* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Modes of Foreign Bank Entry and the Effects on
Survey
Document related concepts
Transcript
Modes of Foreign Bank Entry and the Effects on Interest Rate Spreads: Theory and Evidence∗ Sophie Claeys†and Christa Hainz‡ Preliminary and Incomplete Please do not quote without permission November 26, 2004 Abstract Market entry by foreign banks raised considerable concern in the host country, especially in emerging markets. We study the effect of different entry modes on the interest rate for loans in a model where domestic banks possess private information about their incumbent clients but foreign banks have better screening skills. Our model predicts that interest rate spreads are higher if banks enter through acquisition rather than greenfield investment. This result is supported by the results of a regression analysis for a sample of 10 transition countries of Eastern Europe over the period 1995-2002. Keywords: Banking, Foreign Entry, Interest Rate Spread, Asymmetric Information JEL-Classification: D4, G21 ∗ The authors would like to thank Monika Schnitzer and seminar participants at the University of Munich and Ghent University for helpful comments and suggestions. Christa Hainz gratefully acknowledges financial support by FOROST. This research was started while Christa Hainz was visiting SITE. Sophie Claeys acknowledges financial support from the Programme on Interuniversity Poles of Attraction of the Belgian Federal Office for Scientific, Technical and Cultural Affairs, contract No. P5/2. This paper progressed while Sophie Claeys was visiting the Seminar for Comparative Economics at the University of Munich as a Marie Curie fellow. † Department of Financial Economics and CERISE, Ghent University, W. Wilsonplein 5D, B-9000 Ghent, Tel.: 32-9-264 34 91, Fax.: 32-9-264 89 95, e-mail: [email protected]. ‡ Department of Economics, University of Munich, Akademiestr. 1/III, 80799 Munich, Tel.: 49-892180 3232, Fax.: 49-89-2180 2767, e-mail: [email protected]. 1 1 Introduction In many emerging markets, market entry by foreign banks raised considerable concern. On the one hand, emerging markets could benefit from the better technologies that foreign banks use through learning and spillover effects. On the other hand, there was the fear that foreign banks would engage in cherry picking, leaving the domestic banks with a majority of bad loans in their portfolio. Consequently, the policy towards foreign bank entry has been subject to major changes. In many countries the government restricted entry by only allowing foreign investors to buy formerly state-owned banks. Other governments conducted an entry policy which favored acquisition over greenfield investment. Governments encouraged the acquisition of their domestic banks, provided that the foreign investor was willing to take over its bad loan portfolio and induce the sometimes long-awaited sanitation. In this paper, we evaluate bank entry policy by comparing the modes of entry (acquisition versus greenfield investment) and analyzing how this affects the lending rate in emerging bank markets. We investigate how foreign ownership and the mode of entry impact interest spreads both theoretically and empirically. First, we set up a model where banks have different aptitudes to acquire information. Domestic banks possess information about their incumbent firms. Both the domestic and the foreign bank have the same degree of information about firms which have just entered the credit market. However, foreign banks own better screening skills than the domestic bank. A foreign bank enters through a greenfield investment only if its advantage in screening new applicant firms compensates its disadvantage of having no information about incumbent firms. If a foreign bank enters via acquisition, it acquires the credit portfolio which contains information about the quality of incumbent firms. In addition to this acquired information, the bank can generate information by screening applicants. The mode of entry thus influences the distribution of information between domestic and foreign banks, which in turn determines the position of the weaker bank which always is the domestic bank. The stronger the advantage of the foreign bank, the weaker the position of the domestic bank, and the higher the repayment set by the domestic bank. A higher repayment by the domestic bank gives the foreign bank scope to extract rents from the firms. The theoretical model predicts that the average lending rate is higher in the case of acquisition compared to the case of a greenfield investment. Second, we empirically test the predictions of the model for a sample of 10 Eastern European countries for the period 1995-2002. We use the classification by de Haas and van Lelyveld (2002) to analyze the effects of the mode of entry on the aggregate interest rate spread. Consistent with previous results, we find that foreign banks have a negative impact on spreads. Depending on the mode of entry, however, this effect is more pronounced for greenfield compared to acquired banks. This paper is related to both the theoretical and empirical literature of foreign bank entry. The theoretical literature has pointed out the problems of asymmetric information 2 between incumbent banks and new entrants. In Gehrig (1998), the incumbent bank is a monopolist. Both the incumbent bank and the new entrant can screen the firms applying for credit, but the incumbent bank has the first move in the screening game. It can also adjust the precision of its screening procedure. Accordingly, in the case of financial liberalization, the incumbent bank has an incentive to improve its screening technology in order to deter entry by foreign banks. Hence, credit markets are not easily contestable. In Dell’Ariccia et al. (1999), there exists a barrier to entry because banks possess different levels of information which creates an adverse selection problem. In their model, there are two banks with private information about the customers whom they financed in the past. When new firms enter the credit market, neither bank has information about their type. In a first step, Dell’Ariccia et al. (1999) demonstrate that the smaller of the two banks makes zero expected profit. This result is used to show that a new entrant would make an expected loss, because it faces a higher share of unprofitable firms switching from the incumbent bank to the new entrant which has less information. Dell’Ariccia and Marquez (2004) extend the model which also features two banks and now assume that one of the lenders possesses an informational advantage. They study the case where the bank with less information capital has a cost advantage in extending credit. They further show that spreads are higher in markets characterized by more severe information asymmetries. As a consequence of the higher spreads, it is profitable to finance borrowers of lower profitability. If an uninformed lender enters, the incumbent bank reacts and finds it more profitable to lend to firms in more opaque sectors. The empirical facts of bank market entry differ substantially between developing and industrialized countries. In Europe, for instance, the market share of foreign banks does not exceed ten per cent. This is surprising, because there are no formal restrictions on market entry. Interestingly, foreign owned banks have a lower profitability than domestic banks (Claessens et al., 2001)1 . Rather the opposite situation is found in developing countries. Foreign ownership in these countries increased significantly during the last decade and a majority of assets is now owned by foreign banks. Furthermore, foreign banks have a higher profitability than domestic banks in developing markets (Claessens et al., 2001). Some authors have investigated the mode of bank entry empirically for emerging markets. Martinez Peria and Mody (2004) analyze how foreign bank participation and market concentration affect bank spreads in a sample of five Latin American countries during the late 90ies. Their results suggest that foreign banks are able to charge lower spreads compared to their domestic counterparts. Furthermore, they find that acquired banks have relatively higher spreads than greenfield (de novo foreign) banks. The authors give two possible explanations of why greenfield banks charge even lower spreads compared 1 See also De Young and Nolle (1996) and Berger et al. (2000) who find that foreign banks are less efficient than domestic banks in developed markets. 3 to the acquired banks. First, de novo banks might be more concerned with gaining market share compared to the acquired domestic banks, which leads them to set prices more agressively, benefiting domestic borrowers. Second, due to differences in initial informational conditions, both types of banks might not be able to charge the same spreads. More specifically, as is argued by Dell’Ariccia and Marquez (2004), de novo banks will have little information on domestic borrowers and will thus focus on more transparent (and thus more competitive) market segments. Furthermore, Martinez Peria and Mody (2004) find evidence that acquired banks have higher costs than other foreign banks, making it difficult to compete by offering lower spreads. Eastern Europe is a region with one of the highest shares of foreign participation in the banking sector (Papi and Revoltella, 2000). Some studies therefore have focused on the effects of foreign entry in the transition economies of Eastern Europe. Bonin et al. (2004) study bank efficiency in six