Doubling Exports Under the NEI

8/23/2019 Doubling Exports Under the NEI

1/17

DOUBLING U.S. EXPORTS UNDER THE PRESIDENTS NATIONAL

EXPORT INITIATIVE: IMPLICATIONS OF SUCCESSFUL

IMPLEMENTATION

PETER B. DIXON and MAUREEN T. RIMMER

President Obamas National Export Initiative (NEI) is targeted at doubling U.S.exports between 2010 and 2015. We apply USAGE to quantify what the NEI wouldneed to do to foreign import-demand curves and domestic export-supply curvesto achieve this target. USAGE is a dynamic economy-wide model of the U.S.incorporating recession-relevant factor market specifications including excess capacityand wage/labor-demand elasticities that vary with the level of employment. In ourcentral simulation, export-promotion policies compatible with the Presidents targetreduce the cost of the current recession from about 70 million 1-year jobs for the

period 2008 2020 to 45 million jobs. (JEL E17, C68, E62, E65, F16)

I. INTRODUCTION

In his State of the Union address in January2010, President Obama set the goal of dou-bling U.S. exports in 5 years. Subsequently heannounced a group of export promotion policiestitled the National Export Initiative (NEI). Toadminister and pursue the NEI he formed theExport Promotion Cabinet (EPC). In Septem-ber 2010, the EPC reported on export-promotionpolicies adopted under the NEI and under otherinitiatives of the administration.1 These policiescan be grouped in four categories:

(C1) Shifting foreign demand curves for U.S.exports. These are policies designed to induceforeigners to buy more U.S. goods and servicesat any given purchasers price. NEI and otherAdministration policies in this category include:U.S. sponsorship of both trade missions toforeign countries and visits to U.S. trade showsby potential foreign buyers; and streamlining ofvisa regulations to facilitate visits to the United

We thank the U.S. International Trade Commissionand the U.S. Department of Commerce for supporting thecreation of the USAGE model and for encouraging ourresearch on the National Export Initiative. Neither of theseorganizations is responsible for the contents of the paper.

Dixon: Centre of Policy Studies, Monash University, Clay-ton, Vic 3800, Australia. Phone 61 3 9905 5464, Fax 613 9905 2426, E-mail [email protected]

Rimmer: Centre of Policy Studies, Monash University,Clayton, Vic 3800, Australia. Phone 61 3 9905 5464, Fax61 3 9905 2426, E-mail [email protected]

1. See Export Promotion Cabinet (2010).

States by foreign tourists, students, and businesspeople.

(C2) Shifting U.S. export-supply curves.These are policies that reduce the cost of export-ing. NEI and other policies in this categoryinclude: granting credit through the Export-Import Bank to foreign buyers of U.S. com-modities and to businesses in the United Statesto facilitate their export activity; provision by

the government of foreign-market information toU.S. businesses together with training in dealingwith foreign trade regulations; and improvementof infrastructure required for moving U.S. goodsto and through ports.

(C3) Pursuing trade-expanding policies ininternational forums. The NEI includes rein-forced efforts to remove trade barriers throughnew bilateral and multilateral trade agreementsand through cases in the World Trade Organi-zation designed to force compliance by foreigncountries with existing agreements.

(C4) Improving the macro competitiveness ofthe U.S. economy through cuts in public expen-diture. The EPC report is principally aboutmicropolicies. Nevertheless, it emphasizes theimportance for U.S. exports of strong, sus-tainable, and balanced growth. We interpret at

ABBREVIATIONS

BOTE: Back-of-the-Envelope

EPC: Export Promotion Cabinet

NEI: National Export Initiative

1

Contemporary Economic Policy (ISSN 1465-7287)doi:10.1111/j.1465-7287.2012.00314.x 2012 Western Economic Association International


2/17

2 CONTEMPORARY ECONOMIC POLICY

least part of this as referring to the need toimprove U.S. competitiveness (lowering the realexchange rate2) through fiscal restraint.

We assume that the Presidents goal is inreal terms: the quantity of U.S. exports in

2015 should be twice that in 2010. This isan ambitious target. There is no 5-year periodafter 1950 in which exports have doubled. Thefastest 5-year growth was 68% between 1985and 1990.3 Thus, it is reasonable to ask whetherthe Presidents target is realistic.

In this article, we give a partial answer. Usingan economic model, we quantify the extentto which policies in categories (C1) to (C4)would need to be successful to double exportsin approximately 5 years. That is, we quantifya combination of foreign-demand shift, export-

supply shift, trade expansion, and expenditurereduction that would deliver the Presidentstarget. We do not make a judgment on thelikelihood of achieving these levels of success.

Two key points emerge from our analy-sis. First, export stimulation through policiesin (C1) to (C3) could make a major contribu-tion to bringing the U.S. economy rapidly tonormal levels of employment. This justifies giv-ing export promotion a prominent position inU.S. economic policy. Second, policies in (C4)

implemented in the recessionary conditions pre-vailing in the United States would contributerelatively little to the aim of stimulating exportswhile at the same time they would delay thereturn to normal levels of employment.

This article is organized as follows. Section IIdescribes our model, particularly the idea ofbaseline and perturbation runs. The Appendixgives technical details on our treatment of fac-tor markets. We assume that prices for laborand capital (wage and rental rates) respondsluggishly to increases in demand when there

are high levels of unemployment and excesscapacity. At normal levels, factor prices in ourmodel respond relatively strongly to increases indemand. This is critical to understanding howresponses generated by our model to policiesin (C1) through (C4) differ between recessedand normal conditions. Our model behaves ina Keynesian manner under recessed conditions

2. We define the real exchange rate as the ratio of theprice level in the home country (the United States) to anaverage of the price levels in trade-partner countries, with

all prices translated into a common currency.3. See National Income and Product Accounts (NIPA)

Table 1.1.6 available at http://www.bea.gov/national/nipaweb/SelectTable.asp?Popular=Y.

but is neo-classical under normal conditions.Section III describes our results. Concludingremarks are in Section IV.

II. THE USAGE MODEL

A. Background

In conducting our simulations, we useUSAGE.4 This is a dynamic, CGE model of theUnited States developed at the Centre of Pol-icy Studies, Monash University, in collaborationwith the U.S. International Trade Commission.5

USAGE includes three types of dynamic mecha-nisms: capital accumulation, liability accumula-tion, and lagged adjustment processes. Capitalaccumulation is specified separately for eachindustry. An industrys capital stock at thestart of period t+ 1 is its capital at the startof period t plus its investment during period tminus depreciation. Investment during period tis determined as an increasing function of theexpected rate of return on the industrys capi-tal.6 Liability accumulation is specified for thepublic and foreign sectors. Public sector liabilityat the start of period t+ 1 is public sector liabil-ity at the start of period t plus the public sectordeficit incurred during period t. Net foreign lia-bilities at the start of period t+ 1 are specified

as net foreign liabilities at the start of period tplus the current account deficit in period t plusthe effects of revaluations of assets and liabil-ities caused by changes in price levels and theexchange rate. As discussed in the Appendix,lagged adjustment processes are specified for theresponses of wage rates and rental rates to gapsbetween the demand for and the supply of laborand capital.

