1−a.

• Proceeding in the same way for the Foreign economy shows that P ∗ = p∗(h)ap∗(f )1−a.

114

Appendix B. Obtaining the Demands of Individual Goods

• The main text tells you that, given

C =c(h)ac(f )1−a

aa (1− a)1−a

and the associated CPI expression, the demands of good h and good f are determined by:

c(h) = a

(p(h)

P

)−1C and c(f ) = (1− a)

(p(f )

P

)−1C.

• How do we obtain these expressions?

• They are obtained by choosing c(h) and c(f ) to maximize C subject to a spending constraint.

115

Appendix B. Obtaining the Demands of Individual Goods, Continued

• Put differently, suppose that spending is constrained by the exogenous amount S.

• Then, the demand expressions follow from solving the problem:

maxc(h),c(f)

C subject to p(h)c(h) + p(f )c(f ) = S,

or:

maxc(h),c(f)

c(h)ac(f )1−a

aa (1− a)1−asubject to p(h)c(h) + p(f )c(f ) = S.

• The Lagrangian for this problem is:

L =c(h)ac(f )1−a

aa (1− a)1−a+ λ (S − p(h)c(h)− p(f )c(f )) .

116


• Taking the derivatives of the Lagrangian with respect to c(h) and c(f ) and setting them equal

to 0 yields the FOCs:

∂L

∂c(h)= a

c(h)a−1c(f )1−a

aa (1− a)1−a− λp(h) = a

C

c(h)− λp(h) = 0,

∂L

∂c(f )= (1− a)

c(h)ac(f )1−a−1

aa (1− a)1−a− λp(f ) = (1− a)

C

c(f )− λp(f ) = 0.

• These FOCs yield:

c(h) =aC

λp(h)and c(f ) =

(1− a)C

λp(f ),

respectively.

• Now substitute these expressions into the expression for C

C =

aaCa

λap(h)a(1−a)1−aC1−aλ1−ap(f)1−a

aa (1− a)1−a.

117


• The last equation implies:

λ =1

p(h)ap(f )1−a=

1

P,

where we used the expression of the CPI that we obtained in the previous appendix.

• Substituting λ = 1/P into c(h) = aC/ (λp(h)) and c(f ) = (1− a)C/ (λp(f )) yields:

c(h) = a

(p(h)

P

)−1C and c(f ) = (1− a)

(p(f )

P

)−1C.

118

Appendix C. Details on the History of Bretton Woods, 1946-1958

• The international monetary system that began after WWII was far different from the system

that the architects of BW envisioned.

• The transition period from war to peace was much longer and more painful than anticipated.

• Full convertibility of the major industrial countries was not achieved until the end of 1958,

although the system had started functioning normally by 1955.

• Two interrelated problems dominated the first post-WWII decade: bilateralism and the dollar

shortage.

119

Appendix C. Details on the History of Bretton Woods, 1946-1958, Continued

Bilateralism

• For virtually every country except the U.S., the legacy of WWII was one of pervasive

exchange controls (restrictions on the ability to convert currencies) and controls on trade.

• Except the dollar, no major currencies were convertible.

• Under Article XIV of the IMF Articles of Agreement, countries could continue to use

exchange controls for an indefinite transition period after the establishment of the IMF on

March 1, 1947.

– Under Article XIV, three years after this date, the IMF would begin reporting on the

countries with existing controls.

– Two years later, it would begin consulting with individual members, advising them on

policies to restore BOP equilibrium and convertibility.

– Countries that did not make satisfactory progress would be censored and ultimately

asked to leave the Fund.

– In actual fact, the Fund always accepted the member’s reason for remaining under Article

XIV.

• In conjunction with exchange controls, every country negotiated a series of bilateral

payments agreements with each of its trading partners.120


• The rationale given for continued use of controls and bilateralism was a shortage of

international reserves.

• The economies of Europe and Asia were devastated after the war.

• New and improved capital was required to produce the exports needed to generate foreign

exchange revenue.

• There was an acute shortage of key imports, both foodstuffs to maintain living standards

and raw materials and capital equipment.

• Controls were used to allocate the scarce reserves.

121


• The bilateral arrangements in each country typically consisted of licenses and quotas for

imports and exports and the allocation of foreign exchange through the central bank, with

commercial banks acting as agents.

• Each central bank typically negotiated an agreement with its trading partners providing an

overdraft facility in its currency (called the “swing”) up to a specified limit, with settlement in

foreign exchange beyond that.

• The UK developed a particularly complicated set of arrangements.

• The sterling was convertible for all transactions within the sterling area.

• Certain privileged countries were allowed transferable account status, whereby they could

settle foreign balances in sterling earned from their exports, and, in 1946-47, these balances

could be used to settle dollar payments.

• Another group of countries had bilateral account status.

122


The Dollar Shortage

• By the end of WWII, the U.S. held two-thirds of the world’s monetary gold stock.

• The massive inflow of gold into the U.S. in the 1930s was the consequence of both the dollar

devaluation in 1934 (when the Roosevelt administration raised the price of gold from $20.67

to $35.00 per ounce) and capital flight from Europe, pushing the U.S. BOP into persistent

surplus.

• During WWII, gold flows into the U.S. continued to finance the wartime expenditures by the

Allies.

• At the end of WWII, Europe’s (and Japan’s) gold and dollar reserves were depleted.

• Europe ran a massive current account deficit.

• Combined with other countries’ deficits, this matched the U.S. current account surplus.

123


• The dollar shortage facing deficit countries was likely aggravated by overvalued official

parities set by the major European industrial countries at the end of 1946.

• The IMF pressured its members to declare par values as soon as possible.

• It was argued that the BOP deficits facing most countries reflected the structural incapacity

of their export industry rather than lack of competitiveness.

• The argument was that, if the chosen parity was inappropriate, it could be corrected later.

• The crucial test was the ability to export at all.

• Most countries adopted their pre-WWII dollar parities on the assumption that wartime and

postwar inflation did not seriously disrupt their competitive positions relative to the U.S.

124


• The immediate postwar experience of massive deficits and depleted international reserves

in Europe and the opposite situation in the U.S. led many scholars and policymakers to

believe that the dollar shortage was permanent.

• The key explanation for a permanent shortage given at the time was that the rate of

productivity growth in the rest of the world would never catch up to that of the U.S.

• Alternative explanations included inadequate raw materials, lower savings rates, political

instability, and lack of entrepreneurial drive.

• Advocates of these theories recommended policies including discrimination against U.S.

exports, massive U.S. aid, the encouragement of private capital flows to Europe, and

devaluation.

125

Appendix D. More on The EPU and the Return to Convertibility

• Settlement of transactions within the EPU was made in dollars, gold, or credit.

• The division between credit and dollars/gold depended on the quota of the Union’s capital

allocated to each member, which in turn depended on its volume of trade.

• The Union was started with an initial working capital fund of $350 million provided by the

U.S.

• The EPU became the center of a worldwide multilateral settlement area, including the

countries of the sterling and franc zones.

126

Appendix D. More on The EPU and the Return to Convertibility, Continued

• In 1954, the UK extended transferable account status to all countries with which it had

bilateral agreements.

• Thus, by 1955, the world was essentially divided into two different convertible areas

separated by exchange controls:

– An area based on the EPU and the sterling and an area based on the dollar.

• The final steps in closing the gap were achieved in February 1955, when the Bank of

England extended its exchange market operations to pegging the ER on transferable

account sterling, and eight countries declared their currencies convertible for current

account transactions on December 27, 1958.

127

Appendix E. The Decline of the Pound Sterling

• The sterling shared the role of key currency with the dollar in the interwar period.

• But its importance declined throughout the BW years.

• At the end of WWII, Britain ran a massive BOP deficit, especially in gold and dollars, as did

the other European countries.

• In addition, Britain had an outstanding sterling debt of £3.7 billion amassed during the war

by borrowing largely from its empire.

• Much of this balances were “blocked,” that is, made inconvertible into dollars.

128

Appendix E. The Decline of the Pound Sterling, Continued

• In December 1945, the U.S. and Britain negotiated the Anglo-American Loan, ratified on

July 15, 1946.

• In exchange for $3.75 billion from the U.S. and a further $1.25 billion from Canada, the

British ratified the BW Articles and promised to restore current account convertibility in

dollars, except for existing sterling balances, within one year.

• In 1946-47, transferable account sterling status was extended to all countries with bilateral

agreements, and convertibility for current account transactions was restored on July 15,

1947.

• The ensuing run on the sterling (the consequence of deficit and convertibility at an

overvalued parity) depleted the UK’s reserves by $1 billion within a month, and convertibility

was suspended on August 20, 1947.

• Transferable account sterling again became subject to exchange controls.

• As a consequence of this unsuccessful experience, the return to convertibility by Western

European countries was likely delayed longer than it otherwise would have been.

• The sterling’s role as a reserve currency was further weakened and that of the dollar (by

default) strengthened.

129

Appendix E. The Decline of the Pound Sterling, Continued

• A second event that further weakened the sterling’s credibility as a reserve currency, yet

helped stabilize the international system by reducing both European deficits and the U.S.

surplus (and also strengthen the UK economy), was the sterling devaluation in 1949.

• Speculators evaded exchange controls by means of leads and lags and took a strong

position against the sterling in the summer of 1949.

– Foreign buyers of goods from the sterling area delayed their purchases and payments,

while importers in the sterling area speeded up their payments.

• Two weeks before the event, Sir Stafford Cripps, the Chancellor of the Exchequer, denied

that devaluation was imminent.

• On September 18, 1949, twenty-four hours after the IMF was informed, the pound sterling

was devalued by 30.5%.

• The devaluation improved the current account deficit and the overall BOP of the UK,

although it deteriorated again in the following two years of inflation generated by the funding

of the Korean War in the U.S.

• Importantly, the 1949 devaluation dealt another blow to the role of the pound sterling as

international reserve currency.

130

Appendix F. The Role of the IMF

• A number of developments and events in the pre-convertibility period had great significance

for the prestige and subsequent role of the IMF.

• The IMF, by intention, was not equipped to deal with the postwar reconstruction problem.

• Although limited drawings occurred before 1952, most of he structural BOP assistance in this

period was provided by the Marshall Plan and other U.S. aid, including the Anglo-American

Loan of 1945.

• The consequence was that new institutions such as th OEEC and existing institutions such

as the BIS (the agent for the EPU) emerged as competing sources of international monetary

authority.

• Had the Keynes ICU plan been adopted at BW, the difference between the proposed

resources of $26 billion and the original IMF endowment of $8.8 billion would have nearly

equalled the $13 billion given in Marshall Plan assistance.

131

Appendix F. The Role of the IMF, Continued

• Moreover, although multilateralism was a key precept of the Articles of Agreement, the IMF

did little to achieve that aim before 1952, when, under Article XIV, it began consultations

with individual members.

• The IMF did very little to speed up the process of achieving multilateralism because it felt

that it was not so empowered.

• As a consequence, another agency, the EPU, was set up to provide the clearinghouse that

Keynes envisioned in the ICU plan.

• Also, in part because of its opposition to floating rates, and in part because of its eagerness

to get the system going, the Fund had been criticized for seeking a declaration of ER par

values too soon.

• The resulting fixed parities then set in motion forces within each country to resist devaluation

until it was too late, and the changes that did finally occur in 1949 were larger than

necessary.

• The crisis associated with the 1949 sterling devaluation in turn created further resistance

by monetary authorities to changes in parity, which ultimately changed the nature of

the international monetary system from the adjustable peg intended by the Articles of

Agreement to a fixed rate regime, with devaluation viewed only as a last resort action.

132


• The IMF’s prestige was dealt severe blows by three events in the pre-convertibility period.

• The first was the French devaluation of January 1948, when France created a multiple ER

system in an attempt to economize on scarce hard currency.

• The arrangement consisted of a dual rate for hard currencies, with the official rate of 214.39

francs per dollar for basic imports and a floating rate for tourist and financial transactions.

• The effective rate for the dollar was this 260.26 francs, while for soft currencies it was the

official rate.

• Under Article IV.5, the IMF censured France for creating broken cross rates between the

dollar and the pound, thereby diverting exports to be re-exported to the U.S. via France:

– Since the effective franc rate for the dollar was weaker than the official rate, there was an

incentive for Americans to buy British goods from France rather than directly from Britain,

taking advantage of the dollar price reduction (relative to the price implied by the official

dollar per pound rate) implied by the depreciated effective franc per dollar rate.

133


• To understand this, consider the options for an American trying to buy a British good.

• The American could do this directly from Britain at the official dollar per pound ER.

– Note that the official dollar per pound ER is tied to the official dollar per franc and franc

per pound ERs by:

officialdollar

pound=

(official

dollar

franc

)•(

officialfranc

pound

).

• Or the American could buy the good from France after France had imported it from Britain.

– France would import the British good at the official franc per pound ER.

– The American could then buy the good from France at the effective dollar per franc ER.

– The transaction from the original pound pricing of the good to the dollar price paid by the

American would then be based on an effective dollar per pound ER determined by:

effectivedollar

pound=

(effective

dollar

franc

)•(

officialfranc

pound

).

– Since

effectivedollar

franc< official

dollar

franc,

the transaction going through France would be convenient for the American, because it

would happen at a more convenient ER than the official dollar per pound rate.

134


• Because of this broken cross rates issue, France was denied access to the IMF’s resources

until 1952.

• France ended the broken cross rates in October 1948 and adopted a unified rate in the

devaluation of 1949.

• But since France had access to Marshall Plan aid, the IMF’s actions actually had little effect.

135


• The second event that reduced the prestige of the IMF was the sterling devaluation of

September 1949.

• Although the IMF staff had earlier advised the British to devalue, the IMF was given only

twenty-four hours notice, and the size of the devaluation was larger than suggested.

• This was in marked contrast to Article IV.5, which required a member to consult with the IMF

when a devaluation greater than 10% was being considered, with the IMF to be given more

than seventy-two hours to concur or object.

• This event revealed the IMF’s inability to deter a major power from following its sovereign

interest.

136


• The third event was the decision by Canada in September 1950 to float its currency.

• Faced with a massive capital inflow from the U.S., the Bank of Canada decided to float the

Canadian dollar rather than risk an inappropriate devaluation from the $.909 parity.

– Canada had originally set its parity at $1 in 1946 but devalued with the UK to $.909 in

1949.

• The IMF was highly critical of the action.

• The Canadian monetary authorities assured the Fund that the float was only temporary and

that a new parity would be declared when a new equilibrium had been reached.

• The Canadian dollar floated until 1961.

• The fact that its movements were small and there was no evidence of destabilizing

speculation significantly weakened the case made against floating by the IMF in several

annual reports.

137


• Finally, the Fund’s resources were inadequate to solve the emerging liquidity problem of the

1960s.

• The difference between the growth of international reserves required to finance the growth

of real output and trade and avoid deflation and the growth in the world’s monetary gold

stock was met largely by an increase in official holdings of U.S. dollars (Figure 10) resulting

from growing U.S. BOP deficits.

• By the time full convertibility for current account transactions was achieved, the U.S. dollar

was serving the buffer function for which the Articles of Agreement intended the IMF’s

resources.

• Had the Keynes ICU plan been adopted, and had the U.S. undertaken postwar aid through

the ICU rather than through Marshall Plan aid, the U.S. would have accumulated sufficient

overdraft facilities to finance most of its deficits in the 1950s and 1960s.

• However, the extra liquidity would have likely fueled a higher rate of world inflation than

actually occurred.

138

Appendix G. Special Drawing Rights

• SDRs were created as an alternative reserve asset—special drawing accounts that

countries held with the IMF.

• SDRs were a fiat obligation, not backed by gold.

• Its acceptability stemmed from the obligation of all countries to accept SDRs—similar to the

legal tender nature of domestic fiat money.

• Limits were in place to protect countries from having to absorb a disproportionate amount of

this (inferior) asset.

• One SDR was defined as equivalent to one gold dollar.

• SDRs could be used only to finance BOP deficits and members had to hold an average

balance of 30% of their allocations over 5-year periods.

• The scheme was activated on January 1, 1970:

– The French did not want it to be in place before the elimination of U.S. BOP deficit; the

U.S. ran BOP surpluses in 1968 and 1969.

• The initial allocation of SDRs was quite large ($9.5 billions over 3 years) based on the

projection that the decline in international reserves since 1965 would persist.

139

Appendix G. Special Drawing Rights, Continued

Issues with SDRs

• Acceptability :

– It was argued that SDRs were doomed because less acceptable as reserve assets than

gold and dollars.

· SDRs use was limited to official BOP transactions and could not be used as private

international money.

· SDR bore low rate of interest.

· It was less acceptable than gold because it did not have gold’s intrinsic properties.

· Ultimately, its main function became that of unit of account.

• Reserve growth and reserve center deficits:

– SDR scheme was designed only to increase reserve growth, but did not include a

mechanism to restrain reserve growth through deficits of the reserve centers.

– In actuality, by the time SDRs were introduced, they ended up adding to the inflationary

pressures of the early 1970s.

140

Appendix G. Special Drawing Rights, Continued

• Seigniorage:

– By economizing on gold reserves, issuing SDRs created a social saving, distributed as

seigniorage.

– Seigniorage distribution was neutral to the extent that it reflected IMF member quotas.

• Confidence:

– By restricting use of SDRs to BOP transactions, architects avoided another source of

speculative asset switches.

– But any extension of SDRs to being a closer substitute for dollars would aggravate

confidence problems.

– Also, the architects underestimated the SDRs’ contribution to the inflationary pressures

started in the late 1960s, as the SDRs ultimately contributed to surplus liquidity.

141

Appendix H. The London Gold Pool

• The key problem of the convertible BW period was the constant risk of dollar confidence

crisis.

• As Triffin noted, as long as the U.S. monetary gold stock continue to decline relative to U.S.

dollar liabilities, pressure on U.S. gold reserves would mount.

• A first rush to gold happened in 1960 for fear of inflationary policies in the U.S. under a

Democratic administration.

• Policy response included the creation of the London Gold Pool—an agreement between

the Fed and seven European central banks to intervene cooperatively in the London gold

market to try and keep the market price of gold near the Bretton Woods parity of 35 dollars

per ounce.

142

Appendix H. The London Gold Pool, Continued

• In the 1960s, growing gold scarcity, U.S. inflation, and soaring private demand put continued

pressure on the system.

• In 1966 the London Gold Pool became a net seller of gold.

• As noted in the main text slides, U.S. monetary policy became increasingly expansionary to

finance the Vietnam War, with a negative effect on U.S. competitiveness.

• After a sterling devaluation in 1967 (terminating the role of the sterling as “first line of

defense” for the dollar), pressure mounted:

– The London Gold Pool lost $3 billion in gold between 12/67 and 03/68 (with U.S. share at

$2.2 billions).

143

Appendix H. The London Gold Pool, Continued

• Under increasing pressure, the Pool was disbanded on March 17, 1968, and a two-tier

arrangement was put in place.

• Monetary authorities of the Gold Pool agreed not to transact in gold on the market, but only

among themselves at the official $35 price.

144

Appendix I. Obtaining the Uncovered Interest Rate Parity Condition

• Consider the problem of a Home investor that has 100 units of Home currency at time t

and is debating whether to invest them in Home bonds (denominated in Home currency) or

Foreign bonds (denominated in Foreign currency).

• Investing in Home bonds will yield the nominal interest rate it at time t+ 1. This interest rate

is known with certainty at time t.

• Hence, if the agent invests her/his 100 units of Home currency in Home bonds, she/he will

receive 100 (1 + it) at time t + 1.

• In real terms, this will be worth 100 (1 + it) /Pt+1, where Pt+1 is the Home CPI at time t + 1.

• 100 (1 + it) /Pt+1 is the amount of consumption the investment in Home bonds allows the

investor to purchase at time t + 1, certainly more important to the investor’s decision than

just the nominal payoff 100 (1 + it).

145

Appendix I. Obtaining the Uncovered Interest Rate Parity Condition

• But the Home investor does not just care about the real value of the return to his investment.

• She/he cares about the increment that this will generate in her/his well-being (or welfare).

• Assuming that the Home investor evaluates her/his welfare in terms of a utility function that

depends on consumption (Ct) and leisure (1−Nt, where Nt is time spent working, and 1 is

the total endowment of time in each period), Home investor welfare at t + 1 is measured by

U (Ct+1, 1−Nt+1).

• This implies that the increment in welfare generated by an additional unit of consumption at

t + 1 is equal to UC (Ct+1, 1−Nt+1), where the subscript C denotes that we are taking the

derivative of U (Ct+1, 1−Nt+1) with respect to Ct+1, i.e., UC (Ct+1, 1−Nt+1) is the marginal

utility of consumption at t + 1.

146

Appendix I. Obtaining the Uncovered Interest Rate Parity Condition, Continued

• It follows that the increment of Home investor welfare generated at t + 1 by investing her/his

100 units of Home currency in Home bonds at time t is 100 (1 + it)UC (Ct+1, 1−Nt+1) /Pt+1.

• But at the time when she/he is taking her/his decision (time t), the Home investor does not

know Ct+1, Nt+1, and Pt+1.

• Hence, the investor will be forming an expectation of the payoff from investing in Home

bonds based on the information at time t:

100 (1 + it)Et

(UC (Ct+1, 1−Nt+1)

Pt+1

), (10)

where 100 and 1 + it are outside the expectation operator (Et (· · · )) because they are

information available at time t.

147


• Now consider the alternative of investing in Foreign bonds.

• Investing in Foreign bonds will yield the nominal interest rate i∗t at time t + 1. Also this

interest rate, like the Home interest rate it, is known with certainty at time t.

• However, investing in Foreign bonds requires first converting the 100 units of Home currency

into Foreign currency in order to be able to buy the Foreign bonds.

• If the exchange rate εt is in units of Home currency per unit of Foreign, it follows that 100

units of Home currency are worth 100/εt units of Foreign currency.

• Investing this amount into Foreign bonds at time t will thus deliver 100 (1 + i∗t ) /εt units of

Foreign currency at time t + 1.

• Or, converting this amount back into Home currency, investing in Foreign bonds will yield

100 (1 + i∗t ) εt+1/εt units of Home currency at time t + 1.

148


• In real terms, this will be worth 100 (1 + i∗t ) εt+1/ (εtPt) units of consumption.

• This will yield a welfare increment equal to [100 (1 + i∗t ) εt+1/ (εtPt)]UC (Ct+1, 1−Nt+1).

• Just as when contemplating the investment in Home bonds, at time t, the investor does not

know Ct+1, Nt+1, εt+1, and Pt+1.

