How large are fiscal multipliers? An empirical assessment for the Euro Area

How large are fiscal multipliers? An empirical assessment for the Euro Area Ricardo Jorge Maia Martins da Silva Master Dissertation in Economics Supervised by: Vitor Manuel Carvalho Ana Paula Ribeiro 2012 i Vita Ricardo Jorge Maia Martins da Silva was born on the 6th of December of 1987, in Vila Nova de Gaia, Portugal. He obtained his degree in Economics from the Faculdade de Economia da Universidade do Porto – Portugal (FEP) in 2010, with a final grade of 15/20. Also in 2010, and at the same university, he began attending the Master course in Economics - Specialization in Economic Analysis. He has already completed the curricular part of the master course, with a final grade of 16/20. This dissertation is the final contribution in order to obtain the Master degree. ii Acknowledgements I would like to acknowledge all who made this dissertation possible through their advices, help and support. Specially, I would like to express my gratitude to my supervisors Prof. Vitor Manuel Carvalho and Prof. Ana Paula Ribeiro by their clever ideas, encouragement, monitoring and patience. I also thank my friends and colleagues for their comments and suggestions on this dissertation. iii To my parents, iv Abstract In the current context where the limited role for monetary policy instruments apparently endows fiscal policy with higher effectiveness, European fiscal policy authorities are rather constrained by the fact of most countries being struggling against recessions together with the need to put public finances in a sustainable path. In this context, we assess how large are fiscal multipliers in Europe, for both aggregated and disaggregated spending and revenue variables. Moreover, we analyze how cycle phases and fiscal consolidation episodes shape the size of fiscal multipliers. We present evidence for the Euro area, relying on a VAR model with pooled annual data for the current seventeen country-members, from 1998 to 2008. Estimation results show that, on average, public spending in the Euro area impinges negatively on impact, but the cumulative impact on output is positive while taxes, as expected, have (both impact and cumulative) negative impacts on output. On the expenditure side, transfers are the main driving force for the overall expenditure dynamics; moreover, wages exhibit negative impacts on output while positive effects are strongly driven by shocks in public investment and, to a lesser extent, by intermediate consumption. On the revenue side, all items impinge negatively on output growth. Additionally, openness reduces the effectiveness of public spending while it positively affects the size of the taxes multiplier. We have also found that the intertemporal budget constraint shapes a “more Ricardian” behavior on the European economic agents. As regards to how the size of fiscal multipliers changes across cycle phases, our results show that public spending multiplier is positive in recessions while in expansions is smaller, inclusively, negative. Similarly, the effectiveness of the taxes multiplier is, also, higher in recessions. Last but not the least, we have found that consolidation phases affect negatively the size of the multipliers. Keywords: Fiscal policy, Fiscal multipliers, Fiscal shocks, Business-cycle fluctuations, Public debt, Euro area, VAR analysis. JEL Codes: E32, E62, E65; H60. v Resumo No contexto atual, em que o papel dos instrumentos de política monetária se encontra limitado, seria de esperar que a política orçamental tivesse uma maior eficácia. Porém, as autoridades de política estão limitadas na sua atuação pelo facto de muitos países estarem a sofrer recessões e pressão para garantir a sustentabilidade da dívida. Neste contexto, avaliamos a magnitude dos multiplicadores orçamentais na Europa, em termos agregados e desagregados, para as variáveis da despesa e da receita. Analisamos também se as fases do ciclo e os episódios de consolidação orçamental influenciam a dimensão dos multiplicadores orçamentais. Apresentamos evidência para a área do Euro, com base num modelo VAR, com dados anuais (1998-2008) para todos os países-membros. Os resultados das estimações mostram que, para a área Euro, a despesa pública afeta negativamente o produto no impacto mas esse efeito é positivo em termos acumulados. Os impostos, como esperado, exibem (tanto no impacto como em termos acumulados) efeitos negativos sobre o produto. Do lado da despesa, as transferências são as responsáveis pela dinâmica da despesa global; os salários apresentam efeitos negativos no produto, enquanto os efeitos positivos são fortemente impulsionados por choques no investimento público e, em menor medida, pelo consumo público intermédio. Do lado da receita, todos os instrumentos influenciam negativamente o crescimento do produto. Adicionalmente, o grau de abertura comercial reduz a eficácia da despesa pública mas aumenta o impacto da receita sobre o produto. Concluímos também que a restrição orçamental intertemporal do setor público é ativa, tornando o comportamento dos agentes europeus mais ricardiano. Os resultados mostram ainda que o multiplicador da despesa pública é positivo em recessões, sendo menor em expansões, inclusivamente, negativo. Da mesma forma, a eficácia do multiplicador dos impostos é superior em recessões. Finalmente, conclui-se que as fases de consolidação afetam negativamente o tamanho dos multiplicadores. Palavras-chave: Política orçamental, Multiplicadores orçamentais, Choques de política orçamental, Ciclos económicos, Dívida pública, Área Euro, modelos VAR. Códigos JEL: E32, E62, E65; H60. vi Table of Contents Vita .................................................................................................................................... i Acknowledgements ......................................................................................................... ii Abstract ........................................................................................................................... iv Resumo ............................................................................................................................. v Table of Contents ........................................................................................................... vi List of Tables ................................................................................................................. vii List of Figures ................................................................................................................. ix 1. Introduction ............................................................................................................. 1 2. Theoretical mechanisms and empirical evidence on the size of fiscal multipliers – a review ..................................................................................................... 3 2.1. General determinants of the size of multipliers ................................................. 3 2.2. Empirical evidence on the size of multipliers .................................................... 9 2.3. Are multipliers pro or counter cyclical? ........................................................... 16 2.4. Do multipliers change during fiscal consolidation periods? ............................ 22 3. Methodology and data ........................................................................................... 28 3.1. VAR methodology ........................................................................................... 29 3.2. Model specification and data ........................................................................... 33 4. Estimation results .................................................................................................. 38 4.1. Global Analysis ................................................................................................ 39 4.2. Multipliers across cycle phases ........................................................................ 53 4.3. Multipliers during debt-consolidation periods ................................................. 63 5. Conclusions ............................................................................................................ 69 Appendix ........................................................................................................................ 72 Keynesian spending and tax multipliers (closed economy) ........................................ 72 References ...................................................................................................................... 74 Data Sources .................................................................................................................. 81 Annexes .......................................................................................................................... 82 A.1. List of variables used in estimations ................................................................... 82 A.2. Identification of cycle phases – Euro area countries .......................................... 84 A.3. Identification of consolidation periods – Euro area countries ............................ 85 vii List of Tables Table 1: Fiscal multipliers in literature ........................................................................... 10 Table 2: Government spending multiplier across different scenarios ............................ 14 Table 3: Government spending multiplier according to the state of economy ............... 18 Table 4: Fiscal multipliers in zero lower bound vs. positive interest rate ...................... 21 Table 5: Unit-root tests on core variables of the model - aggregate fiscal variables, full sample ............................................................................................................................. 36 Table 6: Optimal number of lags - aggregate fiscal variables, full sample .................... 36 Table 7: VAR stability test - aggregate fiscal variables, full sample ............................. 37 variables, holding similar (not reported) results. ........................................................... 37 Table 8: Estimation results - aggregated fiscal variables, full sample. ........................... 39 Table 9: Cumulative fiscal multipliers, full sample - public spending vs. taxes. ........... 42 Table 10: Estimation results - disaggregated fiscal variables, full sample. .................... 43 Table 11: Cumulative fiscal multipliers, full sample - disaggregated public spending vs. disaggregated taxes. ........................................................................................................ 46 Table 12: Estimation results controlling for openness degree - aggregated fiscal variables. ......................................................................................................................... 49 Table 13: Cumulative fiscal multipliers, controlling for openness degree – public spending vs. taxes. .......................................................................................................... 