Fiscal consolidation using the example of Germany

Finance and Economics Discussion Series Divisions of Research & Statistics and Monetary Affairs Federal Reserve Board, Washington, D.C. Fiscal consolidation using the example of Germany Tobias Cwik 2012-80 NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminary materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors. References in publications to the Finance and Economics Discussion Series (other than acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers.

∗ Fiscal consolidation using the example of Germany TobiasCwik BoardofGovernorsoftheFederalReserveSystem October11,2012 Abstract After the run up in debt-to-GDP ratios around the world in the aftermath of the financial crisis and the associated lower fiscal space, the question of prudent fiscal consolidation is back on the agenda. In this paper, I study the macroeconomic implications of fiscal consolidation triggered by the newly introduced "debt brake" in Germany, which dampens the accumulation ofdebt. Iaddressthisquestionusingamedium-sizenewKeynesianDSGEmodelforGermany. Themodelincludesthegovernmentdebt-to-GDPratio,governmenttransfers,labourincometax, consumptiontaxandcapitaltaxrevenues.Ifindthatthe"debtbrake"enforcesfiscalconsolidation intimesofeconomicexpansionswithoutconstrainingfiscalpolicymakersintimesofrecessions. Ialsofindthatthedebtbrakeraisesthegovernmentspendingmultiplierinitiallybutnotovertime. Finally,thedebtbrake,withafiscalconsolidationonthegovernmentspendingandtransfersside, leadstoasignificantstabilizationoftheprivatesectorwithoutincreasingthevolatilityofthefiscal instruments. Keywords: DSGEmodelling,BayesianEstimation,Fiscalpolicy, Fiscalconsolidation,Debtbrake JEL-Codes: C11,E62,H63 ∗Theanalysisandconclusionssetfortharethoseoftheauthoranddonotindicateconcurrencebyothermembersofthe researchstaffortheBoardofGovernors.Pleaseaddresscorrespondencetocwik@wiwi.uni-frankfurt.de

1 Introduction Duringthefinancialcrisis,advancedeconomieshavetakensizablemeasurestostabilizethebanking sectoranddomesticdemand.1 Thishasledtoalargebuildupofdebt. InGermany,thedebt-to-GDP ratio has increased to 83.2% of GDP after 65.3% in the end of 2007, exceeding the reference value of the Maastricht Treaty by 23%.2 In the United States, the general government gross debt-to-GDP ratioincreasedfrom62.1%in2007to97.6%in2011. Blanchard et al. [2010] propose in their paper "Rethinking Macroeconomic Policy" to reduce debttargets. Theywrite: "Still,thelessonfromthecrisisisclearlythattargetlevelsshouldbelower than those observed before the crisis. The policy implications for the next decade or two are that, when cyclical conditions permit, major fiscal adjustment is necessary and, should economic growth recover rapidly, it should be used to reduce debt-to-GDP ratios substantially, rather than to finance expenditureincreasesortaxcuts." To anchor the indebtedness, Germany’s government introduced a "debt brake" in 2011.3 This "debtbrake"hasfivecharacteristics. First,ithasnoexplicitdebttarget. Itisadeficitrule,attempting tolowerthegovernmentdeficitandallowingthedebt-to-GDPratiotodeclinethroughGDPgrowth. Second, the rule enforces a maximum average structural deficit of 0.35% of GDP. The target is the structuraldeficitnottheactualdeficit. Attheexpenseofcomplication,thisconcepthastheadvantage of allowing a countercyclical fiscal stance in recessions. Third, deviations of the structural deficit from the target level of 0.35% are booked on an adjustment account, which must be cleared over time. This smooths the fiscal consolidation over a longer period, leading to less abrupt changes in fiscalpolicyinstruments. Fourth,fiscalconsolidationtriggeredbythedebtbrakeiscappedtoallowa maximum adjustment offiscalinstruments of 0.35% ofGDP annually. Andlast, the consolidatonis state-dependent, sothattheGermangovernmentwillonlyreducespendingandtransfersorincrease taxrateswhentheeconomyisnotinrecession. In this paper, I study the macroeconomic implications of fiscal consolidation triggered by the debtbrakedescribedabove. IconstructaDSGEmodelwithadetailedgovernmentsector. Themodel builds on New-Keynesian DSGE models à la Smets and Wouters [2007] or Christiano et al. [2005], which have shown to fit the data well. Because the focus of this paper is on government debt, I allow for rich dynamics induced by the interactions between policy variables and debt. Following Kollmann [1998] and Leeper et al. [2010], government spending, transfers, income tax, capital tax, 1SeeCwikandWieland[2009]foranoverviewofthefiscalpackagesoftheelevenbiggesteuroareacountries. 2TheMaatrichtTreatyobligesMemberStatesoftheEuropeanUniontoavoiddeficitsexceeding3%ofGDPanddebt levelsabove60%ofGDP. 3TheGermandebtbrakewasproposedbythe"FöderalismuskommissionII",signedintolawfromtheGermanparliament onthe29thMay2009andbecomesoperativein2011.However,therulewillbelegallybindingforthefederalgovernment from2016onwardsandforthestategovernmentsfrom2020onwards. 2

and consumption tax can respond to government indebtedness. In addition, the income and capital taxratesadjusttothestateoftheeconomytoallowprogressivetaxation. Sincetheliteratureonfiscal policy has emphasized the importance of non-Ricardian consumers in DSGE models, I introduce a share of rule-of-thumb consumers as described in Gali et al. [2007] together with forward-looking households in the model. The model is estimated on German data from 1983Q1 up to 2009Q4 with Bayesiantechniquesusing12timeseries. Myestimationsampleendsbeforetheintroductionofthe debt brake. To analyze the implications of the debt brake, I supplement the estimated fiscal reaction functionswiththedebtbrakepolicy. I have three main findings. First, the debt brake works asymmetrically. In general, after expansionary shocks, government revenues increase, whereas spending and transfers initially drop. This improves the government budget and leads government debt to fall. But this additional fiscal space has a second round effect on the fiscal instruments. Lower debt creates incentive for fiscal policy makerstolowertaxratesandincreasestructuralspending. Both leadtoanincreaseinthe structural deficit. Thedebtbrakeenforcesafasterfiscalconsolidationintimesofeconomicexpansionwithout constraining fiscal policy makers in times of contractionary shocks. Second, the debt brake initially increases the government spending multiplier independent of the type of fiscal consolidation. Althoughaconsolidationofspendingandtransfersleadstoamoreoutputfavourableoutcome. Agents expect that through the debt brake, the increase in discretionary spending will be offset faster in the future by lower spending and transfers or by higher taxes, both of which lowers the wealth effect and encourages private consumption. But after two years, when fiscal consolidation takes place, the governmentspendingmultiplierislowerwiththedebtbrake. Andthird,thedebtbrakestabilizesthe economy,reducingthevolatilityofkeymacroeconomicvariableswithoutincreasingthatoffiscalinstruments. Thestabilizationismorepronouncedforadebtbrakewithspending-basedconsolidation, whichisstate-dependentandthereforeonlyleadstolowergovernmentspendingandlowertransfers tohouseholdswhentheeconomyisgrowing. Theremainderofthepaperisorganizedasfollows. Section2providesadescriptionoftheDSGE model. Section 3 explains the estimation procedere, section 4 shows the out-of-sample forecast for thedebt-to-GDPratioandsection5discussestheimplicationsofthedebtbrake. Section6concludes. 2 Model Inthissection,IdevelopaNew-KeynesianDSGEmodelthatincludesthemaintransmissionschannels of fiscal policy. Most of the model features are standard and familiar from so-called medium scale DSGE models, as put forward, for instance, in Christiano et al. [2005] or Smets and Wouters [2007]. A fraction of the representative households are forward-looking, smooth their consumption overtimebybuyingdomesticgovernmentbonds,ownthecapitalstock,whichisrentedtogetherwith 3

laborservicestointermediategoodsproducersonaperiod-by-periodbasis,andpaysalestax,labour income tax and capital income tax. The other fraction, the rule-of-thumb consumers, consume their after tax labour income. They work an equal proportion of their time as Ricardian households and receivethesamewage. Adjustinginvestmentiscostly. Iassumethereisacontinuumofintermediategoodproducersoperatingundermonopolisticcompetition and being constrained in price setting à la Calvo. They produce using a Cobb-Douglas production function and rent capital and labour services in competitive factor markets. Labour services are provided by an intermediate labour union sector, which pools labour services from Ricardian and non Ricardian households and set wages. Wages are sticky á la Calvo. Final goods firms combine intermediate goods to provide final goods for private consumption, investment and government consumption. I introduce policy rules for all government instruments in the model. Fiscal authorities are assumed to adjust the tax rates, government spending, and transfers with respect to the current state of the economy and government indebtedness. The debt brake supplements the policy rules in the following way. I define the structural government deficit as the deficit, which is present when there is no adjustment of fiscal instruments to output and the tax base is at their equilibrium value. If the structural government deficit is greater than zero, the amount is booked on an adjustment account, which is cleared over time by appropriately adjusting government revenues or spending. I close the model with a characterization of monetary policy in terms of an interest rate feedback rule. In the followingIgiveaformalexpositionofthemodel. 2.1 Firms Finalgoodsarecompositesofintermediategoodsproducedbyacontinuumofmonopolisticcompetitivefirmsandareusedfordomesticconsumption,C ,investment,I ,andgovernmentspending,G . t t t Iusei ∈ [0,1]toindexintermediategoodfirmsaswellastheirproductsandprices. Finalgoodsfirms operate under perfect competition and purchase intermediate goods, Y (i). They use the following t Kimball aggregation technology with the elasticity of demand being an increasing function of the relativepriceoftheintermediategoodP (i)andthefinalgoodP t t 1 Y (i) t G ;λ di = 1 (1) p,t Y Z0 (cid:18) t (cid:19) G is a concave and increasing function such that G(1) = 1 and (1+λ ) is the time-varying price p,t markupwith lnλ p,t = (1−ρ p )lnλ p +ρ p lnλ p,t−1 +θ p ǫ p,t−1 +ǫ p,t (2) 4

