Fiscal policy and private investment in Greece

International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Contents lists available at ScienceDirect

International Economics journal homepage: www.elsevier.com/locate/inteco

Fiscal policy and private investment in Greece Maria Th. Kasselaki, Athanasios O. Tagkalakis n Economic Analysis and Research Department, Bank of Greece, 21 E. Venizelos Ave., 10250 Athens, Greece

a r t i c l e i n f o

JEL classification: E62 E22 E44 O52 Keywords: Fiscal consolidation Investment Output Financial markets Economic sentiment

abstract This paper investigates the effects of fiscal policy on private nonresidential investment and output in Greece. Besides examining the direct effects of fiscal consolidation, we investigate the role of financial markets and economic sentiment in the transmission of fiscal policy shocks. A tax based fiscal consolidation has more pronounced and more protracted negative effects on output and private non-residential investment relative to an expenditure based fiscal consolidation. A government spending-based fiscal consolidation improves financial markets and boosts economic sentiment. This in turn mitigates the direct negative effects of fiscal consolidation on private investment and output leading to a more rapid recovery. On the other hand, a tax hike fails to induce this positive confidence effect magnifying the negative effects of fiscal adjustment. & 2016 CEPII (Centre d’Etudes Prospectives et d’Informations Internationales), a center for research and expertise on the world economy. Published by Elsevier B.V. All rights reserved.

1. Introduction After the loss of market access in early 2010 Greece applied for EU-IMF assistance. The financial assistance was provided in exchange for the implementation of an ambitious fiscal consolidation and structural reform programme, i.e., the Economic Adjustment Programme (EAP) for Greece which covered the period from May 2010 till the end of June 2015. Since August 2015 Greece has obtained a new three-year financial assistance programme from the European Stability Mechanism (ESM). The goals of these financial assistance programmes, besides covering Greece’s financing needs, are to correct fiscal imbalances, and to facilitate the rebalancing of the economy from the non-tradable to n

Corresponding author. Tel.: þ 30 210 3202442; fax: þ30 210 3232025. E-mail addresses: [email protected] (M.Th. Kasselaki), [email protected] (A.O. Tagkalakis).

http://dx.doi.org/10.1016/j.inteco.2016.03.003 2110-7017/& 2016 CEPII (Centre d’Etudes Prospectives et d’Informations Internationales), a center for research and expertise on the world economy. Published by Elsevier B.V. All rights reserved.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

2

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

the tradable sector by means of structural reforms. The correction of internal and external imbalances is expected to boost market confidence leading to a rapid and sustained increase in private nonresidential investment alongside net exports. The substantial progress achieved so far in correcting fiscal imbalances has come at a huge cost in terms of output loss and rising unemployment.1 The cumulative output loss is close to 25% in the period 2009–2014, whereas unemployment has skyrocketed to 26%, with youth unemployment being close to 55–60%. Forecasts at the start of the fiscal consolidation program in 2010 were pointing to a shallow recession and a much more rapid recovery in 2012.2 However, things turn out to be worse than expected; the recession was much deeper and more protracted.3 A mild recovery (0.8%) was achieved in 2014 but a new recession is projected for 2015. This dramatic output contraction reflects, among other things, extreme developments in private investment, which from about 20.4% of GDP in the period 2000–2009 declined to 9.7% of GDP in 2014. This development primarily reflects the dramatic contraction of the construction sector and to a lesser extent the fall in private non-residential investment. Residential investment declined to 1.0% of GDP in 2014 from an average of 9.0% of GDP in the 2000–2009 period. Private non-residential investment declined to a bit less than 8.6% of GDP in 2014 from an average of 11.4% of GDP in the period 2000–2009. Taking into account that one of the goals of both the EU-IMF financial assistance and the EMS programme is the gradual rebalancing of economic activity from the non-tradable to the tradable sector, the downsizing of the residential investment and consequently of the construction sector is not necessarily that much harmful for long term growth. However, the continued decline in private non-residential investment is detrimental for long term growth prospects. In view of the recent adverse effects of fiscal consolidation on economic activity, and building on the SVAR approach of Blanchard and Perotti (2002) and Perotti (2005) this paper investigates the effects that fiscal policy changes have on private non-residential investment and output growth in Greece. Besides examining the direct effects on fiscal policy on private investment and output, building on Ardagna (2010), Laubach (2009), Schuknecht et al. (2010), Barrell et al. (2012) and Bachmann and Sims (2012) we investigate the role of financial markets and economic sentiment in the transmission of fiscal policy shocks. We investigate the widely held view (incorporated in the EAP) that improvements in fiscal conditions will generate a positive financial markets and confidence response, mitigating (or even negating) the direct negative effects of fiscal consolidation on output. Ardagna (2010) has shown that financial markets respond positively to large fiscal consolidations. Other studies (e.g., Arghyrou and Kontonikas (2011), Attinasi et al. (2009), Barrell et al. (2012), Bernoth and Erdogan (2012), De Grauwe and Ji (2012), Schuknecht et al. (2010), and Laubach (2009)) have shown that government borrowing premia and long term interest rates increase following debt expansions. Konstantinou and Tagkalakis (2011) have shown that fiscal policy can affect consumer and business confidence. On the other hand, as pointed out in Barsky and Sims (2012) surprise changes in consumer confidence are associated with long lasting movements in macroeconomic aggregates, i.e., confidence measures are reflecting changes in future fundamentals. Bachmann and Sims (2012) show that in standard SVARs consumer and business confidence does not react significantly to government spending shocks. However, in non-linear VARs confidence rises following an increase in government spending leading to higher output multipliers in particular in recession times (as in Auerbach and Gorodnichenko (2012)).4 1 In the context of an EU-IMF financed programme Greece achieved a primary surplus of 1.2% of GDP in 2013 and was in primary balance of about 0% in 2014 from a primary deficit of 10.4% of GDP in 2009. 2 According to the first EU-IMF programme output growth was expected to deteriorate until 2011 and return to positive territory in 2012, i.e., the growth rates were expected to be -2.0% 2009 Greece ,  4.0% in 2010,  2.6% in 2011 and 1.1% in 2012 (European Commission, 2010). 3 According to European Commission (2015) the decline in real GDP was 0.4% in 2008, 4.4% in 2009 5.8% in 2010, 8.9% in 2011, 6.6% in 2012 and, 3.9% in 2013. 4 This important confidence effect is explained as reflecting future productivity improvements driven by government spending (in particular on investment) shocks in downturns, i.e., the confidence effect reflects beliefs about long term fundamentals (as in the “news” literature) and not pure “sentiment” (i.e. movement in confidence unrelated to fundamentals).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

3

According to our findings a tax based fiscal consolidation has more pronounced and more protracted negative effects on output and private non-residential investment relative to an expenditure based fiscal consolidation. In addition, increases in direct household taxes, in other indirect taxes and cuts in the government wage bill and in government investment followed by increases in direct business taxes have the most sizeable and long lasting negative effects on output. Nevertheless, a government spending-based fiscal consolidation (in particular focusing on current transfers and the wage bill) improves financial markets (i.e., reduces government bond spreads and increases share prices) and boosts economic sentiment. This in turn mitigates the direct negative effects of fiscal consolidation on private investment and output growth inducing a more rapid recovery. On the other hand, a tax hike (in particular increases in direct business taxes and indirect taxes) fails to induce this positive confidence effect, which in turn magnifies the negative effects of fiscal consolidation on private investment and output. As part of robustness exercise we examined the effects of fiscal policy under different assumptions for the output elasticity of taxes as well as the share price elasticity of taxes and we presented some preliminary findings for the effects of fiscal policy in periods of expansions and contractions. The remainder of the paper is organised as follows. The next section reviews briefly the relevant literature on the effects of fiscal policy on private investment. In Section 3 we present data information and discuss in more detail the econometric methodology. Section 4 presents the main empirical findings. In Section 5 we consider the effects of individual government spending and revenue components and present the relevant empirical findings. In Section 6 we present various robustness exercises. The last section includes a brief summary of the results and concluding remarks. The Appendix provides additional data information.5

2. Relevant literature The typical Real Business Cycle (RBC) model predicts that a government spending shock (financed by future lump-sum taxes) will decrease consumption and will increase employment (through the negative wealth effect), which, in turn will raise the return to capital and will boost investment. The Keynesian model predicts that private consumption will increase after a government spending shock financed by future lump-sum taxes, because disposable income increases. Investment would be crowded out if the increase in consumption could raise the interest rate. However, given that the short term interest rate is a policy variable controlled and set exogenously by the Central Bank (for example along the lines of a Taylor rule) an automatic rise in the interest following an increase in consumption is hardly the case in real life. Therefore, if the central bank holds the interest rate steady in the face of the increase in government spending, the implied effect on investment is nil. Nevertheless, the financing of the expenditure increase matters as well. For example, investment could decline if the expenditure increase is accompanied by a (corporate income) tax hike. Moreover, the reduction in private investment could be due to debt financed government expenditure which implies higher government borrowing and higher long term interest rates. As pointed out by Alesina et al. (2002) there is also a cost or labour market channel that could depress investment. More specifically, increases in public wages, public employment and government transfers increase the wage pressures in the private sector, both in unionized and competitive labour markets. This reduces profits and private investment. Alesina et al. (2002) show that taxes (in particular labour taxes) lower investment, however, the effects of government spending on investment are larger than those of taxes. A recent study by Traum and Yang (forthcoming) building on the New Keynesian framework shows that private investment is crowded-in in the short term when a debt expansion is triggered by cutting capital tax rates or increasing productive government investment, because they both raise the net return to capital. In addition, at longer horizons distortionary financing of debt expansions leads 5 A Supplementary material appendix provides additional information on various econometric tests (e.g. unit root tests, cointegration tests etc.).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

4

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

to a negative response of investment to debt. Increases in government consumption and government transfers lead to higher real interest rates and increase future fiscal adjustment needs crowding out private investment. Forni et al. (2009) examine the economic effects of fiscal policy using an estimated dynamic stochastic general equilibrium model for the Euro area with both Ricardian and non-Ricardian agents. They show that positive shocks in government purchases, government employment and government transfers increases aggregate private consumption, while investment drops due to the increase in the rental rate of capital. Similarly, Coenen and Straub (2005) estimating a New-Keynesian DSGE model of the euro area featuring non-Ricardian households show that private investment responds negatively following an expansionary government spending shock. Moreover, Forni et al. (2009) show that a reduction in capital income tax rate leads to reallocation from labour to capital (whose utilization spikes) and favours investment and output in the medium term. A reduction in labour income tax rate leads to an outward shift of labour supply and to an increase in non-Ricardian disposable income and consumption. Output and employment increase, while investment increases mildly due to the improvement in the marginal product of capital. A decrease in consumption tax rate lowers inflation inducing a decrease in the real interest rate. Consumption of the cheaper goods basket substantially increases more and faster for non-Ricardians than for Ricardians. Firms increase output to meet the additional demand and they do so by increasing employment and capital utilization (investment has not become cheaper as it is not subject to consumption taxes). The smoother increase in Ricardian consumption gradually shifts resources away from investment. A number of recent studies have relied on the SVAR methodology to investigate the effects of fiscal policy on private investment in the US. In most cases the empirical evidence is in favour of the crowding out effect. Blanchard and Perotti (2002) identify fiscal shocks based on the assumption that discretionary fiscal policy does not respond to output within the quarter, while also estimating elasticities of fiscal variables with respect to some macroeconomic variables in order to net out the automatic effects of fiscal policy. They show that an increase in government spending stimulates economic activity but decreases investment (as was also the case in Galí et al. (2007)). Mountford and Uhlig (2009) find a weak positive effect on output while both residential and non-residential investment is crowded out. The response of investment to a government spending shock is found to be insignificant in Fatás and Mihov (2001). Perotti (2005) studies the effect of fiscal policy on economic activity and its sub-components in 5 OECD countries (Australia, Canada, Germany, UK and the US). Considering the pre- and post-1980 periods he finds significant evidence of a negative response of private investment following a government spending shock in the post 1980 period in all countries considered (except in Australia, where he finds positive response in the pre-1980 period). When accounting for possible negative correlation between government purchases and private sector inventories, he shows that the private investment response to a government spending shock was positive (or 0) in the pre-1980s, while it remained negative in the post 1980 period. Following a tax cut private investment responds positively and significantly 3 years after the shock in the US and the UK (in the pre-1980 period), while it is basically zero in the other 3 countries. The response of private investment turns from positive to negative in the post 1980 period in the US, the UK and Australia (in line with their GDP response). Burriel et al. (2009) find that private investment declines in response to higher government spending or net taxes in the US, whereas in the EMU only tax increases seem to entail a negative reaction of private investment. Afonso and Sousa (2011) show that a government spending shock crowds out investment in Portugal. Bouthevillain and Dufrenot (2011) studying the case of France show that government spending increases crowds out business investment, while cuts in corporate income taxes improves business investment in particular in periods of expansion. However, other studies have found a positive response of private investment following a government purchases shock; e.g., Biau and Girard (2005) in case of France and Giordano et al. (2007) in case of on Italy. Whereas, some other studies have shown that investment could even increase following a positive tax shock. Kuismanen and Kämppi (2010) find that a positive tax shock in Finland (or a policy that increases public sector revenues) has a positive effect on investment. Similarly Afonso and Sousa (2011) show that a positive government revenue shock leads to a positive impact effect on investment. Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

7.000

1.500

6.000

1.000

5.000

5

0.500

4.000

0.000

3.000 -0.500

2.000

log(SP_GR)

1.000

log(SP_US)

-1.000

log(SP_GR_net of world developments) (RHS)

-1.500 2014q1

2013q2

2012q3

2011q4

2011q1

2009q3

2010q2

2008q4

2007q2

2008q1

2006q3

2005q4

2005q1

2003q3

2004q2

2002q4

2002q1

2001q2

2000q3

1999q1

1999q4

0.000

Fig. 1. Logarithm of share prices in Greece and the US.