50 Table 14: Estimation results controlling for debt dynamics - aggregated fiscal variables. ........................................................................................................................................ 52 Table 15: Cumulative fiscal multipliers, controlling for debt dynamics – public spending vs. taxes. .......................................................................................................... 53 Table 16: Estimation results - aggregated fiscal variables, expansions. ......................... 54 Table 17: Estimation results - aggregated fiscal variables, recessions. .......................... 55 Table 18: Cumulative fiscal multipliers, expansions vs. recessions - public spending vs. taxes. ............................................................................................................................... 57 Table 19: Estimation results - disaggregated fiscal variables, expansions. .................... 58 Table 20: Estimation results - disaggregated fiscal variables, recessions. ..................... 59 Table 21: Cumulative fiscal multipliers, expansions vs. recessions - disaggregated public spending. .............................................................................................................. 61 Table 22: Cumulative fiscal multipliers, expansions vs. recessions - disaggregated taxes. ........................................................................................................................................ 62 Table 23: Estimation results controlling for consolidation periodsaggregated fiscal variables. ......................................................................................................................... 64 Table 24: Cumulative fiscal multipliers, controlling for consolidation – public spending vs. taxes. .......................................................................................................................... 65 viii Table 25: Estimation results controlling for consolidation periods – disaggregated fiscal variables. ......................................................................................................................... 66 Table 26: Cumulative fiscal multipliers, controlling for consolidation – disaggregated public spending. .............................................................................................................. 67 Table 27: Cumulative fiscal multipliers, controlling for consolidation – disaggregated taxes. ............................................................................................................................... 67 6 pressure on the exchange and the interest rates. 5 As money supply increases as to defend the fixed exchange rate parity, an increase in output occurs with a smaller crowding-out effect. Hence fiscal policy under this scenario is more effective in stimulating output than for a large, rather closed, country. In contrast, in a small open economy under a flexible exchange rate regime, fiscal policy is less effective. A fiscal expansion puts upward pressure on the interest rate; as capital flows into the economy, the demand for domestic currency increases and the domestic currency appreciates. Because prices are sticky, this nominal appreciation is mapped to a real appreciation, and consequently net exports decline. This negative effect on the trade balance limits the expansionary effect of fiscal policy, and so fiscal policy in small open economies under flexible exchange rate regimes affects output by less than in large, rather closed, economies. Moreover, the effects of the exchange-rate channel are reinforced for higher degrees of capital mobility. Perfect mobility of capital implies that assets denominated in domestic currency and in foreign currency are perfect substitutes - as long as domestic interest rate compensates for foreign interest rate and for expected exchange rate changes, assets denominated in both currencies are hold. In this scenario, currency exchange is unlimited, costless and immediate. So, under a fixed exchange rate regime and perfect capital mobility, capital flows are reinforced in response to interest rate differentials, offsetting them, in the sequence of a fiscal policy. Thus, no crowding-out (in) applies and fiscal policy has maximum impacts on output. On the other hand, under flexible exchange rates, exchange-rate crowding-out (in) perfectly offsets the impacts of fiscal policy on output – fiscal multiplier tends to zero. Another factor from which the size of the fiscal multipliers depends on is the degree of financial market development. It has, however, ambiguous effects on the size of multipliers, depending on how the degree of financial development affects liquidity constraints and on the government’s ability to finance its fiscal deficit. Usually, a lower degree of financial development imposes liquidity constraints which limit the ability for consumption (and investment) smoothing. The values of the fiscal multipliers are, thus, expected to be larger (Spilimbergo et al., 2009). 5 In this dissertation we assume that an increase in exchange rate means an appreciation of the domestic currency. 7 Regarding government’s ability to finance the fiscal deficit, the impacts of government financing on interest rates also depends on the degree of financial development. In countries with limited access to financial markets, governments can issue debt to finance the deficit only at very high interest rates, which dampens the size of multipliers. However, if in those countries governments issue bonds to “captive” domestic savers, thereby lowering the costs of financing, the size of the decrease of fiscal multipliers can be restricted. For the sake of concreteness, collect the above-described mechanisms into the analysis of policy and “spillover” effects arising from interdependent economies. Consider, as an example of policy co-ordination, the scenario of a monetary union with high capital mobility and well developed financial markets. In this context, fiscal policy externalities may arise, although moving in opposite directions. A fiscal stimulus will increase domestic output but it will lead to additional imports from trading partners boosting the income of the latter (trade channel). At the same time, if the country (or group of countries) where the fiscal shock occurs is large enough, the policy puts upward pressure on the union’s interest rates. This, in turn, has a contractionary effect on both domestic and foreign output (interest rate channel). Additionally, as the exchange rate of the union currency floats against the rest of the world, a fiscal expansion in a (large) member economy causes an appreciation of the common currency, worsening all member-countries’ trade balances. Cwik and Wieland (2011) find that spillovers between Euro area countries are negligible or even negative, because direct demand effects are offset by the indirect effect of the adjustment in the euro exchange rate. In the same veil, a fiscal expansion in the Euro area, through changes in the exchange rate, boosts the non-member countries’ exports and output. There is a locomotive effect between economies under flexible exchange rates, with both domestic and foreign output expansion in the sequence of a fiscal stimulus. In contrast, under a fixed exchange rate regime, between its currencies an expansionary fiscal policy on one country requires an increase in the foreign interest rate, leading to a decrease in the output of the partner countries. Summing up, there is a negative transmission effect between economies, usually known as a beggar-thyneighbor effect. While the output of a country increases, output of the anchored 8 countries decreases. This effect can be smaller or, eventually, reverted if the income feedback effect through increase of exports is strong enough. A more recent generation of models includes dynamic models with rational expectations and infinitely-lived agents. The crucial contribution of these models is that the range of potential transmission channels for fiscal policy to aggregate demand is broader under a longer time horizon. In this scenario, agents form expectations regarding future developments in public finances and budget policies, and hence their future disposable income and wealth. The intertemporal optimization implies complex and non-linear relationships for consumption and investment, which depend, among other things, on how economic agents form their expectations. Due to wealth effects on consumption, this aggregate demand component is one of the most affected by these assumptions. Considering forward looking agents, and no liquidity constraints, the Ricardian Equivalence proposition may hold 6 - there is a precautionary behavior of economic agents that fully offset fiscal policy changes, and the value of the multiplier may converge to zero. Dynamic Stochastic General Equilibrium (DSGE) models, as the NewKeynesian (NK) and the Real Business Cycle (RBC) ones, are examples of this type of framework. Both models predict a short-run expansion in output in the sequence of discretionary expansionary fiscal policies but a decrease in private consumption operating through a negative wealth effect attached to higher current government expenditures. The difference is that RBC models predict an increase in investment that more than compensates the fall in consumption (e.g., Baxter and King, 1993) while NK models tends to exhibit investment crowding-out. In RBC models the negative wealth effect leads to a decline in consumption whereas interest rate, saving and labor supply increase. The rise in hours worked, in turn, causes real wages to fall whereas investment increases. This described fiscal policy transmission is in force whether the rise in government spending is permanent or persistent. The intuition behind the increase in 6 The concept of Ricardian Equivalence was revived by Barro (1974) and means that a tax cut/expenditure increase financed by issuing government debt may fail to stimulate private consumption, because consumers discount the future tax burden required to debt service and repayment, and so increase their savings accordingly and no impact on output occurs. 9 investment is that a higher steady-state level of hours worked requires initially higher investment to build up additional capital stock. In contrast, in a NK model, after a positive shock to government consumption real wages increase. This is because the resulting increase in output raises the demand for labor which, in the NK setup, offsets the increase in the labor supply due to the negative wealth effect. As regards to investment, in NK models with capital accumulation, an increase in government spending raises the interest rate and crowdsout private investment. Increases in output result from the effects of fiscal stimulus on the supply side in RBC models, while they operate under the demand side in NK models. Additionally, both models predict only temporary output effects from fiscal stimulus. In the NK models, as time passes, prices adjust upwards and, in the medium-to-long run, fiscal shocks affect nominal but no real variables. In RBC models, only in long run permanent effects on output occur through changes in supply-side. These effects are captured by the cumulative fiscal multipliers. 