The representative final goods firms produce Y while minimizing expenditures. The resulting det mandfunctionforanindividualintermediategoodiisgivenby P (i) 1 Y (i) Y (i) Y (i) = Y G ′−1 t G ′ ( t ) t di (3) t t P Y Y (cid:20) t Z0 t t (cid:21) Theproductionofintermediategoods,Y (i),isgovernedbyaCobb-Douglasproductionfunction t Y (i) = Z Ks(i)α γtL (i) 1−α −γtΦ, (4) t t t t (cid:2) (cid:3) whereL (i)andKs(i)respectivelydenotelabourandcapitalemployedbyfirmiandΦdenotefixed t t costs of production, which are set to ensure that profits are zero in steady state. γt represents the labour-augmentingdeterministicgrowthrateintheeconomyandZ governstotalfactorproductivity, t whichisgivenbythefollowingshockprocess lnZ t = ρ z lnZ t−1 +ǫ z,t (5) whereǫ isiiddistributed. LetW andRkdenotethenominalwagerateandtherentalrateofcapital, z,t t t respectively. Minimizing the costs of producing intermediate goods implies for (nominal) marginal costs MC = α −α(1−α) −(1−α)W1−α(Rk)αγ −(1−α)t(Z ) −1, (6) t t t t which are independent of the level of production and identical across firms, because both factors of productioncanbeadjustedfreelyacrossfirms. Iassumethatpricesettingisconstrainedexogenouslybyadiscretetimeversionofthemechanism suggestedbyCalvo[1983]. Eachfirmhastheopportunitytochangeitspricewithagivenprobability 1−ξ p . Firmsthatdonotreoptimizeinacertainperiodindextheirpricetosteadystateinflationπ∗. The optimal price P (i) set by the firm that is allowed to re-optimize at time t results from the t followingoptimizationproblem e ∞ Ξ P maxE t (ξ p β)s t+s t P t (i)(π∗)s−MC t+s Y t+s (i) (7) Ξ P t t+s X s=0 h i e subjecttothedemandfunctiondefinedby(3).4 2.2 Households Thereisacontinuumofhouseholdsindexedbyh ∈ [0,1]. Ashare1−ω ofthesehouseholdsmakes optimizing, forward-looking decision. They are indexed by j ∈ [0,1−ω). These households have access to financial markets. They buy and sell government bonds B (j) and accumulate physical t 4ProfitsarediscountedwiththestochasticdiscountfactoroftheRicardianhouseholds, βsΞt+sPt,whoaretheowners ΞtPt+s ofthefirms. 5

capital K (j). Given capital utilization costs a(u (j)), they decide how much of the accumulated t t capitaltheyrenttofirms. TheyreceivewageincomeWh(j)L (j),dividendpaymentsfromthefirms, t t and profits from the labour unions and pay capital tax τk, consumption tax τc and labour income t t tax τn to the government. Their decisions are made so as to maximize a utility function that is t separable in consumption C (j) and labour supply L (j). Specifically, I use the log utility function t t forconsumptiontoensureabalancedsteadystategrowthpathasputforwardinKingetal.[1988].5 ∞ 1 E t βslog(C t+s (j))− L t+s (j)1+σ l (8) 1+σ l s=0 X subjecttothebudgetconstraint B (j) Wh (j)L (j) (1+τc )C (j)+I (j)+ t+s ≤ (1−τn ) t+s t+s t+s t+s t+s d P t+s P t+s t+s t+s +(1−τ t k +s ) R t k +s u t+s P (j)K t+s−1 (j) −a(u t+s (j))K t+s−1 (j) (cid:20) t+s (cid:21) + B t+s−1 P (j)R t+s−1 +τ t k +s δK t+s−1 (j)+ D P iv t+s t+s t+s thecapitalaccumulationequation I (j) K t (j) = (1−δ)K t−1 (j)+µ t 1−S t I t (j) (9) (cid:20) (cid:18) I t−1 (j) (cid:19)(cid:21) andthedefinitionofemployedcapital K t s(j) = u t (j)K t−1 (j) (10) where δ denotes the depreciation rate. I assume for simplicity that the utilisation costs of physical capital and physical capital depreciation are exempted from taxation as in Coenen et al. [2008]. It may be costly to adjust the level of investment, I . Here S is the adjustment cost function, with t ′ ′′ S(γ) = S (γ) = 0 and S > 0 ensures that the steady state capital stock is independent of the investment adjustment costs.6 d is an exogenous premium in the return to bonds as in Edge et al. t [2008]givenby lnd t = ρ d lnd t−1 +ǫ d,t (11) andµ isastochasticshocktothepriceofinvestmentrelativetoconsumptiongoodsofthefollowing t form lnµ t = ρ µ lnµ t−1 +ǫ µ,t (12) 5I estimated the model with external habits, price indexaton and wage indexation and found values close to 0.1. To keep the parameter space sparse I excluded the exogenous persistence parameters. Kiley [2010] finds that introducing rule-of-thumbconsumerscanexplainalowerhabitparameter. 6SeeChristianoetal.[2005]. 6

The remaining share ω of non Ricardian households is indexed by s ∈ [1 − ω,1]. They simply consume their disposable income, which is given by the after-tax wage income and the transfer paymentsfromthegovernment. WhL (s) (1+τc)C (s) = (1−τn) t t +T (s) (13) t t t P t t Aggregating over all households implies that overall consumption is a weighted average of the consumptionfunctionofrationalandrule-of-thumbconsumers 1 C = C dh = (1−ω)C +ωC (14) t h,t j,t s,t Z0 Labourpackersbuythelaboursuppliedbythehouseholdsfromlabourunionsl ∈ [0,1],package L usingthefollowingaggregationtechnologyandresellittotheintermediategoodsproducers. t 1 1+λ w,t 1 L t = L t (l)1+λw,tdl (15) (cid:20)Z0 (cid:21) (1+λ )denotesthetimevaryingwagemarkupinthelabourmarketdrivenbythemovingaverage w,t processoftheform lnλ w,t = (1−ρ w )lnλ w +ρ w lnλ w,t−1 +θ w ǫ w,t−1 +ǫ w,t (16) Thelabourpackersmaximizeprofitsinaperfectlycompetitiveenvironmentleadingtothefollowing demandforindividuallabourl. −(1+λw,t) W t (l) λw,t L (l) = L (17) t t W t (cid:18) (cid:19) where W is the aggregate wage index. Labour unions allocate and differentiate the labour services t fromthehouseholdsandsetthesamenominalwagerateforbothtypesofhouseholds. Theychoose thewagegivennominalrigiditiesálaCalvosubjecttothelabourdemandequation(17)andthelabour supplydecisionoftheRicardianhouseholds Wh (1+τc) t = t C (j)Lσ l (18) P (1−τn) t t t t Labour unions that cannot adjust the wage in a given period index their wage to the deterministic growthrateγ andsteadystateinflation. Thewagesettingproblemforaunionthatcanadjustitswage inperiodt,W (l),becomes t ∞ f maxE t (ξ w β)s Ξ Ξ t+ P s P t W t (l)(γπ∗)s−W t h +s L t+s (l) (19) t t+s X s=0 h i f 7

2.3 Fiscalauthorities The government raises consumption, capital, and labour income taxes and decides on government expendituressubjecttothefollowingbudgetconstraint R t−1 B t−1 B t G +T + = Ψ + (20) t t t P P t t whererealgovernmentrevenuesaregivenby WhL Rku Ψ t = τ t n P t t +τ t kK t−1 P t t −a(u t )−δ +τ t cC t (21) t t (cid:18) (cid:19) AsinKollmann[1998], Iassumethattaxesareadjustedtostabilizethegovernmentdebt-to-gdp ratio. Additionally, the labour income and capital tax rates respond to the state of the economy to allow for automatic stabilization i.e. progressive taxation. For illustrative purposes, I linearize the reactionfunctions,wherex t = x t x −x definesthepercentagedeviationofavariablefromtrend. 7 Y (1−τn)Y τ t n = ρnτ t n −1 +φn b B bB t−1 −Y t−1 +φn y Y t +τ t n,ǫ + τnWhL Φ(∆Y t ,AC t−1 ;ρ) (22) (cid:18) (cid:19) (cid:16) (cid:17) b b Y b b b b γY b d τ t k = ρkτ t k −1 +φk b B B t−1 −Y t−1 +φk y Y t +τ t k,ǫ + τkK(RK −δ) Φ(∆Y t ,AC t−1 ;ρ) (23) (cid:18) (cid:19) (cid:16) (cid:17) b b τ t c = ρcτ t c − b 1 +φc b b B Y B t−1 −Y t b −1 + b τ t c,ǫ + τ Y cC Φ(∆Y t ,AC t−1 b ;ρ d ) (24) (cid:18) (cid:19) where φn b , φc b a b nd φk b m b easure the elabsticity obf labour i b ncome, consumpbtiodn and capital tax to government debt and φn and φk the elasticity of the tax rates w.r.t. output. ρn, ρk and ρc capture the y y persistence in the adjustment of tax rates by policy makers. τ n,ǫ, τ k,ǫ and τ c,ǫ are iid shocks to pick t t t uptheresidualvolatilityinthetaxrates. ThefunctionΦ(∆Y t ,AC t−1 ;ρ)governsfiscalconsolidation b b b triggered by the debt brake, which depends on the value on the adjustment account AC t−1 , output b d growth ∆Y and the speed of clearing on the adjustment account ρ. The term in front of the funct d tion Φ(∆Y t ,AC t−1 ;ρ) is the inverse of the tax elasticity of the respective revenues in steady state b or in other words by how much policy makers need to adjust the respective tax rates to increase tax b d revenuesby1%ofGDP.Iwillexplainthedebtbrakemechanisminmoredetailinthenextsubsection. Thefiscalrulesforgovernmentspendingandtransferpaymentsaredefinedinasimilarway. Y G t = ρGG t−1 −φG b B B t−1 −Y t−1 −φG y Y t +Gǫ t −Φ(∆Y t ,AC t−1 ;ρ) (25) (cid:18) (cid:19) (cid:16) (cid:17) b b b b b b b d Y T t = ρTT t−1 −φT b B B t−1 −Y t−1 −φT y Y t +T t ǫ−Φ(∆Y t ,AC t−1 ;ρ) (26) (cid:18) (cid:19) (cid:16) (cid:17) 7Althoughbthereactiobnfunctionsoffiscbalpolicymbakersareexpresbsedintebrmsoflineabrizeddvariables, theupperand lowerboundontheadjustmentaccountandthestate-dependentclearingrenderstheadjustmentofthegovernmentinstrumentsnonlinear,asshowninthenextsubsection. 8

