Fiscal Consolidation in an Open Economy

Board of Governors of the Federal Reserve System International Finance Discussion Papers Number 1046 April 2012 Fiscal Consolidation in an Open Economy Christopher J. Erceg and Jesper Lindé NOTE: International Finance Discussion Papers are preliminary materials circulated to stimulate discussion and critical comment. References in publications to International Finance Discussion Papers (other than an acknowledgment that the writer has had access to unpublished material) should be cleared with the author or authors. Recent IFDPs are available on the Web at www.federalreserve.gov/pubs/ifdp/.

Fiscal Consolidation in an Open Economy (cid:3) Christopher J. Erceg Jesper LindØ (cid:3)(cid:3) Federal Reserve Board Federal Reserve Board and CEPR April 2012 Abstract This paper uses a New Keynesian DSGE model of a small open economy to compare how the e⁄ects of (cid:133)scal consolidation di⁄er depending on whether monetary policy is constrained by currencyunionmembershiporbythezerolowerboundonpolicyrates. Weshowthatthereare important di⁄erences in the impact of (cid:133)scal shocks across these monetary regimes that depend bothonthedurationofthezerolowerboundandonfeaturesthatdeterminetheresponsiveness of in(cid:135)ation. JEL Classi(cid:133)cation: E52, E58 Keywords: Monetary Policy, Currency Union, Fiscal Policy, Zero Lower Bound Constraint, New Keynesian Small Open Economy DSGE Model. (cid:3)Thispaperwaspresentedinthesession(cid:147)CurrencyUnionsandMacroeconomicPolicies(F4)(cid:148)organizedbyCharles EngelandchairedbyKennethD.West. WearegratefultoourdiscussantJordiGal(cid:237)forveryhelpfulcomments. The views expressed in this paper are solely the responsibility of the authors and should not be interpreted as re(cid:135)ecting theviewsoftheBoardofGovernorsoftheFederalReserveSystemorofanyotherpersonassociatedwiththeFederal Reserve System. (cid:3)(cid:3) Corresponding Author: Address: Division of International Finance, Federal Reserve Board, Mailstop 20, 20th and C Streets NW, Washington, D.C., 20551, USA. Telephone: 202-452-3055. Fax: 202-263-4850 E-mail addresses: christopher.erceg@frb.gov and jesper.l.linde@frb.gov

1. Introduction Given heightened concerns about debt sustainability, many countries are implementing ambitious (cid:133)scal consolidation plans in which government spending reductions often play a major role. The usual presumption is that the e⁄ects of government spending cuts on output are smaller when a country conducts an independent monetary policy (IMP) than when constrained by membership in a currency union, re(cid:135)ecting that interest rate cuts and currency depreciation appear to dampen the adverse impact on aggregate demand. While econometric analysis (e.g. Ilzetzki, Mendoza and Vegh, 2010) supports this view, it is unclear whether an IMP retains its comparative advantage if constrained by the zero lower bound, especially in light of (cid:147)closed economy(cid:148)analysis showing how a liquidity trap can amplify the government spending multiplier.1 This paper uses a New Keynesian DSGE model of a small open economy to compare the e⁄ects ofagovernmentspendingcutundertwoalternativeconstraintsonmonetarypolicy: currencyunion (CU) membership and the zero lower bound on interest rates (for an economy with an IMP). Given that adjustment of the policy rate is precluded in both monetary regimes for at least some time, the output e⁄ects of (cid:133)scal contraction are larger than under an unconstrained IMP. But we show that there are important di⁄erences across the two constrained regimes, re(cid:135)ecting that the relative size of the output contraction is highly sensitive to structural features which determine how the real exchange rate and long-term real interest rate respond to the (cid:133)scal consolidation. If in(cid:135)ation is fairly sensitive to the output gap (e.g., the Phillips Curve has substantial upward slope), output contracts more deeply under an IMP than a CU if policy rates are constrained from adjusting for a sustainedperiodofroughlytwoyearsormore. Importantly,theanchoringofthenominalexchange rateinaCUturnsouttobeablessinginsofarasitavoidsthelargeappreciation oftherealexchange rate that would occur in a persistent liquidity trap, and implies a smaller rise (if any) in long-term 1See e.g., Erceg and LindØ (2010a) and the references therein. 1

