Baca Cee Currency

DO FOREIGN BANKS DRIVE FOREIGN CURRENCY LENDING IN CENTRAL AND EASTERN EUROPE? Andreas Paulhart, Wolfgang Rainer and Peter Haiss1 THIS VERSION AS OF FEBRUARY 15, 2008 – COMMENTS WELCOME PAPER FOR PRESENTATION AT THE 36TH ANNUAL EFA MEETING (EUROPEAN FINANCE ASSOCIATION), 19-22 AUGUST 2009, NORWEGIAN SCHOOL OF ECONOMICS AND BUSINESS ADMINISTRATION, BERGEN, NORWAY Abstract Foreign direct investment (FDI) by banks into Central and Eastern Europe (CEE) has increased considerably, resulting in foreign bank ownership of up to 90% in some countries in the region. Most CEE countries also recorded a substantial rise in foreign currency (FX) denominated credit, leading to concerns about the potential risks of asset deterioration in the domestic banking sector and of accusations about foreign banks “exporting” risks. While there is a lot of research on dollarization in emerging markets, empirical evidence about the role of foreign banks in promoting fx-lending in CEE is scarce. This paper aims at filling this gap, analyzing whether there exists a significant relationship between foreign banks’ asset share and the level of fx-lending. Using a linear regression model with data on 16 transition countries for the period 1999 to 2006 we find that foreign banks do not increase the risk of foreign currency lending. Our results show that foreign currency denominated lending seems to be a function of the level of fx-deposits, changes in the real exchange rate and the concentration of the banking sector.

KEY WORDS: FOREIGN BANKS, FOREIGN CURRENCY LENDING, CREDIT EUROIZATION, LOAN GROWTH, CURRENCY REGIME

JEL CLASSIFICATION CODES: F31, F36, G15, G21

1

Peter Haiss ([email protected]) is a lecturer at EuropeInstitute, Vienna University of Economics and Business Administration, Althanstrasse 39-45, A-1090 Wien. Wolfgang Rainer ([email protected]; corresponding author) and Andreas Paulhart are graduate student at the Vienna University of Economics and Business Administration. The opinions expressed are the authors’ personal views and not necessarily those of the institutions the authors are affiliated with. The authors are indebted to helpful comments by Gerhard Fink and the FinanceGrowth Nexus-Team at WU-Wien, http://www.wu-wien.ac.at/europainstitut/forschung/nexus, and for data support by Ralph deHaas, EBRD.

1 Electronic copy available at: http://ssrn.com/abstract=1343557

1 INTRODUCTION AND CONCEPTUAL FRAMEWORK Foreign direct investment (FDI) by banks into Central and Eastern Europe (CEE) and more recently in South-Eastern Europe (SEE) and transition economies further East has risen considerably (Breuss, Fink and Haiss, 2004; Fink, Haiss and von Varendorff, 2007). In 2007 total banking assets in general continued to grow at a high pace and majority foreign owned banks are the dominant players: In the Czech Republic, Slovakia, Croatia and Bosnia and Herzegovina, for example, their market shares exceed the 90% mark, while in Albania and Estonia nearly 100% of the banking assets are owned by foreign institutes (RZB Group, 2007).

CHART 1 Foreign currency loans as % of total loans to the private sector for selected transition countries. Source: Own calculations, data from RCB Group 2004, 2005, 2006, 2007.

At the same time credit levels in these countries surged, which has led some researchers to point to the threats of a possible credit boom (ECB, 2005b, 5). It is particularly interesting to note that credits denominated in foreign currency account for a substantial part of total credit volume in some of these countries: Over the period 1999 to 2006 foreign currency loans in Slovenia have grown at an average annual compounded rate of 21.4%, accounting for 55.4% of total domestic credit in 2006. With the notable exception of the Czech Republic, all New EU Member States were in the 30-70% range of foreign currency credit in total loans during the last years, and several national banks took regulatory steps to halt further growth (EBRD, 2005, 41; Herzberg and Watson, 2007, 33; RCB Group, 2007).

2 Electronic copy available at: http://ssrn.com/abstract=1343557

CHART 2 Asset share of foreign banks with foreign ownership above 50% for selected transition countries, Source: own calculations, data from RCB Group 2004, 2005, 2006, 2007.

Foreign currency risk taking, however, entails a number of additional direct and indirect risk factors to the lender, the borrower and the financial stability as a whole: exchange rate risk, interest rate risk, performance risk of repayment vehicle, and economic risks, among others (Waschiczek, 2002, 91f; Bokor and Pellényi, 2005). These practices – borrowing cheaply in one country to invest in a higher-yielding asset somewhere else - in essence resemble financial transactions known as “carry trade” by more financial institutions (Karmin and Perry, 2007). The magnitude of foreign currency related lending risks depend on the borrower (with or without “natural hedge”), the chosen currency (given or implied stability relative to the domestic currency), and the currency regime (Got and Ross, 2006). Considering these two parallel developments, several authors suggest that there might be a nexus between financial sector FDI (FSFDI) and the increasing popularity of fx-credits: The European Commission (2004) asserts that cross-ownership in the banking sector might contribute to the high levels of fxdeposits and fx-loans issued in the new member states. In line with this argument Farnoux/ Lanteri/ Schmidt (2004, 21) reveal that in Poland foreign banks tend to issue fx-credits more often than domestic institutions, illustrating that this discrepancy might stem from more aggressive marketing strategies applied by foreign institutions. Boissay, Calvo-Gonzalez and Kozluk (2006) believe that foreign bank entry is boosting credit growth in CEE because foreign banks use better risk management techniques than local institutions. In addition, competition is enhanced, which in turn leads to lower interest rates and thus accelerates credit growth. As regards fx-lending in particular, foreign banks may profit from superior funding possibilities thanks to their parent banks (Boissay/ Calvo-Gonzalez/ Kozluk, 2006, 3; ECB, 2006a, 60). Luca and Petrova (2007) and Basso, Calov-Gonzalez and Jurgilas (2007) blame the foreign banks´ desire for currency-matched portfolios beyond regulatory requirements for the rise in foreign currency-denominated lending in transition economies. The OeNB (2007, 43) and the IMF (Tieman,