USAGE is a detailed model designed forpolicy analysis. It is impossible to lay it outin a journal-length article: the computer codealone takes about 80 pages. Complete techni-cal documentation is available,7 but this doesnot provide a practical way for readers tounderstand and assess results from a particu-lar application. In this article, we explain theresults via back-of-the-envelope (BOTE) meth-ods. We have developed these methods over

4. U.S. Applied General Equilibrium.5. Prominent applications of USAGE by the Commis-

sion include USITC (2004, 2007 and 2009).6. The investment specification for Australias

MONASH model, adopted in USAGE, is discussed indetail in Dixon et al. (2005).

7. See, for example, Dixon and Rimmer (2002) andhttp://www.monash.edu.au/policy/monbook2.htm.


3/17

DIXON & RIMMER: U.S. NATIONAL EXPORT INITIATIVE 3

many years.8 A BOTE explanation consists ofa stylized model that highlights the particularassumptions, economic mechanisms, and datathat are important for the application under con-sideration. Sometimes we are faced with theargument that if results can be explained by aBOTE model, why do you need the full model?The answer is that the full model often givesresults that could not have been anticipated.The skill of the BOTE technique is identify-ing why the full model gives those results. Awell-constructed BOTE explanation allows us toassess results by pinpointing the major underly-ing theory and data.

USAGE can be run at various levels of disag-gregation. The master database identifies morethan 500 industries. In this article, our focus

is on the macroeconomic effects of successfulimplementation of the NEI in an economy inwhich both labor and capital are underemployed.This requires a computationally difficult specifi-cation involving complementarity conditions torepresent excess capacity at an industry level. Inthese circumstances, it has been computationallyconvenient to cut back on the industry dimen-sion. We use a version in which there are only38 industries and in keeping with our focus wereport results only at the macro level.9

B. USAGE Simulations and Closures: Baselineand Perturbation Runs

In a USAGE simulation of the effects ofpolicy and other shocks, we need two runs ofthe model: a baseline or business-as-usual runand a perturbation run. The two runs usuallyhave different closures (division of variablesbetween exogenous and endogenous). If theexogenous variables in the perturbation run areset at the same values that they had eitherendogenously or exogenously in the baseline,then the perturbation run produces the sameresults as the baseline. However, in perturbationruns, we shock some of the exogenous variablesaway from their baseline values. The effects ofthese shocks are revealed as differences betweenthe perturbation and baseline values of variablesthat are endogenous in the perturbation run. Inthis article, we have a single baseline run andseven main perturbation runs.

8. See for example, Dixon, Parmenter, and Powell(1984).

9. Industry level results are available from the authors.

Baseline Run B: Bland and Non-Recessionary.The baseline run (B) in all our simulations is anon-recessionary, business-as-usual forecast for2008 to 2020. It is the sort of bland forecast thatwas available in 2007, before the financial crisisthat precipitated the present U.S. recession. Ingenerating the baseline, we set most macroeco-nomic variables exogenously on almost smoothpaths with annual growth rates in accordancewith forecasts from the Bureau of Labor Statis-tics (2007) and the U.S. Department of Agricul-ture (2007). In the baseline, employment growsat about 1.1% a year, GDP at 3.3%, capital at4.0%, private consumption at 3.2%, public con-sumption at 2.0%, exports at 6.0%, and importsat 5.6%. Consumer prices and the GDP deflatorgrow by about 2.4% a year.

So that the macroeconomic scenario could beset exogenously, several macroeconomic shiftvariables were made endogenous in the base-line closure. As indicated in Table 1, theseincluded: the average propensity to consumewhich adjusted to accommodate the forecastfor private consumption; a confidence variablewhich moved the position of supply-of-fundscurves to capital creators to accommodate theforecast for investment; a variable which movedexport-demand curves to accommodate the fore-

cast for exports; and an import/domestic twist inpreferences which accommodated the forecastfor imports.

The most difficult part of the baseline clo-sure concerns factor markets and the price level.In our baseline for 2008 to 2020, we intro-duce bland forecasts for the aggregate level ofemployment, the real wage rate, and the pricelevel. A useful framework for seeing how theseforecasts are accommodated is the followingstylized model:

Y = (1/Aprim) F(K,L/Alab)(1)

Aggregate production function

W/P = (1/[Aprim Alab]) F2(K, L/Alab)(2)

Real wage equals marginal product of L

Q/P = (1/Aprim) F1(K, L/Alab)(3)

Real rental equals marginal product of K

In this stylized model, Equation (1) relatesoutput (Y) to capital (K) and labor (L) inputs


4/17


TABLE 1

Macroeconomic Closure Choices

Baseline (B) Recession (P1) Perturbation Runs P2 to P7

20082020 2008 2010 20112020 20112020

Real consumption (C) X, bland forecast X, outcome N NAve. propens. to consume N N X, recovery path X, recovery path

Real investment (I) X, bland forecast X, outcome N N

Shift, supply of funds N N X, recovery path X, recovery path

Public expenditure (G) X, bland forecast X, outcome X, bland forecast X, bland forecast or shocked path

Real exports (X) X, bland forecast X, outcome N N

Shift, export-demandcurves

N N X, recovery path X, recovery or shocked path

Real imports (M) X, bland forecast X, outcome N N

Twist, import/domesticpreferences

N N X, recovery path X, recovery or shocked path

Real GDP N N N N

Employment (L) X, bland forecast N X, recovery path N

Primary-factor techchange (Aprim) N X, baseline X, baseline X, baseline

Real wage rate (W/P) X, bland forecast N N N

Labor-saving techchange (Alab)

N X, baseline X, baseline X, baseline

Nominal wage rate (W) N N, sticky N N, sticky

Shifter in sticky wageequation

N X, baseline N X, P1 path

Price level (P) X, bland forecast N N N

Nominal exchange rate X, bland forecast X, Outcome X, bland forecast X, bland forecast

Notes: N denotes endogenous and X denotes exogenous.

via a constant-returns-to-scale production func-tion, F. Movements in Aprim and Alab canbe used to introduce primary-factor-saving andlabor-saving technical change. In Equations (2)and (3), P is the price level (the price deflatorfor GDP). The real wage rate (W/P) and thereal rental on capital (Q/P) are equated to themarginal products of labor and capital.

In the baseline Y is determined by the expen-diture aggregates and K in each year is predeter-mined via capital and investment in the previousyear. Exogenous values for L and W/P are then

accommodated via Equations (1) and (2) withAprim and Alab endogenous.

10 Given that P isexogenous in the baseline, we can then deduceW. Q can be determined from Equation (3).

Perturbation Run P1: Recession and Recovery.Table 2 shows the growth rates in macroeco-nomic expenditure variables for 2008 to 2010.The left-hand panel contains pre-recession fore-casts, adopted in our baseline. The right-hand

10. Given values for Y and W/P, Equations (1) and (2)

can be solved for Aprim and Alab provided the productionfunction is not CobbDouglas. In USAGE, the productionfunctions are CES with substitution elasticities significantlyless than one.

panel shows outcomes. In our first perturbationrun, P1, we set the scenario for the expen-diture aggregates for 2008 to 2010 at theoutcomes for these years (right-hand panel).Consequently, in P1 the macroeconomic shiftvariables have different values from those inthe baseline: the average propensity to consumemoves lower in the perturbation run than inthe baseline; investor confidence moves lower;export-demand curves move inward; and, con-sistent with the often observed phenomenonof strong growth of import shares in domes-tic markets in the upswing and decline in thedownswing, preferences twist against imports.

