The Real Effect of Financial Crises in The European Transition Economies

Economics of Transition Volume 19(1) 2011, 125 DOI: 10.1111/j.1468-0351.2010.00395.
The real effect of financial crises in the European transition economies1

Davide Furceri* and Aleksandra Zdzienicka**
*OECD; University of Palermo. E-mail: davide.furceri@oecd.org **GATE-CNRS, University of Lyon. E-mail: azdzieni@ens-lsh.fr
Abstract
The aim of this work is to assess the impact of nancial crises on output for 11 European transition economies (CEECs). The results suggest that nancial crises have a signicant and permanent effect, lowering long-term output by about 1217 percent. The effect is larger in smaller countries in which the banking sector presented more important nancial disequilibria. We also found that scal policy has been the most efcient tool in dealing with the crises, whereas the effect of monetary policy has been rather modest. Flexible exchange rates are found to attenuate the impact of the crises in the short and medium term, but to amplify the effect in the long run. International Monetary Fund support is found to moderate the effect in the long run. Finally, the effect in the CEECs is considerably larger than in the EU advanced economies. JEL classifications: E6, G1. Keywords: Output growth, nancial crisis, CEECs.
Received: October 6, 2009; Acceptance: February 26, 2010

1 We would like to thank an anonymous referee, Graciela Kaminski and other participants at the ESRC Development Economics Conference, Manchester (2010). The views expressed in this article are those of the authors and do not necessarily represent those of the OECD or its member countries.
2010 The Authors Economics of Transition 2010 The European Bank for Reconstruction and Development. Published by Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA
Furceri and Zdzienicka
1. Introduction
The current nancial crisis that started to spread over the world in 2007 has led to a rapid and sharp deterioration of economic activity. The severity of the crisis has renewed the discussion about the real economic effects of nancial crises. In the literature, only few studies report a null or modest impact of nancial crises on economic output (Boyd et al., 2005; Hutchison, 2001) and in general these ndings concern developed economies after minor crises.2 In contrast, nancial crises are, in general, associated with severe economic downturns (Reinhart and Rogoff, 2009). Financial crises affect the real economy through sharp currency depreciations and increasing imported factor prices and production costs. Financial crises also affect private agents behaviour, increasing uncertainty about future gains and decreasing the level of investment and consumption. In addition, banking crises reduce investment via the distress of credit intermediation and through the payment system by decreasing collateral values.3 These effects seem to be more important and persistent in emerging countries. In fact, emerging economies are more vulnerable to the factors that lead to crises, such as: banks and private agents exposure to currency and maturity mismatch, disruption in international capital markets, bank panics (Chang and Velasco, 1998) and sudden stops in capital inows (Calvo, 2006). The empirical ndings in the literature corroborate this thesis and suggest that the effect of nancial crises on economic activity in emerging countries is greater and more persistent than in developed economies.4 In this context, the European transition economies (CEECs hereafter) constitute a very interesting group of countries. Indeed, despite increasing similarities with developed economies, they still show important features of emerging markets. In addition, structural reforms, which could enhance the degree of resilience to shocks, are yet to be completed making it more likely that the effects of nancial crises will be greater and long lasting. The main objective of this work is to analyse the effect of nancial crises on real output for 11 CEECs economies and to assess whether output losses are
Some traditional views on the real effect of currency crises suggest that, under the existence of nominal rigidities, a real depreciation increases exports and stimulates employment and output. For example, Demirguc-Kunt et al. (2006) nd a positive impact in the case of 40 percent of the currency crises examined. 3 See, for example, the Financial accelerator theory in Bernanke and Gertler (1989), Kiyotaki and Moore (1997). For a more detailed study on the propagation of crises, see Blot et al. (2009). 4 For example, Hutchison and Noy (2005) analyzing the impact of currency and banking crises on output for a large set of countries nd that while in a sample of 24 emerging economies output losses are about 8 percent and last 2 years, when developed countries are included in the study the reduction in output is about 2 percent and lasts only one year. Similarly, DellAriccia et al. (2008) nd that the real effect of banking crises is about 1.5 percentage points larger for developing countries. Moreover, the recent Reinhart and Rogoffs (2009) historical analysis underlines that a decline in real GDP is greater for emerging than developed countries.
2010 The Authors Economics of Transition 2010 The European Bank for Reconstruction and Development
2
Financial Crises in the European Transition Economies
permanent or if they are made up in the medium and long term. In doing so, we follow the methodology initiated by Romer and Romer (1989) to assess the impact of monetary shocks on output. More precisely, we estimate an ARDL equation (autoregressive with distributed lags) of nancial crises and output growth, and we derive the corresponding impulse response functions (IRFs) for real output. This method completes previous attempts to measure the output cost of the crises by assessing their short and long-term impact. Although the literature has documented that several banking and currency crises are associated with sizeable short-term output losses, few works have analysed whether these losses are made good in the medium and long term. In fact, traditional approaches (initiated by Kaminsky and Reinhart, 1999; Calvo and Reinhart, 2000) consist of output regressions, including several controls, contemporaneous and lagged variables (real growth, real GDP per capital) and nancial crisis dummies in a static panel framework of developed and emerging economies.5 Cerra and Saxena (2008) are the rst, to our knowledge, to add IRFs to distinguish between short and long-term effects of crises on real GDP. The same methodology was used by Furceri and Mourougane (2009) to assess the impact of nancial crises on potential output for a panel of 30 OECD economies, and more recently, by the European Commission to assess the effect of the current nancial crisis on potential output and potential growth (European Commission, 2009). A similar approach has also been adopted by the IMF World Economic Outlook (2009) to assess medium-term output performance and its driving forces following banking crises. Our approach completes and improves existing studies by: 1. Using two different methodologies to identify nancial crises. First, we use a market pressure index (MPI) that takes into account the pressure on the real exchange rate and on foreign exchange reserves (FX), and which accounts for several nancial (mostly currency) crises. Second, we use the recent IMF crisis database (Leaven and Valencia, 2008) to construct a dummy for nancial (banking, debt and currency) crises. Finally, given the importance of the ongoing nancial crisis for CEEC economies, we include it in both indexes as starting in 2008. 2. Comparing the real effects of nancial crises between the CEEC economies and advanced European countries (EU-15 hereafter). 3. Controlling for countries structural heterogeneities and the role of the macro-economic policy response. The rest of the chapter is organized as follows. Section 2 describes methodological and data issues. Section 3 presents the results, and Section 4 the conclusions.
5 See, for example, Milesi-Ferretti and Razin (1998), Barro (2001), Bordo et al. (2001), Hutchison (2001), Hutchison and Noy (2005), Demirguc-Kunt et al. (2006).
2. Methodology and data