3 Electronic copy available at: http://ssrn.com/abstract=1343557

2007; IMF, 2008) speculate about the role of Austrian banks in the diffusion process of fx-loans: Since Austrian institutes can draw on considerable experience concerning fx-lending from their home market, they may have built up specific competencies to market fx-loans also abroad, thus fostering the fx-credit boom. Apart from foreign bank entry, the literature suggests several other factors which might drive fx-credit growth, located on the demand side as well as on the supply side of the market: Luca and Petrova (2007) argue that the level of fx-deposits has an influence on the banks’ willingness to issue fx-denominated loans. The rationale behind this consideration is that financial institutions may prefer holding currencymatched portfolios in order to be hedged against the currency risk. According to Guidotti and Rodriguez (1992, quoted in Uzun 2005, 7), foreign currencies tend to be used as a unit of account in countries with high inflation rates. Accordingly Delgado et al. (2002, quoted in Uzun 2005, 17) claim that also the uncertainty regarding the future inflation rate encourages banks to issue loans denominated in a foreign currency. In line with that, Ize and Parrado (2002, quoted in Uzun 2005, 17) point out that even historically high inflation rates may increase fx-lending over a longer time period due to persistence and hysterisis effects. Within the “dollarization literature” (see e.g. Levy Yeyati, 2006, for a recent overview), other authors argue that the surge in fx-credit is mainly demand driven: Got and Ross (2006) detect three principal reasons: First, demand is fostered by extremely low interest rates in the Eurozone, the US and Switzerland. Moreover, high economic openness and a lack of confidence in the domestic currency contributed to the increase as well (Got/ Ross, 2006). The interest rate argument is partly supported by the ECB (2006b, 39), which agrees that an interest rate gap is a precondition for taking out a fx-loan, arguing at the same time that such a discrepancy is not a sufficient motivation for taking out an fx-loan. In a recent study, Basso/Calvo-Gonzalez/Jurgilas (2007) find a positive relationship between interest rate differentials and the change in loan dollarization, i.e. the change in fx-lending. The World Bank (2007, 12) attaches importance to the exchange rate. According to this publication a stable or even appreciating trend of local currency units (LCUs) is another factor contributing to the growth of fx-lending. In line with that Boissay/ Calvo-Gonzalez/ Kozluk (2006, 17) presume that the existence of fixed or close-pegtype exchange rate regimes lowers the risk perceived by the customers, thus encouraging fx-denominated lending. Backé/ Ritzberger-Grünwald/ Stix (2007, 115) argue that demand for fx-credits may be connected with the level of fx-deposits. In fact, they observe that the Euro is used in several CEE countries already today (“de-facto euroization”), which might promote fx-lending. Similarly, an ECB survey (2006b, 41) shows that a country’s export orientation may have a considerable impact as well, since income denominated in foreign currencies may lead clients to taking out fx-loans. For Austria, Epstein and Tzanninis (2005) show that herd behavior played a major role in the rapid growth of loans in foreign currency. Research so far mainly concerned itself with the general impact of foreign banks on economic

4

development in the host counties (see e.g. Eller, Haiss and Steiner, 2006, for recent empirical evidence). With regard to foreign banks or other factors determining the level of fx-denominated loans, little empirical research has been conducted so far, with Basso et al (2007) and Luca and Petrova (2007) as recent exceptions. This paper is meant to fill this gap by presenting the results of several regression analyses, which investigate the impact of the various determinants of foreign exchange lending suggested in the literature. Our contribution lies in combining the analysis of foreign banks and foreign bank origin with demand-side parameters (interest rate differential, exchange rate, exchange rate regime, foreign trade) and supply-side parameters (foreign currency deposits, inflation differential, and bank concentration) to shed more light on the cause of fx-lending in transition economies. As can be seen in chart 2, the research framework represents a two-step approach. In a first step, the question of whether foreign bank entry has a significant influence is examined. Subsequently, the impacts of further environmental factors, which are located on the demand side as well as on the supply side, are tested as well.

FURTHER VARIABLES: DEMAND SIDE:  INTEREST RATE GAP

FSFDI

 EXCHANGE RATE  EXCHANGE RATE REGIME

(FINANCIAL

 EXPORT ORIENTATION

LEVEL OF

SECTOR FOREIGN

FX-CREDITS DIRECT INVESTMENT)

SUPPLY SIDE:  LEVEL OF FX-DEPOSITS  INFLATION RATE  BANK CONCENTRATION

CHART 3 Framework for the investigation of the factors influencing the level of FX-lending. Source: authors

The remainder of the article is structured as follows. In chapter 2 we describe our research methodology. Chapter 3 gives details about the data sample used in the empirical part of our work. Chapter 4 presents the empirical findings from our regression analyses. Chapter 5 concludes.

2 METHODOLOGY We analyze the underlying hypotheses using an ordinary least square linear regression model with the level of fx-credit as the dependent variable. In general the calculations differ in the use of diverse explanatory variables. Furthermore we decided to modify the various models by changing the underlying

5

assumptions: For those models investigating the influence of the market share of foreign banks we ran the regression for two scenarios: In the original analysis we processed the data from all the countries under investigation whereas in the second regression model we introduced a 20% threshold regarding market share. Therefore, only those countries where foreign banks held a market share exceeding this 20% hurdle rate were included in the calculations. The rationale behind this adaptation is the following: Waschiczek (2002, 89) argues, pointing to the example of Austria, that herding behavior contributes to the growing popularity of fx-credits. In the case of fx-lending this theory suggests that a potential borrower imitates other borrowers, assuming that they have additional pieces of information. Similarly, Tzanninis (2005, 7) attributes high importance to herding and spreading the practice through word of mouth. As a certain threshold might be required in order to trigger these effects, the 20% hurdle was introduced in the second regression model. Another modification concerns the selection of the countries according to their currency regime. The introduction of three different categories (Float, Peg/hard peg/currency board and Exchange Rate Mechanism II) is meant to clarify possible differences between the various types of currency regimes. The countries have been classified according to the IMF classification framework (IMF, 2006). In line with Égert (2007) and Herzberg and Watson (2007) we include Croatia as a “quasi peg” in the Peg-group. Although Hungary is officially not an ERM II member, we include it in this group, as its currency regime de-facto was very similar to the Exchange Rate Mechanism II over several years (Table 1). Float

Peg

ERM II

Albania

Belarus

Estonia

Czech Republic

Bosnia & Herzegovina

Hungary

Kazakhstan

Bulgaria

Latvia

Poland

Croatia

Lithuania

Romania

Ukraine

Slovakia

Russian Federation

Slovenia

Serbia TABLE 1 Classification of the sample countries according to exchange rate regime

In addition, we conduct an Analysis of Variance (ANOVA). In contrast to the independent variable used in the linear regression analysis, the independent variable in the ANOVA is not metrical, but a qualitative attribute. The one-way ANOVA serves to identify significant differences between various groups (Martens, 2003, 158-159). Therefore, it enables us to analyze the influence of purely qualitative features on the level of fx-credit.