• Therefore, she/he will form an expectation of the welfare payoff from investing in Foreign

bonds:

100

εt(1 + i∗t )Et

(εt+1UC (Ct+1, 1−Nt+1)

Pt+1

), (11)

where 100, εt, and 1 + i∗t are outside the expectation operator because they are information

available at time t.

• When is it that the investor is happy with her/his portfolio decision and feels no need to make

further adjustments?

149


• When the expected payoffs (10) and (11) are equal, i.e.,

100 (1 + it)Et

(UC (Ct+1, 1−Nt+1)

Pt+1

)=

100

εt(1 + i∗t )Et

(εt+1UC (Ct+1, 1−Nt+1)

Pt+1

),

or, simplifying the constant 100:

(1 + it)Et

(UC (Ct+1, 1−Nt+1)

Pt+1

)=

1

εt(1 + i∗t )Et

(εt+1UC (Ct+1, 1−Nt+1)

Pt+1

). (12)

• This is a no-arbitrage condition in the market for Home and Foreign bonds:

– When this condition holds, there are no unexploited profit opportunities from trading

between the two bonds.

150


• Now, condition (12) looks awfully complicated.

• If we make assumptions about the distribution of uncertainty to which investors are subject

to that are usually made in the international finance literature, and if we use a technique

called log-linearization, equation (12) becomes the much simpler:

ıt = ı∗t + Et (εt+1)− εt,

or:

ıt − ı∗t = Et (εt+1)− εt,the UIP condition that we introduced in the main text, which requires that the interest rate

differential between Home and Foreign bonds must be compensated by the expectation of

depreciation of the Home currency relative to the Foreign one.

• In order for the investor to be indifferent between Home and Foreign bonds that yield

different interest rates, the investor must be compensated by the expected capital gain from

exchange rate movement.

151


• For instance, if ıt > ı∗t , in order to be indifferent between Home and Foreign bonds,

the investor wants to be compensated by the expectation that the Foreign currency will

strengthen (i.e., the Home currency will depreciate), so that Et (εt+1) > εt, and the difference

between ıt and ı∗t is made up for by Et (εt+1)− εt.

• If ıt > ı∗t + Et (εt+1) − εt, investors will shift their demand toward Home bonds, bidding up

their price (and therefore causing ıt to fall), and will reduce their demand for Foreign bonds

(lowering their price and causing ı∗t to rise).

• The shift in demand across currencies will cause εt to become smaller (the Home currency

strengthens).

• And the process continues until ıt = ı∗t + Et (εt+1)− εt.

• In a frictionless world, this happens instantaneously, and we never see deviations from

ıt − ı∗t = Et (εt+1)− εt.

• As noted in the main text, there is plenty of evidence of deviations from UIP, but we still use

it as a convenient simplification for our analyses.

152

Appendix J. Interest Rate Policy for Fixed Exchange Rates

• The main text tells you that if UIP holds and the Home central bank follows the policy

ıt = ı∗t + τ εt with τ > 0, combined with a commitment of the government to never letting

the currency become worthless, the exchange rate will be fixed, i.e., it will be εt = 0 in all

periods.

• To show this, begins with substituting the policy ıt = ı∗t + τ εt for ıt in the UIP condition

ıt − ı∗t = Et (εt+1)− εt.

• This implies:

τ εt = Et (εt+1)− εt,or:

εt =1

1 + τEt (εt+1) . (13)

153

Appendix J. Interest Rate Policy for Fixed Exchange Rates, Continued

• Since UIP holds in each period and the Home central bank follows the same policy in each

period, it follows that an equation like (13) holds also at t + 1:

εt+1 =1

1 + τEt+1 (εt+2) .

• Substituting this for εt+1 into (13) yields:

εt =

(1

1 + τ

)2Et [Et+1 (εt+2)] .

• But the law of iterated expectations implies that Et [Et+1 (εt+2)] = Et (εt+2).

• Hence:

εt =

(1

1 + τ

)2Et (εt+2) . (14)

154


• Now, equation (13) must hold also at t + 2. Hence:

εt+2 =1

1 + τEt+2 (εt+3) ,

and substituting this into (14) yields:

εt =

(1

1 + τ

)3Et [Et+2 (εt+3)] ,

which becomes

εt =

(1

1 + τ

)3Et (εt+3) ,

by the law of iterated expectations.

155


• If we repeat this procedure until a future date T for equation (13), we get:

εt =

(1

1 + τ

)TEt (εt+T ) .

• And if we let T go to infinite, we have:

εt = limT→∞

(1

1 + τ

)TEt (εt+T ) .

• Now,

limT→∞

(1

1 + τ

)T= 0,

because τ > 0.

156


• This implies that the only way in which it can be εt = limT→∞ [1/(1 + τ )]T Et (εt+T ) 6= 0 is if

Et (εt+T ) becomes infinite.

• But this would mean that it takes an infinite number of units of the Home currency to buy

one unit of the Foreign currency:

– The Home currency has become worthless.

• We assumed that the government is committed to never letting the currency become

worthless, therefore ruling out this scenario.

• It follows that

limT→∞

(1

1 + τ

)TEt (εt+T ) = 0,

or εt = 0, i.e., the exchange rate is fixed.

157


• Note that the policy ıt = ı∗t + τ εt with τ > 0 is not the only one that successfully fixes the

exchange rate, but it is the simplest.

• For a proof of the result in this Appendix in a more general, non-linear environment, see

my 2007 article with Gianluca Benigno and Pierpaolo Benigno in the Journal of Economic

Dynamics and Control.

• In that article, we also relate the properties of the policy and the result to other literature on

policy through interest rate setting.

158

ECON 425Topics in Monetary Economics:

The International Monetary Systemfrom the Gold Standard to Globalization

Monetary Policy Interactions under Alternative Exchange Rate Regimes

Fabio GhironiUniversity of Washington

Exchange Rate Regimes since 1950

(from Ilzetzki Reinhart Rogoff QJE 2019)

Anchor Currencies since 1950

(from Ilzetzki Reinhart Rogoff QJE 2019)

The Model

• I will start by focusing on a model in which the world economy consists of two countries:core and periphery, denoted below by superscripts C and P , respectively.

• Each of these countries specializes in production of a single traded good, and the two goodsare imperfect substitutes:

– Consumers in the two countries want to consume both the domestic and the foreigngood.

• All variables in the model represent deviations of actual values from equilibrium values inthe absence of shocks and, except interest rates, all variables are in logs.

• I omit time subscripts to economize on notation: Variables without a time subscript denotecurrent-period variables. Variable with a +1 (−1) subscript denote variables next (last)period.

9

The Model, Continued

• Output in each country (yj, j = C,P throughout) depends on employment (nj) and aworldwide productivity shock (x):

yj = (1− α)nj − x, (1)

where α is the elasticity of output to employment, 0 < α < 1.

• The worldwide productivity shock x is identically and independently distributed with zeromean.

• The labor demand of firms is determined by the profit maximization condition:

wj − pj = −αnj − x, (2)

where wj is the nominal wage and is the price of country j’s good (both in units of itscurrency).– We can rearrange (2) as:

nj = − 1α

¡wj − pj + x

¢.

Firms demand less labor if the real wage (in terms of their product price) is higher (for given real wage) and more labor if labor is more productive (because marginal cost of production is lower).

10

Digression: Obtaining Labor Demand

• How do we obtain (2)?

• Write the production function in levels (rather than logs), denoted with upper-case letters:

Y =N1−α

X,

where I omit the country superscript for simplicity.

• Firms choose how much labor they employ to maximize profits, or revenues from outputsales minus labor costs,

PY −WN

subject to the production function Y = N1−α/X.

11

Digression: Obtaining Labor Demand, Continued

• Thus firms, choose N to maximize

PN1−α

X−WN.

• The first-order condition for optimal labor demand sets the first derivative of this expressionwith respect to N to zero:

(1− α)PN−α

X−W = 0,

or

(1− α)PN−α

X=W.

– The properties of the profit maximization problem we are solving ensure that thefirst-order condition is both necessary and sufficient for optimality.

• If you take logs of both sides of this equation and recognize that the deviation of log (1− α)

from its zero-shock equilibrium value is zero, you get (2).

12


• Consumer price indexes (denoted with qj) are weighted averages of the prices of the core’sand periphery’s goods.

• Consumers in the core allocate a fraction a of their spending to the domestic good and 1− a

to the good produced by the periphery.

• The parameter a, which takes values between 0 and 1, describes the economic size of thetwo countries.

– When a is small, the core is small, while the periphery is large.

– As a increases, the core becomes larger.

– When a = 1, the periphery is reduced to a small, open economy, whose policies, as weshall see, have no effect on the core.

13


• The consumer price index (CPI) in the core is therefore:

qC = apC + (1− a)¡pP − e

¢.

– This averages the price of the core’s good and the price of the periphery’s good (in unitsof the core’s currency) with weights equal to the expenditure shares.

– e is the exchange rate: units of the periphery’s currency per one unit of the core’s.

– Given the price of the periphery’s good PP in periphery’s currency, we must divide it byE (obtaining pP − e in logs) to translate it into core’s currency.

14


• Define the terms of trade between the two countries as:

z ≡ e + pC − pP .

• This tells us how many units of the periphery’s good it takes to buy one unit of the core’sgood.

• To see this, consider the level (rather than logs) version of the definition:

Z ≡ EPC

PP=

³Periphery’s currency

Core’s currency

´³Core’s currency

Core’s good

´Periphery’s currency

Periphery’s good

=Periphery’s good

Core’s good.

• If z increases (decreases), the core’s good becomes relatively more expensive (cheaper)in international trade, because it takes more (less) units of the periphery’s good to buy oneunit of it.

• Ghironi and Giavazzi refer to z as the real exchange rate (consistent with Matthew Canzoneriand Dale Henderson, 1991).

• But the real exchange rate (RER) is most often defined as the relative price of the twocountries’ consumption baskets rather than the relative price of the goods they produce andtrade (the terms of trade).

• I stick to this more frequent convention in the slides.15


• Given the definition of z, we can rewrite the core’s CPI as:

qC = pC − pC + apC + (1− a)¡pP − e

¢= pC − (1− a) z. (3)

• The core’s CPI increases if the price of the core’s good rises and if importing the periphery’sgood becomes more expensive (z decreases).

• The effect of terms of trade changes depends on the share of the periphery’s good in thecore’s consumption basket, 1− a.

• If a = 1, the consumption basket consists only of the core’s good, and changes in z have noeffect on qC.

16


• We assume that consumers in the periphery are characterized by the same consumptionpattern as those in the core, and they allocate the fraction a of their spending to the core’sgood and 1− a to the periphery’s good.

• Hence, the periphery’s CPI is:

qP = (1− a) pP + a¡e + pC

¢= pP + az. (4)

17


• Note that each country’s CPI is expressed in units of the country’s currency.

• Suppose we want to compare CPI’s evaluated in the same currency:

• Translate qC into periphery’s currency and consider the difference relative to qP (or, in levels,consider EQC/QP )

e + qC − qP .

– This tells us how many units of the periphery’s consumption basket it takes to buy oneunit of the core’s consumption basket:

EQC

QP=

³Periphery’s currency

Core’s currency

´³Core’s currency

Core’s consumption basket

´Periphery’s currency

Periphery’s consumption basket

=Periphery’s consumption basket

Core’s consumption basket.

– This is the standard definition of the real exchange rate.

18


• Note: Using (3), (4), and z ≡ e + pC − pP :

e + qC − qP = e + pC − (1− a) z − pP − az

= e + pC − (e + pC − pP )− pP

= 0,

or EQC/QP = 1.

• It takes exactly one unit of the periphery’s consumption basket to buy one unit of the core’sconsumption basket!

• This is purchasing power parity (PPP), which we already encountered in this course.

• It holds in our model because there is no friction in trade in goods (and thus arbitrageensures that the price of each good is the same in the two countries once we evaluate it inthe same currency) and consumers in the two countries have identical spending patterns(or preferences), i.e., they have identical consumption baskets.

19


• Equality between actual and planned expenditure on the two goods requires:

yC = −δ (1− a) z + ε£ayC + (1− a) yP

¤− νr, (5)

yP = δaz + ε£ayC + (1− a) yP

¤− νr. (6)

• Residents of each country allocate an equal fraction ε (0 < ε < 1) of an increase in incometo the domestic and the foreign good based on the spending share a for the core’s good and1− a for the periphery’s.

20


yC = −δ (1− a) z + ε£ayC + (1− a) yP

¤− νr,


¤− νr.

• Depreciation of a country’s terms of trade (i.e., its good becoming cheaper relative to thecounterpart’s – referred to as real depreciation of the country’s currency in Ghironi-Giavazzi)shifts world expenditure toward that country’s good:

– If z increases, the core’s good becomes relatively more expensive (the core’s terms oftrade appreciate), reducing planned demand of the core’s good and increasing planneddemand of the periphery’s.

– The effect of the terms of trade on planned expenditure depends on the positive elasticityparameter δ and is weighted by the share of the imported good in the consumptionbasket (1− a in the core’s consumption basket, a in the periphery’s).

– When a = 1 (i.e., when the size of the periphery is negligible), a depreciation of theperiphery’s terms of trade (or an appreciation of home’s terms of trade) has no impact onplanned expenditure on the core’s good, while it has the largest impact on yP .

21


yC = −δ (1− a) z + ε£ayC + (1− a) yP

¤− νr,


¤− νr.

• A higher ex ante interest rate (r) reduces planned expenditure on both goods equally:

– Residents of each country decrease spending by the same amount (0 < ν < 1) for eachpercentage point increase in the ex ante real interest rate facing them.

· Note: We could have written the real interest rate terms in (5) and (6) as arC +

(1− a) rP , assuming that agents can borrow and lend freely across countries.

· As we will verify below, perfect capital mobility and identity of the consumptionpatterns imply rC = rP = r.

22


• Ex ante real interest rates are given by:

rj = ij −³E³qj+1

´− qj

´, (7)

where ij is the nominal interest rate on bonds denominated in country j’s currency andE³qj+1

´is the expected value of country j’s CPI one period ahead based on the currently

available information.

– Thus, E³qj+1

´− qj is expected CPI inflation in country j, and ex ante real interest rates

are equal to the difference between nominal interest rates and expected CPI inflation.

23


• Agents can hold bonds denominated in both currencies.

– Even if there were controls on capital flows under the BW regime, they were increasinglyless pervasive and effective over time.

– To simplify, we just assume that capital mobility is perfect.

• As we know, optimal behavior by bond holders choosing how to allocate their wealthbetween the two bonds implies that the nominal interest rate differential is equal to expectedER depreciation:

iP − iC = E (e+1)− e. (8)

– This is known as the uncovered interest rate parity condition (UIP).

24


• Now consider the difference between rP and rC implied by (7):

rP − rC = iP − iC −£E¡qP+1¢−E

¡qC+1¢−¡qP − qC

¢¤.

• We proved above that e + qC − qP = 0 (PPP), which implies that qP − qC = e.

• Since PPP holds in all periods in the model, it is also the case that E¡qP+1¢− E

¡qC+1¢=

E (e+1).

• Using these results and (8), it follows that:

rP − rC = E (e+1)− e− [E (e+1)− e] = 0,

proving the statement we made above that free capital mobility and PPP imply equal exante real interest rates.

25


• While residents of each country hold bonds denominated in both currencies, they hold onlytheir domestic money.

• Demands for real money balances are given by:

mj − pj = yj − λij, (9)

E−1where λ > 0 is the interest rate semi-elasticity of money demand (semi-elasticity asopposed to elasticity because interest rates are not in logs in this model).

– Real money demand increases with the amount of transactions that agents make (thisexplains the positive effect of yj) and it decreases with the nominal interest rate.

· A higher interest rate makes it more attractive to hold bonds rather than money (i.e.,there is a higher opportunity cost of holding money), thus reducing the demand formoney.

– The model defines real money balances by deflating nominal money with the domesticoutput price.

– The results are not affected if real money balances are defined by deflating nominalmoney with the CPI (replacing pj with qj in (9)).

26


• Substitute yj = (1− α)nj − x (equation (1)) into (9):

mj − pj = (1− α)nj − x− λij.

• Next, note that wj − pj = −αnj − x (equation (2)) implies pj = wj + αnj + x.

• Hence,

mj −¡wj + αnj + x

¢= (1− α)nj − x− λij,

which implies:

nj = mj − wj + λij. (10)

– Since λ > 0, a higher nominal interest rate (other things given) causes labor demand torise.

– This is a consequence of money market equilibrium (equation (9), where we know thatmoney demand has to be equal to money supply in equilibrium):

· Holding mj and pj constant, if ij rises, a higher yj is required to keep equilibrium inthe money market.

– Of course, as we shall see, changes in monetary policy affect mj, pj, and ij, so that amonetary policy expansion will imply both lower ij and higher nj.

27


• Nominal wages are predetermined according to contracts signed before the beginning ofthe current period by competitive unions and firms.

• We assume that unions choose nominal wages to minimize a weighted average ofthe expected deviations of employment and real wages (in units of consumption) fromequilibrium wages subject to the constraint given by equation (10).

• Hence, unions solve:

minwj

1

2

nωE−1

h¡mj − wj + λij

¢2i+ (1− ω)E−1

h¡wj − qj

¢2io. (11)

• The first-order condition for this problem is both necessary and sufficient for optimality (sincewe are minimizing a quadratic objective function subject to a linear constraint).

28


• The first-order condition sets the first derivative of (11) with respect to wj to zero:

−ωE−1¡mj − wj + λij

¢+ (1− ω)E−1

¡wj − qj

¢= 0,

which implies:

wj = ωE−1¡mj + λij

¢+ (1− ω)E−1

¡qj¢. (12)

– Since wj is set at the time when expectations of events in the current period were formed,we can take it outside the expectations operator.

• Nominal wages are thus a weighted average of the expected indicators of monetary policystance (money supply and the interest rate) and the CPI.

29


• We are going to assume that the goal of policymakers in each country is to minimizeweighted averages of the deviations of CPI and employment from their zero-shockequilibrium values.

• This is going to imply that there is no domestic source of time inconsistency in monetarypolicy behavior:

– Policymakers do not have employment objectives above the zero-shock equilibrium levelsthat would induce agents to expect them to implement monetary expansion to driveemployment above the zero-shock level.

• In the model, policymakers maneuver their instruments only to stabilize the economy againstshocks.

• Since productivity shocks are unexpected (they have zero mean), this implies that unionsrationally expect policy instruments to remain at their zero-shock equilibrium values.

• In turn, this implies that agents rationally expect that all variables remain at their zero-shocklevels, and wage setting simplifies to:

wj = 0. (13)

30


• Given wj = 0, equations (10) and (2) imply, respectively:

nj = mj + λij, (14)

and

pj = αnj + x. (15)

• Central banks choose their policy instruments to minimize the following loss functions:

LCBj

=1

2

hγ¡qj¢2+ (1− γ)

¡nj¢2i

, 0 < γ < 1,

where γ measures the weight the authorities attach to the CPI versus employment objective.

– The central bank in each country is minimizing a weighted average of the deviations ofthe CPI and employment from their zero-shock levels.

– Since a deviation of the CPI from the zero-shock level is a movement of the CPI, we referto the CPI portion of the loss function as an inflation objective.

31


• The instruments available to the central banks depend on the ER regime between the coreand the periphery.

• Under MERs, the core’s central bank sets its money supply, while the periphery’s centralbank sets the ER.

• Under FlERs, each central bank sets its own money supply.

• In both cases, we assume that monetary policies are set non-cooperatively, i.e., we focuson the Nash equilibrium of the game between the central banks, in which each policymakeroptimizes his/her objective function taking the other policymaker’s behavior as given.

– Parity realignments under the BW regime (or under the European Monetary System) didinvolve some form of cooperation, but certainly not the joint minimization of central bankloss functions that the standard modeling of policy coordination assumes.

32

Exchange Rate Determination under Flexible Exchange Rates

• Before plunging into the analysis of policy tradeoffs and the game between policymakers, itis instructive to study how the model determines the ER under the FlER regime.

• We use a superscript D to denote cross-country differences between periphery and corevariables below.

• If we subtract the core’s money demand equation from the periphery’s, we have:

mD − pD = yD − λiD.

• Subtracting the core’s expenditure equation from the periphery’s and using the definition ofthe terms of trade yields:

yD = δz = δ¡e− pD

¢.

• Using equations (14) and (15),

pD = αnD = α¡mD + λiD

¢.

• Hence,

yD = δ£e− α

¡mD + λiD

¢¤.

33

Exchange Rate Determination under Flexible Exchange Rates, Continued

• Substitute yD = δ£e− α

¡mD + λiD

¢¤and pD = α

¡mD + λiD

¢into mD − pD = yD − λiD, and

use UIP (iD = E (e+1)− e):

mD − α£mD + λ (E (e+1)− e)

¤= δ

©e− α

£mD + λ (E (e+1)− e)

¤ª− λ (E (e+1)− e) .

• This equation implies that the current ER depends on the current money supply differentialand the expected future ER:

e =1− α (1− δ)

λ [1− α (1− δ)] + δmD +

λ [1− α (1− δ)]

λ [1− α (1− δ)] + δE (e+1) . (16)

• The restriction 0 < δ < 1, combined with 0 < α < 1, is sufficient to ensure that thecoefficients of mD and E (e+1) are positive:

– The periphery’s currency depreciates if the periphery’s money supply increases relativeto the core’s and if it is expected to depreciate in the future.

34

Exchange Rate Determination under Flexible Exchange Rates, Continued

e =1− α (1− δ)

λ [1− α (1− δ)] + δmD +

λ [1− α (1− δ)]

λ [1− α (1− δ)] + δE (e+1) .

• You should recognize this ER equation as a discrete-time variant of the ER equation weused when discussing Krugman’s model of target zones.

– Specifically, note that we can rewrite the ER as function of the money supply differentialand the expected rate of depreciation between this period and the next (as in theKrugman model) as:

e =1− α (1− δ)

δmD +

λ [1− α (1− δ)]

δ(E (e+1)− e) .

• Since expected future variables are at their zero-shock equilibrium levels in the Ghironi-Giavazzi model, E (e+1) = 0, and we have the solution for the ER under the FlER regimesimply as:

e =1− α (1− δ)

λ [1− α (1− δ)] + δmD. (17)

– Technical note: The assumption that E (•+1) = 0 rules out speculative bubbles.

35

The Reduced Form Solution

• In the reduced form solution of the model under FlER or MER, endogenous variables arelinear functions of the policy instruments and the world shock x.