real interest rates. By contrast, if the Phillips Curve is very (cid:135)at, the output contraction under an IMP tends to be smaller than under a CU, re(cid:135)ecting that the real exchange rate depreciates even in a prolonged liquidity trap and long-term real interest rates fall. We conclude by arguing that the responsiveness of in(cid:135)ation during the recent global recession should be highly informative in discriminating between these contrasting predictions of the theory. 2. A New Keynesian Open Economy Model Our benchmark model is very similar to the small open economy model of Clarida, Gal(cid:237), and Gertler (2001). Households consume a domestic and foreign good that are imperfect substitutes. To rationalize Calvo-style price rigidities, the domestic good is assumed to be a comprised of a continuum of di⁄erentiated intermediate goods, each of which is produced by a monopolistically competitive (cid:133)rm. The government consumes some of the domestic good and (cid:133)nances itself through lump-sum taxes. The home economy is small in the sense that it does not in(cid:135)uence any foreign variables, and (cid:133)nancial markets are complete. To save space, we present only the log linearized model in which all variables are expressed as percent or percentage point deviations from their steady state levels, and we omit all foreign variables. Under an independent monetary policy, the key equations are given by: x = E x (cid:27)^open(i E (cid:25) r pot ); (1) t t t+1 t t t+1 t (cid:0) (cid:0) (cid:0) (cid:25) = (cid:12)E (cid:25) +(cid:20) x ; (2) t t t+1 x t i = max( i;(cid:13) (cid:25) +(cid:13) x ); (3) t (cid:25) t x t (cid:0) 2

y = (cid:27)^open(cid:28) +g g +(1 g )(1 !)(cid:23) (cid:23) (4) t t y t y c t (cid:0) (cid:0) 1 pot y = [g g +(1 g )(1 !)(cid:23) (cid:23) ] (5) t (cid:30) (cid:27)^open y t (cid:0) y (cid:0) c t mc 1 1 pot (cid:28) = (1 )[g g +(1 g )(1 !)(cid:23) (cid:23) ] (6) t (cid:0)(cid:27)^open (cid:0) (cid:30) (cid:27)^open y t (cid:0) y (cid:0) c t mc pot pot pot r = E (cid:28) (cid:28) ; (7) t t t+1(cid:0) t where (cid:27)^open = (1 g )[(1 (cid:23) )(1 !)2(cid:27) + !(2 !)" ] and the superscript (cid:145)pot(cid:146)denotes the level y c P (cid:0) (cid:0) (cid:0) (cid:0) that would prevail under completely (cid:135)exible prices. As in Clarida et al, the (cid:133)rst three equations represent the New Keynesian open economy IS curve, Phillips Curve, and monetary rule, respectively, that jointly determine the output gap pot (x = y y ); price in(cid:135)ation ((cid:25) ); and the nominal policy rate (i ); with the key di⁄erence t t t t t (cid:0) that equation (3) requires the policy rate to remain above its lower bound ( i). Thus, the output (cid:0) gap x depends inversely on the deviation of the real interest rate (i E (cid:25) ) from the potential t t t t+1 (cid:0) real interest rate r pot , with the sensitivity parameter (cid:27)^open varying positively with the household(cid:146)s t intertemporal elasticity of substitution in consumption (cid:27) and substitution elasticity " between P foreign and domestic goods (the relative weight on the latter rises with trade openness !): The Phillips curve slope (cid:20) in equation (2) is the product of parameters determining the sensitivity x of in(cid:135)ation to marginal cost (cid:20) and of marginal cost to the output gap (cid:30) , i.e. (cid:20) = (cid:20) (cid:30) . mc mc x mc mc From equation (5), a contraction in government spending g (g is the government spending share t y of steady state output) or negative taste shock (cid:23) ((cid:23) is a scaling parameter) reduces potential t c pot output y : Even so, both of these exogenous shocks, if negative, cause the the potential terms of t 3