Several studies rely on the so-called “narrative” approach of fiscal policy of Ramey and Shapiro (1998), Edelberg et al. (1999), Burnside et al. (2004), Romer and Romer (2010) and Ramey (2011). Fiscal policy shocks are identified based either on the military build up of the Korean and Vietnam wars (defence budget expansions) or on announced and implemented changes in tax legislation (“narrative” of Presidential speeches and Congressional records). Burnside et al. (2004) and Edelberg et al. (1999) find a mostly positive private investment response following a government spending shock.

3. Data information and SVAR methodology We use quarterly data from 2000 Q1 to 2014 Q1.6 Data were taken from the International Financial Statistics of the IMF (2014) and the OECD Economic Outlook (OECD, 2014); see Appendix. In view of the small sample size we consider a parsimonious specification which is a variant of those used in Blanchard and Perotti (2002) and Perotti (2005). In order to examine the effects of government spending and tax shocks we consider a 5-variableSVAR: the change in log (Δlog) of real total government spending excluding interest payments and other capital transfers, the change in log (Δlog) of real tax burden (direct, indirect taxes, social security contributions and capital taxes), the change in log (Δlog) of real share prices, the change in log (Δlog) of real GDP, and the change in log (Δlog) of real private non-residential investment.7 Following Perotti (2005) private investment is ordered after GDP. Building on Ardagna (2010) we use share price movements as a proxy of the response of financial markets to fiscal policy changes. However, the share prices movements of the Athens Stock Exchange reflect not only domestic but also international developments. To control for this effects we regress the (log of the) Greek stock market index on the (log of the) US stock market index which captures world wide developments and international investors' responses. The (first difference of the) residual 6 We use ESA 1995 quarterly data for the period 2000 Q1–2014 Q1 in order to focus on the period that Greece is part of the euro area. Greece became part of the euro area on 1st January 2001 but its euro entry was already decided in 2000, therefore we start our data set in 2000 because expectations that euro area entry would occur had already been formed. Another reason to look at the post-2000 era relates to the fact it is only since 2000 that the statistical authorities of Greece have started the production and dissemination of quarterly non-interpolated fiscal and macroeconomic data. It should be noted that all data have been approved by Eurostat, i.e., the data used in the analysis are not subject to any statistical deficiencies Nevertheless, it should be pointed out that since autumn 2014 the Greek statistical authorities (and Eurostat) are producing ESA 2010 fiscal and national account data. However, the ESA 2010 quarterly fiscal series that are currently available start in 2006 but run until 2015 Q1. Given the lack of sufficient ESA 2010 data to conduct the analysis we are constrained to rely on ESA 1995 macroeconomic and fiscal data. 7 Unit root and cointegration tests that are reported in the Supplementary material appendix justify the choice of a SVAR model in log-differences.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

6

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.4 0.2 0 -0.2 Dlog(SP_GR) -0.4 Dlog(SP_US) -0.6

2013q3

2012q4

2011q2

2012q1

2010q3

2009q4

2008q2

2009q1

2007q3

2006q1

2006q4

2005q2

2004q3

2003q4

2002q2

2003q1

2000q1

2000q4

1999q2

2001q3

Dlog(SP_GR_net of world developments)

-0.8

Fig. 2. ΔLog of share prices in Greece and the US.

from this regression is then employed in SVAR. It reflects share price movements net of any international economic and financial developments. This variable better captures the response of financial market to domestic fiscal policy changes (see Figs. 1 and 2). All variables are transformed in real terms using the GDP deflator. The lagged value of the public debt to GDP ratio is included as an exogenous variable to capture the constraints imposed on fiscal policy by debt developments in line with Favero and Giavazzi (2007). In addition, we control for a series of policy changes and s undertaken in recent years by incorporating the shift dummy variable EAP taking values 1 from 2010 Q1 onwards and zero otherwise; EAP stands for Economic Adjustment Programme.8 The SVAR specification includes an intercept and the lag length is set to 1.9 The SVAR we estimate is of the form: X t ¼ A1 X t1 þ Ct þ B  Dt1 þut

ð1Þ

where Xt ¼[G, T, SP, Y, I] is the vector of endogenous variables, G stands for government spending (excluding interest payments and other capital transfers), T stands for tax burden (which is the sum of direct, indirect, capital taxes and social security contributions), SP stands for the share price variable we described above, Y stands for output and I stands for private non-residential investment. Ct contains the deterministic terms and Dt  1 is the debt to GDP ratio. ut are the VAR innovations. Building on the Blanchard and Perotti (2002) SVAR approach we identify the structural shocks to G and T by imposing on the matrices A and B that determine the mapping from the VAR innovations u to the structural shocks ε (Aut ¼Bεt) the following restrictions:

8 The EU-IMF financed Economic Adjustment Programme for Greece was initiated in May 2010, but a number of fiscal measures were already taken in early 2010. That is why EAP takes value 1 from 2010 Q1 onwards. The inclusion of this dummy is also in line with the unit root and cointegration tests presented in the Supplementary material appendix. 9 The lag length was chosen based on the information provided by relevant lag-length criteria, LM test on no autocorrelation, and the need to work with a parsimonious specification in view of the small sample size (see Supplementary material appendix).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

7

Employing information from Price et al. (2014) we set αgy ¼0.07 and αty ¼ 0.94.10 In addition, we set αti ¼ 0 and αgi ¼ 0. Hence, we assume that neither government spending nor government revenue respond on impact to contemporaneous changes in private non-residential investment. On the other hand, there is a contemporaneous change in both taxes and government spending (through the unemployment benefit component) following changes in real GDP. Moreover, we assume that: αgsp ¼ αtsp ¼0, i.e., there is no contemporaneous (automatic) response in government spending and taxes following a change in share prices. This assumption might be relatively strong given that some tax revenue components are affected contemporaneously by stock market movements; we shall relax this assumption in Section 6.1. In line with Blanchard and Perotti (2002) we set β12 ¼0 and estimate β21.11 The above specification is used to examine the effect of a government spending and a tax burden shock on output and private investment. Following Ardagna (2010) we will examine the role of financial markets in the transmission of fiscal policy shocks and their effects on output and private investment. Thus we will disentangle the effect of fiscal variables on output and investment through share price movements. If share prices respond on impact to fiscal shocks then α31 and α32 should be non zero. The coefficient α43 captures the response of output to share price movements. Hence α31*α43 captures the effect of government spending on output though the share price channel; α32*α43 captures the effect of tax burden on output though the share price channel. The coefficients α41 and α42 reflect the direct responses of output to spending and tax shocks, respectively. By setting α31 ¼ α32 ¼ 0 we assume that financial markets (share prices) does not react on impact to the spending and tax shocks. Moreover, by restricting the AR coefficients of lagged output, spending, tax burden and private non-residential investment in the share price equation to zero we impose the constraint that share prices does not respond to fiscal shocks at any horizon. This so-called “restricted share price” SVAR specification will only capture the direct effect of fiscal shocks on output and investment. The indirect effect of fiscal shocks on output and investment through financial markets (share price changes) is being isolated.12 We'll also examine the alternative “unrestricted share price” SVAR specification, where the abovementioned restrictions do not apply. As a next step, building on Bachmann and Sims (2012), we examine the role of economic sentiment indicator (Δlog of ESI) in the transmission of fiscal policy shocks, i.e., ESI is a used as a proxy of the expectations of economic agents. Economic sentiment is a composite indicator reflecting the outlook in the industrial, services, retail and construction sectors, as well as consumer confidence. Hence, ESI captures explicitly the expectations of real economy actors (involved in private investment decisions) regarding future economic prospects. On the contrary, share price movements are assumed to incorporate the views of financial markets on the long term prospects of the economy. Last but not least, we consider an alternative specification where we use the spread between the rate of return of the 10 year Greek Government Bond and the rate of return of the 10 year German “Bund”. This follows the work of Ardagna (2010), Laubach (2009), Schuknecht et al. (2010) and Barrell et al. (2012).13

10 According to Price et al. (2014) the output elasticity of personal income tax (PIT) is 2.22, the output elasticity of corporate income tax (CIT) is 1.90, the output elasticity of social security contributions is 0.58, and the output elasticity of indirect taxes is 1. Using these elasticities and the sample averages of the revenue shares of each tax category we set αty ¼0.94 in specification (2). Analogous elasticity estimates for PIT and CIT have been reported in Tagkalakis (2015a, 2015b).The elasticity αgy ¼ 0.07 is obtained as follows: we multiply the elasticity of unemployment with respect to the output gap (it is  3.15 from Price et al. (2014)) with the average share of unemployment spending on government spending over the sample which is about 2.3%. 11 We have also considered the opposite case as Blanchard and Perotti (2002) and Favero and Giavazzi (2012), but the results are invariant to the ordering because the correlation between shocks is low enough and insignificant. 12 A similar exercise was conducted in Bachmann and Sims (2012) with the focus being on the role of confidence (see footnote 5 in page 240) and in Tagkalakis (2014a) with the focus being on the role of credit. 13 We use the government bond spread instead of just the rate of return of the 10 year Greek government bond on the secondary market in order to account for Eurosystem monetary policy developments.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

8

0.05 0.00 0

2

4

6

8

10 12 14 16 18 20

-0.05 -0.10 -0.15 -0.20 -0.25

Investment response Investment response _restricted share prices 68% CI

Fig. 3. The response of private non-residential investment to a cut in government spending (share prices SVAR).

4. Main findings 4.1. The role of share prices The findings are presented in Fig. 3–8. The solid black line represents the response of the variable of interest to the fiscal policy shock in the unrestricted SVAR, the two round-dotted black lines are the 68% confidence bands, and the solid red line represents the response of the variable of interest to the fiscal policy shock in the restricted SVAR.14 Following Blanchard and Perotti (2002) the responses of output and investment to government spending and tax shocks are reported in terms of (non-accumulated) multipliers. The original impulse responses of the responding variable (e.g. output, investment) are divided by the impact response (at t¼ 0) of the government spending (tax) variable and the result is divided by the (sample average of the) ratio of government spending (tax) variable and the responding variable. The impulse responses measure the reaction of the responding variable in percent of real GDP to a government spending (tax) shock of 1% of real GDP. The responses of share prices and economic sentiment are in percentage changes and the response of the government bond spread is in percentage point. A cut of 1 percentage point (p.p.) of GDP in government spending decrease real GDP on impact by about 0.34 percentage points and private non-residential investment by about 0.15 p.p. of GDP. The impact responses in the restricted and unrestricted SVARs are of comparable magnitude, with the response of investment being slightly bigger in the restricted SVAR (0.165 pp; see Figs. 3 and 5). The cut in government spending induces a positive stock market response in the unrestricted SVAR (see Fig. 4).15 This implies that an expenditure-based fiscal consolidation can boost financial markets’ confidence as regards the success of fiscal consolidation and the outlook of the Greek economy. This finding is in line with the evidence reported in Ardagna (2010). Nevertheless, both investment and output decline following the government spending contraction. This is the so-called “accelerator” effect, i.e., a fall in real GDP signals lower profit, reduced sales and cash flow, and smaller use of existing capacity. Consequently profit expectations fall and business investment are discouraged. This positive correlation between government spending changes and private investment is in line with the finding reported by Biau and Girard (2005) for France and Giordano et al. Giordano et al., (2007) for Italy, as well as Burnside et al. (2004) and Edelberg et al. (1999) for the US, and Perotti (2005) for Australia (in the pre 1980 period). In the unrestricted share prices SVAR both private non-residential investment and output recover (or even become positive) after the fiscal policy shock at a much faster pace. Private investment 14 The confidence bands correspond to the unrestricted SVAR; standard errors have been calculated by bootstrapping the residuals (1.000 replications were performed). 15 The results are qualitatively similar if in the construction of the share price variable we use as proxy of international developments the German stock market index instead of the US stock market index.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4.00 Share prices response

3.50

Share prices response_restricted 68% CI

3.00 2.50 2.00 1.50 1.00 0.50 0.00

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

-0.50 Fig. 4. The response of share prices to a cut in government spending (share prices SVAR).

0.20 0.10 0.00 0

2

4

6

8 10 12 14 16 18 20

-0.10 -0.20

Output response

-0.30 -0.40

Output response_restricted share prices 68% CI

-0.50

68% CI

Fig. 5. The output response to a cut in government spending (share prices SVAR).