– For instance, in the solution for the ER under FlER above, the ER is a linear function ofmoney supplies (specifically, of the money supply differential).

– The world shock does not appear because it affects both countries identically.

• Given linearity of endogenous variables in policy instruments and x, when x = 0, zero valuesof the policy instruments ensure zero losses for both central banks.

• This proves the rationality of expecting zero values of policy instruments under theassumption that shocks have zero mean.

36

The Reduced Form Solution, Continued

Managed Exchange Rates• The reduced form for employment and the CPI in each country under MER can be written

as:

qC = AmC −B (1− a) e + Σx,

qP = AmC + [1−B (1− a)] e + Σx,

nC = ΛmC +∆ (1− a) e−Hx,

nP = ΛmC + (Ω−∆a) e−Hx, (18)

where upper-case Greek letters denote parameters that are functions of the structuralparameters of the model (i.e., of the parameters in the structural relations we positedabove).

• Reduced form parameters (the coefficients on policy instruments and x) are written so as tohighlight the role of country size:

– When a = 1, i.e., if the periphery is a small open economy, changes in the ER have noeffect on the core, but they have a one-to-one effect on the periphery’s CPI.

37


Flexible Exchange Rates• The reduced form for employment and the CPI in each country under MER can be written

as:

qC = [A+ Φ (1− a)]mC − Φ (1− a)mP + Σx,

qP = (A + Φa)mP − ΦamC + Σx,

nC = [Λ− Γ (1− a)]mC + Γ (1− a)mP −Hx,

nP = (Λ− Γa)mP + ΓamC −Hx. (19)

• If the periphery is a small open economy, its policy choices have no effect on the core.

• If the two countries have equal size (a = .5), symmetry of the ER regime implies symmetricreduced forms.

38


The Effect of the World Shock• Irrespective of ER regime and country size, a positive realization of x causes inflation and

unemployment in both countries.

• The choice of focusing on central banks’ interactions after this type of supply disturbance(neglecting other types of shocks) is motivated precisely by the fact that these disturbancesare the most typical example of shocks that present policymakers with a tradeoff betweenemployment and inflation stabilization.

• For all plausible parameter values, a monetary policy contraction aimed at stabilizinginflation further decreases employment, whereas a monetary expansion designed to boostemployment has inflationary consequences.

• Understanding the determinants of a country’s employment-inflation tradeoff is thusimportant for analyzing the country’s policymaking.

39

Size, ER Regimes, and the Employment-Inflation Tradeoff

• The first result that we show is that the tradeoff facing the periphery under MER becomessteeper as the size of the country gets smaller.

• A steeper tradeoff allows the central bank to trade a larger inflation gain for a smalleremployment loss.

• If the central bank is sufficiently averse to inflation, a steeper tradeoff is also a morefavorable tradeoff.

• Thus, when we argue that a steeper tradeoff is better, we are implicitly assuming thatcentral banks care more about inflation than employment (γ > .5 in the central banks’ lossfunctions).– The assumption that central banks care more about inflation than unemployment became

increasingly realistic across the industrial world after the early 1970s.

• This assumption also implies that central banks will respond to positive realizations of x bycontracting monetary policy.

40

Size, ER Regimes, and the Employment-Inflation Tradeoff, Continued

The Periphery’s Tradeoff under Managed Exchange Rates• The periphery’s employment-inflation tradeoff under MER is defined by:

∂qP

∂nP≡ ∂qP/∂e

∂nP/∂e.

• The expression ∂u/∂v denotes the derivative of variable u with respect to variable v for anypair of variables u and v.

• In the right-hand side of the tradeoff’s definition, the numerator tells us by how much achange in the periphery’s policy instrument causes its CPI to change; the denominator tellsus how much the same change in the periphery’s instrument causes its employment tochange.

• Numerator and denominator together tell us how much the CPI changes in association withthe employment change implied by a given change in the policy instrument.

41


• Using the reduced form solutions under MER, we have:

∂qP

∂nP=1−B (1− a)

Ω−∆a.

• The numerator of this expression is an increasing function of a:– As the periphery becomes small, and correspondingly reduces its consumption of

domestic goods, its CPI increasingly depends upon the price at which peripheryresidents can buy core products, i.e., on the ER.

– As we mentioned above, in the limit case in which a=1, the periphery consumes onlycore goods and e has a one-to-one impact on qP

• To complete the proof that the periphery’s tradeoff improves as its size becomes smaller,it is thus sufficient to show that the denominator of the above expression is a decreasingfunction of a.

42


• From equation (14), nP = mP (e) + λiP , where mP depends on e through the endogeneityconstraint imposed by the MER regime (since e is the instrument of the periphery’s centralbank under MER, its money supply becomes an endogenous variable).

• This constraint has the form mP = mC + (1/η) e, where η is a parameter independent of a.– Note 1: You can recover the constraint and the expression of η as function of structural

parameters from equation (17), recalling that mD ≡ mP −mC, and that, under MER, e isa policy instrument and mP is endogenous.

– Note 2: The constraint mP = mC + (1/η) e highlights the fact that, for given ER (i.e., whene = 0) the periphery imports the core’s monetary policy (i.e., mP = mC).

• From the fact that η does not depend on a, it follows that the impact of e on nP varies withthe size of the periphery because the impact of e on iP depends on a.

43


• Using the uncovered interest parity condition (8) allows us to write iP = iC − e (sinceE (e+1) = 0).

• Therefore, if there is a change de in the ER:

∂iP

∂ede =

∂iC

∂ede− de

• When a = 1, changes in e have no impact on the core, so that ∂iP/∂e = −1, and

∂nP

∂e=

∂mP

∂e+ λ

∂iP

∂e=1

η− λ.

• For a < 1, we have:∂nP

∂e=

∂mP

∂e+ λ

∂iP

∂e=1

η+ λ

µ∂iC

∂e− 1¶.

• For the latter expression to be greater than the corresponding expression evaluated at a = 1,it has to be ∂iC/∂e > 0.

• The absolute value of ∂iC/∂e is intuitively decreasing in a, because the impact of theperiphery on the core is smaller the larger the core.

• Showing that ∂iC/∂e > 0 would conclude the proof that the periphery’s tradeoff improves asa increases, since it would imply that ∂nP/∂e decreases as the size of the core increases.

44


• But note now that:∂iC

∂e= 1 +

∂iP

∂e.

• Hence, ∂iC/∂e > 0 if and only if ∂iP/∂e > −1.

• But a = 1 is the situation in which changes in e have the greatest impact on the periphery’svariables:

• In such situation ∂iP/∂e = −1.

• Therefore, ∂iP/∂e cannot decrease below −1 and is strictly above if a < 1.

• Thus, the tradeoff for the country that controls the ER improves as its size gets smaller.

45


• To further understand why, when a < 1, ∂iC/∂e > 0, observe that, in general, the UIPcondition implies iP − iC = E (e+1)− e.

• For any given value of E (e+1) (including zero), if e decreases, the expected depreciation ofthe periphery’s currency (E (e+1) − e) increases, and so does the interest rate differential,iP − iC, to preserve portfolio equilibrium.

• Unless the size of the periphery is negligible, iP − iC widens when e decreases becauseinterest rates fall in the core and rise in the periphery.

• A larger expected depreciation makes the periphery’s bonds less attractive.

• Investors substitute away from the periphery’s bonds into the core’s.

• Hence the price of the former decreases, so that iP increases, and the price of the latterincreases, so that iC decreases (i.e., ∂iC/∂e > 0).

• If a = 1, holdings of core’s bonds relative to periphery’s bonds do not change and ∂iC/∂e = 0.

46


• To summarize, the employment-inflation tradeoff facing the periphery under MER steepens,as the size of the periphery becomes smaller, for two reasons.

1. First, due to our assumptions about the pattern of trade, a relatively smaller peripheryconsumes more goods produced in the core.– Thus, the fall of the CPI induced by (say) an ER appreciation (a decrease in e) is larger.

2. The impact of the appreciation on employment becomes smaller because interest ratesin the core are less affected, while iP rises by more, thus reducing the fall in employmentrequired to restore equilibrium in the money market.– Recall that nominal appreciation implies a decrease in mP via the constraint mP =

mC + (1/η) e.– Given nP = mP (e) + λiP , a larger increase in iP reduces the fall in employment implied

by lower mP .

• This result is illustrated in Figure 1, which shows the employment-inflation tradeoff facingthe periphery for two values of a.– The steeper line corresponds to the case where the core is relatively large, and the

periphery is relatively small.– Along this line the tradeoff is more favorable if the central bank is averse to inflation.

47

268 F. Ghironi, F. Giavazzi / Journal of International Economics 45 (1998) 259 –296

To summarize, the employment–inflation tradeoff facing the periphery undermanaged exchange rates steepens, as the size of the periphery becomes smaller, fortwo reasons. First, due to our assumptions about the pattern of trade, a relativelysmaller periphery consumes more goods produced in the core: thus, the fall of theCPI induced by (say) an exchange rate appreciation is larger. At the same time, theimpact of the appreciation on employment becomes smaller because interest rates

Pin the core are less affected, while i rises by more, thus reducing the fall in11employment required to restore equilibrium in the money market. This result is

illustrated in Fig. 1, which shows the employment–inflation tradeoff facing theperiphery for two values of a. The steeper line corresponds to the case where thecore is relatively large, and the periphery is relatively small; along this line thetradeoff is also more favourable if the central bank is averse to inflation.

We next consider how size affects the periphery’s tradeoff under flexibleexchange rates. One might expect that for any given relative size the tradeoffdepends on the monetary regime. This is not true: we shall show that for any givenrelative size, the periphery faces the same tradeoff independently of the exchange-rate regime.

Fig. 1. The employment–inflation tradeoff of the periphery.

11 PRecall that a nominal appreciation implies a decrease in m via the managed exchange ratesconstraint.


The Periphery’s Tradeoff under Flexible Exchange Rates• Next consider how size affects the periphery’ s tradeoff under FlER.

• One might expect that for any given relative size the tradeoff depends on the ER regime.

• But this is not true: We show that for any given relative size, the periphery faces the sametradeoff independently of the ER regime.

– See page 269 of the Ghironi-Giavazzi article for the proof.

48


The Core’s Tradeoff• What about the core country?

• Note first that under MER, the core, contrary to the periphery, always faces the sameemployment-inflation tradeoff, independently of its size.

• This is apparent if we observe that under MER the reduced forms imply:

∂qC

∂nC≡ ∂qC/∂mC

∂nC/∂mC=

A

Λ,

where A and Λ are parameters that do not vary with a.

49


• Instead, under FlER (a symmetric ER regime), the definitions of “core” and “periphery”become arbitrary (we can just switch the names as we wish).

• Hence, if (as we have shown) the periphery faces the best tradeoff when it is small, if weswitch the names and we call “core” the country that was previously “periphery” and viceversa, it follows that also the core faces the best tradeoff when it is small and both countriesface identical tradeoffs when they are exactly identical (a = .5).

• When a = 1, the core faces the same tradeoff it would face under MER, as it is easy tocheck.

• The intuition is straightforward:– When the periphery is a small open economy, it has no impact on the core.– Therefore, the tradeoff facing the core must be unaffected by the ER regime.

50


• The common intuition that underlies all the results above is that the tradeoff facing a centralbank depends on the size of the area for which the central bank sets its instrument (inaddition to the other structural parameters).

• As we noted at the beginning, in the case of the periphery this size is independent of the ERregime:– The central bank of the periphery always sets its instrument only for its own economy,

independently of the ER regime.

• This is not the case for the core.– Under FlER the core central bank sets the money supply only for the core itself, but in

the MER regime the central bank of the core sets money supply for the entire world (forgiven ER).

– Thus, the economy that is relevant to determine the tradeoff facing the core centralbank is the entire world, core plus periphery, regardless of the relative size of the twoeconomies.

• Changing the ER regime changes the size of the economy for which the core central banksets its instrument, but not the size of the economy for which the periphery’s central banksets its instrument.

51


• The more favorable the tradeoff a country faces relative to its neighbors, the larger theamount of inflation that such a country can shift upon them with its policy choice in responseto the common shock x, independently of the ER regime.

• Under MER, the periphery always faces a better tradeoff than the core regardless of itsrelative size, because the core faces the worst possible tradeoff.

• Under FlER, the country facing the best tradeoff is the one whose relative size is smaller.– The tradeoff depends on a set of structural characteristics of the country, not only its size.– But since all other characteristics are symmetric across countries in the model, size is the

crucial determinant of how favorable a country’s tradeoff is relative to the other country’s.

• Of course, if the country that exports inflation is much smaller than that which suffers theconsequences of its partner’s beggar-thy-neighbor policy, the impact of exported inflation onthe latter economy will be correspondingly reduced.

• One implication is that, under MER, the periphery can export inflation to the core also whena = .5.

• Under FlER, the periphery can manage to export inflation only if it is smaller than the core.

52


• Giavazzi and Giovannini (1989) argue that the employment–inflation tradeoff facing thecountry that controls the ER under MER is superior to that facing the country that sets theworld-wide money supply.

• Their intuition is that this happens because, under MER, the country that controls the ERcan improve its tradeoff by exporting inflation abroad via ER appreciation.

• But if this interpretation were correct, one would expect the periphery to face a bettertradeoff under MER than under FlER independently of its size, which, as we have shown, isnot true.– Although it is true that if the periphery’s central bank controls the exchange rate – and

thus the inflation differential qP − qC (since PPP implies qP − qC = e) – the central bank ofthe periphery will affect the position of its tradeoff relative to that of the core by “driving”the latter to the worst situation under FlER.

53


• The correct way to think about the result is that countries successfully run beggar-thy-neighbor policies when their (more favorable) tradeoffs allow them to do so.

• This is different from thinking that countries face more favorable tradeoffs because theysuccessfully run beggar-thy-neighbor policies.

• The result presented in Giavazzi and Giovannini (1989) according to which the central bankof the core ranks FlER above MER unless its size is much bigger than the periphery’s cannow be reinterpreted correctly as follows.

• For any value of a smaller than one, the core faces a better tradeoff under FlER than underMER.

• Instead, when a = 1, the tradeoff facing the core is the same irrespective of the ER regime.

54

Monetary Policy Interactions

• The policy tradeoffs of core and periphery were determined independently of the policymak-ers’ preferences.

• These mattered only insofar as we assumed central banks to be relatively more concernedabout inflation.

• Both structural constraints (the determinants of the tradeoffs) and policymakers’ preferencesbecome relevant in determining the outcome of policy interactions because central banksminimize the respective loss functions subject to the constraints given by their tradeoffs.

• Using the results above, the reduced form equations for the CPI’s and employment can berewritten as follows.

55

Monetary Policy Interactions, Continued

Managed Exchange Rates

qC = AmC −µ1− A

η

¶(1− a) e + Σx,

qP = AmC +

∙a +

A

η(1− a)

¸e + Σx,

nC = ΛmC +∆ (1− a) e−Hx,

nP = ΛmC +

µΛ

η−∆a

¶e−Hx. (20)

56


Flexible Exchange Rates

qC = [Aa + η (1− a)]mC − (η −A) (1− a)mP + Σx,

qP = [(A (1− a) + ηa)]mP − (η −A) amC + Σx,

nC = [Λ− η∆ (1− a)]mC + η∆ (1− a)mP −Hx,

nP = (Λ− η∆a)mP + η∆amC −Hx. (21)

57


• In the non-cooperative MER regime, the central bank of the core chooses its money supplyto minimize its loss function taking the ER as given.

• The central bank of the periphery chooses the ER to minimize its loss function taking thecore’s money supply as given.

• Taking the first derivative of each central bank’s loss function with respect to its instrumentand setting it equal to zero, this results in the first-order conditions:

γqC∂qC

∂mC+ (1− γ)nC

∂nC

∂mC= 0 for the core’s central bank, and (22)

γqP∂qP

∂e+ (1− γ)nP

∂nP

∂e= 0 for the periphery’s central bank. (23)

• These conditions are necessary and sufficient for optimality because we are minimizing aquadratic function (the loss function) subject to linear constraints (the reduced forms for CPIand employment).

• The derivatives of CPI and employment with respect to instruments in these conditions canbe easily recovered from the reduced form solutions in (20).

58


• In the non-cooperative FlER regime both central banks use money supply as theirinstrument, taking their counterpart’s money supply as given.

• In this case the first-order condition for the core’s central bank is unchanged, and thefirst-order condition for the periphery’s central bank is similar, with superscript P replacing C

everywhere.

• However, under FlER, the derivatives of CPI and employment with respect to instruments inthe first-order conditions are recovered from the reduced form solutions in (21).

59


• Under each regime, a central bank’s first-order condition defines that central bank’s reactionto the shock x and to what the other central bank is doing.

• For instance, consider the first-order condition of the core’s central bank under MER:

γqC∂qC

∂mC+ (1− γ)nC

∂nC

∂mC= 0.

• Using the reduced forms in (20), this becomes:

γ

∙AmC −

µ1− A

η

¶(1− a) e + Σx

¸A + (1− γ)

£ΛmC +∆ (1− a) e−Hx

¤Λ = 0.

• If you solve this equation with respect to mC (i.e., move all the terms involving x and e to theright-hand side and collect coefficients), the resulting equation will tell you how the core’scentral bank sets mC in response to the shock x and the periphery’s choice of e.

• If a < 1, the core’s central bank responds to the periphery’s choice of e because this affectsqC and nC (i.e., the variables that the core’s central bank is trying to stabilize).

• If a = 1, the periphery’s actions have no effect on the core, and you will find that the core’scentral bank responds only to x.

• You can similarly analyze the first-order condition for the periphery’s central bank underMER, and both central banks’ first-order conditions under FlER.

60


Indifference Curves and Reaction Functions• The central banks’ loss functions that we posited above are analogous to the policymakers’

objectives that we had posited in Barry Eichengreen’s model of policy interactions under theinterwar GS (quadratic functions defined over two objectives).

• Therefore, we could represent them in a diagram by means of a set of indifference curves, or“altitude contours” (where each contour gives us the combinations of CPI and employmentthat imply the same loss level).

• Since CPI and employment in each country are functions of the policy instruments, as wehad done with Eichengreen’s model, we could re-draw the indifference curves in a diagramwith the policy instruments of the two central banks on the horizontal and vertical axes.

• In this case, for each central bank, an indifference curve would give us the combinations ofits policy instrument and the other central bank’s instrument that imply the same loss levelfor the central bank we are considering.

• The indifference curves for each central bank would be centered around the respective blisspoint, which is the zero-loss point that the central bank would choose if it controlled both itsinstrument and the other central bank’s instrument.

61


• Since central banks control only one instrument under non-cooperation, they cannot achievethe bliss point, and we know they face a tradeoff between their objectives:

– A positive shock to x causes inflation and unemployment;

– monetary contraction reduces inflation, but it increases unemployment;

– monetary expansion boosts employment, but it increases inflation.

• Under the assumption that central banks care more about inflation than employment(γ > 1/2 in the loss functions), both policymakers will respond to the shock x > 0 bycontracting monetary policy.

62


• Additionally, we could plot in the diagram each central bank’s reaction function implied by itsfirst-order condition, which tells us how the central bank responds with its instrument to theother central bank’s instrument setting.

• Given the assumption that central banks care more about inflation than employment, thereaction functions will both be positively sloped under FlER:

– Since a monetary policy contraction in the periphery (lower money supply, mP ↓) causesqC to increase, the core’s central bank responds by lowering mC.

– Similarly, since a monetary policy contraction in the core (mC ↓) causes qP to increase,the periphery’s central bank responds by lowering mP .

• Under MER, a monetary policy contraction in the periphery (ER revaluation, e ↓) causes qC

to increase. Hence, the core’s central bank responds by lowering mC.

• Lower mC reduces qP for given e, so the central bank of the periphery would respond tolower mC by increasing e (or by reducing the extent of the revaluation implied by x > 0).

• In other words, while the core’s reaction function is positively sloped, the periphery’s reactionfunction under MER is negatively sloped.

63


• In either regime, we would have a diagram akin to the figure we used to analyze thepolicy game in Eichengreen’s article, and the intersection of the two central banks’ reactionfunction would be the Nash equilibrium of the game (the point where both central banksare responding optimally to what the other central bank is doing) and would give us thecorresponding, Nash-equilibrium (or non-cooperative) levels of the policy instruments.

– Stability of the Nash equilibrium under FlER requires the periphery’s central bank’sreaction function to be flatter than the core’s if we have mC on the horizontal axis and mP

on the vertical axis.

– Stability of the Nash equilibrium under MER requires the absolute value of the periphery’scentral bank’s slope to be smaller than the core’s if we have mC on the horizontal axisand e on the vertical axis.

– We assume that these conditions are satisfied.

64


Cooperation versus Non-Cooperation• If central banks coordinate their policies, they act together to minimize jointly a weighted

average of the respective loss functions LCBC and LCBP .

• As we learned from Eichengreen’s paper, this allows them to achieve a point on the contractcurve – the locus of the points such that neither central bank can be made better off withoutmaking the other worse off (or the locus of tangency points between the two central banks’indifference curves).

• Which point is reached depends on the central banks’ bargaining power, captured in themodel by the weight attributed to its loss function in the joint loss minimization problem.

• Since the core has economic size a in our model and the periphery has size 1− a, a naturalchoice for weights is a for the core and 1− a for the periphery, so that coordination impliesthat central banks jointly minimize aLCBC

+ (1− a)LCBP , or:a

2

hγ¡qC¢2+ (1− γ)

¡nC¢2i

+1− a

2

hγ¡qP¢2+ (1− γ)

¡nP¢2i

. (24)

65


• For instance, under flexible exchange rates, central banks jointly choose mC and mP tominimize expression (24) subject to the reduced form solutions in (21).

• This results in the first-order conditions:

a

∙γqC

∂qC

∂mC+ (1− γ)nC

∂nC

∂mC

¸+ (1− a)

∙γqP

∂qP

∂mC+ (1− γ)nP

∂nP

∂mC

¸= 0,

a

∙γqC

∂qC

∂mP+ (1− γ)nC

∂nC

∂mP

¸+ (1− a)

∙γqP

∂qP

∂mP+ (1− γ)nP

∂nP

∂mP

¸= 0, (25)

where the derivatives involved in the equations are easily obtained from the reduced formsolutions in (21).

66


• Note the key difference between the cooperative first-order conditions in (25) and non-cooperative behavior:– Under non-cooperation, the optimal choice of mC is determined by:

γqC∂qC

∂mC+ (1− γ)nC

∂nC

∂mC= 0.