To assess the impact and the persistence of nancial crises on output we follow the methodology proposed by Romer and Romer (1989) and adopted by Cerra and Saxena (2008). In particular, the approach consists of estimating an ARDL (4, 4) equation and deriving the relative IRFs: Dyit ai
4 X j1
bj Dyi;tj
4 X j0
dj Di;tj eit ;
where y is the log of real output, D is a dummy variable, which takes a value equal to one in the case of the occurrence of a nancial crisis and zero otherwise, ai are country xed effects to control for unobservable country-specic characteristics and eit is the error term. The numbers of lags has been tested, and the results suggest that the inclusion of up to four lags produces the best specication. We correct for heteroskedasticity, when this exists, using White robust standard errors, while the problem of autocorrelation related to our dependent variable is solved using its lags as explanatory variables. The IRFs are obtained by simulating a 1 year crisis and by computing the response of output over time through the estimated coefcients. In particular, the simultaneous response will be d0, the one-ahead cumulative response will be d0 + (d1 + b1d0) and so on. Then, 95-percent level condence bands are derived via MonteCarlo simulations using 1,000 trials. Since we are also interested in the effect of structural and macroeconomic policy variables on growth performance during times of crises, we re-estimate Equation (1) adding these variables as controls and as interaction with the crisis dummy and the lagged output growth.6 The augmented specication takes the following form: Dyit ai
4 X j1
bj Dyi;tj
4 X j0
dj Di;tj uDyi;t1 Xi;t mXi;t eit ;
where Xi,t stands for structural or macroeconomic policy variables, u and c capture the effects of Xi,t on output persistence and on the direct impact of nancial crises on output, respectively. m captures the impact of control variables on output. In all, the simultaneous impact of the crisis on output as measured by the impulse
6 From a technical point of view, it would have been more appropriate to include the interaction terms for each of the coefcients of Equation (1). This, however, leads to serious problems of multicollinearity without qualitatively changing the results.
response functions will be d0 + cX, the one period-ahead cumulative response will be d0 + cX+d1 + (b1 + uX)[(d]0 + cX)) and so on. To deal with endogeneity issues, since some control variables can be affected by the occurrence of a crisis, we consider their lagged values. In addition, to capture the impact of some macroeconomic policy measures, such as changes in government spending or in interest rates, we focus on their changes in the year after the onset of the crisis. All data are taken from the IMF International Financial Statistics and World Economic Outlook databases, except for government spending and central banks interest rates that are taken from OECD Economic Outlook 85 (2009) and central bank statistics. The dataset consists of an unbalanced panel of annual observations from 1989 to 2008 for 11 CEECs economies (Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia and Slovenia).7 Financial crisis episodes are identied using two approaches.8 The rst is based on the work by Valencia and Leaven (2008), which lists nancial crises (currency, debt and banking crises) for all the countries in our sample from 1970 to 2007. The second approach is based on a MPI. Since nancial crises are usually associated with severe pressure on the domestic currency, we follow the approach proposed by Kaminsky et al. (1998) and we adopt a criterion for the denition of a crisis that accounts for pressures occurring in the real effective exchange rate (REER) market and in ofcial FX.9 In particular, the crisis is considered to occur when the market pressure indicator, dened as x1DREER ) x2DFX,10 is d standard deviations above its mean. We adopt a gridsearch methodology which consists of seeking a value of d in the interval [0, 3]11 which best reproduces the historical crises of the countries examined. Using this methodology, we nd that the value of 1.5 correctly identies major crises in the CEECs.12 For both indexes, the current nancial crisis is assumed to have taken place in 2008.
The EU-15 countries included in the analysis for comparison are: Austria, Belgium, Denmark, France, Finland, Germany, Greece, Italy, Ireland, Luxembourg, the Netherlands, Portugal, Spain, Sweden and the United Kingdom. 8 See Table 1 for a detailed description of crisis episodes. 9 The reason we use the real exchange rate rather than the nominal one is that it gives a more realistic view of crisis episodes in the CEEC economies since these countries have adopted many different exchange rate regimes since the beginning of their transition process. We use REER and ofcial FX reserves quarterly data from the IMF International Financial Statistics database. 10 For a detailed description of the indicator and the identication of crisis episodes, see Andreou et al. (2009). 11 The conventional choice of d is between 2.5 and 3 (Edison, 2000). 12 Using the same method, we set the value d at 2 for the EU-15 countries.
Table 1. Financial crises episodes

MPI Bulgaria Bulgaria Bulgaria Bulgaria Croatia Croatia Czech Rep Czech Rep Estonia Estonia Estonia Estonia Hungary Hungary Hungary Latvia Latvia Latvia Lithuania Lithuania Lithuania Poland Poland Poland Romania Romania Romania Slovakia Slovakia Slovakia Slovenia Slovenia 1990 1994 1996 2008 1998 2008 1996 2008 1992 1998 2001 2008 1991 1997 2008 1992 1994 2008 1992 1996 2008 1992 2000 2008 1990 1996 2008 1999 2003 2008 1992 2008 IMF 1994 1996 2008 1998 2008 1996 2008 1992 2008 1991 2008 1992 1995 2008 1992 1995 2008 1992 2008 1990 1996 2008 1998 2008 1992 2008
Source: Leaven and Valencia (2008) and Andreou et al. (2009).
3. Empirical analysis 3.1. Baseline

3.1.1 Baseline results We start by estimating Equation (1) using both the MPI index and the IMF nancial crisis dummy. The results of these regressions are presented in Table 2. Looking at the table, we can see that both nancial crisis dummies have a signicant contemporaneous effect on growth. Moreover, while in the case of the MPI index the rst, the second and the third lags are statistically signicant, in the case of the IMF dummy all lags are statistically signicant. These results, together with the signicance of the coefcient of output growth persistence, imply that the occurrence of nancial crises have a long-lasting and statistically signicant effect on output. This is conrmed by the impulse response functions derived by the estimated coefcients presented in the table. In particular, when our MPI is applied (Figure 1A), a nancial crisis lowers output by 12 percent in the long run. The impact is even greater with the IMF crisis indicator, for which the cumulative output loss is close to 17 percent (Figure 1B). This is consistent with the nding that the impact of banking and twin (currency and banking) crises on output is bigger than that of currency crises (Cerra and Saxena, 2008; Goldstein et al., 2000; Kaminsky and Reinhart, 1999). Overall, the magnitude of the long-term effect is also consistent with that obtained by Cerra and Saxena (2008) for a larger sample of transition economies, Table 2. The effect of nancial crises on output-OLS
MPI Growth ()1) Growth ()2) Growth ()3) Growth ()4) Crisis Crisis ()1) Crisis ()2) Crisis ()3) Crisis ()4) F-test total (long term) effect N R2 0.398 (5.44)*** )0.118 ()1.55) 0.002 (0.04) )0.021 ()0.52) )2.0602 ()4.05)*** )3.909 ()5.73)** )1.698 ()2.58)** )1.558 ()2.59)** )0.146 ()0.26) 33.30*** 156 0.60 IMF 0.329 (4.68)*** )0.121 ()1.70)* 0.022 (0.40) )0.043 ()1.10) )1.729 ()3.26)*** )5.621 ()6.78)** )2.700 ()3.85)*** )2.069 ()3.19)*** )1.453 ()2.38)** 49.03*** 156 0.63
Note: t-statistics in parentheses. ***, **, * Signicance at 1, 5 and 10 percent, respectively.