0.10 0.05 0.00 -0.05 0 -0.10 -0.15 -0.20 -0.25 -0.30 -0.35 -0.40 -0.45

2

4

6

8 10 12 14 16 18 20

Investment response Investment response _restricted 68% CI

Fig. 6. The response of private non-residential investment to an increase in tax burden (share prices SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

9

10

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

2.00 1.00 0.00 -1.00 0 2 4 6 8 10 12 14 16 18 20 -2.00

-5.00

Share prices response to a 1 p.p. of GDP tax burden shock Share prices response_restricted

-6.00

68% CI

-3.00 -4.00

-7.00 Fig. 7. The response of share prices to an increase in tax burden (share prices SVAR).

0.10 0.00 -0.10 0

2

4

6

8 10 12 14 16 18 20

-0.20 -0.30 Output response

-0.40 -0.50 -0.60

Output response_restricted share prices 68% CI

-0.70 Fig. 8. The output response to an increase in tax burden (share prices SVAR).

crosses the zero line at around the 5th quarter and then becomes marginally positive (see Fig. 3). On the contrary, when we shut down the financial markets’ channel private investment approaches the zero line at around the 10th quarter. The response of output crosses the zero line around the 3rd quarter and becomes positive thereafter in the unrestricted SVAR, whereas in the restricted SVAR it returns to zero line at around the 6th quarter (see Fig. 5). Overall, the response profile of private non-residential investment to a cut in government spending follows closely the response of output (in line with the “accelerator” model). Hence, there is no crowding-in following a government spending consolidation. Nevertheless, confidence effects through improvements in financial markets can support a more rapid recovery of economic activity and investment spending. In particular, the first year cumulative multiplier of output to a government spending cut is  0.46 in the unrestricted model while it reaches to  0.56 in the restricted one.16 As far as investment are concerned the cumulative multiplier of a government spending cut is  0.46 in the unrestricted and  0.53 in the restricted model.

16 First year cumulative multipliers are measured as the sum of the GDP (investment) response over the first four quarters divided by the sum of the corresponding spending or revenue response. The ratio is divided by the sample average of the shocked fiscal variable and the responding variable. We do not report cumulative multipliers for subsequent years given that point estimates in outer quarters are (in most cases) not statistically significant.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

11

0.05 0.00 -0.05

0

-0.10

2

4

6

8 10 12 14 16 18 20

Investment response

-0.15 -0.20

Investment response _restricted 68% CI

-0.25 Fig. 9. The response of private non-residential investment to cut in government spending (ESI SVAR).

Turning now to tax policy changes, a 1 p.p. of GDP increase in tax burden decreases real GDP on impact by about 0.42 (0.43) p.p. and private non-residential investment by about 0.27 (0.25) p.p. of GDP in the unrestricted (restricted) SVAR (see Figs. 6 and 8). Following a tax burden hike both private investment and output decline, with the fall being more pronounced and protracted in the unrestricted vis-à-vis the restricted SVAR. This is explained by the fact that share prices decline following a tax burden hike (see Fig. 7). Although a tax based fiscal consolidation is expected to improve fiscal balances it does not improve financial markets’ sentiment, either because it is not considered effective or because it is perceived as being more detrimental to growth from a private sector perspective. Therefore, in the unrestricted SVAR specification a tax burden hike reduces investment on account of both lower output (accelerator effect) and because of the deterioration in financial markets’ confidence. In particular, the first year cumulative multiplier of output to a tax hike is  1.33 in the restricted and 1.59 in the unrestricted case. As regards investment the cumulative multiplier of a tax hike is  0.28 in the restricted and 0.45 in the unrestricted model. The negative investment response to a tax burden hike is in line with the evidence reported in other studies such as Alesina et al. (2002), Perotti (2005) for the US and the UK (in the pre-1980 period) and Burriel et al. (2009) for the EMU countries and Bouthevillain and Dufrenot (2011) for France. 17 Overall, based on the evidence presented above the impact effect of a tax-based fiscal consolidation on output (  0.42 to 0.43) and investment ( 0.25 to  0.27) is more pronounced compared to a similar size spending-based fiscal consolidation (i.e.,  0.33 to  0.34 for output and 0.15 to  0.16 for investment). Additionally, based on the abovementioned first year cumulative multipliers, the negative effects of a tax burden hike on output and private investment are more protracted compared to the effects of a government expenditure shock. 4.2. The role of economic sentiment Following Bachmann and Sims (2012) we examine the role of economic sentiment indicator (ESI) in the transmission of fiscal policy shocks. The findings are reported in Figs. 9–11 and 12–14, respectively, for a cut in government spending and a tax burden hike. A government spending cut lowers economic sentiment on impact. Thereafter ESI improves up until the 4th quarter, but the point estimates are not statistically significant. Both output and investment decline as a consequence of the fiscal consolidation shock, with the economic sentiment having no particular effect on their impulse response profile (see Figs. 9–11). The impact multipliers following a government spending cut are 17 The impulse responses of the other variables in the SVAR are not shown here due to space limitations but can be found in the supplementary material appendix.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

12

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.60 0.40 0.20 0.00 0 2 4 6 8 10 12 14 16 18 20 -0.20 -0.40

ESI response ESI response_restricted 68% CI

-0.60

Fig. 10. The ESI response to a cut in government spending (ESI SVAR).

0.10 0.00 0 2 4 6 8 10 12 14 16 18 20 -0.10 -0.20

Output response

-0.30

Output response_restricted

-0.40

68% CI

-0.50 Fig. 11. The output response to an increase in tax burden (ESI SVAR).

0.35 for output and  0.14 for private investment. The first year cumulative multiplier of output to a government spending cut is about 0.56 while the one for investment is  0.43. A tax hike depresses economic sentiment (the point estimates of the impulse response are not statistically significant). Consequently, this translates into a relatively slower recovery of output and private non-residential investment in the unrestricted vis-à-vis the restricted SVAR (see Figs. 12–14). The impact multipliers following a tax burden hike are about  0.37 in case of output and  0.25 in case of private investment in both SVAR specifications. However, the first year cumulative multiplier of output following a tax hike is  0.98 in the restricted and  1.07 in the unrestricted specification. The cumulative multiplier of investment is 0.26 in the restricted and  0.31 in the unrestricted model. In line with the findings reported in the previous section the negative impact effects of a tax burden shock on output and private investment are more pronounced. In addition, a tax based fiscal consolidation delays the recovery due to its negative effect on economic sentiment. 4.3. The role of government bond spreads In this section we examine a variant of (1) which uses the spread between the rate of return of the 10 year Greek government bond and the rate of return of the 10 year German “Bund” as the third Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.15 0.10 0.05 0.00 -0.05 0 -0.10 -0.15 -0.20 -0.25 -0.30 -0.35 -0.40

2

4

6

13

8 10 12 14 16 18 20 Investment response Investment response _restricted 68% CI

Fig. 12. The response of private non-residential investment to an increase in tax burden (ESI SVAR).

0.80 0.60 0.40 0.20 0.00 -0.20 0 2 4 6 8 10 12 14 16 18 20 ESI response -0.40 -0.60

ESI

-0.80

response_restricted 68% CI

-1.00 Fig. 13. The ESI response to an increase in tax burden (ESI SVAR).

0.10 0.00 -0.10

0 2 4 6 8 10 12 14 16 18 20

-0.20 -0.30 Output response

-0.40 -0.50 -0.60

Output response_restricted 68% CI

Fig. 14. The output response to an increase in tax burden (ESI SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

14

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.05 0.00 -0.05 -0.10

0

2

4

6

8 10 12 14 16 18 20 Investment response

-0.15

68% CI

-0.20

Investment response_restricted

-0.25

Fig. 15. The response of private non-residential investment to a cut in government spending (spread SVAR).

0.20 0.10 0.00 -0.10 -0.20 -0.30 -0.40

0 2 4 6 8 10 12 14 16 18 20 Spread response Spread response_restricted 68% CI

-0.50 Fig. 16. The spread response to a cut in government spending (spread SVAR).

variable in the model (see Figs. 15–20). Therefore, in this specification the Greek government bond spread reflects the financial markets’ views on Greece’s economic outlook in line with Ardagna (2010), Laubach (2009), and Schuknecht et al. (2010). A cut in government spending lowers the government bond spread– this reflects the fact that fiscal consolidation is perceived as a positive action by financial markets (see Fig. 16). In a previous version of the paper we showed that long term government bond yield declines on account of a spending based fiscal consolidation. The negative response of interest rates and spreads and the positive effect of share prices are in line with the portfolio effects that operate in the economy, i.e., market participants operate portfolio choices and arbitrage between assets with different rates of return. Hence, when the rate of return on bonds decreases market participants have the incentive to buy other assets in the economy among which private equity. In the restricted and unrestricted SVARs, both output and private non-residential investment decline following a government spending containment (see Figs. 15 and 17). Comparing the two cases we see that: (a) a cut in government spending reduces private investment by  0.16 pp. of GDP on impact, while the first year cumulative multiplier is  0.47 in the restricted vis-à-vis  0.46 in the unrestricted SVAR. (b) Real GDP declines by  0.37 p.p. on impact in the restricted SVAR compared to 0.35 pp in the unrestricted. The first year cumulative multiplier of output is  0.59 in the restricted

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

15

0.20 0.10 0.00 -0.10

0

2

4

6

8 10 12 14 16 18 20 Output response

-0.20 -0.30 -0.40

Output response_restricted 68% CI

-0.50 Fig. 17. The output response to a cut in government spending (spread SVAR).

0.10 0.05 0.00 -0.05 0 2 4 6 8 10 12 14 16 18 20 -0.10 -0.15 -0.20 Investment response -0.25 -0.30 Investment response -0.35 _restricted -0.40 68% CI -0.45 Fig. 18. The response of private non-residential investment to an increase in tax burden (spread SVAR).

0.80 0.60 0.40 0.20 0.00 -0.20

0 2 4 6 8 10 12 14 16 18 20 Spread response

-0.40 -0.60

Spread response_restricted 68% CI

Fig. 19. The spread response to an increase in tax burden (spread SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

16

0.10 0.00 -0.10

0

2

4

6

8 10 12 14 16 18 20

-0.20 -0.30 -0.40 -0.50

Output response Output response_restricted 68% CI

-0.60 Fig. 20. The output response to an increase in tax burden (spread SVAR).

compared to  0.56 in the unrestricted SVAR.18 Overall, the negative real GDP response is marginally more pronounced in the restricted SVAR but only for about one year, while the response of investment is almost identical. Hence, it is primarily the GDP response that formulates the response profile of investment, while at the same time there are limited differences between the restricted and unrestricted spread specifications. In case of a positive tax shock (tax hike), the spread increases on impact, but it then oscillates around the zero line and finally turns positive and remains so until the end of the 20 quarter horizon (see Fig. 19). The positive response (on impact and after the first year) could reflect doubts on the effectiveness of tax based consolidation in correcting fiscal imbalances. The tax hike reduces both output and private investment (see Figs. 18 and 20). The increase of the spread (on impact) leads to a slightly more pronounced negative investment (  0.26 pp) and output (  0.39 pp) response in the unrestricted SVAR vis-à-vis the restricted SVAR (  0.25pp and  0.36 pp, respectively). Thereafter, given that the spread response oscillates around the zero line, there is no particular difference between the output and investment responses in the two specifications.

5. Composition effects of fiscal policy Earlier literature has shown that the composition of fiscal policy matters for macroeconomic outcomes (e.g. Ardagna (2001); Lane and Perotti (2003)). Therefore, as a next step, we want to assess the private non-residential investment and output effects of changes in individual government expenditure and revenue components. We re-estimate specifications (1)–(2) examining 4 different SVAR specifications, i.e., one for each individual government spending component that we consider: government wage bill, government intermediate consumption, government investment, and current government transfers (including social payments, i.e., pensions, inability benefits etc). In each case the government spending variable is replaced by one of the abovementioned individual government spending components and enters before the tax variable. However, in these cases we need to control for possible correlations between budgetary items. We control for possible correlation between the budgetary items by incorporating in the SVAR (after the tax variable) the remaining government spending components, i.e., we consider 6-variableSVARs. When the variable of interest is the government wage bill (intermediate consumption or investment) we control for net taxes (tax burden excluding current government transfers) as well as other government spending i.e., government consumption and investment minus the wage bill 18

After the 4th quarter this effect turns the other way around, but the point estimates are not statistically significant.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

17

0.3 0.2 0.1 0 -0.1 0

2

4

-0.2

6

8 10 12 14 16 18 20 Investment response

-0.3 -0.4 -0.5

Investment response _restricted 68% CI

-0.6 Fig. 21. The response of private non-residential investment to a cut in government wage bill (share prices SVAR).