· This does not take into account the effects of mC (core monetary policy) on qP andnP .· In other words, there is no internalization of the externalities generated by policy.

– Under cooperation (or coordination), the optimal choice of mC is determined by the firstequation in (25):

a

∙γqC

∂qC

∂mC+ (1− γ)nC

∂nC

∂mC

¸+ (1− a)

∙γqP

∂qP

∂mC+ (1− γ)nP

∂nP

∂mC

¸= 0.

· The second part of this equation reflects the internalization of policy externalities, bytaking into account the effects of mC on qP and nP (with weight 1− a).

• Similarly for mP .

67


• When central banks do not cooperate, if they care more about inflation than employment,they try to use their policy instruments to manipulate the terms of trade z to their advantage,in order to generate lower domestic CPI inflation by inducing a terms of trade appreciation(making their imports cheaper via a stronger ER) in response to a shock x > 0 that causesinflation and unemployment.

• By doing so, each central bank is de facto trying to export some of its inflation to the othercountry by contracting its monetary policy.

• As we learned above, a central bank is successful at this when it faces a more favorabletradeoff than its counterpart.

– Under MER, this is always the case for the periphery.

– Under FlER, it is the smaller country’s central bank that actually manages to exportinflation by taking advantage of its more favorable tradeoff.

68


• If a = 1/2, under FlER, the central banks are facing identical tradeoffs, and neither of themmanages to export inflation to the other:

– Both central banks choose exactly the same money supply (since the two countries areidentical when a = 1/2, and they are affected by the same shock x), and the ER actuallydoes not move at all in equilibrium.

• But the effort to try and export inflation results in excessive monetary contraction (much ascompetition for gold resulted in excessive monetary contraction in Eichengreen’s model ofthe interwar GS), which causes excessive employment loss in both countries.

• If central banks cooperate, they give up trying to export inflation to one another (byinternalizing the external effects of policy).

• The excessive monetary contraction is removes, and both central banks are better off byfacing lower employment losses.

• This is the same analysis of cooperation versus non-cooperation as in Matthew Canzoneriand Dale Henderson’s (1991) MIT Press book, although Canzoneri and Henderson do notfocus on the role of policy tradeoffs as in Ghironi-Giavazzi.

69

Monetary Policy Interactions: A Numerical Example

• In Ghironi-Giavazzi, you find a numerical exercise (under non-cooperation) that comparesMER and FlER for different relative sizes of the two countries.

• We consider three possible relative sizes of the two regions: a = .5, a = .75, and a = 1.

• When a = .5, the two countries are identical; when a = .75, the periphery is one-third thesize of the core; and when a = 1 the size of the periphery is negligible.

– Symmetry of the model makes the cases in which a is smaller than .5 redundant.

• The values we assign to the other structural parameters of the model are: α = .34, ε = δ = .8,ν = .4, and λ = .6.

• These parameter values are empirically plausible and result in shocks x having a significantimpact on employment.

• They also imply a non-negligible external effect of domestic policies on foreign employmentunder flexible exchange rates.

• Numerical values of the reduced forms implied by these parameter values are in Table A1.

70

292 F. Ghironi, F. Giavazzi / Journal of International Economics 45 (1998) 259 –296

with the rest of the world. Our results have implications for EMU and itsconsequences for the international monetary system.

Acknowledgements

¨We thank for valuable comments Mario Blejer, Alex Cukierman, Jurgen vonHagen, Olivier Jeanne, Peter Kenen, Maury Obstfeld, David Romer, Andy Rose,Bent Sørenson, and an anonymous referee. We are also grateful to the participantsin the Conference on Optimum Currency Areas held at the Sapir Center, Tel AvivUniversity (December 4–6, 1996), and in the Berkeley international economicsseminar. Ghironi gratefully acknowledges financial support from CEPR, Ente‘‘Einaudi,’’ and the Institute of International Studies of the University ofCalifornia, Berkeley.

Appendix A

Numerical solutions to the stabilization games

Table A1. Reduced forms in a two-region world

(1) Asymmetric regime

C Cq 5 0.2425m 2 0.7585(1 2 a)e 1 0.9272x

P Cq 5 0.2425m 1 (0.2415 1 0.7585a)e 1 0.9272x

C Cn 5 0.7133m 1 0.3145(1 2 a)e 2 0.2140x

P Cn 5 0.7133m 1 (0.7101 2 0.3145a)e 2 0.2140x

(2) Symmetric regime

C C Pq 5 (1.0044 2 0.7619a)m 2 0.7619(1 2 a)m 1 0.9272x

P P Cq 5 (0.2425 1 0.7619a)m 2 0.7619am 1 0.9272x

C C Pn 5 (0.3974 1 0.3159a)m 1 0.3159(1 2 a)m 2 0.2140x

P P Cn 5 (0.7133 2 0.3159a)m 1 0.3159am 2 0.2140x

Monetary Policy Interactions: A Numerical Example, Continued

• Finally, we assume that central banks care much more about CPI inflation than employment,setting γ = .9.

• The solution for the optimal (non-cooperative) policy instrument choices under the two ERregimes, together with the implied values of endogenous variables, loss functions, andinflation-employment ratios, is in Tables A2 and A3.

71

F. Ghironi, F. Giavazzi / Journal of International Economics 45 (1998) 259 –296 293

aTable A2.

Non-cooperative Periphery’s Periphery’s Periphery’sasymmetric size negligible size small size equal toregime (a51) (a50.75) core’s (a50.5)

C˜Core’s money (m ) 21.8026 21.9464 22.0669˜Nominal periphery /core (e ) 20.4169 20.4173 20.3834

˜Real periphery /core (z ) 20.3608 20.3612 20.3318C˜Core’s CPI (q ) 0.4901 0.5343 0.5714

P˜Periphery’s CPI (q ) 0.0732 0.1170 0.1880C˜Core’s employment (n ) 21.4997 21.6351 21.7485

P˜Periphery’s employment (n ) 21.6646 21.8002 21.9002Loss core’s CB 0.2205 0.2622 0.2998Loss periphery’s CB 0.1410 0.1682 0.1964

C C˜ ˜q /n 20.3268 20.3268 20.3268P P˜ ˜q /n 20.0439 20.0650 20.0990

a In this table and in the following ones, the values of the policy instruments and of the endogenous2variables should be multiplied by x, while the values of the loss functions should be multiplied by x . A

positive realization of x is a negative supply-side shock, which lowers employment and raises the CPI.Given that variables are in logs, the numbers we report in the tables are the elasticities of policyinstruments and endogenous variables with respect to the supply shock implied by the relevantpolicymaking regime and size of the currency area. We then calculate the losses implied by thoseelasticities.

Table A3.

Non-cooperative Periphery’s Periphery’s Periphery’ssymmetric size neglibible size small size equal toregime (a51) (a50.75) core’s (a50.5)

CCore’s money (m ) 21.8026 22.5860 22.8939PPeriphery’s money (m ) 22.2177 22.7828 22.8939

Real periphery /core (z) 20.3608 20.1710 0CCore’s CPI (q ) 0.4901 0.3376 0.2254

PPeriphery’s CPI (q ) 0.0732 0.1399 0.2254CCore’s employment (n ) 21.4997 22.0741 22.2781

PPeriphery’s employment (n ) 21.6646 22.1522 22.2781Loss core’s CB 0.2205 0.2664 0.2823Loss periphery’s CB 0.1410 0.2404 0.2823

C Cq /n 20.3268 20.1627 20.0990P Pq /n 20.0439 20.0650 20.0990


• Letting Â denote “preferred to,” Table 1 summarizes the central banks’ preference rankingsover the size of the core country (i.e., this tells us what is the size-situation that centralbanks find preferable).

• Both monetary authorities have preference rankings over the size of the core such that:1 Â .75 Â .5.

• The large core outcome is the most preferred for both policymakers.

72

F. Ghironi, F. Giavazzi / Journal of International Economics 45 (1998) 259 –296 277

16policies on foreign employment under flexible exchange rates. Numerical valuesof the reduced forms are shown in Appendix A (Table A1). Finally, we make therealistic assumption that central banks care much more about CPI inflation thanabout employment, choosing g 50.9. The solution of the system under the tworegimes, together with the implied values of endogenous variables, loss functions,and inflation–employment ratios, is shown in Tables A2 and A3. Letting s denote‘‘preferred to,’’ Table 1 summarizes the central banks’ preference rankings overthe size of the core region.

Both monetary authorities have preference rankings over the size of thecurrency area of the type: 1s0.75s0.5. The larger core outcome is the first bestfor both policymakers. These results are consistent with the insights obtainedanalyzing marginal changes. The intuition is straightforward, although here oneneeds to be more careful in considering the impact of changes in a on theequilibrium values of the policy instruments. Consider first the periphery’s centralbank. As a increases, this central bank faces an improving tradeoff under bothregimes. Under the asymmetric regime, the core’s central bank always faces thesame tradeoff, regardless of the value of a. However, as the periphery shrinks, theimpact of imported inflation on the core economy becomes smaller. Other thingsbeing given, because central banks care more about inflation than about employ-ment, facing a more favourable tradeoff as a increases strengthens the periphery’sincentives to behave aggressively. Thus, for any given policy adopted by thecore’s central bank, the smaller impact on core inflation of inflation imported fromthe periphery as a increases (a structure-related effect), must be weighed againstthe fact that the periphery is induced to shift a larger amount of inflation upon thecore (a preference-related effect), when determining the overall impact of theperiphery’s actions on the core economy. In principle, there may be situations inwhich inflation in the core rises as a increases as a consequence of the periphery’s

Table 1Summary of preference rankings in a two-region world

Core’s central bankAsymmetric regime 1s0.75s0.5Symmetric regime 1s0.75s0.5

Periphery’s central bankAsymmetric regime 1s0.75s0.5Symmetric regime 1s0.75s0.5

16As shown by Canzoneri and Henderson (1991), l50 would imply no effect of monetary policy onforeign employment under flexible exchange rates and no impact of the supply shock on employmentanywhere in the world.


• As a increases, the periphery’s central bank faces an improving tradeoff under both ERregimes.

• Under MER, the core’s central bank always faces the same tradeoff, regardless of the valueof a.

• However, as the periphery shrinks, the impact of imported inflation on the core economybecomes smaller.

• Other things given, because central banks care more about inflation than employment,facing a more favorable tradeoff as a increases strengthens the periphery’s incentives tobehave aggressively.

• Thus, for any given policy adopted by the core’s central bank, the smaller impact on coreinflation of inflation imported from the periphery as a increases (a structure-related effect),must be weighed against the fact that the periphery is induced to shift a larger amount ofinflation upon the core (a preference-related effect), when determining the overall impact ofthe periphery’s actions on the core economy.

73


• In principle, there may be situations in which inflation in the core rises as a increases as aconsequence of the periphery’s behavior.

– Although this will not happen for values of a sufficiently close to 1, i.e. when the peripheryhas no impact on the core.

• This notwithstanding, one needs to remember that the final outcome of the stabilizationgame is not uniquely determined by the periphery’s incentives and choices.

• In particular, the fact that, other things given, facing a better tradeoff strengthens theincentives to act aggressively does not mean that the periphery automatically does so in theequilibrium of the game.

• For example, the periphery’s aggressiveness is reduced as a goes from .75 to 1.

• Strategic interactions with other players can induce lower aggressiveness as equilibrium out-come when the periphery’s policymaker optimally trades control of inflation for employmentstabilization given the other player’s reaction.

• In the case we are studying, the combined effect of these considerations is such thatequilibrium inflation in the core decreases monotonically as a goes from .5 to .75 and 1.

74


• Under FlER, the two central banks face identical tradeoffs when a = .5, and none of them isable to export inflation to its counterpart.

• When the core becomes larger, the tradeoff facing its central bank worsens and equilibriuminflation increases.

• However, the smaller employment loss implied by a less aggressive monetary policy morethan offsets the worse inflationary outcome.

– Note that also under FlER the periphery is less aggressive as a increases from .75 to 1.

75


• The results in tables A2 and A3 allow us to address also the question of how the two centralbanks rank the ER regimes we have considered.

• Consistent with the intuition, the periphery’s central bank prefers MER when a is smallerthan 1, whereas the core’s central bank would choose FlER over MER when a = .5.

• Both central banks are indifferent about the ER regime when a = 1.

– That the core’s central bank is indifferent between ER regimes when a = 1 is intuitive.– Indifference by the periphery’s central bank follows from the fact that the core’s

policymaker chooses exactly the same policy irrespective of the regime when a = 1.– As a consequence, given an unchanging tradeoff across regimes, the periphery’s central

bank finds it optimal to select exactly the same point along its tradeoff under bothregimes.

• A relative size of the periphery as small as .25 is sufficient for the core’s central bank to findMER preferable, even if it is characterized by a less favorable tradeoff.

• The inflation-employment ratio is higher under MER, but employment is considerably morestable and this more than offsets the loss from inflation.

76

Policy Interactions and Exchange Rate Regimes Wrap-Up

• As we mentioned, the results above give us some insights into the functioning of asymmetricER regimes like BW, providing a further reason for the likely collapse of the regime even if ithad survived past 1971.

• At that point, the increased economic size of the regime’s periphery would have made itpreferable for the core (the U.S.) to move to FlER.

• In my article with Giavazzi, we used the results to offer some insight on the state ofEuropean monetary integration in the late 1990s.

• If we think of Europe’s Economic and Monetary Union (EMU) as the core in the model, ourresults suggest that its enlargement may be desirable and that, given the desirability of alarge EMU, having chosen a MER-type regime to govern interactions between insiders andoutsiders may actually prove optimal given the outsiders’ small economic size.

• Of course, our exercise does not address the issues posed by asymmetric shocks acrosscountries and their effect on the desirability of alternative ER arrangements.

• This will be part of our discussion of EMU in the coming lectures.

77

ECON 425




Monetary Policy Commitment Versus Discretion

and the Advantage of Tying One’s Hands

Fabio Ghironi


Introduction

• We saw that the experience of the Gold Standard and then of the Bretton Woods system

was ultimately characterized by the collapse of currency pegs.

• Nevertheless, in the 1970s and beyond, including recent times, policymakers in many

countries found it desirable to pursue various types of monetary regimes characterized by

limiting the flexibility of the exchange rate.

– In 1979, several European countries committed to a regime (the European Monetary

System—EMS) of adjustable parities and target zones relative to a European basket

currency unit (the ECU—European Currency Unit) that quickly evolved into parities and

target zones versus the Deutschemark.

– Many Asian and Latin American countries implemented various types of pegs relative to

the U.S. dollar.

• All these regimes involve giving up monetary policy independence—the more so the tighter

the peg and the higher the degree of international capital mobility.

• But given the long history of peg collapses, why is it the case that countries may still find a

peg desirable?

• The answer has to do with the credibility of institutions and their ability to stay committed to

policies of low and stable inflation.

1

Monetary Expansion in the AS-AD Diagram

• Consider a standard aggregate supply-aggregate demand (AS-AD) diagram, and suppose

the central bank increases nominal money supply from M to M ′.

• The following figure describes the short-run effect of this policy action.

• Assume that, before the change in M , AS intersected AD at point A: Y = Yn (the trend, or

zero-shock, level of output) and P = P e (the price level P was where it was expected to be

when all price and wage contracts in the economy had been signed).

– We are using a diagram with P on the vertical axis, but we could make the same

arguments below with inflation instead of the price level on the vertical axis.

• Think of point A as the pre-shock flexible-price equilibrium of the economy.

• The monetary police expansion causes the AD curve to shift to the right to AD′.

– Output increases to Y ′ and the price level increases to P ′.

2

2

Figure 1 • Over time, adjustment of expectations comes into play: Y > Yn ⇒ P > Pe ⇒ Wage setters revise the expectations incorporated in their wage contracts, and the AS curve shifts up over time. • The economy moves up along AD’. Adjustment stops when Y = Yn again and price level = P’’ = Pe (> initial Pe). In the medium run, AS is given by AS’’, and the economy is at A’’, with Y = Yn and P = P’’. • The next figure shows the entire transition from the short run equilibrium to the medium run:

Monetary Expansion in the AS-AD Diagram, Continued

• Over time, adjustment of expectations comes into play: Y > Yn ⇒ P > P e ⇒ Wage and

price setters that did not get to adjust to the shock immediately revise the expectations

incorporated in their contracts, and the AS curve shifts up.

• The economy moves up along AD′.

• Adjustment stops when Y = Yn again and price level = P ′′ = P e′ (> P e).

3

3

Figure 2 * Note:

- We can pin down the exact size of the eventual increase in P. If Y is back at Yn, M/P must also be back at its initial value. - Then, it must be the case that the proportional increase in P equals the initial proportional increase in M: if M ↑ by 10%, P eventually ↑ by 10%.

The Neutrality of Money in the Medium Run

• In the short run, M ↑ ⇒ Y ↑, i ↓, P ↑. • How much of the effect of the monetary expansion falls initially on Y depends on the slope of the AS curve.

Monetary Expansion in the AS-AD Diagram, Continued

• In the long run, AS is given by AS ′′, and the economy is at A′′, with output back at trend and

the effect of the monetary expansion fully reflected into prices.

– We are implicitly assuming that long-run neutrality of money holds in the diagram, i.e.,

that the economy behaves as if there is a vertical long-run AS curve at Y = Yn.

– This notion has become a subject of discussion since the Global Financial Crisis of

2007-08 and the Great Recession that followed, with scholars arguing that the trend

position of the economy is itself endogenous to the management of aggregate demand

in the short to medium run.

– Many modern models with nominal rigidity, unless modified in specific ways, imply that

monetary policy is not neutral in the long run.

– But the effects would be small under plausible assumptions and it would still be the case

that a welfare-maximizing policymaker acting under commitment would choose a zero

inflation rate because of the costs that inflation would impose on the economy.

4

Expectations

• Now let us think about expectation formation.

• If all agents in the economy of the figure form expectations in a forward looking manner, use

the information at their disposal optimally, and can renegotiate their contracts at the same

time after the initial surprise without incurring additional costs to implementing price and

wage changes, the transition from the short-run equilibrium to the long-run position following

a monetary policy expansion will take just one round of expectation and contract revision.

• Why is that? Because if all agents are forward-looking and they know the structure of the

model, as soon as they observe the policy shock and find themselves in the short-run

equilibrium, they know that the economy must eventually converge to the long-run position.

• As soon as they get a chance to renegotiate contracts based on their revised expectations,

they will immediately set the expected price level at the long-run equilibrium level.

• This will cause the economy to move from the short-run equilibrium to the long-run

equilibrium in just one round of expectation and contract revision, thus shortening the

amount of time during which monetary policy causes output to be above trend.

5

Expectations, Continued

• If all agents are backward or current-looking, expectation revisions reset expectations

(at best) to the currently observed price level, resulting in a transition to the medium run

equilibrium that may take several rounds of expectation and contract revision.

• If forward- and backward-looking agents coexist, as it is plausible, the transition will be faster

than in the fully backward-looking case, but slower than in the fully rational one.

• Now, monetary policy is effective in terms of generating a level of output that differs from

trend to the extent that it generates a price level that differs from the expected price level

embedded in contracts—either because of a full surprise effect at the time of the policy

action or because, even if agents knew that a policy change was coming, something (like

costs of adjusting prices or wages) prevented them from resetting prices fully before the

policy change.

• Unexpected policy has a larger impact because no agent will have had a chance to

renegotiate her/his contract.

• Fully credible, anticipated policy will have no real effect (no effect on output) and will only

affect prices and wages if all agents have a chance adjust prices and wages fully between

the announcement of the policy and the time when the policy change actually happens.

6

Expectations, Continued

• Now suppose that, once Y has returned to Yn, the central bank increases M again.

• The process will be repeated (taking the long-run equilibrium following the previous

monetary expansion as the new initial position):

• Y will rise above Yn for some time but eventually return to Yn, with P increasing to match the

further increase in M over time.

• If the central bank plays this game repeatedly, even if agents are not fully forward-looking

and optimizing to begin with, they will eventually recognize the central bank’s behavior and

incorporate it in their own.

• Expectation and contract revisions will become more and more frequent, so that the

deviations of Y from Yn caused by monetary expansion will become shorter and shorter

lived.

• Eventually the AS curve will become de facto vertical at Yn, and all the policymaker will

accomplish by increasing M will be to increase P immediately.

• Think of this as an environment in which the scale parameter of costs of price adjustment is

becoming smaller and smaller, as frequent, sizable price changes have become the norm.

7

Commitment Versus Discretion

• Now let us use the AS-AD model and our discussion of the role of expectations to talk about

a theory of inflationary consequences of policymaker incentives and agents’ expectations

that was proposed by Finn Kydland and Edward Prescott in 1977 (Journal of Political

Economy ) to explain how high inflation can arise as outcome of the interaction (the “game”)

between the central bank and private agents.

– The theory was then reformulated by Robert Barro and David Gordon (1983, Journal of

Political Economy ) and has become known as the Barro-Gordon model.

– However, Kydland and Prescott were the first to propose it. Kydland and Prescott

received the Nobel Prize in 2004 for their work in macroeconomics.

• Suppose that the policymaker has a target of output Y ∗ above the natural level Yn because

the latter is inefficiently low (for instance, as a consequence of monopoly power or other

possible sources of distortion in how the economy functions).

• Suppose that, when price and wage contracts are being negotiated, the policymaker—who

knows that she/he can only drive Y above Yn for some time—announces that she/he will

follow a rigorous monetary policy aimed at preserving price stability.

• Assume however that the policymaker has no way of actually precommitting herself/himself

in a credibly binding way to implementing the announcement.

8

Commitment Versus Discretion, Continued

• Suppose that private agents have forward-looking, rational expectations.

• Moreover, they know the structure of the economy (the AS-AD model), the fact that

the policymaker has an objective Y ∗ > Yn, and that the policymaker cannot precommit

herself/himself in a binding way to actually implementing her/his announced policy.

• If private agents believe the central bank’s announcement that monetary policy will preserve

a stable price, they will embed this in their expectations and set prices and wages

accordingly.

• Now, once private agents have signed their contracts believing the policymaker’s announce-

ment, it is no longer optimal for her/him to actually implement it.

• She/he has an incentive to expand monetary policy, exploiting nominal rigidity, to drive Y to

Y ∗ at least for some time.

9


• However, private agents know that the policymaker has this incentive.