pot pot trade (cid:28) to depreciate (a rise in (cid:28) in equation 6) because they depress the marginal utility of t t consumption (noting (cid:30) (cid:27)^open > 1). If both shocks follow stationary AR(1) processes, and hence mc pot have front-loaded e⁄ects, a reduction in government spending or negative taste shock reduces r . t Finally, the nominal exchange rate e equals p +(cid:28) where p = p +(cid:25) . t t t t t 1 t (cid:0) Giventhattheformoftheequationsdeterminingoutput,in(cid:135)ation,andinterestratesisidentical tothatinaclosedeconomy(cid:150)asemphasizedbyClaridaetal(cid:150)resultsfromextensiveclosedeconomy analysis,e.g.,ErcegandLindØ(2010a)aredirectlyapplicableforassessingtheimpactofgovernment spending shocks in a liquidity trap. We next consider how the model is modi(cid:133)ed for the CU case (largely following the analysis of Corsetti et al. 2011). A CU member takes the nominal exchange rate as (cid:133)xed, so that the terms of trade (cid:28) is simply the gap between home and foreign price levels, i.e., (cid:28) = (p p ) = p .2 t t t (cid:3)t t (cid:0) (cid:0) (cid:0) Moreover, the home economy is assumed to be small enough that the policy rate is e⁄ectively exogenous. Given that equation (4) implies that the output gap is proportional to the terms of trade gap, i.e., x = (cid:27)^open((cid:28) (cid:28) pot ); the price setting equation (2) may be expressed as a second t t t (cid:0) order di⁄erence equation in the terms of trade (cid:28) (cid:28) = (cid:12)((cid:28) (cid:28) )+(cid:20) (cid:27)^open((cid:28) (cid:28) pot ); (8) t t 1 t+1t t x t t (cid:0) (cid:0) j (cid:0) (cid:0) yielding a solution of the form: (cid:21) (cid:28) = (cid:21)(cid:28) +(cid:20) (cid:27)^open (cid:28) pot ; (9) t t (cid:0) 1 x 1 (cid:12)(cid:26)(cid:21) t (cid:0) The persistence parameter (cid:21) = 0:5(a a2 4=(cid:12) ), where a = (1)(1+(cid:12) +(cid:20) (cid:27)^open); lies between 0 (cid:0) (cid:0) (cid:12) x p andunity,and(cid:26)isthepersistenceoftheshockprocesses(assumedtobethesameforthetasteshock and government spending). Equation (9) has two important implications. First, because (cid:21) > 0; pot a contraction in government spending (cid:150)which raises (cid:28) by equation (6) (cid:150)moves (cid:28) in the same t t 2As the real exchange rate is proportional to (cid:28) , we use the terms interchangeably. t 4

direction, implying a depreciation. Together with equation (4), this implies that the government spendingmultiplierm isstrictlylessthanunity,i.e.,m = 1 dyt = 1+(cid:27)^open d(cid:28)t d(cid:28)p t ot < 1(recalling t t gy dgt gy d(cid:28)pot dgt t that d(cid:28)p t ot < 0): Second, as (cid:20) (cid:27)^open becomes very small, (cid:21) rises toward unity and the coe¢ cient dgt x pot pot on (cid:28) shrinks, implying very gradual adjustment of the terms of trade to (cid:28) (and hence to a t t change in government spending); conversely, the terms of trade adjustment is more rapid if (cid:20) (cid:27)^open x is larger. In economic terms, the terms of trade adjusts more quickly if the Phillips Curve slope is higher (high (cid:20) ), or if aggregate demand is relatively sensitive to the terms of trade (high (cid:27)^open): x 2.1. Simulation Results The left panel of Figure 1 shows the e⁄ects of a 1 percent of baseline GDP cut in government spending under a calibration in which the Phillips Curve parameter relating in(cid:135)ation to marginal cost (cid:20) = :025. This calibration is towards the higher side of empirical estimates, while the right mc panel shows a calibration which sets (cid:20) = :007, towards the very low end of empirical estimates. mc If factors were completely mobile, these calibrations would imply mean price contract durations of about 7 and 12 quarters, respectively, but (cid:150)as emphasized by an extensive literature (e.g., Altig et al., 2011) (cid:150)the reduced form slopes could be regarded as consistent with much shorter contract durations under reasonable assumptions about strategic complementarities.3 As seen in the upper pot panels, the potential terms of trade (cid:28) depreciates (rises) initially, and then dies out slowly at t the rate (cid:26) = 0:95: This fall in the relative price of domestically-produced goods re(cid:135)ects that the government spending cut boosts home consumption relative to foreign consumption. Moreover, 3Weadoptastandardquarterlycalibrationbysettingthediscountfactor(cid:12) =0:995;andsteadystatenetin(cid:135)ation (cid:25) = :005 so that i = :01. We set (cid:27) = 1 (log utility); the capital share (cid:11) = 0:3; the Frisch elasticity of labor supply 1 = 0:4; the government spending share g = 0:2; and the taste shock parameter (cid:23) = 0:01 (implying (cid:30) (cid:31) y c mc = (cid:31) + 1 + (cid:11) = 5.1). In the absence of the ZLB, monetary policy completely stabilizes output and in(cid:135)ation 1 (cid:11) (cid:27)^open 1 (cid:11) (cid:0) (cid:0) (achieved by making (cid:13) in eq. 3 arbitrarily large). Finally, the open economy parameters ! = 0:3, and " = 1:5. (cid:25) p 5