Aggregate employment and real wages aretreated endogenously in P1 for 2008 to 2010.Correspondingly, primary-factor- and labor-saving technical changes are exogenous, and,as indicated in Table 1, they are set on theirbaseline paths. The bland baseline reveals devel-opments in technology which we assume takeplace independently of the state of the macroe-conomy. If we continue to assume that K is

predetermined, then in terms of our stylizedmodel L is determined via Equation (1), W/Pvia Equation (2), and Q/P via Equation (3).


5/17


TABLE 2

Percentage Growth Rates in Macroeconomic Variables

Pre-recession Forecast (B) Recession Outcomes (P1)

2008 2009 2010 2008 2009 2010

Private consumption 3.2 3.2 3.2 0.3 1.2 17Public consumption 2.0 2.0 2.0 2.5 1.9 1.6

Investment 5.3 5.3 4.8 7.3 18.5 16.1

Exports 6.0 6.0 6.0 6.0 9.5 11.9

Imports 5.6 5.6 5.6 2.6 13.8 14.6

GDP 3.3 3.3 3.2 0.0 2.5 2.9

However, as explained in Dixon and Rimmer(2011), this approach leads CGE models to gen-erate highly unrealistic results for the effectsof recession on factor prices. With Y stronglyreduced relative to the bland baseline, if K

is assumed to be still on it baseline value,then K/L is sharply increased implying a largeincrease in W/P and a correspondingly largedecrease in Q/P. Such movements in factorprices are not generally descriptive of recessionsand are certainly not descriptive of the currentU.S. recession.

To achieve realistic recession effects onfactor prices, we need to drop the assump-tion that existing capital is fully used. Tothis end, we added sticky rental adjustment

equations to USAGE perturbation computa-tions. We also included sticky wage adjustmentequations. Details of these equations are givenin the Appendix. In stylized form, they can bewritten as:

(W(t)/Wb(t)) (W(t 1)/Wb(t 1)) = (L(t)/Lb(t)) + other terms, 0, (1) = 0(4)

Sticky nominal wage, perturbation runs

(Q(t)/Qb(t)) (Q(t 1)/Qb(t 1)) = (KE(t)/K(t)) + other terms, 0, (1) = 0(5)

Sticky capital rental, perturbation runs.

Under Equation (4), we specify that the ratioof the perturbation to baseline level of nomi-nal wages (W/Wb) adjusts in a sticky fashionto the gap between the perturbation level ofemployment (L) and the baseline level (Lb): ifthe perturbation level for employment in yeart is below the baseline level, we assume thatthe perturbation level for the nominal wage rate(W) declines relative to its baseline level (Wb).

11

11. In the version of USAGE used in this article, weassume that there is just one labor market: Equation (4) has

Under Equation (5), we specify that the ratiosof the perturbation to baseline levels of nom-inal rentals (Q/Qb) adjust in a sticky fashionto the gap between perturbation levels of capi-tal in existence (KE) and capital in use (K): if

the perturbation level for capital in existence inyear t is above that of capital in use, we assumethat the perturbation level for the nominal rentalrate relative to its baseline level declines. Putanother way, in Equation (5) we assume that ifthere is excess capacity [(KE(t) > K(t)], thenprofit margins, reflected by rentals, adjust downsluggishly.

Equations (4) and (5) allow USAGE to gen-erate recession results with realistic factor-market behavior. In terms of the stylized model,

we can visualize the P1 solution for 20082010in the following way. With the recession levelsof Y given by the expenditure aggregates,we can restrict attention to combinations ofcapital in use (K) and employment (L) that

are compatible with the aggregate produc-tion function (1). Given these combinations, wecan use the sticky wage and rental equations(Equations (4) and (5)) to trace out a down-ward sloping curve between K/L and W/Q, andwe can use the marginal productivity conditions(Equations (2) and (3)) to trace out an upwardsloping curve between K/L and W/Q. The inter-section of these two curves determines the factor

no occupation or industry dimension. On the other hand,Equation (5) has an industry dimension.


6/17


input and factor price ratios. Knowing K/L andY is sufficient to tie down K and L. The abso-lute factor prices (W and Q) can be determinedfrom Equations (4) and (5), thereby determin-ing the domestic price level (the price deflatorfor GDP).

Beyond 2010, the expenditure aggregates C,I, X, and M are treated endogenously in P1(Table 1). The corresponding shift variables areexogenous. For 2011, we assume in P1 that theseshift variables remain at 2010 values. Then overthe next 5 years (2012 to 2016), they returnsteadily to the values that they had in the base-line, that is, we assume that there is a gradualreturn of the average propensity to consume,investor confidence, export-demand curves andimport/domestic preferences to normal non-

recessionary levels. From 2017 onwards, theshift variables follow their baseline paths.12

By 2010, employment in P1 is 9.6% below itsbaseline forecast value. Because the role of P1 inour analysis is simply to provide the backgroundsituation in which to analyze export promotionpolicies, we want it to portray a typical non-controversial recovery from recession. For thisreason, we decided to impose on P1 a smoothrecovery in employment. We assume that after2010 employment moves gradually toward thebaseline, almost reaching it by 2016. A shifter

in the sticky-wage equation was used as theendogenous instrument for getting employmentonto its assumed path. This is an effective instru-ment because with the exchange rate exogenous(Table 1), adjustments in the nominal wage rateaffect U.S. competitiveness (the costs of U.S.goods in foreign currency) and thereby influ-ence the simulated outcome for employment byaffecting exports and imports.

Perturbation Runs P2 to P5: Effects of Export

Promotion Policies, One At a Time. We conductfour perturbation runs (P2 to P5) to show the

12. Some readers of a preliminary version of this paperthink we are too optimistic. They argue that consumerand investor confidence could take longer than 5 years toreturn to normal and that depressed R&D expenditure couldpermanently reduce U.S. technological progress. As we willshow, successful export promotion is particularly valuablein a recessed economy. Thus, if the U.S. recession lastslonger than we are assuming, then the NEI would be evenmore beneficial than implied by our analysis. We return tothis issue in Section IV. Another pessimistic view is thatthe current recession may permanently increase the non-

recessed unemployment rate. In this case, the U.S. economyis currently closer to capacity constraints than we haveassumed. This would mean that we are over-estimating thebenefits of a successful NEI.

effects of policies in the four categories outlinedin Section I. Each of these runs is identical toP1 for 2008 to 2010.

In P2, we illustrate the effects of (C1) policiesby phasing in over 5 years a 15% outwardmovement in export-demand curves for all U.S.

goods and services. The shift is introducedrelative to the position of the export-demandcurves in the recessionary situation depictedin P1. That is, instead of the export-demandshift variable following the path it had in P1,it follows a path that allows exports to be3% higher at any given foreign-currency pricein 2011 than they would be with the export-demand curves in P1; 6% higher in 2012; 9%higher in 2013; 12% higher in 2014; and 15%higher in 2015 and beyond. All other exogenous

shift variables for expenditure aggregates followthe same paths in P2 as in P1 (Table 1).Rather than fixing the post-2010 path for

employment as in P1, we allow the P2 employ-ment path to move away from the P1 path inresponse to the increased demand for exports.Technically, we determine the P2 path foremployment by fixing the nominal wage-shiftvariable on its P1 path, thereby allowing wagerates to respond in a sticky fashion to move-ments in aggregate employment.

In P3, we illustrate the effects of (C2) policies

by phasing in over 5 years a technologicalchange that reduces costs per unit of exportsfor all U.S. goods and services by 5%. Weadopt the same approach as in P2. Apart fromthe export-cost-reducing technological change,all exogenous variables in P3 follow the samepaths as in P1.