Figure 1. The impact of nancial crises on the CEECs output

(A)
0
The MPI Index

0 1 2 3 4 5 6 7 8 9 10
(B)
0
The IMF crisis dummy

0 1 2 3 4 5 6 7 8 9 10
5 10 15 20 25 30
5 10 15 20 25 30
and is in the range of estimates (condence bands between 5 and 17) reported in the IMF, World Economic Outlook (2009). 3.1.2 Robustness check and endogeneity test To check for the robustness of our results, we replicate the analysis including other time shocks (time dummies) and oil shocks (oil price). The results even in this case are statistically signicant and suggest that the nancial crises have permanent effects on output. In detail, we nd that, even if the impact of other shocks on annual growth is statistically signicant, nancial crises continue to have a negative long-term impact on output. The cumulative loss is about 11 percent in the case of the MPI (Figure 2A) and 17 percent in the case of the IMF indicator (Figure 2B). Subsequently, we control for possible endogeneity of nancial crises dummies. In fact, it could be the case that economic slowdowns lead to crises onset, and in that case our OLS estimates will be biased and inconsistent. However, the GMM estimates obtained using the MPI (Figure 3A) and IMF index (Figure 3B) point to an almost identical impact than the one obtained with OLS. As an additional test, the Probit estimates of the probability of a crisis as a function of past growth performance and previous nancial crises corroborate the absence of reverse causality (Table 3).13 Finally, as an additional test, we replicate our analysis using quarterly data for GDP growth until the second quarter of 2009. The advantage of using quarterly data is that the estimates are based on a considerably higher number of observations, and therefore degrees of freedom. However, since quarterly data are available for many countries only starting in the late 1990s, several episodes of nancial crises are excluded from the analysis. In particular, the only nancial episodes left in the analysis are: Estonia (1998), Poland (2000), the Slovak Republic (2003) and the
13 The endogeneity test carried out also for the EU-15 conrms the validity of the assumption of exogeneity of nancial crises. Results are available from the authors upon request.
Figure 2. The impact of nancial crises controlling for other shocks

(A)
0 5 10 15 20 25 30
The MPI Index

0 1 2 3 4 5 6 7 8 9 10
(B)
0 5 10 15 20 25 30

0 1 2 3 4 5 6 7 8 9 10
Figure 3. The impact of nancial crises-GMM

(A)
0 0 5 10 15 20 25 30 1 2 3 4 5 6 7 8 9 10 5 10 15 20 25 30
The MPI index
(B)
0 0 1

2 3 4 5 6 7 8 9 10
Note: dotted lines = condence bands at 95%; line = estimated response.
crisis in 2008. Since we do not have information on the precise occurrence of the crisis within a given year, we assume a value of one for all quarters in the year in which a nancial crisis occurred. The IRFs are obtained estimating an ARDL (16, 16) of annualized quarterly growth rates of real GDP and a dummy for nancial crises. Looking at Figure 4 it is possible to see that even given the restriction in the number of episodes, nancial crises lower output both in the short and in the long term. The effect is similar for the MPI and IMF dummy. In particular, nancial crises identied by the MPI dummy are found to lower output in the long term by about 28 percent, whereas with the IMF dummy, the long-term effect is about 26 percent. However, it has to be stressed that given the different composition of nancial crisis episodes between the annual and the quarterly estimates, the results of the two approaches are not directly comparable. Given the strong robustness of results, the next sections will perform the country-heterogeneity analysis using the baseline specication. In addition, due to the high statistical signicance of the results, we will consider annual observations to capture the full set of crisis episodes.
10
Figure 4. The impact of nancial crises - quarterly data

0 5 10 15 20 25 30 MPI 35 IMF
Quarters
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
Table 3. The effect of past growth on the probability of nancial crisis (Probit)
MPI Crisis ()1) Crisis ()2) Crisis ()3) Crisis ()4) Growth ()1) Growth ()2) Growth ()3) Growth ()4) N
Note: z-statistics in parentheses.
IMF )6.632 ()0.00) )0.170 ()0.27) )6.510 ()0.00) )0.270 ()0.44) )0.076 ()1.20) 0.135 (1.72) )0.056 ()1.05) 0.022 (0.58) 156
)6.484 ()0.00) )0.318 ()0.58) )0.445 ()0.00) )0.039 ()0.09) )0.002 ()0.05) )0.006 ()0.09) 0.003 (0.06) 0.020 (0.58) 156
3.1.3 Are nancial crises more pernicious in the EU transition economies? We test whether CEEC countries are more affected by nancial crises than other EU countries. To this purpose we estimate Equation (1) for the EU-15 for the period 19802008. When we compare the results for CEEC and EU-15 countries, we can see that the effect is bigger and more persistent in the case of the CEECs economies. More precisely, with the MPI dummy (Figure 5A), the short-term effect of nancial crisis on the EU-15 output is negligible while in the CEECs output decreases by 2 percent. The long-term crisis impact in the EU-15 is greater than the one in the shortrun (almost 3 percent of cumulative loss), but still much less important than in the CEECs (12 percent). The estimates based on the IMF indicator (Figure 5B) conrm these results. In particular, the long-term effect for the EU-15 is about 2 percent, compared with 17 percent for the CEECs. These results are in line
11
Figure 5. The impact of nancial crises: CEECs vs. EU-15

(A)
0 2 4 6 8 10 12 14 CEECs EU-15 1 2 3
The MPI Index

4 5 6 7 8 9 10 11
(B)
0 2 4 6 8 10 12 14 16 18 20 1 2

3 4 5 6 7 8 9 10 11
CEECs
EU-15
with previous ndings. For example, according to European Commission, the longterm reduction in growth rate is expected to be considerably larger in the CEEC countries (about 7 percent) than in the EMU countries (about 4 percent).
3.2. Macroeconomic features and policy variables

So far, we have studied how nancial crises affect output in the CEECs economies. Now, we attempt to determine whether the countries structural heterogeneity and macroeconomic policy variables have a signicant effect in shaping the impact of nancial crises. More precisely, we explore the role of country size, openness, capital and FDI inows, nancial development, credit boom, banks foreign liabilities and a set of macroeconomic policy variables: exchange rate regimes, government spending, monetary policy, and IMF credits and loans.14 3.2.1 Size and openness First, we examine whether the countrys economic size (measured by the log of real GDP)15 and trade openness affect the impact and persistence of nancial crises on output growth. It is reasonable to expect that smaller economies are more exposed to external shocks (Aghion et al., 1999), since they are less diversied and face resource constraints. It is also reasonable to assume that in economies with a higher degree of trade openness, that are per se more exposed to external shocks (Rodrik, 1998), the impact of nancial crises may be more important. However, while our results indicate that the impact of nancial crises decreases with the size of the economy (Table 5), trade openness (measured as the GDP share of total exports and imports) is found to be statistically insignicant (Table 5). This puzzling nding for trade openness is also reported by other studies (Cerra et al., 2009), and may be partially explained by the reduced effect of scal stimuli in open economies.
14 15
Table 4 provides descriptive statistics concerning these variables. Similar results are obtained when we consider the log of total population.
12
Table 4. The sample means of structural and macroeconomic policy variables