2.2. Empirical evidence on the size of multipliers In the remainder of this subsection we propose to review most of the empirical results in the literature, while comparing them with the theoretical expected outcomes described above. There is a huge amount of empirical work on the assessment of the size of fiscal multipliers. Deliberately, we will focus on those using the most standard methodology, namely the vector autoregression (VAR) framework. Moreover, as several literature reviews have been frequently produced (see, for instance, Briotti (2005), Spilimbergo et al. (2009) and Hebous (2011)), we will focus mainly on more recent papers. Table 1 briefly reviews some of the most relevant papers of this strand of literature, complemented by some recent studies on the impacts on output from fiscal stimulus. 10 Table 1: Fiscal multipliers in literature Fiscal Instrument / Composition Effects on output Size-range for multipliers Author Methodology Sample Impact Cumulative Public spending (total) Positive 1.68 1.21 (10 years) Pereira and Sagalés (2011) VAR (Cholesky decomposition) Portugal (19802005) 0.1 0.2 (5 years) Fatás and Mihov (2001) VAR (Cholesky decomposition) U.S. (19601996) 0.75 0.02 (5 years) Burriel et al. (2010) VAR (Structural approach) U.S. (1981Q12007Q4) < 1 Afonso and Sousa (2012) VAR (Bayesian structural approach) U.S. (1970Q32007Q4) U.K. (1964Q22007Q4) Germany (1980Q32006Q4) Italy (1986Q22004Q4) Negative 0.65 -2.24 (5 years) Mountford and Uhlig (2009) Sign restrictions U.S. (19552000) Ambiguous 0.8 -0.24 (5 years) Blanchard and Perotti (2002) VAR (Structural approach) U.S. (1947Q11997Q4) 0.4 -0.2 (4 years) Agnello et al. (2011) Econometric approach (2SLS) Panel of 132 11 countries (19602008) Taxes (total) Positive Negative 0.00 -1.83 (10 years) Pereira and Sagalés (2011) VAR (Cholesky decomposition) Portugal (19802005) -1.4 -0.4 (5 years) Blanchard and Perotti (2002) VAR (Structural approach) U.S. (1947Q11997Q4) Public Investment Positive 2.44 4.69 (10 years) Pereira and Sagalés (2011) VAR (Cholesky decomposition) Portugal (19802005) Wage expenditure Negative 0.33 -2.68 (10 years) Pereira and Sagalés (2011) VAR (Cholesky decomposition) Portugal (19802005) < 0 7 Alesina and Ardagna (2010) Case study OECD countries (19702007) -0.48 -2.56 (5 years) Alesina et al. (2002) VAR (Cholesky decomposition) OECD countries (19601996) Direct taxes Negative -0.1 -2.78 (10 years) Pereira and Sagalés (2011) VAR (Cholesky decomposition) Portugal (19802005) Fatás and Mihov (2001) compare the dynamic impacts of fiscal policy on macroeconomic variables implied by a typical RBC model with the empirical results 7 Alesina and Ardagna (2010) do not compute the size of fiscal multipliers, they only compare the effects of different compositions of major fiscal changes. 12 from an identified vector autoregression, using quarterly data from 1960Q1 to 1996Q4 for the U.S. They find strong and persistent increases in private consumption and employment in response to rises in government expenditures. This is inconsistent with theoretical benchmark of RBC models where private consumption falls in response to rises in government expenditures. Another discrepancy between the model and the empirical results is that while empirical results show a positive conditional correlation between consumption and employment, in the theoretical benchmark these variables move in opposite directions. Overall, their results show that increases in government spending are expansionary with residential investment being the main driving force for the persistent rise in output. In a more disaggregated analysis they find that the increase in government spending is followed by a persistent rise in all components of consumption (durable goods, nondurables goods and services). Investment also increases, but only within a lag of six quarters, returning to trend after three years. This evidence on the dynamics of investment goes against the Keynesian view and favors the classical (RBC) approach. In contrast, Alesina et al. (2002), using a panel of OECD countries and covering data from 1960 to 1996, show that there is a negative effect of government expenditures on investment (crowding-out effect of fiscal policy), being this negative effect mainly associated to increases in government wages. A possible reason for this effect is that increases in wages in the public sector stimulate unions pressure for wage increases in the private sector, which have negative impact on profit margins and, consequently, on private investment. Afonso and Sousa (2012) analyze recent empirical evidence for the U.S., the U.K., Germany, and Italy, respectively, for the periods 1970Q3-2007Q4, 1964Q22007Q4, 1980Q3-2006Q4, and 1986Q2-2004Q4. They use a bayesian structural vector autoregression approach with a recursive identification scheme to identify fiscal policy shocks. They conclude that government spending shocks have, in general, a small effect on GDP; they do not impact significantly on private consumption and have a negative effect on private investment (evidence for the NK approach). Another recent paper in this domain is Pereira and Sagalés (2011) who, besides analyzing the effects of expenditure-side fiscal policies, as the previous recorded studies 13 do, also analyze the impacts, both at aggregate and disaggregate levels, of revenuebased fiscal policies. The authors consider four components on the spending side: current transfers, intermediate public consumption, wages and public investment; and two components on revenue side: direct and indirect tax revenues. Using the Cholesky decomposition for shock identification, the cumulative impulse-response functions are drawn for Portugal and rely on a sample covering the period from 1980 to 2005. At the aggregate level, results show that revenue-side is more effective than spending-side fiscal policies in affecting output. By definition, an increase in taxes does not produce contemporaneous effects in output (impact multiplier is 0); although, an intertemporal accumulated effect in output is estimated in -1.83 (cumulative multiplier). In contrast, an increase in public spending produces contemporaneous effects in output (impact multiplier is of 1.68) while the cumulative multiplier is 1.21. This suggests important short-term demand effects and negligible longer-term effects. Indeed, the response of output becomes insignificant after five years. Moreover, these results go against the Keynesian theory which postulates that the tax multiplier is (in absolute value) smaller than the spending multiplier. At a more disaggregated level, results in Pereira and Sagalés (2011) show that, among the spending-side components, public investment, wages, current transfers and intermediate public consumption exhibit, respectively, the highest through the lowest multiplier value. However, and in spite of the barely positive impact on output, wages have a negative cumulative multiplier estimated in 2.68. For the remaining expenditures, impact (and cumulative) multipliers range from 2.44 (4.69) for public investment to 0.27 (0.62) for intermediate consumption. As in the case of expenditure on wages, transfers also imply transition negative impacts on output. These results are with the Keynesian theory with exception of those observed for the compensation of employees; in this case, results go in line of those found by Alesina et al. (2002). On the revenue side, direct taxes have stronger impacts on output than indirect taxes, with impact (and cumulative) multipliers of -0.1 (-2.78) and -0.06 (-0.18), respectively. The smaller effect of indirect taxes in output is explained by the large weight of consumption taxes in indirect taxes in Portugal together with consumption patterns and habits relatively price-inelastic; thus, indirect taxes have not significant effects on aggregate demand. 14 As for evidence on how structural or institutional factors shape the size of multipliers, we found evidence covering for the exchange rate regime, the degree of openness and the level of public debt-to-GDP ratio. Table 2 show the results found by Ilzetzki et al. (2011) for government spending multiplier across different scenarios: exchange rate regimes, openness trade degree and financial fragility. Table 2: Government spending multiplier across different scenarios Fiscal Instrument Transmission Channel Cumulative Multiplier Public spending Exchange rate regime Fixed Floating 1.5 0 Openness trade degree “Closed” Economy (<60% GDP) “Open Economy” (> 60% GDP) 1.2 -0.47 Financial fragility High debt (debt-to-GDP ratio > 60% GDP) for 3 (or more) consecutive years -2.3 Source: Ilzetzki et al. (2011). Ilzetzki et al. (2011) analyze the role of the exchange rate regime in size of the fiscal multipliers. They find that the choice of the exchange rate regime is relevant to the effectiveness of fiscal policy. Using a panel VAR framework, they estimate sizeable (long-run) multipliers in countries under fixed exchange rates (1.5), while multipliers are zero in countries under floating exchange rates. This empirical evidence supports the conclusions of the Mundell-Fleming model which predicts that the effectiveness of fiscal policy increases as the exchange rate regime becomes more rigid. In another recent study, Born et al. (2012) estimate a VAR that suggests a shortrun multiplier of about 1.2 under fixed exchange rates and 0.75 under floating exchange rates. Hence, the multiplier differs across exchange rate regimes – but to a lesser extent 15 than the reported in earlier studies. However, the dynamics of the exchange rate and net exports provide little support to the fiscal transmission mechanism at the heart of the Mundell-Fleming model. In their analysis, under the assumption that government debt is riskless, a small open economy framework cannot account for cross-country spillover effects. In the presence of sovereign risk the multiplier is likely to be smaller relative to “normal” times 8 if monetary policy is constrained (Corsetti et al., 2012a) as such spillovers may be sizeable, notably within monetary unions (Corsetti et al., 2011). Also, the dynamics of the real exchange rate may differ systematically across currency regimes. A slightly appreciation of the real effective exchange rate followed by a depreciation over time is a behavior described by the literature after a rise in government spending; recent studies, however, document a fall in the real exchange rate after a rise in government spending in flexible exchange rate regimes. In contrast, studies focusing on fixed exchange rate regimes tend to document real appreciation in response to positive spending shocks (Corsetti et al., 2012a). Regarding the degree of openness, Ilzetzki et al. (2011) conclude that it is also a relevant variable for the effectiveness of fiscal policy. In line with the theoretical models, more closed economies, i.e., economies with a degree of openness 9 below 60% of GDP, present higher multipliers than small open economies. They found that the government spending multiplier is larger in more closed (large) economies relative to small open economies, with an impact multiplier of 0.02 in the former and -0.19 in the latter and a long-run multiplier of 1.29 in the former and -0.47 in the latter. This difference is statistically significant on impact, but not at longer horizons. Last but not the least, Ilzetzki et al. (2011) analyze how the level of public debtto-GDP influences the multiplier. Results obtained show that for countries with a debtto-GDP ratio above 60%, the cumulative, expenditure-side, expansionary fiscal policies have negative effects on output. They estimate a long run multiplier of -2.3. Recently, Corsetti et al. (2012b) found that in “good” times, fiscal balances and private consumption co-move negatively while in “bad” times 10 , with high levels of public debt, the co-movement becomes positive - a set of results partially supporting nonKeynesian effects of fiscal policy. 