P4 illustrates the effects of (C3) policies. Wephase in over 5 years export-demand shifts andimport-domestic preference twists that wouldexpand both U.S. exports and U.S. imports by

15% at any given set of prices. As with P2 andP3, apart from the relevant policy variables, allexogenous variables in P4 follow the same pathsas in P1.

In P5, we illustrate the effects of (C4) policiesby phasing in over 5 years a 10% cut in publicconsumption and investment. Since we holdtax rates constant, this represents a significanttightening of fiscal policy.

Perturbation Runs P6 and P7: Effects of ExportPromotion Policies, Altogether. In P6, we apply

all the export-expansion shocks from P2 toP5 and find that in combination they wouldapproximately double exports in 5 years.


7/17


As will become apparent from our results forP5, cuts in government spending in recessionaryconditions have a strongly negative effect onemployment. Thus it seems reasonable to delaythe cuts until employment and capital utilizationhave returned to normal levels. In P7, we

illustrate this approach by applying all theshocks from P2 to P5 but we delay the start ofthe phase in of cuts in public expenditure until2015.

III. RESULTS

A. Baseline and Recession

Figures 1 to 6 give results for the baseline Band the recession run P1. For each variable,

we plot an index number with the 2010 valuein P1 at 1.0.13 The depth of the current reces-sion is measured by the gaps in 2008 to 2010between the B and P1 lines. For example,Figure 1 shows that employment measured inhours was reduced by the recession in 2010by about 10% [=100 (1/1.11 1)]. Simi-larly, Figures 2 and 4 show that the recession-ary levels of exports and GDP in 2010 wereabout 10% and 9% lower than the levels theywould have had under non-recessionary condi-tions. In the case of imports and investment, the

recession caused reductions in 2010 of about18% and 24% (Figures 5 and 6). Despite thesharp recession-induced reduction in investment,capital in existence in 2010 was only about3% below its non-recessionary baseline level(Figure 3). Capital in existence is a stock vari-able and shows a muted response to changes ininvestment, its flow variable. As explained in theAppendix, USAGE allows for the emergence ofexcess capacity in response to sharp reductionsin demand for capital. While there is no excess

capacity in the baseline, P1 implies that excesscapacity in 2010 was about 3%. This can beseen in Figure 3 where the P1 level of capital inuse in 2010 is about 3% below the P1 level ofcapital in existence.

Beyond 2010, all the variables shown inFigures 1 to 6 for the recession run move backtoward their baseline paths. By 2017, the gapsbetween the P1 and B paths for employment andfor GDP are substantially eliminated. The P1paths for exports and capital are still below theirB paths by 2020 but the gaps are small. The P1

13. An exception is capital in use. In Figure 3, it iscapital in existence that is given a 2010 value in P1 of 1.0.

FIGURE 1

Baseline (B) and Recession (P1): Employment

0.95

1

1.05

1.1

1.15

1.2

1.25

no recession (B)

recession (P1)

index

1.00

1.11

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 2Baseline (B) and Recession (P1): Exports

0

0.5

1

1.5

2

2.5

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

no recession (B)

recession (P1)

index

1.00

1.29

1.46

1.11

FIGURE 3

Baseline (B) and Recession (P1): Capital

0.9

1

1.1

1.2

1.3

1.4

1.5

no recession (B), capital in use and existence

recession (P1), capital in use

recession (P1), capitalin existence

index

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

paths for investment and imports overshoot thebaseline paths. The overshoot for investment isexplained by post-recession shortage of capital


8/17


FIGURE 4

Baseline (B) and Recession (P1): GDP

0.9

1

1.1

1.2

1.3

1.4

1.5

1.6

no recession (B)

recession (P1)

index

1.00

1.10

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 5Baseline (B) and Recession (P1): Imports

0.8

1

1.2

1.4

1.6

1.8

2

2.2

no recession(B)

recession (P1)

index

1.00

1.22

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 6

Baseline (B) and Recession (P1): Investment

0.8

1

1.2

1.4

1.6

1.8

2

no recession(B)

recession (P1)

index

1.00

1.32

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

and the overshoot for imports is associated withstrong post-recession investment and consequentreal appreciation.

As can be seen from the figures, the P1 pathsimply strong growth rates after 2010, consistentwith catch up. Under usual definitions, thismeans that the recession is over by 2011.However, the figures show that the recessionhas been costly and will continue to be costly.Integrating the gap between the B and P1 pathsin Figure 4 over the period 2008 to 2020 putsthe cost of the recession at about 45% of1 years GDP. A similar calculation applied toFigure 1 puts the cost at about 44% of 1 yearsemployment, that is about 70 million 1-year

jobs.A final noteworthy feature of the P1 run is

the export increases to 2015 and 2016. Figure 2shows export growth on the P1 line of 29%between 2010 and 2015 and 46% between 2010

and 2016. These increases are generated withno new export policies. They imply that toachieve the Presidents target in 2015, newexport policies would need to boost exportsby about 55% beyond where they otherwisewould have been [55 = 100 (2/1.29 1)]. Ifwe are prepared to tolerate a bit of slippage andwait until 2016 for exports to double, then therequired boost from new policies is about 37%[37 = 100 (2/1.46 1)].

B. Effects of Export-Promotion Policies

Figures 7 to 16 contain results for employ-ment and exports in policy runs P2 to P6. Thefigures also contain baseline paths and pathsfrom the recession run, P1. The headline resultsare in Figures 15 and 16. Figure 16 impliesthat in combination the four sets of shocks(run P6) that we have adopted to representthe impacts of policies in (C1) to (C4) wouldallow exports to expand by 106% between 2010and 2016, approximately meeting the Presi-dents target. Figure 15 implies that meeting thePresidents target in this way would make apositive but minor contribution to returning U.S.employment to non-recessionary levels over theperiod up to 2013. After 2013, the employmentcontribution of the combined policies is moresubstantial (the gap between the P1 and P6 pathsin Figure 15 widens).

Contributions of Each of the Four Policies tothe Presidents export target. Figures 8, 10, 12,

and 14 show that policies in each of the fourcategories would help achieve the Presidentstarget.


9/17


FIGURE 7

B, P1, and Export-Demand Shift (P2):

Employment

0.95

1

1.05

1.1

1.15

1.2

1.25

no recession (B)

recession (P1)

recession (P2), 15% export demand

1.11

1.13

1.00

index

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 8

B, P1, and Export-Demand Shift (P2): Exports

0

0.5

1

1.5

2

2.5

no recession (B)

recession (P1)

recession (P2), 15% export demand

1.00

1.38

1.29

1.56

1.46

index

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

Concentrating on 2016, we see that the15% export-demand shift adopted to illus-

trate (C1) policies elevates exports by 6.8%[= 100 (1.56/1.46 1), Figure 8]. Why not15%? A little over half the effect of thehorizontal demand shift is dissipated in priceincreases14: the outward movement in export-demand induces both upward movements ofexport-supply curves (caused by, e.g., wageincreases) and upward movements along export-supply curves (reflecting capital scarcity leadingto higher rental rates).

14. With the nominal exchange rate treated as exoge-nous (determined independently of the export-expansionpolicy), these price increases are equivalent to realappreciation.