6

3 DATA Our sample includes data on 16 countries from Central and Eastern Europe (i.e. the New EU Member States from CEE), South-Eastern Europe (SEE), as well as the former CIS region2, covering the period from 1999 to 20063. The data have been obtained from the database of the Economist Intelligence Unit, the various editions of the RZB Group´s CEE Banking Sector Reports, the Erste Bank (2007) Sector Report on SEE Banks and the EBRD Banking database. For descriptive statistics for the variables used in the regressions see Table 13 in the appendix.

4 EMPIRICAL RESULTS This section presents the empirical results of our regression analyses. In different scenarios we test several explanatory variables to their impact on the level of fx-lending.

4.1 MARKET SHARE OF FOREIGN BANKS As outlined above, several recent papers discuss the impact of foreign banks on the level of fx-lending in transition economies. Basso et al (2007) argue that better access by foreign banks to foreign funds leads to a higher level of foreign-currency denominated loans. As subsidiaries of foreign banks, they have ample access to foreign resources from their parent banks. To keep the banks´ exposures matched, the foreign subsidiaries grant loans in foreign currencies. Luca and Petrova (2007), based on the analysis of 21 transition countries over the 1990-2003 period similarly argue that foreign banks shift the currency risk onto their borrowers by lending in foreign currency, in essence trading currency risk for currencyinduced default risk. Bokor and Pellényi (2005) attribute that rather to currency board regimes, arguing that in countries like Estonia and Bulgaria, foreign banks can access cheap foreign resources from their parent companies and lend on without foreign exchange risk. Thus evidence on whether rising foreign bank involvement drives higher foreign exchange lending is scant, and Luca and Petrova (2007) accordingly call for studying the behavior of foreign banks more research in this field. We try to fill this gap by directly relating the market share of foreign banks (i.e. the percentage of assets held by foreign banks with more than 50% foreign ownership to the level of foreign exchange denominated loans.

2 Albania, Belarus, Bosnia & Herzegovina, Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Kazakhstan, Latvia, Lithuania, Poland, Romania, Russia, Serbia, Slovakia, Slovenia, Ukraine 3 2007 for Albania, Bulgaria and Romania,

7

Dependent Variable Independent Variable

Countries All countries FLOAT PEG ERM II PEG & ERM II All countries FLOAT PEG ERM II PEG & REM II

Threshold on Asset Share no no no no no yes (20%) yes (20%) yes (20%) yes (20%) yes (20%)

FX-Loans (% of total loans) Asset share of foreign banks (% of total assets)

Adjusted R² 0.012 0.031 0.348 0.050 -0.008 0.031 0.041 0.505 -0.150 0.011

Beta -0.147 -0.321 -0.610 0.276 -0.085 -0.204 -0.263 -0.728 0.126 -0.171

Significance 0.127 0.131 0.000 0.098 0.498 0.053 0.127 0.000 0.478 0.212

TABLE 2 Market share of foreign banks as explanatory variable

Our empirical results show that contrary to earlier findings by Basso et al (2007) and Luca and Petrova (2007) in broader samples, foreign banks do not increase the risk of foreign currency lending in Central and Eastern Europe. In fact, the asset share of foreign banks is not significantly correlated with the level of fx-lending (the Beta of -0.147 even points to a negative relationship, although not statistically significant on a 95% confidence interval). Introducing a 20% threshold on the asset share of foreign banks improves the significance of the general regression model. The group of countries with a fixed peg currency regime shows the most significant and highest negative correlation. Only the ERM II countries exhibit a positive correlation, although not statistically significant. While Lucan and Petrova (2007) analyzed 21 transition countries (i.e. a CIS majority) over the 1990-2003 period (i.e including very early years of transition), we analyze 16 transition countries (mainly the New EU Member States and Balkan Accession Countries) over the more recent 1999-2006 period. Countries in our sample are thus rather more developed. The 1990-1999 period was characterized by hyperinflation, economic volatility and bank crises in most transition economies which may cause data distortions. Besides Hungary, most countries only allowed for de novo (greenfield-startup) banks during the early years of transition, and bank licenses were handed out quite openhandedly in several countries. Only the privatizations towards the end of the 1990s brought about large-scale foreign entry in the transition economies. Thus we think our sample coverage more appropriate for the research issue. Basso et al (2007) similarly analyze 24 transition economies over the 2000-2006 period, using monthly data. They find a strong positive correlation between the level of foreign liabilities in the banking sector and both the number of foreign banks and their share in aggregate bank assets. The broad country sample notion again applies. Moreover, neither of the two previous studies included threshold effects. Drawing on the finding by Epstein and Tzanninis (2005) that herding was a major trigger for spreading the fx-loan practice across Austria, we argue that for herding and cascading to work threshold levels of foreign impact need to be included. In summary, we can not confirm the accusation that foreign banks increase the risk of foreign currency lending. Much more, it seems as if the asset share of foreign banks is negatively correlated with the level

8

of fx-lending. The results point to a stabilizing influence of foreign banks in CEE, possibly an effect of foreign banks’ superior risk management techniques. This is in line with the findings of Eller et al (2006) of a rather positive impact of financial sector FDI on economic development in CEE.

4.2 MARKET SHARE OF AUSTRIAN BANKS The second variable we tested was the asset share of Austrian banks in CEE. The OenB (2007) and the IMF attribute Austrian banks a special role in the context of the rapid credit expansion in CEE, arguing that Austrian banks account for a disproportionately high share of foreign currency lending in some of the countries (Oenb, 2007, 43; Tieman, 2007).