8 In this context “normal” times refers to a non-recession period. 9 Ilzetzki et al. (2011) calculate degree of openness as . 10 In this context “good” times mean an expansion period and “bad” times a recession period. 22 the context of a theoretical New Keynesian model calibrated to match a large number of features of postwar U.S. data. Cogan et al. (2010) also argue that an increase in government purchases during a period in which the interest rate is zero, which is expected to last for the current quarter only - implying no change in expected future income or inflation -, has a multiplier of exactly 1. So, a value of around 2.3 for the temporary government spending obtained by Eggertsson (2011) and Cogan et al. (2010) means that 1.0 of this is due to the increase in government purchases during the current quarter, while the other 1.3 results from higher anticipated government purchases in the future. More recently Christiano et al. (2011) and Woodford (2011) through NK models obtained similar results, confirming that multipliers are considerably larger if monetary policy is constrained by the zero lower bound. 2.4. Do multipliers change during fiscal consolidation periods? The study of how periods of fiscal consolidation shape the size of fiscal multipliers is meaningful because such scenario represents an example of a structural policy regime shift and also because potential non-Keynesian effects may arise. Moreover, in the sequence of the recent global economic and financial crisis, most of developed countries are now struggling against unsustainable public debt paths. In an environment characterized by intertemporal optimization, large fiscal imbalances and high risk premium on interest rates can make fiscal expansions to have contractionary effects: consumers’ and investors’ confidence decreases as fiscal expansion reinforces fiscal sustainability concerns. Similarly, a credible fiscal consolidation - that aims at lowering the public debt-to-GDP ratio in a permanent way - can lead to an improvement of agents’ expectations about future fiscal policy and, thus, an improvement in expected future income. In such scenario, fiscal multipliers value are expected to be negative or close to zero. Moreover, a fiscal consolidation scenario is an example of a negative permanent shock. Literature shows that permanent measures deliver higher multipliers than temporary measures in interventions that work through income (e.g., changes in direct 23 taxes), while the reverse is true for interventions operating through prices (changes in indirect taxes or in investment taxes) because changes in relative intertemporal prices are more likely to affect intertemporal consumption patterns, (Spilimbergo et al., 2009). So, one should expect different size (even different sign) for multipliers in periods of fiscal consolidation relative to those when no consolidation applies. Moreover, as some different consolidation strategies impinge differently on the success of reducing debt permanently, it is expectable that the underlying fiscal multipliers differ for successful consolidation processes relative to unsuccessful ones. 16 Based on the arguments previously reviewed, during a fiscal consolidation the size of first year multipliers is larger if the fiscal consolidation is based on government expenditures – and government investment in particular –, if the measures taken are not credible and of temporary nature, if agents are not financially constrained and if no reduction on real interest rates occurs alongside with the fiscal shock. Moreover, these effects are also larger if consolidations are implemented at the same time worldwide. The composition of consolidation is non-neutral to long-term output, with taxbased consolidations being less supportive of long-term growth (European Commission, 2012). Also, relying on the literature reviewed in the previous subsection, when a consolidation process delivers a recession, as automatic stabilizers represent a larger weight, multipliers for discretionary fiscal policy would turn out to be smaller during debt-correcting periods. Additionally, a recessive environment feeds-back to the debtto-output ratio in proportion of the latter (the larger the debt, the larger the impacts of growth on the debt service); thus larger indebtedness is expected to produce smaller fiscal multipliers during consolidations. The same effect is expected if confidence breaks occur during recessions; in contrast, nominal and real rigidities and zero lower bound scenarios are expected to amplify multipliers. 16 In the literature, there are several ways to identify periods of successful fiscal consolidation. For instance, Alesina and Ardagna (2010) define a successful fiscal consolidation episode if it brings down the debt to GDP ratio by at least 4.5% of GDP in the three years following a consolidation episode. Instead, according to the European Commission (2007), a fiscal consolidation is successful if the following condition applies: in the three years after the end of the consolidation episode the CAPB does not deteriorate by more than 0.75% of GDP in cumulative terms compared to the level recorded in the last year of the consolidation period. In other words, at least half of the overall minimum fiscal correction required to qualify as consolidation has to be safeguarded three years after. 24 However, fiscal adjustments have, in general, a negative, but small, impact on economic activity. Even expansionary non-Keynesian effects from expenditure cuts can emerge in the medium run as a result of anticipated effects of higher future disposable income or profitability (e.g., Giavazzi and Pagano, 1990; Alesina and Ardagna, 2010). 17 This suggests that non-Keynesian responses of private consumption are more likely when changes in fiscal policy are large and persistent. Expansionary effects of fiscal consolidations can go through both demand and supply-side channels. Demand-side channel works through positive wealth effect (expectations of an increase in lifetime disposable income, relishing liquidity constraints) or through interest rate effect (Alesina and Ardagna, 2010). The implementation of a credible consolidation program (e.g., subject to supra-national rules, under the surveillance of a fiscal council or with substantial political costs), reducing risk premium on government securities, may lead to general fall in interest rates and crowding-in effects from private investment. Falling real interest rates lead to a decrease in public interest payments which can eventually lead to a decrease in future tax rates; supply-side effects may arise, contributing positively to support growth. Moreover, the responsiveness of long-term interest rates to substantial consolidation is likely to be stronger at high debt levels (OECD, 2010). Alesina and Ardagna (2010) argue that supply-side channels entail expansionary effects of fiscal consolidations to work via labor market: tax increases and/or spending cuts impact on the individual labor supply as it affects the unions’ fall-back position under imperfectly competitive labor markets - lower wage bargaining power of unions impinge positively on profits, investment and competitiveness of the private sector. Supply-side effects are expected to be larger when consolidations operate through cuts in the government wage bill, while tax-based consolidations are found to be less supportive of long-term growth, (European Commission, 2012). Alesina and Ardagna (2010), using annual data for OECD countries from 1970 to 2007, identify episodes of expansionary effects on output after a fiscal consolidation (non-Keynesian effects of fiscal policy). Considering that an episode of fiscal 17 Non-Keynesian effects of fiscal policy have been re-interpreted in sequence of the recent global financial crisis. Recently, several authors as Perotti (2011) concluded that this output expansion typically reflects exchange rate depreciation and a relaxation of monetary conditions rather than confidence effects per se arising from fiscal tightening. 