Transfers and spending are reduced, when debt increases and are increased when output falls below trend to capture higher transfer payments in recessions i.e. higher unemployments benefits. Again, I account for persistence in the adjustment of the fiscal instruments and there is possible downward pressureonspendingandtransfersthroughfiscalconsolidation. 2.4 Structuralgovernmentdeficit For the evaluation of the fiscal stance, the debt brake uses the structural government deficit, that is the deficit which would be present when the economy was on the balanced growth path, instead of the actualdeficit. The advantageofthis conceptis thatit allowsfor countercyclicalfiscalpolicybut restrainsexcessivedeficitsinthemediumrun. Themodeloutlinedinthesubsectionsaboveprovides anaturalmeasureofthestructuraldeficit DS = GS +TS + R t−1 B t−1 −ΨS − B t−1 (27) t t t P t t P t−1 Structuralgovernmentspending,transfers,andtaxrevenues(GS,TS,ΨS)areobtainedbysetting t t t theoutput-elasticitiesoftaxes,spendingandtransferstozero(φn = 0,φk = 0,φ g = 0,φt = 0). For y y y y example, Y GS t = ρGGS t−1 −φG b B B t−1 −Y t−1 +Gǫ t −Φ(∆Y t ,AC t−1 ;ρ) (28) (cid:18) (cid:19) Additionally, I asbsume thabt the tax base isbin steadby state.8 bStructuralbrevdenues in percent deviation fromsteadystatearethendefinedasfollows τnWhL τkK(Rk −δ) τcC ΨS = τ S,n + τ S,k + τ S,c . (29) t Y t Yγ t Y t b b b b 2.5 Thedebtbrake The debt brake incorporates an adjustment account, which is definedas the sum of all past and contemporaneousdeviationsofthestructuraldeficitfromzerominustheadjustmentaccountclearing. 0 ≤ AC t = (1−ρ)AC t−1 +D t S ≤ 1.5 (30) The lower bound is introduced todcorrect only exdcessive strbuctural deficits not surpluses, whereas the upper bound caps the maximum amount of fiscal consolidation every period. The parameter ρ determines the fraction of the adjustment account, which is cleared every period and calibrated to allow for a maximum clearing of 0.35% of GDP annually, which is stated explicitly in the relevant law. ThefunctionΦisthereforegivenby Φ(∆Y t ,AC t−1 ;ρ) = ρAC t−1 (Y t −Y t−4 ) ≥ 0. (31) 8AsIwilldiscussinsection3,thesteadystateenforceslong-runbudgetbalance. b d d b b 9

I multiply the clearing amount by the Q4-Q4 GDP growth rate. Together with the non-negativity constraint, this ensures that there is only fiscal consolidation taking place when the economy is not inarecession. Excludingthelastterm(Y t −Y t−4 )intheequationallowsforastricterversionofthe debtbrakeandtriggersafiscalconsolidationindependentofthestateoftheeconomy. b b 2.6 Monetaryauthorities I assume that monetary policy is characterized by an interest rate feedback rule as in Smets and Wouters[2007]. Specifically,Iassumefortheinterestrate 1−ρ ρ ψ ψ R ψ R t R t−1 R π t 1 Y t 2 Y t /Y t−1 3 = r (32) ∗ ∗ ∗ t R∗ (cid:18) R∗ (cid:19) " (cid:18) π∗ (cid:19) (cid:18) Y t (cid:19) # (cid:18) Y t /Y t−1(cid:19) ∗ where R∗ is the steady state gross nominal interest rate and Y t is the natural output level defined as the output level in absence of nominal frictions. The parameter ρ ∈ [0,1] captures interest rate R smoothing. Finally, r represents a shock to the short-term interest rate not-accounted for by the t systematicfeedbackrule. Itthusrepresentsamonetarypolicyshock lnr t = ρ r lnr t−1 +ǫ r,t . (33) Thefollowingresourceconstraintclosesthemodel Y t = C t +I t +G t +Nex t +a(u t )K t−1 (34) whereNex isanautocorrelatedstochasticshockreflectingtheeffectsofchangesinforeigndemand t onnetexports,whichisgivenby Nex t = ρ nex Nex t−1 +ǫ nex,t . (35) Iassumethatdebt,governmentrevenues,transfersandspendinggrowinsteadystateatthesamerate asoutput. ThereforeIdetrendthevariablesincludingthefiscalpolicyvariableswithafixedcommon growthtrendγt. Thentheequationsarelinearizedaroundthedeterministicsteadystate. 3 Estimation I use quarterly German data for the period 1983Q1-2009Q4 to estimate the model with Bayesian estimation techniques and match the following 12 variables: real consumption, real investment, real GDP,realwages,hoursworked,theGDP-Deflator,theshort-runinterestrate,governmentdebt,government labour income tax revenues and social security contributions received by the government, consumptiontaxrevenues,capitaltaxrevenues,andsocialsecuritybenefitspaidbythegovernment.9 9AdetaileddescriptionofthedataandtheexactmeasurementequationsarestatedinAppendixA. 10

Thedebtbrakeisoperativesince2011andwasnotaffectingthedecisionsoftheGermangovernment until 2010. Therefore I will assume during estimation that the adjustment account is zero and also Φ(∆Y t ,AC t−1 ;ρ) = 0.10 The estimated model can thus be interpreted as a description of the behavior of the government, private agents and firms before the introduction of the debt brake and b d servesasagoodbenchmarktocomparemacroeconomicoutcomesbeforeandafterthedebtbrake. I estimate the mode of the posterior distribution by maximizing the log posterior kernel, which combines the prior information on the parameters with the likelihood of the data. Then the Metropolis-Hastings algorithm is used to evaluate the whole posterior distribution and the marginal likelihoodofthemodel.11 Afewparametersarekeptfixedthroughouttheestimation. Thedepreciationrate,δ,issetto2.5% andthesteadystatewagemarkup,(1+λ ),to1.5. ThecurvatureparameteroftheKimballaggregator w ′′′ inthegoodsmarketη, whichisgivenbyη = 1/λ p (cid:20) 2 1 + + G G G G ′ ′ ′ ′ ′ −1 (cid:21) , isnotidentified. ThereforeIfixit to10analogoustoSmetsandWouters[2007]. Isetthequarterlytrendgrowthrateγ to0.3,whichis equivalenttotheaveragegrowthrateofrealGDPinthedata. Table1: Calibratedparameters Parameter Value δ Depreciationrate 0.025 (1+λ ) Wagemarkup 1.5 w η CurvatureparameterPhillipsCurve 10 γ Trendgrowthrate 0.32 b∗ SSdebttoGDPratio 0.52 Ψn ∗ Incometaxrevenuesratio 0.29 Ψc ∗ Consumptiontaxrevenuesratio 0.11 Ψk ∗ Capitaltaxrevenuesratio 0.004 t∗ TransferstoGDPratio 0.18 τn SSlabourincometaxrate 0.46 τc SSconsumptiontaxrate 0.15 τk SScapitaltaxrate 0.10 G∗ GovernmentspendingtoGDPratio 0.21 The steady-state government debt to GDP ratio, b∗, is calibrated to 0.52, which corresponds to the mean of this variable using the whole estimation sample from 1983 up to now. I calibrate the steady state tax revenues to GDP ratios in the same way. The income, consumption and capital tax rates in steady state are calculated to match the tax revenues to GDP ratios stated above. Finally, the government spending to GDP ratio in steady state is pinned down to ensure that the government 10TheparameterρandtheupperboundontheadjustmentaccountcanbeparametrizedfromtheGermanlawanddoesn’t needtobeestimated. 11All estimations are executed with Dynare (http://www.cpremap.cnrs.fr/dynare). A sample of 250,000 Metropolis- Hastingsdrawswascreated(neglectingthefirst50,000draws). 11

