Why in the World Are We All Keynesians Again? The Flimsy Case for Stimulus Spending

7/30/2019 Why in the World Are We All Keynesians Again? The Flimsy Case for Stimulus Spending

1/20

Executive Summary

The U.S. federal government responded tothe financial crisis and recession that beganin 200708 with unprecedented fiscal stimu-lus. Passed in February of 2009, the AmericanRecovery and Reinvestment Act (ARRA) camewith a price tag of $831 billion. Yet the economyhas not returned to a path of robust economicgrowth and unemployment has stubbornly re-mained high; only in September of 2012 did

it dip (barely) below 8 percent. This has notstopped the Obama administration from push-ing for further fiscal stimulus.

Whether or not fiscal spending stimulusis effective hinges critically on the size of thespending multiplier, which is dependent on sever-al factors. In particular, if individuals anticipatethe future tax liabilities associated with deficitspending and/or are crowded out by the defi-cit spending, then the multiplier is likely to beless than one; that is, each dollar of stimulus

increases total spending in the economy by lessthan one dollar. Ultimately, whether the govern-ment spending multiplier is less than or greaterthan one is an empirical question.

While no one seriously debates the long-runcosts associated with exploding public debt,the evidence suggesting significant short-runbenefits of stimulus spending is weak. Studiessuggesting that stimulus spending is effective

assume that current economic activity does notaffect fiscal policy. This assumption is question-able and, most important, the studies resultsare very sensitive to it.

Alternatively, studies using a narrative ap-proach based on historical and contemporaryaccounts of U.S. military buildups to iden-tify government spending shocks find thatthe short-run effects of stimulus spending aresmall. These latter studies are grounded in his-torical reality and are thus more compelling.

Why in the World Are WeAll Keynesians Again?

The Flimsy Case for Stimulus Spending

by Andrew T. Young

No. 721 February 14, 2013

Andrew T. Young is an associate professor of economics at West Virginia University and the Colleges BB&TScholar. His papers have been published in Public Choice; Journal of Money, Credit and Banking;Review of Economics and Statistics; Managerial and Decision Economics; Review of EconomicDynamics; andReview of Austrian Economics.


2/20

2

The return tofashion of fiscal

stimulus wasbased more on

wishful thinkingthan any new

preponderanceof evidence.

Introduction

In 1965 Milton Friedman was famously(or infamously, depending on your point ofview) quoted as saying We are all Keynes-

ians now.

1

Though he later remarked thathe had been taken out of context, Friedmanaccurately expressed a consensus amongeconomists and policymakers at the time:countercyclical fiscal policy is an appropriateand effective means toward combating thebusiness cycle. Specifically, in response to arecession, the federal government should in-crease its expenditures and/or provide tax re-lief to make up for the shortfall in the privatesectors aggregate demand for goods and ser-vices. Nearly three decades after his publica-

tion of The General Theory of Employment, In-terest, and Money, the policy prescriptions ofLord Keynes were dogma.

However, by the beginning of the new mil-lennium all that had changed. Countercycli-cal fiscal policy had fallen into disrepute. Anew consensus held that fiscal stimulus inthe wake of a recession was politically in-feasible and, even if it could be enacted in atimely fashion, stabilization policy was likelyto be ineffective. Alan Blinder, the formervice chairman of the Federal Reserves Board

of Governors, went so far as to say that vir-tually every contemporary discussion of sta-bilization policy by economistswhether itis abstract or concrete, theoretical or prac-ticalis about monetary policy, not fiscalpolicy.2

But it would seem that the fashionable-ness of countercyclical fiscal policy is, well,countercyclical. Beginning in late 2007, re-cession and financial crisis led to a rehabili-tation of Keynesian fiscal policy. Under theGeorge W. Bush administration, $152 bil-

lion of tax rebates and other relief was passedin February of 2008. Almost exactly a yearto the day later, President Barack Obamasigned the $831 billion American Recoveryand Reinvestment (ARRA) stimulus packageinto law.3 Christina Romer, at that time thechair of Obamas Council of Economic Ad-visers, remarked that the ARRA was simply

the biggest, boldest countercyclical stimulusin American history and that she firmlybelieve[ed] that fiscal policy will have a cru-cial beneficial impact.4 Pro-stimulus aca-demic macroeconomists Antonio Fats and

Ilian Mihov declared that, properly execut-ed empirical studies support the view that afiscal expansion will help the U.S. economyrecover.5 And even after nearly two years ofcontinued unemployment above 9 percentNobel laureate Paul Krugman opined thatthe ARRAs only problem was that it wasmuch too small.6

But the return to fashion of fiscal stim-ulus was based more on wishful thinkingthan any new preponderance of evidenceNational Bureau of Economic Research

(NBER) macroeconomist Jonathan Parkernoted that when the Obama administra-tion turned to economistssignificantlyacademic economiststo help craft the sizeand details of the [ARRA] stimulus pack-age what they provided was a wide rangeof answers . . . almost as useless as no an-swer. Indeed, the Cato Institute published afull-page New York Times ad arguing againstgovernment spending stimulus and carry-ing the signatures of over 200 economists,including three Nobel laureates.7 Lacking

any consensus from the experts, policy-makers resorted to using a multiequationmacroeconomic forecasting model, one in-consistent with the best practice of modernmacroeconomics.8 Less than 15 years afterRobert Lucas won the Nobel Prize in eco-nomics for forcefully arguing against theappropriateness of using such large scalemacroeconometric models to design policy,the Obama administration placed an $800billion bet that Lucas had been wrong.

This paper evaluates, particularly in re-

gard to government spending, what confi-dence we should have in fiscal stimulus poli-cies to restore the U.S. economy to robustexpansion. My focus on spending is to makethe subject manageable and also reflects thecurrent administrations demonstrated pref-erence for government spending relative totax relief. As I will argue, there is noconsensus


3/20

3

The governmentspendingmultiplier issignificantlyless than one.

that fiscal stimulus spending is at all effective. In-deed, while one can find studies to supportalmost any position in the policy debate,the most compelling efforts to isolate exog-enous changes in government spending and

estimate their effect on gross domestic prod-uct (GDP) report that each additional dollarof stimulus yields less than one additionaldollar in total GDP. In the terminology ofeconomists: the government spending multiplieris significantly less than one. Increases in govern-ment expenditures are associated with de-creases in private spending. The net changein total expenditures, then, falls short of thefiscal stimulus itself.