pot the positive wealth e⁄ect reduces potential output y (lower panels). A country with an IMP (cid:150)if t unconstrained by the zero lower bound (cid:150)couldachievethis(cid:135)exiblepriceallocationsimplythrougha monetaryrule(3)thatrespondedveryaggressivelytoin(cid:135)ationand/ortheoutputgap. Undersuch pot a rule, the terms of trade (cid:28) would track (cid:28) exactly, and given that in(cid:135)ation remains unchanged t t pot from baseline, both the real and nominal interest rate would decline in line with r , re(cid:135)ecting t pot that consumption would be expected to fall after its initial rise. Thus, output would track y t irrespective of the degree of price stickiness. With the price level constant, the jump in the real exchange rate would be achieved through nominal exchange rate depreciation. IntheCUcase,thenominalexchangerateis(cid:133)xed,sothatthegovernmentspendingcutinitially pot boosts (cid:28) by more than the actual terms of trade (cid:28) (upper panels): The negative terms of trade t t pot gap (cid:28) (cid:28) which may be regarded as an (cid:147)overvalued(cid:148)terms of trade (cid:150)causes output to t t (cid:0) (cid:0) fall persistently below potential. The negative output gap causes in(cid:135)ation to fall persistently (cid:150) implying a progressive depreciation of the terms of trade (cid:150)and the associated narrowing of the termsoftradegapeventuallymovesoutputtowardspotential. Asnotedpreviously,theadjustment process proceeds more quickly with shorter-lived price contracts, which explains why the output contraction in the left panel is smaller and less persistent than in the right panel. In addition, factors that raise the sensitivity (cid:27)^open of the output gap to the terms of trade gap (cid:150)such as a higher trade elasticity " (cid:150)would also speed-up the adjustment. Importantly, although the terms p of trade adjusts sluggishly in line with the price level, it does at least move in the (cid:147)right direction(cid:148) for narrowing the output gap. Moreover, as highlighted by Corsetti et al. (2011), the ex ante long-term real interest rate falls in response to a persistent fall in government spending: although in(cid:135)ation declines in the near-term, the terms of trade (and hence the price level) must eventually revert to steady state, implying some rise in long-run expected in(cid:135)ation. While greater price (cid:135)exibility cushions the impact of a government spending cut in a CU, more 6

