HOW POLICY ACTIONS AFFECT SHORT-TERM POST-CRISIS RECOVERY?

Branimir Jovanovic (Tor Vergata, Rome)

BEM, June 2012, Bratislava

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 1 / 20 Many papers investigated why - Berglof et al. (2009), Berkmen et al. (2009), Giannone et al. (2010), Lane and Milesi-Ferretti (2009), Rose and Spiegel (2009), Blanchard et al. (2010a), IMF (2010), Crespo and Feldkircher (forthcoming)... Right now, it is obvious that di erent countries are recovering di erently from the crisis. Still, no research on the issue.

Background

Di erent countries were a ected in di erent ways by the Great Recession.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 2 / 20 Right now, it is obvious that di erent countries are recovering di erently from the crisis. Still, no research on the issue.

Background

Di erent countries were a ected in di erent ways by the Great Recession. Many papers investigated why - Berglof et al. (2009), Berkmen et al. (2009), Giannone et al. (2010), Lane and Milesi-Ferretti (2009), Rose and Spiegel (2009), Blanchard et al. (2010a), IMF (2010), Crespo and Feldkircher (forthcoming)...

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 2 / 20 Still, no research on the issue.

Background

Di erent countries were a ected in di erent ways by the Great Recession. Many papers investigated why - Berglof et al. (2009), Berkmen et al. (2009), Giannone et al. (2010), Lane and Milesi-Ferretti (2009), Rose and Spiegel (2009), Blanchard et al. (2010a), IMF (2010), Crespo and Feldkircher (forthcoming)... Right now, it is obvious that di erent countries are recovering di erently from the crisis.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 2 / 20 Background

Di erent countries were a ected in di erent ways by the Great Recession. Many papers investigated why - Berglof et al. (2009), Berkmen et al. (2009), Giannone et al. (2010), Lane and Milesi-Ferretti (2009), Rose and Spiegel (2009), Blanchard et al. (2010a), IMF (2010), Crespo and Feldkircher (forthcoming)... Right now, it is obvious that di erent countries are recovering di erently from the crisis. Still, no research on the issue.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 2 / 20 It will investigate which factors a ect short-term GDP growth after economic crises. It will focus on the policy actions:

1 Monetary policy 2 Fiscal policy 3 Banking regulation and supervision policy

This paper...

This paper will do that.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 3 / 20 It will focus on the policy actions:

1 Monetary policy 2 Fiscal policy 3 Banking regulation and supervision policy

This paper...

This paper will do that. It will investigate which factors a ect short-term GDP growth after economic crises.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 3 / 20 1 Monetary policy 2 Fiscal policy 3 Banking regulation and supervision policy

This paper...

This paper will do that. It will investigate which factors a ect short-term GDP growth after economic crises. It will focus on the policy actions:

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 3 / 20 2 Fiscal policy 3 Banking regulation and supervision policy

This paper...

This paper will do that. It will investigate which factors a ect short-term GDP growth after economic crises. It will focus on the policy actions:

1 Monetary policy

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 3 / 20 3 Banking regulation and supervision policy

This paper...

This paper will do that. It will investigate which factors a ect short-term GDP growth after economic crises. It will focus on the policy actions:

1 Monetary policy 2 Fiscal policy

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 3 / 20 This paper...

This paper will do that. It will investigate which factors a ect short-term GDP growth after economic crises. It will focus on the policy actions:

1 Monetary policy 2 Fiscal policy 3 Banking regulation and supervision policy

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 3 / 20 The scal policy e ect is even stronger in highly-leveraged countries In other words, public saving during crises hurts post-crisis recoveries substantially

This paper... Findings

All three policies have signi cant positive in uence on recovery

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 4 / 20 In other words, public saving during crises hurts post-crisis recoveries substantially

This paper... Findings

All three policies have signi cant positive in uence on recovery The scal policy e ect is even stronger in highly-leveraged countries

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 4 / 20 This paper... Findings