budgetconstraintisfulfilled. SeeTable1foranoverviewofthecalibratedparameters. Table2: Priorandposteriordistributionsofthestructuralparameters Parameter Priordistribution Posteriordistribution type mean std.dev. mode std.dev. mean 5% 95% ξ Calvowages beta 0.50 0.10 0.57 0.07 0.56 0.44 0.67 w ξ Calvoprices beta 0.50 0.10 0.92 0.02 0.91 0.88 0.94 p SSpricemarkup norm 1.25 0.13 1.91 0.09 1.89 1.78 2.03 ′′ S Investmentadj. cost norm 4.00 1.50 1.73 0.72 2.96 0.98 4.91 ω nonRicardiancons. beta 0.30 0.10 0.28 0.08 0.27 0.15 0.40 σ Laboursupplyelas. norm 2.00 0.75 0.69 0.16 0.71 0.44 0.99 l χ Capitalutilization beta 0.50 0.15 0.27 0.05 0.29 0.20 0.37 α Capitalshare norm 0.30 0.05 0.14 0.02 0.13 0.09 0.17 ρ Int. ratesmoothing beta 0.75 0.10 0.90 0.02 0.90 0.86 0.94 R ψ Inflationresponse norm 1.75 0.20 1.61 0.21 1.59 1.24 1.92 1 ψ Outputgapresponse norm 0.13 0.05 0.03 0.02 0.05 0.01 0.09 2 ψ Diff. outputgapresp norm 0.13 0.05 0.26 0.03 0.26 0.21 0.31 3 g φ Governmentspending invg 0.05 2.00 0.02 0.00 0.02 0.01 0.02 b φn Labourtax invg 0.05 2.00 0.01 0.00 0.01 0.01 0.02 b φc Consumptiontax invg 0.05 2.00 0.01 0.00 0.02 0.01 0.02 b φk Capitaltax invg 0.05 2.00 0.03 0.01 0.03 0.01 0.05 b φt Transfers invg 0.05 2.00 0.01 0.00 0.01 0.01 0.01 b g φ Governmentspending invg 0.10 2.00 0.03 0.01 0.04 0.02 0.05 y φn Labourtax invg 0.10 2.00 0.07 0.02 0.09 0.05 0.12 y φk Capitaltax invg 0.10 2.00 0.05 0.02 0.08 0.03 0.14 y φt Transfers invg 0.10 2.00 0.02 0.00 0.02 0.02 0.03 y β Discountrate gamm 0.25 0.10 0.58 0.14 0.59 0.37 0.83 π SSinflationrate gamm 0.63 0.10 0.53 0.08 0.54 0.41 0.66 Notes: Prior and posterior distributions for the estimated structural parameters. Parameter estimates obtained from BayesianestimationoftheDSGEmodelusingGermandatafrom1983:1-2009:4. 3.1 Priordistributionsoftheestimatedparameters Iestimate37parametersand12standarddeviationsofshockinnovations. Theparameterχgoverning the capital utilisation costs is given by χ = 1 , where a’ and a” are the first and second oder 1+ a a ′ ′ ′ derivativesofthecapitalutilisationcostfunctionw.r.t. capitalutilisationevaluatedatthesteadystate. The estimated parameter β, which drives the discount factor is defined as follows β = 100(1 −1). β Tables 2 and 3 show the assumptions for the prior distributions, which are similar to Smets and Wouters [2007] for the standard model parameters. For the share of rule-of-thumb consumers ω, I assume a beta distribution with mean 0.3 and standard deviation 0.1, with the mean being close to thefindingofCoenenandStraub[2005],whoestimatedtheshareofrule-of-thumbconsumersinthe 12

euroarea. Iassumeforthedebtelasticitiesofthetaxratesaninversegammadistributionwithmean 0.05andstandarddeviation2andfortheoutputelasticitiesaninversegammadistributionwithmean 0.1andstandarddeviation2,implyingsluggishadjustmentbythegovernment. Table3: Priorandposteriordistributionsoftheshockprocesses Parameter Priordistribution Posteriordistribution type mean std.dev. mode std.dev. mean 5% 95% ρ Technologyshock beta 0.50 0.20 0.98 0.00 0.98 0.97 0.99 z ρ Pricemark-upshock beta 0.50 0.20 0.90 0.04 0.89 0.82 0.95 p ρ Investmentshock beta 0.50 0.20 0.23 0.10 0.25 0.09 0.40 µ ρ Riskpremiumshock beta 0.50 0.20 0.93 0.03 0.94 0.89 0.99 d ρ MPshock beta 0.50 0.20 0.24 0.07 0.25 0.14 0.36 r ρ Wagemark-upshock beta 0.50 0.20 0.96 0.03 0.94 0.89 0.99 w ρ Gov. spend. shock beta 0.50 0.20 0.70 0.05 0.69 0.61 0.77 υ ρ Netexportsshock beta 0.50 0.20 0.96 0.02 0.95 0.92 0.98 nex ρ τn Labourtax beta 0.50 0.20 0.92 0.02 0.91 0.87 0.94 ρ τc Consumptiontax beta 0.50 0.20 0.89 0.04 0.88 0.83 0.95 ρ Capitaltax beta 0.50 0.20 0.95 0.02 0.94 0.90 0.97 τk ρ Transfers beta 0.50 0.20 0.94 0.03 0.93 0.89 0.97 T θ MApricemark-up beta 0.50 0.20 0.93 0.02 0.93 0.89 0.97 p θ MAwagemark-up beta 0.50 0.20 0.67 0.10 0.61 0.43 0.79 w ǫ Technologyshock invg 0.10 2.00 0.50 0.04 0.51 0.45 0.58 z ǫ Pricemark-upshock invg 0.10 2.00 0.49 0.04 0.51 0.44 0.57 p ǫ Investmentshock invg 0.10 2.00 1.48 0.16 1.45 1.19 1.71 µ ǫ Riskpremiumshock invg 0.10 2.00 0.14 0.03 0.15 0.10 0.19 d ǫ MPshock invg 0.10 2.00 0.14 0.01 0.15 0.12 0.17 r ǫ Wagemark-upshock invg 0.10 2.00 0.55 0.09 0.58 0.43 0.73 w ǫ Gov. spend. shock invg 0.10 2.00 0.87 0.06 0.89 0.79 0.99 υ ǫ Netexportsshock invg 0.10 2.00 0.82 0.06 0.83 0.73 0.92 nex ǫ τn Labourtax invg 0.10 2.00 0.58 0.04 0.61 0.53 0.69 ǫ τc Consumptiontax invg 0.10 2.00 1.14 0.08 1.17 1.03 1.30 ǫ Capitaltax invg 0.10 2.00 2.26 0.16 2.31 2.04 2.58 τk ǫ Transfers invg 0.10 2.00 0.26 0.02 0.28 0.24 0.31 T Notes: Priorandposteriordistributionsfortheestimatedshocksprocesses. EstimatesobtainedfromBayesianestimationoftheDSGEmodelusingGermandatafrom1983:1-2009:4. 3.2 Posteriordistributionsoftheestimatedparameters The results of the posterior maximization can be find in Tables 2 and 3, which display the posterior mode and mean together with the 90% confidence bands for all 49 estimated parameters. Figures 4 and 5 in the Appendix show graphically the prior and posterior distributions of the structural model parameters and the parameters of the shock processes. The AR(1) coefficients of the fiscal instru- 13

ments are quite high with values around 0.9, which capture the persistence in the decision making of government policy. The standard structural parameters of the model also look reasonable. The Calvo parameter in the goods market has a posterior mean of 0.91 despite strategic complementaritiesthroughaKimballtypeaggregatorforfinalgoods. ButSmetsandWouters[2003]findwith0.91 avalueofsimilarmagnitudeintheirestimatedmodelfortheeuroareaandinflationpersistencewas even higher in Germany compared to other euro area countries. The estimate for the coefficient on inflation in the monetary policy rule is 1.59. Thus monetary policy reacts moderately to movements in inflation. I estimate with a 27% share of non Ricardian consumers, in line with estimates of CoenenandStraub[2005]fortheeuroareaandCoganetal.[2010]fortheUS.Thedebtelasticitiesof tax rates, spending, and transfers are estimated to be below 0.03, i.e. tax rates are adjusted slowly to the debt-to-GDP ratio in Germany. An increase in the debt-to-GDP ratio of 10% of GDP, for example, would cause a reduction in government spending of 0.45% of GDP after one year.12 I find higherestimatesfortheoutputelasticitiesoftaxratesandspending,especiallyforthelabourincome tax rate with a posterior mean of 0.09 and for the capital tax rate with a value of 0.08. This points to significant progressivity in both tax rates. The dynamics of fiscal policy in the model are quite complex, withalmostallfiscalinstrumentsreactingtodebtandthebusinesscycle. AsUhlig[2010] pointsout,theeffectofdiscretionaryfiscalpolicydependsverystronglyonthereactionoftheother fiscalinstrumentstoahigherlevelofgovernmentspending. Iwillshowinthenextsectionhowthis fiscalsetupinteractswiththedebtbrake. 4 Forecasting Performance Ifirstevaluatethemodel’sout-of-sampleforecastingability. Inparticular,Iusetheestimatedmodel describedabovetoconstructanout-of-sampledensityforecastforthedebt-to-GDPratiofromthefirst quarter of 2010 until 2016 and compare it to the actual debt-to-GDP ratio in 2010 and 2011 and the projectionoftheGermangovernmentprovidedinthe"GermanstabilityProgramme2012"fortheEU commission. The adjustment of fiscal instruments in practise depends on political negotiations and resultsoftheelections,whichleadstovariabilityintheparametersofthefiscalpolicyrulesovertime. I take this uncertainty regarding the parameters of the model into account and pick randomly 2,500 Metropolis-Hastings draws of the posterior parameter distributions (Figure 4 in the Appendix). For eachparametercombination,aforecastforthedebt-to-GDPratioiscomputedusingthelinearmodel solution. Additionally policy makers don’t have perfect foresight and don’t know certainly how the economywillevolveinthefuture. TocopewiththisuncertaintyIcreateeveryperiodrandomdraws 12Takingthesmoothingofgovernmentspendingintoaccountandassumingallelseequal,onecancomputetheadjustment ofgovernmentspendingtoa10%increaseinthedebt-to-GDPratioasfollows0.45=0.182+0.182∗0.6911+0.182∗ 0.69112+0.182∗0.69113. 14