• They know that, if they sign contracts based on expectations of a stable price, since the

policymaker is not bound by any credible precommitment device to implementing her/his

announced rigorous policy, she/he will fool them with a monetary expansion that erodes

markups and real wages.

• Rational, optimizing private agents recognize this, do not believe the policymaker’s

announcement, and incorporate her/his expected behavior in their contract negotiation,

setting price expectations at the level that corresponds to the long-run equilibrium to which

the economy would have eventually converged if the policymaker had actually managed to

fool private agents for some time.

• Optimal behavior by rational private agents will thus cause the AS curve to shift up even

before the policymaker actually takes any action.

10


• At this point, if the policymaker sticks to her/his announcement of stable monetary policy,

the short run equilibrium will be not only below Y ∗, the policymaker’s target, but also below

Yn!

• What is the optimal thing to do for the policymaker if she/he wants at least to avoid having

Y < Yn for some time?

• It is to validate agents’ expectations and do what they expected her/him to do in the first

place: expand monetary policy, shift the AD curve to the right, and drive the equilibrium

where agents expected it to be: at Yn, but with a higher price level.

– If the policymaker does not validate agents’ expectations and surprises them by sticking

to her/his prior announcement, Y will be below Yn for as long as it takes for expectations

and contracts to be revised and the AS curve to return to its original position.

– This is how the central banker would establish her/his credibility for being “tough”: at a

potentially significant cost for the economy.

• Given our assumptions, high-price expectation and monetary policy expansion are the best

responses of the private sector and the central bank, respectively, to each other’s behavior.

11


• Thus, as a consequence of the inability of the policymaker to precommit to a stable-price

policy, her/his incentive to generate Y above Yn, and private agents’ recognition of the

policymaker’s lack of commitment and incentives, all that we see is no deviation of Y from

Yn and a higher price level.

• This is the Nash equilibrium of the game between the central bank and the private sector.

• Monetary expansion is the only possible equilibrium policy in this game.

• Inability to commit credibly to the stable-price policy—i.e., the fact that policy is conducted

under discretion—results in a high-price-level (or high-inflation) outcome for the economy.

12

The Time Inconsistency of Optimal Policy

• Central to the theory is the concept of time inconsistency of the optimal policy:

– Ex ante, when private sector expectations embedded in wage and price contracts are

being formed, the rigorous, stable-price policy is optimal.

– Announcing anything different would simply result in an immediate change in price

expectations.

– Ex post, once private agents have committed to contracts based on believing the

policymaker’s announcement of a stable-price policy, the policy is no longer optimal.

– It is optimal for the policymaker to fool agents with a monetary expansion.

• Note that the policymaker is benevolent:

– If effective on the real economy, the monetary expansion increases Y above the

inefficient level Yn for some time.

– But the agents’ recognition of the policymaker’s incentive and their desire to protect

themselves from inflation prevent the policymaker from accomplishing this goal.

13

The Time Inconsistency of Optimal Policy, Continued

• The ex ante optimal policy is thus time inconsistent.

• The time consistent policy (expected by rational agents and implemented in equilibrium) is

to expand monetary policy, which ends up only resulting in high prices without any output

gain.

14

Help from Other Policies

• The problem would be removed if other policies (fiscal policy and regulatory policy) acted

on the distortions in the economy that cause Yn to be suboptimally low.

• In this case the central bank would not face the temptation to expand and the outcome

under discretion would not be different from that under commitment.

15

A Repeated Interaction

• The fact that the interaction between the central bank and the private sector is repeated

over time rather than just played once in a one-shot game such as the one described above

provides a possible mechanism that would support a stable-price equilibrium in the absence

of “help” from other policies.

• Suppose that the game between the policymaker and the private sector is played each

period over an infinite horizon.

• Suppose that the private sector “tells” the policymaker today: “Ok, we believe your stable-

price announcement. But if you cheat on us later on, we will never believe you again, and

we will always set expectations consistent with the one-shot-game, monetary expansion

behavior.”

• In this case, when deciding what to do, the policymaker is trading off the gains from surprising

the private sector in the short run against the losses from high price expectations (and high

actual price since, as we discussed, it will be optimal to validate those expectations) for the

infinite future.

• If the policymaker cares enough about the future, she/he will stick to the announced policy

of monetary rigor.

16

A Repeated Interaction, Continued

• The problem with this mechanism is that policymakers definitely do not have an infinite

horizon.

– Note that the fact that central bankers are generally appointed for long periods is another

way to strengthen their independence, and thus their credibility.

• If the game is played a finite number of times, the mechanism that sustains the stable-price

policy in the infinite horizon case unravels, and the only possible policy equilibrium is the

monetary expansion of the one-shot game.

• Why?

• Because, in the last period of the game (call it period T ), the interaction reverts to the

single-period interaction in which the policymaker has a clear incentive to cheat on the

announcement of rigorous policy (since there is no future game to be played next period

involving that policymaker).

• Thus, monetary expansion is the only equilibrium policy in period T .

17

A Repeated Interaction, Continued

• But in period T − 1, agents know that this will happen in period T , i.e., they know that the

policymaker will expand monetary policy in period T .

• They will respond to this by setting high price expectations already at T − 1, which the

policymaker will find it optimal to validate to avoid a recession then.

• The same mechanism will apply at T − 2, T − 3, and so on.

• By backward induction, the only equilibrium policy outcome in all periods from T back to the

current time will be monetary expansion and a high price level.

18

Commitment Mechanisms

• Bottom line: The ability (or inability) of the policymaker to precommit credibly to a course of

policy is crucial for observed outcomes.

– Think of Ulysses and the sirens: It was only through the precommitment devices of filling

his sailors’ ears with wax and having himself tied to the mast of the ship that Ulysses was

able to escape the sirens.

• Policies that act directly on Yn (for instance, enforcement of anti-trust legislation or

appropriate tax setting) are better suited than monetary policy to resolve the problem of an

inefficiently low Yn.

• Absent such policies, precommitment mechanisms that bind policymakers to implement

announced policies are a remedy for the inflationary consequences of time inconsistency,

by giving the policymaker a way to “resist the sirens” of an output target above Yn.

• The theory proposed by Kydland and Prescott and Barro and Gordon has been used

to explain several high inflation episodes across countries and over time due to lack of

precommitment, including high inflation in the U.S. in the 1970s.

• The argument fit stylized facts in several countries, and it has been (and still is) central to the

focus of monetary policy and the design of monetary institutions to establish and preserve

anti-inflationary credibility.

19

Commitment Mechanisms, Continued

Inflation Targeting

• A commitment mechanism that several countries have implemented beginning in the early

1990s is the adoption of an inflation targeting regime that clearly specifies the central bank’s

target and responsibilities with respect to inflation and puts an institutional setup in place to

ensure central bank independence in the pursuit of this target and accountability in case of

failure.

• Inflation targeting regimes are generally characterized by much transparency in commu-

nication between the central bank and the public to help the establishment of necessary

credibility.

• Several countries are operating variants of such regime, including industrial countries and

emerging markets.

20


• As Michael McLeay and Silvana Tenreyro have argued in an NBER Macroeconomics Annual

2019 article on “Optimal Inflation and the Identification of the Phillips Curve,” the pursuit

of inflation targeting by central banks over a number of years may be responsible for the

inability to easily identify the Phillips Curve (or its New Keynesian variant) in the data that

has led some to argue that it is no longer a relevant concept.

• The Phillips Curve relation may be alive and well as part of the mechanisms of the economy.

• But if the central bank has successfully kept inflation low and stable over a number of years,

the data will not show it “transparently.”

21


• Now, some view the failure of central banks like the Federal Reserve or the European

Central Bank to accomplish their 2 percent inflation targets consistently since the Global

Financial Crisis of 2007-08 as an indictment of the failure of inflation targeting.

• Remembering the times when inflation in the U.S. was well into double digits and, in Italy, it

was above 20 percent, I was (and still am) much more lenient than these colleagues toward

an inflation rate stubbornly at, say, 1.7 percent when the target is 2 percent.

• Central banks have had to contend with a variety of pressures toward low-flation or even

de-flation since 2007-08 (and some, like the Bank of Japan, since the early 1990s).

• The impact of the current COVID-19 crisis on inflation remains to be seen, but I certainly

expect lively debates on the topic among policymakers and academics.

22


Exchange Rate Pegs

• Limiting exchange rate flexibility is another possible precommitment device—provided the

commitment to pegging the currency is credible—when the domestic institutional setup does

not support a country’s independent, stable policies.

• The idea is that a strategy of pegging the currency is easier to communicate transparently

to the public, contributing to credibility that would be harder and take longer to establish in

absence of the appropriate domestic institutional setup.

• For many countries, the strategy of adopting a fixed (or semi-fixed) exchange rate that tied

their monetary policies to those of established, low-inflationary central banks was indeed

the precommitment device that would otherwise have been missing.

• By pegging the exchange rate, a country can “import” the low inflation policy of the country

to which it pegs.

• The center country’s monetary policy may not be the best over the business cycle from the

perspective of the pegger (because the center country’s policymakers do not take decisions

with the pegger’s economy in mind), but the cost of giving up independent stabilization of

the cycle is compensated by the benefit of accomplishing a lower, more stable inflation on

average.

23


• This argument is the core of Francesco Giavazzi and Marco Pagano’s (European Economic

Review, 1988) analysis of the European Monetary System (EMS) in place between 1979

and the advent of the euro:

– European central banks that could not implement a credible commitment to low inflation

policies of their own found it better to commit to shadowing the Bundesbank through the

EMS constraint rather than finding themselves in the high-inflation, discretionary-policy

outcome often generated by the inflationary pressures from their governments.

• The desirability of the strategy to lower inflation by importing credibility through a peg in fact

had its roots in institutional setups that made it hard for central banks to keep inflation low

and stable.

24


• For instance, by law, the Bank of Italy had to purchase all bonds issued by the government

that went unsold on the market until laws were adjusted in 1981.

• Monetary financing of deficits by spending-prone governments was embedded in laws until

then, making it de facto impossible for the Bank of Italy to pursue low-inflation policies

without the “protection” for this objective that the EMS created.

• Even after the 1981 “divorce” between the Bank of Italy and the government, the EMS

contributed to importing credibility for low inflation that would have taken much longer to

establish independently.

• In sum, a commitment device in the form of an explicit inflation targeting regime (if credible

institutions for it are in place) or an exchange rate peg can be key to delivering price stability.

– Nominal GDP targeting and price-level targeting have also been receiving attention in

debates.

25

Conclusion: Policy Rules and Institutions

• Inflation targeting and exchange rate pegs are examples of policy rules.

• Kydland and Prescott’s and Barro and Gordon’s work started a huge literature on the

advantages of rules versus discretion.

• The policymaker we looked at in our discussion of the time inconsistency problem operates

under discretion.

• Unless there is a commitment device that binds policy to the announced one, the policymaker

is free to reoptimize her/his behavior at each point in time.

• The fact that what is optimal ex ante is no longer optimal ex post and the private sector’s

forward-looking behavior then result in the unfavorable equilibrium with monetary expansion

and no output gain.

• Thus, it would be better if the policymaker could commit to a rule that forces her/him to

implement the ex ante optimal policy by removing discretion.

26

Conclusion: Policy Rules and Institutions, Continued

• As we mentioned, inflation targeting and fixed exchange rate regimes are examples of policy

rules to which the policymaker can precommit in order to establish independent domestic

anti-inflationary credibility (under a properly designed inflation targeting regime) or to import

the monetary policy credibility of the center country (under an exchange rate peg).

• The so-called Taylor rule (a rule for interest rate setting in response to inflation and

output movement first studied by John Taylor in a 1993 article in the Carnegie-Rochester

Conference Series on Public Policy ) is another example of a rule to which the policymaker

could be committed in some binding form.

– Note an important difference: Inflation targeting or an exchange rate peg are targeting

rules (they are rules about policy targets), the Taylor rule is an instrument rule (it is a rule

for the setting of a policy instrument).

• Even if some macroeconomists went as far as suggesting that, say, the Federal Reserve

should be committed to automatically implementing the Taylor rule in each period, Taylor

himself in his 1993 article interprets it more as a benchmark guideline for policymaking in

normal conditions, from which departures should be allowed in special circumstances.

27


• He views the rule as constrained discretion, whereby the policymaker operates under

discretion, but subject to the constraint of a benchmark guideline for policymaking in normal

times.

• Clearly, such constrained discretion is feasible for institutions that do not lack credibility in

the pursuit of a stable monetary environment.

• It is much less feasible for central banks that lack credibility, for which a more binding

commitment can be the best way to bring inflation under control.

28


• In the end, policy rules are just examples of institutions broadly defined, and the discussion

in these slides highlights the importance of institutional design.

• The work by the scholars we mentioned—and by Alberto Alesina (Harvard University),

Allan Drazen (University of Maryland), Torsten Persson (Institute for International Economic

Studies, Stockholm), and Guido Tabellini (Bocconi University, Milan), and many more over

the years—was crucial for us to understand the importance of central bank independence

and credibility, the importance of how institutional structures shape economic outcomes,

and the interaction of political incentives and macroeconomic dynamics.

• Large crises—such as the Global Financial Crisis of 2007-08 or the current crisis created by

COVID-19—are moments in which any simple rule designed for policymaking under normal

economic conditions must be abandoned.

• But these are also the moments in which the credibility of institutions in charge of

policymaking becomes even more crucially important, and with it the institutional setup that

helped establish it.

29

ECON 425




Europe

Fabio Ghironi


Introduction

• The euro is part of a broader agenda of European integration that began shortly after WorldWar II.

• The recent European crisis and the responses of European policymakers to the events ofthe last few years cannot be understood without placing the recent experience in thelonger-run context of European integration.

• I will use the first part of this presentation for this purpose, focusing on the efforts ofEuropean countries to stabilize their exchange rates (ERs).

1

Limiting Exchange Rate Flexibility

• In 1979, Western European countries created the European Monetary System (EMS) tostabilize ERs vis-a-vis each other.

– These countries sought to avoid trade disruptions associated with “competitive deval-uations,” and the strains that ER fluctuations would put on the European EconomicCommunity’s Common Agricultural Policy.

• The EMS featured bands of fluctuation around parities that could be realigned (in principle,in cooperative fashion) in the presence of persistent imbalances.

• Germany emerged as the center of the system, with other countries de facto pegging to thedeutsche mark (D-mark) and, therefore, shadowing the Bundesbank’s monetary policy.

2

The Advantage of Tying One’s Hands

• Limiting ER flexibility imposes a constraint on policy, as it becomes more difficult (oraltogether impossible if the ER is completely fixed) to use monetary policy to address theconsequences of asymmetric shocks.

• What made participation in the EMS optimal from the perspective of central banks otherthan the Bundesbank?

• Francesco Giavazzi and Marco Pagano (European Economic Review, 1988) highlight thevalue of the EMS as a policy commitment device.

– Countries other than Germany that had problems with independent implementation oflow inflation policies were able to commit to such policies by pegging to the D-mark andbeing forced to shadow the Bundesbank.

3

The Advantage of Tying One’s Hands, Continued

• In related work, Carlo Carraro and Francesco Giavazzi (1989) highlight the role ofinteractions between monetary and fiscal policymakers in weak currency countries.– In the 1980s the Bank of Italy would have often liked to implement monetary contraction

to lower inflation in Italy.– However, governments were running large deficits over the years and, by law, the Bank

of Italy had to buy all government bonds issued to finance those deficits that would notbe bought by markets.

– Clear lack of central bank independence implied pressure for monetary finance ofgovernment deficits.

– By being in the EMS, the Bank of Italy was able to play the following game:· Government goes to Bank of Italy and says: “I need you to print more money to buy

these bonds”· Bank of Italy replies: “I’d gladly do that, but it would run against my EMS obligations

to shadow the Bundesbank, and, if I don’t do that, I am running against the Europeanintegration agenda that you care so much about. So, there is a limit to how much I canhelp you.”

– Thus, the Bank of Italy was able to use the EMS as a shield to protect the anti-inflationarypolicy that it would have liked to do but could not have done due to lack of independence.(The “divorce” of the Bank of Italy from the government was achieved only at thebeginning of the 1990s.)

4

-5

0

5

10

15

20

25

France Germany Italy UK

CPI Inflation (% per year, Quarterly Data, SA)

The EMS Experience

• Central parities were modified on average once every eight months in the first four years ofthe EMS.

• Over the next four years, through January 1987, the frequency of realignments declined toonce every twelve months.

– This reflected, among other things, gradual relaxation of capital controls, which madeorderly realignments more difficult to execute.

5

400

500

600

700

800

900

1000

1100

1200

1300

0

0.5

1

1.5

2

2.5

3

3.5

4

French franc Pound sterling US dollar Italian lira (right scale)

Exchange Rates vs. German mark (Daily Data)

400

500

600

700

800

900

1000

1100

1200

1300

0

0.5

1

1.5

2

2.5

3

3.5

4

French franc Pound sterling US dollar Italian lira (right scale)

Exchange Rates vs. German mark (Monthly Data)

The Single European Act

• The Single Market Program, embodied in the Single European Act of 1986, sought to createa single European market in goods and factors of production.

• Eliminating currency conversion costs was viewed as key to a truly integrated market.

• Moreover, removing the possibility for countries to manipulate ERs was deemed necessaryto prevent opposition to trade liberalization.

• These arguments found institutional expression in the Maastricht Treaty adopted by theEuropean Council in December 1991.

6

The Removal of Capital Controls and the “Hard EMS”

• Integral to the creation of the Single Market was the removal of capital controls.

• But the elimination of controls made periodic parity realignments difficult to implement.

• After the beginning of 1987 there were no more parity realignments: “hard EMS.”

7

On the Way to Crisis in the Early 1990s

• Starting in 1990, a series of shocks happened.

– Global recession (associated with the First Gulf War) elevated unemployment rates inEurope;

– Dollar depreciation undermined European competitiveness;

– German unification (1989) raised interest rates throughout the European Community asthe Bundesbank tightened monetary policy in response to the German government’sfiscal expansion and the forced one-to-one conversion of Ostmarks into D-marks.

• National political leaders began to question the Maastricht blueprint.

• Markets began to question the commitment of political leaders to the defense of EMS pegs.

• Pressure mounted, and the EMS collapsed in 1992-93.

8

Understanding the Crisis

• German reunification in 1990 caused the Bundesbank to contract its policy aggressively asreunification put pressure on inflation in Germany relative to the rest of Europe (because ofa combination of fiscal expansion and the one-to-one conversion of Ostmarks into D-marks,which the Bundesbank had opposed).

• The EMS constraint required other central banks to shadow the Bundesbank, but this wasbecoming increasingly costly given economic conditions in the respective countries.

– For instance, Italy was undergoing fiscal contraction in an effort to satisfy the Maastrichtconvergence criteria for entry into Economic and Monetary Union (EMU).

– Adding further monetary contraction would have sent the country into more costlyrecession than it was already facing as a consequence of fiscal consolidation and theglobal recession of 1990-91.

• Absent a speculative attack, the political will to satisfy the Maastricht criteria may have beenenough to keep following the Bundesbank even at increased cost, but the probability thatthis was not the case increased enough that markets found it optimal to test the credibility ofthis commitment.

9

Understanding the Crisis, Continued

• Markets thus launched a speculative attack, selling the currencies of countries whosecommitment they decided to challenge (for instance, Italian lira and British pound) inexchange for D-marks in expectation that the D-mark would appreciate.

• To defend the ER, Bank of Italy and Bank of England (and other central banks under attack)had to sell D-marks and buy liras and pounds.

• The obvious problem facing these central banks is that they did not have access to unlimitedreserves of D-marks.

• Under EMS rules, the Bundesbank should have opened unlimited, short-term D-markcredit lines to the Bank of Italy, the Bank of England, and other central banks under attack,intervening on their behalf by injecting D-marks in currency markets.

• The Bundesbank indeed did this, but eventually withdrew from this arrangement.

– When the EMS was set up, the Bundesbank had negotiated a document (known as theEmminger Letter) that allowed it to opt out of its intervention obligations if it was perceivedthat the implied increased supply of D-marks would caused inflation in Germany.

– The Bundesbank, concerned about inflation, invoked the Emminger Letter for the firsttime in this occasion.

10

Understanding the Crisis, Continued

• The speculative attack succeeded.

• Bank of Italy and Bank of England ran out of D-mark reserves, and lira and pound wereforced to devalue and float.

11

The Crisis and the Incentives for EMU

• Pound and lira lost value between 1992 and 1993.

• But this ended up boosting the competitiveness of exporters in Italy and the UK when sellingoutput in Germany and France.

• Producers in these countries complained loudly about the competitive advantage of Italianand British producers and asked for trade measures to be introduced to compensate that.

• Ultimately, the trade measures—which would have run squarely against the Single Marketproject—were not introduced because it could not be argued that the weakening of lira andpound was intentionally driven by policy.

– It was driven by markets after Bank of Italy and Bank of England had done all they couldto defend the ER.

• But it was also to remove the risk that policy-induced ER movements could eventuallydestabilize the Single Market that countries decided to remove the ER option altogether byrenewing their commitment to adopting a single currency: the euro.

12

Meeting the (Fiscal) Convergence Criteria

• At German insistence, the Maastricht Treaty had set targets for inflation, interest rates, ERstability and fiscal stability for countries seeking to qualify for participation in the monetaryunion.

• The fiscal criteria—a budget deficit of no more than 3 per cent of GDP and a public debt ofnot more than 60 per cent—were key.

• The expectation was that these criteria would effectively filter out countries lacking therequisite fiscal stability culture.

• They would bar from participation countries inclined to press for loose monetary policy tomake it easier to finance their budget deficits.

13

Meeting the (Fiscal) Convergence Criteria, Continued

• But these fiscal convergence criteria proved a less effective bar than its German architectshad hoped.

• Faster growth after 1993, which augmented public-sector revenues, meant that deficitsdeclined even in the absence of policy initiatives.

• Governments, including Italy’s, could take one-off measures—typically in the form ofadditional taxes—to temporarily squeeze under the 3 per cent bar.

• Some, including Italy’s, resorted to accounting gimmicks.

• Thus, all EU members that aspired to participate in the monetary union when it began in1999 could claim that they satisfied the deficit criterion, aside only from Greece.

14

-12

-10

-8

-6

-4

-2

0

2

4

Germany France Italy United Kingdom

Government Deficit (% of GDP, Annual Data)

A Large EMU

• Moreover, the fact that important decisions were made by consensus made it difficult to barmember states in a dubious position.

• Since important decisions in the political process of the European Union (EU) require theunanimous consent of members, countries left out of the monetary union could threaten toretaliate by obstructing progress in other areas.