FIGURE 9

B, P1, and Export-Supply Shift (P3):

Employment

0.95

1

1.05

1.1

1.15

1.2

1.25

no recession (B)

recession (P1)

recession (P3), 5% export cost reduction

1.11

1.13

1.00

index

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 10

B, P1, and Export-Supply Shift (P3): Exports

0

0.5

1

1.5

2

2.5

no recession (B)

recession (P1)

recession (P3), 5% export cost cut

1.00

1.45

1.29

1.64

1.46

index

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

Because export-demand elasticities inUSAGE for all commodities are set at 3,15

we expected on the basis of partial equilib-rium reasoning that export stimulation via a15% rightward movement in foreign demandcurves (P2) would have approximately the sameimpact on export volumes as a 5% downwardmovement in export-supply curves (P3). Thisis the result suggested by a demand and sup-ply diagram (Figure 17). However, as shownin Figure 10, the P3 run generates a notice-ably larger export expansion than P2. Whereasthe demand shift induced a 6.8% expansionin exports in 2016, the cost reduction induces

15. Justification for numbers like 3 is provided inDixon and Rimmer (2010; and 2002, section 27).


10/17


FIGURE 11

B, P1, and Trade Expansion (P4): Employment

0.95

1

1.05

1.1

1.15

1.2

1.25

no recession (B)

recession (P1)

recession (P4), 15% trade expansionindex

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 12

B, P1, and Trade Expansion (P4): Exports

0

0.5

1

1.5

2

2.5

no recession (B)

recession (P1)

recession (P4), 15% trade expansion

1.00

1.45

1.29

1.63

1.46

index

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

a 12.3% expansion [=100 (1.64/1.46 1),Figure 10]. The explanation involves generalequilibrium effects. In P3, less labor and cap-ital are required to produce any given level ofexports than in P2. Extra resources available tothe economy in P3 relative to P2 mean that fac-tor costs are lowered in P3 relative to P2.16 Thisis an extra downward force on the supply curve(beyond the 5% assumed in P3) that operates inP3 but not in P2.

Figure 12 shows that the trade-expandingpolicies assumed in P4 generate an increase inexports in 2016 of 11.6% [=100 (1.63/1.46 1)]. In view of the P2 results, we can attribute

16. In this context, we should think of the exchangerate as the numeraire. Thus we are saying that factor costsrelative to the international price level are lower in P3 thanP2, that is the real exchange rate is lower in P3 than in P2.

FIGURE 13

B, P1, and Public Expenditure Cuts (P5):

Employment

0.95

1

1.05

1.1

1.15

1.2

1.25

no recession (B)

recession (P1)

recession (P5), 10% budgetary cuts

1.11

1.08

index

1.14

1.10

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 14

B, P1, and Public Expenditure Cuts (P5):

Exports

0

0.5

1

1.5

2

2.5

no recession (B)

recession (P1)

recession (P5), 10% budgetary cuts

1.00

1.35

1.29

1.53

1.46

index

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

approximately 6.8 percentage points of this

to the 15% rightward shift in export-demandcurves assumed in P4. The rest of the P4expansion in exports can be attributed to theimport-creating effects of the assumed twist inimport-domestic preferences. Extra demand forimports in P4 lowers the U.S. real exchange rateand thereby stimulates exports.

In Figure 14, the 10% cut in public expendi-ture assumed in P5 causes an increase in exportsin 2016 of 5% [=100 (1.53/1.46 1)]. Thisexport response may seem rather weak to peo-ple used to looking at CGE results generated

under the assumption of full capacity utilization.However, we do not adopt this assumption forthe recessionary conditions currently prevailing


11/17


FIGURE 15

B, P1, and All Policies (P6): Employment

0.95

1

1.05

1.1

1.15

1.2

1.25

no recession (B)recession (P1)

recession (P6), with all export policiesindex

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 16B, P1, and All Policies (P6): Exports

0

0.5

1

1.5

2

2.5

3

no recession (B)

recession (P1)

recession (P6), with all export policies

1.00

1.82

1.29

2.06

1.46

index

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

in the United States. With excess capacity andunemployment allowed for in our simulations,changes in demand for capital and labor causemuted responses in factor prices. This limits theextent of real devaluation, and the consequent

increase in exports, associated with cuts in pub-lic expenditure.

Contributions of Each of the Four Policies toEmployment Recovery. Figures 7 and 9 indi-cate that (C1) and (C2) policies would helpthe United States to return speedily to base-line levels of employment. In both figures, ourassumed policies raise employment strongly inthe recessionary period, by 2 percentage pointsin 2013 for example. These employment effectsreflect the Keynesian nature of our underly-

ing assumptions. When employment is wellbelow its baseline path, our model impliesthat output and employment respond strongly

FIGURE 17

Export-Demand and Supply

Price

Quantity

1.00

1.05

0.95

1.151.00 Qf

ab

c

Initial demand curve

Demand curve after 15%rightward shift

Initial supply curve

Supply curve after 5% costreduction

Note: The initial price/quantity point is at a. A 15%rightward shift in the demand curve moves the price quantitypoint to b. A 5% downward shift in the supply curve movesthe price/quantity point to c. With the demand elasticity at3, the quantity at points b and c is the same, Qf.

to increases in aggregate demand, includingexport-demand.

Figure 11 shows the employment effects of(C3) policies. These policies move export-demand curves to the right, leading to anexpansion of exports and, in an underemployedeconomy, a consequent increase in employment.On the other hand, they lead to a replacementof domestic products by imported products onthe domestic market. In an underemployed econ-omy, this has a negative effect on employment.In our P4 run, which illustrates the C3 policy,we assume that the United States enters into bal-anced free trade agreements, that is agreementswhere the impact effects on exports and importsare the same. Such policies have little net effecton employment.

Figure 13 again reflects Keynesian mecha-nisms, showing strongly negative employmenteffects from cuts in public expenditure. This

may seem controversial: there is a considerableliterature implying that cuts in public expendi-ture have only minor effects on employment.This literature is usually framed in terms ofincreases in public expenditure and emphasizespotential crowding-out effects on: consumption(Ricardian equivalence, Barro 1974); net exports(real appreciation, Mundell 1963 and Fleming1962); and investment (increased interest rates).Crowding-out arguments are developed largelyin theoretical models that assume business-as-usual or steady-state conditions. Such assump-tions are also adopted in the DSGE modelingliterature, see for example Ratto, Roeger, andint Veld (2009). However, Figure 13 shows


12/17


effects under recessionary conditions. In theseconditions, tighter fiscal policies are likely tohave a more deleterious effect on employ-ment than in business-as-usual conditions. Thisis illustrated in Figure 18. The low path inFigure 18 is an alternative presentation of the P5

results in Figure 13: it shows the gaps betweenthe P5 and P1 lines expressed as percentagesof the P1 levels. The other path in Figure 18was computed by imposing the expenditure cutsas shocks to an economy that would otherwisehave been evolving as in the business-as-usualbaseline (run B).

With 10% of government expenditure beingworth 1.94% of GDP, the recession results inFigure 18 imply an employment multiplier forgovernment expenditure17 of 1.77 (=3.44/1.94).

On the other hand, the business-as-usual resultsimply a multiplier of 0.43 (=0.83/1.94). Thislower multiplier is broadly consistent withthe recent study by the IMF (2010) whichfound that a fiscal consolidation worth 1%of GDP increases the unemployment rate by0.3 percentage points. The IMF multiplier canbe considered an average over a large num-ber of consolidation episodes (time periodsand countries), the bulk of which reflect non-recessionary conditions.18 Cuts in public expen-diture in non-recessionary conditions restrainwage increases thereby causing the price levelto be lower than it otherwise would have beenwith beneficial effects for investment, exports,and import-competing industries. In recession-ary conditions, cuts in government spendinghave little restraining effect on wages and prices:there is little wage/price pressure to restrain.Consequently, in these conditions there is lit-tle to offset the direct loss of jobs generated byexpenditure cuts, and much higher multipliersare realistic. As alluded to earlier, a corollary

is that tight fiscal policy stimulates exports lessstrongly in recessed conditions than in normalconditions (see Figure 19).