95 90 85 80 75 70 65 60 Baseline Baseline + Fiscal package Government actual debt−to−GDP ratio 55 Government projection Government projection w/o european stability measures 50 2006 2008 2010 2012 2014 2016 Figure1: Forecastdebt-to-GDPratioNotes: 2500subdrawsoftheMetropolisHastingsalgorithmanddrawsfrom theshockdistributionsareusedtogeneratedensityforecast.Historicdebt-to-GDPseriesfrom2005until2010isdisplayed withasolidblackline,themeanforecastbyadashed-dottedblackline,the10thuptothe90thpercentilesaredisplayed bygreyshadedareasandthemeanforecastwithadditionaldiscretionaryfiscalpackagein2010isshownbythedashed blue line. The actual debt-to-GDP ratio in 2010 and 2011 is plotted in a red solid line. The projections of the German governmentoutofthe"GermanstabilityProgramme2012"providedfortheEUcommissionwithandwithouteuropean stabilitymeasuresareshownbyareddashedlineandareddashed-dottedline. from the distributions of the 12 shock innovations in the model and add them to the model solution whilecomputingtheforecasts. Figure 1 shows the historical debt-to-GDP series up to 2010 (black solid line), followed by the meanmodelforecast(blackdashed-dottedline)andthecorresponding10thuptothe90thpercentiles (greyshadedareas). Themodelpredictsanincreaseinthedebt-to-GDPratiofrom73.4%inthelast quarter of 2009 until 79% in 2010 and afterwards a gradual decline due to the expected economic recoveryleadingtohigherexpectedlabourincomeandahigheraveragelabourincometaxrateboth stimulatingtaxrevenues. SomepartsofthefiscalstimuluspackagesenactedbytheGermangovernmentbecameoperative in2010. Thereforethesemeasures,whichincreasethedebt-to-GDPratioin2010,arenotincludedin thebaselineforecasts. InAppendixBIlistindetailthefiscalmeasuresandthesizeofbothprograms and group them to the fiscal instruments in the model. Table 4 contains an overview. Measures to lowertheincometaxwedge,whichsumupto29.6billioneuroor1.24%of2009GDP,buildthemajor part of the fiscal package in 2010, which consists of the reduction in health insurance contributions, taxdeductibilityofprofessionelcommuteordirectincometaxcuts. Additionalgovernmentspending hasasizeof10.1billioneuromainlyduetotheinfrastructureinvestmentprogram. Themajorpartof 15

Table4: Germany‘sfiscalstimulusmeasuresin2010 Stimuluspackages GrowthAccelerationAct Totalfiscalpackages blnEuro %ofGDP blnEuro %ofGDP blnEuro %ofGDP Incometax 29.6 1.24 0 0 29.6 1.24 Consumptiontax 0.4 0.02 1 0.04 1.4 0.06 Capitaltax 4.7 0.20 2.4 0.10 7.1 0.30 Spending 10.1 0.42 0 0 10.1 0.42 Transfers 4.5 0.19 5 0.21 9.5 0.40 Allinstruments 49.3 2.06 8.4 0.35 57.7 2.41 Source: GermanTreasury: BrotundButterBrief"DerWirtschaftskriseentgegensteuern",and"Wachstumsbeschleunigungsgesetz". the "Growth Acceleration Act" are higher tax exemptions for dependent children and child benefits with a value of 4.6 billion euro, which I groupedas transfers. Changesin the legislation of business tax, which are estimated to lower capital tax revenues by 2.4 billion euro in 2010 are the second biggestpart. Overallbothprogramsareexpectedtolowerthetaxbaseby57.7billioneuroor2.41% of 2009 GDP in 2010. To take the effect of these fiscal changeson the forecast for the debt-to-GDP ratio into account I compute conditional forecasts. I add in every quarter in 2010 a discretionary government spending shock in the size of 0.105% of GDP and a transfer shock in the size of 0.10% of GDP to the model solution, together with the random shock realisations, while computing the forecasts.13 Additionally,Iusethebaselineforecastsforthedifferenttaxrevenuesin2010andlower therespectivetaxratestodecreasethelabourincometax,consumptiontaxandcapitaltaxrevenuesin theamountof0.31%,0.015%and0.075%ofGDPeveryquarterin2010. Theconditionalforecastis shownwithabluedashedlineinFigure1. Onecanseethatafteraddingthefiscalpackagein2010, themodelforecastpredictsthepeakinthedebt-to-GDPratioof83%intheendof2010right, when comparing the conditional forecast to the actual data of the German government (red solid line).14 The medium-term projection of the German government also lies within a reasonable confidence bandespeciallywheneuropeanstabilitymeasuresareexcluded(reddashed-dottedline). Overallthe modelseemstopredicttheevolutionofthedebt-to-GDPratioinGermanywell. 5 Implications of the Debt Brake 5.1 Implicationsfortheresponsestoapositivedemandshock Toshedmorelightonthemodeldynamics,IcomputetheImpulseResponsestoaonestandarddeviationexogenousincreaseindemandof0.83%ofGDP.Figure2showsthereactionofkeymacroeco- 13Thesizeofthequarterlydiscretionarygovernmentspendingandtransfershockis1/4ofthevaluestatedinTable4. 14Theannualdebt-to-GDPdataseriesoftheGermangovernmentisinterpolatedtoquarterlyfrequencybyacubic-spline function. 16

GDP Incometaxrate Consumptiontaxrate 0 0.6 −0.05 0.1 −0.1 0.4 −0.15 0.2 0.05 −0.2 −0.25 0 0 5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 25 30 Consumption Incometaxrevenues Consumptiontaxrevenues 0 0.15 0 −0.05 −0.01 0.1 −0.1 −0.02 −0.15 0.05 −0.03 −0.2 0 5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 25 30 Governmentspending Transfers Budgetdeficit 0.1 0.05 0.04 0 0.05 0.02 −0.05 0 −0.1 0 −0.02 −0.15 5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 25 30 Structuraldeficit Adjustmentaccount Debt 0.4 1 0 −0.2 0.3 0.5 −0.4 0.2 0 −0.6 0.1 −0.5 −0.8 −1 0 −1 5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 25 30 Figure 2: Positive demand shock (0.83% of GDP) Notes: Impulseresponsesofkeymodelvariablestoaone standarddeviationdemandshock.Variablesaredefinedinpercentdeviationfromsteadystate. nomicvariablestothisexpansionaryshockinthemodelwithoutthedebtbrake. Theincreaseindemandmakeslabourincomeincreaseandgenerateshigherlabourincometaxrevenues(middlepanel second row). Due to a progressive labour income tax rate, an increase in labour income also raises theaveragelabourincometaxrateintheeconomy(middlepanelfirstrow),whichfurtherfostersthe improvement of the government budget. On the other hand, the expansion in GDP and the subsequent upward pressure on prices leads monetary policy to raise the interest rate, creating incentives for Ricardian households to save more and consume less. The crowding-out of Ricardian household’sconsumptiondominatestheconsumptionincreaseofnon-Ricardianconsumers,whoreacttoa higher labour income by spending more. Therefore consumption tax revenues are slightly subdued. As might be expected given the estimates of φG and φT, government spending and transfers react y y countercyclicallyinthefirst5-10quarters. Bothhigheroveralltaxrevenuesandlowerspendinglead toa budgetsurplusandlower governmentdebt. Thereductionin governmentindebtednesscausesa second-roundeffectinthefiscalinstruments. Thegreaterfiscalspacecreatesincentivesforfiscalauthoritiestoincreasespendingandlowertaxrates. Thereforetheaveragelabourincometaxrate,and theconsumptiontaxratefallandspendingandtransfersincreaseafter5quarters. Thesesecond-round effectsleadtoanincreaseinthecyclicaladjustedorstructuraldeficit(bottomleftpanel). Themodel 17