price (cid:135)exibility (cid:150)or more generally, a larger Phillips Curve slope (cid:20) (cid:150)can greatly deepen the x contraction that occurs under an IMP subject to the zero bound constraint, and imply an output multiplier much larger than in a CU. In this vein, Figure 1 shows the e⁄ects of the government spending contraction under an IMP against the backdrop of initial conditions which imply an ten quarterliquiditytrap(i.e., anegativetasteshockthatisscaledtoinducealiquiditytraplastingten pot quarters in the absence of the (cid:133)scal shock). As the government spending shock reduces r while t the policy rate remains (cid:133)xed, the output gap would contract even if expected in(cid:135)ation remained constant. However, the output contraction is reinforced by a persistent decline in in(cid:135)ation that is particularly large when price adjustment is relatively rapid (the left panel). Thus, the peak output decline is 1.5 percent under the IMP, compared with 0.8 percent in a CU. Importantly, the large output decline under the IMP re(cid:135)ects two factors. First, long-term ex ante real interest rates rise substantially, in contrast to the decline that occurs in a CU. Second, the rise in the real interest rate under an IMP implies a (cid:147)perverse(cid:148)initial appreciation of the terms of trade (as seen in the upper left panel). Thus, although the CU precludes the nominal exchange rate from adjusting, the lack of adjustment better cushions output than the appreciation that occurs under an IMP. Under more sluggish price adjustment, the multiplier is only 0.7 under an IMP, smaller than the multiplier of 0.9 in a CU. With in(cid:135)ation much less responsive, long-term real interest rates fall under an IMP, allowing a front-loaded terms of trade depreciation to cushion the impact on output. Overall, our results underscore that the same conditions which mitigate the e⁄ects of (cid:133)scal consolidation in a CU (cid:150)namely, an upward-sloping Phillips Curve (cid:150)exacerbate the e⁄ects under an IMP constrained by the ZLB; and conversely, a (cid:135)atter Phillips Curve tends to make an IMP look more attractive, since the real exchange rate can adjust immediately to lessen the bite on aggregate demand. While the results in Figure 1 consider the speci(cid:133)c case of a ten quarter liquidity trap, it is 7

natural to ask how long a liquidity trap is required for (cid:133)scal consolidation to produce a more contractionary e⁄ect under an IMP than a CU. To address this, Figure 2 plots the output response to a 1 percent of GDP contraction under di⁄erent assumptions about the duration of the liquidity trap under an IMP (generated by appropriately-sized adverse taste shocks). As in Figure 1, the left panel adopts the calibration in which price adjustment is relatively faster, while the right panel assumes that price adjustment is slower. In the former case, the output contraction becomes much more pronounced as the liquidity trap lengthens, with the multiplier in the case of a eight quarter liquidity trap exceeding the multiplier under a CU of 0.8. With a three year liquidity trap, the spending multiplier is nearly 3, as a sharp rise in long-term real interest rates (caused by lower expected in(cid:135)ation) causes a large improvement in the terms of trade (lower panel). In this environment, the anchoring of the long-run price level provided by a CU is very bene(cid:133)cial in insulating the economy from the potential pressures that can arise in a liquidity trap. By contrast, with slower price adjustment, the liquidity trap must last 12 quarters for the multiplier under an IMP to exceed that under a CU; for liquidity traps of less than two years, the front-loaded depreciation of the terms of trade signi(cid:133)cantly mitigates the e⁄ects of the spending cut on output. As the slope of the Phillips Curve becomes (cid:135)atter, the liquidity trap duration required to produce a larger output downturn than in a CU becomes progressively longer. Thusfar,ouranalysishasfocusedonasimplemodelthatabstractsfromanarrayofempiricallyrelevant nominal and real frictions. Even so, our key points continue to hold in more realistic open economy settings. Figure 3 shows responses to a 1 percent of GDP (cid:133)scal contraction in a larger-scale open economy model used in Erceg and LindØ (2010b) that embeds nominal wage and price rigidities, endogenous capital accumulation, rule-of-thumb consumers, and incomplete exchange rate passthrough in the short-run. The left panel with (cid:147)faster price adjustment(cid:148)adopts a calibration of the price and wage contract duration parameters that is broadly representative of 8