transition economies with a special focus on ownership structure. Their results suggest that foreign owned banks are most efficient and government owned banks are least efficient. Majnoni et al. (2003) investigate whether and how foreign ownership has affected banks’ cost and profit efficiency in Hungary. Using a sample of 26 commercial banks over the period 1994-2000, they analyze how strategic acquisition versus greenfield investment and the adopted management style after acquisition affect efficiency gains. They find that greenfield banks are much more profitable than acquired banks; but not through the lending spread. More specifically, the authors suggest that the acquired bank might have an information advantage which enables it to maintain higher spreads compared to newly established banks. Greenfield banks, however, have higher profits due to a wider range of financial services offered. The authors further suggest that greenfield banks might exhibit higher efficiency in monitoring and screening, which enhances the overall quality of their loan portfolio and thus increases their profits. In contrast to Dell’Ariccia and Marquez (2004), we assume that all banks have identical cost structures. However, foreign banks distinguish themselves from domestic banks by screening applicants. Thus, we model the advantage of foreign banks more explicitly. Furthermore, we compare different modes of entry whereas Dell’Ariccia and Marquez (2004) restrict their analysis to greenfield investment. We will argue that the degree of asymmetric information between banks is much higher in the case of greenfield investment than in the case of acquisition. Moreover, we provide an alternative explanation for the findings of Martinez Peria and Mody (2004). If a foreign bank enters via greenfield investment, its only advantage compared to the domestic bank is the screening capability. This has to compensate the disadvantage the foreign bank has because it does not have information about incumbent firms. If the foreign bank enters via acquisition, it has the advantage of a better screening technology in addition to the information about the incumbent firms which were debtors of the acquired bank. Thus, the foreign bank possesses more information in the case of acquisition. This stronger position of the foreign bank renders competition less intensive. The reason is that the domestic bank is left with a credit portfolio of decreasing quality as the foreign bank’s informational 4 capital increases. As a result, the domestic bank needs a higher interest rate in order to make zero expected profits. This allows the foreign bank to demand a higher interest rate as well. Next to a new theoretical explanation which takes banks’ aptitude of information access into account, depending on the mode of entry, we provide empirical evidence that support our results. We find evidence of a differential impact of foreign banks and domestic banks on interest rate spreads and the mode of entry for a sample of 10 transition countries. This paper is organized as follows. In section 2, the model of bank market entry is presented. We derive the credit contract offered in the case of greenfield investment and in the case of acquisition. The welfare implications, i.e., the effects on the expected repayments, are discussed as well. Section 3 presents the empirical analysis. The testable hypotheses are derived and data and methodology are introduced before interpreting the results. 2 A Model of Bank Market Entry 2.1 Setup of the Model We study the market entry decision of a bank in a one-period framework. We take the bank-firm relationships which have been established in the past as given. To capture these effects, we use a setup that is similar to that of Dell’Ariccia et al. (1999). Before starting the analysis, we describe the characteristics of the borrowers and the banking sector. Firms There are two groups of borrowers. Those borrowers that have established a bank-relationship in the past will be called “old firms”.2 Among these old firms a share p will be profitable in the future and will be referred to as “good firms”. A share (1 − p) will fail, they are called “bad firms”. Through the bank-relationship, the incumbent bank has perfect information about the profitability of the old firms and knows which firm is good or bad respectively. Moreover, there are “new firms” that enter the credit market. No bank has information about the type of an individual firm. However, it is common knowledge that there is a share of q good and a share of (1 − q) bad firms among the new firms. Banks do not know whether a firm is old or new when it applies for credit. The total number of firms is normalized to 1; the share of old firm is µ and those of new firms is (1 − µ). 