observationsareinlinewithempiricalresultsbyGaliandPerotti[2003],whoestimatetheoutputgap elasticitiesofthecyclicalcomponentofthegovernmentbudgetdeficitandthecyclicaladjustedcomponent for a sample of OECD countries from 1980-2002. They find a procyclical structural deficit andacountercyclicalfiscalstanceonautomaticstabilizersinGermany. GDP Incometaxrate Consumptiontaxrate 0.6 0.25 0.1 0.2 0 0.4 0.15 −0.1 0.1 0.2 0.05 −0.2 0 −0.3 0 5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 25 30 Consumption Incometaxrevenues Consumptiontaxrevenues 0 0 0.15 −0.05 −0.01 −0.1 0.1 −0.02 −0.15 0.05 −0.03 −0.2 −0.04 0 5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 25 30 Governmentspending Transfers Budgetdeficit 0.1 0.08 0.05 0.06 0 0.05 0.04 −0.05 0.02 −0.1 0 0 −0.02 −0.15 5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 25 30 Structuraldeficit Adjustmentaccount Debt 0.4 3 0 0.3 2 −0.5 0.2 0.1 1 −1 0 0 5 10 15 20 25 30 5 10 15 20 25 30 5 10 15 20 25 30 Figure 3: Positive demand shock (0.83% of GDP) Notes: Impulse response of key model variables to a one standarddeviationdemandshock. Variablesaredefinedinpercentdeviationfromsteadystate. Theblacksolidlineshows theresultswithoutdebtbrakemechanism,thebluelinesthecasewithaspending-basedconsolidation(50%spendingcuts and50%transfercuts)andtheredlinesthecasewithatax-basedconsolidation(71%labourincometax,28%consumption taxand1%capitaltax).Thedashedlinescorrespondtoamaximumadjustmentaccountclearingof0.35%ofGDPannually, thedottedlinestoamaximumclearingof0.70%ofGDP. InFigure3,Icomparethebaselineresultswithtwoconsolidationscenarios. Theempiricalliteraturetypicallytreatsfiscalconsolidationthroughcutsinspendingorthroughtaxincreasesseparately, as in Alesina and Ardagna [2010]. I take the same approach and first look at a spending-based consolidationwith50%oftheconsolidationtakingplacethroughcutsintransfersand50%throughcuts ingovernmentpurchases(bluelines)andsecond,atax-basedscenariowith71%oftheconsolidation takingplacethroughanincreaseinthelabourincometax,28%throughanincreaseintheconsumptiontaxand1%throughanincreaseinthecapitaltax(redlines). Taxesareraisedaccordingtotheir correspondingsteady-staterevenue-to-GDPshares. Thedashedlinescorrespondtothebaselinespecificationofthedebtbrake,wherefiscalconsolidationiscappedatamaximumof0.35%ofGDP.The dotted lines show a more ambitious case, where the adjustment account is capped at 3% of GDP, 18

leading to a maximum fiscal consolidation of 0.70% of GDP annually; the speed of adjustment ρ is keptunchanged. Thenonlinearmodelequationsassociatedwiththeimplementationofthedebtbrake requirenonlinearsolutiontechniques;IusethemethodologydescribedinJuillard[1996]. Under either approach to consolidation following a demand shock, the structural deficit leads to an accumulation on the adjustment account until it reaches the maximum of 1.5% (3%) of GDP at around10(18)quarters. Withspending-basedconsolidation,governmentspendingandtransfersare below baseline starting around the fifth quarter. In scenario 2, the income tax rate and consumption taxrateareincreasedcomparedtobaseline. Inbothscenarios,thisimprovesthebudgetdeficitorleads toahigherbudgetsurplus(lastpanelthirdrow)anddebtfallssubstantiallymorethaninthebaseline case. With the spending-based consolidation and the baseline debt brake specification, debt peaks at -1.35 percent, 0.25% of GDP lower than in the case without debt brake. The greater fall in debt creates second-round effects in the fiscal instruments. In scenario 1, fiscal policy lowers taxes after around10quarters,whereasinthetax-basedconsolidationcasespendingandtransfersareincreased. These fiscal dynamics have important implications for the private sector. As can be seen in the second panel on the left side, the drop in private consumption after the exogenous demand shock is lowerinthecaseofaspending-basedconsolidationandlargerinscenario2. Withlogutilityandno consumptionhabits,thelinearizedeulerequationoftheRicardianhouseholdscanbesimplywritten asastreamoffuturerealinterestratesandchangesintheconsumptiontaxrate.15 ∞ τc C = −E (R −π )+ E (τc −τc ) (36) t t t+s t+1+s 1+τc t t+1+s t+s s=0 X Higher contempobraneous and exbpectedbfuture real induces hobuseholdsbto consume more today. Herethefirsteffectdominatesthelast. Inscenario1(2)thelower(higher)expectedrealinterestrates compared to baseline overcompensates the expected drop (increase) in the consumption tax rate. A smallerdropinconsumptioninscenario1associatedwithalowermarginalutilityofconsumptionfor Ricardianhouseholdscomparedtobaselineandscenario2raisestheirdesiredrealwage. Thisleads to the higher-than-baseline income tax revenues (middle panel second row) in the case of spendingbased consolidation in the first 10 quarters. Afterwards, the movements in the tax rates determine the evolution of wages. Both the increase in the income tax rate and the consumption tax rate in scenario2raisesthewedgebetweenthemarginalrateofsubstitutionofleisureforconsumptionand the real wage in the labor supply decision of the Ricardian households and induces households to demand higher wages. Therefore in the medium-run wages are higher in scenario 2, which pushes 15Departingfromtheassumptionoflogutilityandassumingahighercoefficientofrelativeriskaversioninanonseparable utilityfunctionalaKingetal.[1988],doesn’tchangethequalitativeresults. Consumptionfallslessinallscenariosdue toagent’slowerintertemporalelasticityofsubstitutionanddebtfallsmoreduetogreaterincometaxrevenuesgenerated through higher wages in the economy. Assuming external habits in consumption doesn’t change the qualitative results either,althoughthereactionofRicardianhousehold’sconsumptionismorehump-shapedandpersistent. 19

marginalcostsandinflationupandleadsmonetarypolicytoraiseinterestrates. Thisinturnraisesthe financingcostsofgovernmentdebtandleadstoamoresubdueddropingovernmentdebtcomparedto the spending consolidation case. Overall, the debt brake with spending-based consolidation reduces government debt most given the same amount of adjustment account clearing relative to the case of tax-basedconsolidationandleadstoslightyhigherGDPduetolessofadropinprivateconsumption, whichispartlyoffsetbylowerpublicdemand. Allowingformorefiscalconsolidation(dottedlines) reinforces the effects described above. Angeloni et al. [2011] come to the same conclusions, when lookingatdifferentexitstrategiesoutofdebtintheaftermathofarecession. Inparticular, theyfind thatpre-announcedspendingbasedfiscalconsolidationismostpromising. 5.2 Implicationsfortheresponsestoanegativedemandshock Howdotheresultschangeifacontractionaryshockoccurs? Inthecaseofanegativedemandshock, governmentrevenuesdropduetoalowertaxbase, andspendingandtransfersincrease, leadingtoa jump up in the budget deficit and debt, but to a structural budget surplus. The results are the mirror imageofFigure2. Thedebtbrakeissetupinawaythatitworksasymmetricallyi.e. onlystructural deficitsarebalancedthroughtaxincreasesorspendingcutsbutnostructuralsurplusiscleared. Asa consequence,theadjustmentaccountstaysatzeroandfiscalinstrumentsarenotmodified. Thus,the additionalfiscalconsolidationassociatedwiththedebtbraketakesplaceonlyineconomicexpansions without constraining fiscal policy in recessions and even creates room for additional fiscal measures inrecessions. 5.3 Implicationsforgovernmentspendingmultipliers Does the debt brake change the effectiveness of discretionary government spending? I compute the impulseresponsestoa1%ofGDPgovernmentspendingshocksettingtheoutputanddebtelasticities of government spending to zero (φ g = φ g = 0) to render the process for government spending exy b ogenousandtomaketheresultscomparabletotheliterature,whichusuallystudiesanautocorrelated 1% increase in government spending. Table 5 shows the initial multipliers and the present value multipliers after 2 years for the baseline case without debt brake, the spending-based and tax-based scenariosequivalenttothesubsectionaboveandamixedscenario,whichcorrespondstothesavings packagetheGermangovernmentannouncedin2010tofulfillthedebtbrake.16 17 The main finding in Table 5 is that the implications of the debt brake for government spending 16ThePresentvaluemultipliersarecalculatedusingthefollowingformular,wheregovernmentspendingandoutputare discountedwiththeSSnominalinterestrate:PV = P8 j=0(R−j)∆Yt+j. P8 j=0 (R−j)∆Gt+j 17The savings package contains mainly transfer cuts and spending cuts, but also additional taxes like a ticket tax for flights.Theweightsforthefiscalinstrumentsassociatedwiththispackageare:Transfers45%,governmentspending25%, incometaxrevenues21.3%,consumptiontaxrevenues8.4%andcapitaltaxrevenues0.3%. 20

Table5: Governmentspendingmultipliers Baseline Spending Tax Mixed Initialmultiplier 0.95 0.97 0.96 0.97 RelativetoBaseline(in%) 2.73 1.06 1.86 Presentvaluemultiplier 0.57 0.57 0.55 0.56 RelativetoBaseline(in%) -0.12 -3.98 -2.14 Notes: Initial government spending multipliers and present value multipliers after 2 years. Outputanddebtelasticitiesofgovernmentspendingaresettozero(φg =φg = y b 0). Row2and4showthepercentchangeofthespendingmultiplierswithdebtbrake relativetothebaselinecasewithoutdebtbrake(( PV(with_debt_brake) −1)∗100). PV(without_debt_brake) multipliers are small. Perhaps surprisingly, the initial multiplier increases when the debt brake is in place, no matter which consolidation scheme the government chooses. But the stimulation in GDP is greatest with a spending-based consolidation. This finding supports the results by Corsetti et al. [2012],whoshowthattheinitialgovernmentspendingmultiplierincreasesifagentsexpectaspending reversal in the future i.e. the government announces spending cuts in the future. The mechanism is similar with the debt brake in place. Agents know that the government is credibly committed to keeping the structural deficit stable over time, and that a discretionary increase in government spending raises the structural deficit. Therefore they will expect lower public spending or higher taxesinthenearfutureandlessofanincreaseingovernmentdebt. Thislowersthewealtheffectand leadsagentstoincreasetheirconsumptioncomparedtothecasewithoutthedebtbrake. But the present value multipliers after two years are lower with the debt brake in place than without, independent of the consolidation scheme. If the economy is not in a recessionary episode, thefiscalconsolidationwillkickinimmediatelyaftertheworseninginfiscalconditionsandwilllead either to an increase in distortions in the economy through a tax increase or to lower public demand ortolowerhouseholddisposableincome. 5.4 Stabilization Does the debt brake stabilize or destabilize the economy? To answer this question, I simulate the modelwithandwithoutthedebtbrakefor120,000periods,droppingthefirst20,000observationsas atrainingperiod. Duetothenonlinearitiesassociatedwiththedebtbrake,Icannotsimplyiteratethe policy function of the linear model forward adding a draw from the shock distribution each period; instead I use an extended shooting algorithm described in Braun and Koerber [2011]. Starting from steady state in period 0, agents experience a draw from the shock distribution in period 1 and given theseshocks,solvethesetofnonlinearequationsthatdescribetheirrespectivedecisionrulesforward for 500 periods until the economy is back at steady state. Agents use the outcome in period 1 as initial conditions and experience a new set of shocks in period 2 and again solve the model forward 21