Improving the Employment Effects by Delay-ing the Cuts in Public Expenditure. The resultsin Figures 13, 14, 18, and 19 suggest that theP5 shocks (cuts in public expenditure) could

17. Percentage effect on employment of an increase ingovernment expenditure worth 1% of GDP.

18. The IMF study does not deal directly with the effects

of recessions on multipliers. However, the study impliesthat multipliers are inversely related to interest rates. This issupportive of our results because interest rates are typicallylow in recessionary conditions.

FIGURE 18

10% Public Expenditure Cuts: Effects on

Employment (% deviations)

-4

-3.5

-3

-2.5

-2

-1.5

-1

-0.5

0

0.5

non-recessionary conditions

recessionaryconditions

% deviation

-3.44

-0.83

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 19

10% Public Expenditure Cuts: Effects on

Exports (% deviations)

0

2

4

6

8

10

12

14

non-recessionary conditions

recessionary conditions

% deviation

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

be excluded from the export promotion pack-age (P6) with little cost on the export targetand major benefits for employment. This isconfirmed by the P7 run in which we delay

the commencement of the public expenditurecuts until 2015. As shown in Figure 20, exportgrowth between 2010 and 2016 is reduced from106% in P6 to 95% in P7. Figure 21 showsthe employment compensation allowed by thisslippage on the export target. In Figure 21, theemployment line for the P7 package lies wellabove that for the P6 package throughout theperiod of seriously depressed employment, up to2014. In the long term, beyond 2016, employ-ment performance under P6 is better than thatunder P7: in this period employment under P7 is

damped by cuts in public expenditure. However,the P7/P6 tradeoff for employment is clearlyin favor of P7. Adding up the gaps between


13/17


FIGURE 20

B, P1, P6, and Delayed Cuts (P7): Exports

0

0.5

1

1.5

2

2.5

3

no recession (B)

recession (P1)

recession (P6), with all export policies

1.00

1.73

1.29

1.95

1.46

1.82

2.06

recession (P7), with all exportpolicies but delayedbudgetary cuts

index

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

FIGURE 21B, P1, P6, and Delayed Cuts (P7):

Employment

0.95

1

1.05

1.1

1.15

1.2

1.25

no recession (B)recession (P1)

recession (P7) , with all export policiesbut delayed budgetary cuts

recession (P6), with all export policiesindex

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

the P7 and B lines in Figure 21 gives the costof the recession under the P7 package at 28%of a years aggregate employment, down from44% in P1, without the package. Under P6, the

cost of the recession is 32% of a years aggre-gate employment. The advantage of P7 over P6would be accentuated if we weighted jobs cre-ated early in the simulation period, when jobsare desperately needed, more highly than jobscreated later in the period, when the labor mar-ket is tight.

IV. CONCLUDING REMARKS

The United States is suffering from its mostserious recession since the 1930s. Using the

USAGE model, we have traced out a pictureof the economy, run P1, in which there is a nat-ural recovery starting in 2011, with employment

approximately returning to its baseline (norecession) path by 2015. Given this picture, thecost of the recession accumulated for the years2008 to 2020 is about 44% of a years aggregateemployment or 70 million 1-year jobs.

Implementation of policies to achieve the

Presidents target under the NEI of doublingexports by 2015 could substantially reduce thiscost. In our P7 run, we found that policies imple-mented over the period 2011 to 2015 which

shifted export-demand curves to the rightby 15% (C1),

reduced costs per unit of exports by 5%(C2), and

generated a 15% impact on exports andimports via trade agreements (C3),

would approximately double exports by 2015.At the same time, they would reduce the cost ofthe recession to 28% of a years employment or45 million 1-year jobs, a saving of 25 million

jobs. In P7, (C4) policies (cuts in public expen-diture of 10%) are also undertaken. However,their implementation is delayed until 2015 whenemployment has returned approximately to itsbaseline (no recession) path. Our simulationsshow that tighter fiscal policies implementedwhen employment is well below its baselinelevel contribute relatively little to the goal of

expanding exports while at the same time theyhave strongly negative effects on employment.

The value of export promotion policies in(C1) to (C3) as a means of stimulating employ-ment is sensitive to the state of the economy.The worse the recession and the slower thenatural recovery, the more valuable are thesepolicies. In a simulation not reported in this arti-cle, we created an alternative no-policy reces-sion scenario (P1alt) in which the start of therecovery is delayed until 2012. In this alterna-

tive scenario, the cost of the recession for theperiod 2008 to 2020 is 55% of a years aggregateemployment or 88 million 1-year jobs. Imple-mentation of the P7 package in this situationreduces the cost of the recession to 35% of ayears employment or 56 million jobs, a savingof 32 million jobs.

Against these benefits must be put thecosts of the NEI. One relevant figure is thatthe budget for implementing the NEI for FY2012 is $78.5 million.19 Another figure is loans

19. See Commerce Departments 2012 budget at http://www.commerce.gov/print/blog/2011/02/14/commerce-department%E2%80%99s-proposed-2012-budget, down-loaded May 25, 2011.


14/17


outstanding by the U.S. governments Export-Import Bank. The Bank makes loans to U.Sexporters and to importers of U.S. productswith the objective of boosting U.S. exports. In2011 outstanding loans by the Bank stood at$75 billion.20 It can be argued that this imposes

a cost on U.S. taxpayers because the Bankcharges low interest rates that do not fullyaccount for the riskiness of their loans.21 How-ever, the extent of this cost is likely to be small.In normal years the Bank is self-financing. Evenin the crisis year of 2009, the Banks losseswere less than $1 billion and by 2010 the Bankrecorded a small profit. Assuming that the Banknormally breaks even but that a fully commer-cial bank would achieve an interest rate spreadof 1%, then the extent of the annual subsidy

provided to exporters through the Export-ImportBank is about $750 million. Only a small frac-tion of this is associated with the NEI: mostof the loans predate the NEI. Thus, overall ourimpression is that the annual costs of the NEIcan be no more than a few hundred milliondollars. These costs are negligible in relationto the potential benefits which on an annualbasis are several percent of GDP (1% of GDP isabout $140 billion). On these figures, we wouldneed to assign an extremely low probability toeven partial success to conclude that the NEI

is a bad idea. Even in the absence of an esti-mate of this probability, our analysis of theNEI has proved useful in policy circles in twoways. First, it has provided policy people, whohave a feel for how export promotion programsaffect sales for specific products, with an esti-mate of how much policy effort will be requiredto achieve the NEI target. This is importantin their budgeting and goal setting. Second,by demonstrating the benefits of successfulimplementation of the NEI in recessionary con-

ditions, the analysis has helped proponentsto argue for the program against alternativeproposals.

In analyzing the Presidents export target,we have used a one-country model and lookedat the issue purely from the point of view ofthe United States. It is worth asking whetherexport-promotion by the United States to speedrecovery from recession is really a 1930sbeggar-thy-neighbor policy that will provokeself-defeating retaliation.

20. See Export-Import Bank of the U.S., 2010 AnnualReport, available at http://www.exim.gov/about/reports/ar/2010/, downloaded October 31, 2011.