The debate over the appropriateness ofKeynesian stimulus spending is ongoing,

and the potential consequences are large. De-spite virtually no prima facie evidence thatthe ARRA has been effective in restoring ro-bust economic growth and full employment,the Obama administration is prepared toroll the dice again. Since September of 2011,the administration has been advocating anadditional $447 billion of stimulusabouthalf of which would be new government ex-pendituresunder the proposed AmericanJobs Act.9 With the U.S. federal governmentdebt already standing at more than 70 per-

cent of our GDP, we should demand criticalevaluation of such a proposal from our poli-cymakers.

In what follows I begin by putting recentfiscal stimulus spending in quantitativeperspective relative to recent U.S. history.At least relative to recent U.S. experience,the ARRA was a very large spending pack-age. Next, I discuss the intuition behind theconcept of a government spending multi-plier. Deciding on the appropriateness offiscal stimulus hinges on estimating the

size of this multiplier using U.S. data. I thendiscuss the problems involved in estimat-ing the government spending multiplier. Ireview the existing estimates and concludethat while a wide range of estimates ex-ist, the most plausible of those are signifi-cantly less than one. Last, I note that evenadvocates of fiscal stimulus, such as Fats

and Mihov, admit that the long-run con-sequences for the national debt are oftendire. Indeed, the findings of their own aca-demic research suggest that the long-runcosts are ominous. It is for these (negative)

consequences that we truly have consensusamongst macroeconomists.

Recent Fiscal Stimulus inPerspective

Paul Krugman criticized the ARRA, andin particular its government spending com-ponent, for not being big enough to do thejob.10 This certainly could be true and it can-not simply be dismissed. What I argue here

is that, relative to recent U.S. experience, thenew government spending initiated by theARRA was quite large. This cannot rule outthe possibility that countercyclical respons-es of U.S. government spending have been ingeneraltoo small. However, it does move thediscussion toward something more concretethan the otherwise vague characterizationsof too big or too small.

Figure 1 plots the quarterly percentagedeviations of real (i.e., inflation-adjusted)U.S. federal government expenditures from

their trend values. The basic data are 1947 to2012, and a HodrickPrescott filter was usedto compute a smoothly evolving trend.11 Tolend some historical perspective, verticallines indicate the NBER beginning-of-re-cession dates and shaded areas correspondto the Korean War years as well as years ofmajor U.S. involvement in the Vietnam War.A horizontal line at just over 40 percent cor-responds to the peak deviation in the wakeof the post-2008 fiscal stimulus policies.Outside of the large and positive deviations

in federal spending during buildups for ma-jor military efforts, the increased spendingthat peaks in 2010 is extraordinary. Noth-ing like it has been previously seen duringpeacetime.

Additional perspective based on nonmili-tary federal government expenditures is giv-en in Figure 2. The ratio of federal nonmili-


4/20

4

tary expenditures to GDP is plotted, againon a quarterly basis, from 1947 to 2012. TheKorean War and Vietnam War eras are againshaded; a vertical line marks the beginning ofthe recent recession in the fourth quarter of2007. The horizontal line in this case marksthe 2010 peak of the series at over 2.7 per-

cent of GDP. This is the highest since 1992.More important, measured from 2007, it isthe largest single increase since the 1960s.In general, aside from the upward spike in1992, nonmilitary expenditures as a percentof GDP had been trending downward sincethe 1970s. The ARRA and plans for addi-tional federal stimulus, along with the loom-

ing entitlements associated with Obamacaremay permanently reverse that trend.

From Figures 1 and 2 we can see that thefiscal stimulus in the wake of the Great Re-cession was remarkable. Federal governmentspendingeven nonmilitaryhas certainlybeen a larger share of the U.S. economy at

times. However, in terms of fiscal stimuluspolicy what matters is the increase in ex-penditures. The federal governments fiscalresponse to the recession reversed basicallyall of the decreases in nondefense spendingthat had been achieved during the years ofthe Clinton administration and the Con-tract with America Congress.

60

80

100

120

0

20

40

80

60

40

120

100

Figure 1Deviations of Real Federal Government Expenditures from Trend, 19472012

Source: U.S. Bureau of Economic Analysis. The federal component of government consumption expenditures, in billions of seasonally adjustedchained 2005 U.S. dollars. Observations are quarterly but stated at annual rates. Federal expenditures are detrended using the filter provided bRobert J. Hodrick and Edward C. Prescott, Postwar U.S. Business Cycles: An Empirical Investigation,Journal of Money, Credit and Banking29, no. 2

(February 1997): 116. The smoothing parameter value is set according to Morten O. Ravn and Harald Uhlig, On Adjusting the Hodrick-PrescotFilter for the Frequency of Observations,Review of Economics and Stati stics 84, no. 5 (May 2002): 37176. Shaded areas mark the Korean War andmajor U.S. involvement in the Vietnam War. Vertical lines mark National Bureau of Economic Research beginning-of-recession dates.


5/20

5

Fiscal Multipliers

Of course, if you believe in the Keynesianmodel then a remarkable increase in govern-ment expenditures is precisely what the doc-tor ordered for the slumping U.S. economy.Furthermore, you hope that the effect on to-tal spending in the economy is even more re-markable than the initial fiscal stimulus. The

whole point of stimulus policy is tostimulateincreases in total spending and incomes be-yond the size of the stimulus package itself.In other words, the idea is for the initial injec-tions into the economy to multiply. Whetherthe size of this effectthe multiplieris largewill depend on several factors, including in-dividuals marginal propensity to spend out

of their incomes and their expectations offuture tax liabilities, the extent to which def-icit-financed government spending raisesinterest rates and crowds out private spend-ing, and the amount of slack in the econo-my. These factors are not independent fromone another. However, understanding eachof their separate roles will facilitate thinkingabout why fiscal stimulus may or may notbe effective.

Before considering each of the factors inturn, it will be helpful to place fiscal stimu-lus in the context of a simple framework. Aswould be common in an introductory mac-roeconomics course, think of an economystotal expenditures as the sum of three com-ponents: private consumption expenditures,private investment expenditures, and gov-

0.04

0.044

0.032

0.036

0.02

0.024

.

0.016

1947 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 2002 2007 2012

Figure 2Ratio of Federal Non-Military Expenditures to GDP, 19472012

Source: U.S. Bureau of Economic Analysis. The federal component of government consumption expenditures minus national defense expenditures divided by real gross domestic product, in billions of seasonally adjusted chained 2005 U.S. dollars. Observations are quarterly but statedat annual rates. Shaded areas mark the Korean War and major U.S. involvement in the Vietnam War. Vertical line marks National Bureau ofEconomic Research beginning-of-recession date in the fourth quarter of 2007.