All three policies have signi cant positive in uence on recovery The scal policy e ect is even stronger in highly-leveraged countries In other words, public saving during crises hurts post-crisis recoveries substantially

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 4 / 20 Park and Lee (2003) - which factors in uence short-term growth after 176 currency crises? Depreciation of exchange rate and supportive monetary and scal policy. Endogeneity problems - output and policy variables measured over the same period. Hong and Tornell (2005) - recoveries from 100 currency crises. Credit expansion and reserve inadequacy before the crisis depress growth after the crisis. No clear relationship between policy variables and recovery. Again, endogeneity - policy variables contemporaneous with recovery. Abiad et al. (2009) - medium-term output after 88 banking crises. Expansionary scal policy correlated with higher growth, but not the monetary policy. Again, endogeneity - policy variables measured before the growth, but they both refer to post-crisis period.

Existing literature Existing literature

Very few papers investigate determinants of post-crisis recoveries.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 5 / 20 Hong and Tornell (2005) - recoveries from 100 currency crises. Credit expansion and reserve inadequacy before the crisis depress growth after the crisis. No clear relationship between policy variables and recovery. Again, endogeneity - policy variables contemporaneous with recovery. Abiad et al. (2009) - medium-term output after 88 banking crises. Expansionary scal policy correlated with higher growth, but not the monetary policy. Again, endogeneity - policy variables measured before the growth, but they both refer to post-crisis period.

Existing literature Existing literature

Very few papers investigate determinants of post-crisis recoveries. Park and Lee (2003) - which factors in uence short-term growth after 176 currency crises? Depreciation of exchange rate and supportive monetary and scal policy. Endogeneity problems - output and policy variables measured over the same period.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 5 / 20 Abiad et al. (2009) - medium-term output after 88 banking crises. Expansionary scal policy correlated with higher growth, but not the monetary policy. Again, endogeneity - policy variables measured before the growth, but they both refer to post-crisis period.

Existing literature Existing literature

Very few papers investigate determinants of post-crisis recoveries. Park and Lee (2003) - which factors in uence short-term growth after 176 currency crises? Depreciation of exchange rate and supportive monetary and scal policy. Endogeneity problems - output and policy variables measured over the same period. Hong and Tornell (2005) - recoveries from 100 currency crises. Credit expansion and reserve inadequacy before the crisis depress growth after the crisis. No clear relationship between policy variables and recovery. Again, endogeneity - policy variables contemporaneous with recovery.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 5 / 20 Existing literature Existing literature

Very few papers investigate determinants of post-crisis recoveries. Park and Lee (2003) - which factors in uence short-term growth after 176 currency crises? Depreciation of exchange rate and supportive monetary and scal policy. Endogeneity problems - output and policy variables measured over the same period. Hong and Tornell (2005) - recoveries from 100 currency crises. Credit expansion and reserve inadequacy before the crisis depress growth after the crisis. No clear relationship between policy variables and recovery. Again, endogeneity - policy variables contemporaneous with recovery. Abiad et al. (2009) - medium-term output after 88 banking crises. Expansionary scal policy correlated with higher growth, but not the monetary policy. Again, endogeneity - policy variables measured before the growth, but they both refer to post-crisis period.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 5 / 20 Endogeneity Endogeneity, i.e. reverse causality

When regressing output on scal policy, for example, positive coecient can mean that scal policy a ects output positively, but also that higher output allows more supportive policy. That would bias the policy coecient upwards. Even when scal policy is dated before the output, positive coecient can be obtained because policy makers increase spending today in anticipation of higher output in the future...

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 6 / 20 Endogeneity How to avoid it?