(intermediate consumption or investment) which enters as a third variable in the SVAR. In the case of current government transfers we incorporate (after the tax burden variable) the sum of government consumption and investment as a third variable in the SVAR.19 Turning to the tax burden subcomponents we consider 4 different 6-variable-SVAR specifications, i.e., one for each of the following variables of interest: direct household taxes, direct business taxes, VAT, and other indirect taxes. In each case one of the abovementioned individual revenue subcomponents enters in the SVAR after the government spending variable. Furthermore, in order to control for changes in other tax revenue variables we incorporate in the SVAR (as a third variable) the remaining tax revenue components. For example, when the variable of interest is direct household taxes (direct business taxes or VAT or other indirect taxes) the third variable in the SVAR is the difference between tax burden and direct household taxes (direct business taxes or VAT or other indirect taxes). 5.1. Government spending sub-components20 A cut in the government wage bill lowers domestic demand and consequently reduces both output and investment (see Figs. 21–26). Contrary to Alesina et al. (2002) and Traum and Yang (forthcoming) we do not find robust evidence of crowding-in of private investment (except on impact) in case of cuts in the government wage bill (see Figs. 21 and 24). 21 A cut in the government wage bill improves share prices and economic sentiment for about 3–4 quarters (except on impact – see Figs. 22 and 25). However, the positive effect of the fiscal consolidation on share prices and sentiment indicators is not sufficient to counterbalance the negative demand effect which exerts on output and consequently on private investment. Nevertheless, the recovery of output and investment following the downsizing of the government wage bill is faster in the unrestricted share price SVAR (Figs. 21–23) due to the positive share price response. In particular, the first year cumulative multiplier of output is  1.5 in the unrestricted share prices SVAR and  1.67 in the restricted one. The first year cumulative multipliers for investment are 19

See Appendix for the elasticity assumptions. In the remaining part we focus on the share prices and economic sentiment SVARs. However, as mentioned in Section 4.3 the findings for share prices should be considered in line with those reported earlier for the government bond spreads, i.e., portfolio effects are in operation in the economy. 21 The impact multiplier on output following a government wage bill cut ranges from  0.52 to  0.70 in the restricted (share prices/economic sentiment) SVAR to  0.52 to  0.72 in the unrestricted (share prices/ economic sentiment) SVAR. The investment impact response is positive on impact pointing to a crowding-in effect i.e., following a cut on government wage bill investment increase by about 0.04 in the share prices SVAR and by about 0.02 to 0.16, respectively, in the unrestricted and restricted ESI SVAR 20

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

18

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

5 4

Share prices

3

Share prices response_restricted 68% CI

2 1 0 -1 0

2

4

6

8 10 12 14 16 18 20

-2 -3 Fig. 22. The response of share prices to a cut in government wage bill.

0.4 0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20

-0.4 Output response _resticted Output response

-0.6 -0.8 -1

68% CI

-1.2 Fig. 23. The response of output to a cut in government wage bill (share prices SVAR).

0.3 0.2 0.1 0 -0.1 0

2

4

6

8 10 12 14 16 18 20

-0.2 -0.3

Investment response

-0.4 -0.5 -0.6

Investment response _restricted

Fig. 24. The response of private non-residential investment to a cut in government wage bill (ESI SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

19

1 0.5 0 0 2 4 6 8 10 12 14 16 18 20 -0.5 -1 ESI response

-1.5

ESI response_restricted 68% CI

-2

Fig. 25. The response of economic sentiment to a cut in government wage bill.

0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20

-0.4 -0.6 -0.8

Output response _resticted

-1 Output response

-1.2 -1.4

68% CI

Fig. 26. The response of output to a cut in government wage bill (ESI SVAR).

0.51 and 0.71, respectively, in the unrestricted and restricted share prices SVAR specifications. In the ESI-SVAR (Fig. 24–26) there is no evidence of this “confidence” channel.22 A spending cut targeting government intermediate consumption has a negative impact effect on share prices and economic sentiment, they both return to positive territory about 2 quarters after the shock (Figs. 28 and 31). Output and private investment fall on impact in both SVAR specifications and remain negative for more than year (Figs. 27–32).23 The sizeable impact decline in share prices and economic sentiment indicates that the output and investment responses are more pronounced in the unrestricted specifications, in particular in the ESI SVAR. Therefore, the recovery is delayed when 22 The first year cumulative multiplier of output is  1.8 in the restricted ESI SVAR and  2.3 in the unrestricted one; investment multipliers are  0.15 and  0.59, respectively, in the unrestricted and restricted ESI SVAR specifications. 23 The impact multiplier on output following a government intermediate consumption cut ranges from  0.61 to  0.72 in the restricted (share prices/ economic sentiment) SVAR to  0.56 to  0.70 in the unrestricted (share prices/ economic sentiment) SVAR. The investment impact multiplier is  0.21 to  0.22 in the unrestricted specification and  0.23 to  0.24 in the restricted ones.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

20

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.1 0 0

2

4

6

8 10 12 14 16 18 20

-0.1 -0.2 Investment response

-0.3 Investment response _restricted

-0.4

68% CI

-0.5 Fig. 27. The response of private non-residential investment to a cut in government intermediate consumption (share prices SVAR).

1 0.5 0 -0.5

0

2

4

6

8 10 12 14 16 18 20

-1 Share prices response

-1.5 -2

Share prices response_restricted 68% CI

-2.5 Fig. 28. The response of share prices to a cut in government intermediate consumption.

fiscal consolidation targets exclusively intermediate consumption. The first year cumulative multiplier of output ranges from  0.44 in the restricted to  1.18 in the unrestricted economic sentiment SVAR. The first year cumulative multiplier of investment is  0.50 in the restricted ESI SVAR and 0.77 in the unrestricted one. Cuts in government investment have a negative (except on impact) and persistent effect on share prices and economic sentiment (Figs. 34 and 37). Moreover, in line with Traum and Yang (forthcoming) a cut in government investment reduces output and private non-residential investment, indicating the presence of strong complementarities between private and public investment (see Figs. 33–38). Cuts in government investment reduce future public capital, which in turn lowers the marginal product of private capital. The negative response of share prices and economic sentiment (except on impact) implies that the output and investment responses are more pronounced in the unrestricted specifications. For example, in the ESI-SVAR (Figs. 36–38) the first year cumulative multipliers of output and investment are, respectively,  1.20 and  0.56 in the unrestricted model. In the restricted specification their values are  1.06 and  0.50, respectively, for output and investment. Thus, a fiscal adjustment relying on government investment induces adverse confidence effects,

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

21

0.4 0.2 0 -0.2

0 2 4 6 8 10 12 14 16 18 20

-0.4

Output response_resticted

-0.6

Output response

-0.8 68% CI

-1 Fig. 29. The response of output to a cut in government intermediate consumption (share prices SVAR).

0.1 0 -0.1

0

2

4

6

8 10 12 14 16 18 20

-0.2 -0.3

Investment response

-0.4

Investment response_restricted

-0.5

68% CI

-0.6 Fig. 30. The response of private non-residential investment to a cut in government intermediate consumption (ESI SVAR).

0.2 0.1 0 -0.1 0

2

4

6

8 10 12 14 16 18 20

-0.2 -0.3 -0.4 -0.5

ESI response ESI response_restricted 68% CI

-0.6 -0.7

68% CI

Fig. 31. The response of economic sentiment to a cut in government intermediate consumption.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

22

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.4 0.2 0 -0.2 0

2 4

6

8 10 12 14 16 18 20

-0.4 Output response_resticted Output response

-0.6 -0.8 -1

68% CI

-1.2 Fig. 32. The response of output to a cut in government intermediate consumption (ESI SVAR).

0.3 0.2 0.1 0 -0.1 0

2

4

6

8 10 12 14 16 18 20

-0.2 -0.3 -0.4 -0.5 -0.6

Investment response Investment response_restricted 68% CI

-0.7 Fig. 33. The response of private non-residential investment to a cut in government investment (share prices-SVAR).

1.5 Share prices response

1 0.5

Share prices response_restricted 68% CI

0 0 2 4 6 8 10 12 14 16 18 20 -0.5 -1 -1.5 Fig. 34. The response of share prices to a cut in government investment.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

23

0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20

-0.4 -0.6 Output response_resticted

-0.8 -1

Output response

-1.2 -1.4

68% CI

-1.6 Fig. 35. The response of output to a cut in government investment (share prices-SVAR).

0.2 0.1 0 -0.1 0

2

4

6

8 10 12 14 16 18 20

-0.2 -0.3 Investment response

-0.4 -0.5 -0.6

Investment response _restricted

-0.7

68% CI

-0.8 Fig. 36. The response of private non-residential investment to a cut in government investment (ESI-SVAR).

0.15 0.1 0.05 0 -0.05 -0.1 -0.15 -0.2

0 2 4

6 8 10 12 14 16 18 20 ESI response

ESI response_restricted 68% CI

-0.25 Fig. 37. The response of economic sentiment to a cut in government investment.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

24

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20

-0.4 -0.6 -0.8 Output

-1

response_resticted Output response

-1.2 -1.4

68% CI

-1.6 Fig. 38. The response of output to a cut in government investment (ESI-SVAR).

0.1 0.05 0 -0.05 0

2

4

6

8 10 12 14 16 18 20

-0.1 -0.15 -0.2 -0.25

Investment response Investment response_restricted 68% CI

-0.3 Fig. 39. The response of private non-residential investment to a cut in government transfer (share prices-SVAR).

magnifying the negative effects of the fiscal shock. 24 Expenditure saving in government transfer programs depresses domestic demand and decreases output and private investment on impact (see Fig. 39–44). 25 However, it generates a positive stock market response and boosts economic sentiment contributing to a faster recovery in terms of both output and private investment (in the unrestricted SVAR). The first year cumulative multiplier of output is  0.46 (  0.41) in the restricted share prices (ESI) SVAR and declines to zero in both unrestricted specifications. Similarly, the first year cumulative multiplier of investment is  0.82 (  0.60) in the restricted share prices (ESI) SVAR and declines to  0.41 (  0.32) in the unrestricted share prices (ESI) SVAR.

24 The impact multiplier of output following a government investment cut ranges from  0.90 in the share prices SVAR to  1.01 in the ESI SVAR. The investment impact response ranges from  0.23 in the ESI SVAR to  0.18 to  0.28 in the share prices SVAR (respectively, for the unrestricted and restricted specifications). 25 The impact multiplier of output following a government transfer cut ranges from  0.39 in the restricted (economic sentiment/share prices) SVAR to  0.36 to  0.38 in the unrestricted (economic sentiment/ share prices) SVAR. The investment impact response is negative on impact i.e., following a cut on government transfers private investment declines by about  0.21 to  0.25 in the restricted (economic sentiment/ share prices) SVAR to  0.17 to  0.19 p.p. of GDP in the unrestricted (economic sentiment/ share prices) SVAR.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

25

4.5 4 Share prices response

3.5 3

Share prices response_restricted 68% CI

2.5 2 1.5 1 0.5 0

-0.5 0 2 4 6 8 10 12 14 16 18 20 Fig. 40. The response of share prices to a cut in government transfer.

0.2 0 0

2

4

6

8 10 12 14 16 18 20

-0.2 Output response_resticted

-0.4 Output response

-0.6

68% CI

Fig. 41. The response of output to a cut in government transfer (share prices-SVAR).

0.1 0.05 0 -0.05 -0.1 -0.15 -0.2

0

2

4

6

8 10 12 14 16 18 20

Investment response Investment response_restricted 68% CI

-0.25 Fig. 42. The response of private non-residential investment to a cut in government transfer (ESI-SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

26

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 0.8 0.6

ESI response

0.4

ESI response_restricted

0.2

68% CI

0 -0.2

0

2

4

6

8 10 12 14 16 18 20

Fig. 43. The response of economic sentiment to a cut in government transfer.

0.4 0.3 0.2 0.1 0 -0.1 0 -0.2 -0.3

2

4

6

8 10 12 14 16 18 20 Output response_resticted Output response

-0.4 -0.5

68% CI

-0.6 Fig. 44. The response of output to a cut in government transfer (ESI-SVAR).

Alesina et al. (2002) and Traum and Yang (forthcoming) have shown that cuts in government transfer programmes induce a positive effect on private investment (possibly because they imply smaller consolidation needs in the future). This negative correlation is not fully at odds with our findings. Actually we do find evidence of a private investment increase after the 3rd quarter (see Figs. 39 and 42) following a cut in the government transfer programmes. 5.1.1. Assessment of the government spending findings The positive share price response in case of cuts in aggregate government spending is driven by cuts in the government wage bill and social transfers. Spending cut in government investment and intermediate consumption are not associated with a sustained increase in share prices. Turning to economic sentiment, it is only cuts in government transfers that induce a positive sentiment response. Hence, a fiscal adjustment that targets current transfers and the government wage bill is likely to generate positive sentiment and market reaction effects facilitating a faster recovery of output and investment. On the other hand, a fiscal effort primarily based on government investment and intermediate consumption induces a negative sentiment effect and market reaction magnifying the negative effects of fiscal consolidation and delaying the recovery. Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

27

Table 1 Impact and first year cumulative multipliers based on the unrestricted share price and ESI SVAR specifications following a cut in government spending.

Output

Multiplier

Government spending

Government Wage bill

Government intermediate consumption

Government investment

Current transfers

Impact

 0.34 to  0.35  0.46 to  0.56  0.14 to  0.15  0.43 to  0.46

 0.52 to  0.72  1.5 to  2.3

 0.56 to  0.70

 0.91 to  1.01

 0.85 to 1.18

 1.13 to  1.20

 0.38 to  0.39 0

 0.21 to  0.22

 0.18 to  0.23

 0.76 to 0.77

 0.51 to  0.56

First year (cumulative) Investment Impact First year (cumulative)

 0.02 to  0.04  0.51 to  0.59

 0.17 to  0.19  0.32 to  0.45

0.5 0 0 2 4 6 8 10 12 14 16 18 20 -0.5 Investment response

-1 -1.5

Investment response_restricted 68% CI

-2 Fig. 45. The response of private non-residential investment to an increase in direct household taxes (share prices-SVAR).