the estimates of the slopes of price and wage Phillips Curves based on data prior to the (cid:133)nancial crisis; speci(cid:133)cally,itadoptstheestimateofAltigetal. (2011)of(cid:20) = 0:014:Therightpanelshows mc estimates under an alternative calibration that imposes an extremely (cid:135)at price (and wage) Phillips Curve of (cid:20) = 0:002.4 The unconstrained IMP follows a Taylor rule, while the constrained policy mc is derived under the assumption that the liquidity trap lasts ten quarters. Under the calibration with relatively faster price adjustment, output declines over 2 percent after 4 quarters under the constrained IMP, compared with only about 0:8 percent in a CU. The larger output decline in the former case re(cid:135)ects a larger fall in in(cid:135)ation (middle left panel) (cid:150)which pushes up long-term real interest rates (cid:150)and a sizeable real appreciation of the exchange rate. Thus, the (cid:133)scal shock is ampli(cid:133)ed by a sharp contraction in private domestic demand and real net exports. In a CU, the real exchange rate depreciates slightly even in the near-term, and the long-term real interest rate is about constant. By contrast, under very slow price adjustment (cid:150)the right panel (cid:150)the e⁄ects of (cid:133)scal consolidation on output are modestly smaller under a constrained IMP than CU. The smaller output decline under an IMP re(cid:135)ects both a front-loaded exchange rate depreciation and fall in long-term real interest rates (since in(cid:135)ation barely moves, and policy rates fall after two years). For a shorter-lived liquidity trap, the advantages of an IMP are even larger.5 4Under (cid:147)faster price adjustment,(cid:148)the contract duration parameters for prices and wages are (cid:24) = 0:86 and p (cid:24) =0:82; respectively, while (cid:24) =0:95 and (cid:24) =0:90 under (cid:147)slower price adjustment.(cid:148)The model and calibration w p w of the other parameters are described in Erceg and LindØ (2010b). 5A permanent reduction in government spending would have very similar e⁄ects in the larger model to those shown in Figure 3. By contrast, a permanent cut would have no e⁄ect on the output gap (even in a liquidity trap) in the stylized model, as the absence of real rigidities allows for strong and immediate crowding-in e⁄ects. 9

3. Implications and Open Questions Conditional on some key structural parameters, including those highlighted above, our modeling framework can potentially help gauge whether the output e⁄ects of (cid:133)scal consolidation in an economy such as the United Kingdom (cid:150)where policy rates are constrained by the ZLB (cid:150)are likely to be larger than in a CU member such as Belgium. But given that even the qualitative answer hinges on factors that determine the responsiveness of in(cid:135)ation, which view does the evidence favor? There is substantial econometric evidence estimating the sensitivity of price in(cid:135)ation to marginal cost; as noted, the calibration in the left side of Figure 3 seems squarely in line with such evidence. On this basis, (cid:133)scal consolidation would have a deeper contractionary impact under an IMP provided that the ZLB was binding for over two years; and the seeming strictures of a CU would in fact ameliorate the output contraction. However,theresilienceofin(cid:135)ationduringtherecentglobalrecessionsuggeststhepossibilitythat theresponsivenessofin(cid:135)ationmaybeconsiderablylowerthanimpliedbymostexistingeconometric evidence. As seen in the left panel of Figure 3, the 1 percent of GDP (cid:133)scal contraction reduces in(cid:135)ation by 2 percentage points. Moreover, under the same calibration of price adjustment, a fall in output of say 6-8 percent or more below its pre-crisis trend path (cid:150)as was experienced by the UnitedStatesandEuropeduringtherecession(cid:150)wouldcausein(cid:135)ationandone-yearaheadexpected in(cid:135)ation to fall more than 4 percentage points below the central bank(cid:146)s in(cid:135)ation target if mainly driven by aggregate demand shocks. This implied decline in in(cid:135)ation is much larger than actually occurred either in the United States, where core in(cid:135)ation and market expectations of core in(cid:135)ation remained above 1 percent, or in major economies in Europe. It is quite conceivable that in(cid:135)ation behavior during the past few years can be rationalized as consistentwitheconometricevidencebasedonpre-crisisobservations. Forexample,(cid:133)nancialshocks and other shocks may have adversely impacted the supply side of the economy enough to square 10