But, in times of crises policy makers do not anticipate higher output in the future. They just don't expect output to start growing soon. So, policy during crises does not depend on output after crises. Hence, by measuring policy during crises, when policy makers do not expect output to start growing soon, reverse causality will be avoided.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 7 / 20 The model The model

recovery = c1 + c2 IR + c3 money + c4 NER + c5 cap adeq     + c6 gov cons + c7 pub debt + c8 ext debt    + c9 FDI + c10 exp orts + c11 portfolio    Recovery is de ned as average annualized GDP growth in the 3 years after the end of the crisis, minus the average GDP growth since 1980 (or since there are GDP data). In this way all factors that cause di erent countries to have di erent average growth rates (omitted variables) are wiped out. The remaining variation can be attributed to the policy variables (or other short-term factors, like the capital ows). Since no omitted variables and no reverse causality, can be consistently estimated by OLS.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 8 / 20 The model

Five groups of explanatory variables:

1 variables measuring monetary policy (IR, money, NER) - measured during crises, to avoid reverse causality; 2 variable measuring the scal policy (government consumption) - measured during crises; 3 variable measuring the banking supervision policy (capital adequacy) - measured before crises, to avoid worsening in adequacy due to crises; 4 variables measuring the constraints for policy action (public debt and external debt) - measured before crises; 5 variables measuring short-term trade or nancial ows that can a ect short-run post-crisis recovery (FDI, exports and portfolio investment ows) - measured during recovery; 6 Three dummies - for high public debt, for currency crises and for the 2008 crisis.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 9 / 20 Crises Crises

Taken from one of the most cited papers in the literature, Laeven and Valencia (2008). Only crises for countries with available quarterly GDP taken.

Argentina 1994, 1998, 2008 Korea 1998, 2008 Belarus 1995, 2008 Malaysia 1998, 2008 1996, 2008 Mexico 1982, 1995, 2008 Chile 1981, 2008 Norway 1991, 2008 Colombia 1998 Peru 1982, 1988, 2008 1998, 2008 Philippines 1983, 1998, 2008 1997, 2008 Russia 1998, 2008 Finland 1990, 2008 1998, 2008 Indonesia 1998 Sweden 1991, 2008 Jamaica 1996, 2008 Thailand 1997, 2008 Japan 1997, 2008 Turkey 1994, 2001, 2008

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 10 / 20 Crises

Crises dated similarly to the conventional peak-to-trough logic: a crisis starts when GDP falls for two consecutive quarters; a crisis ends when GDP records the rst positive growth rate after which there are no consecutive declines. The 12 quarters after a crisis ends is the recovery period. Example: Argentina 1994 crisis

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 11 / 20 Crises Argentinian GDP 1991-1998

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 12 / 20 Crises The 1994 crisis

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 13 / 20 Crises The recovery after the 1994 crisis

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 14 / 20 General-to-speci c approach used, popularised by David Hendry. Starting with a model that includes all the potential determinants of recovery, we exclude the most insigni cant variables one by one, until we reach a parsimonious model.

Data and methodology Data and methodology

Data sources are standard - IMF's IFS, World Bank's WDI and central bank

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 15 / 20 Starting with a model that includes all the potential determinants of recovery, we exclude the most insigni cant variables one by one, until we reach a parsimonious model.

Data and methodology Data and methodology

Data sources are standard - IMF's IFS, World Bank's WDI and central bank General-to-speci c approach used, popularised by David Hendry.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 15 / 20 Data and methodology Data and methodology