for 500 periods. This is repeated for a total of 120,000 periods. Afterwards I compute the variances of key macroeconomic variables and derive the following measure of stabilization S to compare the volatilityofthevariable(x)inthecasewithandwithoutdebtbrake: var(x)(with_debt_brake) S = −1 ∗100. (37) var(x)(without_debt_brake) (cid:18) (cid:19) Two results emerge from Table 6. First, a spending-based consolidation is preferable in terms of stabilization. Although a tax-based consolidation still leads to lower volatility in consumption and partly in output compared to the baseline scenario without the debt brake, a spending-based consolidationclearlyoutperformsboth. Second,astate-dependentconsolidation(Columns3and5), whichonlytriggersanadjustmentaccountclearingintheabsenceofarecession, ispreferredovera state-independentconsolidation(Columns2and4). Table6: Stabilizationproperties Variable Spending Spending Tax Tax (State-dependent) (State-dependent) Output -2.11 -2.81 0.00 -0.35 Hours -1.31 -1.31 0.00 0.00 Consumption -4.20 -4.66 -0.47 -0.94 Investment -2.24 -2.61 0.00 -0.37 Structuraldeficit -1.75 -2.44 0.00 0.00 Debt-to-GDP -0.13 -0.19 0.57 1.09 Revenues -0.98 -2.92 0.98 1.97 Spending -2.41 -3.60 0.00 1.22 Transfers -1.69 -1.69 1.70 3.52 Incometaxrate -5.28 -6.31 2.15 2.15 Notes: Model simulation with and without debt brake for 120,000 periods dropping the first 20,000observations. Thetableshowsthepercentchangeinthevarianceoftheindicatedvariables, when moving from the baseline case without debt brake to the respective debt brake scenarios. Surprisingly,inthecaseofaspending-basedconsolidation,thedebtbrakestabilizestheeconomy without leading to more volatility in the fiscal instruments. One might expect that more adjustment in the fiscal instruments is necessary to achieve a stabilization of the private sector. There are three forces at work. First, the consolidation triggered by the debt brake through the adjustment account clearing (function Φ) creates additional volatility in the instruments. But this additional volatility is eased by the smoothed consolidation via the adjustment account. The overall impact of this first force depends therefore on the maximal clearing and on the smoothing parameter in the adjustment accountρ. Second,thedebtbrakemakesdebtitselfmorevolatilebutnotnecessarilythedebt-to-GDP ratio,andthefiscalinstrumentsareallafunctionofthedebt-to-GDPratio. Inthecaseofaspendingbasedconsolidation,theoutputstabilizationdominatesthedebtdestabilizationandthedebt-to-GDP 22

ratio becomes less volatile; in the case of tax based consolidation, it’s the other way round. Third, the automatic stabilizers react endogenously to the state of the economy. If there is less volatility in output, tax revenues, spending, and transfers are also less volatile. In the case of spending-based consolidation,thelattertwoforcesclearlydominatethefirst. Inthecaseoftax-basedconsolidation, thefirsttwoeffectsdominatethelast. 6 Conclusion Inthispaper,IstudythemacroeconomicimplicationsofGermany’srecentlyintroduced"debtbrake" in a New Keynesian DSGE model à la Christiano et al. [2005] or Smets and Wouters [2007] integrating non-Ricardian consumers as well as Ricardian consumers, adding distortionary taxes like consumption tax, capital tax, and labor income tax, and fiscal rules for transfers and spending. I estimate the model on 12 German data series including the debt-to-GDP ratio, tax revenues, and transfers. Thedebtbrakeconstrainsfiscalpolicymakersinboomsbutnotinrecessions,forcingthem toreducethegovernmentbudgetdeficitthroughincreasesintaxesorexpenditurecuts. Additionally, thedebtbrakeraisesinitialgovernmentspendingmultipliersindependentofthetypeoffiscalconsolidationandstabilizestheprivatesectorwithoutleadingtoagreatervolatilityinthefiscalinstruments. That is especially true when the fiscal consolidation triggered by the debt brake is executed through expenditurecutsandthefiscalconsolidationtakesplaceonlyintimesofpositiveoutputgrowth. Inthemodel,Idonottakeintoaccounthowtheevolutionofdebtaffectstheinterestrate. Laubach [2010] estimates a significant positive impact of expected budget deficits and the expected debtto-GDP ratio on long-term real interest rates in the US. Taking this channel into account, a faster consolidationofthegovernmentbudgetthroughthedebtbrakewouldlowerfutureinterestpayments onoutstandinggovernmentdebtfurtherandwouldhaveevenmorefavourableoutcomesintermsof stabilizationandgovernmentspendingmultipiers. 23

A Appendix estimation Data The data I employ are from the OECD Economic Outlook Database. For the period 1983:1-1991:1 I use data for West Germany, which are scaled so that the observation in the first quarter of 1991 matches the observation for reunified Germany in this quarter. Precisely I use the following OECD time series: Gross domestic product, volume, market prices (DEUGDPV, WGRGDPV), Gross domesticproduct,deflator,marketprices(DEUPGDP,WGRPGDP),Privatefinalconsumptionexpenditure, value (DEUCPAA, WGRCPAA), Private non-residential gross fixed capital formation, volume (DEUIBV,WGRIBV),Privatenon-residentialfixedcapitalformation,deflator(DEUPIB,WGRPIB), Total employment (DEUET, WGRET), Short-term interest rate (DEUIRS, WGRIRS), Trend labour force (DEULFS, WGRLFS), Hours worked per employee, total economy (DEUHRS, WGRHRS), Compensation rate, total economy (DEUWSST, WGRWSST), Capital tax and transfers receipts, value (DEUTKTRG, WGRTKTRG), Indirect taxes, value (DEUTIND, WGRTIND), Total direct taxes,value(DEUTY,WGRTY),Socialsecuritycontributionreceivedbygeneralgovernment,value (DEUSSRG, WGRSSRG), Social security benefits paid by general government, value (DEUSSPG, WGRSSPG),Generalgovernmentgrossfinancialliabilities,asapercentageofGDP(WGRGGFLQ) for the period 1983:1-1994:4 and Gross public debt according to the Maastricht criterion as a percentageofGDP(DEUGGFLMQ)fortheperiod1995:1-2009:4. DataonGrossdebtaccordingtothe MaastrichtcriterionarenotavailableforGermanybefore1995. Theconceptsdifferintworespects. GrossdebtaccordingtotheMaastrichtcriteriondoesnotincludetradecreditsandadvancesandgovernmentbondsarevaluedattheirmarketvaluenotnominalvalue. Timesseriesinmonthlyfrequency areconvertedtoquarterlyfrequencybyusingtheaverageofthemonthlyrates. Annualfiscaldataare convertedtoquarterlydatabyassumingthatthevariablegrowswiththesamerateduringtheyear. Measurementequations I use quarterly German data for the period 1983Q1-2009Q4 to estimate the model with Bayesian estimation techniques and match the following 12 variables: the log difference of real consumption (dlCons ), real investment (dlInv ), real GDP (dlGDP ), real wages (dlWage ) and hours t t t t worked (dlHours ), the log difference of the GDP-Deflator (dlP ), the quarterly short-run interest t t rate(Interest ),thelogofgovernmentdebt(ldebt ),theQ4-Q4logdifferenceofgovernmentlabour t t income tax revenues and social security benefits received (dlinctax ), consumption tax revenues t (dlconstax ), capital tax revenues (dlcaptax ) and social security benefits paid (dltransfer ). All t t t government variables are expressed relative to GDP. I removed a linear trend in hours worked to 24

make the data series stationary and consistent with the model assumptions. The following measurementequationsareemployedtolinkthemodelvariablestothedata. dlGDP t γ y t −y t−1  dlCons t  γ  c t −c t−1  b b  dlInv t   γ   i t −i t−1       b b    dlWage t     γ     w b t − b w t−1      dlHours t         l t −l t−1        b b     dlP t    =    π   +   b π b t       Interest t     r   r t     ldebt t       b       b b b ∗ t b −y t     b    dlinctax t         Ψn t /Ψn ∗ −y t − Ψn t−4 /Ψn ∗ −y t−4        b     dlconstax t            (cid:16) b Ψc t /Ψc ∗ −y b t (cid:17)− (cid:16)Ψ b c t−4 /Ψc ∗ −y b t−4 (cid:17)       dlcaptax t            (cid:16)Ψ b k t /Ψk ∗ − b y t (cid:17)−(cid:16)Ψ b k t−4 /Ψk ∗ − b y t−4 (cid:17)      dltransfer t       (cid:16) t t /t∗ −y t (cid:17)− (cid:16)t t−4 /t∗ −y t−4 (cid:17)      b b b b    (cid:0) (cid:1) (cid:0) (cid:1) b b b b where dl stands for 100 times the log difference and l for 100 times the log. x t = x t x −x defines thepercentagedeviationofavariablefromtrend. γ = 100(γ−1)denotesthecommonrealquarterly b trend growth rate, π the quarterly steady state inflation rate, r the steady state nominal interest rate, b∗thesteadystatedebt-to-GDPratio,withb = 100ln(b∗),Ψn ∗ thesteadystateratiooflabourincome tax revenues to GDP, Ψc ∗ the steady state ratio of consumption tax revenues to GDP, Ψk ∗ the steady stateratioofcapitaltaxrevenuestoGDPandt∗ theratioofsteadystatesocialsecuritybenefitspaid bythegovernmenttoGDP. 25