Dependence on external nancing (% change) Total foreign ows over GDP (%) Ratio of FDI over GDP (%) Financial development (% of GDP) Number of credit boom episodes IMF credit and loans (millions USD) Change in CB interest rate (p.p)* )2.72 )1.50 )0.82 0.07 )1.25 )2.36 )4.68 )2.60 )2.98 )0.55 )3.96 34.91 38.53 46.91 48.78 38.21 38.91 26.69 27.63 13.88 42.25 41.00 4 3 2 3 2 2 2 0 1 2 1 )4.04 1.76 )69.98 0.25 278.07 48.20 )0.54 )7.05 )10.95 )34.44 )0.06 34.44 30.71 33.71 56.70 66.52 27.30 21.17 19.96 14.05 29.77 16.00 98.29 91.01 69.67 122.31 131.42 96.46 59.50 57.28 40.99 75.08 62.34 67.63 15.05 10.25 57.42 19.61 34.77 45.35 35.00 66.08 14.81 32.17 Change in government spending (%)* )1.67 1.50 0.03 2.25 0.59 2.74 1.51 1.24 0.15 )0.43 3.48 Furceri and Zdzienicka
Size (log GDP)
Openness (%)
Bulgaria Croatia Czech Rep Estonia Hungary Latvia Lithuania Poland Romania Slovakia Slovenia
0.78 4.90 7.71 4.62 9.50 1.64 3.92 6.52 4.83 6.87 2.93
112.30 87.42 125.01 154.74 110.98 98.62 111.56 58.55 63.86 141.48 118.43
Note: *Means decrease. Source: IMF, OECD, Central Banks Statics, Authors Computations.
13
The effect of country size in shaping the response of output to nancial crises can be seen in Figure 6A (MPI index) and Figure 6B (IMF dummy). The gures show the response of output in the case (i) small countries (rst quartile of the country size distribution), (ii) average countries, and (iii) larger countries (third quartile of the distribution). The charts show that with the MPI index (Figure 6A) the short-term and long-term effects of nancial crises on output are, respectively, about 1 and 8 percent for the larger countries, while for smaller countries, the losses are, respectively, about 3 and 12 percent. The estimates based on the IMF dummy (Figure 6B) show that the short-term and long-term effects of nancial crises on output are, respectively, about 2 and 12 percent for the larger countries, while for smaller countries the losses are respectively about 3 and 15 percent. 3.2.2 Foreign capital ows and FDI In this section, we analyse whether the effect of nancial crises on output depends on foreign capital ows (FDI plus short-term ows) and foreign direct investments
Table 5. The impact of nancial crises on output controlling for: size and openness
Size MPI Growth ()1) Growth ()2) Growth ()3) Growth ()4) Crisis Crisis ()1) Crisis ()2) Crisis ()3) Crisis ()4) Growth ()1) X Crises*X X F-test total (long term) effect N R2 0.433 (3.95)*** )0.133 ()1.85)* )0.016 ()0.28) )0.092 ()2.04)** )4.945 ()4.14)*** )3.500 ()5.32)*** )1.228 ()1.95)* )1.156 ()2.01)** 0.017 (0.32) )0.027 ()1.17) 0.051 (2.59)** 4.582 (3.66)*** 14.61*** 156 0.65 IMF 0.328 (3.11)*** )0.121 ()1.79)* 0.000 (0.00) )0.110 ()2.58)** )4.916 ()4.37)*** )5.226 ()6.49)*** )2.208 ()3.29)** )1.758 ()2.80)*** )1.048 ()1.79)* )0.021 ()0.96) 0.055 (2.90)*** 4.507 (3.70)*** 17.97*** 156 0.68 MPI 0.531 (2.72)*** )0.118 ()1.63) )0.024 ()0.43) )0.055 ()1.38) )0.578 ()0.34) )0.386 ()5.94)*** )0.169 ()2.72)*** )0.128 ()2.24)** )0.079 ()0.15) )0.002 ()1.06) )0.013 ()0.94) 0.062 (4.35)*** 2.95* 154 0.65 Openness IMF 0.450 (2.33)** )0.118 ()1.68)*** 0.001 (0.02) )0.064 ()1.62)* )2.940 ()1.67)* )5.239 ()6.33)*** )2.555 ()3.69)*** )1.761 ()2.72)*** )1.164 ()1.92)** )0.001 ()0.87) 0.011 (0.72) 0.043 (2.93)*** 4.45** 154 0.66

14
Figure 6. The impact of nancial crises controlling for economic size

(A)
2 1 3 8 13 18 Average Small (1st quartile) Large (3rd quartile) 2 3 4 5 6 7 8 9 10 11 3
The MPI index
(B)
2