25 adjustment is expansionary if the average growth rate of GDP, in difference from the G7 average (weighted by GDP weights), in the first period of the episode and in the two years after, is greater than the value of 75th percentile of the same variable empirical density in all episodes of fiscal adjustments, they identified 26 episodes of expansionary periods during fiscal adjustments (3.7% of the observations of the entire OECD sample). OECD (2010) provides an analysis of the dynamics of public debt of the Member States and makes recommendations to correct this situation when countries face an unsustainable path. Results show that, for countries with no stabilized debt, policy authorities must compromise to achieve a primary fiscal surplus quantified in 0.5% of GDP. This means an improvement of the primary fiscal balance for the OECD countries quantified in 5.25% of GDP, on average, until 2025. This adjustment would allow the stabilization of the public debt-to-GDP at 110% (average of OECD countries). The OECD Global Model simulations suggest that fiscal consolidations are typically contractionary in the short run and expansionary only after two to four years. Sustainable output increases are expectable between 4 and 5 years after the beginning of the fiscal consolidation, due to a decrease of risk premium and, consequently, a decrease in average interest rates. Considering expenditure ranking, results confirm those of general literature: 2 years after the beginning of the process, multipliers between 0.9 and 1.1 are found for public investment, 0.4 to 0.7 for public transfer to households and 0.5 to 0.9 for public consumption. As for the revenue side, up to 2 years after starting consolidation, the multiplier for indirect taxes is comprised between -0.2 and -0.4 and between -0.4 and -0.7 to personal income taxes. These results support most of empirical evidence, which suggests that consolidations based on expenditure side rather than on the revenue side tend, in general, to be more long-lasting and more growth-supporting in the medium-term, but more recessive in the short-term (Alesina et al., 2012). There is a trade-off between short-run pain and long-run gain. The pain arises from the negative multiplier effects of lower spending or higher taxes, while the gain stems from the lower world interest rates and lower distortionary taxes associated with lower debt levels. The results on both pain and gain are subject to important qualifications such as the design of a fiscal package; if the tightening is well designed with favorable long-run incentives to investment and 26 labor supply, then the short-term pain only arises in the presence of an initial lack of credibility and only lasts for the non-credibility period. If, on the other hand, the fiscal tightening is badly designed, for example, sharply raising taxes on income or cutting essential government investment, the long run gain could be much lower or even nonexistent as higher distortions and/or productivity losses offset the gains from lower real interest rates (European Commission, 2012). Typically, long-term growth is what matters to ensure fiscal sustainability; however, recently, financial markets are focusing in short-term growth due, for instance, to markets’ beliefs that a country facing a sizeable decline in GDP is unlikely to sustain its fiscal adjustment effort over time or due to some degree of short-termism by market agents following several years of heightened market uncertainty, (European Commission, 2012). This behavior contributes to smaller fiscal multipliers. OECD (2010) also emphasizes the importance of several features that influences the consolidation effort and, also, the size of the fiscal multipliers: privatization revenues or the implementation of structural reforms to promote economic growth in the medium-term. In fact, fiscal consolidation processes are often associated with the implementation of structural reforms (e.g., enhancing competition in goods, service and labor markets that result in improvements in efficiency and competitiveness of the economy – supply-side effects). Since the purpose of structural reforms is to allow an increase in potential output, potentially leading to higher output growth rates in the medium/long-term, and given that successful consolidations may have other structural impacts on affecting interest-rate risk premium or expectations, fiscal multipliers may crucially differ (being smaller) during these episodes. Favero et al. (2011) estimate a Global VAR (GVAR) for a sample of fifteen countries, using annual data from 1978 to 2009. In this paper they identify cases in which the government implements tax hikes or spending cuts (at the general government level) in order to primarily reduce the budget deficit and put public finances on a sustainable path. The strategy to identify cases in which the government implements meaningful tax hikes or spending cuts is designed by the “narrative approach” to identify the fiscal shocks. This approach identifies episodes based on fiscal policy actions motivated by deficit reduction, as described in several policy documents, irrespective of the outcomes. "Narrative shocks" are identified on the exam of intentions 27 and actions as described in policy documents to identify measures motivated primarily by deficit reductions. Such fiscal actions represent a response to past decisions and past economic conditions rather than to prospective conditions. This paper includes also variables, such as the initial public debt-to-GDP ratio, trade openness and spillovers which are not included in others papers. The main conclusion from their results is that there is not an unconditional fiscal multiplier. The effect of fiscal policy on output is different according these several factors, i.e., during consolidations, depending on different debt dynamics, different degrees of openness, etc. At last but not the least it is important refer the effect of consecutive consolidations. If consolidations are repeated, especially in periods where multipliers are large and persistent, in presence of continued myopic behavior of financial markets, can have counterintuitive dynamics, i.e., in this situation it is possible that the scenario consolidation-debt increase-consolidation-further debt increase takes place as long as the current multiplier is high enough to induce further short-term debt-to-GDP increases in response to consolidation (European Commission, 2012). 28 3. Methodology and data Among the relevant literature we found three main methodologies that can be used to assess the effectiveness of fiscal policy: large structural macroeconomic models such as DSGE models, case studies as in Romer and Romer (2010) and Vector Autoregression (VAR) approaches. Macroeconomic model simulations, frequently, have an underlying aggregate demand/aggregate supply structure characterized with little forward looking behavior as far as decisions by households and firms that anticipate future changes in government policies is concerned. These frameworks deliver positive impacts on output by construction in response to an increase/decrease in public spending/revenue. The multipliers obtained are small or negative only if fiscal sustainability is in question, economic agents are substantially forward looking or when monetary policy is not accommodative enough. See, for example Smets and Wouters (2003), Laxton and Pesenti (2003) and Ratto et al. (2009). Case studies are based in the identification of real experiments, e.g., episodes of truly exogenous fiscal expansions. Romer and Romer (2010) is one of most reputational papers using this approach, applied to study the impacts of changes in tax policy. An important feature of case studies is that the results obtained are specific to the type of fiscal measure studied and, for example, the prevailing macroeconomic conditions at the time of implementation. Another methodology usually found in the literature for the assessment of the effectiveness of fiscal policy is, as already mentioned, the VAR approach. Such as in the case studies, the VAR methodology is based in the correct identification of exogenous movements in public expenditure or taxes. VARs give the response of the economy, taking implicitly into account the monetary policy response and thus the effects on the interest rates (whether or not interest rates are included in the VAR), to fiscal shocks. See, for example, Fatás and Mihov (2001), Blanchard and Perotti (2002) and Burriel et al. (2010). In this work we intend to estimate the fiscal multipliers for the Euro area as a whole through using a VAR approach. In this context, and in what follows, we will present the model and model extensions to be estimated, after previously characterizing, 29 in brief, the VAR methodology. Additionally, we provide the description of all variables and as well as the respective data sources. 3.1. VAR methodology The usual representation of a reduced VAR model takes the following form: ∑ where the n-dimensional vector X includes the endogenous variables of interest (e.g., public spending, output, taxes) and is the nxn matrix of coefficients. The optimal number of included lags, k, can be determined by a priori evidence on the behavior of the variables or through some information criteria such as the Schwarz or Akaike criteria. The vector of reduced-form residuals, , is n-dimensional with the variancecovariance matrix ∑e, where E[ . The reduced-form residuals capture three components: automatic stabilizers, discretionary fiscal policy responses (active changes in the tax rates or in public expenditure for stabilization purposes) and random discretionary fiscal policy responses. The latter is what a structural fiscal shock is meant to capture. The structural VAR model takes the following form: ∑ where the matrix describes the contemporaneous relationships among the variables in the vector . The matrix B describes the relation between the reduced-form residuals and the structural-form residuals : 30 The dynamics of the variables following a one unit increase in the current value of the structural residual in the equation respecting a fiscal variable, holding all other residuals fixed (that is a “structural fiscal shock”), can be summarized in the impulse responses of the variables included in the system. The crucial challenge is how to pin down the structural shocks. To compute the impulse responses of the variables in the system, the matrices , B and the variancecovariance matrix of the structural residuals (∑v) must be estimated. However, the system can be identified only if some coefficients in the matrices are restricted to take certain values typically justified by theoretical considerations (“identifying assumptions”). Even with the assumption that the matrix B is the identity matrix, this exclusion restriction is not enough to identify the system. The diagonal matrix ∑v can be used to express ∑e as follows ∑v ( ∑e. Still, this relation does not identify a unique solution. As previously referred the crucial point to estimate VAR is identify structural fiscal policy shocks. In the literature there are four approaches to identify structural fiscal shocks. The recursive formulation (Cholesky decomposition) approach proposed by Sims (1980); the structural identification (SVAR) approach proposed by Blanchard and Perotti (2002); the sign restriction proposed by the approach developed by Uhlig (2005) and the narrative (or dummy variables) approach proposed by Ramey and Shapiro (1999). According the recursive formulation approach, the first variable ordered in the system responds only to its own exogenous shock. In the simplest three dimensional VAR model used to assess fiscal policy multipliers, including spending, output and taxes, spending is ordered first, meaning that government spending is assumed not to contemporaneously react to shocks to the economy. The second variable, output, responds to the first variable (spending) and to its own shock. The third variable, taxes, reacts to its own shock and shocks to the previous two variables. Following this approach, the matrix is a triangular matrix with ones on the main diagonal, and the matrix B is the identity matrix. [ ] [ ] [ ] [ ] 31 In this context, the ordering of the variables plays a crucial role because it defines the direction of causal relationship between them. Although, there is no theoretical guide for ordering the variables, and the assumptions behind any ordering are nontrivial. For instance, positioning taxes after output excludes a priori contemporaneous