Dependent Variable Independent Variable

Countries All countries FLOAT PEG ERM II PEG & ERM II All countries FLOAT PEG ERM II PEG & REM II

Threshold on Asset Share no no no no no yes (20%) yes (20%) yes (20%) yes (20%) yes (20%)

FX-Loans (% of total loans) Asset Share of Austrian Banks (% of total assets)

Adjusted R² 0.029 0.181 0.846 0.281 0.713 -0.390 0.635 0.289 0.821 0.774

Beta -0.217 0.465 -0.924 -0.589 -0.850 -0.064 0.818 -0.625 -0.930 -0.890

Significance 0.109 0.022 0.000 0.044 0.000 0.762 0.001 0.098 0.022 0.000

TABLE 3 Market share of Austrian banks as explanatory variable

We do not find support of the theory that Austrian banks are responsible for the high share of fx-lending in CEE. The general model reveals no statistically significant relationship. Breaking the analysis further down, however, reveals some interesting results. Considering only countries with a floating exchange rate regime, the relationship between the asset share of Austrian banks and the level of fx-lending becomes statistically significant and positive (Beta=0.465, R²=0.181). In the group of countries with a fixed peg regime the asset share of Austrian banks seem to be highly negatively correlated with fx-lending, with a Beta of -0.924 and R² of 0.846. In countries which joined the ERM II Austrian banks also seem to have a negative impact on fx-lending (Beta= -0.589, R²=0.281).

4.3 FX-DEPOSITS If the level of fx-loans is not matched with a similar amount of fx-deposits, the resulting currency mismatches constitute considerable risks to banks and the financial system. These mismatches are generally seen as a source of financial fragility (Arteta, 2002, 1). Following this argument, it would make sense for banks to align the amounts of fx-loans granted with the level of fx-deposits they hold. We would

9

therefore expect a positive relationship between the level of fx-deposits and the level of fx-lending. With regards to the impact of the exchange rate regime on currency mismatches, most economists agree that a fixed exchange rate regime encourages currency mismatches as banks and firms do not hedge their foreign currency liabilities (Arteta, 2002, 1). Consequently, we expect a stronger relationship between fxdeposits and fx-lending in those countries with a floating exchange rate regime, as banks and firms seek to limit their exposure to currency fluctuations.

Dependent Variable Independent Variable

FX-Loans (% of GDP) FX-Deposits (% of GDP)

Countries All countries FLOAT PEG ERM II PEG & ERM II

Beta 0.314 0.691 0.305 0.359 0.151

Adjusted R² 0.090 0.466 0.043 0.105 0.005

Significance 0.001 0.000 0.191 0.027 0.259

TABLE 4 FX-Deposits as explanatory variable

At a 95% confidence interval the regression analysis covering the data from all the countries under examination confirms that there is a positive relationship between the level of fx-deposits and fx-loans. However, beta equals only 0.314 and the adjusted R² amounts to only 0.090. The classification of the various countries according to their currency regimes allows gaining deeper insight: As expected, in those countries with a floating exchange rate the level of fx-deposits appears to be highly relevant, as beta reaches 0.691 and the adjusted R² equals 0.466. The analysis of ERM II countries reveals a significant relationship between the level of fx-deposits and the level of fx-credit. Yet, a beta of 0.359 and an adjusted R² of 0.105 indicate that this nexus is considerably weaker than in countries with a floating exchange rate regime. In those countries with a pegged exchange rate there does not seem to be a significant relationship at all, which would confirm the theory that fixed exchange rate regimes encourage currency mismatches.

10

CHART 4 Scatter Plot for the regression of fx-deposits on fx-loans for floating regime countries. Source: Own calculations

4.4 EXPORT ORIENTATION For this analysis we used the total exports as a percentage of the GDP as a measure for a country’s export orientation. Contrarily to the underlying theory, the general regression model unveils a significant negative correlation, as beta is -0.199. Yet, the adjusted R² value of 0.032 is rather low. As for the analyses of the subsets, results are mixed. While the influence of export orientation seems to be insignificant in ERM II countries, there appears to be a negative correlation in those countries with a floating exchange rate (beta = - 0.564) and those with a pegged exchange rate (beta = -0.461).

Dependent Variable Independent Variable

FX-Loans (% of GDP) Total Exports as % of GDP

Countries All countries FLOAT PEG ERM II PEG & ERM II

Beta -0.199 -0.564 -0.461 -0.188 -0.047

Adjusted R² 0.032 0.283 0.188 0.012 -0.011

11

Significance 0.025 0.000 0.006 0.222 0.685

TABLE 5 Total Exports as % of GDP as explanatory variable

Our results should be treated carefully, as we do not have data distinguishing household from corporate fx-lending. This distinction, however, would be very important, as the export orientation intuitively should be linked with corporate fx-lending, but not household fx-lending. Basso/Calvo-Gonzalez/Jurgilas (2007) find that a country's openness to the international economy is contributing to corporate but not to household financial dollarization.

4.5 INTEREST RATE DIFFERENTIAL Theoretically, uncovered interest parity suggests that any interest differential between two countries can be explained by an expected depreciation or appreciation of the currency (Basso/Calvo-Gonzalez/Jurgilas, 2007, 53). Therefore, if uncovered interest parity holds, it should not be possible to take advantage of lower interest rates on foreign currency loans. However, the decisions of households or firms to finance in foreign currency might not be based on interest parity considerations. As outlined above, the attraction of lower nominal interest rates, coupled with herd-behavior, might have a considerable influence on fxlending decisions. Dependent Variable Independent Variable

FX-Loans (% of GDP) Interest Differential Germany

Countries All countries FLOAT PEG ERM II PEG & ERM II

Beta 0.064 0.183 0.422 -0.561 -0.031

Adjusted R² -0.004 0.013 0.153 0.298 -0.012

Significance 0.475 0.204 0.013 0.000 0.784

TABLE 6 Interest rate differential to German interest rate (US interest rate for Belarus, Kazakhstan, Russian Federation and Ukraine) as explanatory variable