There is some limited evidence of crowding-in of private investment following cuts in the government wage bill (but only on impact) and government transfers (except on impact). In most cases, cuts in government spending sub-components lead to substantial private investment and output decline. Based on the evidence from the unrestricted specifications we find that cuts in the government investment program produce the most pronounced negative impact responses on output (  0.91 to  1.01), followed by intermediate consumption ( 0.56 to  0.70), wage bill (  0.52 to 0.72) and current transfers (  0.36 to  0.38) cuts (see Table 1). The fall in private investment is more pronounced in case of expenditure cuts in intermediate consumption ( 0.21 to 0.22) followed by cuts in government investment (  0.18 to 0.23), and in current transfers (  0.17 to  0.19). A reduction in the government wage bill has a positive impact effect on investment. According to first year cumulative multipliers, and based on the unrestricted specifications, it is wage bill cuts that induce the more pronounced effect on output (  1.5 to  2.3), followed by government investment (  1.13 to  1.2), and intermediate consumption (  0.85 to  1.18), while in case of current transfers the effect is nil (see Table 1). The first year cumulative multiplier of investment is more sizeable in case of cuts in government intermediate consumption (  0.76 to  0.77), followed by the wage bill (  0.51 to  0.59), government investment (  0.51 to  0.56) and current transfers cuts (  0.32 to  0.45). However, it is the containment of the government investment programme that has the most long lasting effects on output and investment (Fig. 33–38).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

28

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4 Share prices response Share prices response_restricted 68% CI

3 2 1 0 0

2

4

6

8 10 12 14 16 18 20

-1 -2 Fig. 46. The response of share prices to an increase in direct household taxes.

0.5 0 -0.5 0

2

4

6

8 10 12 14 16 18 20

-1 -1.5 -2 -2.5

Output response_resticted

-3

Output response

-3.5

68% CI

Fig. 47. The response of output to an increase in direct household taxes (share prices-SVAR).

0.4 0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20

-0.4 -0.6

Investment response

-0.8 -1 -1.2

Investment response _restricted 68% CI

-1.4 Fig. 48. The response of private non-residential investment to an increase in direct household taxes (ESI-SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

29

1 0.8 0.6 0.4 0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20

-0.4

ESI response

-0.6

ESI response_restricted

-0.8

68% CI

Fig. 49. The response of economic sentiment to an increase in direct household taxes.

0.5 0 -0.5 0 -1 -1.5

2

4

6

8 10 12 14 16 18 20 Output response_resticted Output response

-2 -2.5

68% CI

-3

68% CI

-3.5 Fig. 50. The response of output to an increase in direct household taxes (ESI-SVAR).

5.2. Government revenue components A direct household tax hike reduces on impact both output and private investment (see Fig. 45–50). 26 The decline in investment is in line with the evidence presented in Traum and Yang (forthcoming), Forni et al. (2009), and Yang (2005). A labour tax hike decreases the demand for labour reducing the marginal product of capital, thus, decreasing private investment. Nevertheless, both share prices and economic sentiment respond positively (except on impact for ESI).27 This has a beneficial effect on investment and output, because they recover at a faster pace after the direct tax hike shock in the unrestricted relative to the restricted SVAR specifications. In more detail, the first year cumulative multiplier of output ranges from  4.38 in the unrestricted to 4.58 in the restricted ESI specification. The first year cumulative multiplier of investment ranges from  1.25 in the unrestricted to  1.43 in the restricted ESI specification,

26 The impact multiplier of output following a direct household tax shock ranges from  2.41 to  2.43, respectively, in the unrestricted and restricted ESI SVAR to  2.67 in the (restricted and unrestricted) share price SVAR. The impact multiplier of investment following a direct household tax shock ranges from  0.88 in the (restricted and unrestricted) ESI SVAR to  1.15 and  1.17, respectively, in the unrestricted and restricted share prices SVAR. 27 Note that economic sentiment decline on impact because of the likely fall in consumer confidence due to the direct household tax hikes (see Konstantinou and Tagkalakis (2011)).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

30

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.1 0 -0.1

0

2

4

6

8 10 12 14 16 18 20

-0.2 -0.3

Investment response

-0.4

Investment response_restricted

-0.5

68% CI

-0.6 Fig. 51. The response of private non-residential investment to an increase in direct business taxes (share prices-SVAR).

0.8 0.6 0.4 0.2 0 -0.2 0 2 4 6 8 10 12 14 16 18 20 -0.4 -0.6 -0.8

Share prices response Share prices response_restricted 68% CI

Fig. 52. The response of share prices to an increase in direct business taxes.

0.2 0.1 0 -0.1 0 2 4 6 -0.2

8 10 12 14 16 18 20

-0.3

Output response

-0.4

Output response

-0.5

68% CI

-0.6 -0.7 -0.8 Fig. 53. The response of output to an increase in direct business taxes (share prices-SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

31

0.1 0 -0.1 0 2 4 6 8 10 12 14 16 18 20 -0.2

Investment response

-0.3

Investment response_restricted 68% CI

-0.4 -0.5 -0.6

Fig. 54. The response of private non-residential investment to an increase in direct business taxes (ESI-SVAR).

0.05 0 0 2 4 6 8 10 12 14 16 18 20 -0.05 -0.1 -0.15

ESI response ESI response_restricted 68% CI

-0.2 Fig. 55. The response of economic sentiment to an increase in direct business taxes.

0.1 0 -0.1 0 2 4 6 8 10 12 14 16 18 20 -0.2 -0.3 -0.4 -0.5 -0.6 -0.7

Output response_resticted Output response 68% CI

-0.8 Fig. 56. The response of output to an increase in direct business taxes (ESI-SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

32

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.8

Investment response

0.6

Investment response _restricted 68% CI

0.4 0.2 0 0

-0.2

2

4

6

8 10 12 14 16 18 20

-0.4 Fig. 57. The response of private non-residential investment to an increase in VAT (share prices-SVAR).

1 0 -1

0

2

4

6

8 10 12 14 16 18 20

-2 -3 -4 -5

Share prices response Share prices response_restricted 68% CI

-6 Fig. 58. The response of share prices to an increase in VAT.

A tax hike driven by direct business taxes depresses both output and investment (see Fig. 51–56). 28 Relevant theoretical literature (e.g. Traum and Yang (forthcoming), Forni et al. (2009) and Yang (2005)) has shown that an increase in the business and capital tax rate reduces the after-tax-rate of return to investment. The stock market (after a fall on impact) perceives this tax hike as improving fiscal balances. However, the negative effects of fiscal consolidation on output and investment are not particularly mitigated in the unrestricted vis-à-vis the restricted SVAR (Figs. 51–53). In line with Konstantinou and Tagkalakis (2011) the economic sentiment indicator declines following a direct business tax hike, reflecting primarily the negative responses of the business sector, i.e., the industrial, retail, services and construction sectors (Fig. 55). This implies that the real economy actors perceive the direct business tax hike as a negative development, which will further depresses investment and output, on top of the direct fiscal consolidation effects (see Figs. 54–56). The first year cumulative multiplier of output ranges from  0.86 in the restricted to  1.00 in the unrestricted ESI SVAR, i.e., the effects of fiscal consolidation are aggravated due to the negative confidence effect.29 28 The impact multiplier of output following a direct business tax shock ranges from  0.21 in the ESI SVAR to  0.26 in the share prices SVAR. The impact multiplier of investment following a direct business tax shock ranges from  0.20 in the ESI SVAR to  0.21 to  0.24, respectively, in the restricted and unrestricted share prices SVAR. 29 The first year cumulative multiplier is 1.3 in the share prices SVAR.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 0.8 0.6 0.4

33

Output response_resticted Output response 68% CI

0.2 0 -0.2 0 2 4 6 8 10 12 14 16 18 20 -0.4 -0.6 -0.8 Fig. 59. The response of output to an increase in VAT (share prices-SVAR).

0.8 Investment response

0.6 0.4 0.2

Investment response_restricted 68% CI

0 -0.2

0 2 4 6 8 10 12 14 16 18 20

-0.4 -0.6 Fig. 60. The response of private non-residential investment to an increase in VAT (ESI-SVAR).

0.8 0.6 0.4 0.2 0 -0.2 0 2 4 6 8 10 12 14 16 18 20 -0.4 -0.6 -0.8

ESI response ESI response_restricted 68% CI

-1 Fig. 61. The response of economic sentiment to an increase in VAT.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

34

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1 0.8 0.6 0.4

Output response_resticted Output response 68% CI

0.2 0 -0.2 0 2 4 6 8 10 12 14 16 18 20 -0.4 -0.6 -0.8 -1 Fig. 62. The response of output to an increase in VAT (ESI-SVAR).

0.2 0.1 0 -0.1 0 2 4 6 8 10 12 14 16 18 20 -0.2 -0.3 Investment response -0.4 -0.5 Investment -0.6 response_restricted 68% CI -0.7 -0.8 Fig. 63. The response of private non-residential investment to an increase in other indirect taxes (share prices-SVAR).

An increase in the Value Added Tax (VAT) reduces output and investment on impact. Both output and investment become positive for few quarters, but they turn negative again (see unrestricted SVAR specifications; Figs 57–62). The VAT increase leads to a negative share price and economic sentiment response (except on impact for ESI).30 Consequently, both the (restricted) share price and the ESI SVARs indicate that the increase in investment and output would have been more pronounced in the absence of the negative share price and economic sentiment response. Hence, a VAT based consolidation although it has smaller negative or even positive effects on economic activity it generates adverse confidence effects which delay the recovery. The first year cumulative multiplier of output is  0.31 in the unrestricted share prices SVAR and turns to 0.78 (i.e., positive) in case of the restricted share prices SVAR. Thus, a VAT tax hike decreases output in the unrestricted specification, but it increases it in the restricted model.31 As discussed in Traum and Yang (forthcoming) an increase in consumption taxes leads to a negative investment response on impact because consumer prices increase inducing the monetary 30 The impact multiplier of output following VAT shock ranges from  0.27 to  0.29 in the (restricted versus unrestricted) share prices SVAR to  0.40 to  0.34 in the (restricted versus the unrestricted) ESI SVAR. The impact multiplier on investment following an indirect tax shock ranges from  0.01 to  0.04 in the (restricted versus unrestricted) share prices SVAR to  0.14 to  0.15 in the (unrestricted versus the restricted) ESI SVAR. 31 The cumulative multiplier of investment is 0.18 in the unrestricted case and becomes 0.32 in the restricted ESI SVAR, i.e., a VAT tax hike increase investment in both cases, but the effect is bigger in the restricted specification.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

3

35

Share prices response

2.5 2 1.5 1

Share prices response_restricted 68% CI

0.5 0 -0.5 0 2 4 6 8 10 12 14 16 18 20 -1 -1.5 -2 Fig. 64. The response of share prices to an increase in other indirect taxes.

0.2 0 -0.2 0 2 4 6 8 10 12 14 16 18 20 -0.4 -0.6 -0.8 -1 -1.2 -1.4

Output response_resticted Output response 68% CI

-1.6 Fig. 65. The response of output to an increase in other indirect taxes (share prices-SVAR).

0.1 0 -0.1

0 2 4 6 8 10 12 14 16 18 20

-0.2 -0.3 -0.4 -0.5

Investment response

-0.6

Investment response_restricted

-0.7

68% CI

-0.8 Fig. 66. The response of private non-residential investment to an increase in other indirect taxes (ESI-SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

36

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.3 0.2 0.1 0 -0.1 0 2 4 6 8 10 12 14 16 18 20 -0.2

ESI response

-0.3

ESI response_restricted 68% CI

-0.4 -0.5

Fig. 67. The response of economic sentiment to an increase in other indirect taxes.

0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20

-0.4 -0.6 -0.8 -1 -1.2 -1.4 -1.6

Output response_resticted Output response 68% CI

-1.8 Fig. 68. The response of output to an increase in other indirect taxes (ESI-SVAR).

policy maker to raise interest rates. However, after the immediate negative impact effect private investment gradually becomes positive (as future consolidation needs are reduced). In the specification of Forni et al. (2009) that allows for capital utilization, following a consumption tax hike and the subsequent monetary policy reaction, investment responds positively in a gradual manner. A tax hike related to other indirect taxes (which includes excise taxes and financial and capital transaction taxes) and affects productions costs (e.g. excise taxes on energy) has a negative effect on both output and investment (see Fig. 63–68).32 The increase in other indirect taxes leads to a decline in share prices on impact; next they turn positive for 2–3 quarters, before becoming negative again. On the other hand, economic sentiment responds negatively to the increase in other indirect taxes. As a consequence the investment and output responses in the unrestricted ESI SVAR point to a delayed recovery compared to those in the restricted ESI SVAR. The evidence on the share prices specification is not as clear cut. Therefore, the first year cumulative multiplier of output is  2.9 in the

32 The impact multiplier of output following a shock on other indirect taxes ranges from  0.40 to  0.41 in the (unrestricted versus restricted) share prices SVAR to  0.44 to  0.45 in the (unrestricted versus the restricted) ESI SVAR. The impact multiplier on investment following an indirect tax shock ranges from  0.35 to  0.38 in the restricted versus unrestricted share prices SVAR to  0.35 in both ESI SVAR specification.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

37

Table 2 Impact and first year cumulative multipliers based on the unrestricted share price and ESI SVAR specifications following a tax hike.