1 2 3 4 5 6 7 8 9 10 11
13 18
Average Small (1st quartile) Large (3rd quartile)
(FDI) alone. From a theoretical point of view, since countries relying hardly on external nancing have private sector balance sheets more exposed and are more vulnerable to sudden reversals in foreign capital ows, we should expect that the more a country depends on external nancing the larger will be the effect of nancial crises on its growth performance. To test for this hypothesis, we estimate Equation (2) using these two variables. The results are reported in Table 6. Looking at the table, we can see that although the coefcients associated to the interaction terms for the crisis dummies are negative, they are not statistically signicant, suggesting that total capital ows and foreign direct investments do not signicantly inuence output growth during nancial crises. At the same time, we nd that the effect of these variables during normal times is positive and statistically signicant in stimulating growth. 3.2.3 Financial development, credit boom and banks foreign liabilities We also examine how nancial development affects the impact of the crises on the real economy. A priori, the inuence of nancial development on growth performance during times of crises is uncertain. On the one hand, nancial development has a positive impact on economic growth through saving and investment channels (Levine, 1997). On the other hand, nancial development can increase nancial instability leading to an unsustainable credit expansion, deterioration of borrower capacity and balance sheet disequilibria (Bordo et al., 2001). In this context, previous works found a short-term detrimental effect and a long-term positive impact of nancial development on growth performance (Loayza and Ranciere, 2006). When we look at nancial developments in the EU transition economies, we note that the real impact of nancial crises is greater for countries with higher levels of private credit over GDP (Table 7). The effect is statistically signicant when we consider the IMF dummy. This can be also
15
Table 6. The impact of nancial crises on output controlling for: foreign ows to GDP, FDI to GDP
Total foreign ows MPI Growth ()1) Growth ()2) Growth ()3) Growth ()4) Crisis Crisis ()1) Crisis ()2) Crisis ()3) Crisis ()4) Growth ()1) X Crises X X F-test total (long term) effect N R2 0.488 (3.15)*** )0.091 ()1.06) )0.026 ()0.37) )0.067 ()q.41) )1.193 ()0.94) )3.278 ()4.58)*** )1.102 ()1.34) )1.231 ()1.97)** )0.096 ()0.24) )0.002 ()0.84) )0.006 ()0.49) 0.025 (1.74)* 4.90** 139 0.58 IMF 0.253 (1.68)* )0.090 ()1.16) )0.001 ()0.01) )0.086 ()1.90)* )0.062 ()0.49) )5.145 ()5.59)*** )2.677 ()3.09)*** )2.097 ()3.05)*** )1.670 ()2.57)** 0.001 (0.38) )0.004 ()0.38) 0.007 (0.52) 9.21*** 139 0.62 MPI 0.434 (4.18)*** )0.091 ()1.09) )0.038 ()0.54) )0.077 ()1.68)* )1.140 ()2.01)** )3.089 ()4.42)*** )0.975 ()1.32) )1.169 ()1.88)** )0.013 ()0.22) )0.001 ()1.46) )0.003 ()0.70) 0.049 (3.40)*** 10.13*** 138 0.60 FDI/GDP IMF 0.347 (3.42)*** )0.089 ()1.14) )0.009 (0.14) )0.081 ()1.84)* )0.087 ()0.91) )4.883 ()5.23)*** )2.390 ()2.78)*** )1.918 ()2.77)*** )1.582 ()2.33)** )0.001 ()0.86) )0.003 ()0.20) 0.028 (1.86)* 13.61*** 138 0.63
seen in Figure 7A and B. In particular, looking at the chart relative to the IMF dummy we can notice that, while the long-run effect is about 15 percent for countries with a relative small level of credit over GDP (rst quartile), the effect is about 19 percent for countries with a higher level of private credit over GDP (third quartile). These ndings can be partially explained by the fact that nancial development in some CEECs transition economies can be considered as excessive and unsustainable compared with their economic development (Zdzienicka, 2010). Indeed, when we test for credit boom,16 we can observe that the countries that have been characterized by excessive banking credit developments are more heavily affected by nancial crises (Table 7). The results are statistically signicant for the IMF dummy (Figure 8A and B).
16
Episodes of credit boom are taken from Zdzienicka (2010). See Table 4 for descriptive statistics.
16
Table 7. The impact of nancial crises on output controlling for: nancial development, credit booms and banks foreign liabilities
Credit booms MPI 0.321 (4.08)*** )0.098 ()1.30) )0.011 ()0.19) )0.032 ()0.81) )1.389 ()2.48)** )3.969 ()5.83)*** )1.698 ()2.59)** )1.637 ()2.76)*** )0.041 ()0.07) 0.137 (0.86) )2.956 ()1.40) 0.031 (0.03) 24.38*** 0.247 (3.49)*** )0.113 ()1.70)* )0.004 ()0.07) )0.052 ()1.44) )0.667 ()1.55) )5.764 ()7.43)*** )2.384 ()3.56)*** )2.232 ()3.67)*** )1.337 ()2.36)** )0.060 ()0.39) )9.399 ()3.82)*** 1.295 (1.31) 43.74*** 0.365 (4.31)*** )0.090 ()1.17) )0.007 ()0.09) )0.021 ()0.48) )0.972 ()1.76)* )3.699 ()5.37)*** )0.940 ()1.29) )1.275 ()2.10)** )0.284 ()0.58) 0.225 (2.02)** )1.756 ()2.40)** )0.038 ()0.06) 25.02*** IMF MPI Foreign liabilities IMF 0.277 (3.41)*** )0.112 ()1.63)* 0.022 (0.33) )0.042 ()1.05) )0.590 ()1.07) )5.712 ()7.22)*** )2.143 ()2.78)*** )2.40 ()3.29)*** )1.331 ()2.20)** 0.225 (2.15)** )2.038 ()2.94)*** 0.046 (0.08) 39.16***
Financial development IMF 0.132 (1.32) )0.127 ()1.79)* 0.010 (0.18) )0.458 ()1.15) )0.091 ()0.08) )5.736 ()7.01)*** )2.688 ()3.89)*** )2.312 ()3.27)*** )1.414 ()2.28)** 0.006 (2.63)*** )0.032 ()1.62)* )0.018 ()1.07) 18.05***
MPI
Growth ()1) 0.288 (2.75)** Growth ()2) )0.128 ()1.65)* Growth ()3) )0.010 ()0.17) Growth ()4) )0.009 ()0.23) Crisis )0.978 ()0.89) Crisis ()1) )3.870 ()5.63)*** Crisis ()2) )1.653 ()2.49)** Crisis ()3) )1.665 ()2.69)*** Crisis ()4) )0.153 ()0.26) Growth ()1) X 0.003 (1.49) Crises X )0.199 ()0.95) X )0.018 ()0.97) 10.28** F-test total (long term) effect N 154 2 R 0.61 154 0.66 156 0.62 156 0.69 148 0.65
148 0.70
17
Figure 7. The impact of nancial crises controlling for the degree of nancial deepening
(A)
0 1 2 3 4 5 6 7 8 9 10 11 5 10 15 20 Average Small (1st quartile) Large (3rd quartile) 5 10 15 20 Average Small (1st quartile) Large (3rd quartile)
The MPI index
(B)
0

1 2 3 4 5 6 7 8 9 10 11
Figure 8. The impact of nancial crises controlling for credit booms

(A)
0 1 2 3 4 5 6 7 8 9 10 11 5 10 15 20 25 Credit boom No credit boom 5 10 15 20 25 Credit boom No credit boom
The MPI index
(B)
0

1 2 3 4 5 6 7 8 9 10 11
Figure 9. The impact of nancial crises controlling for banks foreign liabilities
(A)
0 1 5 10 15 20 Average Small (1st quartile) Large (3rd quartile) 2 3 4 5 6 7 8 9 10 11
The MPI index
(B)
0