21. See James (2011).

We think that retaliation is unlikely for tworeasons. First, the United States intends to abideby bilateral and multilateral trade agreements,and provided it does this, the rest of the worldmay regard U.S. export promotion as a legit-imate contribution to redressing global imbal-

ances. Second, the present global downturn isless internationally pervasive and symmetricthan that in the 1930s. If the United States andits trade partners were equally recessed then itis difficult to see how U.S. export promotionpolicies could be of much value even withoutexplicit retaliation. In a symmetrically recessedworld, initial increases in U.S. exports wouldreduce activity in other countries causing reduc-tions in their demands for imports, eventuallynegating the initial increase in U.S. exports.

However, with the current downturn concen-trated in the United States, a switch in demandtoward the United States and away from therest of the world (an increase in U.S. exportsand a corresponding increase in rest-of-worldimports) is likely to have a relatively large pos-itive impact on output in the under-employedU.S. economy and only a small negative impacton output in the more fully employed rest-of-world economy. This will allow U.S. exportpromotion to benefit the United States with-out significant reductions in output and importdemands in other countries. In our simulations,export promotion policies expand U.S. activitysufficiently to generate increases in U.S. importsthat almost match the increases in U.S. exports,implying that U.S. export promotion will imposelittle macroeconomic adjustment on its tradingpartners.

APPENDIX: ASYMMETRIC TREATMENT OF FACTORMARKETS IN THE USAGE MODEL

In analyzing demand-stimulating policies such as exportpromotion, we need to recognize that their effects dependon the state of the economy on which they are imposed.If there is considerable unemployment and excess capacitythen demand stimulation is likely to generate strong quantitymovements (increases in employment and output) and weakprice movements. On the other hand, if the economy is nearfull employment then demand stimulation is likely to havelittle effect on quantities and strong effects on prices. Thispoint is overlooked by modelers who report single numbermultiplier effects.

In the version of USAGE used in this article, theresponses of wages and rental rates on capital to changes indemand are dependent on the levels of unemployment and

excess capacity. Thus our model shows quite different macroresponses to a given policy when that policy is imposed inrecessionary conditions from the responses when the policyis imposed in business-as-usual conditions.


15/17


This Appendix outlines the specifications in USAGE ofwage and rental behavior. In Section A1, we describe howUSAGE allows, in perturbation runs, the responsiveness ofwage movements to changes in employment to depend onthe level of employment relative to the baseline level. InSection A2, we describe how USAGE allows the respon-siveness of rental movements to changes in demand for

capital to depend on the level of capital utilization. By intro-ducing conditional responsiveness of factor prices, we giveUSAGE the ability to differentiate between the effects ofpolicies in recessionary conditions compared with normalconditions.

A1. Labor-Market Specification: Asymmetric Wage

Adjustment

In perturbation runs P1 to P7 we assume that the averagewage rate for the economy adjusts according to the equation:

{W(t)/Wb(t) 1} = {W (t 1)/Wb(t 1) 1}(A1)

+ F(L(t)/Lb

(t)) + S(t).

In this equation, the subscript b indicates a baselinevalue, that is, a value in the non-recessionary run, runB. Wb(t) and Lb(t) are the wage rate and the level ofemployment in year t in run B. W(t) and L(t) are the wagerate and the level of employment in year t in the perturbationrun. F is a function with a positive first derivative and avalue of zero when L(t)/Lb(t) = 1. S is a shift variable.

For understanding Equation (A1), it is easiest to start byassuming that S is fixed on zero. Then under Equation (A1),we assume in perturbation runs that the deviation in the wagerate from its baseline level in year t will be the same as thatin year t 1 if employment in year t is at its baseline level.If employment in the perturbation run in year t is above

(below) its baseline level then the deviation in the wage ratewill increase (decrease), driving employment back towardits baseline level.

However, in the perturbation runs reported in this article,we do not assume that S is fixed on zero. In P1, S movesendogenously to allow (A1) to be satisfied with L movingalong its exogenously imposed no-policy recovery path.In P2 to P7, S is set exogenously on its P1 path andemployment adjusts endogenously.

With S on its P1 path, (A1) will ensure that W followsits P1 path if L follows its P1 path. If in a perturbation runL moves above its P1 path in year t, then wages in this runwill increase at a quicker rate (or decrease at a slower rate)in year t than in the P1 run. This means that if the policy in

a run P2 to P7 makes the year-t labor market tighter than itwas in P1, then the year-t increase in wages in that run willbe greater than that in P1.

The simplest example of an F function is

F(L(t)/Lb(t)) = (L(t)/Lb(t) 1),(A2)

where is a positive parameter. However, for this article wehave found that Equation (A2) leads to an unrealisticallyfavorable picture for employment as the economy comesout of a recession. In some preliminary perturbation runs,employment during the recovery phase (2011 to 2016)moved above its baseline path by nearly as much as ithad earlier (2008 to 2010) moved below its baseline path.

This unrealistic result reflects the symmetric nature ofEquation (A2). For example, if is 1 and S is set on 0,then wages will rise by 2% when employment is 2% aboveits baseline level and fall by 2% when employment is 2%

below its baseline level. With the bland, close-to-NAIRU-employment baseline (the B run) adopted for this article wewould expect wages to rise more than 2% relative to baselineif employment were 2% above baseline, which would be avery tight labor market. On the other hand, with downwardstickiness, we would expect wage rates to fall by less than2% relative to baseline when employment is 2% below

baseline.To capture these asymmetric responses, we specified F

according to

F(L(t)/Lb(t)) = (/{exp[( L(t)/Lb(t))] 1}) 1.

(A3)

We calibrate Equation (A3) by setting

= 1.02(A4)

F (1) = 0.00 and(A5)

F (0.9) = 0.02.(A6)

These last two conditions give = 0.007192 and =

1.643908.Figure A1 shows the F function with these parameter

values. Condition (A4) imposes an upper bound on theemployment deviation in the positive direction of 2%.Condition (A5) ensures that the wage deviation is constant ifemployment is at its baseline value and S is zero. Condition(A6) limits the fall in wages relative to baseline to 2%when employment is 10% below its baseline value and Sis zero.

A2. Capital-Market Specification: Asymmetric Rental

Adjustment and Excess Capacity

Most CGE calculations are conducted under the assump-

tion of full capacity utilization. It is assumed that the rentalrate on capital in each industry adjusts in each year toensure that demand for capital equals the supply of cap-ital. The supply of capital is predetermined, and dependson investment in previous years. Dixon and Rimmer (2011)demonstrate that this approach is unsuitable for perturbationruns designed to simulate the onset of a serious recessionor the effects of policy changes in recessionary conditions.They show that the full capacity assumption leads to unreal-istically large simulated reductions in rental rates in responseto reductions in demands for capital of the magnitude that areevident in the current U.S. recession. Unrealistic reductionsin rentals generate unrealistic reductions in the real exchange

FIGURE A1

Asymmetric Wage Adjustment Function, F[L(t)/Lb(t)].

F(L(t)/Lb(t))

L(t)/Lb(t)

1.021.00

0.900.00

-0.02


16/17


FIGURE A2

Demand and Supply Curves for Capital: Effects of aReduction in Demand.

.

.

Rental oncapital

KU=KE

KUKE

Perturbed solution withexcess capacity specification

Perturbed solutionunder full capacity

D (perturbed)

D (baseline)

Baseline solution

.

excess capacity

rate which in turn lead to unrealistic simulated responses forexports and imports.