6/20

6

Increases ordecreases ingovernment

spending andtaxes can be

contemplatedand enacted

independently ofone another.

ernment expenditures. There are also expen-ditures associated with the import of foreigngoods and the export of our own goods toforeigners, but we can have a meaningful dis-cussion of fiscal policy without considering

them explicitly. None of the fundamentalideas and intuitions below would be changedby doing so, and focusing on consumption,investment, and government expenditureskeeps it simple.

The choices by individuals and policymak-ers that result in consumption, investment,and government expenditures are collec-tively referred to as the economys aggregatedemand for goods and services. The point offiscal stimulus is to provide initial increasesto aggregate demand. This is why such a

policy to combat the business cycle if oftenreferred to as aggregate demand management.The initial increases in aggregate demand areeither directly from policymakers increasingthe amount of government expenditures, orindirectly from their decreasing taxes. In thelatter case, by decreasing the amount of taxescollected from individuals and businesses inthe economy, net (or disposable) private in-comes are higher. The hope is that individu-als and businesses, finding themselves withmore of their incomes and profits left over

after paying taxes, increase their private con-sumption and investment expenditures.

Because the government has recourse tofinancial markets and can run deficits or(much less commonly in recent U.S. histo-ry) surpluses, a balanced-budget constraintis not binding when policymakers considershort-run fiscal policy changes. Policymak-ers can borrow to finance a deficit or usesurplus revenues to buy financial assets (in-cluding the governments own outstandingdebt).Therefore, increases or decreases in

government spending and taxes can be con-templated and enacted independently of oneanother. For example, the ARRA providedfor both increases in government spendingand decreases in taxes. This, of course, wouldnot have been possible if legislators wereconstrained by law or necessity to keep thefederal governments budget balanced.

The Marginal Propensity to ExpendBut the initial increases in expenditures

need not be the whole story; indeed, theyare almost certainly not. Any expendituremade in the economy generates income for

some individuals. Those individuals thenface a decision about what to do with theincrease in their incomes. To spend or notto spend? That is the question. At the heartof multiplier effects, then, are individualsmarginal propensities to expend. The mar-ginal propensity to expend is the percent ofan additional (or marginal) dollar of incomereceived that will be spent rather than saved.

Consider, for example, that the federalgovernment undertakes a new highwaymaintenance program by spending $10 mil-

lion. Government spending (and aggregatedemand) immediately increases by $10 mil-lion. But when that money, it creates $10 mil-lion in new income for asphalt sellers, roadcrews, etc. These recipients of new incomecan save all or some of it, and then spendthe rest. What they spend leads to a furtherincrease in aggregate demand through anincrease in private consumption expendi-tures and/or investment expenditures. If, forexample, the marginal propensity to spendis 0.8, then individuals spend 80 percent of

their new incomein this case, $8 millionand save the rest. In turn, that $8 million ofnew spending creates $8 million of new in-come for other individuals. If a marginal pro-pensity to spend of 0.8 is typical, these otherindividuals also spend 80 percent of theirnew incomes, or $6.4 million. And the pro-cess continues as yet others receive the $6.4million as income. With each round of newincome generated, the dollar value of new ex-penditures falls because some part of that in-comein this example, 20 percentis saved

rather than spent.If one wants to know, for a given mar-

ginal propensity to expend, the total amountof new expenditures that can result from aninitial fiscal stimulus, a bit of simple math ishelpful. Denote the typical Americans mar-ginal propensity to spend byMPS. We candefine the government spending multiplier


7/20

7

If the decreasein privateexpendituresexactly matches

the newgovernmentexpenditures,then the negativand positivemultiplier effectwill exactlyoffset. Themultiplier willbe zero.

as 1/(1MPS). It is called a multiplier becauseit tells us by how much some initial amountof new government expenditures (G) canmultiply in total new expenditures (Y). Theprocess described in the previous paragraph

can be expressed mathematically as

Y= (1/(1MPS)) G.

Plug in the initial stimulus (G) on the righthand side of the equation and, given peo-ples tendency to spend out of additional in-come (MPS), one can easily solve for the totalamount of new spending and incomes (Y)that can be generated.

Now, lets go back to our example wherethere is an initial stimulus in the form of

$10 million of new government spending onhighway maintenance. If theMPS is 0.8 thenthe government spending multiplier is five.Each dollar of fiscal stimulus can yield $5 oftotal increase in aggregate demand. A $10million highway maintenance program willresult in $50 million in total new expendi-tures and income. If, alternatively, the MPSis 0.9, then the multiplier is 10, and there istwice the bang for the stimulus buck. Intui-tively, if at each round of the process, indi-viduals spend a greater portion of their new

incomes, then a greater portion of their newincomes subsequently becomes new incomefor additional individuals.

Note that for a given marginal propensityto expend, the government spending mul-tiplier tells us how much a dollar of initialstimulus can yield in terms of total new ex-penditures; not necessarily how much it willyield. Most U.S. citizens spend more than80 percent of the income that they receive.U.S. personal savings rates are actually quitelow, so the marginal propensity to expend

is likely to be quite high. However, not eventhe most ardent advocates of fiscal stimulusbelieve that the government spending mul-tiplier is five or more. Typically, studies citedin support of fiscal stimulus report govern-ment spending multipliers above one butnot more than two.12 A multiplier of twowould correspond to a marginal propen-

sity to spend of 0.5. Personal savings ratesin the U.S. are nowhere near 50 percent, sofor the empirical multiplier to be two or lessthere must be other factors associated with astimulus that cause counteracting decreases

in private consumption and investment ex-penditures.13 In other words, there must beother important considerations not includ-ed in the simple Keynesian model.

Expectations of Future Tax Liabilitiesand Ricardian Equivalence

The federal government can certainly in-crease its expenditures beyond its revenuesin the short run, but this does not mean thatthe bills will not come due. The governmentruns spending deficits by borrowing; it sells

its treasury securities on financial markets toU.S. and foreign individuals. Those treasur-ies represent promises of future paymentsfrom the federal government (amountingto an outstanding total of about $16 trillionas of late). The government will eventuallyhave to raise revenue with a present valueequal to those liabilities.

Macroeconomist Robert Barro put forththe argument in the 1970s that governmentspending multipliers will be close to zeroprecisely because when government runs

a deficit, taxpayers will rationally add thevalue of that deficit to their expected futuretax liabilities.14 In anticipation of having tomake good on these liabilities, they increasetheir savings, which implies that theydecreasetheir expenditures. If the decrease in privateexpenditures exactly matches the new gov-ernment expenditures, then the negative andpositive multiplier effects will exactly offset.The multiplier will be zero. For every deficit-financed dollar of government expenditures,taxpayers hold a dollar back in anticipation

of having to pay higher taxes in the futureto retire the new government debt. In a ref-erence to the classical economist David Ri-cardo, who in the 1800s discussed similarinsights in relation to British war bonds,the hypothesis that taxpayers will internal-ize governments intertemporal budget con-straint is called Ricardian equivalence.