ξ ξ 1+λ w p p 6 20 4 4 2 10 2 0 0 0 0.2 0.4 0.6 0.8 0.4 0.6 0.8 1 1.5 2 S‘‘ ω σ l 0.3 4 2 0.2 0.1 2 1 0 0 0 0 5 10 0 0.2 0.4 0.6 0.8 0 1 2 3 4 χ α ρ R 8 20 6 15 4 10 10 2 5 0 0 0 0 0.2 0.4 0.6 0.8 0 0.1 0.2 0.3 0.4 0.4 0.6 0.8 1 ψ ψ ψ 1 2 3 2 15 10 10 1 5 5 0 0 0 1 1.5 2 2.5 0 0.1 0.2 0 0.1 0.2 0.3 0.4 φg φn φc b b b 200 100 100 50 100 50 0 0 0 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25 0 0.05 0.1 0.15 0.2 0.25 φk φt φg b b y 50 500 20 0 0 0 0 0.05 0.1 0.15 0.2 0.25 0.05 0.1 0.15 0.2 0.25 0 0.1 0.2 0.3 0.4 0.5 φn φk φt y y y 100 10 10 50 0 0 0 0 0.1 0.2 0.3 0.4 0.5 0 0.2 0.4 0.1 0.2 0.3 0.4 0.5 πm βm 4 4 2 2 0 0 0.2 0.4 0.6 0.8 1 0 0.5 1 1.5 Figure4: Priorandposteriordistributionsofthestructuralparameters. Notes:Prior(solidgrey)vs.posterior (solidblack)distributionsfortheestimatedstructuralparameters.EstimatesobtainedfromBayesianestimationoftheDSGE modelusingGermandatafrom1983:1-2009:4. 26

ρ z ρ p ρ µ 80 10 4 60 40 5 2 20 0 0 0 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 −0.2 0 0.2 0.4 0.6 0.8 ρ ρ ρ d r w 6 15 10 4 10 5 2 5 0 0 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8 1 ρ υ ρ nex ρ τn 6 8 20 15 4 10 10 2 5 0 0 0 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 ρ τc ρ τk ρ T 10 20 15 10 5 10 5 0 0 0 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 θ θ ε p w z 4 15 15 10 2 10 5 5 0 0 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 ε p ε µ ε d 15 15 15 10 10 10 5 5 5 0 0 0 0.2 0.4 0.6 0.8 0.5 1 1.5 2 2.5 0 0.1 0.2 0.3 0.4 0.5 ε r ε w ε υ 30 15 15 20 10 10 10 5 5 0 0 0 0.1 0.2 0.3 0.4 0.5 0 0.5 1 1.5 0.2 0.4 0.6 0.8 1 1.2 ε nex ε τn ε τc 15 15 15 10 10 10 5 5 5 0 0 0 0.2 0.4 0.6 0.8 1 1.2 0.2 0.4 0.6 0.8 1 0.5 1 1.5 ε τk ε T 15 20 10 10 5 0 0 0.5 1 1.5 2 2.5 3 0.1 0.2 0.3 0.4 0.5 Figure 5: Prior and posterior distributions of the shock processes. Notes: Prior(solidgrey)vs. posterior (solidblack)distributionsfortheestimatedshocksprocesses. EstimatesobtainedfromBayesianestimationoftheDSGE modelusingGermandatafrom1983:1-2009:4. 27

B Details of Germany´s fiscal stimulus measures in 2010 Stimuluspackages Instrument Measure (blnEuro) Taxdeductibilityofprofessionelcommute 4.00 Packagefortaxburdenreduction,stabilisationof socialsecuritycontributionsandinvestmentinfamilies 11.91 Incometax Incometaxcut 6.04 Statepaymentof50%socialinsuranceforshort-timeworkers forshort-timeworkers 1.15 Reductioninhealthinsurancecontributions 6.50 Suspensionofcartaxonnewvehicles 0.13 Consumptiontax Reformofcartax 0.17 DecreaseoftaxonBiodiesel 0.13 Capitaltax Highertax-freeallowancesforcompanies 0.37 Degressivedepreciationdeduction 4.33 Investmentsintotransportinfrastructure 1.00 Spending Infrastructureinvestmentprogramme 8.68 Innovationsupportprogramme 0.45 Retrainingandstrongerjobservice 1.59 Transfers Increasedchildbenefits 2.84 Increasedhousingbenefits 0.06 49.31 Source:BrotundButterBrief"DerWirtschaftskriseentgegensteuern",GermanTreasury GrowthAccelerationAct Instrument Measure (blnEuro) Consumptiontax DecreaseofVATonHotelsandrestaurants 1.00 Capitaltax Changesinthelegislationofbusinesstax 2.40 Transfers Highertaxexemptionfordependentchildrenandchildbenefits 4.60 Lowerinheritancetax 0.40 8.40 Source:Wachstumsbeschleunigungsgesetz,GermanTreasury 28

References A. Alesina and S. Ardagna. Large changes in fiscal policy: Taxes versus spending. Tax Policy and theEconomy,forthcoming,2010. I.Angeloni,E.Faia,andR.Winkler. Exitstrategies. CFSWorkingPaper2010/25,2011. O.Blanchard,G.Dell’Ariccia,andP.Mauro. Rethinkingmacroeconomicpolicy. IMFStaffPosition Note,February12,2010,2010. R.A.BraunandL.M.Koerber. Newkeynesiandynamicsinalowinterestrateenvironment. Federal ReserveBankofAltantaWorkingPaper2011-10,2011. G.Calvo. Staggeredpricesinautilitymaximizingframework. JournalofMonetaryEconomics,12: 383–398,1983. L. J. Christiano, M. Eichenbaum, and C. L. Evans. Nominal rigidities and the dynamic effects of a shocktomonetarypolicy. JournalofPoliticalEconomy,113:1–45,2005. G. Coenen and R. Straub. Does government spending crowd in private consumption? theory and empiricalevidencefortheeuroarea. InternationalFinance,8:435–470,2005. G.Coenen,P.McAdam,andR.Straub. Taxreformandlabour-marketperformanceintheeuroarea: A simulation-based analysis using the new area-wide model. Journal of Economic Dynamics & Control,32:2543–2583,2008. J.F.Cogan,T.Cwik,J.B.Taylor,andV.Wieland. Newkeynesianversusoldkeynesiangovernment spendingmultipliers. JournalofEconomicDynamicsandControl,34:281–295,2010. G.Corsetti,A.Meier,andG.Mueller. Fiscalstimuluswithspendingreversals. ReviewofEconomics andStatistics,2012. IMFWorkingPaper09/106. T.CwikandV.Wieland. Keynesiangovernmentspendingmultipliersandspilloversintheeuroarea. CEPRDiscussionPaper7389,2009. R.Edge,M.T.Kiley,andJ.-P.Laforte. Naturalratemeasuresinanestimateddsgemodeloftheu.s. economy. JournalofEconomicDynamics&Control,32:2512–2535,2008. J. Gali and R. Perotti. Fiscal policy and monetary integration in europe. Economic Policy, pages 533–572,2003. J. Gali, J.D.Lopez-Salido, and J. Valles. Understanding the effects of government spending on consumption. JournaloftheEuropeanEconomicAssociation,5(1):227–270,2007. 29

M. Juillard. Dynare: a program for the resolution and simulation of dynamic models with forward variablesthroughtheuseofarelaxationalgorithm. CEPREMAPWorkingPaper9602,1996. M. T. Kiley. Habit persistence, nonseparability between consumption and leisure, or rule-of-thumb consumers: which accounts for the predictability of consumption growth? The Review, 92(3): 679–683,2010. R. G. King, C. I. Plosser, and S. T. Rebelo. Production, growth and business cycles - i. the basic neoclassicalmodel. JournalofMonetaryEconomics,21:195–232,1988. R. Kollmann. US trade balance dynamics: the role of fiscal policy and productiviy shocks and of financialmarketlinkages. JournalofInternationalMoneyandFinance,17:637–669,1998. T. Laubach. New evidence on the interest rate effects of budget deficits and debt. Journal of the EuropeanEconomicAssociation,forthcoming,2010. E.M.Leeper,M.Plante,andN.Traum. Dynamicsoffiscalfinancingintheunitedstates. Journalof Econometrics,forthcoming,2010. F. Smets and R. Wouters. An estimated stochastic dynamic general equilibrium model of the euro area. JournaloftheEuropeanEconomicAssociation,1:1123–1175,2003. F. Smets and R. Wouters. Shocks and frictions in US business cycles: A bayesian DSGE approach. AmericanEconomicReview,97:586–606,2007. H.Uhlig. Somefiscalcalculus. AmericanEconomicReview,100(2):30–34,2010. draft. 30