To solve this problem, Dixon and Rimmer introduced asticky adjustment specification for rental rates in perturba-tion runs. Under this specification, full capacity utilizationis maintained in an industry when there is a sharp fall indemand for capital only if this is possible with no morethan a limited fall in the rental rate. If an unrealisticallylarge fall in the rental rate would be required to maintainfull capacity utilization then capital in use (KU) drops belowcapital in existence (KE). Figure A2 is an illustration of the

DixonRimmer specification.In mathematical terms, the sticky rental specificationadopted for the current paper is as follows:

{(Q(j, t)/Qb(j,t)) 1} = {(Q(j, t 1)/Qb (j,t 1)) 1}(A7)

EXC(j,t) MF(j,t) + S(j,t) for all t,

EXC(j,t) = (KE(j,t)/KU(j,t)) 1 for all t,(A8)

MF(j,t) = 500 ((RQ(j,t)/RQb(j,t)) 1)(A9)

+ FMF(j,t) for all t,

Q(j,t) = nj (KU(j,t),.. .) for all t,(A10)

EXC(j,t) 0, S(j, t ) 0,(A11)

EXC(j,t) S(j,t) = 0 for all t,

MF(j,t) 0.05 0, FMF(j,t) 0,(A12)

(MF(j,t) 0.05) FMF(j,t) = 0 for all t,

where Q(j,t) and Qb (j,t) are the rental rate on capital inindustry j in year t in the perturbation and baseline runs;EXC(j ,t) is excess capacity in industry j in year t; MF(j, t)

is a variable that limits the fall between years t 1 and t inrental in the perturbation run relative to baseline; RQ(j , t)and RQb(j, t) are the perturbation and baseline values ofthe real rental on capital in industry j and year t, defined as

rental rate deflated by the price of a unit of capital; nj is thedemand function for capital inverted to express the rentalas a decreasing function of capital in use; is a positiveparameter; and S(j,t) and FMF(j, t) are complementaryslack variables whose role will be defined shortly.

For understanding Equations (A7) to (A12) a good start-ing strategy is to ignore Equation (A9) and assume that

MF(j, t) is a parameter set at 0.05. Then consider twopossibilities. The first possibility is that there is no excesscapacity, KU(j, t) = KE(j, t). With KU(j, t) given in thisway, the rental rate Q(j, t) can be determined from thedemand curve for capital, see Equation (A10). Once weknow Q(j,t) we can check to see if Equation (A7) issatisfied with a non-negative value for S(j,t). If so, wecan conclude that industry j will operate at full capacityin year t. If not, then we must consider the other possi-bility, that KU(j, t) < KE(j, t). In this case, we must setS(j,t) at 0, see Equation (A11). Then starting KU(j, t) atKE(j, t) we can consider successively lower values. As welower KU(j, t) the left-hand side of Equation (A7) rises andthe right-hand side falls. In this way the KU(j, t) satisfying

Equation (A7) can be determined. Notice that Equation (A7)does not limit the extent to which rental rates can increasewhen capital is fully used. What it does is to limit the extentto which rentals can fall. With set at a small value (we used0.2) and MF(j, t) at 0.05, Equation (A7) limits the extent towhich the rental rate in industry j can fall relative to itsbaseline value to a little over 5%.

However, falls of more than 5% are realistic if rentals arehigh. For example, ift 1 in the perturbation run is a year ofcapital scarcity in industry j , allowing a temporary elevationin profitability (reflected by a high real rental on capital),then it is realistic to suppose that profitability can fall sharplyin year t in response to an increase in capacity generated bystrong investment in year t 1. We allow for this possibility

by treating MF(j ,t) as a variable. Its lower bound is 0.05(see Equation (A12)), but it will take higher values (allowingfor larger falls in rental rates) if RQ(j ,t)/RQb(j,t) 1 isgreater than 0.0001. In this case FMF(j, t) must be zero(see Equation (A12)) implying that MF(j, t) > 0.05. Thusif RQ(j, t) is significantly higher than RQb(j,t), then rentalscan fall sharply without necessitating excess capacity.

REFERENCES

Barro, R. Are Government Bonds Net Wealth? Journal ofPolitical Economy, 81, 1974, 109517.

Bureau of Labor Statistics. The U.S. Economy to 2016:

Slower Growth as Boomers Begin to Retire, inMonthly Labor Review, Vol. 130(11), edited by BettyW. Su. November, 2007. Accessed January 28, 2011.http://www.bls.gov/opub/mlr/2007/11/art2full.pdf.

Dixon, P. B., and M. T. Rimmer. Dynamic General Equilib-rium Modelling for Forecasting and Policy: A Practi-cal Guide and Documentation of MONASH, Contribu-tions to Economic Analysis 256. Amsterdam: North-Holland Publishing Company, 2002, pp. xiv+338.

. Optimal Tariffs: Should Australia Cut Automo-tive Tariffs Unilaterally? Economic Record, 86(273),2010, 14361.

. You Cant Have a CGE Recession without ExcessCapacity. Economic Modelling, 28(12), 2011,60213.

Dixon, P. B., B. R. Parmenter, and A. A. Powell. TheRole of Miniatures in Computable General Equilib-rium Modelling: Experience from ORANI. Economic

Modelling, 1(4), 1984, 42128.


17/17


Dixon, P. B., K. R. Pearson, M. R. Picton, and M. T.Rimmer. Rational Expectations for Large CGE Mod-els: A Practical Algorithm and a Policy Application.

Economic Modelling, 22(6), 2005, 100119.Export Promotion Cabinet. Report to the President on the

National Export Initiative: The Export Promotion Cab-inets Plan for Doubling U.S. Exports in 5 Years, 2010.Accessed January 6, 2011. http://www.whitehouse.gov/sites/default/files/nei_report_9-16-10_full.pdf.

Fleming, J. Domestic Financial Policy under Fixed andFloating Exchange Rates. IMF Staff Papers, 9(3),1962, 36979.

International Monetary Fund. World EconomicOutlook: Recovery, Risk and Rebalancing, WorldEconomic and Financial Survey, IMF, October, 2010,pp. 124+xvii. Accessed January 27, 2011. http://www.imf.org/external/pubs/ft/weo/2010/02/pdf/text.pdf.

James, S. Time to X Out the Ex-Im Bank. Trade PolicyAnalysis paper no. 47, Cato Institute, Washington

DC, July 2011, p. 20. Accessed January 28, 2011.http://www.cato.org/pubs/tpa/tpa-047.pdf.

Mundell, R. Capital Mobility and Stabilization Policyunder Fixed and Flexible Exchange Rates. Canadian

Journal of Economics and Political Science, 29(4),1963, 47585.

Ratto, M., W. Roeger, and J. int Veld. QUEST III: AnEstimated DSGE Model of the Euro Area with Fiscaland Monetary Policy. Economic Modelling, 26, 2009,22233.

U.S. Department of Agriculture. USDA Agricultural Projec-tions to 2016, February, 2007. Accessed January 28,2011. http://usda.mannlib.cornell.edu/usda/ers/94005/2007/oce20071.pdf

U.S. International Trade Commission, USITC. The Eco-nomic Effects of Significant U.S. Import Restraints:Fourth, Fifth and Sixth Updates, 2004, 2007 & 2009,Investigation No. 332-325, Publications 3701, 3906and 4094.

Doubling Exports Under the NEI

Documents

Transcript of Doubling Exports Under the NEI