Dependent Variable Independent Variable

FX-Loans (% of GDP) Interest Differential Eurozone

Countries All countries FLOAT PEG ERM II PEG & ERM II

Beta 0.054 0.169 0.396 -0.570 -0.051

Adjusted R² -0.005 0.008 0.131 0.309 -0.011

Significance 0.548 0.240 0.020 0.000 0.658

TABLE 7 Interest rate differential to eurozone interest rate (US interest rate for Belarus, Kazakhstan, Russian Federation and Ukraine) as explanatory variable

In order to investigate the influence of the interest rate differential we used two different data sets: In the first calculation the differential between the local interest rate and the German interest rate was used

12

while in the second regression we observed the gap between the local rate and a Eurozone Proxy lending rate.4 Still, the outcomes of both approaches are similar: The calculations covering all the countries do not point at a statistically significant nexus applying the 95% confidence interval. By contrast, the examinations of the “Peg”-countries and the ERM II countries exclusively revealed significant relationships. It is interesting to note that the existence of an interest rate differential in the “Peg”-subset appears to be positively correlated with the level of fx-credit (beta = 0.422 and 0.396 respectively) whereas the calculation for the ERM II countries results in a negative correlation (beta = -0.561 and -0.570 respectively).

4.6 INFLATION DIFFERENTIAL As high levels of inflation typically imply high expected interest rates, it might seem rational in such a scenario to switch to foreign currency loans, to benefit from lower interest rates. Accordingly, we would expect to find a positive relationship between the inflation differential and the level of fx-loans.5

Dependent Variable Independent Variable

FX-Loans (% of GDP) Inflation Differential Eurozone

Countries All countries FLOAT PEG ERM II PEG & ERM II

Beta 0.137 0.285 0.494 -0.632 0.001

Adjusted R² 0.011 0.047 0.220 0.385 -0.013

Significance 0.122 0.071 0.003 0.000 0.996

TABLE 8 Inflation differential to Eurozone inflation (US inflation for Belarus, Kazakhstan, Russian Federation and Ukraine) as explanatory variable

The regression models investigating the inflation rate differential as explanatory variable lead to astonishing results: While the general model, comprising data from all the countries, does not provide evidence that the inflation rate differential has an impact on the level of fx-lending, the analyses of the subsets “Peg” and ERM II show a significant correlation at a 95% confidence interval. The correlation revealed in the countries with a currency peg is positive (beta = 0.494, adjusted R² = 0.220) and thus

4

In both versions, however, for Russia, Kazakhstan, Ukraine and Belarus the interest rate gap between the local currency and the US lending rate was used, as these countries rather belong to the extended Dollar zone.

5

However, it has to be mentioned that if Purchasing Power Parity (PPP) holds, any change in the expected inflation differential between two countries will be compensated by changes in the exchange rate. Therefore it should not be possible to take advantage of lower interest rates in other countries.

13

consistent with the underlying theory. By contrast, the relationship detected in the ERM II countries is significantly negative with a beta of -0.632 and an adjusted R² of 0.385.

4.7 REAL EXCHANGE RATE The impact of changes in the real exchange rate on foreign currency borrowing is ambiguous. On the one hand, an appreciation of a currency reduces the interest and redemption payments agents have to make on loans denominated in foreign currencies. Hence, we would expect firms and households to increase their fx-lendings in times of local currency appreciations. On the other hand, however, Catao and Terrones (2000) argue that banks which seek to maximize their profits in dollars (or any other foreign currency for that matter) will decrease the foreign currency part of their loan portfolio in the wake of an expected appreciation of the local currency. Following Luca and Petrova (2003), who show that bank specific factors, as opposed to firm specific factors, are the main driving force of credit dollarization, we would expect the relationship between the real exchange rate and the level of fx-loans to be negative.

Dependent Variable Independent Variable

FX-Loans (% of GDP) Real Effective Exchange Rate Change (%)

Countries All countries FLOAT PEG ERM II PEG & ERM II

Beta -0.196 -0.379 0.026 -0.231 -0.064

Adjusted R² 0.031 0.123 -0.031 0.031 -0.009

Significance 0.030 0.011 0.882 0.131 0.577

TABLE 9 Real effective exchange rate change as explanatory variable

The results of the general model confirm our expectations of a negative relationship between the real exchange rate and fx-lending. However, the adjusted R² amounts only 0.031. Considering countries with floating exchange rate regimes exclusively, this negative relationship is somewhat stronger (beta = -0.379 and adjusted R² 0.123). The investigations of the subsets “Peg” and ERM II have not pointed at a significant influence of the real exchange rate change, though.

4.8 MARKET CONCENTRATION Additionally to the variables researchers have suggested so far, we introduce a new explanatory variable. We test whether the level of fx-lending can be explained by the level of bank concentration. The rational is that the access of banks to foreign exchange, and thus the possibility to grant fx-loans, might be a function of the size of banks.

14

Dependent Variable Independent Variable

FX-Loans (% of GDP) Asset share of 5 largest banks (%)

Countries All countries FLOAT PEG ERM II PEG & ERM II

Beta 0.422 0.522 -0.418 0.812 0.384

Adjusted R² 0.171 0.256 0.147 0.651 0.136

Significance 0.000 0.000 0.017 0.000 0.001

TABLE 10 Market share of the 5 largest banks as explanatory variable

In this regression model the asset share of the five largest banks in every country was used as a proxy for market concentration. In general, we find support of the theory that fx-lending is a function of the size of banks. The general model reveals a strong positive relationship with a beta of 0.422 and an adjusted R² of 0.171. The analysis of ERM II countries exclusively shows the strongest nexus: beta is 0.812 and also the adjusted R² of 0.651 is comparatively high. By contrast, the impact of market concentration in those countries with a currency peg appears to be negative – the adjusted R² value of 0.147 is rather low though.

CHART 5 Scatter plot for Asset share of the 5 largest banks as explanatory variable. Source: Own calculations

15

4.9 EXCHANGE RATE REGIME Boissay, Calvo-Gonzalez and Kozluk (2006) suggest that the existence of exchange rate regimes might be a driver for the rising level of fx-credit. They assume that due to exchange rate regimes the exchange rate risk is perceived as very low, which might in turn encourage foreign currency lending (Boissay / CalvoGonzalez / Kozluk, 2006, 16-17). In order to investigate whether there is a link between the type of exchange rate regime and the level of fx-credit we conduct an ANOVA. The observed countries are assigned either the variable 0, which indicates that their currency is floating freely, the variable 1, meaning that the country’s currency is pegged against another currency, or 2, which stands for all the countries participating in ERM II. The calculation shows that on average the level of fx-denominated loans is the lowest in the “Peg” subset (34.96%), while the ERM II countries exhibit the highest average share (44.86%). In those countries with a floating exchange rate the mean equals 39.93%. According to the ANOVA, however, the type of exchange rate regime is not the reason for these differences – even if considering a 90% confidence interval, the influence is statistically insignificant.