8/20

8

Politicalvariables, such

as U.S. House ofRepresentatives

committeememberships,

were betterpredictorsof funding

levels from thestimulus.

Just as even the most ardent advocates ofKeynesian stimulus do not believe that thespending multiplier is five, not even RobertBarro claims that the short-run governmentspending multiplier is zero. A year after the

ARRA, Barro stated that his best reading ofthe evidence and recent experience suggests amultiplier of 0.4 in the first year and 0.6 overtwo years.15 A multiplier significantly lessthan one means that every $1 of governmentstimulus buys the economy less than a $1 in-crease in total income. If the spending multi-plier is 0.6, for example, the ARRAs approxi-mately $500 billion in new spending boughtonly about $300 billion in total new expendi-tures and income; if the multiplier is 0.4, onlyabout $200 billion. One doesnt need to be an

economist to see how thats a raw deal.

Crowding OutTo believe that there may be spending de-

flators, rather than multipliers, associatedwith stimulus spending, one does not haveto believe that U.S. citizens are experts onpublic finance, poring over CongressionalBudget Office projections and forming ra-tional expectations of their future tax liabili-ties. Largely one just has to believe that theyrespond to market prices and, in particular,

market interest rates.When the federal government pursues

spending in excess of its current tax reve-nues, it has to turn to financial markets andcompete with private borrowers for funds.The increased demand for funds will, all elseequal, put upward pressure on interest rates,making borrowing more costly. By raisingthe cost to private borrowersboth indi-viduals and businessesgovernment deficitspending tends to crowd out private invest-ment expenditures and consumption ex-

penditures that are sensitive to interest rates(e.g., car purchases).

Furthermore, higher costs of borrowingalso mean higher returns to saving. As thegovernment increases its demand for funds,this must be satisfied by increases in privatesavings. This also discourages private con-sumption expenditures.

Targeting Stimulus to Where TheresSlack

Another factor upon which the effectiveness of fiscal stimulus depends is theamount of slack in the economy where it

is targeted. This is acknowledged in the stan-dard textbook exposition of the Keynesianmodel where fiscal policy is conceived of asshifting aggregate demand along the short-run aggregate supply curve (SRAS) (Figure3). The SRAS curve represents firms willing-ness to produce various amounts of goodsand services (Y) at various levels of prices (P)As the economy approaches full employmentand its potential level of output, the slope ofthe SRAS curve increases, approaching verti-cal. This implies that when the economy is

running on all cylinders increases in aggre-gate demand will largely translate into ris-ing prices (inflation) rather than higher pro-duction. Alternatively, the flatter portion ofSRAS represents levels of (Y) considerably be-low potential. There is considerable slack interms of the available capital and labor, andincreases in aggregate demand can prime thepump and bring those resources into use inproduction of good and services.

Obviously the ARRA was enacted duringa time of considerable slack in the economy.

Unemployment peaked at 10 percent in Julyof 2009 and remained (and remains) stub-bornly above 8 percent. However, the re-cession did not hit all areas of the countryequally. Considering the allocation of fundsacross U.S. states and based on availablemeasures of slack (e.g., unemployment ratesand state GDP levels), my own research withRuss Sobel indicates that the ARRA has notbeen consistently targeted to areas operatingbelow potential. (Nor, for that matter, wasit targeted to states where the marginal pro-

pensity to expend is particularly high.) Rath-er, political variables, such as U.S. House ofRepresentatives committee membershipswere better predictors of funding levels fromthe stimulus. And the strongest and mostrobust predictor of a states ARRA alloca-tion is the ratio of a states federal grants andpayments to its federal taxes paid previous


9/20

9

In the realeconomy,stimuluspackages areproducts of thepolitical process

to the stimulus. In other words, states thatpreviously captured large amounts of federalfunds also managed to capture larger por-tions of the ARRA stimulus.16

While my research with Sobel does notspeak directly to the effectiveness of fiscalstimulus, it serves as a reminder that fiscalpolicy is never a matter of simply increas-

ing spending or decreasing taxes. In the realeconomy, stimulus packages are products ofthe political process and objects of the stan-dard rent-seeking games that politicians andspecial interests play. This not only bringsinto question whether stimulus spendingwill be well-targeted to areas of slack in theeconomy. It should also lead us to question

the non-stimulus value of the investmentsthat are being made and services that are be-ing provided. Even if a rush to find shovel-ready projects does not itself lead to unwiseexpenditures, the political competition tocapture funds and please constituents islikely to.

The (Lack of) Evidence

Finally we come to the big question: DoesKeynesian fiscal stimulus actually work? Isthe spending multiplier greater than one?My own reading of the literature draws fivegeneral conclusions:

Y

P

SRAS

AD AD*

Y Y*

AD

AD*

Y Y*

stimulus stimulus

Figure 3The Effect of Stimulus in the Keynesian SRAS/AD Mode Is Contingent on HowClose the Economy Operates to Its Potential

Source: Authors creation; standard aggregate supply/aggregate demand graph.


10/20

10

Even though theU.S. economy

has not roaredback following

the AmericanRecovery andReinvestment

Act, one mightargue that it

would be doingfar worse now

had it notbeen for the

federal stimulusspending.

There are numerous studies findingthat the U.S. government spendingmultiplier is greater than one; there arealso numerous studies finding that itis less than one.

The primary challenge tackled by all ofthese studies is to identify exogenouschanges in government spending andthen trace out their effects on totalspending over time.

The results are very sensitive to howthese exogenous government spend-ing changes (or shocks) are identified.

The most plausible identificationstrategiesbased on increases in gov-ernment spending associated withmilitary buildupsresult in multipli-

er estimates that are less than one. Even researchers who report largespending multipliers acknowledge thatthe long-run costs of deficit-financedfiscal stimulus are likely to be large.

At best, our confidence in the effectivenessof fiscal stimulus should be small relative tothe price tags attached to a package like theARRA. At worst, I will argue contra AntonioFats and Ilian Mihov that studies with thegreatest claim to being properly executed

suggest that a stimulus bang is quite a bitsmaller than its buck.

Identifying Exogenous Changes inSpending

Even though the U.S. economy has notroared back following the ARRA, one mightargue that it would be doing far worse nowhad it not been for the federal stimulusspending. Fiscal stimulus may get us a pe-riod of weak growth and employment in ex-change for one where we plunge into a deep

depression. Alternatively, if stimulus pack-ages tend to be passed late in a recessionorafter the recession has actually endedthenthey will tend to be positively correlated withgrowth and rising employment. In eithercase, we will not be getting a clear picture ofthe macroeconomic effects of the stimulusitself.