effects of taxes on any component of output, including private consumption, which is a rather strong assumption. According the structural identification approach, some elements of the matrix can be obtained by using information on the elasticities of output relative to spending and taxes. These elasticities correspond to certain elements in the matrix . These values are exogenously imposed in the VAR model. The system takes the following form: [ ] [ ] [ ] [ ] where , and are, respectively, the value of the elasticities of output relative to government spending and net taxes, and the elasticity net taxes relative to output. The crucial assumption needed in this identification approach is that government spending does not react within a period time, typically assumed to be a quarter, to shocks to the economy. This is based on the idea that fiscal actions are subject to the decision lags, e.g., the time needed for fiscal policy makers to respond to the status of the economy. Also, estimating while setting = 0 implies that spending decisions come before tax decisions. In contrast with the two previous approaches, the sign restriction approach does not require imposing zero contemporaneous effects on some variables, but requires restrictions on the sign of the impulse responses of the fiscal variables. For example, Mountford and Uhlig (2009) identify a government spending shock as a shock that is orthogonal to monetary and business cycle shocks, and for which government revenue increases for four quarters after the shock.. Hence, they impose also sign restrictions on revenue increases to identify monetary and business cycle shocks. The fourth and last approach to identify structural fiscal shocks is the narrative (or dummy variables) approach, which is particularly implemented in studies focusing 38 4. Estimation results In this section we compute fiscal multipliers resulting from VAR estimations under the specifications with both aggregated and disaggregated fiscal variables. Multipliers are computed using all observation in the sample and also considering subsample: recession vs. expansion. Moreover, we control, to aggregated and disaggregated fiscal variables, for the effect of fiscal consolidation and, to aggregated fiscal variables, for the effect of the degree of openness and the debt dynamics on the size of fiscal multipliers. Results are critically analyzed by comparing them with those predicted by the literature reviewed in section 2, above. 39 4.1. Global Analysis Results from VAR estimation using aggregated fiscal variables are reported in Table 8, below. Table 8: Estimation results - aggregated fiscal variables, full sample. 40 At a 5% significance level, all variables are globally significant for all the four equations. Specifically, F-statistic > F critical value (4,175) = 2.423. 21 In VAR models, interpretation of the estimated coefficients is not trivial because each endogenous variable is explained by other lagged variables that result from other equations of the model. Indeed, the most appealing interpretation is given through the analysis of the impulse response functions. Figure 1 shows the accumulated impacts on DELTA_GDP to shocks in DELTA_SPENDING and DELTA_TAXES, respectively. 21 F critical value was computed by using the function = @qfdist(0.95, 4,175) in Eviews 7.1 software. 41 Figure 1: Accumulated impulse response functions of DELTA_GDP to shocks in DELTA_SPENDING (left-hand side) and DELTA_TAXES (right-hand side), full sample. Public spending, overall (excluding debt service expenditure) impinges negatively on impact, but the cumulative impact on output is positive. Taxes, as expected, have (both impact and cumulative) negative impacts on output. By normalizing the impulse response functions considering a one percentage point shock to public spending and taxes, we compute the accumulated fiscal multipliers on aggregate fiscal variables. In particular, multipliers are computed by dividing the -.015 -.010 -.005 .000 .005 .010 .015 12345678910 Accumulated Response of DELTA_GDP to DELTA_SPENDING -.015 -.010 -.005 .000 .005 .010 .015 12345678910 Accumulated Response of DELTA_GDP to DELTA_TAXES Accumulated Response to Cholesky One S.D. Innovations ± 2 S.E. 42 accumulated responses of GDP to a 1 standard deviation shock in public spending (or taxes) by the impact response of public spending (or taxes) to its own shock – this gives the elasticity of output relative to public spending (or taxes). Thus, public spending (or taxes) multiplier results from dividing the corresponding elasticity by the average weight of public spending (or taxes) on output. Table 9 shows the results and they respect to the changes in output induced by a euro unit shock in the fiscal instrument. 22 Table 9: Cumulative fiscal multipliers, full sample - public spending vs. taxes. Period Public Spending Taxes 1 -0.077 23 0 2 -0.065 -0.108 3 -0.031 -0.185 4 -0.003 -0.3 5 0.017 -0.255 6 0.03 -0.269 7 0.038 -0.279 8 0.044 -0.285 9 0.048 -0.289 10 0.05 -0.291 Source: own calculations Table 10 shows the estimation results for the VAR using disaggregated fiscal variables. 22 Throughout the text, similar computations were made for different disaggregation of fiscal instruments or for different periods of analysis. 23 E.g., ⁄. IRF values of output and public spending to a 1 standard deviation shock in public spending are available upon request. 43 Table 10: Estimation results - disaggregated fiscal variables, full sample. 44 At a 5% significance level, all variables are globally significant for all the nine equations. Specifically, F-statistic > F critical value (9,175) = 1.934. 24 Figure 2 and Figure 3 show the cumulative responses of GDP growth to shocks in the growth rate of different items of public spending and revenue, respectively. Figure 2: Accumulated impulse response functions of DELTA_GDP to shocks in disaggregated items of DELTA_SPENDING, full sample. 24 F critical value was computed by using the function = @qfdist(0.95, 9,175) in Eviews 7.1 software. -.012 -.008 -.004 .000 .004 .008 .012 .016 .020 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GTRA -.012 -.008 -.004 .000 .004 .008 .012 .016 .020 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GICO -.012 -.008 -.004 .000 .004 .008 .012 .016 .020 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GWAG -.012 -.008 -.004 .000 .004 .008 .012 .016 .020 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GINV Accumulated Response to Cholesky One S.D. Innovations ± 2 S.E. 45 Figure 3: Accumulated impulse response functions of DELTA_GDP to shocks in disaggregated items of DELTA_TAXES, full sample. On the expenditure side, transfers are the main driving force for the overall expenditure dynamics (cf. responses to DELTA_GTRA with those to DELTA_SPENDING, above in Figure 1). Moreover, wages exhibit negative impacts on output while, positive effects are driven by shocks in public investment and intermediate consumption. On the revenue side, all items impinge negatively on output growth. Table 11 shows the computed multipliers for different spending and revenue items. Values refer to cumulative multipliers per period. -.020 -.015 -.010 -.005 .000 .005 .010 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_SS -.020 -.015 -.010 -.005 .000 .005 .010 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_TDIR -.020 -.015 -.010 -.005 .000 .005 .010 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_TIND Accumulated Response to Cholesky One S.D. Innovations ± 2 S.E. 46 Table 11: Cumulative fiscal multipliers, full sample - disaggregated public spending vs. disaggregated taxes. Period Public Spending Taxes GTRA GICO GWAG GINV SS TDIR TIND 1 -0.353 0.25 -0.6 1.62 0 0 0 2 -0.391 0.245 -0.382 1.579 -0.313 -0.369 0.171 3 -0.256 0.499 -0.163 1.663 -0.56 -0.636 -0.108 4 -0.14 0.609 -0.137 1.845 -0.628 -0.762 -0.383 5 -0.081 0.632 -0.136 1.993 -0.66 -0.849 -0.522 6 -0.042 0.657 -0.121 2.095 -0.7 -0.92 -0.595 7 -0.009 0.686 -0.103 2.171 -0.735 -0.975 -0.648 8 0.016 0.709 -0.089 2.228 -0.76 -1.014 -0.689 9 0.035 0.726 -0.079 2.272 -0.778 -1.043 -0.72 10 0.049 0.739 -0.072 2.304 -0.792 -1.065 -0.742 Source: own calculations. As previously referred, the literature frequently points that government spending multiplier is higher than tax multiplier and that an increase in GDP can be obtained through an increase in government spending or a decrease in taxes. Our results, using aggregated fiscal data, do not satisfy the first stylized fact. Tax cumulative multiplier is higher than government spending cumulative multiplier (in modulus 0.291 vs. 0.05); the same applies to impact multipliers with values (in modulus) of 0.108 and 0.077, respectively. The second stylized fact is only partially supported by the results. While, on impact and cumulative terms, an increase in GDP can be obtained by a decrease in taxes, as far as government spending multiplier is concerned, an increase in total public spending produces positive impacts on output only in cumulative terms, on impact the effects are negative. Theoretical literature and empirical evidence, ceteris paribus, do not support this last result, especially, because many government spending packages are directed to agents that face liquidity constraints and so, use this increase in disposable income to smooth consumption or pay debts. It is also important to refer that the values of the multipliers are low comparatively with others obtained through similar methodology (see Table 1, section 2). However, the analysis regarding the effects of disaggregated expenditure and revenue on output provides some insights to previous results. As far as government spending multiplier is concerned, the component with the highest positive effect on output is public investment (GINV), delivering impact and cumulative multipliers of 1.62 and 2.304, respectively. Public investment is, frequently, 47 directed to infrastructures or to the qualification of human resources that bring positive effects, not only at the time when these measures are implemented but, also, in the longer run, either directly or indirectly, by increasing the productivity of private inputs. 