0 1 2 Total

N 50 34 44 128

Std. Deviation 20.61283 17.52291 22.22565 20.63653

Mean 39.9340 34.9559 44.8575 40.3041

Std. Error 2.91509 3.00515 3.35064 1.82403

95% Confidence Interval for Mean Lower Upper Bound Bound 34.0759 45.7921 28.8419 41.0699 38.1003 51.6147 36.6947 43.9136

Minimum 12.00 10.10 13.70 10.10

Maximum 82.10 64.90 13.70 82.10

TABLE 11 Descriptives of ANOVA for Fx-loans, Factor: Exchange rate regime, 0 = Float, 1 = Peg, 2 = ERM II

Between Groups

Sum of Squares 1891.637

Within Groups Total

df 2

Mean Square 945.819

52193.403

125

417.547

54085.041

127

F 2.265

TABLE 12 ANOVA for Fx-loans, Factor: Exchange rate regime

16

Sig. .108

CHART 6 Error bar showing the means of fx-loans for the three groups of exchange rate regime countries (95% confidence interval). Source: Own calculations

5 CONCLUSIONS Financial markets in transition economies differ from mature markets by their low level of intermediation, high growth rates, a high level of foreign ownership and a high level of foreign exchange borrowing. Previous research blames foreign banks for imposing extra risk on the host economy by foreign exchange (fx) lending. The goal of this article is to empirically investigate causes for foreign currency lending in transition economies. We analyze annual data over the 1999 to 2006 period for 16 transition countries, including mainly the New EU Member States, South-Eastern European (Balkan) and a few countries further East. We contribute by investigating the impact of three groups of parameters on fx-lending: bank ownership and origin; demand side; and supply side factors. According to the regression model developed in this article, the market share of foreign banks does not increase the level of foreign currency denominated credit. We thus cannot confirm the findings by Basso et al (2007) and Luca and Petrova (2007) who analyzed broader samples of transition countries, in the case of Luca and Perova also including the turbulent 1990-1999 period. We attribute the difference in findings (1) to sample differences (early transition years; low vs. intermediate level of development) and (2) to the fact that no threshold levels for the possible impact of foreign banks were considered by these previous studies which according to herding theory evoked by Epstein and Tzanninis (2005) for

17

explaining the rise of fx-loans in Austria is important for herding to work. Our analyses of the market share of Austrian banks, the interest rate differential and the inflation differential led to ambiguous results. With regards to export orientation, there seems to be even a negative correlation with the level of fx-credit. According to our analysis, the investigation of fx-deposits and their impact on foreign currency lending deserve particular attention: In line with the theoretical assumption there seems to be positive relationship in countries with floating exchange rate regimes and ERM II countries. Hence, banks could indeed try to avoid currency mismatches and shift the exchange rate risk to the borrowers. In countries with a currency peg this relationship could not be confirmed, though. This deviation could indicate that exchange rate regimes promote currency mismatches. To a certain degree the regression models also support the theoretical nexus between the real exchange rate development and the level of fx-credit. Whereas the link is rather weak in the aggregate model, it is considerably negative in those countries with a floating exchange rate. This correlation could point at the banks’ ability to successfully market higher shares of fx-credits in times of local currency depreciation, thus increasing their income. With regard to the impact of market concentration, in general, the regression model meets the initial expectations, as a concentrated market appears to foster the issuance of fx-loans. This finding is important for regulation and supervision of currency risk in the transition economies. Interestingly, in the countries with a currency peg a significantly negative correlation can be observed. While some of these results confirm the underlying theories, other assumptions may have to be reconsidered. Especially the occurrence of inhomogeneous findings for the different types of exchange rate regimes raise new questions which may have to be analyzed in greater detail as these currency regimes evolve over time. Data limitations kept us from separating fx-loans to households from fx-loans to corporations, i.e. borrowers with different abilities to hedge the fx-risk. Further research might also use different measures for foreign bank involvement, e.g. foreign share in total or sectoral lending, and crossborder-lending. The mode of foreign entry may also make a difference, i.e. whether foreign investors take over an existing bank in the host country (M&A), or whether they enter via de novo greenfield-startup. For household loans, the impact of remittances which may provide a currency risk hedge if high might also be analyzed. Therefore further research on this recent topic is highly desirable.

18

6 LITERATURE Arteta, Carlos Óscar (2002), Exchange Rate Regimes and Financial Dollarization: Does Flexibility Reduce Bank Currency Mismatches? CIDER Working Paper No. C02-123. Backé, Peter, Ritzberger-Grünwald, Doris, and Stix, Helmut (2007), The Euro on the Road East, Monetary Policy and the Economy Q1/07, 114-127, OeNB. Basso, Henrique, Calvo-Gonzalez, Oscar and Jugilas, Marius (2007), Financial Dollarization: The Role of

Banks

and

Interest

Rates,

ECB

Working

Paper

No.