Both of the above scenarios arise becausefiscal stimulus is what economists refer toas an endogenous event. Stimulus funds donot randomly fall onto the economy like somuch manna from heaven. Rather, they are

the result of legislation that is designed be-cause of and in response to macroeconomicconditions. We cannot just look at subse-quent economic events and determine thatthey were caused by a stimulus, because astimulus is itself caused by the unfoldingchain of economic events.

What researchers would like to identifyare exogenous changes in government spend-ingchanges that were not caused by un-folding events in the economy. Economistsrefer to such changes as shocks since they

arise from outside of the specific system ofvariables that one wants to analyze. Tracingout the effects of such shocks to governmentspending is the ideal way to estimate thegovernment spending multiplier. But iden-tifying such shocks is easier said than doneFiscal stimulus is countercyclical policy; itis pretty much by definition endogenous tothe business cycle, and so are variations inspending associated with automatic stabiliz-ers such as unemployment insurance thatby design, increase their payments when eco-

nomic activity falls. Other legislated changesin government spending are also generallyinfluenced by economic conditions.

Economists who study the effects of fis-cal stimulus have addressed this challengingproblem using one of two approaches (or acombination of the two). The first approachinvolves the estimation of structural vectorautoregressions (SVARs). A vector autore-gression (VAR) is a system of equations thatdescribes how certain economic variables(e.g., GDP and government spending) evolve

over time and affect one another. In regardto the government spending multiplier, aresearcher will be interested in using macro-economic data to estimate a VAR and quan-tify the effects of a shock to governmentspending on GDP over time.

The challenge in working with VARs is toidentify the shocks (or exogenous changes)


11/20

11

The results ofestimating astructural vectorautoregressionare very sensitivto the particularordering thatone assumes.

to government spending. A researcher can-not simply examine all changes to govern-ment spending. Many of those spendingchanges will be the result of (at least in part)changes in economic activity. To the extent

that the government spending changes areassociated with changes in GDP over time,the researcher will be unable to tell to whatextent those GDP changes are caused by thegovernment spending rather than merely as-sociated with previous changes in economicactivity. The researcher wants to identify ex-ogenous shocks to government spending sothat he or she can be confident that the es-timated effects on GDP are the result of thestimulus.

To solve this identification problem, a re-

searcher must impose additional structureon the VAR (hence, an SVAR). A commonstrategy is to order the shocks of the differ-ent variables included in the SVAR. Specifi-cally, a researcher will assume that shocks togovernment spending can affect economicactivity and GDP immediately, but not viceversa. The technical details about the esti-mation of SVARs would be inappropriatehere, but a simple example is provided inthe appendix of this paper for the interestedreader. What is important to note is that the

results of estimating a SVAR are very sensi-tive to the particular ordering that one as-sumes.

As means of illustration, I estimated a sim-ple bivariate SVAR in U.S. real GDP and fed-eral government spending. I included bothvariables in equations containing six laggedvalues of each. The data I used to estimatethe system is quarterly data for both GDPand government expenditures from 1947 Q3through 2012 Q1, and I took the natural logsof both series, as is standard in most stud-

ies. Figure 4 presents the resulting impulseresponses of U.S. real GDP to a federal govern-ment spending shock. An impulse responseplots how a shock to one variable affects an-other variable in the system over time. Theimpulse response in panel (a) is computedassuming that government spending is or-dered first; the one in panel (b) is computed

assuming, instead, that real GDP is orderedfirst. The difference is like night versus day.In panel (a), a shock to government spendinghas positive effects on GDP for a full year.In panel (b), the effect of a federal spending

shock on GDP is immediately zero and thenactually turns negative after another quarterpasses.

Numerous papers estimating the govern-ment spending multiplier using U.S. dataemploy a SVAR approach based on the or-dering and placement of other restrictionson the shocks. For example, papers by An-tonio Fats and Ilian Mihov in 2001andOlivier Blanchard and Roberto Perotti in2002 assume that government spending ispredetermined with respect to macroeco-

nomic shocks. In other words, the shockto government spending is ordered first; ina given quarter only it, and not exogenouschanges to GDP, will increase or decreasegovernment spending. This is precisely thetype of assumption that I made in comput-ing the panel (a) impulse response whereGDP responds positively to a governmentspending shock. According to Fats and Mi-hov, this assumption is a plausible, but un-fortunately untestable hypothesis.17 Theyreport that the U.S. government spending

multiplier is greater than one. A 2007 paperby macroeconomists Jord Gali, Javier Valls,and J. David Lpez-Salido echoed these ear-lier findings, reporting that the governmentspending multiplier reaches 1.74 two yearsafter a spending shock.18

My own little SVAR analysis (Figure 4) is anaive one and, admittedly, does little justiceto the published studies by other scholars.For instance, other variables in addition togovernment spending and GDP should beincluded in any serious study of the effects

of fiscal stimulus. However, my simple ex-ample highlights the point I wish to makehere: the results are very sensitive to thestructure that one chooses to impose on theSVAR. And while the predetermined natureof government spending certainly is untest-able, there are reasons to doubt whether ornot it is really all that plausible. First and


12/20

12

0.001

0.002

0.002

0.001

0.003

1 2 3 4 5 6 7 8

0.001

0

0.001

0.003

0.002

0.004

1 2 3 4 5 6 7 8

Figure 4Estimated Effects of Real GDP to a Federal Government Expenditures Shock

Source: Source: Authors calculations.Note: Both impulse responses are based on a VAR incorporating quarterly real GDP and federal governmentexpenditures from the BEA for 1948 Q3 to 2012 Q1. Six are lags are assumed for each equation and natural logare taken of each variable. In the left panel (a) federal spending is ordered prior to real GDP; in the right pane(b) real GDP is ordered first. Depending on which restriction is assumed, the estimated response of real GDP toa federal spending shock is (a) positive for an entire year or (b) always negative. The vertical axis measures thchange in the natural log of real GDP to a standard deviation innovation to the natural log of federal government expenditures.