2 This differentiation is independent of the activity of the firm on the product market. It only refers to the previous relationship with a bank. 5 Firms both old and new can invest in new projects. However, only good firms will be successful, which means that they generate a payoff of X with certainty. Bad firms will always fail. In order to undertake the project, firms need to invest an amount I. Since they do not have own liquid funds, the investment has to be credit-financed. Banks We assume that on the credit market there is Bertrand competition between two banks3 . We will study different scenarios of entry into the banking sector. In each scenario, the foreign bank has a given screening technology and therefore receives a signal about the profitability of the firm, which is correct with probability s. We assume that the foreign bank has a comparative advantage in screening skills. More specifically, we assume that the domestic bank does not possess any screening technology. The assumption that the domestic bank cannot screen can be interpreted as a normalization according to which we measure how much better a foreign bank is able to evaluate credit proposals than the domestic bank4 . The opposite is true for the information about old firms. Here the domestic bank, which was the incumbent bank of the old firms, has perfect information about this group in the population. We assume that no formal information sharing on borrowers’ credit histories occurs. Finally, the cost of raising funds is assumed equal for both domestic and foreign banks and is normalized to 0. We assume that banks do not have any constraints in lending capacity. The credit contract offered specifies a repayment R in the case of success and a repayment 0 in the case of failure. In order to study how information asymmetries affect domestic and foreign banks, we restrict our analysis to uncollateralized credit contracts. Focusing on uncollateralized credit contracts should not be a severe restriction of the contracting space in an emerging market setting. First, in many of these countries, the firms’ asset endowments are often too low such that collateralization would not solve problems of adverse selection. Second, collateralization entails substantial costs for the bank since institutions and the legal system as a whole often function only very imperfectly5 . Finally, we assume that the foreign bank receives the signal without incurring any additional costs. The idea is that the foreign bank uses the screening technology built in the home market, in order to limit the losses in the market it has just entered. Incorporating the screening costs would not change the results but would add an additional term in each calculation. 3 Competition for primary deposits could play a role in the structure of the credit market (Besanko and Thakor, 1987). In many transition economies, former savings banks are still the most important collectors of deposits, which they transfer to the credit-granting banks through the money market (Dittus and Prowse, 1996). Since most credit-granting banks in Eastern Europe are not competing for primary deposits, we neglect the market situation for deposits and focus on the credit market. 4 In the empirical analysis we focus on a sample of transition countries where the screening skills of domestic banks are relatively low. 5 We do not expect any differences between domestic and foreign banks with respect to collateralization. Collateralization would solve the adverse selection problem independent of bank ownership. Consequently, there would not be an effect of the entry mode on social welfare. 6 Timing The timing works as follows. Before the foreign bank enters, the incumbent bank learns the type of all old firms. There are two rounds in which the banks offer credit. First, all banks make offers to new applicants. Second, the incumbent bank makes offers to good old firms. Then, firms choose which bank to lend from and invest. If both banks demand the same repayment, firms apply in proportion to their share in the population. Old firms stay with their incumbent bank if both the incumbent and the outside bank demand the same repayment. Finally, the payoffs realize and firms repay if they are successful. The timing is summarized in the following figure: [Figure 1] There is no equilibrium in pure strategies for the repayment terms. The reason is that a marginal change in the repayments can lead to a discontinues change in the bank’s profits. This is due to the fact that we focus on asymmetric information between banks. If we have incorporated, for example, switching costs, we would be able to derive an equilibrium in pure strategies (as an example see Boukaert and Degryse, 2002). We do not incorporate this here, since we want to restrict our attention to the question on how different entry modes influence the expected repayment through differences in the distribution of information and differences in screening skills. 