1 2 3 4 5 6 7 8 9 10 11
5 10 15 20 Average Small (1st quartile) Large (3rd quartile)
Finally, we test whether the share of foreign liabilities of the banking sector has an impact in shaping the response of output to nancial crises. Since in countries with a larger share of foreign liabilities the banking sector is more vulnerable to
18
sudden reversal ows, we should expect that the more a countrys banking sector depends on external nancing the larger is the effect of nancial crises on its growth performance. This hypothesis is conrmed by our results (Table 7). In particular, looking at Figure 9A and B, we can notice that the long-run effect on output is 3 percent higher in countries characterized by a higher share of banks foreign liabilities (third quartile) compared with countries with a smaller share (rst quartile). 3.2.4 Macroeconomic policy variables So far, we have focused on the CEECs structural heterogeneity in explaining the impact of nancial crises on output, but also macroeconomic policy variables can inuence the effect of nancial crises. Here, we examine the effect of the following macroeconomic policy variables: the exchange rate regime, changes in government spending17 and in central bank policy rates, and foreign aid approximated by IMF credits and loans. First, we analyse the impact of exchange rate regimes. Traditionally, exible exchange rate regimes are reported to perform better during nancial crises (Cerra et al., 2009; Goldstein et al., 2000; Kaminsky and Reinhart, 1999). Indeed, it is reasonable to think that under a exible exchange rate regime, monetary authorities are less constrained and dispose of at least one additional policy instrument to deal with the crisis (Tong and Wei, 2009). Our results reported in Table 8 only partially conrm these ndings. In detail, the results show that both coefcients associated to the interaction term with the crisis dummy and the lagged growth are positive. This suggests that exible exchange rates attenuate the impact of the crises in the short and medium term, but tend to amplify the effect in the long run. This can be better seen in Figure 10A and B. When we look at scal policy, we notice a positive and signicant effect of government spending increases in attenuating the negative effect of the crises on growth (Table 8).18 The magnitude of the effect can be also seen in Figure 11A (MPI index) and Figure 11B (IMF dummy), where the response of output to nancial crises is shown for (i) a 3 percentage points reduction in government spending, (ii) no change in government spending, and (iii) a 3 percentage points increase in government spending. In detail, with the IMF dummy (Figure 11B), the long-term impact of nancial crises on output for a 3 percentage point increase in spending is almost 2 percent less important than for a situation of no change in government spending. When we use the MPI methodology (Figure 11A), the effect of a 3 percent government spending increase is lower and about 1 percent.
17 Similar results are obtained using the government decit. However, we consider only government spending, since government spending is less responsive to output changes than government revenue and public decits (Afonso et al., 2010). 18 Previous ndings corroborate a positive impact of government spending on growth recovery in developing and developed countries (Cerra et al., 2009; Claessens et al., 2004).
Table 8. The impact of nancial crises on output controlling for: exchange rate regime, changes in government spending, changes in monetary policy rates and IMF loan and credit
Change in government spending MPI IMF MPI IMF MPI Change in policy rate IMF credit and loan IMF
Exchange rate regime IMF
MPI
Growth ()1)
Growth ()2)
Growth ()3)
Growth ()4)
Crisis
Crisis ()1)
Crisis ()2)
Crisis ()3)
Crisis ()4) Growth ()1) X Crises X
F-test total (long term) effect N R2 156 0.64 156 0.62
0.355 (4.83)*** )0.137 ()1.65)* 0.019 (0.25) )0.014 ()0.36) )2.541 ()3.05)*** )3.832 ()3.75)*** )1.566 ()1.75)* )1.627 ()2.10)** )0.014 ()0.27) 0.213 (1.77)* 1.689 (1.76)* )1.717 ()2.85)*** 19.00***
0.294 (3.46)*** )0.131 ()1.53) 0.033 (0.48) )0.036 ()1.04) )2.013 ()2.40)** )5.465 ()4.76)*** )2.497 ()2.27)** )2.064 ()4.13)*** )1.383 ()1.64)* 0.212 (1.65)* 0.677 (0.65) )1.267 ()2.11)** 25.22***
0.331 (4.24)*** )0.100 ()1.33) )0.001 ()0.03) )0.023 ()0.58) )2.214 ()3.94)*** )3.750 ()5.53)*** )1.672 ()2.58)** )1.534 ()2.58)** )0.033 ()0.06) )0.004 ()0.36) 0.230 (1.14) 0.144 (2.07)** 26.69***
0.245 (3.25)*** )0.111 ()1.59) 0.016 (0.30) )0.048 ()1.26) )2.250 ()3.78)*** )5.418 ()6.49)*** )2.788 ()4.06)*** )2.200 ()3.41)*** )1.502 ()2.52)** 0.002 (0.20) 0.461 (2.10)** 0.125 (1.88)* 41.04***
0.342 (4.19)*** )0.139 ()1.72)* 0.006 (0.10) )0.069 ()1.57) )3.400 ()4.55)*** )5.007 ()7.23)*** )1.998 ()2.92)*** )2.229 ()3.59)*** )0.207 ()0.36) )0.002 ()0.34) 0.144 (2.96)** )0.012 ()0.42) 31.13***
0.313 (3.87)*** )0.139 ()1.79)* 0.023 (0.35) )0.066 ()1.46) )3.616 ()3.73)*** )5.997 ()6.72)*** )2.119 ()2.87)*** )2.376 ()3.32)*** )1.447 ()2.21)** 0.004 (0.79)* 0.099 (1.96)* 0.025 (0.82) 30.36***
0.423 (5.27)*** )0.126 ()1.47) )0.009 ()0.13) 0.002 ()0.06) )2.372 ()3.12)*** )3.892 ()3.81)*** )1.483 ()1.67)* )1.693 ()2.18)** )0.216 ()0.42) )0.001 ()2.16)** 0.000 (0.18) 0.000 (1.02) 29.81***
0.343 (3.74)*** )0.120 ()1.36) 0.013 (0.19) )0.030 ()0.81) )1.978 ()2.86)*** )5.679 ()5.10)*** )2.572 ()3.25)** )2.227 ()4.43)*** )1.542 ()1.84)* )0.001 ()1.75)*** 0.000 (0.34) 0.000 (0.96) 43.95***
156 0.62
156 0.66
126 0.73
126 0.71
148 0.61
148 0.64
19
20
Figure 10. The impact of nancial crises controlling for exchange rate regimes
(A)
0 1 2 3 4 5 6 7 8 9 10 11 5 10 15 20 25 Flexible Fixed 5 10 15 20 25 Flexible Fixed
The MPI index
(B)
0

1 2 3 4 5 6 7 8 9 10 11
Figure 11. The impact of nancial crises controlling for changes in government spending
(A)
0 1 2 5 10 15 20 No change Decrease 3 pp Increase 3 pp 3 4 5 6 7 8 9 10 11 5 10 15 20 No change Decrease 3 pp Increase 3 pp
The MPI index
(B)
0 1 2

3 4 5 6 7 8 9 10 11
In the case of monetary policy, the effect on growth performance during periods of crisis is less obvious, especially for developing economies. On the one hand, there is no unique recommendation concerning the appropriate actions to take. Some economists advocate an increase in interest rates to defend the domestic currency and attract foreign capital back into the country, whereas others encourage the opposite action to stimulate the economy (Krugman, 1999; Christiano et al., 2002). On the other hand, the impact of monetary stimulus on growth performance also remains uncertain. More precisely, while a positive impact is reported for industrialized economies, for developing countries monetary policy has been found to have only a weak effect on economic growth (Cerra et al., 2009; Romer and Romer, 1989). For the CEEC economies we report a statistically signicant and positive impact of changes in central bank interest rates on output in periods of crisis (Table 8). However, the magnitude of the effect is relatively small (Figure 12A and B). Finally, the last variable that we analyse is IMF credits and loans. There is a large body of the literature debating the impact of foreign aid on economic activity
21
Figure 12. The impact of nancial crises controlling for changes in central bank policy rates
(A)
0 1 2 3 4 5 6 7 8 9 10 11 5 10 15 20 No change Decrease 3 pp Increase 3 pp 5 10 15 20 No change Decrease 3 pp Increase 3 pp
The MPI index
(B)
0