748,

May

2007,

http://www.ecb.eu/pub/pdf/scpwps/ecbwp748.pdf (6.4.2008). Boissay, Frederic, Calvo-Gonzalez, Oscar and Kozluk, Tomasz (2006), Is Lending in Central and Eastern Europe developing too fast? Preliminary draft, paper presented on the Conference on European Economic

Integration,

14-15th

November

2005,

Vienna,

http://www.eu-financial-

system.org/fileadmin/content/Dokumente_Events/second_conference/Boissay_CalvoGonzales_Kozluk.pdf (9.4.2008). Bokor, László and Pellényi, Gábor (2005), Foreign Currency Denominated Borrowing in Central Europe: Trends, Factors and Consequences, ICEG EC Opinion Nr. 5, February 205, International Center for Econmic Growth – European Center Breuss, Fritz, Fink, Gerhard, and Haiss, Peter (2004), How well prepared are the New Member States for the European Monetary Union? Journal of Policy Modeling 26, 769-791 Catao, L, and Terrones, M. (2000), Determinants of Dollarization: The Banking Side, IMF Working Papers WP/00/146, http://www.imf.org/external/pubs/ft/wp/2000/wp00146.pdf (8.4.2008). Delgado, F. L., Kanda, D. S., Casselle, G. M. and R.A. Morales (2002), Domestic Lending In Foreign Currency in Building Strong Banks through Surveillance and Resolution, Enoch, C., Marston, D. and M. Taylor (eds.), International Monetary Fund, Washington D.C., 40-61. EBRD (2005), Transition Report 2005, European Bank for Reconstruction and Development, London. ECB (2005b), Lending Booms in the New EU Member States: Will Euro Adoption matter? ECB Working Paper No. 543, http://www.ecb.int/pub/pdf/scpwps/ecbwp543.pdf (9.4.2008). ECB (2006a), Macroeconomic and Financial Stability Challenges for Acceding and Candidate Countries, ECB Occasional Paper series, No. 48, http://www.ecb.int/pub/pdf/scpops/ecbocp48.pdf (9.4.2008).

19

ECB

(2006b),

EU

Banking

Sector

Stability,

November

2006,

www.ecb.int/pub/pdf/other/eubankingsectorstability2006en.pdf (9.4.2008). EIU Economist Intelligence Unit (2008) EIU Country Data, https://eiu.bvdep.com/version2008226/cgi/template.dll?product=101 (1.3.2008). Eller, Markus, Haiss, Peter, and Steiner, Katharina (2006), Foreign direct investment in the financial sector and economic growth in Central and Eastern Europe: The crucial role of the efficiency channel, Emerging Markets Review 7, 300-319. Égert, Balázs (2007), Central Bank Interventions, Communication and Interest Rate Policy in Emerging European Economies, OeNB Working Paper Nr. 134, http://www.oenb.at/de/img/wp134_tcm1451029.pdf (6.4.2008). Erste Bank (2008), SEE Banks Boom or bust? CEE Equity Research Sector Report, February 19, 2008. European Commission (2004), Financial Market Integration and the Use of the Euro in the New Member States, Quarterly Note on the Euro-Denominated Bond Markets, No. 75, September 2004, http://ec.europa.eu/economy_finance/publications/bond_markets/2004/bondq0304_en.pdf (7.12.2007). Farnoux, Marc, Lanteri, Marc, and Schmidt, Jerome (2004), Foreign direct investment in the polish financial sector, Case study prepared for the CGFS Working Group on Financial Sector FDI, http://www.bis.org/publ/cgfs22bdf.pdf (9.4.2008). Fink, Gerhard, Haiss, Peter, and von Varendorff, Mina (2007),

Serbia´s Banking Sector Reform:

Implications for Economic Growth and Financial Development, Southeastern European and Black Sea Studies 7(4), 609-636. Got, Xavier and Ross, Alise (2006), The Foreign Currency Gamble – Rising Risks For Banks In Central And Southeast Europe, Standard & Poor’s Ratingsdirect, Aug. 23, 2006. Guidotti, P. and C. Rodriguez (1992), Dollarization in Latin America: Gresham’s Law in Reverse?, International Monetary Fund Staff Papers, 39(3), 518-544. Herzberg, Valerie and Watson, Max (2007), Growth, risks and governance: The role of the financial sector in southeastern Europe, European Economy Occasional Paper No. 29, April 2007, European Commission Directorate-General for Economic and Financial Affairs, Brussels. IMF (2006) De Facto Classification of Exchange Rate Regimes and Monetary Policy Framework. http://www.imf.org/external/np/mfd/er/2006/eng/0706.htm (9.4.2008).

20

IMF (2008), Austria-2008 Article IV Consultation, Preliminary Conclusions of the Mission, March 17, 2008, http://www.imf.org/external/np/ms/2008031708.htm (18.3.2008). Ize, A. and E. Parrado (2002) Dollarization, Monetary Policy and the Pass-Through, IMF Working Paper, No.188. Karmin, Craig and Perry, Jiellen (2007), Homeowners Abroad Take Currency Gamble in Loans, The Wall Street Journal, May 29, 2007 Levy Yeyati, Eduardo (2006), Financial dollarization: evaluating the consequences, Economic Policy, Jan. 2006, 61-118. Luca, Alina and Petrova, Iva (2007), What drives credit dollarization in transition economies? Journal of Banking and Finance, doi: 10.1016/j.jbankfin.2007.06.003. Tieman, Alexander (2007), Cross-Border Banking Issues for the Austrian Banks and Their Supervisors, in IMF (ed.): Austria: Selected Issues, IMF Country Report No. 07/143, April 2007, 30-43. Martens, Jul (2003), Statistische Datenanalyse mit SPSS für Windows, 2. Auflage, München und Wien, Oldenbourg. OeNB (2007), Austrian Financial Intermediaries Benefit from the Benign Economic Climate, Financial Stability Report 13, 37-65, http://www.oenb.at/en/img/fsr_13_tcm16-57616.pdf (9.4.2008). RZB Group (2004), CEE Banking Sector Report, Raiffeisen Research, October 2004. RZB Group (2005), CEE Banking Sector Report Part I, Raiffeisen Research, October 2005. RZB Group (2006), As good as it gets… CEE Banking Sector Report, Raiffeisen Research, September 2006. RZB Group (2007), The Heat goes on. CEE Banking Sector Report, Raiffeisen Research, October 2007. Tzanninis, Dimitri (2005), What Explains the Surge of Foreign Currency Loans to Austrian Households? In Epstein, Natan and Tzanninis, Dimitri (eds).: Austria: Selected Issues, IMF Country Report No. 05/249: 3-37, http://www.imf.org/external/pubs/ft/scr/2005/cr05249.pdf (6.4.2008). Uzun, Arzu (2005), Financial Dollarization, Monetary Policy Stance and Institutional Structure: The Experience of Latin America and Turkey, Thesis submitted to the Graduate School of Social Sciences

of

Middle

East

Technical

University,

http://etd.lib.metu.edu.tr/upload/12606739/index.pdf (5.4.2008). Waschiczek, Walter (2002), Foreign Currency Loans in Austria - Efficiency and Risk Considerations, in:

21

OeNB, Financial Stability Report 4, Wien, 83-99, http://www.oenb.at/en/img/fsr_04_tcm168061.pdf (9.4.2008). World Bank (2007), World Bank EU 8+2 Regular Economic Report, Part II: Special Topic January 2007, http://siteresources.worldbank.org/INTECA/Resources/EU8+2_SpecialTopic.pdf (30.3.2008).