Output

Multiplier

Tax burden

Direct household taxes

Direct business taxes

VAT

Other indirect taxes

Impact

 0.37 to  0.42  1.07 to  1.5  0.25 to  0.27  0.31 to  0.41

 2.41 to  2.67

 0.21 to  0.26

 0.41 to  0.45

 4.38 to  4.54

 1.32 to  1.43

 0.29 to  0.34 0 to  0.31

 0.89 to  1.15

 0.20 to  0.24

 1.25 to  1.67

 0.98 to  1.00

First year (cumulative) Investment Impact First year (cumulative)

 0.04 to  0.14 0.18 to 0.36

 2.38 to  2.92  0.35 to  0.38  0.91 to  1.02

unrestricted compared to  2.5 in the restricted ESI SVAR. Similarly, the first year cumulative multiplier for investment is  1.02 in the unrestricted and  0.84 in the restricted ESI SVAR.33

5.2.1. Assessment of the government revenue findings The negative share price response in case of an increase in the tax burden is driven by increases in VAT, while increases in the other tax components are associated positively with share price changes except on impact. Economic sentiment deteriorates following a tax hike in all cases considered (except for direct household taxes). Moreover, in most cases considered a tax hike leads to negative output and investment response on impact. However, after a direct household tax hike economic sentiment and share price increase, which leads to a more rapid recovery of output and investment. On the other hand, in line with the baseline finding for tax burden, a fiscal adjustment effort that is based on direct business taxes and other indirect taxes has a negative impact on economic sentiment which in turn depresses investment and output, i.e., it aggravates the negative effects of fiscal consolidation. Similarly, VAT based consolidations generate adverse confidence effects which delay the recovery. Based on the unrestricted specifications we find that a direct household tax hike has the most sizeable impact effect on output (  2.41 to  2.67), followed by other indirect taxes ( 0.41 to 0.45), VAT (  0.29 to  0.30), and direct business taxes ( 0.21 to  0.26); see Table 2. As regards private non-residential investment the most pronounced impact effect occurs in case of increases in direct household taxes (  0.89 to 1.15), followed by other indirect taxes (  0.35 to 0.38), direct business taxes (  0.20 to  0.24), and VAT (  0.04 to  0.14). According to first year cumulative multipliers from the unrestricted specifications, it is household taxes that induces the biggest and most long lasting effect on output ( 4.38), followed by other indirect taxes ( 2.38), direct business taxes (  1.32 to 1.43), and VAT tax hikes (  0.31). Turning to private investment, the first year cumulative multiplier is more sizeable in case of direct household taxes (  1.25 to 1.67), followed by direct business taxes (  0.98 to 1.00), and other indirect taxes (  0.91 to  1.02). VAT has positive effects on investment one year after the initial shock. 34

33 The first year cumulative multipliers of output and investment are, respectively,  2.38 and  0.91 in the share prices SVAR. 34 We have also considered a specification with capital taxes (i.e., estate, inheritance and gift taxes) and actual social security contributions being the variable of interest (controlling also for the difference between the tax burden and capital taxes or social security contributions). An increase in social security contributions is associated negatively with output with an impact multiplier of about  0.14 and a first year cumulative multiplier of  0.60 (  0.86) in the unrestricted ESI (share price) specifications and  0.5 in both restricted SVARs. Capital taxes are associated positively with output in line with Arnold et al. (2011). However, in both cases the private investment responses were not significant.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

38

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.05 0 -0.05

0

2

4

6

8 10 12 14 16 18 20

-0.1 -0.15 -0.2 -0.25 -0.3 -0.35

Investment response_restricted Investment response Investment response_αtsp=0.01 Investment response_αtsp=0.026

Fig. 69. The response of private investment to an increase in the tax burden with positive share price elasticity of taxes αtsp.

5.3. Summary of findings: tax revenue versus expenditure based fiscal consolidation The negative effect of an expenditure-based fiscal consolidation on output and investment are of smaller magnitude compared to a tax burden based fiscal consolidation both on impact ( 0.34 to 0.35 versus  0.37 to  0.42) and based on first year cumulative multipliers ( 0.46 to 0.56 versus 1.07 to  1.5). Moreover, the negative effects of a tax burden hike on output and private investment last longer relative to those of a spending cut. Turning to individual budgetary components we find evidence that direct household taxes, as well as cuts in government investment exert the most pronounced negative impact effect on output and investment. Increases in direct household taxes, other indirect taxes and cuts in the government wage bill, and in government investment followed by increases in direct business taxes, and cuts in intermediate consumption have the most sizeable and long lasting negative effects on output. Tax hikes in direct household and business taxes and in other indirect taxes, followed by cuts in government intermediate consumption, investment, wage bill and current transfers have the most sizeable and protracted negative effect on investment. A fiscal consolidation driven primarily by tax hikes will depress private investment and output to a greater extend compared to a government spending based fiscal consolidation. This is due both to its direct effect on output and investment and its indirect effect through the fall in share prices and the deterioration in economic sentiment. Some of the spending and revenue components (e.g. increases in direct household taxes, cuts in current transfers and the government wage bill) generate some positive confidence effect which in turn mitigate the negative effects of fiscal consolidation. On the other hand, increases in direct business taxes and indirect taxes as well as cuts in government investment and intermediate consumption generate a negative confidence effect which delays the recovery.

6. Robustness checks 6.1. The role of a positive share price elasticity of taxes (αtsp) In the baseline specification (2) we assumed that the share price elasticity of taxes is αtsp ¼ 0. However, earlier studies have shown that tax revenue are sensitive to asset price changes (e.g., Morris and Schucknecht (2007), Price and Dang (2011), Tagkalakis (2011), Liu et al. (2015)). In such a case a tax revenue shock at quarter t¼ 0 would lead to an immediate change in share prices, which in turn

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

39

0 0

2

4

6

8 10 12 14 16 18 20

-2 -4

Share price response_resticted

-6

Share prices response

-8

Share price response_αtsp=0.01

-10

Share price response_αtsp=0.026

-12

Fig. 70. The response of share prices to an increase in the tax burden with positive share price elasticity of taxes αtsp.

0 0

2

4

6

8 10 12 14 16 18 20

-0.1 -0.2 -0.3

Output response_restricted Output response

-0.4 -0.5 -0.6

Output response_αtsp=0.01 Output response_αtsp=0.026

Fig. 71. The response of output to an increase in the tax burden with positive share price elasticity of taxes αtsp.

would feed back to tax revenue within the same quarter. This casts doubt on the identification of the tax policy shock and the assessment of its effects on share prices and economic activity. In order to account for this contemporaneous impact we follow earlier studies such as Liu et al. (2015) and Price and Dang (2011). Based on a group of advanced economies Liu et al. (2015) have estimated a short run elasticity (that captures contemporaneous changes) between share prices and tax revenue of αtsp ¼0.01. On the other hand, Price and Dang (2011) using disaggregated information on various tax categories have found a short run elasticity between share prices and tax revenue of about αtsp ¼0.026 for Greece.35 Next we re-estimate the baseline share prices SVAR taking into account these two elasticity assumptions; the findings are reported in Fig. 69–71. After an increase in the tax burden share prices decline to a much greater extent once we account for the contemporaneous elasticity between share prices and tax revenue. The impact effect on share prices is more pronounced the bigger is the share price elasticity of taxes. The impulse response of output and investment indicate that once the 35 Both Liu et al. (2015) and Price and Dang (2011) have also estimated long run elasticties between share prices and tax revenue. In both cases the values are close to 0.1%, i.e., 0.095 in case of Liu et al. (2015) and 0.08 in case of Price and Dang (2011). For the case of our exercise short run elasticities are more relevant because we want to capture the contemporaneous (within the quarter) response of tax revenue to share price changes.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

40

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0.5 0 0

2

4

6

8 10 12 14 16 18 20

-0.5 Investment response_restricted Investment response

-1 -1.5

Investment response_αtsp=0.07

-2

Fig. 72. The response of private investment to an increase in direct household taxes with positive share price elasticity of taxes αtsp.

4 2 0 -2

0 2 4 6 8 10 12 14 16 18 20

-4

Share price response_restricted Share price response

-6 -8

Share price response_αtsp=0.07

-10

Fig. 73. The response of share prices to an increase in direct household taxes with positive share price elasticity of taxes αtsp.

0 -0.5 0

2

4

6

8 10 12 14 16 18 20

-1 -1.5 -2 -2.5 -3 -3.5

Output response_restricted Output response Output response_αtsp=0.07

Fig. 74. The response of output to an increase in direct household taxes with positive share price elasticity of taxes αtsp.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

41

0 -0.1 0

2

4

6

8 10 12 14 16 18 20

-0.2 -0.3

Investment response_restricted

-0.4 -0.5

Investment response

-0.6 -0.7 Investment response_αtsp=0.09

-0.8 -0.9

Fig. 75. The response of private investment to an increase in other indirect taxes with positive share price elasticity of taxes αtsp.

2 0 0

2

4

6

8 10 12 14 16 18 20

-2 Share price response_restricted

-4

Share price response

-6 Share price response_αtsp=0.09

-8

Fig. 76. The response of share prices to an increase in other indirect taxes with positive share price elasticity of taxes αtsp.

0 0

2

4

6

8 10 12 14 16 18 20

-0.5

-1

-1.5

-2

Output response_restricted Output response Output response_αtsp=0.09

Fig. 77. The response of output to an increase in other indirect taxes with positive share price elasticity of taxes αtsp.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

42

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

1.2 Tax burden response_αty=0.94

1 0.8 0.6

Tax burden response_αty=1.5

0.4

Tax burden response_αty=2

0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20

-0.4 Fig. 78. The response of tax burden to an own shock with higher output elasticity of taxes αty. (share price SVAR).

0 -0.05 0 2 4 6 8 10 12 14 16 18 20 -0.1 -0.15 -0.2 -0.25 -0.3 -0.35 -0.4

Investment response_αty=0.94 Investment response_αty=1.5 Investment response_αty=2

-0.45 Fig. 79. The response of private investment to an increase in the tax burden with higher output elasticity of taxes αty. (share price SVAR).

0 0 2 4 6 8 10 12 14 16 18 20 -0.2 -0.4 -0.6 -0.8 -1

Output response_αty=0.94 Output response_αty=1.5 Output response_αty=2

Fig. 80. The response of output to an increase in the tax burden with higher output elasticity of taxes αty. (share price SVAR).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

43

0 1

-2

3

5

7

9 11 13 15 17 19 21

Share price response αty=2 αtsp=0.026

-4 -6

SP response_0,94

-8 SP response_2

-10 -12

Share price response_αty=0.94 αtsp=0.026

-14

Fig. 81. The response of share prices to an increase in the tax burden under different elasticity assumptions (αty ¼ 0.94 and 2, αtsp ¼0 and 0.026) (share price SVAR).

0 -0.05 1 -0.1 -0.15 -0.2 -0.25 -0.3 -0.35 -0.4 -0.45 -0.5

3

5

7

9 11 13 15 17 19 21 Investment response_αty=2 αtsp=0.026 Investment response_αty=0.94 Investment response_αty=2 Investment response_αty=0.94 αtsp=0.026

Fig. 82. The response of private investment to an increase in the tax burden under different elasticity assumptions (αty ¼ 0.94 and 2, αtsp ¼ 0 and 0.026) (share price SVAR).

contemporaneous share price elasticity of taxes is accounted for the negative investment and output responses following a tax burden shock are more sizeable and last longer. We have repeated a similar exercise for two tax revenue components: (a) direct household taxes, because this revenue category incorporates the capital gains taxes on individuals and (b) other indirect taxes, because this revenue category incorporates taxes on financial and capital transactions. The elasticity assumptions follow the short term elasticity assumptions of Price and Dang (2011). In the case of direct household taxes it is set to αtsp ¼ 0.07, while in the case of other indirect taxes it is set to 0.092.36 The findings are reported in Fig. 72–74 for direct household taxes and Figs. 75–77 for other indirect taxes. The share price, investment and output responses are more pronounced and more protracted once we account for the contemporaneous correlation between share prices and tax revenue. In particular, in the case of other indirect taxes the findings are substantially more sizeable relative to the baseline specification.

36 The short run elasticity of share prices to taxes on financial and capital transactions is estimated to be 0.36 in Price and Dang (2011). Taking into account that financial and capital transactions taxes are about ¼ of other indirect taxes the share price elasticity of other indirect taxes is set to 0.092.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

44

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0 -0.1 1

3

5

7

9 11 13 15 17 19 21

-0.2 -0.3 -0.4 -0.5

Output response_αty=0.94

-0.6 Output response_αty=2 αtsp=0.026 Output response_αty=2

-0.7 -0.8 -0.9 -1

Output response_αty=0.94 αtsp=0.026

Fig. 83. The response of output to an increase in the tax burden under different elasticity assumptions (αty ¼ 0.94 and 2, αtsp ¼ 0 and 0.026) (share price SVAR).

0.02 0 -0.02 0 2 4 6 8 10 12 14 16 18 20 -0.04 -0.06

Investment response_αgy=-0.07

-0.08 -0.1 -0.12 -0.14 -0.16

Investment response_αgy=-0.2 Investment response_αgy=-0.3

-0.18 Fig. 84. The response of private non-residential to an increase in the government spending under different elasticity assumptions (αgy ¼  0.07,  0.2 and  0.3) (ESI-SVAR).