• When the Maastricht Treaty was signed, the expectation had been of a small monetary unioncentered on France and Germany and including perhaps Austria, Belgium, Luxembourg,and the Netherlands.

• The decision, taken at the May 1998 Economic Council in Brussels, was instead for a largemonetary union that included also Ireland, Italy, Spain, Portugal and Finland.

• Exchange rates between these countries were irrevocably locked as of January 1, 1999,and euro notes and coins began circulating on January 1, 2002.

15

Large and Getting Larger

• Greece joined on January 1, 2001.

• Since then, Slovenia, Cyprus, Malta, Slovakia, and Estonia have joined.

• Some EU countries had negotiated opt-out clauses: UK, Denmark.

• All other EU members are supposed to join EMU when they meet the required convergencecriteria.

– Some countries strategize over these criteria to remain outside (for instance, Sweden).

16

Interest-Rate Convergence

• The first 10 years of the euro and the European Central Bank (ECB) were generally viewedas a success.

• Interest rates on same-maturity bonds issued by different EMU countries converged, andspreads relative to the German Bund were small until 2008.

– For nearly a decade, markets treated bonds issued by different EMU countries as nearlyperfect substitutes, resulting in what was perceived as a major success of EMU.

17

0

1

2

3

4

5

6

7

8

9

France Italy United Kingdom

Interest Rate Spread vs German 10-year bund (%, Monthly Data)

But Not All Euro-Denominated Bonds Are Created Equal

• Spreads between PIIGS (Portugal, Ireland, Italy, Greece, Spain) and Germany widened in2009 and again, increasingly, in 2010-2011.

• The spread on Greek bonds continued to increase notwithstanding EU-IMF rescue plans.

– It remains large, notwithstanding the reduction due to the implementation of austeritymeasures and rescue actions.

• Spreads on Irish and Portuguese bonds rose similarly in 2010-2011, with Spanish andItalian bonds coming under increasing pressure next.

– All these countries were forced to implement various forms of austerity policies, withvarying success.

• Note: Data in figures from OECD unless noted. Tables from Giavazzi and Spaventa (2011).

18

-2

0

2

4

6

8

10

12

14

16

18

20

22

24

26

28

1999 2001 2003 2005 2007 2009 2011 2013

Portugal Ireland Italy Greece Spain

Interest Rate Spread vs. German 10-Yr Bund

(Percent per Year, Monthly Data)

0

5

10

15

20

25

30

35

2008 2009 2010 2011 2012 2013

Portugal Ireland Italy Greece Spain

Interest Rate Spread vs. German 10-Yr Bund

(Percent per Year, Daily Data, Eurostat)

The Current Account Matters... Even in a Monetary Union

• PIIGS have been running persistent current account deficits—of different sizes—in contrastto Germany’s persistent surpluses, resulting in large foreign debt positions.

• Giavazzi and Spaventa (2011): “Conventional wisdom” downplayed the importance of theexternal balance of EMU members in both academic and policy debates.

– No mention in the Maastricht convergence criteria or European Commission’s evaluationof members’ performance.

– Little concern by the ECB (until recently).

– Treaty on the Functioning of the EU (Treaty of Lisbon, 2007): Only member states thathave not adopted the euro can receive assistance for BOP problems.

• Perception: MU blurs the concept of external surplus or deficit.

– After all, we do not hear much about external (cross-state) positions of individual statesin the U.S.

• Problem: CA matters—even in a MU—and markets have been punishing unsustainable CAs(among other things).

19

-20

-15

-10

-5

0

5

10

Germany Portugal Ireland Italy Greece Spain

Current Account: Germany vs. PIIGS

(Percent of GDP, Quarterly Data, SA)

Table 1 - General Government Balance and Debt, % of GDP

Balance Debt Average 2008 2009 2008 2009

2000-2007

Euroarea -2,3 -2 -6,3 69,7 79 Ireland -1,0 -7,3 -14,3 43,9 64 Greece -6,1 -7,7 -13,6 99,2 115,1 Spain -1,3 -4,1 -11,2 39,7 53,2 Portugal -4,1 -3,7 -7,1 66,3 76,8 Italy -3,1 -2,7 -5,3 106,1 115,8

Source: Eurostat

Table 2 - Cumulated current accounts - 1999-2008, % of GDP

Ireland -19,2 Germany 31,5 Spain -59 Netherlands 53,7 Greece -85,1 Finland 59,1 Italy -13 France 3,1 Portugal -90,7 Euroarea 22,2

Source: Eurostat

Competitiveness

• Some attention was paid to dynamics of competitive positions of key EMU members, whichcontributed to their external positions.

20

-8

-6

-4

-2

0

2

4

6

8

Q1-1999 Q1-2001 Q1-2003 Q1-2005 Q1-2007 Q1-2009 Q1-2011 Q1-2013


CPI: Germany vs. PIIGS

(YoY Percent Change, Quarterly Data, SA)

60

70

80

90

100

110

120

130

140

150

160


Unit Labor Cost: Germany vs. PIIGS

(Index: 2005=100, Quarterly Data, SA)

Growth and Capital Flows

• Capital should flow to faster growing countries.

– If this were happening, we should not worry about CA deficits.

• If we look at the growth performance of PIIGS vs. Germany and the determinants of growth,we can begin to understand the role of the CA and its determinants (and “uses”) in EMU’scrisis.

21

-8

-6

-4

-2

0

2

4

6

8

10

Q1-1999 Q1-2001 Q1-2003 Q1-2005 Q1-2007 Q1-2009 Q1-2011


Real GDP Growth: Germany vs. PIIGS

(YoY Percent, Quarterly Data, SA)

Convergence?

• Not really.

• GDP per capita:

– Convergence for Greece and Spain;– Portugal barely moved;– Ireland already above average in 1998;– Italy? Downward divergence...

• Relative labor productivity displays consistent picture:

– Upward in Greece, flat in Portugal, downward in Italy;– Ireland? Slower than per capita GDP;– Spain? GDP catch-up with nearly no productivity gain!

22

Table 3 - Per capita income and labour productivity

GDP per capita Labour productivity Labour productivity per person employed per hour worked

(a) (b) (a) (b) (a) (b) Ireland

1998 106,1 99,2 108,1 111,4 2000 94,9 89,0 2008 123,8 116,4 118,7 121,7 104,2 94,7

Greece 1998 72,8 68,0 78,4 80,7 2000 64,2 60,2 2008 86,2 81 93,2 95,5 71 64,5

Spain 1998 83,3 77,9 92,9 95,7 2000 87,2 81,7 2008 94,5 88,8 94,5 96,9 92,4 84

Portugal 1998 69,3 64,8 60,4 62,3 2000 52,9 49,6 2008 71,6 67,2 67,1 68,8 56,2 51,1

Italy 1998 105,3 98,4 112,2 115,6 2000 98,5 92,3 2008 93,6 87,9 99,8 102,3 88,8 80,8

(a) Euro-area =100

(b) Germany = 100

Source: Eurostat

Deteriorating Government Budgets

• To different degrees, the deteriorating external position and resulting crisis are explained bydeteriorating government budgets.

• This is certainly the case of Greece, which came under massive attack when it becameknown that budget numbers had been “cooked.”

• The situation of government finances is also key for Italy, which accumulated relatively smallCA deficits, but has a public debt-GDP ratio near 120%.

– Lack of growth contributes to the gravity of the Italian case.

23

-35

-30

-25

-20

-15

-10

-5

0

5

1999 2001 2003 2005 2007 2009 2011 2013 2015


Government Deficit: Germany vs. PIIGS

(Percent of GDP, Annual Data)

Table 1 - General Government Balance and Debt, % of GDP

Balance Debt Average 2008 2009 2008 2009

2000-2007

Euroarea -2,3 -2 -6,3 69,7 79 Ireland -1,0 -7,3 -14,3 43,9 64 Greece -6,1 -7,7 -13,6 99,2 115,1 Spain -1,3 -4,1 -11,2 39,7 53,2 Portugal -4,1 -3,7 -7,1 66,3 76,8 Italy -3,1 -2,7 -5,3 106,1 115,8

Source: Eurostat

Table 2 - Cumulated current accounts - 1999-2008, % of GDP

Ireland -19,2 Germany 31,5 Spain -59 Netherlands 53,7 Greece -85,1 Finland 59,1 Italy -13 France 3,1 Portugal -90,7 Euroarea 22,2

Source: Eurostat

Capital Inflow and Resource Allocation

• Ireland and Spain—and, to a lesser extent, Portugal—are prime examples of a problematicallocation of borrowed resources.

• Their external borrowing went largely into financing construction booms fueled by risinghouse prices.

• Put differently, the current account deficits were used to a significant extent for expansion ofthe non-traded sector relative to the traded sector.

24

0

2

4

6

8

10

12

14

16

1998 2000 2002 2004 2006 2008 2010

Germany Portugal Ireland Italy Greece Spain Euro area

Value added of construction as % of total value added

(Germany vs. PIIGS, annual data, Eurostat)

Capital Inflow, Resource Allocation, and Market Response

• Intertemporal sustainability of a country's net foreign debt requires that current external deficits should be matched by the expectation of future surpluses.

• But it is hard to generate the required future surpluses if borrowed resources are allocatedto expansion of the non-traded sector and consumption, as was the case in Ireland, Spain,and, to some extent, Portugal, where domestic credit and household debt burgeoned.

• Markets understood the consequences of resource allocation for intertemporal budgetconstraints and are responding accordingly.

– There would be no problem in a “frictionless” economy, but that is not how the real worldoperates.

• The current account matters...even in a MU!

25

EMU’s Crisis and the Euro’s Exchange Rates

• Not surprisingly, pressure on euro-denominated bonds was associated with weakening of the euro vs. U.S. dollar and other currencies.

• Silver lining? Improved competitiveness vs. non-euro trading partners (but much of PIIGS’trade is within EMU).

26

Politics

• Politics dictated outcomes and will continue to do so.

• The conditions of EU/ECB and IMF rescue plans for PIIGS put enormous pressure on

policymakers.

• Austerity became very costly. It went too far in some cases.

• In Italy, the crisis led in 2011 to the creation of an interim, so-called “technical government,”

charged with the tasks of improving government finances (most of which was accomplished

by raising taxes) and implementing market reforms (very little of which was accomplished).

27

Options Then and Now

• Leaving the euro: Very complicated to do. Potentially very bad idea.

– Low credibility, high inflation, inability to borrow in own currency.

– No solution to underlying problems: Devaluations would be “shots in the arm,” but countries would be back to “square 1” relatively quickly if the root causes of weakness (excessive dependence on government sector, sclerotic economies) are not addressed.

• Europe “muddled through,” largely thanks to the resolve of Mario Draghi and the ECB.

Progress was made toward banking union and true capital markets union, but not there yet.

Mechanisms were put in place to prevent local crisis situations from threatening the stability of the entire EMU and to support crisis countries (subject to their making policy adjustments and implementing reforms).

Signs of significant progress in other key dimensions of integration (such as fiscal policy cooperation) are now happening in response to the COVID-19 crisis, but there is still a lot to do before the long-run survival of EMU is ensured.

28

What Do Weak Countries Really Need? The Case of Italy

• Reforms facilitating entrepreneurial activity, enhancing competition, increasing the flexibility of labor markets, reducing the burden of bureaucracy, improving the performance of the judiciary...

• Former ECB Presidents Jean-Claude Trichet and Mario Draghi, current President Christine Lagarde, the European Commission, and many influential economists have been calling for such reforms (so-called structural reforms) since the crisis started (and before).

• The consensus among academics is that more flexible markets improve economic

performance.

• Will reforms happen? Some steps did (example, the so-called Jobs Act in Italy), but more isneeded.

• Next are some figures from the World Bank’s Doing Business database that highlight how

badly Italy needed reforms (data from 2013, improvement has happened since, but a lot more is needed).

29

7 ItalyDoing Business 2013

THE BUSINESS ENVIRONMENT

For policy makers, knowing where their economy

stands in the aggregate ranking on the ease of

doing business is useful. Also useful is to know how

it ranks relative to comparator economies and

relative to the regional average (figure 1.2). The

economy‘s rankings on the topics included in the

ease of doing business index provide another

perspective (figure 1.3).

Figure 1.2 How Italy and comparator economies rank on the ease of doing business

Source: Doing Business database.

8 ItalyDoing Business 2013

THE BUSINESS ENVIRONMENT

Figure 1.3 How Italy ranks on Doing Business topics


16ItalyDoing Business 2013

STARTING A BUSINESS

Globally, Italy stands at 84 in the ranking of 185

economies on the ease of starting a business (figure

2.2). The rankings for comparator economies and the

regional average ranking provide other useful

information for assessing how easy it is for an

entrepreneur in Italy to start a business.

Figure 2.2 How Italy and comparator economies rank on the ease of starting a business



STARTING A BUSINESS

Equally helpful may be the benchmarks provided by

the economies that over time have had the best

performance regionally or globally on the procedures,

time, cost or paid-in minimum capital required to start

a business (figure 2.3). These benchmarks help show

what is possible in making it easier to start a business.

And changes in regional averages can show where

Italy is keeping up—and where it is falling behind.

Figure 2.3 Has starting a business become easier over time?

Procedures (number)

Time (days)


STARTING A BUSINESS

Cost (% of income per capita)

Paid-in minimum capital (% of income per capita)

Note: Ninety-one economies globally have no paid-in minimum capital requirement.


Market Reforms and AD Policy

• Reforms may imply short-term costs, and appropriate

management of aggregate-demand-side policies may be required during the transition.

• This point was made in IMF documents in 2012 and beyond.

• The appropriate management of ECB policy in response to market reforms is the topic of

a paper that Matteo Cacciatore, Giuseppe Fiori, and I published in the JIE in 2016.

We developed a broad research agenda on reforms of product and labor markets.

It provided the background to the analysis in the IMF's April 2016 World Economic Outlook andto much advice the IMF has been giving since.

30

ECON 425




Global Financial Crisis and Response

Fabio Ghironi


Introduction

• From the late 1990s, international economic integration resulted in global imbalances on a scale never witnessed previously in modern international monetary history.

• These imbalances became one of the most important topics of academic and policy discussions in international macroeconomics in the late 2000s.

• After the onset of the ongoing global financial crisis in 2007-08, the focus shifted toward understanding the crisis and responding to it, but imbalances and the crisis were tightly connected.

• These slides discuss the origin of the global imbalances of the 2000s, their contribution to the crisis of 2007-08, and the response to it.

The first part of the slides reproduces material from Eichengreen's (2019) Globalizing Capital.

1

China’s Growth and Savings

• China, which had been largely unscathed by the Asian crisis, grew at a very fast pace fora number of years before the current crisis on the back of investment in excess of 40% ofGDP.

• Chinese saving exceeded even these high levels of Chinese investment.

• Saving by households alone approached 25% of GDP.

– This was entirely consistent with the life-cycle model of savings behavior.

– That model emphasizes the incentive for those of working age to save for retirement.

– It observes that net saving by households will be the difference between saving by theyoung and dissaving by the old.

– In an economy like China’s, which has sustained a growth rate of 10% per year, theincomes of current labor force participants will be a multiple of those once earned by theelderly.

– Hence, saving by the young will be significantly higher than dissaving by the old.

2

China’s Growth and Savings

• A study by Shang-Jin Wei and Xiaobo Zhang (“Sex Ratios and Savings Rates: Evidence from ‘Excess Men’ in China,” 2011, JPE) suggests that the growing men/women ratio in the Chinese population contributes to high household savings.

– Men and/or men’s parents increase their savings as accumulated wealth makes the menmore competitive on the “marriage market.”

– Wei and Zhang show that the effect of this behavior on aggregate household savingis robust to controlling for possible downward adjustment in saving by women and/orparents of women (as their “competitive” position is enhanced by relative scarcity).

– In fact, increased saving by men and/or men’s parents and their demand of housing as awealth signal may induce also women and/or women’s parents to increase their savingsto be able to afford houses at prices that have been bid up by men’s demand.

• If household saving was not enough, another 25% of GDP was saved by Chineseenterprises, which enjoyed enormous revenue growth and felt little pressure to pay outdividends.

• With national saving approaching 50% of GDP and thereby exceeding even China’sextraordinarily high investment rates, the country ran continuous current account surpluses.

3

The Rest of the Emerging World

• With the ASEAN (Association of South Eastern Asian Nations) economies no longerencouraging investment at all cost, their national savings exceeded their investment as well.

• In Latin America, more stable policies similarly encouraged saving.

• For countries that had experienced crises in the late 1990s and early 2000s, increasedsaving was also a precautionary response to the volatility experienced with the crises andassociated reversals of capital flows:

– As households and firms in these countries increased their precautionary savings,several Asian and Latin American countries changed their positions from “traditional”borrowers to lenders.

• With strong growth in China and India pushing up energy prices, Middle East oil exportersearned more than they could invest at home; they too ran current account surpluses.

4

The U.S.

• All this excess saving had to go somewhere.

• If all these countries were in current account surplus, in other words, someone else had tobe in deficit, and that someone was the U.S.

• The U.S. had long-run current account deficits (Figure below).

5

U.S. CURRENT ACCOUNT DEFICIT AND REAL EFFECTIVE EXCHANGE RATE OF THE DOLLAR (1973-2007)

-7%

-6%

-5%

-4%

-3%

-2%

-1%

0%

1%

2%

Q11980

Q11982

Q11984

Q11986

Q11988

Q11990

Q11992

Q11994

Q11996

Q11998

Q12000

Q12002

Q12004

Q12006

0

20

40

60

80

100

120

140

CA (% of GDP) REER

CA

as

% o

f GD

P

RE

ER (C

PI B

ased

)

Source: Bureau of Economic Analysis and IFS

The Exorbitant Privilege and Valuation Effects

• In effect, other countries purchased financial claims on the U.S., and the U.S. purchasedmerchandise from other countries.

• Foreign central banks and governments were prepared to accumulate financial claims onthe U.S. because U.S. securities were traded in deep and liquid markets.

• Ricardo Caballero, Emmanuel Farhi, and Pierre-Olivier Gourinchas (2008, AmericanEconomic Review) highlight the comparative advantage of the U.S. in supplying financialassets as a key source of global imbalances.

• According to their argument, savings flow to the U.S. because the U.S. market provides awider range of attractive and, in many cases, safer investment opportunities.

6

The Exorbitant Privilege and Valuation Effects, Continued

• A related argument by Enrique Mendoza, Vincenzo Quadrini, and José Victor Ríos-Rull(2007, Journal of Political Economy) highlights asymmetry in financial development (proxiedby the quality of financial contract enforcement) as the reason why the U.S. market attractsthe world’s saving.

• Matteo Cacciatore, Viktors Stebunovs, and I (“The Domestic and International Effects of Interstate U.S. Banking,” 2015, JIE) focus on relative deregulation of the banking sector (which makes borrowing easier) as a source of U.S. external borrowing.

• All these arguments share a common focus on asymmetries in some key feature of thefinancial sector as the reason why the U.S. absorbs so much of the world’s savings: bettersupply of assets, better enforcement of contracts, easier access to borrowing in a morederegulated environment.

7


• The U.S. was able to place dollar-denominated debt securities with foreign central banksand governments while paying a lower interest rate than other borrowers.

• This ability to borrow in dollars at a lower interest rate than other borrowers was the“exorbitant privilege” of which the French had complained in the 1960s and that was broughtback to the center of discussions in international macroeconomics by the global imbalancesof the last decade.

• Moreover, whereas other countries accumulated U.S. debt securities, U.S. investorsacquired foreign equity:

– They purchased shares in foreign companies or even purchased those companiesoutright.

8


• While this meant that American investors took more risk, they earned higher returns on theirforeign assets in consequence.

• The U.S. could thus run continuing deficits without seeing its net foreign financial obligations explode due to the beneficial effects of returns on its external investments – what became known in the literature as valuation effects (Figure below).

• Cumulating current account data to evaluate the U.S. net foreign asset (or net international investment) position yields a significantly worse situation than it actually is because it does not take into account the effect of asset prices and returns on the market value of the net asset position.

9

U.S. NET INTERNATIONAL INVESTMENT POSITION AND CUMULATIVE CURRENT ACCOUNT DEFICIT (1993-2006), $ BILLION

-6000

-5000

-4000

-3000

-2000

-1000

0

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

$ bi

llion

NIIP Cumulative CA Deficit

Source: Bureau of Economic Analysis


• The U.S. runs a leveraged portfolio:

– Short in dollar-denominated U.S. assets (U.S. equity, corporate and government debt,inward direct investment, etc.) and long in foreign currency-denominated foreign assets(Japanese equity, direct investment in China, etc.).

– Weakening of the dollar and the excess return on foreign equity and FDI improve the U.S. asset position, as U.S. liabilities (the assets held by foreigners) lose value and the U.S. receives strong returns from its gross asset position.

10


• To think about the order of magnitudes involved, consider also the following figures.

• Note the boom of U.S. gross asset and liability positions as international financial integrationincreased in the 1990s.

• The next figure compares the standard way of constructing net foreign asset positions(cumulating current account data) to the U.S. actual net foreign asset position.

• The difference is due to valuation effects generated by asset price and ER movements that are not accounted for in the current account data, unless the data are appropriately adjusted.

• The net foreign asset position resulting from comparing gross assets and liabilities at market values was significantly better in 2005-2007 than suggested by cumulating the current account data as computed then.

11

U.S. Gross Foreign Assets and Liabilities(Percentage of GDP)

0

20

40

60

80

100

120

140

16019

76

1978

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

2002

2004

2006

Assets Liabilities

U.S. Net Foreign Assets and Cumulative Current Account(Percentage of GDP)

-25

-20

-15

-10

-5

0

519

90

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

Cumulative CA NFA

Deficits Grow

• In the second half of the 1990s, small current account deficits gave way to large currentaccount deficits – large absolutely and as a share of U.S. GDP.

• This was the era of the “New Economy”:

– Productivity growth accelerated in the United States as the country’s prior investments ininformation and communications technologies came to fruition.

– Faster productivity growth promised a higher return on capital, encouraging investment.

– The effects were most clearly evident in the NASDAQ boom.

– High share prices reflected hopeful expectations of high future profits and encouragedadditional investment.

– With investment rising relative to savings, that additional investment was necessarilyfinanced by foreigners.

12

But There Is Little Disquiet

• As yet there was little disquiet over the U.S. deficit:

– The current account being the difference between saving and investment, the deficit wasgrowing, it was said, because investment in America was becoming more attractive.