22

7 APPENDIX Detailed statistics on all 55 regression analyses have been omitted for space considerations, but are available from the authors upon request (please contact the corresponding author).

Variable name FX_Loans_pc_of_total_loans

N 128

Minimum 10.10

Maximum 82.10

Mean 40.3041

Std. Deviation 20.63653

FX_Loans_pc_of_GDP FX_Deposits_ps_of_GDP Asset_Share_Foreign_Banks

117 131 123

.14 .78 .30

27.85 25.10 100.00

4.7999 7.0447 56.5033

4.40740 5.81661 32.09626

Asset_Share_Austrian_Banks

56

1.10

62.60

24.3393

17.67158

Asset_Share_5_largest_banks

144

35.32

99.45

65.8949

15.68209

Exports_in_pc_of_GDP Interest_Differential_Germany

158 161

2.24 -4.17

53.16 68.30

17.6214 6.0658

11.06771 12.68509

Interest_Differential_Eurozone

161

-3.60

71.56

8.7616

12.29795

Inflation_Differential Real_Exchange_Rate_Change

162 153

-1.17 -50.13

293.73 24.09

12.0663 2.0734

28.94528 8.14839

TABLE 13 Descriptive statistics of the regression variables

23

Correlations

FX_Loans_pc_of_total_loa ns

Pearson Correlation

FX_Loans_pc_ of_total_loans

FX_Loans_pc_ of_GDP

FX_Deposits_p s_of_GDP

Asset_Share_F oreign_Banks

Asset_Share_A ustrian_Banks

Asset_Share_5 _largest_banks

Exports_in_pc_ of_GDP

Interest_Differe ntial_Germany

Interest_Differe ntial_Eurozone

Inflation_Differ ential

Real_Exchang e_Rate_Chang e

1

.336(**)

-.175

-.147

-.217

.422(**)

-.199(*)

.064

.054

.137

-.196(*)

Sig. (2-tailed) N FX_Loans_pc_of_GDP

Pearson Correlation Sig. (2-tailed) N

FX_Deposits_ps_of_GDP

Asset_Share_Foreign_Ban ks

Asset_Share_Austrian_Ban ks

Interest_Differential_Germa ny

Interest_Differential_Euroz one

Inflation_Differential

Real_Exchange_Rate_Cha nge

.127

.109

.000

.025

.475

.548

.122

.030

116

110

56

124

127

128

128

128

122

.336(**)

1

.314(**)

.141

-.078

.292(**)

.525(**)

-.394(**)

-.384(**)

-.264(**)

-.001

.001

.149

.580

.002

.000

.000

.000

.004

.989

.000 115

117

106

106

52

115

115

117

117

117

111

.314(**)

1

.220(*)

.586(**)

.178(*)

.097

-.287(**)

-.248(**)

-.223(*)

.002

Sig. (2-tailed)

.060

.001

.020

.000

.041

.272

.001

.004

.010

.978

N

116

106

131

111

52

131

130

130

130

131

123

-.147

.141

.220(*)

1

.697(**)

.152

.305(**)

-.463(**)

-.405(**)

-.356(**)

.149

Sig. (2-tailed)

.127

.149

.020

.000

.095

.001

.000

.000

.000

.110

N

110

106

111

123

56

121

121

122

122

123

116

-.217

-.078

.586(**)

.697(**)

1

.543(**)

.072

-.279(*)

-.160

-.399(**)

-.024

.109

.580

.000

.000

.000

.599

.038

.238

.002

.868

56

52

52

56

56

56

56

56

56

56

52

.422(**)

.292(**)

.178(*)

.152

.543(**)

1

.100

-.116

-.094

.029

.028

Sig. (2-tailed)

.000

.002

.041

.095

.000

.237

.169

.263

.731

.743

N

124

115

131

121

56

144

141

143

143

144

136

-.199(*)

.525(**)

.097

.305(**)

.072

.100

1

-.483(**)

-.468(**)

-.222(**)

.147

Sig. (2-tailed)

.025

.000

.272

.001

.599

.237

.000

.000

.005

.074

N

127

115

130

121

56

141

158

157

157

158

149

.064

-.394(**)

-.287(**)

-.463(**)

-.279(*)

-.116

-.483(**)

1

.994(**)

.620(**)

-.084

Sig. (2-tailed)

.475

.000

.001

.000

.038

.169

.000

.000

.000

.301

N

128

117

130

122

56

143

157

161

161

161

152

.054

-.384(**)

-.248(**)

-.405(**)

-.160

-.094

-.468(**)

.994(**)

1

.603(**)

-.077

Sig. (2-tailed)

.548

.000

.004

.000

.238

.263

.000

.000

.000

.345

N

128

117

130

122

56

143

157

161

161

161

152

Pearson Correlation

.137

-.264(**)

-.223(*)

-.356(**)

-.399(**)

.029

-.222(**)

.620(**)

.603(**)

1

-.150

Sig. (2-tailed)

.122

.004

.010

.000

.002

.731

.005

.000

.000

N

128

117

131

123

56

144

158

161

161

162

153

-.196(*)

-.001

.002

.149

-.024

.028

.147

-.084

-.077

-.150

1

Sig. (2-tailed)

.030

.989

.978

.110

.868

.743

.074

.301

.345

.065

N

122

111

123

116

52

136

149

152

152

153

Pearson Correlation

Pearson Correlation

N

Exports_in_pc_of_GDP

.060

115

-.175

Pearson Correlation

Sig. (2-tailed) Asset_Share_5_largest_ba nks

.000 128

Pearson Correlation

Pearson Correlation

Pearson Correlation

Pearson Correlation

Pearson Correlation

TABLE 14 Cross-correlations of the variables used in the regression analyses ** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).

24

.065

153