Concluding, failure to account for the contemporaneous correlation between share prices and taxes could underestimate the negative effect of tax policy changes on output and investment, in particular, in the case of revenue components that are sensitive to asset price changes.

6.2. Higher output elasticity of taxes As a robustness check we consider the effect of a higher output elasticity of taxes αty, i.e., we consider αty ¼ 1.5 and 2 instead of αty ¼0.94 in the baseline model. Previous studies, such as Poghosyan (2011), Sancak et al. (2010) and Tagkalakis (2014b) have shown that tax revenue elasticities increase in crisis years. If we fail to account for this development we shall underestimate the effect of the tax shock on output. Building on the baseline share prices SVAR we present the relevant impulse responses in Fig. 78–80. The bigger is the contemporaneous elasticity of taxes the more sizeable are the output and investment responses to the tax burden shock. Therefore, if tax elasticities increased in recent crisis years above the current elasticity estimates in Price et al. (2014), then the actual effect of a tax policy change might have Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

45

0.05 0 -0.05 0 2 4 6 8 10 12 14 16 18 20 -0.1 -0.15

Output response_αgy=-0.07

-0.2 -0.25

Output response_αgy=-0.2

-0.3 -0.35

Output response_αgy=-0.3

-0.4 -0.45

Fig. 85. The response of output to an increase in the government spending under different elasticity assumptions (αgy ¼  0.07,  0.2 and  0.3) (ESI-SVAR).

0.2 0 -0.2 -0.4 -0.6

1

3

5

7

9 11 13 15 17 19 21 Government spending response_1st part

-0.8 -1

Government spending response_2nd part

-1.2 Fig. 86. The response of government spending to an own shock in the first and second part of the sample.

been underestimated leading to more dramatic output decline than initially expected (at the time that the tax policy change was planned).

6.3. Output elasticity of taxes versus share price elasticity of taxes In previous subsections we show that the underestimation of output elasticity of taxes (αty) and the failure to account for the contemporaneous correlation of share prices and tax revenue (αtsp) lead to an underestimation of the output and investment responses to a tax shock. In order to examine which effect matters the most we consider the baseline SVAR specification which assumes αty ¼0.94 and αtsp ¼0 and another high output and share price elasticity SVAR which assumes αty ¼2 and αtsp ¼0.026. We find that it is the share price elasticity of taxes (αtsp) that mostly affects the share prices response to a tax policy shock (see Fig. 81). However, it should be emphasized that it is the output elasticity of taxes (αty) that matters the most for the magnitude of the output and investment responses (see Fig. 82 and 83).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

46

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4 3 2 1 0 0

-1

2

4

6

8

10 12 14 16 18 20

Share price response_restricted

-2

Share price response_1st part

-3

Share price response_2nd part

Fig. 87. The response of share prices to a government spending shock in the first and second part of the sample.

0.2 0.1 0 0

2

4

6

8 10 12 14 16 18 20

-0.1 -0.2

Output response_1st part

-0.3

Output response_restricted_1st part Output response_restricted_2nd part Output response_2nd part

-0.4 -0.5

Fig. 88. The response of output to a government spending shock in the first and second part of the sample.

0 -0.05 0

2

4

6

8 10 12 14 16 18 20

-0.1 -0.15 -0.2

Investment response_1st part

-0.25 -0.3 -0.35

Investment response_2nd part

Fig. 89. The response of private investment to a government spending shock in the first and second part of the sample.

6.4. Higher output elasticity of government spending As a robustness check we consider the effect of a higher output elasticity of government spending

αgy, i.e., we consider αgy ¼  0.2 and  0.3 instead of αgy ¼  0.07 in the baseline. This assumption follows the same logic as the one for taxes, i.e., the contemporaneous output elasticity of government

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

47

0.8 0.6 0.4 0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20 ESI response_restricted

-0.4

ESI response_1st part

-0.6

ESI response_2nd part Fig. 90. The response economic sentiment to a government spending shock in the first and second part of the sample.

1.2 1 0.8 Tax burden response_1st part Tax burden response_2nd part

0.6 0.4 0.2 0 -0.2

0

2

4

6

8 10 12 14 16 18 20

Fig. 91. The response of tax burden to an own shock in the first and second part of the sample.

spending could increase more than anticipated during high unemployment recession years. For example, the unemployment rate in Greece increased from below 9–10% before the crisis to about 26% in 2014. The relevant impulse responses are reported in Fig. 84 and 85. The bigger is the contemporaneous output elasticity of government spending the more sizeable are the output and investment responses to the government spending shock. Failure to account for the higher unemployment related spending during crisis years could underestimate the effect of the government spending shock on output and investment. However, this effect is much less pronounced compared to the one for the (underestimation of the) output elasticity of taxes. 6.5. Different sub-samples One of the particularities of the data that are available is the fact that from the beginning of the sample in 2000 Q1 till 2008 Q2 Greece registered positive growth rates (year-on-year), whereas from 2008 Q3 till the end of the sample in 2014 Q1 Greece registered negative growth rates. As a robustness exercise we consider the output effects of the government spending and tax shocks in the two sub-samples. We consider two parsimonious SVAR specifications; both involve 4 variables in first differences with one lag and the lagged debt ratio as exogenous variable. The first specification includes: government spending, tax burden, share prices or ESI and output, while the second has

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

48

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

4 2 0 0

2

4

6

8 10 12 14 16 18 20

-2

Share price response_restricted

-4

Share price response_1st part

-6

Share price response_2nd part

Fig. 92. The response of share prices to a tax burden shock in the first and second part of the sample.

0 -0.1 0 2 4 6 8 10 12 14 16 18 20 -0.2

Output response_1st part

-0.3 -0.4 -0.5 -0.6 -0.7 -0.8

Output response_restricted_1st part Output response_restricted_2nd part Output response_2nd part

-0.9 Fig. 93. The response of output to a tax burden shock in the first and second part of the sample.

0 -0.05 0 2 4 6 8 10 12 14 16 18 20 -0.1 -0.15 -0.2

Investment response_1st part

-0.25 -0.3 -0.35

Investment response_2nd part

-0.4 Fig. 94. The response of private investment to a tax burden shock in the first and second part of the sample.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

49

0.6 0.4 0.2 0 -0.2 0

2

4

6

8 10 12 14 16 18 20

-0.4 ESI response_restricted

-0.6 -0.8

ESI response_1st part

-1

ESI response_2nd part

-1.2 Fig. 95. The response of economic sentiment to a tax burden shock in the first and second part of the sample.

0 -0.05 0

2

4

6

-0.1

8 10 12 14 16 18 20 Investment response_1st part

-0.15

Investment response_2nd part

-0.2 -0.25

Investment response_2nd part_αty=1.5

-0.3 -0.35

Investment response_2nd part_αty=2

-0.4

Fig. 96. The response of private investment to a tax burden shock in the first and second part of the sample under different assumptions about the output elasticity of taxes (αty).

0 -0.1 0 -0.2 -0.3 -0.4 -0.5 -0.6

2

4

6

8 10 12 14 16 18 20 Output response_1nd part Output response_2nd part Output response_2nd part_ αty=1.5 Output response_2nd part_ αty=2

-0.7 Fig. 97. The response of output to a tax burden shock in the first and second part of the sample under different assumptions about the output elasticity of taxes (αty).

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

50

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

0 0

2

4

6

8 10 12 14 16 18 20

-0.1 Output response_1st part

-0.2 -0.3 -0.4 -0.5

Output response_2nd part Output response_2nd

part_α ty=2 Output response _2nd

part_α ty=2, αtsp=0.026 -0.6 -0.7

Output response _1st

part_α ty=0.94, αtsp=0.01

Fig. 98. The response of output to a tax burden shock in the first and second part of the sample under different assumptions about the output and share price elasticity of taxes (αty and αtsp).

government spending, tax burden, private investment and output.37 Given that the SVARs involve limited number of observations, i.e., 34 in the first part and 23 in the second part of the sample, the findings have to be considered as preliminary and have to be taken with a grain of salt (that is also why we do not report confidence bands). Following a cut in government spending share prices and economic sentiment respond positively in the second part of the sample (a government spending based consolidation induces a positive confidence effect) and negatively in the first part (Figs. 86, 87 and 90). The output decline is bigger and more protracted in the second half of the sample (Fig. 88), whereas the fall in private investment is more pronounced in the first sub-sample (Fig. 89).38 The positive share price and ESI response implies that output recovers at a faster pace in the second part of the sample (see restricted vis-à-vis the unrestricted specification in Fig. 88). In particular, one year after the shock the (non-accumulated) expenditure multiplier is at  0.14 in the unrestricted compared to 0.20 in the restricted specification. In addition, the expenditure multiplier returns to zero at the 8th quarter after the shock in the unrestricted model, whereas it takes until the 14th quarter to return to zero line in the resticted model. This effect is absent in the period prior to 2008 Q3. During the period 2008 Q3–2014 Q1 output declines to a greater extent but share prices and sentiment improve counterbalancing the negative demand effect on investment. In the first part of the sample output declines to a lesser extent. However, both share prices and ESI fall magnifying the negative investment response. A tax hike in the first part of the sample increases share prices and economic sentiment (Figs. 91, 92 and 95) and decreases both output and investment (Fig. 93–94). In the second part of the sample the tax hike lowers ESI and share prices, i.e., a tax based fiscal consolidation generates a negative confidence effect. Consequently, the fall in output is more sizeable in the unrestricted vis-à-vis the restricted specification in the 2nd part of the sample for the first 5-6 quarters after the shock (see Fig. 93). In more detail, one year after the shock the (non-accumulated) tax multiplier is  0.27 in the unrestricted versus  0.19 in the restricted specification.39 Nevertheless, the decline in output and investment is much more pronounced in the first rather than the second sub-sample. During the period 2000 Q1–2008 Q2 the more pronounced fall in output determines the response profile of investment despite the fact that stock market and economic sentiment increase after the tax shock. In the second part of the sample, investment declines in line with the fall in output and share prices and the deterioration in economic sentiment. 37 We have also considered a simpler 3 variable SVAR with government spending, tax burden and output. The output responses are qualitatively similar. 38 The impulse response of investment is obtained from the second SVAR specification without share prices or ESI. 39 The tax multiplier returns to zero at the 10th quarter after the shock in the unrestricted model, whereas it takes until the 13th–14th quarter to return to zero line in the restricted model.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

51

Interestingly, although the effects of a government spending shock on output are more pronounced in the recession years in line with relevant literature (e.g. Auerbach and Gorodnichenko (2012)), we do not find the same for the tax shock. As noted above it might be the case that this is related to the underestimation of the output and share prices elasticities of taxes in recession years. Therefore, we re-estimate the SVAR specifications considered before assuming αty ¼ 1.5 and αty ¼2 in the second part of the sample and αty ¼ 0.94 in the first. According to the evidence reported in Fig. 96 and 97 the output and investment responses increase in size in the second part of the sample once we assume higher output elasticities of taxes. Nevertheless, the impulse responses are still not as pronounced as in the first sample. As a next step we consider an alternative specification that takes also into account the share price elasticity of taxes. We assume αty ¼ 2 and αtsp ¼0 (or αtsp ¼0.026) in the second part and αty ¼ 0.94 and αtsp ¼0 (or αtsp ¼0.01) in the first part. As displayed in Fig. 98 the combination of substantially higher output and share prices elasticity of taxes in the second half of the sample (αty ¼2 and αtsp ¼0.026) compared to the first half (αty ¼0.94 and αtsp ¼ 0) could explain most of the remaining difference in the output response to a tax shock.40 Overall, it should be borne in mind that the evidence presented in this section is preliminary and is based only on limited data information. Further analysis will be required in the future (once more data become available) in order to better understand the workings of the fiscal policy in the periods of expansion and recession in Greece.