· The U.S. was disproportionately responsible for developing the new generation ofmicroprocessor-based technologies.

· Of all the advanced countries, it had the most flexible markets.

· Its firms were thus well positioned to reorganize their operations to capitalize on theopportunities afforded by high-speed computation, broadband, and the Internet.

– It was no wonder then that investment surged or that foreigners willingly financed it.

– Nor would there be a problem of repaying these obligations to foreigners, since a morerapidly growing economy would have a correspondingly greater capacity to service itsdebts.

13

The Bubble Bursts, but Deficits Keep Growing

• This happy scenario grew less plausible after the turn of the century, although it took sometime for popular and even professional commentary to cotton onto the fact.

• Once investors discovered that the New Economy was over-hyped and the NASDAQ bubbleburst, it became harder to argue that U.S. current account deficits were investment drivenand benign.

• But the deficit nonetheless continued to grow, from a bit more than 4% of GDP, the level thateconomists customarily took as the safe upper bound, to 5% in 2003, 6% in 2005, and 7 percent in 2006.

• The source, or the culprit as it was increasingly seen, was low U.S. saving.

14

Sources of Low U.S. Savings

• The Bush Administration cut taxes upon assuming office in 2001.

• A federal budget that had swung into surplus in the 1990s now swung back into deficit.

• But where the explanation for the fall in government savings was obvious – with federalspending as a share of GDP holding steady, it was the fall in tax take – explaining the fall inhousehold savings was less straightforward.

• Personal savings rates first fell to the low single digits and then turned negative around themiddle of the decade.

• Diehard proponents of the New Economy argued that households were spending morebecause U.S. economic fundamentals were so strong.

• That households could look forward to higher future incomes justified more spending now.

• But this view became harder to sustain after the NASDAQ crash and especially afterproductivity growth showed signs of slowing.

15

Sources of Low U.S. Savings, Continued

• The alternative explanation focused on the series of dramatic interest rate cuts by the Fedin the 2001 recession.

• Reversing out those cuts without choking off the subsequent recovery had to be donegradually.

• In the meantime, low interest rates and the availability of funds in the U.S. banking industrygenerated by the inflow of capital from abroad fueled an unprecedented housing boom.

• Higher real estate prices made households feel wealthier.

• Feelings aside, low interest rates enabled them to refinance their mortgages and divert theinterest savings to consumption.

• These observations made for less optimism about the sustainability of the deficit, however,since unusually low interest rates would not last forever and house prices could not just goup but would also eventually come down...

16

It Takes Two to Tango

• As always, it took two to tango.

• The U.S., in other words, was able to run large deficits only because other countries werewilling to run large surpluses.

• The U.S. was able to save less than it invested because other countries saved more.

• Federal Reserve Chairman Ben Bernanke described global imbalances, not without reason,as reflecting a “global savings glut.”

• But it was a global savings glut superimposed on a U.S. savings drought.

17

Reasons for This to Persist

• One view was that this situation was likely to persist for some time.

• Growth in China centered on manufacturing industry, which exported much of what itproduced, and much of this to the U.S.

• Keeping the ER down against the dollar was part of selling those additional manufacturesinto foreign markets.

• And as China grew, its central bank demanded additional foreign-currency reserves tosmooth the flow of international payments and insulate the economy from financial volatility.

• It was able to accumulate those reserves only because Chinese exports grew faster thanChinese imports.

• Meanwhile the U.S., as the source of those reserves, was happy to import more than itexported and consume more than it produced.

• Thus, this status quo was in the common interest.

• According to this argument, the status quo was likely to continue for 20 years, whichwas how long it could take to absorb an additional 200 million Chinese peasants into themanufacturing sector.

18

Bretton Woods II

• This situation also resembled that of the 1950s and 1960s, which is why some came to referto it as Bretton Woods II.

• Then as now, there had been a key-currency country at the center of the system runningdeficits and supplying the rest of the world with international liquidity.

• At the periphery had been a set of fast-growing economies exporting their way to higherincomes, running surpluses, and accumulating the additional reserves appropriate for theirnow larger economies.

• The country with the exorbitant privilege of supplying the reserves had been the same: theU.S.

• The only difference was the identity the catch-up economies running chronic surpluses andaccumulating reserves.

• Back then it had been Europe and Japan; now it was China and other Asian countries.

• But the implication was the same:

– If the original Bretton Woods System had lasted the better part of 20 years, then so toomight its successor.

19

Bretton Woods II, Continued

• Left to market forces, ERs in catch-up economies tend to appreciate.

• Since productivity is growing relatively fast, currency appreciation is needed to preventdisequilibrium from developing between the growth of exports and imports.

• Currency appreciation avoids the development of that disequilibrium by increasing thecommand of consumers over traded goods.

• This is one way in which higher productivity translates into higher living standards.

• But under BW this mechanism had been suppressed:

– European and Japanese currencies had been pegged to the dollar and, with fewexceptions, were prevented from moving.

• Now there was no formal agreement to stabilize ERs against the dollar, but the catch-upeconomies could still intervene in the market to prevent their currencies from appreciating.

20

Market Forces and Interest Compatibility

• Market pressures do not stay bottled up forever.

• In the case of the original BW system, they exploded in the early 1970s.

• The ongoing crisis suggests that this Bretton Woods II “regime” might reach its end evenmore quickly.

• Recall that the state of affairs ostensibly rested on the compatibility of U.S. and Chineseinterests:

– The U.S. was happy to consume more than it produced.

– China was interested in saving and exporting its way to prosperity and in accumulatingthe international reserves needed to smooth a larger volume of international transactions.

21

Signs of Strain

• But by 2005, both officials and investors had developed second thoughts:

– U.S. politicians saw the flood of merchandise imports from the developing world asunfairly burdening manufacturing industry.

– They blamed the reluctance of China and its neighbors to let their currencies rise andthreatened trade sanctions in response to this supposed manipulation.

– For China, saving 50% of GDP and investing nearly that much were not sustainableeconomically or politically.

– It simply was not possible to deploy that much additional capital year after year – to buildthat many new factories and dams – without significant inefficiencies.

– And it was not socially palatable for households to defer that much consumptionindefinitely.

– As Chinese savings fell, something that would happen even more quickly as thepopulation aged, the country’s external surplus would shrink.

· And this phenomenon of population ageing was not limited to China; it was presentalso in other East Asian countries, such as Japan and South Korea.

22

Signs of Strain, Continued

• Foreign reserves, meanwhile, had risen far beyond the levels needed to smooth internationaltransactions.

• The standard rules of thumb for reserve adequacy were the equivalent of three months ofimports or the cost of interest and principal payments on the foreign debt for a year.

• By 2005 reserves not just in China but in emerging markets generally far exceeded thosebenchmarks.

23

Policy Adjustments

• This pointed to the desirability of stimulating domestic demand to narrow the externalsurplus and slow reserve accumulation while allowing the currency to appreciate to preventthat additional demand stimulus from fanning inflation.

• With these goals in mind, and to head off trade sanctions by the United States, Chinaannounced in July 2005 that it was revaluing the renminbi by 2.1% and that, henceforth, itwould allow the currency to appreciate against the dollar.

• But 2.1% paled in comparison with the change in the Chinese ER needed to contribute toan orderly correction of global imbalances, which observers put at 20, 30 or even 40%.

• Letting the currency appreciate against the dollar by 5% per year, the pace now signaled bythe Chinese authorities, was barely enough to keep the problem from worsening.

• And with China reluctant to move faster, other countries hesitated to allow their owncurrencies to appreciate.

24

China’s Reluctance

• China’s reluctance had several sources:

– Officials hesitated to mess with success, as the currency peg had served the countrywell.

– There were limits on how quickly spending on infrastructure, education, and socialservices could be increased.

– There were questions about whether the country’s troubled banks could cope with thebalance-sheet effects of a more volatile ER.

· Officials warned that the country still lacked hedging markets on which banks andfirms could protect themselves from unpredictable ER swings.

• There were also worries that curtailing intervention in the foreign exchange market mightlead not just to a modest appreciation of Asian currencies against the dollar; it mightprecipitate a dollar crash.

25

The Late 1990s vs. More Recent Years

• In the late 1990s, finance for the U.S. current account deficit had originated with foreigninvestors lured by the siren song of the New Economy.

• Now the main foreign purchasers of U.S. assets were central banks and governments, andtheir purchases mainly took the form of the debt securities that were the favored form ofreserves.

– The U.S. federal budget deficits caused by the Bush tax cuts and the Iraq War ensuredplentiful supply of such debt securities.

• If foreign central banks and governments now curtailed their purchases, the dollar would fallsharply.

• This might catch investors wrong footed, causing financial disruptions and threatening globalgrowth.

• And it would cause those same central banks and governments to suffer capital losses ontheir existing reserves, the majority of which were denominated in dollars.

26

The Ideal Scenario

• In the ideal scenario, central banks and governments would curtail their accumulation ofdollars only gradually.

• Any effort to diversify their existing reserve holdings so as to protect their portfolios from adecline in the dollar would also proceed gradually.

• And if smaller capital inflows into the U.S. meant slower growth of demand in the U.S., thisshould be offset by measures to stimulate demand in other countries.

27

But...

• But realizing this outcome presupposed international cooperation.

• While it was in the collective interest for central banks to diversify out of dollars onlygradually, it was in the individual interest of each central bank to diversify quickly if it couldget away with it – if it could do so surreptitiously to avoid exciting the markets.

• But if enough central banks succumbed to the temptation, investors would catch on, and thedollar would come crashing down.

• What was in the collective interest, in other words, was not obviously in the individualinterest.

• Similarly, in return for undertaking currency and spending adjustments in the global interest,Chinese authorities wanted something back from the U.S.

28

The IMF

• The IMF had been established in 1944 to help organize collective action on internationalmonetary matters, and it now sought to organize solutions to these problems.

• It pushed central banks to release more information on the currency composition of theirforeign exchange reserves through its Special Data Dissemination Standard, the idea beingthat greater transparency meant less scope for surreptitious portfolio adjustments.

• It brought together the United States, Japan, China, and euro area, and Saudi Arabia(as a representative of the oil exporters) to discuss mutually-advantageous macro-policyadjustments.

• But progress on reserve transparency was slow.

• Only a couple of dozen countries participated, and even they released information onreserve composition only with a lag, leaving considerable scope for opportunistic portfolioadjustments.

29

The IMF, Continued

• The IMF’s multilateral consultations, for their part, produced much talk but little action.

• The Fund has no ability to compel action by large countries that did not borrow from it.

• It was especially feeble when dealing with surplus countries – in the present instance China.

• The IMF’s membership agreed to strengthen the Fund’s authority for ER surveillance, andspecifically its authority to warn of significantly undervalued currencies.

• A new decision on ER surveillance was agreed by its Executive Board, with the dissent onlyof China (not surprisingly).

• But only time would tell whether the IMF is finally prepared to use its bully pulpit and whetherits calls will be heard.

30

Crisis in 2007-2009

• By late 2007 the issues associated with imbalances and their underlying dynamics assumeda growing urgency.

• U.S. house prices peaked in 2006 and started declining as excess supply of accumulatedresidential construction in the boom years contributed to the unraveling of the bubble.

• By 2007 residential construction was in decline.

• As house prices declined, the underpinning of a large number of mortgages extended byU.S. financial institutions collapsed and crisis erupted in the second half of 2007:

– Subprime borrowers, who had had easy access to liquidity also thanks to the globalsavings glut that had flooded the U.S. financial sector, found themselves unable to makemortgage payments as interest rates adjusted upward.

– Financial institutions that tried to recoup insolvent mortgages by seizing and selling theunderlying assets (the houses) were unable to do so at the expected prices and saw theirasset balances drastically reduced.

31

Crisis in 2007-2009, Continued

• Securitization of mortgages (the creation of so-called mortgage-backed securities bypackaging mortgages, “slicing,” and selling the resulting claims to the underlying assets ininternational financial markets spread the crisis outside the U.S., as international holders ofthese securities found themselves holding what became known as “toxic” assets.

• Massive use of credit default swaps (swap contracts in which the buyer makes payments tothe seller and receives in exchange a payoff if a credit instrument – such as a loan – goesinto default) further contributed to the propagation of the financial shock, bringing downinsurance giant AIG.

• Financial institutions in the U.S. and elsewhere collapsed or were saved by governmentintervention in the largest crisis since the Great Depression.

• Not surprisingly, with balance sheet news becoming bleaker by the day, stock markets wentinto a tailspin.

32


• Notwithstanding the coordinated attempt of central banks to inject liquidity in the bankingsector, the crisis resulted in a freezing of credit.

• With financial institutions fearful of lending to each other in the interbank market, access tocredit became increasingly difficult for households and firms.

• As a consequence, the financial shock propagated to the real economy, with employmentand output losses as businesses reduced production or completely shut down.

• U.S. consumption demand – historically the key driver of U.S. GDP with a 70 percent share– dropped, partly also a consequence of the negative wealth effect of huge stock marketlosses absorbed by many American households.

33


• In the early stages of the crisis, the presumption of its implications for the U.S. dollar was ofa loss in value of the U.S. currency.

• The thinking was that if there was less demand at home, then more U.S. products wouldhave to be sold abroad, and the dollar would have to fall to price those U.S. goods intoforeign markets.

• In fact, the dollar had already begun falling in anticipation of this eventuality.

• With the onset of the crisis, U.S. markets appeared less attractive as a destination for foreignfunds.

• Capital inflows slowed, and market participants began to talk of a dollar crash.

34


• According to this argument, the incentive to scramble out of dollars to avoid losses is all thegreater insofar as there is something to scramble into, namely euros.– The euro area also has deep and liquid financial markets, which make it an increasingly

attractive place for central banks to hold international reserves.

• But if investors shift into euros in large numbers, the result is also an uncomfortably strongeuro exchange rate – uncomfortably strong for European exporters in particular.

• In fact, while the dollar did lose ground relative to the Japanese yen (although it reboundedmost recently), it appreciated against the euro, and it appreciated against the currencies ofmajor trading partners.

• As global financial turmoil spread, international investors still found that dollar assets – inparticular, U.S. T-bills – remained the safest haven for their money.

35

http://research.stlouisfed.org/fred2/graph/?id=EXJPUS&printgraph&load_default_graph

1 of 1 4/30/2010 5:05 PM

http://research.stlouisfed.org/fred2/graph/?id=EXUSEU&printgraph&load_default_graph

1 of 1 4/30/2010 5:09 PM

http://research.stlouisfed.org/fred2/graph/?id=TWEXMMTH&printgraph&load_default_graph

1 of 1 4/30/2010 5:10 PM

Policy Responses

• Central banks around the world were the first policymakers to respond aggressively to theglobal crisis, by injecting liquidity in the economy in a coordinated fashion.

– Stephen Cecchetti’s 2009 Journal of Economic Perspectives article in the syllabusdescribes this response in detail, with special focus on the Federal Reserve.

• Importantly, the Fed was very creative in its approach to the crisis.

• Not only it lowered the Federal Funds Rate to essentially zero and supplied liquidity throughthe traditional channels.

• It also engaged in acquisitions of assets (as collateral for lending in the form of cash andsecurities) that drastically altered the composition of the asset side of its balance sheet.

36

Policy Responses, Continued

• The acquisition of assets of questionable value by the Fed was more akin to fiscal policyactions than to standard monetary policymaking, and it poses a very important question forthe long-term future of the Federal Reserve.

• As the Fed engages in actions that are akin to fiscal policy, it exposes itself to increasedinterference from the government in monetary policymaking, i.e., it runs the risk of reducedindependence.

• Much literature has documented that low central bank independence tends to result in highinflation over time, as governments have a “natural” tendency to put pressure on centralbanks to engage in inflationary, monetary financing of deficit spending.

– If a government can choose to finance spending by raising taxes or printing money, it willoften choose the latter for its lower short-term political cost.

– This is a key reason why successful central banks are not run by elected officials (althoughtheir governors are accountable to elected officials) and are highly independent.

• With the Fed’s incursion into “fiscal” policy, will we see a politicization of U.S. monetarypolicy in the future, with potentially destabilizing effect on inflation?

37


• Given this long-term risk, why did Chairman Bernanke push the Fed in the direction of thesenon-standard policy actions?

• Bernanke faced the following tradeoff:

– Either I take the risks implied by a foray into “fiscal” policy, or I take the risk of not doingenough and letting the crisis truly explode into a new Great Depression.

• My view is that Bernanke and the Fed did the right thing.

• But the consequences of this choice for the interaction between U.S. politics and U.S. monetary policy remain unresolved.

38


• Monetary policy was not the only tool deployed to respond to the crisis.

• Governments also intervened by using a variety of fiscal policy instruments and programs,of which you have had ample coverage in the media and whose merits and success (orfailure) will be determined in the coming months and years.

– Fiscal policy moved slower than monetary policy due to the inherent relative inflexibility offiscal policymaking, which involves political debate, compromise, and legislative action.

• A global debate on international financial regulation was also started, with the Financial Stability Board, chaired by then Bank of Italy Governor Mario Draghi, as the key international umbrella organization.

• And the resources for international institutions such as the IMF were bolstered by theagreement reached in the April 2009 G-20 meeting in London.

39


• In London, President Obama pushed for increased, internationally coordinated fiscalstimulus.

• French President Sarkozy and German Chancellor Angela Merkel pushed for strengthenedinternational financial regulation rather than more fiscal expansion.

• As Francesco Giavazzi correctly put it in a newspaper editorial in those days, the U.S.administration recognized that perfect regulation of dead markets does not help much.

• Its first goal is thus to re-start the economy and financial markets, unfreezing credit revivingthe stock market – the main source of U.S. households’ financial wealth.

• It remains to be seen how much of a success the meeting was in reality.

40

Understanding Policy Coordination after the Crisis

• Can we use the tools that we learned in this course to understand why policymakers pushedfor coordinated responses to the crisis?

• We cannot do it with respect to fiscal policy, because the models we reviewed do not includefiscal policy.

• But we can do it with respect to monetary policy using a simple modification of the model you worked on in your homework problem on monetary policy interactions under flexible exchange rates.

41

Understanding Policy Coordination after the Crisis, Continued

• Suppose we modify the money demand equation for the home and foreign countries in themodel as follows:

m+ v = p + y, (1)m∗ + v∗ = p∗ + y∗. (2)

• In these equations, v and v∗ denote the velocity of money in the home and foreign country,respectively – the frequency with which a unit of money is spent for transactions in theeconomy.

– For given money supply, higher velocity translates into higher prices and/or output.

• We treat velocity as an exogenous shock with average value of zero, as we did with theexogenous productivity shock x, and we can think of velocity as an indicator of the situationof credit markets.

– As for other variables, v and v∗ measure the percent deviation of velocity from trend.

– On average – in the zero-shock equilibrium – the percent deviation of velocity from trendis zero.

• A credit market collapse (such as the current crisis) results in a drop in the velocity of money(treat this as a negative realization of v and v∗ in a global credit crisis).

42


• If you repeat the steps you took to solve the homework problem, using the money demand equations (1) and (2) and all the other assumptions in the problem, you will be able to verify the following:

w = E−1 (m + v) = 0,

w∗ = E−1 (m∗ + v∗) = 0,

n = m + v,

n∗ = m∗ + v∗,

y = (1− α) (m+ v)− x,

y∗ = (1− α) (m∗ + v∗)− x

p = α (m + v) + x,

p∗ = α (m∗ + v∗) + x.

• Note: A credit crisis (a drop in velocity) puts downward pressure on employment, output, and prices: precisely what has happened in the U.S. and the rest of the world during the crisis.

43


• Proceeding as in the homework, you can verify that the nominal exchange rate and theterms of trade are determined by:

e =1− α (1− δ)

δ

¡mD + vD

¢,

z = e− pD =1− α

δ

¡mD + vD

¢,

and home and foreign CPIs are:

q =αδ + (1− a) (1− α)

δm+

αδ + (1− a) (1− α)

δv

−(1− a) (1− α)

δm∗ − (1− a) (1− α)

δv∗ + x,

q∗ =αδ + a (1− α)

δm∗ +

αδ + a (1− α)

δv∗ − a (1− α)

δm− a (1− α)

δv + x.

• Note that all these equations are identical to those in the homework except for replacing m

with m+ v and m∗ with m∗ + v∗ everywhere.

• Intuitively, this follows from the fact that the only change we made in the model was toreplace m with m+ v and m∗ with m∗ + v∗ in the money demand equations.

44


• Now, as in the homework, assume equal country size (a = 1/2) and find the optimal moneysupplies for the home and foreign central banks under non-cooperation.

• The first-order condition for the home central bank is:

γ∂q

∂mq + (1− γ)

∂n

∂mn = 0.

• Given the reduced form equations for q and n, this implies:

γ

∙αδ + 1

2 (1− α)

δ

¸" αδ+12(1−α)δ m+

αδ+12(1−α)δ v

−1−α2δ m∗ −1−α2δ v

∗ + x

#+ (1− γ) (m + v) = 0.

• Tedious, but straightforward algebra allows you to re-write this as:(1− γ + γ

∙αδ + 1

2 (1− α)

δ

¸2)m = γ

∙αδ + 1

2 (1− α)

δ

¸1− α

2δm∗

−(1− γ + γ

∙αδ + 1

2 (1− α)

δ

¸2)v

+γ

∙αδ + 1

2 (1− α)

δ

¸1− α

2δv∗

−γ∙αδ + 1

2 (1− α)

δ

¸x. (3)

45


• Define the following coefficients:

HN1 ≡ 1− γ + γ

∙αδ + 1

2 (1− α)

δ

¸2> 0,

HN2 ≡ γ

∙αδ + 1

2 (1− α)

δ

¸1− α

2δ> 0,

HN3 ≡ γ

∙αδ + 1

2 (1− α)

δ

¸> 0.

• Then, we can rewrite equation (3) as:

HN1 m = HN

2 m∗ −HN

1 v +HN2 v∗ −HN

3 x,

or:

m =HN2

HN1

m∗ − v +HN2

HN1

v∗ − HN3

HN1

x. (4)

46


m =HN2

HN1

m∗ − v +HN2

HN1

v∗ − HN3

HN1

x.

• This equation tells us how the home central bank responds to foreign money supply, homeand foreign velocity shocks, and the world productivity shock:

– All the H coefficients in the equations are positive because of our assumptions on theparameters involved. Hence:

· The home central bank contracts its money supply in response to a foreign contractionor a drop in foreign velocity that causes q to rise.

· The home central bank expands its money supply in response to a drop in domesticvelocity that causes q and n to fall.