Data sources are standard - IMF's IFS, World Bank's WDI and central bank General-to-speci c approach used, popularised by David Hendry. Starting with a model that includes all the potential determinants of recovery, we exclude the most insigni cant variables one by one, until we reach a parsimonious model.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 15 / 20 Regression results -1- -2- -3- -4- -5- Constant -1.306 -2.20 -1.15 -1.194 -1.259 (0.756) (-1.239) (-1.083) (-1.148) (-1.333) IR -0.35 -0.25 -0.47* -0.464* -0.505** (0.459) (-0.97) (-1.771) (-1.88) (-2.13) money 0.073 0.078* 0.048* 0.048** 0.048* (0.171) (1.913) (2.025) (2.066) (1.978) NER 0.003 0.002 0.003 0.003 0.004* (0.56) (0.47) (1.104) (1.099) (1.817) cap adeq 0.40* 0.39* 0.42*** 0.42*** 0.432*** (0.07) (2.192) (3.001) (3.036) (3.259) gov cons 0.061 0.047 0.123** 0.124** 0.121** (0.526) (0.563) (2.178) (2.265) (2.289) pub debt 0.19 0.21 (0.625) (0.591) ext debt -0.015 (0.763) FDI 0.12 0.208 0.24* 0.242* 0.245** (0.696) (1.164) (1.823) (1.846) (2.232) exports 0.076 0.10 -0.006 (0.55) (1.497) (-0.117) portfolio 0.381 0.49 0.11 0.111 (0.546) (1.433) (0.459) (0.491) hi pub debt*gov cons 0.191 0.179 0.266 0.259 0.29* (0.535) (0.63) (1.341) (1.336) (1.86) curr crisis*IR 0.409 0.31 0.54** 0.531** 0.565** (0.375) (1.225) (2.066) (2.202) (2.435) dum 2008 -1.23 -0.86 -2.33** -2.327** -2.246** (0.504) (-0.725) (-2.191) (-2.225) (-2.546)

R2 0.73 0.72 0.53 0.52 0.53 Observations 21 21 36 36 37 T statistics in parentheses. *, ** and *** denote signi cance at 1%, 5% and 10%

2 Regression results -1- -2- -3- -4- -5- Constant -1.306 -2.20 -1.15 -1.194 -1.259 (0.756) (-1.239) (-1.083) (-1.148) (-1.333) IR -0.35 -0.25 -0.47* -0.464* -0.505** (0.459) (-0.97) (-1.771) (-1.88) (-2.13) money 0.073 0.078* 0.048* 0.048** 0.048* (0.171) (1.913) (2.025) (2.066) (1.978) NER 0.003 0.002 0.003 0.003 0.004* (0.56) (0.47) (1.104) (1.099) (1.817) cap adeq 0.40* 0.39* 0.42*** 0.42*** 0.432*** (0.07) (2.192) (3.001) (3.036) (3.259) gov cons 0.061 0.047 0.123** 0.124** 0.121** (0.526) (0.563) (2.178) (2.265) (2.289) pub debt 0.19 0.21 (0.625) (0.591) ext debt -0.015 (0.763) FDI 0.12 0.208 0.24* 0.242* 0.245** (0.696) (1.164) (1.823) (1.846) (2.232) exports 0.076 0.10 -0.006 (0.55) (1.497) (-0.117) portfolio 0.381 0.49 0.11 0.111 (0.546) (1.433) (0.459) (0.491) hi pub debt*gov cons 0.191 0.179 0.266 0.259 0.29* (0.535) (0.63) (1.341) (1.336) (1.86) curr crisis*IR 0.409 0.31 0.54** 0.531** 0.565** (0.375) (1.225) (2.066) (2.202) (2.435) dum 2008 -1.23 -0.86 -2.33** -2.327** -2.246** (0.504) (-0.725) (-2.191) (-2.225) (-2.546)

R2 0.73 0.72 0.53 0.52 0.53 Observations 21 21 36 36 37 T statistics in parentheses. *, ** and *** denote signi cance at 1%, 5% and 10%