1 2 3 4 5 6 7 8 9 10 11
Figure 13. The impact of nancial crises controlling for changes in the IMF support
(A)
0 1 2 3 4 5 6 7 8 9 10 11 5 10 15 20 Mean Last two years No credit 5 10 15 20 Mean Last two years No credit
The MPI index
(B)
0

1 2 3 4 5 6 7 8 9 10 11
in times of crisis. Some authors underline their positive impact on growth recovery (Eichengreen and Rose, 2003; Park and Lee, 2003), whereas others nd no signicant effect (Cerra et al., 2009; Honohan and Klingebiel, 2003). In the case of European transition economies, the impact of foreign liquidity support is not significant in the shortrun (Table 8). At the same time, IMF support is found to signicantly affect the persistence of the shock (the interaction coefcient with lagged growth, Table 8), affecting therefore the total impact in the long run (Figure 13A and B).
4. Conclusions
The aim of this work is to evaluate the short and long-term impact of nancial crises on output in the CEEC economies by estimating an ARDL equation of nancial crises and output growth, and deriving the relative impulse response functions for real output.
22
The main results of the paper can be summarized as follows: 1. Financial crises have a signicant impact on output both in the short and in the long run. In particular, nancial crises are found to lower output by 1 percent after one year and by 1217 percent after 5 years. Comparing the effect of nancial crises between the CEEC and the EU-15 economies, our results suggest that the effect is greater for the CEECs. Controlling for structural heterogeneity among CEECs, the impact of nancial crises is larger for smaller countries, in which the banking sector shows greater disequilibria. The impact of nancial crises on growth performance is mostly inuenced by scal policy (in terms of increases in government spending), whereas the effect of monetary policy is rather modest. Flexible exchange rates attenuate the impact of the crises in the short to medium term but tend to amplify the effect in the long run. Finally, foreign nancial aid (in terms of IMF credits and loans) is found to attenuate the effect of the crises in the long run.
2. 3.
4.
References
Afonso, A., Agnello, L. and Furceri, D. (2010). Fiscal policy responsiveness, persistence and discretion, Public Choice, (forthcoming). DOI: 10.1007/s11127-009-9577-x. Aghion, P., Bacchetta, P. and Banerjee, A. (1999). Capital market and instability of open economies, CEPR Discussion Paper 2083. Andreou, I., Dufrenot, G., Sand-Zantman, A. and Zdzienicka, A. (2009). A forewarning indicator system for nancial crises: The case of six central and Eastern European countries, Journal of Economic Integration, 24, pp. 87115. Barro, R. (2001). Economic growth in East Asia after and before the nancial crisis, NBER Working Paper 8330, Cambridge, MA: NBER. Bernanke, B. and Gertler, M. (1989). Agency costs, net worth and business uctuations, American Economic Review, 79(1), pp. 1431. Blot, C., Le Bayon, S., Lemoine, M. and Levasseur, S. (2009). Quantifying the impact of nancial crisis on the real sphere for France and the United States, http://www. euroframe.org/leadmin/user_upload/euroframe/docs/2009/EUROF09_Blot_etal.pdf Bordo, M., Eichengreen, B., Klingebiel, D. and Soledad Martinez-Peria, M. (2001). Is the crisis problem growing more severe? Economic Policy, 16(32), pp. 5182. Boyd, J., Kwak, S. and Smith, B. (2005). The real output loss associated with modern banking crises, Journal of Money, Credit and Banking, 37, pp. 977999. Calvo, G. (2006). Monetary policy challenges in emerging markets: Sudden stop, liability dollarization and lender of last resort, NBER Working Papers, 12788, Cambridge, MA: NBER. Calvo, G. and Reinhart, C. (2000). When capital inows come to a sudden stop: Consequences and policy options, MPRA Papers 6982, Germany: University Library of Munich.
23
Cerra, V. and Saxena, S. (2008). Growth dynamics: The myth of economic recovery, American Economic Review, 98, pp. 439457. Cerra, V., Panizza, U. and Saxena, S. (2009). International evidence on recovery from recessions, IMF Working Paper 09/183, Washington, DC: IMF. Chang, R. and Velasco, A. (1998). The Asian liquidity crisis, NBER Working Paper 6796, Cambridge, MA: NBER. Christiano, L., Gust, C. and Roldos, J. (2002). Monetary policy in a nancial crisis, NBER Working Papers, 9005, Cambridge, MA: NBER. Claessens, C. A., Klingebiel, D. and Laeven, L. (2004). Resolving systemic nancial crises: Policies and institutions, Policy Research Working Paper Series 3377, Washington, DC: The World Bank. DellAriccia, G., Detragiache, E. and Rajan, R. (2008). The real effect of banking crises, Journal of Financial Intermediation, 17(1), pp. 89112. Demirguc-Kunt, A., Detragiache, E. and Gupta, P. (2006). Inside the crisis: An empirical analysis of banking systems in distress, Journal of International Money and Finance, 25(5), pp. 702718. Edison, H. (2000). Do indicators of nancial crises work? An evaluation of an early warning system, Federal Reserves Board of Governors, International Finance Discussion Papers, Number 675. Eichengreen, B. and Rose, A. (2003). Does it pay to defend against a speculative attack? in Michael Dooley and Jeffrey Frankel (eds), Managing Currency Crises in Emerging Markets, Chicago: University of Chicago Press. European Commission (2009). the nancial crises and potential growth: Policy challenges for Europe, Economic Briefs, 3 June 2009. Furceri, D. and Mourougane, A. (2009). The effect of nancial crises on potential output: New empirical evidence from OECD countries, OECD, Economics Department Working Paper 699, Paris: OECD. Goldstein, M., Kaminsky, G. and Reinhart, C. (2000). Assessing Financial Vulnerability: An Early Warning System for Emerging Markets, Washington, DC: Institute for International Economics. Honohan, P. and Klingebiel, D. (2003). The scal cost implication of an accommodating approach to banking crises, Journal of Banking and Finance, 27(8), pp. 15391560. Hutchison, M. (2001). How costly are IMF stabilization programs, Economic Letter, Federal 30, Reserve Bank of San Francisco. Hutchison, M. and Noy, I. (2005). How bad are twins? Output costs of currency and banking crises, Journal of Money, Credit and Banking, 37(4), pp. 725752. International Monetary Fund, World Economic Outlook (2009), Washington, DC: IMF. Kaminsky, G. and Reinhart, C. (1999). The twin crises. The causes of banking and balanceof-payments problems, American Economic Review, 89, pp. 473500. Kaminsky, G., Lizondo, S. and Reinhart, C. (1998). Leading indicators of currency crises, IMF Staff Papers, 45, pp. 148. Kiyotaki, N. and Moore, J. (1997). Credit cycles, Journal of Political Economy, 105(2), pp. 211 248. Krugman, P. (1999). Balance sheets, the transfer problem, and nancial crises, in P. Isard, A. Razin and A. Rose (eds.), International Finance and Financial Crises: Essays in Honor of Robert P. Flood, Jr., Kluwer Academic Publishers: New York.
24
Laeven, L. and Valencia, F. (2008). Systemic banking crises: A new database, IMF Working Paper WP/08/224, Washington, DC: IMF. Levine, R. (1997). Financial development and economic growth: Causality and causes, Journal of Monetary Economics, 46, pp. 3177. Levy Yeyati, E. and Sturzenegger, F. (2005). Classifying exchange rate regimes: Deeds vs. words, European Economic Review, 49(6), pp. 16031635. Loayza, N. and Ranciere, R. (2006). Financial development, nancial fragility and growth, Journal of Money, Credit and Banking, 38(4), pp. 10511076. Milesi-Ferretti, G. M. and Razin, A. (1998). Current account reversals and currency crises Empirical regularities, CEPR Discussion Papers 1921, London: CEPR. Park, Y. C. and Lee, J. W. (2003). Recovery and sustainability in East Asia, in M. P. Dooley and J. A. Frankel (eds), Managing Currency Crises in Emerging Markets, Cambridge, MA: NBER. Reinhart, C. and Rogoff, K. (2004). The modern history of exchange rate arrangements: A re-interpretation, Quarterly Journal of Economics, 119(1), pp. 148. Reinhart, C. and Rogoff, K. (2009). The aftermath of nancial crises, NBER Working Papers, 14656, Cambridge, MA: NBER. Rodrik, D. (1998). Why do more open economies have bigger governments? Journal of Political Economy, 106(5), pp. 9971032. Romer, C. and Romer, D. (1989). Does monetary policy matter? A new test in the spirit of Friedman and Schwartz, NBER Macroeconomics Annual, 4, pp. 121170. Tong, H. and Wei, S. J. (2009). Real effects of the 200708 nancial crisis around the world, http://www.nance-innovation.org/risk09/work/1201504.pdf. Zdzienicka, A. (2008). Financial vulnerabilities in Central and Eastern Europe: Credit growth, GATE Working Paper 09.12, GATE-CNRS: Lyon. Zdzienicka, A. (2010). A re-assessment of credit development in European transition economies, MPRA Working Paper, Munich: MPRA.
Appendix Data and sources