25 Intermediate consumption (GICO) also has a positive effect on GDP at all-time horizons, exhibiting multipliers of 0.25 on impact and of 0.739 in cumulative terms. On the other hand, results obtained relative to compensation of public employees (GWAG) and transfers (GTRA) have negative effects on output, at least on impact. According to the literature, while in the long run increases in government wages can lead to distortions in the labor market between public and private sector with negative impact on employment, and thus on output (confirming the negative cumulative multiplier of 0.072), the negative value obtained for the impact multiplier is counterintuitive because an increase in wages leads to increases in disposable income of beneficiaries and through smoothing consumption behavior we should expect this value to be positive. Negative impact values could result if households acknowledge such expenditures as spurious and to be fully financed, in the very near future, with taxes. Results on transfer multipliers are also counterintuitive: GTRA exhibit a negative impact multiplier (-0.353), but positive contributions arise during the third year after the shock (note that the multiplier starts increasing in period 3, Table 11). Using quarterly data, Burriel et al. (2010) also finds similar evidence for the Euro area: an initial positive impact on output is reversed after some quarters and net positive impacts arise only within 30 to 40 quarters, in accordance to the dynamics in Table 11. An increase in government transfers has, frequently, the same effects of a decrease in direct taxes: transfers increase the level of real disposable income, and hence of consumption. One explanation is that transfers work as automatic stabilizers and, in the very short run may exhibit a strong negative correlation with output (reverse causality); the stabilization role of transfers becomes less pronounced as time passes. The effects of these latter instruments differ, however, if transfers include pension payments and other transfers that are taxable and, hence, some of the increase in incomes is absorbed through an increase in tax payments. According to the literature, a 1% of nominal GDP cut in transfers has a smaller multiplier effect than a 1% of 25 There is an extensive literature on the productivity of public capital, basically following the seminal work of Aschauer (1989). 54 and cumulative fiscal multipliers comparing expansions with recessions are presented in Table 18. Table 16: Estimation results - aggregated fiscal variables, expansions. 55 Table 17: Estimation results - aggregated fiscal variables, recessions. 56 Figure 4: Accumulated impulse response functions of DELTA_GDP to shocks in DELTA_SPENDING (left-hand side) and DELTA_TAXES (right-hand side), expansions. Figure 5: Accumulated impulse response functions of DELTA_GDP to shocks in DELTA_SPENDING (left-hand side) and DELTA_TAXES (right-hand side), recessions. -.02 -.01 .00 .01 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_SPENDING -.02 -.01 .00 .01 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_TAXES Accumulated Response to Cholesky One S.D. Innovations ± 2 S.E. -.02 -.01 .00 .01 .02 .03 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_SPENDING -.02 -.01 .00 .01 .02 .03 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_TAXES Accumulated Response to Cholesky One S.D. Innovations ± 2 S.E. 57 Table 18: Cumulative fiscal multipliers, expansions vs. recessions - public spending vs. taxes. Period Public spending Taxes Expansions Recessions Expansions Recessions 1 -0.167 0.037 0 0 2 -0.192 0.062 -0.076 -0.15 3 -0.18 0.085 -0.121 -0.285 4 -0.167 0.104 -0.139 -0.382 5 -0.158 0.117 -0.144 -0.447 6 -0.152 0.127 -0.144 -0.49 7 -0.148 0.133 -0.143 -0.52 8 -0.146 0.138 -0.141 -0.538 9 -0.144 0.14 -0.14 -0.55 10 -0.142 0.142 -0.138 -0.56 Source: own calculations. Taking into account the two subsamples to distinguish between cycle phases, estimation results show that in recessions government spending multiplier is positive while in expansions is smaller, inclusively, negative (impact multipliers of 0.037 and - 0.167, respectively; cumulative multipliers of 0.142 and -0.142, respectively). This smaller value in expansions is in line with theoretical literature and empirical evidence. Relatively to tax multiplier, the effectiveness of this instrument is, also, higher in recessions: in both cumulative terms and on impact (impact multipliers are -0.15 in recessions, against -0.076 in expansions; cumulative multipliers are, respectively, of - 0.56 and -0.138). Apparently, in the Euro area as a whole, the larger values for fiscal multipliers during recessions suggest that fiscal policy purely targeted to stabilization concerns would, on average, produce a fiscal surplus bias. As expected, the same conclusions can also be drawn using disaggregate fiscal variables. Table 19 and Table 20 show estimation results and Figure 6 and Figure 7 show the corresponding accumulated responses of output in expansions and recessions, respectively, considering disaggregated fiscal data. 58 Table 19: Estimation results - disaggregated fiscal variables, expansions. 59 Table 20: Estimation results - disaggregated fiscal variables, recessions. 60 Figure 6: Accumulated impulse response functions of DELTA_GDP to shocks in disaggregated items of DELTA_SPENDING and DELTA_TAXES, expansions. -.03 -.02 -.01 .00 .01 .02 .03 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GTRA -.03 -.02 -.01 .00 .01 .02 .03 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GICO -.03 -.02 -.01 .00 .01 .02 .03 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GWAG -.03 -.02 -.01 .00 .01 .02 .03 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GINV -.03 -.02 -.01 .00 .01 .02 .03 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_SS -.03 -.02 -.01 .00 .01 .02 .03 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_TDIR -.03 -.02 -.01 .00 .01 .02 .03 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_TIND Accumulated Response to Cholesky One S.D. Innovations ± 2 S.E. 61 Figure 7: Accumulated impulse response functions of DELTA_GDP to shocks in disaggregated items of DELTA_SPENDING and DELTA_TAXES, recessions. Table 21 and Table 22 compare, respectively, disaggregated expenditure and revenue multipliers, under expansions and recessions. Table 21: Cumulative fiscal multipliers, expansions vs. recessions - disaggregated public spending. Period Expansions Recessions GTRA GICO GWAG GINV GTRA GICO GWAG GINV 1 -0.561 -0.04 -0.767 2.554 -0.118 0.43 0.317 0.232 2 -0.669 -0.348 -0.869 3.435 -0.243 0.465 1.047 -1.199 3 -0.586 0.072 -0.876 3.792 -0.063 0.426 1.984 -1.697 4 -0.472 0.119 -0.969 4.223 0.241 0.679 2.235 -2.055 5 -0.429 0.164 -1.047 4.601 0.428 0.834 2.433 -2.287 6 -0.409 0.138 -1.105 4.929 0.548 0.917 2.636 -2.442 7 -0.396 0.146 -1.128 5.17 0.643 0.981 2.813 -2.562 8 -0.38 0.158 -1.14 5.36 0.72 1.04 2.945 -2.65 9 -0.364 0.172 -1.146 5.507 0.779 1.09 3.039 -2.716 10 -0.352 0.182 -1.153 5.63 0.822 1.125 3.112 -2.766 Source: own calculations. -.04 -.02 .00 .02 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GTRA -.04 -.02 .00 .02 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GICO -.04 -.02 .00 .02 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GWAG -.04 -.02 .00 .02 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_GINV -.04 -.02 .00 .02 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_SS -.04 -.02 .00 .02 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_TDIR -.04 -.02 .00 .02 .04 1 2 3 4 5 6 7 8 9 10 Accumulated Response of DELTA_GDP to DELTA_TIND Accumulated Response to Cholesky One S.D. Innovations ± 2 S.E. 62 Table 22: Cumulative fiscal multipliers, expansions vs. recessions - disaggregated taxes. Period Expansions Recessions SS TDIR TIND SS TDIR TIND 1 0 0 0 0 0 0 2 -0.688 -0.258 0.391 0.238 -0.479 -0.081 3 -1.316 -0.519 0.261 0.548 -0.775 -0.453 4 -1.637 -0.64 0.028 0.918 -1.029 -0.709 5 -1.823 -0.689 -0.119 1.268 -1.209 -0.896 6 -1.967 -0.712 -0.147 1.541 -1.366 -1.026 7 -2.101 -0.738 -0.135 1.738 -1.493 -1.126 8 -2.219 -0.765 -0.122 1.885 -1.586 -1.206 9 -2.314 -0.787 -0.119 2 -1.655 -1.266 10 -2.389 -0.804 -0.119 2.088 -1.708 -1.311 Source: own calculations. In disaggregated terms, as far as spending side is concerned, most of the multipliers are in line to what happens in aggregated terms. A single counterintuitive result emerges regarding the GINV multiplier: in expansions, the value of the multiplier is larger than in recessions, reaching 5.63 in the positive phases of the cycle while being even negative in downturns (-2.766). Apparently, during recessions, public investment policies fail to stimulate output, possibly because they are of less productive nature, or even harmful. On the revenue it is important to notice that, in cumulative terms, in recessions, the GDP response to a change in direct taxes is approximately the double than that occurring in expansions (-1.708 vs. -0.804) being this effect even larger in the case of indirect taxes which values are of -1.311 in recessions vs. -0.119 in expansions. However, evidence shows that changes in social security contributions work strongly pro-cyclically: in recessions, multipliers range from 0.238 to 2.088 while, under expansions, a 1 euro rise in SS leads to a decrease in output of 2.389 euros in cumulative terms. One possible explanation for the stronger negative impact during expansions may be the non-linear behavior of labor supply, exhibiting higher wage elasticity in low-unemployment times. 63 4.3. Multipliers during debt-consolidation periods A last empirical question that we aim to answer is how structural breaks, such as fiscal consolidation efforts, are meaningful for the effectiveness of fiscal policy. In this regard, we have proceeded in a similar way as with the control for openness. We have included in the VAR a dummy variable, as exogenous. The dummy takes the value of 1 if the period is considered as one under the effort of a fiscal consolidation process, and is recorded zero otherwise. The assumption is that consolidation is a particular structural environment that may affect, relying on literature, the multipliers. Classification of periods as of consolidation draws on the European Commission (2007) criteria as described in section 3, above. Table 23 shows the estimation results for the VAR using aggregate fiscal data and controlling for consolidation periods. Corresponding fiscal multipliers are presented in Table 24. 