2 Regression results -1- -2- -3- -4- -5- Constant -1.306 -2.20 -1.15 -1.194 -1.259 (0.756) (-1.239) (-1.083) (-1.148) (-1.333) IR -0.35 -0.25 -0.47* -0.464* -0.505** (0.459) (-0.97) (-1.771) (-1.88) (-2.13) money 0.073 0.078* 0.048* 0.048** 0.048* (0.171) (1.913) (2.025) (2.066) (1.978) NER 0.003 0.002 0.003 0.003 0.004* (0.56) (0.47) (1.104) (1.099) (1.817) cap adeq 0.40* 0.39* 0.42*** 0.42*** 0.432*** (0.07) (2.192) (3.001) (3.036) (3.259) gov cons 0.061 0.047 0.123** 0.124** 0.121** (0.526) (0.563) (2.178) (2.265) (2.289) pub debt 0.19 0.21 (0.625) (0.591) ext debt -0.015 (0.763) FDI 0.12 0.208 0.24* 0.242* 0.245** (0.696) (1.164) (1.823) (1.846) (2.232) exports 0.076 0.10 -0.006 (0.55) (1.497) (-0.117) portfolio 0.381 0.49 0.11 0.111 (0.546) (1.433) (0.459) (0.491) hi pub debt*gov cons 0.191 0.179 0.266 0.259 0.29* (0.535) (0.63) (1.341) (1.336) (1.86) curr crisis*IR 0.409 0.31 0.54** 0.531** 0.565** (0.375) (1.225) (2.066) (2.202) (2.435) dum 2008 -1.23 -0.86 -2.33** -2.327** -2.246** (0.504) (-0.725) (-2.191) (-2.225) (-2.546)

R2 0.73 0.72 0.53 0.52 0.53 Observations 21 21 36 36 37 T statistics in parentheses. *, ** and *** denote signi cance at 1%, 5% and 10%

2 Regression results -1- -2- -3- -4- -5- Constant -1.306 -2.20 -1.15 -1.194 -1.259 (0.756) (-1.239) (-1.083) (-1.148) (-1.333) IR -0.35 -0.25 -0.47* -0.464* -0.505** (0.459) (-0.97) (-1.771) (-1.88) (-2.13) money 0.073 0.078* 0.048* 0.048** 0.048* (0.171) (1.913) (2.025) (2.066) (1.978) NER 0.003 0.002 0.003 0.003 0.004* (0.56) (0.47) (1.104) (1.099) (1.817) cap adeq 0.40* 0.39* 0.42*** 0.42*** 0.432*** (0.07) (2.192) (3.001) (3.036) (3.259) gov cons 0.061 0.047 0.123** 0.124** 0.121** (0.526) (0.563) (2.178) (2.265) (2.289) pub debt 0.19 0.21 (0.625) (0.591) ext debt -0.015 (0.763) FDI 0.12 0.208 0.24* 0.242* 0.245** (0.696) (1.164) (1.823) (1.846) (2.232) exports 0.076 0.10 -0.006 (0.55) (1.497) (-0.117) portfolio 0.381 0.49 0.11 0.111 (0.546) (1.433) (0.459) (0.491) hi pub debt*gov cons 0.191 0.179 0.266 0.259 0.29* (0.535) (0.63) (1.341) (1.336) (1.86) curr crisis*IR 0.409 0.31 0.54** 0.531** 0.565** (0.375) (1.225) (2.066) (2.202) (2.435) dum 2008 -1.23 -0.86 -2.33** -2.327** -2.246** (0.504) (-0.725) (-2.191) (-2.225) (-2.546)

R2 0.73 0.72 0.53 0.52 0.53 Observations 21 21 36 36 37 T statistics in parentheses. *, ** and *** denote signi cance at 1%, 5% and 10%