Output Growth: Growth of real GDP. Source: IMF International Financial Statistics. MPI crisis index: Value equal to 1 corresponds to the real effective exchange rate depreciation and loss in foreign reserves above safe threshold. Source: Andreou et al. (2009). IMF crisis dummy: Value equal to 1 in correspondence of currency, banking and debt crises. Source: Leaven and Valencia (2008). Country Size: Calculated as the logarithm of real GDP. Source: IMF International Financial Statistics. Trade Openness: The ratio of exports plus imports to GDP at constant prices. Source: IMF International Financial Statistics.
25
Foreign Liabilities: Calculated as foreign liabilities of domestic banks over GDP. Source: IMF International Financial Statistics and World Economic Outlook. Financial Development: Ratio of banking credit to the private sector over GDP. Source: IMF International Financial Statistics. Credit Boom dummy: Value equal to 1 corresponds to excessive banking credit to the private sector. Source: Zdzienicka (2008). Exchange Rate Regime dummy: Value equal to 1 corresponds to exible and 0 stands for intermediate and xed exchange rate regime. Source: Reinhart and Rogoff (2004), LevyYeyati and Sturzenegger (2005) and after 2004, the IMF de facto classication. Central Bank Interest Rate: Corresponds to the central banks policy rate. Source: Eurostat and Central Banks statistics. Government Spending: Calculated as the ratio of government spending to GDP. Source: OECD database (85). Foreign Aid: Approximated by the IMF credits and loans. Source: IMF International Financial Statistics.

The Real Effect of Financial Crises in The European Transition Economies

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

The Real Effect of Financial Crises in The European Transition Economies

Hochgeladen von

Copyright:

Verfügbare Formate

Economics of Transition Volume 19(1) 2011, 125 DOI: 10.1111/j.1468-0351.2010.00395.

The real effect of financial crises in the European transition economies1

Received: October 6, 2009; Acceptance: February 26, 2010

Furceri and Zdzienicka

Financial Crises in the European Transition Economies

Furceri and Zdzienicka

2. Methodology and data

dj Di;tj uDyi;t1 Xi;t mXi;t eit ;

Financial Crises in the European Transition Economies

Furceri and Zdzienicka

Table 1. Financial crises episodes

Source: Leaven and Valencia (2008) and Andreou et al. (2009).

Financial Crises in the European Transition Economies

3. Empirical analysis 3.1. Baseline

Note: t-statistics in parentheses. ***, **, * Signicance at 1, 5 and 10 percent, respectively.

Furceri and Zdzienicka

Figure 1. The impact of nancial crises on the CEECs output

The MPI Index

The IMF crisis dummy

Financial Crises in the European Transition Economies

Figure 2. The impact of nancial crises controlling for other shocks

The MPI Index

The IMF crisis dummy

Figure 3. The impact of nancial crises-GMM

The MPI index

The IMF crisis dummy

Note: dotted lines = condence bands at 95%; line = estimated response.

Furceri and Zdzienicka

Figure 4. The impact of nancial crises - quarterly data

Financial Crises in the European Transition Economies

Figure 5. The impact of nancial crises: CEECs vs. EU-15

The MPI Index

The IMF crisis dummy

3.2. Macroeconomic features and policy variables

Table 4. The sample means of structural and macroeconomic policy variables

Size (log GDP)

Financial Crises in the European Transition Economies

Note: t-statistics in parentheses. ***, **, * Signicance at 1, 5 and 10 percent, respectively.

Furceri and Zdzienicka

Figure 6. The impact of nancial crises controlling for economic size

The MPI index

The IMF crisis dummy

Average Small (1st quartile) Large (3rd quartile)

Financial Crises in the European Transition Economies

Note: t-statistics in parentheses. ***, **, * Signicance at 1, 5 and 10 percent, respectively.

Furceri and Zdzienicka

Note: t-statistics in parentheses. ***, **, * Signicance at 1, 5 and 10 percent, respectively.

Financial Crises in the European Transition Economies

The MPI index

The IMF crisis dummy

Figure 8. The impact of nancial crises controlling for credit booms

The MPI index

The IMF crisis dummy

The MPI index

The IMF crisis dummy

5 10 15 20 Average Small (1st quartile) Large (3rd quartile)

Furceri and Zdzienicka

Exchange rate regime IMF

Financial Crises in the European Transition Economies

Crisis ()4) Growth ()1) X Crises X

F-test total (long term) effect N R2 156 0.64 156 0.62

Note: t-statistics in parentheses. ***, **, * Signicance at 1, 5 and 10 percent, respectively.

Furceri and Zdzienicka

Note: t-statistics in parentheses. , , Signicance at 1, 5 and 10 percent, respectively.

Note: t-statistics in parentheses. , , Signicance at 1, 5 and 10 percent, respectively.

Note: t-statistics in parentheses. , , Signicance at 1, 5 and 10 percent, respectively.

Note: t-statistics in parentheses. , , Signicance at 1, 5 and 10 percent, respectively.

Note: t-statistics in parentheses. , , Signicance at 1, 5 and 10 percent, respectively.