Quarters

Quarters

(a)

(b)


13/20

13

Using themost generousestimate, thisimplies thatfor each $1 offederal stimulusspending,$0.20 of privateexpenditures iscrowded out, anthe total increasin GDP will beless than thenew governmenexpenditures.

foremost, it assumes that policymakers havea lack of access to, or an ignorance of, eco-nomic data that are being collected through-out a given quarter. (This seems especiallyimplausible when using data samples that

overlap the Internet age.) Also, in any SVARstudy there are numerous variables associ-ated with economic activity that are left outof the analysis. If those variables affect bothGDP and policymakers behavior in a giventime period, then shocks that are identifiedas directly occurring to GDP may also af-fect government spending. For example, as-sume that economic conditions deterioratein the European Union. This could certainlyimpinge on U.S. economic activity, and U.S.policymakers are cognizant of this; their leg-

islative activities could then be directly influ-enced by the overseas developments.When researchers abandon the predeter-

mined government spending assumption infavor of alternative identification strategies,the results are drastically different. For exam-ple, in a 2009 paper Andrew Mountford andHarald Uhlig identified government spend-ing shocks as being associated with increasesin spending for at least a year and businesscycle shocks as ones that caused governmentrevenue, GDP, consumption, and invest-

ment to move in the same direction. They re-port a U.S. government spending multiplierof merely 0.07 after two years.19 (The largesteffect that they report is in the first year, andeven then, the spending multiplier is only0.65.)

Evidence Based on Variation in MilitarySpending

An alternative approach to identifyinggovernment spending shocks was pioneeredby Valerie Ramey and Matthew Shapiro in

a 1998 paper. They adopted a narrative ap-proach and use historical accounts and con-temporary Business Week stories to isolateepisodes of postWorld War II U.S. militarybuildups. As Ramey and Shapiro explain,military buildups occur rapidly and unex-pectedly, so they are naturally modeled asshocks[;] they are driven by imperatives of

foreign policy. . . . [B]ecause they are drivenby geopolitical shocks, military buildups arelikely to be exogenous with respect to mac-roeconomic variables. 20

Researchers following the Ramey and

Shapiro approach report modest spendingmultipliers. Furthermore, they find thatprivate consumption expenditures tend tofall in the wake of a government spendingshock.21 In a 2008 paper, Valerie Ramey re-ports that the postWorld War II U.S. spend-ing multiplier is between 0.6 and 0.8.22 Us-ing the most generous estimate, this impliesthat for each $1 of federal stimulus spend-ing, $0.20 of private expenditures is crowdedout, and the total increase in GDP will be lessthan the new government expenditures. The

least generous estimate implies that the totalincrease in GDP is only 60 percent of the newgovernment expenditures.

The narrative approach based on militarybuildups is by no means a perfect identifi-cation strategy, but it is in at least two waysappealing relative to the SVAR approach.First, shocks to government spending be-come grounded in historical reality. Theyare not simply fluctuations in spending thatare identified when an SVAR is estimated;they rather correspond to historical episodes

where the motives of policymakers can be de-scribed, albeit imperfectly. Second, advocatesof federal stimulus spending often expressa go big or go home approach. Militarybuildups are large increases in governmentspending that are comparable to fiscal stim-ulus packages.

In the Long Run We Are All in DebtThe evidence on the size of the U.S. gov-

ernment spending multiplier is mixed, atbest, and I have argued above that the most

plausible evidence suggests a multiplier sig-nificantly less than one. Meanwhile, the se-verity of the long-term consequences of theARRA seems clear. Figure 5 demonstrates themassive increase in federal government debtheld by the public as a percent of U.S. GDP.In 2007 Q4, this ratio stood at about 0.36. By2012 the federal debt was more than 70 per-


14/20

14

Pro-stimuluseconomists

Antonio Fatsand Ilian Mihov

report thatgovernments

that usefiscal policyaggressively

induce significantmacroeconomic

instability.

cent of the U.S. GDP. The debt-to-GDP ratiohas almost doubled in less than five years; ithas increased by more than 160 percent since1970. The ARRA is clearly not the only causeof the debt explosion. For example, anotherimportant contributor is increasing Medi-care expenditures. However, the $800 billionplus ARRA has exacerbated the problem, aswill any of the proposed subsequent stimu-lus spending.

Whatever its sources, all of this debt en-tails future tax liabilities. While some maydoubt that U.S. citizens always internalizethe federal governments long-run budgetconstraint, it is also hard to believe that suchlarge and looming liabilities will not serve todampen private economic activity. Further-more, there is considerable uncertainty as to

exactly how and when the necessary futurerevenues will be raised. Uncertainty is anath-ema to vibrant, future-oriented business ac-tivity.

Remarkably, in a 2003 study of data from91 countries, pro-stimulus economists Anto-nio Fats and Ilian Mihov report that gov-ernments that use fiscal policy aggressivelyinduce significant macroeconomic instabil-ity.23 Even more remarkably, looking back

on that research in a 2009 paper titled WhyFiscal Stimulus is Likely to Work, they statethat changes in fiscal policy, present and fu-ture, will create instability in the macroeco-nomic environment that can lead back to alack of confidence, uncertainty and hurt in-vestment and growth. . . . But our view is thatthe benefits of a well-executed short-term

0.7

0.8

0.6

0.4

0.5

0.2

0.3

1970 1975 1980 1985 1990 1995 2000 2005 2010

Figure 5Ratio of Federal Debt Held by the Public to GDP, 19702012

Source: Federal Reserve Bank of St. Louis Federal Reserve Economic Database, using federal debt held by thpublic divided by nominal GDP. Vertical line corresponds to 2007 Q4, the beginning of the Great Recessionaccording to National Bureau of Economic Research business cycle dates.


15/20

15

The evidence onthe effectivenessof stimulusspending is

weak.

fiscal stimulus are larger than the long-termcosts.24 Their sanguinity concerning the sizeof the short-run government spending mul-tiplier is unjustified given the existing evi-dence, especially given that no one seriously

disputes the long-run burden that legislationlike the ARRA places on the economy.

Conclusion

Why in the world are we all Keynes-ians again? The stimulus spending of theAmerican Recovery Reinvestment Act hasbeen an apparent failure. The economy hasnot returned to a path of robust economicgrowth. Unemployment has remained stub-

bornly above 8 percent. Yet many pundits,policymakers, and academics continue toencourage additional increases in govern-ment spending. For some, like Nobel laure-ate Paul Krugman, the only problem withthe ARRA is that even more than $800 bil-lion was simply not large enough. Their so-lution, therefore, is to argue for additionaland larger stimulus packages.

But the evidence on the effectiveness ofstimulus spending is weak. Studies findinga government spending multiplier greater

than one typically rely on identifying gov-ernment spending shocks by assuming thatgovernment spending does not respond tocontemporaneous changes in economic ac-tivity. However, if that assumption is incor-rect, then those researchers are attributingindependent changes in GDP to changes ingovernment spending. Data on underlyingactivity is readily available to policymakerswithin a given quarter. Also, factors such aschanges in other countries economic con-ditions can easily influence both domestic

economic activity and fiscal policy decisions.For both reasons, the critical assumption ofthese studies is likely to be violated.