7. Summary and conclusions This paper builds on the SVAR approach proposed by Blanchard and Perotti (2002) and Perotti (2005) and investigates the effects that fiscal policy changes have on private non-residential investment and output in Greece in the post 2000 period. Our primary focus is on private nonresidential investment because according to both the EU-IMF financed Economic Adjustment Programme for Greece and the new three year EMS programme they are expected to be one of the key drivers of economic recovery in Greece in the years to come. Besides examining the direct effects of fiscal policy on investment and output, building on the studies by Ardagna (2010), Laubach (2009), Schuknecht et al. (2010), Barrell et al. (2012) and Bachmann and Sims (2012), we examine the widely held view that improvements in fiscal conditions will contribute to the investment and output recovery through the confidence channel. We find that a fiscal consolidation driven primarily by tax hikes will depress private investment and output to a greater extent compared to a government spending-based fiscal consolidation. Moreover, the negative effects of a tax burden shock on output and private investment are much more protracted. Increases in direct household taxes, other indirect taxes and cuts in the government wage bill, and in government investment followed by increases in direct business taxes, and cuts in intermediate consumption have the most sizeable and long lasting negative effects on output. Tax hikes in direct household and business taxes and in other indirect taxes, followed by cuts in government intermediate consumption, investment, wage bill and current transfers have the most sizeable and protracted negative effect on investment. Nevertheless, there is some evidence of crowding-in of private investment following cuts in the government wage bill (but only on impact) and government transfers (except on impact). Our findings show that a government spending-based-fiscal consolidation (in particular focusing on current transfers and the wage bill) improves financial markets, as well as economic sentiment and lowers government bond spreads. This confidence channel mitigates the negative effects of fiscal consolidation on private investment and output growth inducing a more rapid recovery. On the contrary, a tax hike (in particular increase in direct business taxes and indirect taxes) generates a 40

This exercise is performed using the 4 variable SVAR without investment.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

52

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

negative confidence effect which magnifies the adverse effects of tax based fiscal consolidation on both private investment and output. As part of robustness exercise we examined the effects of fiscal policy in periods of expansions and contractions. Overall, in the more recent recessionary years a fiscal consolidation that relies primarily on expenditure containment is found to be more preferable compared to one that relies on tax increases. This is due to the positive confidence effect it generates, which mitigates the negative effect of fiscal consolidation and facilitates the recovery (i.e., the impulse responses return faster to the zero line). Although an expenditure based consolidation is more preferable we find that it has more pronounced negative effects on output during periods of recession rather than on expansions. On the contrary, the opposite is true in the case of taxes. Nevertheless, the effect of the tax shock could partly be explained by the likely of underestimation of tax elasticities in recent recessionary years. The decline in private investment following a fiscal policy shock (expenditure cut or tax hike) is smaller in the more recent recessionary years. This implies that fiscal consolidation on its own cannot explain the recent fall in private investment spending in Greece. Instead other sources, such as expectations, financing constraints, fears of currency denominations, and delays in reforms shaped the response of private investment spending. The policy message that comes out of the analysis is that a fiscal consolidation programme is more likely to generate positive confidence effects when it is primarily based on government spending containment. Tax-based consolidation efforts (in most cases) worsen economic sentiment and financial markets performance magnifying the negative effects of fiscal consolidation. Hence, given the presence of the abovementioned positive confidence effects policy makers should lean towards a frontloaded fiscal consolidation effort in order for these beneficial effects to materialize sooner rather than later. Nevertheless, delays or backtracking in the implementation of the fiscal adjustment and other structural policy changes, as well as institutional weaknesses that lead to delays in the decision making process at the EU level (i.e., on the eventual resolution of the Greek debt crisis and the public debt overhang problem) could possibly impact negatively on the functioning of the confidence channel postponing or negating the positive confidence effects of fiscal consolidation.41

Acknowledgments We would like to thank Gilles Dufrénot, Valerie Mignon and three anonymous reviewers for their useful comments and suggestions. The views of the paper are our own and do not necessarily reflect those of the Bank of Greece. All remaining errors are ours.

Appendix A. Elasticity assumptions The government spending and revenue elasticity assumptions in (2) still hold in the case of the SVAR specifications examining the output and investment effects of government purchases, government wage-bill, and government investment. In case of government transfers we assume that the output and investment elasticity assumptions with respect to the tax revenue variable still hold, however, we now assume that the output elasticity of government transfers is set to αgy ¼  0.16 (the investment elasticity of government transfers is set to αgi ¼ 0); i.e., we assume that government transfers responds only to output contemporaneously. 42 41 According to a study by Bi et al. (2013) a series of factors would determine the final effect of fiscal consolidation effort. As the authors point out: “the composition of the fiscal consolidation, its duration, the monetary policy stance, the level of government debt, and expectations over the likelihood and composition of fiscal consolidations all matter in determining the extent to which a given consolidation is expansionary or successful in stabilising government debt.”

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

53

In the case of the SVAR specifications examining the output effects of each individual revenue specification, the output elasticity of direct household taxes is set to 2.22, the output elasticity of direct business taxes is set to 1.90, and the output elasticity of VAT and other indirect tax revenue is set to 1 based on the elasticities estimated by Price et al. (2014) and Tagkalakis (2015a, 2015b).43 The investment elasticity of each tax revenue component is set to 0. B. Data information The macroeconomic and fiscal data used in the analysis extent over the period 2000 Q1–2014 Q1 and were all taken from the International Financial Statistics of the International Monetary Fund (IMF, 2014), the OECD Economic Outlook (OECD, 2014), and European Commission (2014).44 The fiscal variables used are: the debt to GDP ratio, government spending (excluding interest payments and other capital transfers), government wage bill (compensation of employees), current government transfers, tax burden (which is the sum of direct, indirect, capital taxes and social security contributions), net taxes (which equals tax burden minus current transfers) direct household and business taxes, VAT and other indirect taxes. The government spending and revenue, GDP, private non-residential investment, share prices and economic sentiment variables were first expressed in logarithmic terms (log) and next in log differences. Fiscal and macroeconomic variables were expressed in real terms after subtracting the log of the GDP deflator. The EAP dummy variable takes value 1 from 2010 Q1 (i.e., the time that Greece lost access to market access which initiated the talks for the adoption of the EU-IMF funded Economic Adjustment Programme for Greece in May 2010) till 2014 Q1.

Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi. org/10.1016/j.inteco.2016.03.003.

References Afonso A, Sousa R. The macroeconomic effects of fiscal policy in Portugal: a Bayesian SVAR analysis. Port Econ J 2011;10:61–82. Alesina A, Ardagna S, Perotti R, Schiantarelli F. Fiscal policy, profits, and investment. Am Econ Rev 2002;92:571–89. Ardagna S. Fiscal policy composition, public debt and economic activity. Pub Ch 2001;109:301–25. Ardagna S S. Financial markets' behavior around episodes of large changes in the fiscal stance. Eur Econ Rev 2010;53:37–55. Arghyrou M, Kontonikas A. The EMU sovereign-debt crisis: fundamentals, expectations and contagion. European Commission. Eur Econ – Econ Pap 2011;436. Arnold JM, Brys B, Heady C, Johansson A, Schwellnus C, Vartia L. Tax policy for economic recovery and growth. Econ J 2011;121:59–80. Attinasi MG, Checherita C, Nickel C. What explains the surge in euro area sovereign spreads during the financial crisis of 2007– 09? ECB Working Paper Series 1131; 2009. Auerbach AJ, Gorodnichenko Y. Measuring the output responses to fiscal policy. Am Ec J: Econ Pol 2012;4:1–27. Bachmann R, Sims ER. Confidence and the transmission of government spending shocks. J Mon Econ 2012;59:235–49. Barrell R, Holland D, Hurst I. Fiscal multipliers and prospects for consolidation. OECD J: OECD Econ Stud 2012;1:71–102. Barsky R, Sims E. Information, animal spirits, and the meaning of innovations in consumer confidence. Am Econ R 2012;102:1343–77. Bernoth K, Erdogan B. Sovereign bond yield spreads: a time-varying coefficient approach. J Int Mon Fin 2012;31:639–56. Bi H, Leeper EM, Leith C. Uncertain fiscal consolidations. Econ J 2013;123:31–63.

42 This is obtained as follows: we multiply the elasticity of unemployment with respect to the output gap (it is  3.15 from Price et al. (2014)) with the share of unemployment spending on government transfers which is around 5%. 43 Price et al. (2014) have estimated revenue and expenditure elasticities with respect to output gap for OECD countries. These elasticities are used to construct the cyclically adjusted balances – i.e., to clean the effect of the cycle from revenue, expenditure and primary and overall budget balances. The output elasticity of personal income tax is 2.22, of corporate income tax is 1.90, of social security contributions is 0.58, and of indirect taxes is 1. Using these and the sample averages of the shares of each tax category we set αty ¼0.94 in specification (2). 44 To correct for seasonal patterns in the quarterly data we have applied the census X12 filter.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i

54

M.Th. Kasselaki, A.O. Tagkalakis / International Economics ∎ (∎∎∎∎) ∎∎∎–∎∎∎

Biau O, Girard E. Politique budgetaire et dynamique economique en France: l’approche VAR structurel. Econ Et Prev 2005;169– 171:1–24. Blanchard O, Perotti R. An empirical characterization of the dynamic effects of government spending and taxes on output. Quar J Econ 2002;117:1329–68. Bouthevillain C, Dufrenot G. Are the effects of fiscal changes difference in times of crisis and non crisis? The French case Rev d’ Econ Pol 2011;121:371–407. Burnside C, Eichenbaum M, Fischer J. Fiscal shocks and their consequences. J Econ Th 2004;115:89–117. Burriel P, de Castro F, Garrote D, Gordo E, Paredes J, Perez JJ. Fiscal policy shocks in the euro area: an empirical assessment. ECB Work Pap 2009;1133. Coenen G, Straub R. Does government spending crowd in private consumption? theory and empirical evidence for the Euro area Int Fin 2005;8:435–70. De Grauwe P, Ji Y. Mispricing of sovereign risk and multiple equilibria in the Eurozone. CEPS Work Doc 2012;361. Edelberg W, Eichenbaum M, Fischer J. Understanding the effects of a shock to government purchases. Rev Econ Dyn 1999;2:166–206. European Commission, The Economic Adjustment Programme for Greece, Occasional Paper 61, Brussels; 2010. European Commission, Economic Forecasts, autumn, European Commission, Brussels; 2014. European Commission, Economic Forecasts, spring, European Commission, Brussels; 2015. Favero C, Giavazzi F. Debt and the effects of fiscal policy. Working Paper. NBER; 2007. Favero C, Giavazzi F. Measuring tax multipliers: the narrative method in fiscal VARs. Am Econ J: Econ Pol 2012;4:69–94. Fatás A, Mihov I. The effects of fiscal policy on consumption and employment: theory and evidence. CEPR Discuss Pap 2001;2760. Forni L, Monteforte L, Sessa L. The general equilibrium effects of fiscal policy: estimates for the Euro area. J Pub Econ 2009;93:559–85. Galí J, López-Salido JD, Vallés J. Understanding the effects of government spending on consumption. J Eur Econ Assoc 2007;5:227–70. Giordano R, Momigliano S, Neri S, Perotti R. The effects of fiscal policy in Italy: evidence from a VAR model. Eur J Pol Econ 2007;23:707–33. IMF, International Financial Statistics, IMF, Washington; 2014. Konstantinou P, Tagkalakis A. Boosting confidence: is there a role for fiscal policy? Econ Mod 2011;28:1629–41. Kuismanen M, Kämppi V. The effects of fiscal policy on economic activity in Finland. Econ Mod 2010;27:1315–23. Lane PR, Perotti R. The importance of composition of fiscal policy: evidence from different exchange rate regimes. J Pub Econ 2003;87:2253–79. Laubach T. New evidence on the interest rate effects of budget deficits and debt. J Eur Econ Assoc 2009;7:858–85. Morris R, Schucknecht L. Structural balances and revenue windfalls: the role of asset prices. ECB Work Pap 2007;737. Mountford A, Uhlig H. What are the effects of fiscal policy shocks? J Appl Econ 2009;24:960–92. Liu EX, Mattina T, Poghosyan T. Corecting “beyond the cycle:’ accounting for asset prices in structural fiscal balances. Working paper. IMF 2015;1009. OECD, Economic Outlook, No 96. OECD, Pari; 2014. Perotti R. Estimating the effects of fiscal policy in OECD countries. Proceedings. Federal Reserve Bank of San Francisco; 2005. Poghosyan T. Assessing the variability of tax elasticities in Lithuania. IMF Work Pap 2011;270. Price R, Dang Th Th. Adjusting fiscal balances for asset price cycles. Working paper. . OECD, Economics Department; 2011. Price R, Dang Th Th, Guillemette Y. New tax and expenditure elasticity estimates for EU budget surveillance. Working Paper. . OECD Economics Department; 2014. Ramey V, Shapiro M. Costly capital reallocation and the effects of government spending. Carn -Roch Conf S Pub Pol 1998;48:145–94. Ramey V. Identifying government spending shocks: it's all in the timing Quar. J Econ 2011;126:1–50. Romer CD, Romer DH. The macroeconomic effects of tax changes: estimates based on a new measure of fiscal shocks. Am Econ Rev 2010;100:763–801. Sancak C, Velloso R, Xing J. Tax revenue response to the business cycle. IMF Work Pap 2010;71. Schuknecht L, von Hagen J, Wolswijk G. Government bond risk premiums in the EU revisited: the impact of the financial crisis. ECB Work Pap Ser 2010;1152. Tagkalakis A. Asset price volatility and government revenue. Econ Mod 2011;28:2532–43. Tagkalakis A. Credit and the transmission of fiscal policy shocks: recent evidence from Greece,. Econ Let 2014a;122:263–7. Tagkalakis A. Assessing the variability of indirect tax elasticity in Greece. Econ Bull 2014b;34:1762–75. Tagkalakis A. Estimating the elasticity of personal income tax to gross earnings from income tax statements in Greece: implications for tax buoyancy. Pub Fin Manag 2015a;15:47–65. Tagkalakis A. Estimating the elasticity of corporate income tax to output gap in Greece: implications for tax buoyancy. Pub Fin Rev 2015b:1–27. Traum N, Yang SCS. When does government debt crowd out investment? J Appl Econom 2016, http://dxdoi.org/10.1002/ jae.2356 Forthcoming. Yang SCS. Quantifying tax effects under policy foresight. J Mon Econ 2005;52:1557–68.

Please cite this article as: Kasselaki MTh, Tagkalakis AO. Fiscal policy and private investment in Greece. International Economics (2016), http://dx.doi.org/10.1016/j.inteco.2016.03.003i