observed in(cid:135)ation behavior with existing econometric evidence. However, future analysis may well point to a somewhat lower degree of in(cid:135)ation responsiveness. If so, outside of a very prolonged liquidity trap, our analysis would indicate that an economy with an IMP may be somewhat better poised to absorb the e⁄ects of (cid:133)scal consolidation than a CU, with real exchange rate and interest movements tending to cushion rather than amplify the impact. References Altig,David,Christiano,LawrenceJ.,Eichenbaum,MartinandJesperLindØ(2011),(cid:147)Firm-Speci(cid:133)c Capital, Nominal Rigidities and the Business Cycle(cid:148), Review of Economic Dynamics 14(2), 225-247. Clarida, Richard, Jordi Gal(cid:237) and Mark Gertler (2001), (cid:147)Optimal Monetary Policy in Open Versus Closed Economics: An Integrated Approach(cid:148), American Economic Review Papers and Proceedings 91, 248-252. Corsetti Giancarlo, Keith Kuester, and Gernot J. M(cid:252)ller (2011), (cid:147)Floats, Pegs, and the Transmission of Fiscal Policy,(cid:148)Journal Econom(cid:237)a Chilena 14(2), 5-38, Erceg, Christopher and Jesper LindØ (2010a), (cid:147)Is There A Fiscal Free Lunch In a Liquidity Trap?(cid:148), International Finance Discussion Papers No. 1003. Erceg, Christopher and Jesper LindØ (2010b), (cid:147)Asymmetric Shocks in a Currency Union with Monetary and Fiscal Handcu⁄s(cid:148), NBER International Seminar on Macroeconomics 2010, 95-135. Ilzetzki, Ethan, Enrique G. Mendoza and Carlos A. VØgh (2010), (cid:147)How Big (Small?) are Fiscal Multipliers?(cid:148), NBER Working Paper Series No. 16479. 11

Figure 1: Persistent Contraction in Government Spending Fast Price Adjustment Terms of Trade 1 0.5 0 −0.5 −1 0 4 8 12 16 Quarters tnecreP 1 0.5 0 −0.5 −1 0 4 8 12 16 Quarters tnecreP Slow Price Adjustment Terms of Trade Ind. Policy, 10q Liq. Trap Currency Union Potential Inflation (APR) 0 −0.5 −1 −1.5 −2 −2.5 0 4 8 12 16 Quarters tnecreP Inflation (APR) 0 −0.5 −1 −1.5 −2 −2.5 0 4 8 12 16 Quarters tnecreP Output 0 −0.5 −1 −1.5 0 4 8 12 16 Quarters tnecreP Output 0 −0.5 −1 −1.5 0 4 8 12 16 Quarters tnecreP

0 −0.5 −1 −1.5 −2 −2.5 −3 12 11 10 9 8 7 6 5 4 3 2 1 0 tnecreP Figure 2: Impact Output Response to Immediate Government Spending Cut With Independent Monetary Policy and in a Currency Union Output 0 −0.5 −1 −1.5 −2 −2.5 −3 12 11 10 9 8 7 6 5 4 3 2 1 0 Liquidity Trap Duration Liquidity Trap Duration tnecreP Fast Price Adjustment Slow Price Adjustment Output Independent Policy Currency Union Potential 1 0.5 0 −0.5 −1 −1.5 −2 12 11 10 9 8 7 6 5 4 3 2 1 0 tnecreP Terms of Trade 1 0.5 0 −0.5 −1 −1.5 −2 12 11 10 9 8 7 6 5 4 3 2 1 0 Liquidity Trap Duration tnecreP Terms of Trade Liquidity Trap Duration

Figure 3: Persistent Government Spending Cut in Large Model Faster Price Adjustment Real Exchange Rate 1 0.5 0 −0.5 −1 −1.5 −2 0 4 8 12 16 Quarters tnecreP Slower Price Adjustment Real Exchange Rate 1 0.5 0 −0.5 −1 −1.5 −2 0 4 8 12 16 Quarters tnecreP Inflation (APR) 0 −0.5 −1 −1.5 −2 0 4 8 12 16 Quarters tnecreP 0 −0.5 −1 −1.5 −2 0 4 8 12 16 Quarters tnecreP Inflation (APR) Ind. Policy, 10q Liq. Trap Currency Union Ind. Policy, Unconstrained Output 0 −0.5 −1 −1.5 −2 0 4 8 12 16 Quarters tnecreP Output 0 −0.5 −1 −1.5 −2 0 4 8 12 16 Quarters tnecreP