· The home central bank contracts its money supply in response to a drop in worldproductivity (x > 0) that causes q to rise.

• Because of symmetry between the two countries when a = 1/2, the foreign central bank’sbehavior is determined by a symmetric equation:

m∗ =HN2

HN1

m− v∗ +HN2

HN1

v − HN3

HN1

x. (5)47


• To find the optimal, non-cooperative m and m∗, substitute (5) into (4) and solve the resultingequation for the Nash equilibrium level of m (denoted mN), obtaining:

mN = −v − HN3

HN1 −HN

2

x.

• If you use the expressions for HN1 and HN

2 , you can verify that:

HN1 −HN

2 = 1− γ + αγ

∙αδ + 1

2 (1− α)

δ

¸.

• Taking the expression for HN3 into account, it follows that:

mN = −v −γhαδ+1

2(1−α)δ

i1− γ + αγ

hαδ+1

2(1−α)δ

ix. (6)

• Then substitute this into (5), and use the expressions for HN1 , HN

2 , and HN3 , to find:

m∗N = −v∗ −γhαδ+1

2(1−α)δ

i1− γ + αγ

hαδ+1

2(1−α)δ

ix. (7)

48


• The values of m and m∗ in (6) and (7) represent the intersection of the reaction functions (4)and (5) of the two central banks.

• Note: In the non-cooperative equilibrium of the game, both central banks fully offset therespective velocity shock (or domestic credit market crisis), with no response to the othercountry’s.

– If you look at the reduced form equations for employments and CPIs, you will see that, ifthe credit market crisis is a pure velocity shock and there is no other shock (i.e., if x = 0),this allows the two central banks to keep the respective employments and CPIs (andoutputs, and domestic product prices) at zero.

– In other words, the policy in which each country fully offsets its own credit crisis allowsboth central bank to achieve the bliss equilibrium of zero loss (since the loss function foreach central bank depends on the squares of the respective CPI and employment).

49


• How is this bliss situation possible?

– This goes back to what we noted above – that modifying the model to include velocityshocks only implies replacing m with m + v and m∗ with m∗ + v∗ in the money demandequations and all reduced forms.

· Since the velocity shock pushes both CPI and employment in the same direction, andit does it by the same amount as a money supply change does, the shock does notface central banks with a tradeoff between objectives.

– It follows that if a central bank is fully offsetting its velocity shock with its money supply,the other will just find it optimal to do the same, and both will achieve the bliss situationof zero loss.

– It also follows that if the credit crisis is a pure velocity shock, there is no gain frominternational coordination, because it is impossible to do better than the bliss equilibrium.

– We verify this point below.

50


• The first-order condition for the optimal choice of m when the two central banks coordinatetheir policies is:

γ∂q

∂mq + (1− γ)

∂n

∂mn+ γ

∂q∗

∂mq∗ + (1− γ)

∂n∗

∂mn∗ = 0,

or:

0 = γ

∙αδ + 1

2 (1− α)

δ

¸" αδ+12(1−α)δ m +

αδ+12(1−α)δ v

−1−α2δ m∗ −1−α2δ v

∗ + x

#+ (1− γ) (m+ v)

−γ1− α

2δ

∙αδ + 1

2 (1− α)

δm∗ +

αδ + 12 (1− α)

δv∗ − 1− α

2δm− 1− α

2δv + x

¸. (8)

51


• Proceeding as in the non-cooperative case, we can rewrite equation (8) as:

HC1 m = HC

2 m∗ −HC

1 v +HC2 v∗ −HC

3 x,

where we define:

HC1 ≡ 1− γ + γ

∙αδ + 1

2 (1− α)

δ

¸2+ γ

µ1− α

2δ

¶2> 0,

HC2 ≡ γ

∙αδ + 1

2 (1− α)

δ

¸1− α

δ> 0,

HC3 ≡ γ

∙αδ + 1

2 (1− α)

δ

¸− γ

1− α

2δ.

• This implies the cooperative setting of m according to:

m =HC2

HC1

m∗ − v +HC2

HC1

v∗ − HC3

HC1

x. (9)

• In turn, symmetry between the two countries implies that the first-order condition for thechoice of m∗ will yield:

m∗ =HC2

HC1

m− v∗ +HC2

HC1

v − HC3

HC1

x. (10)

52


• We can solve the system of equations (9) and (10) for the cooperative responses of thecentral banks to velocity and productivity shocks as we did in the non-cooperative case.

• Substituting (10) into (9) and rearranging, the solution for mC is:

mC = −v − HC3

HC1 −HC

2

x.

• Using the expressions for HC1 and HC

2 shows that HC1 −HC

2 = 1− γ¡1− α2

¢.

• Hence, taking HC3 = γα into account, we have:

mC = −v − γα

1− γ (1− α2)x.

• Substituting this into (10) and using the expressions for HC1 , HC

2 , and HC3 = γα yields:

m∗C = −v∗ − γα

1− γ (1− α2)x.

53


• Note: The optimal, cooperative responses to velocity shocks are the same as in thenon-cooperative case:

– Both central banks fully offset the respective credit crisis (v < 0, v∗ < 0) by expandingmoney supply in matching fashion.

– Regardless of cooperation or non-cooperation, doing so allows central banks to achievethe bliss equilibrium with zero loss if the credit crisis is a pure velocity shock for thereasons discussed above.

– Hence, there is no gain from coordinating policies if the global crisis is a pure internationalvelocity shock.

54


• However, the responses to a productivity shock differ between cooperative and non-cooperative equilibria.

• In particular, we can verify that the cooperative responses are less aggressive than thenon-cooperative ones.

• We can see this already by studying the responses along the reaction functions (6)-(7)versus (9)-(10).

• Note that manipulating the expression for HC3 in (9)-(10) shows that HC

3 = γα > 0.

• You can also verify that the expression for HN3 in (6)-(7) can be rewritten as HN

3 = γα+γ 1−α2δ .

• Hence, HC3 < HN

3 .

• Since it is immediate to see that HC1 > HN

1 , it follows that cooperative responses to anunfavorable world productivity shock (x > 0) are less contractionary than the non-cooperativeones.

55


• In the equilibrium of the policy game (i.e., for the equilibrium money supplies), this point canbe confirmed by verifying that:

γhαδ+1

2(1−α)δ

i1− γ + αγ

hαδ+1

2(1−α)δ

i > γα

1− γ (1− α2).

• If you do the algebra, this inequality reduces to 1 − γ > 0, which is always satisfied underthe assumptions of our model.

56


• This reduction in policy “aggressiveness” in the cooperative response to a productivity shockreflects the intuition we discussed many times:– When an unfavorable, global productivity shock (x > 0) happens, it causes inflation to

rise in both countries.– Central banks face a tradeoff: If they respond by contracting money supply, they reduce

inflation, but this comes at the cost of lower employment.– In the non-cooperative equilibrium, central banks contract more aggressively because

they try to induce their currency to appreciate, thus shifting some of the inflation to theother country.

– Since both central banks choose the same response to x (equal country size impliesidentical tradeoffs), the exchange rate actually does not move, and non-cooperativebehavior results in excess contraction, with a larger employment cost.

– When central banks coordinate their policies, they refrain from trying to manipulate theexchange rate in their favor and contract by less in response to the same shock x > 0.

– As a consequence, even if inflation ends up being somewhat higher, they end up sufferingsmaller losses by imposing less employment cost on the economy.

– Hence, there are gains from coordinating policies in response to the productivity shock x.

57


• So, what did we conclude so far?

1. No gains from coordination after a velocity shock.

2. Gains from coordination after a global productivity shock.

• But if we treat the current credit crisis as a pure velocity shock, why then did central bankwant to coordinate their policies?

58


• The model does offer a possible answer for that, and it is the fact that we should not treatthe crisis only as a pure velocity shock.

• Remember what we learned from Ben Bernanke’s article on the Great Depression:

– Dislocations of credit have negative supply-side effects as asymmetric information issuesmake access to credit harder and prevent firms from operating efficiently.

– In other words, there is a negative productivity effect of credit crises – which has beendocumented empirically for several countries.

• In our model, we can capture that by positing that the shock x, instead of being purelyindependent from v and v∗, depends on these variables: x = x (v, v∗).

59


• In particular, suppose that, when v and v∗ become negative (a global credit crisis), xbecomes positive (a productivity loss).

• In this case, lack of central bank cooperation will result in less monetary policy expansion(and therefore more losses) than when central banks coordinate their policies, because ofthe combined responses of each central bank to the velocity and productivity effects of thecrisis.

• Our model would thus tell us that central banks wanted to coordinate their responses to thecrisis to ensure the larger, optimal global injection of liquidity (monetary expansion).

• Considering that also in the case of fiscal policy the effort at the G-20 meeting in Londonwas to increase the degree of fiscal expansion by coordinating, this is a plausible conclusionfrom the model.

• Thus, you see how we can use the tools we studied in the course as a lens to approacheven the most recent economic events.

60

Some Current, Cutting-Edge Research on the Crisis

• Fabrizio Perri (University of Minnesota) and Vincenzo Quadrini (University of Southern California) published an interesting paper on the global crisis in the AER in 2018 (“International Recessions”).

• The paper begins by presenting some evidence:

– Strong synchronization of 2008-2009 recession across countries—stronger than inearlier recessions.

– Negative correlation between labor productivity and GDP and hours during 2008-2009recession (and 2001 recession); positive during earlier recessions.

61

are also plotted. The figure reveals the strong co-movement that has characterized the current

crisis.

Figure 1: The dynamics of GDP during the 2008 recession: US v/s other G7 countries.

To examine whether the international synchronization of the recent recession differs from

previous contractions, Figure 2 plots the GDP dynamics for the G7 countries in six of the

most recent US recessionary episodes: one recession experienced in the first half of the 1970s,

two in the first half of 1980s, one in the early 1990s and two in the 2000s. A quick glance at

the figure shows that the synchronization of the US GDP with other G7 countries has been

significantly stronger in the latest recession. While the G7 countries experienced very different

GDP dynamics during the previous US recessionary periods, in the most recent contraction

the GDP of all countries have moved in the same direction.

The higher cross-country synchronization of most recent recession can also be seen in Figure

3 which plots the average correlation of US GDP with the GDP of each of the other G7

countries. The correlations are computed on rolling windows of 10 and 20 years. The dates

in the graph correspond to the end points of the window used to compute the correlation.

Although the figure shows that during previous recessions there is an increase in correlation,

the current recession stands out as the one that marks an increase in correlation larger than

2

Figure 2: The dynamics of GDP during the six most recent recessions in the G7 countries.

3

Figure 3: Average rolling correlations of US GDP v/s G7 countries.

-.6

-.4

-.2

.0

.2

.4

.6

60 65 70 75 80 85 90 95 00 05

corr h, ph, 10y

-.1

.0

.1

.2

.3

.4

.5

.6

.7

60 65 70 75 80 85 90 95 00 05

corr us,g7, 10y

-.6

-.4

-.2

.0

.2

.4

.6

60 65 70 75 80 85 90 95 00 05

corr h,ph, 20y

.15

.20

.25

.30

.35

.40

.45

.50

60 65 70 75 80 85 90 95 00 05

corr us,g7, 20y

Figure 14 the increases observed in previous recessions.

1.2 Productivity and Economic Activity

Figure 4 plots labor productivity (output per hour) in the nonfarm business sector of the US

economy for the six most recent recessions. As shown in the last panel, in the recent recession

labor productivity has continued to grow for most of the period. This pattern can also be seen

in the 2001 recession. By contrast, in the first four recessions, labor productivity has declined

and its level at the end of the recession was not higher than before the recession. Therefore,

while earlier recessionary episodes have been associated with significant falls in productivity,

there is not much of a productivity slow down in the last two recessions. This pattern is in

contrast to the dynamics of labor and output. As can be seen in Figure 5, all recessions are

characterized by sizable contractions in working hours and GDP.

The different behavior of productivity and labor during the two most recent recessions

4

Figure 4: Productivity of labor (output per hour) in the private nonfarm sector.

5

Figure 5: Hours and GDP in the private nonfarm sector.

6

Figure 6: Rolling correlations of productivity growth and hours growth in the US.

reflect a more general pattern for which the correlation between productivity and labor has

declined sharply in the US economy. Figure 6 plots rolling correlations of productivity (output

per hour in the private nonfarm business sector) and labor (hours worked in the private nonfarm

business sector) computed on 20 years windows. The Figure shows a drastic drop in the

correlation between productivity and labor starting at the beginning of the 2000s. This pattern

is also documented in Gali and Gambetti (2009) for the US economy.

Is the declining correlation between labor productivity and hours also a feature of other

countries? Figure 7 plots rolling correlations of output per hour and working hours for each of

the G7 countries. Because of comparability issues, these correlations are computed only for the

manufacturing sector and at an annual frequency. Although there are some divergences among

the G7 countries, the average plotted in the bottom panel clearly shows that the correlation

has been declining on average in the group of the seven major industrialized economies.

1.3 Hints from the data and theoretical approach

To summarize, the graphs shown above point out two major changes:

1. Higher international synchronization of recessions.

7

Figure 7: Rolling correlations on a 20 years window of productivity growth with hours growthin the manufacturing sector. Annual data for the G7 countries.

8

Perri and Quadrini’s Theory

• Perri and Quadrini then present a model to explain this evidence:

– 2-country dynamic, microfounded model with financial frictions in the form of contractenforcement constraints and shocks to these constraints (in addition to productivityshocks).

– Shocks to enforcement constraints (credit market shocks) generate cross-country andwithin-country dynamics that nicely match the evidence—at least qualitatively—whenfinancial markets are internationally integrated.

– Shocks to productivity do not generate such matching: no increased GDP comovementacross countries (unless shocks are perfectly correlated); positive correlation betweenproductivity and GDP and labor.

• Conclusion: 2008-2009 was about credit market shocks, not productivity shocks.

62

But Did Productivity Really Increase? And If So, What Measure?

• Following are two figures from recent work by Susanto Basu.

• They show the paths of a standard total factor productivity (TFP) measure and a capital-and-labor-utilization-adjusted measure that builds on Basu’s American Economic Review(2006) work with John Fernald and Miles Kimball and further work by John Fernald withKyle Matoba (Fernald and Matoba, 2009, Federal Reserve Bank of San Francisco).

• Two facts emerge:

1. A standard measure of TFP indicates that TFP dropped in the recent recession—as it didin 1991 and (less so) in 2001.

2. However, utilization-adjusted TFP increased sharply, differently from 1991 and 2001.

63

.

Page 2 of 9

Growth accounting adjusted for utilization of capital and labor

Figure 1 TFP and Utilization-Adjusted TFP

‐3

‐2

‐1

0

1

2

3

4

1989:Q1 1991:Q1 1993:Q1 1995:Q1 1997:Q1 1999:Q1 2001:Q1 2003:Q1 2005:Q1 2007:Q1 2009:Q1

Four‐Qua

rter Percent Cha

nge, %

Note: Appendix I describes the quarterly measure of standard TFP growth constructed by FM, lnTFPΔ . Appendix II describes the BFK method of estimating utilization change, lnUΔ . The

difference between the two gives the estimate of utilization-adjusted TFP growth, ln TFPAΔ . 2009:3 data are from the “advance” estimate.

• Standard TFP, the blue line in Figure 1, typically looks weak in recessions. The current recession is no exception, as TFP growth turned negative through 2009:1 before recovering sharply in the second and third quarters of this year.

• In contrast, utilization-adjusted TFP, the red line, has recovered from its very weak performance in 2006-2007.

• As Table 1 shows, utilization-adjusted TFP has actually been growing faster than its post-war average since the start of the financial crisis in 2008:3.

TFP

Utilization‐adjusted TFP

.

Page 7 of 9

Figure A1 TFP and Utilization-Adjusted TFP

0.95

1.05

1.15

1.25

1.35

1.45

1.55

1.65

1960:Q1

1962:Q1

1964:Q1

1966:Q1

1968:Q1

1970:Q1

1972:Q1

1974:Q1

1976:Q1

1978:Q1

1980:Q1

1982:Q1

1984:Q1

1986:Q1

1988:Q1

1990:Q1

1992:Q1

1994:Q1

1996:Q1

1998:Q1

2000:Q1

2002:Q1

2004:Q1

2006:Q1

2008:Q1

Inde

x Level: 19

60:Q1 = 1

Figure A1, as well as Figure 1, raises the possibility that some of the weak productivity performance in 2006-7 could reflect a “take back” of the strong performance earlier in the 2000s. For example, in 2002-2004, anecdotal evidence suggested that firms faced strong cost pressures, which could have led firms to postpone unobserved intangible investments in organizational capital—raising growth in observed market output relative to true output (which includes both market output and intangible investment). Once the recovery was strongly under way, firms might have resumed these intangible investments that, in the short run, thereby reducing growth in measured market output relative to true output.

Utilization‐ adjusted TFP

TFP

What Does This Imply for the Perri-Quadrini Model?

• The production function in Perri and Quadrini’s work is:

Yt = Zt

¡Kα

t N1−αt

¢σ.

• If σ = 1 (constant returns to scale, CRS), the standard measure of TFP generated by themodel in response to a credit shock is constant (because, by assumption, Zt is not moving).

– The model would miss the drop in standard TFP and the increase in utilization-adjustedTFP during the Great Recession.

• With σ = .9 (the value used in Perri and Quadrini’s exercise), we have decreasing returns toscale (DRS), and TFP would actually rise when inputs fall in response to the shock.

– The usual TFP calculation assumes CRS. DRS makes TFP countercyclical, and IRSmakes it procyclical, even if Zt is constant.

• Perri and Quadrini could say that variable utilization (currently not in their model) wouldexplain the decline in the standard measure of TFP in the data, while utilization-adjustedTFP rises because of DRS.

• Nevertheless, with σ = .9 (or other plausible values of ν), the model would still be quite farfrom capturing the increase in utilization-adjusted TFP documented by Basu.

64

Focusing on Labor Productivity

• Let us just say we want to focus on labor productivity, as Perri and Quadrini do.

• How would our simpler model do on that dimension?

• Labor productivity (measured as output per hour) would be computed as:

y − n = −α (m + v)− x,

where we used the reduced forms for y and n implied by the model.

• Suppose monetary policy does not move. Then:

y − n = −αv − x.

• The unfavorable credit market shock (v < 0) would cause labor productivity to increase,consistent with the evidence in Perri and Quadrini.

• An unfavorable global productivity shock (x > 0) would cause output to fall in both countries,but it would also cause measured labor productivity to fall, contrary to the evidence.

• A combination of the two (a credit market shock, v < 0, that causes also an unfavorableproductivity shock, x > 0, through credit dislocation) could still be consistent with rising y−n

as in the data.65

“Starting from the End”

• The main problem with our simple analysis (and Perri and Quadrini’s paper) is that we areboth “starting from the end.”

• From Perri and Quadrini’s Conclusion: “Although the paper illustrates the importance of[credit] shocks, it does not provide an explanation of what could cause a credit shock. Moreresearch is needed to identify the sources of these shocks.”

• Understanding the source of the Great Recession should be central to understanding its propagation and to the next task of evaluating policy responses.

66

Matteo Iacoviello (2015): “Financial Business Cycles”

• In 2015, Matteo Iacoviello published a model in the Review ofEconomic Dynamics in which recession is initiated by losses of financial institutions, and exacerbated by their inability to extend credit to the real economy.

• The shock that triggers the recession is the default by a small sector of the economy—borrowers who use their homes as collateral—on their loans.

• When banks hold little equity in excess of regulatory requirements, portfolio losses requirethem to recapitalize or deleverage immediately.

• By deleveraging, banks transform the initial shock into a credit crunch, and they amplify andpropagate the financial shock to the real economy.

• In Iacoviello’s experiments, borrowers default resulting in credit losses of about 4 percent ofGDP lead to a 3 percent drop in output.

67

loans. As a whole, the household sector is a net supplier of savings to the rest of the economy.

Entrepreneurs accumulate physical capital (which they rent to a representative firm) and borrow

from the bank, subject to a collateral constraint.

Figure 1: Summary of the Model Structure

Bankers intermediate funds between patient savers on the one hand, and entrepreneurs and sub-

primers on the other. The nature of the banking activity implies that bankers are borrowers when it

comes to their relationship with households, and are lenders when it comes to their relationship with

the credit-dependent sectors (entrepreneurs and subprimers) of the economy. I design preferences in

a way that two frictions coexist and interact in the model’s equilbrium: first, bankers’are credit con-

strained in how much they can borrow from the patient savers; second, entrepreneurs and subprimers

are credit constrained in how much they can borrow from bankers. My interest is in understanding

how these two frictions interact with and reinforce each other.

Finally, the representative firm converts entrepreneurial capital and household labor into the final

good using a constant-returns-to-scale technology.

Patient Household Savers. There is a continuum of measure 1 σ of savers (indexed by H). They

choose consumption C, housing H and time spent working N to solve the following intertemporal

problem:

3

Matteo Iacoviello (2010): “Financial Business Cycles,” Continued

• The shock that triggers the recession is modeled as an exogenous drop in the amount ofloan repayment in the budget constraint of household-borrowers, rather than an exogenousshock in the borrowing constraint.

• To some extent, that is not entirely satisfactory either—ideally, we would like the model toexplain also why some households end up defaulting.

• Nevertheless, I view Iacoviello’s work (and work by others in macro) as very valuable for thefuture development of international macro models that go deeper into the sources of theGreat Recession.

68

Starting “Before the End”

• The 2007-2008 meltdown was the culmination of a process that started some years before the crisis, in which asset price and international imbalance dynamics played crucial roles.

• Global imbalances fueled reckless lending (and borrowing) in the U.S. in the (misguided)expectation of ever increasing house prices and in a world of over-optimism about insurancefrom securitization (as shown in another interesting recent paper by Emine Boz and EnriqueMendoza, “Financial Innovation, the Discovery of Risk, and the U.S. Credit Crisis,” 2009).

• To some extent, modeling the international Great Recession simply as the response to anexogenous credit shock amounts to starting from the end: the end of that process and thenits consequences.

• A deep understanding of the Great Recession (and future policy evaluation exercises) should not be separated from the dynamics that led to it.

Much progress has been made in this area.

This progress and the lessons we learned from episodes like the Great Depression, the 2007-08 Global Financial Crisis, and the Great Recession that followed it are now informing policymakers in their responses to the COVID-19 crisis.

69

Topics in Monetary Economics: The International Monetary...

Documents

Transcript of Topics in Monetary Economics: The International Monetary...