2 Regression results -1- -2- -3- -4- -5- Constant -1.306 -2.20 -1.15 -1.194 -1.259 (0.756) (-1.239) (-1.083) (-1.148) (-1.333) IR -0.35 -0.25 -0.47* -0.464* -0.505** (0.459) (-0.97) (-1.771) (-1.88) (-2.13) money 0.073 0.078* 0.048* 0.048** 0.048* (0.171) (1.913) (2.025) (2.066) (1.978) NER 0.003 0.002 0.003 0.003 0.004* (0.56) (0.47) (1.104) (1.099) (1.817) cap adeq 0.40* 0.39* 0.42*** 0.42*** 0.432*** (0.07) (2.192) (3.001) (3.036) (3.259) gov cons 0.061 0.047 0.123** 0.124** 0.121** (0.526) (0.563) (2.178) (2.265) (2.289) pub debt 0.19 0.21 (0.625) (0.591) ext debt -0.015 (0.763) FDI 0.12 0.208 0.24* 0.242* 0.245** (0.696) (1.164) (1.823) (1.846) (2.232) exports 0.076 0.10 -0.006 (0.55) (1.497) (-0.117) portfolio 0.381 0.49 0.11 0.111 (0.546) (1.433) (0.459) (0.491) hi pub debt*gov cons 0.191 0.179 0.266 0.259 0.29* (0.535) (0.63) (1.341) (1.336) (1.86) curr crisis*IR 0.409 0.31 0.54** 0.531** 0.565** (0.375) (1.225) (2.066) (2.202) (2.435) dum 2008 -1.23 -0.86 -2.33** -2.327** -2.246** (0.504) (-0.725) (-2.191) (-2.225) (-2.546)

R2 0.73 0.72 0.53 0.52 0.53 Observations 21 21 36 36 37 T statistics in parentheses. *, ** and *** denote signi cance at 1%, 5% and 10%

2 Interpretation Interpretation

Public debt, surprisingly, insigni cant. Maybe due to insucient observations? From the short-term ows, only FDI signi cant. Policy variables signi cant, both statistically and economically. INTEREST RATE - if interest rate decreases by 6 percentage points (p.p.) during the crisis (move from the 25th percentile to the 75th percentile of the variable), this stimulates post-crisis GDP growth by roughly 3 p.p., on average, ceteris paribus. Cross product of CURRENCY CRISES dummy and interest rate positive and signi cant, suggesting that in these cases it is optimal to increase the interest rate, not to decrease it, to defend the currency. MONEY - expanding money supply by 23 percent (move from the 25th percentile to the 75th percentile) leads to higher GDP growth by roughly 1 p.p. Since average GDP growth is 1.6, these e ects are very strong.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 16 / 20 Interpretation

EXCHANGE RATE - depreciation of the currency by 45 percent (move from 25th to 75th percentile) leads to a 0.2 p.p. higher growth after the crisis. CAPITAL ADEQUACY - higher capital/assets ratio by 3.5 percentage points (move from 25th to 75th percentile), increases post-crisis recovery by roughly 1.4 percentage points. GOVERNMENT CONSUMPTION - when government consumption growth is higher by 11 p.p. (move from the 25th percentile to the 75th), this leads to 1.3 p.p. higher post-crisis growth. Cross product of HIGH PUBLIC DEBT dummy and government consumption positive and signi cant, suggesting that scal multiplier is even stronger in highly leveraged countries In other words, if 11 p.p. lower government consumption growth leads to a roughly 1.3 p.p. lower output growth in normal countries, it leads to a 4.5 p.p. lower growth in highly indebted countries! That is, insisting on public saving during crises creates substantial damage to the short-run post-crisis GDP growth.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 17 / 20 Interpretation

Dummy for 2008 crisis signi cant and negative - global recovery slower for this crisis No structural break for the 2008 crisis - cross products of the 2008 dummy with the other variables insigni cant

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 18 / 20 Future extension Future extension

Only 37 observations. Sample should be increased. First, by lling in some existing data gaps (refering primarily to the capital adequacy data). Then, by including also other crises from the 80's and 90's. Problem - no quarterly GDP data. Solution - interpolate it. Fiscal stimulus during crises comes mainly from public investment, not consumption. Hence, include public investment as additional explanatory variable. Finally, incorporate medium-term growth issues. The dilemma that policy makers are facing right now is whether e orts to stimulate short-run growth can endanger medium-term growth prospects.

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 19 / 20 Future extension

Thank you for your attention!

[email protected]

Branimir Jovanovic (Tor Vergata, Rome) () BEM, June 2012, Bratislava 20 / 20