I have argued above that studies identify-ing shocks with the changes in governmentspending associated with U.S. military build-ups are more plausible. These studies reportgovernment spending multipliers less than

one. They find that private consumption ex-penditures fall in the wake of a shock. This isconsistent with the insights of Robert Barroconcerning Ricardian equivalence. Individu-als in the economy internalize the govern-

ments intertemporal budget constraint.Responding to new future tax liabilities andhigher interest rates, individuals offset newgovernment spending, at least in part, by de-creasing their own private expenditures.

In the above discussion, I have focusedentirely on government spending as op-posed to tax cuts. This is largely becauseboth the ARRA and the policy proposals ofthe Obama administration moving forwardhave been based on the premise that the fed-eral government can spend its way out of the

economic malaise. Also, the same structuralvector autoregression studies that reporta government spending multiplier greaterthan one find that the U.S. tax multiplier isless than one.25

However, a 2010 paper by Christina Romerand David Romer reports a tax multiplier ofnearly three.26 Romer and Romer pursue anarrative approach to identifying tax shocksthat is similar in spirit to Ramey and Shap-iros for identifying spending shocks. Romerand Romer utilize U.S. presidential speeches

and congressional reports to provide evi-dence of tax changes that were not motivat-ed by changes in macroeconomic conditions.(Given that Christina Romer is on record asan enthusiastic architect and proponent ofthe ARRA, it is interesting to note that thepapers discussion of its findings refers onlyto tax increases and their contractionary ef-fects; never the possible stimulating effectsof tax cuts.) To the extent that the narrativeapproach is more compelling than SVARs,Romer and Romers evidence suggests that

tax cuts are relatively more effective as fiscalstimulus than are government spending in-creases.

Stimulus packages like the ARRA comewith staggering price tags, and even theirproponents admit that they result in pub-lic debt burdens that are likely to dampengrowth in the future and destabilize the


16/20

16

It is hard tojustify stimulus

spendingpackages in light

of the almostcertain and large

long-run costs.

macroeconomy. Given weak evidence ofsignificant short-run benefits, it is hard tojustify stimulus spending packages in lightof the almost certain and large long-runcosts. In September of 2011, the Obama ad-

ministration advocated an additional $447billion of stimulus under the proposedAmerican Jobs Act. The large majority ofthis proposed legislation was blocked byHouse Republicans.27 However, the contin-

ued stagnation of U.S. economic growthand employment, along with the reelectionof President Obama, suggests that furtherstimulus spending could be on the horizonWith federal government debt already in ex-

cess of 70 percent of GDP, we should insistthat our policymakers evaluate such propos-als critically and, lacking solid evidence fortheir effectiveness, put the federal check-book away.


17/20

17

Appendix:Details of a Simple VAR

A simple two-variable VAR might take the form,

tttttt

tttttt

GOVTGOVTGDPGDPGOVT

GOVTGOVTGDPGDPGDP

,2241312110

,1241312110

where the tindexes time periods (say quarters) so that over time, policy GOVTaffects eco-nomic activityGDPand the level of economic activity also affects policy.

A researcher wants to estimate the parameter values (i.e., thes and thes) to understandthe patterns and magnitudes of those interactions. Then, to understand how a shock toGOVTaffects the system, one has to first identify those shocks, and that involves imposingrestrictions on the covariance structure between the components of the systems error terms,1,tand 2,t. The researcher presumes that the components of1,tand 2,tare the shocks to

both GDPand GOVT.The simplest way to identify these underlying shocks is to assume a specific ordering ofthose shocks. Call the underlying shocks to GDPand GOVT, respectively, gdp,t and govt,t.One ordering would assume that shocks to government spending can affect economic activ-ity immediately, but not vice versa:

1,t=gdp,t+ govt,t and 2,t= govt,t.

As assumed above, the shock to government spendinggovt,thas an impact on both the GDPand GOVTequation directly. The parameter determines how strongly a shock to GOVTinitially impacts GDP. On the other hand, the shock to economic activitygdp,tonly entersthe GDPequation directly and it affects GOVTonly indirectly through its effect on GDPover

time. Alternatively, one could order GDPfirst by assuming that 1,t=gdp,tand 2,t= govt,t +gdp,t.

In Figure 4 in the text, the panel (a) impulse response is computed based on the assump-tion that 1,t=gdp,t+ govt,t and 2,t= govt,t. The panel (b) impulse response is computedbased on the assumption that 1,t=gdp,tand 2,t=govt,t+ gdp,t.

,


18/20

18

Notes1. The Economy: We are All Keynesians Now,Time, December 31, 1965.

2. Alan S. Blinder, The Case Against Discre-tionary Fiscal Policy, CEPS Working Paper no.

100 (June 2004).

3. The original estimate of the total tax cuts andexpenditure increases of the ARRA was $787 bil-lion. The Congressional Budget Office has sinceincreased that estimate: http://www.cbo.gov/sites/default/files/cbofiles/attachments/02-22-ARRA.pdf. In a September 16, 2012, Washington Postpiece,Robert Samuelson also argued that the true stimu-lus also includes subsequent tax cuts and spendingincreases plus automatic stabilizers. These refer tothe budgets tendency to swing into deficit duringa recession because tax revenues fall and spendingon unemployment benefits and other safety-netprograms rise. Budget deficits broadly measure

stimulus. From 2009 to 2012, theyre about $5.1trillion. See Robert Samuelson, Bernanke on theBrink, Washington Post, http://www.washingtonpost.com/opinions/robert-j-samuelson-bernanke-on-the-brink/2012/09/16/9c94a66c-fe93-11e1-8adc-499661afe377_story.html.

4. See Christina Romers comments at the Na-tional Association of Business Economists 25thannual Washington Economic Policy Confer-ence, March 3, 2009, at http://www.whitehouse.gov/administration/eop/cea/speeches-testimony

/03032009/.

5. Antonio Fats and Ilian Mihov, Why FiscalStimulus is Likely to Work, International Finance12, no. 1 (Spring 2009): 70.

6. Paul Krugman, Keynes Was Right,New YorkTimes, http://www.nytimes.com/2011/12/30/opinion/keynes-was-right.html.

7. David Boaz, Economists against the Stimu-lus, Cato at Liberty (blog), http://www.cato-at-liberty.org/economists-against-the-stimulus/.

8. Jonathan A. Parker, On Measuring the Ef-fects of Fiscal Policy in Recessions,Journal of Eco-nomic Literature 49, no. 3 (September 2011): 704.

9. Janet Novack, Obama Proposes $447 BillionJobs Package, Forbes, http://www.forbes.com/sites/janetnovack/2011/09/08/obama-proposes-447-billion-jobs-package/.