70 which the output gap is positive (expansions) and the other comprising periods of negative output gap (recessions). Estimation results show that public spending multiplier is positive in recessions while in expansions is smaller, inclusively, negative. Similarly, the effectiveness of the taxes multiplier is, also, higher in recessions. Last but not the least, controlling for consolidation periods, the size of overall public spending and taxes multipliers is higher than in baseline scenario – consolidation phases affect negatively the size of the multipliers. Moreover, cuts in spending appear to be the most adequate instrument for achieving a fiscal consolidation with substantially smaller impacts on output than in times of no-consolidation. Looking at disaggregated fiscal variables, a reduction in indirect taxes should be the measure most recommended to policymakers, on the revenue side. For all the remaining revenue items, the impact on output is estimated to be higher during consolidations. On the spending side, reducing transfers and intermediate consumption has stronger impacts on output during consolidations, while reducing public investment or wages has lower impacts on output during consolidation processes. Reducing public wages exhibits even stronger nonKeynesian effects during consolidation periods. We are aware, however, that these results should be carefully interpreted because of both technical and theoretical limitations of the analysis. On the one hand, we have chosen a parsimonious VAR model with recursive formulation to identify structural shocks; the other strategies, in spite of being more appealing, would have been substantially more time-consuming and would, in certain cases, further reduce the already small number of observations in the sample. Moreover, regime-switching VAR models that allow for smooth transitions across states would probably be more suitable for computing multipliers in recessions and expansions (e.g., Auerbach and Gorodnichenko, 2012). From a theoretical point of view, analyzing fiscal shocks in the context of an international panel entails several drawbacks. Formally, fiscal policy in the Euro area is country-specific; thus, fiscal policy mechanisms operate with different intensity in different countries, fiscal policy authorities have different preferences on fiscal policy mixes, on debt stabilization relative to cycle stabilization and policy responses are not immune to international policy spillovers. For instance, Favero et al. (2011) found that a fiscal consolidation has different impacts whether it is implemented by a single country or contemporaneously by a set of highly integrated countries. The 71 analysis of fiscal multipliers is probably more useful for policy makers if it relies only on observations for a specific country, although of difficult implementation for all the European countries, due to sparse data since the Euro involvement. 72 Appendix Keynesian spending and tax multipliers (closed economy)                               Spending Multiplier     Tax Multiplier     If  , in modulus, Spending Multiplier > Tax Multiplier, since 0<c<1, | | | | c.q.d. 73 where Gross Domestic Product; Planned Expenditure Autonomous Aggregate Expenditure; Private Consumption; Government Spending; Planned Private Investment; Marginal Propensity to Consume; = Income Taxes; t = Income Tax Rate; R = Transfers from Government to private sector. 74 References Afonso, Antonio and Ricardo M. Sousa (2012), “The Macroeconomic Effects of Fiscal Policy”, Applied Economics, Vol. 44, No. 34, pp. 4439-4454. Agnello, Luca, Davide Furceri and Ricardo M. Sousa (2011), “Fiscal Policy Discretion, Private Spending, and Crisis Episodes”, Núcleo de Investigação em Políticas Económicas – Universidade do Minho, NIPE Working Papers, No. 31. Alesina, Alberto, Carlo Favero and Francesco Giavazzi (2012), “The Output Effect of Fiscal Consolidations”, National Bureau of Economic Research, NBER Working Papers, No. 18336, August. Alesina, Alberto and Silvia Ardagna (2010), “Large Changes in Fiscal Policy: Taxes versus Spending”, National Bureau of Economic Research, NBER Chapters, in: Tax Policy and the Economy, Vol. 24, pp. 35-68. Alesina, Alberto, Silvia Ardagna, Roberto Perotti and Fabio Schiantarelli (2002), “Fiscal Policy, Profits and Investment", American Economic Review, Vol. 92, No. 3, pp. 571-589. Ardagna, Silvia (2007), “Fiscal Policy in Unionized Labor Markets”, Journal of Economic Dynamics & Control, vol. 31, pp. 1498-1534. Aschauer, David A. (1989), “ Is Public Expenditure Productive”, Journal of Monetary Economics, Vol. 23, No. 2, pp. 177-200. Auerbach, Alan J. and Yuriy Gorodnichenko (2012), “Fiscal Multipliers in Recession and Expansion”, National Bureau of Economic Research, NBER Chapters, in: Fiscal Policy after the Financial Crisis. 75 Bachmann, Rudiger and Eric R. Sims (2011), “Confidence and the Transmission of Government Spending Shocks”, National Bureau of Economic Research, NBER Working Papers, No. 17063, May. Barsky, Robert B. and Eric R. Sims (2012), “Information, Animal Spirits, and the Meaning of Innovations in Consumer Confidence.”, American Economic Review, Vol. 102, No. 4, pp. 1343-1377. Barro, Robert (1974), “Are Government Bonds Net Wealth?”, University of Chicago Press, Journal of Political Economy, Vol. 82, No. 6, pp. 1095-1117. Baxter, Marianne and Robert G. King (1993), “Fiscal Policy in General Equilibrium”, American Economic Review, Vol. 83, No. 3 pp. 315-334. Blanchard, Olivier and Roberto Perotti (2002), “An Empirical Characterization of the Dynamic Effects of Changes in Government Spending and Taxes on Output”, MIT Press, The Quarterly Journal of Economics”, Vol. 117, No. 4, pp. 1329-1368. Born, Benjamin, Falko Juessen and Gernot J. Muller (2012), “Exchange Rate Regimes and Fiscal Multipliers”, Centre for Economic Policy Research, CEPR Discussion Papers, No. 8986. Bouakez, Hafedh, Foued Chihi and Michel Normandin (2010), “Measuring the Effects of Fiscal Policy”, Centre Interuniversitaire sur le Risque, les Politiques Économiques et l’ Employ, CIRPÉE Working Paper, No. 16, May. Briotti, Maria Gabriella (2005), “Economic Reactions to Public Finance Consolidation: A Survey of the Literature”, European Central Bank, Occasional Papers, No. 38, October. Burda, Michael and Charles Wyplosz (2009), Macroeconomics – a European Text, 5th edition, Oxford University Press. 76 Burriel, Pablo, Francisco de Castro, Daniel Garrote, Esther Gordo, Joan Paredes and Javier J. Pérez (2010), “Fiscal Policy Shocks in the Euro Area and the US – An Empirical Assessment”, Institute for Fiscal Studies, Fiscal Studies, Vol. 31, No. 2, pp. 251-285. Canzoneri, Matthew, Fabrice Collard, Harris Dellas and Behzad Diba (2012), “Fiscal Multipliers in Recessions”, Department of Economics – Universitaet Bern, Discussion Papers, No. 12–04, May. Christiano, Lawrence, Martin Eichenbaum and Sergio Rebelo (2011), “When is the Government Spending Multiplier Large?”, University of Chicago Press, Journal of Political Economy, Vol. 119, No. 1, pp. 78-121. Cogan, John F., Tobias Cwik, John B. Taylor and Volker Wieland (2010), “New Keynesian versus Old Keynesian Government Spending Multipliers”, Journal of Economic Dynamics & Control, Vol. 34, No. 3, pp. 281-295. Corsetti, Giancarlo and Gernot J. Mueller (2011), “Multilateral Economic Cooperation and the International Transmission of Fiscal Policy”, National Bureau of Economic Research, NBER Chapters in: “Globalization in an Age of Crisis: Multilateral Economic Cooperation in the Twenty-First Century. Corsetti, Giancarlo, Keith Kuester, Andre Meier and Gernot J. Muller (2012a), “Sovereign Risk, Fiscal Policy, and Macroeconomic Stability”, International Monetary Fund, IMF Working Papers, No. 33, January. Corsetti, Giancarlo, Andre Meier and Gernot J. Muller (2012b), “What Determines Government Spending Multipliers?”, Centre for Economic Policy Research, CEPR Discussion Papers, No. 9010, March. 77 Cwik, Tobias and Volker Wieland (2011), “Keynesian Government Spending Multipliers and Spillovers in the Euro Area”, Economic Policy, Vol. 26, No. 67, pp. 493-549. Eggertsson, Gauti B. (2011), “What Fiscal Policy is Effective at Zero Interest Rates?”, National Bureau of Economic Research, NBER Chapters, in: NBER Macroeconomics Annual 2010, Vol. 25, pp. 59-112. European Central Bank (2012), Monthly Bulletin - Euro Area Statistics Online, September 2012 European Commission (2012), “Report on Public Finances in EMU 2012”, European Economy Series, April. European Commission (2007), “Public Finances in EMU-2007”, European Economy Series, June Fatás, Antonio and Ilian Mihov (2001), “The Effects of Fiscal Policy on Consumption and Employment: Theory and Evidence”, Centre for Economic Policy Research, CEPR Discussion Papers, No. 2760, April Favero, Carlo and Francesco Giavazzi (2007), “Debt and the Effects of Fiscal Policy”, Centre for Economic Policy Research, CEPR Discussion Papers, No. 6092, February Favero, Carlo, Francesco Giavazzi and Jacopo Perego (2011), “Country Heterogeneity and the International Evidence on the Effects of Fiscal Policy”, International Monetary Fund, IMF Economic Review, Vol. 59, No. 4, pp. 652-682. Giavazzi, Francesco and Marco Pagano (1990), “Can Severe Fiscal Contractions Be Expansionary? Tales of Two Small European Countries”, National Bureau of Economic Research, NBER Chapters, in: NBER Macroeconomics Annual 1990, Vol. 5, pp. 75122. 78 Gordon, Robert J. (2012), Macroeconomics, 12th edition, Pearson, Addison-Wesley. Hebous, Shafik (2011), “The Effects of Discretionary Fiscal Policy on Macroeconomic Aggregates: A Reappraisal”, Journal of Economic Surveys, Vol. 25, No. 4, pp. 674-707. Ilzetzki, Ethan, Enrique G. Mendoza and Carlos A. Végh (2011), “How Big (Small?) are Fiscal Multipliers?”, International Monetary Fund, IMF Working Papers, No. 52, March. Laxton, Douglas and Paolo Pesenti (2003), “Monetary rules for Small Open, Emerging Economies”, Journal of Monetary Economics, Vol. 50, No. 5, pp. 1109-1146 Leeper, Eric M., (2010), “Monetary Science, Fiscal Alchemy”, National Bureau of Economic Research, NBER Working Papers, No. 16510, October Leeper, Eric M., Nora Traum and Todd B. Walker (2011), “Clearing Up the Fiscal Multiplier Morass”, National Bureau of Economic Research, NBER Working Papers, No. 17444, September. Leeper, Eric M., Todd B. Walker and Shu-Chun Susan Yang (2008), “Fiscal Foresight: Analytics and Econometrics”, National Bureau of Economic Research, NBER Working Papers, No. 14028, May. Michaillat, Pascal, (2012), “A Theory of Countercyclical Government-Consumption Multiplier”, Centre for Economic Policy Research, CEPR Discussion Papers, No. 9052, June. Mountford, Andrew and Harald Uhlig (2009), “What Are the Effects of Fiscal Policy Shocks?”, Journal of Applied Econometrics, Vol. 24, No. 6, pp. 960-992. 79 OECD (2010), “OECD Economic Outlook”, Volume 2, Chapter 4 “Fiscal Consolidation: Requirements, Timing, Instruments and Institutional Arrangements”. Paredes, J., D. J. Pedregal and J. J. Pérez (2009), “A Quarterly Fiscal Database for the Euro Area based on Intra-annual Fiscal Indicators”, ECB Working Paper Series, No. 1132, December. Pereira, A. M. and O. Roca-Sagalés (2011), “Long-Term Effects of Fiscal Policies in Portugal”, Journal of Economic Studies, Vol. 38, No. 1, pp. 114-127. Perotti, Roberto (2005), “Estimating the Effects of Fiscal Policy in OECD Countries”, Centre for Economic Policy Research, CEPR Discussion Papers, No. 4842, January. Perotti, Roberto (2011), “ The “Austerity Myth”: Gain Without Pain”, National Bureau of Economic Research, NBER Working Papers, No. 17571, November. Ramey, Valerie A. (2011), “Identifying Government Spending Shocks: It’s All in the Timing”, The Quarterly Journal of Economics, Vol. 126, No. 1, pp. 1-50. Ramey, Valerie A. and Matthew D. Shapiro (1999), “Costly Capital Reallocation and the Effects of Government Spending”, Carnegie-Rochester Conference Series on Public Policy, Vol. 48, No. 1, pp. 145-194 Ratto, Marco, Werner Roeger and Jan in’t Veld (2009), “QUEST III: An Estimated Open-Economy DSGE Model of the Euro Area with Fiscal and Monetary Policy”, Economic Modelling, Vol. 26, No. 1, pp. 222-233 Romer, Christina D. and David H. Romer (2010), “The Macroeconomic Effects of Tax Changes: Estimates Based on a New Measure of Fiscal Shocks”, American Economic Review, Vol.100, No. 3, pp. 763-801.