10. Paul Krugman, The Conscience of a Liberal,New York Times, http://krugman.blogs.nytimes.com/2011/09/05/on-the-inadequacy-of-the-stimulus/.

11. The Hodrick-Prescott filter is a popular

method for fitting a smooth trend through mac-roeconomic data. See Robert J. Hodrick and Ed-ward C. Prescott, Postwar U.S. Business Cycles

Am Empirical Investigation, Journal of Money,Credit, and Banking 29, no. 1 (February 1997)116.

12. For example, see Antonio Fats and Ilian Mihov, The Effects of Fiscal Policy on Consump-tion and Employment: Theory and Evidence,CEPR Discussion Paper no. 2760 (April 2001)and Olivier Blanchard and Roberto Perotti, AnEmpirical Characterization of the Dynamic Ef-fects of Changes in Government Spending andTaxes on Output, Quarterly Journal of Economics117, no. 4 (November 2002): 132968.

13. Strictly speaking, looking at data on personal savings rates indicates (the inverse of) Ameri-cans average propensity to expend; not their mar

ginal propensity to expend. But the differencebetween observed savings rates and the marginal

propensity to expend necessary to imply a mul-tiplier of two (or less) is too large to not suggestother factors that must be taken into account.

14. Robert J. Barro, Are Government Bonds NetWealth?Journal of Political Economy 82, no. 6 (No-

vemberDecember 1974): 1095117.

15. Robert J. Barro, The Stimulus EvidenceOne Year On, Wall Street Journal, February 232010, http://scholar.harvard.edu/barro/files/wsj_10_0223_stimuluspackage.pdf.

16. Andrew T. Young and Russell S. Sobel, Re-

covery and Reinvestment Act Spending at theState Level: Keynesian Stimulus or DistributivePolitics,Public Choice (forthcoming).

17. See Olivier Blanchard and Roberto PerottiAn Empirical Characterization of Changes inGovernment Spending and Taxes on Output,Quarterly Journal of Economics 117, no. 4 (Novem-ber 2002): 132968; and Antonio Fats and IlianMihov, The Effects of Fiscal Policy on Consump-tion and Employment: Theory and Evidence,CEPR Discussion Papers no. 2760 (April 2001)The quotes are from Fats and Mihov, p. 6.

18. Jord Gali, Javier Valls, and J. David Lpez

Salido, Understanding the Effects of Govern-ment Consumption Spending on Consumption,

Journal of the European Economic Association 5, no. 4(March 2007): 22770.

19. See Andrew Mountford and Harald UhligWhat Are the Effects of Fiscal Policy Shocks?

Journal of Applied Econometrics 24, no. 6 (April2009): 96092.

20. Valerie A. Ramey and Matthew D. Shapiro


19/20

19

Costly Capital Reallocation and the Effects ofGovernment Spending, Carnegie-Rochester Con-

ference Series on Public Policy 48, no. 1 (June 1998):14594. Quote is from p. 174.

21. Recent papers include Wendy Edelberg, Mar-tin Eichenbaum, and Jonas D. M. Fisher, Under-

standing the Effects of a Shock to GovernmentPurchases, Review of Economic Dynamics 2, no. 1(January 1999): 166206; and Craig Burnside,Martin Eichenbaum, and Jonas D.M. Fisher, Fis-cal Shocks and Their Consequences, Journal of

Economic Theory 115, no. 1 (March 2004): 89117.

22. Valerie Ramey, Identifying GovernmentSpending Shocks: Its All in the Timing, Quarter-ly Journal of Economics 126, no. 1 (February 2011).

23. Antonio Fats and Ilian Mihov, The Casefor Restricting Fiscal Policy Discretion, Quarterly

Journal of Economics 118, no. 4 (November 2003):141947.

24. Antonio Fats and Ilian Mihov, Why FiscalStimulus is Likely to Work, International Finance12, no. 1 (May 2009): 5773.

25. For example, Blanchard and Perotti (2002)report a tax multiplier around 0.8. (See note 14.)

26. Christina Romer and David Romer, TheMacroeconomic Effects of Tax Changes: Esti-mates Based on a New Measure of Fiscal Shocks,

American Economic Review 100, no. 3: 76381.Mountford and Uhlig, using their alternativeSVAR identification strategy, also report a taxmultiplier of 1.63 after two years, which is consid-erably larger than the spending multiplier of 0.07that they report. (See note 21.)

27. An exception being the Jump Start OurBusiness Start-Ups (JOBS) Act, signed into lawin April of 2012. JOBS is a package of incentivesfor small businesses; it is largely devoid of ARRA-type fiscal stimulus.


20/20

RELATED PUBLICATIONS FROM THE CATO INSTITUTE

Should U.S. Fiscal Policy Address Slow Growth or the Debt? A Nondilemma byJeffrey Miron, Cato Institute Policy Analysis no. 718 (January 8, 2013)

Fiscal and Generational Imbalances and Generational Accounts: A 2012 Update byJagadeesh Gokhale, Cato Institute Working Paper no. 12 (November 13, 2012)Fiscal Policy Report Card on Americas Governors: 2012 by Chris Edwards, CatoInstitute White Paper (October 9, 2012)

The Misuse of Top 1 Percent Income Shares as a Measure of Inequalityby AlanReynolds, Cato Institute Working Paper no. 9 (October 4, 2012)

World Hyperinflations by Steve H. Hanke and Nicholas Krus, Cato Institute WorkingPaper no. 8 (August 15, 2012)

What Made the Financial Crisis Systemic? by Patric H. Hendershott and KevinVillani, Cato Institute Policy Analysis no. 693 (March 6, 2012)

RECENT STUDIES IN THECATO INSTITUTE POLICY ANALYSIS SERIES

720. Liberalizing Cross-Border Trade in Higher Education: The ComingRevolution of Online Universities by Simon Lester (February 5, 2013)

719. How to Make Guest Worker Visas Work by Alex Nowrasteh (January 31, 2013)

718. Should U.S. Fiscal Policy Address Slow Growth or the Debt? ANondilemma by Jeffrey Miron (January 8, 2013)

717. China, America, and the Pivot to Asia by Justin Logan (January 8, 2013)

716. A Rational Response to the Privacy Crisis by Larry Downes (January 7,2013)

715. Humanity Unbound: How Fossil Fuels Saved Humanity from Nature and

Nature from Humanityby Indur M. Goklany (December 20, 2012)

714. On the Limits of Federal Supremacy: When States Relax (or Abandon)Marijuana Bans by Robert A. Mikos (December 12, 2012)

Why in the World Are We All Keynesians Again? The Flimsy Case for Stimulus Spending

Documents

Transcript of Why in the World Are We All Keynesians Again? The Flimsy Case for Stimulus Spending