2.2 Market Entry through Greenfield Investment If the government liberalizes market entry, a foreign bank can undertake a greenfield investment. Establishing a new bank has a fixed cost of K. We now derive the credit contract offered in this scenario. The domestic bank has perfect information about all old firms. Therefore, it will only lend to the good old firms and deny credit to bad old firms since it would mean making an expected loss. Initially, the domestic bank serves all the new firms applying for credit. Since it has no screening skills, the minimal repayment that the domestic bank requires, RD , is determined by the break-even condition when the domestic bank serves the whole market, i.e., if the domestic bank undercuts the repayment demanded by the foreign bank. Formally, this condition can be written as: G qRD − I = 0 or I G = . RD q (1) (2) The minimal repayment of the foreign bank, RFG , is derived by studying the average quality of firms applying when the foreign bank serves the whole market. The foreign bank finances all bad old borrowers that have generated a positive signal. Moreover, it finances all new firms with a positive signal. Since the signal is imperfect, a share of (1 − s) (1 − q) new borrowers receive credit although they are not creditworthy. The break-even condition is given by ¢ ¢ ¡ ¡ (3) µ (1 − s) (1 − p) (−I) + (1 − µ) qs RF − I + (1 − q) (1 − s) (−I) = 0 7 This condition determines the minimal repayment as RFG = I µ (1 − s) (1 − p) + (1 − µ) (qs + (1 − q) (1 − s)) (1 − µ) sq (4) Since the foreign bank has to incur a fixed cost for entering the market, it will do so only if it makes a positive profit on the credit market. This is only possible if its minimal G G repayment satisfies RFG < RD . This implies that, given RD = Iq , the foreign bank has positive profits whenever it serves the whole market. In equilibrium, the banks mix continuously on the range [R, X) or do not bid at all. Given these minimal repayments, banks decide about their required repayment RiG , i = D, F , the cumulative distribution function FiG and the probability of denying credit probG i (D). Proposition 1 shows the resulting equilibrium in mixed strategies. Proposition 1 There exists an equilibrium in mixed strategies with the following features: h i • The domestic bank demands a repayment from the range Iq , X from newly applying firms, according to the following cumulative distribution function: ´ h s I G G FD (R) = (qR − I) qsR−2qsI−I+sI+qI ∀RD ² q , X 2s−1 and does not make an offer with probability probG D (D) = I (1 − q) qsR−2qsI−I+sI+qI . h i • The foreign bank demands a repayment from the range Iq , X from newly applying firms, according to the following cumulative distribution function: ´ h 1 FFG (R) = (qR − I) qsR−2qsI−I+sI+qI ∀RFG ² Iq , X ¡ ¢ and prob RFG = X = qX(−1+s)−I(2qs−s−q) . qsX−2qsI−I+sI+qI Proof. See the Appendix. The foreign bank will only enter the market if it has an absolute advantage in terms of information. When the foreign bank decides to enter, the domestic bank will have an informational disadvantage compared to the foreign bank with respect to the new applicant firms. The domestic bank will therefore stay out of the market with positive G make zero expected profit from the newly probability and with each repayment RD applying customers. For each repayment the foreign bank makes the same expected profit, which is given by: ´ ´ ³ ³ G − I + (1 − q) (1 − s) (−I) (1 − s) µ ((1 − p) (−I)) + (1 − µ) qs RD = I ((1 − µ) (1 − q) (2s − 1) − µ (1 − s) (1 − p)) 8 (5) It is interesting to note that the value of the domestic bank’s cumulative distribution function is always a fraction s of the foreign bank’s cumulative distribution function, i.e., FDA (R) = sFFA (R). The domestic bank stays out of the market of serving new firms with positive probability because it is faced with a so-called “winner’s curse problem”. All new firms that apply to the domestic bank are those which were denied credit by the foreign bank after generating a bad signal. In order to limit the risk to end up with a loss, the domestic bank will deny credit with a positive probability probG D (D). This probability increases as the screening technology of the foreign bank improves. The intuition is that a better screening technology of the foreign bank deteriorates the average quality of firms that apply at the domestic bank. In order to avoid losses, the domestic bank therefore rations credit with a higher probability. Corollary 2 Foreign banks only enter through greenfield investment if its screening skill K +1−(+1−µ)q−µp is high enough, i.e., s > se = I2−µ−µp−2q+2µq . When comparing the minimum interest rate for domestic and foreign banks, we find that the quality of the signal generated by screening has to exceed se such that the foreign bank makes a positive profit. Naturally, the higher the fixed cost of market entry, K, the higher se has to be. The higher I, the amount of credit needed, the lower se has to be. Comparative statics further show that the higher the share of old firms, the higher the screening quality of the foreign banks has to be. This corollary explains why banks find greenfield investment attractive in emerging markets. In these economies, there are many de novo firms which have not yet established a bank-relationship. Therefore, the share of applicants about which neither foreign nor domestic banks know the type is high. With respect to the new firms, the foreign bank has an absolute advantage compared to the domestic bank because it possesses a better screening technology than the domestic bank. Deriving se shows how much better the foreign bank has to be than the domestic bank, which by assumption does not get an informative signal, as we have normalized the quality of the signal of the domestic bank to 0.5, implying that the signal is uninformative. Consequently, better screening skills of domestic banks increase se. This explains why in industrialized countries market entry through greenfield investment is less attractive for foreign banks. 2.3 Market Entry through Acquisition In many cases acquisition happened through the privatization of state-owned banks. We capture acquisition as follows. Assume that initially there exists one state-owned bank. The government splits up the monopolistic bank into two identical banks. One of the banks is sold to a foreign bank for a price P A which is exogenously given6 . When 6 In many emerging markets, the prices were set by the government when the banks were privatized. 9 taking over the bank, the foreign bank acquires information about all the customers of the existing bank. Moreover, it can implement its screening technology without any costs and screening generates the same quality of signals as in the case of greenfield investmentThe domestic bank has information about the quality of its old customers. The idea is that each of the two banks has information about the stock of customers it inherited from the past. One could think of the credit files each successor of the monopolistic bank gets or the staff they continue to employ. The bank staff possesses soft information about the firms that have already established a bank-relationship. Analogously to the case of greenfield investment, we determine the minimal repayments necessary for the domestic and the foreign bank, respectively. When serving the whole market, the break-even condition for the domestic bank is determined by the quality of firms that receive credit. Since the domestic bank does not screen, it serves all customer that apply, i.e., all bad old customers that are switching bank. The domestic bank serves all old bad firms that switch bank, their share in the total population is µ0.5 (1 − p) (−I)), and all new customers. Formally, the break-even condition is given by ³ ´ A µ0.5 (1 − p) (−I) + (1 − µ) qRD − I = 0 or (6) A RD = I 1 − 12 µ (1 − p) . q (1 − µ) (7) In contrast to the domestic bank, the foreign bank screens its applicants. Consequently, the foreign bank finances only those with a positive signal. The break-even condition is ³ ³ ´ ´ µ0.5 (1 − s) (1 − p) (−I) + (1 − µ) qs RFA − I + (1 − q) (1 − s) (−I) = 0. (8) 1 (1−s)µ(1−p)+(1−µ)((1−q)(1−s)+qs) The minimal repayment the foreign bank needs is RFA = I 2 . sq(1−µ) A A It can easily be shown that RD > RF . This implies that the foreign bank has positive A profits whenever it demands exactly RD . Since each repayment has to generate the same expected payoff, the foreign bank makes an expected profit on the credit market. The foreign bank decides to enter the credit market if the expected profit exceeds the acquisition price P A . The following proposition describes the equilibrium in mixed strategies in more detail: Proposition 3 There exists an equilibrium in mixed strategies with the following features: • The domestic ibank demands a repayment from newly applying firms in the range h 1− 12 µ(1−p) I q(1−µ) , X , according to the following cumulative distribution function: h ´ 1− 21 µ(1−p) s(2qR−2µqR−2I+µI+µpI) A FDA (R) = 12 I (qsR−2qsI−I+sI+qI)(1−µ) ∀RD ² I q(1−µ) ,X 10 A and does not make an offer with probability probA D (D) = 1 − FD (X). • The foreign bank i demands a repayment from newly applying firms in the range h 1− 12 µ(1−p) I q(1−µ) , X , according to the following cumulative distribution function: h ´ 1− 12 µ(1−p) (2qR−2µqR−2I+µI+µpI) FFA (R) = 12 I (qsR−2qsI−I+sI+qI)(1−µ) ∀RFA ² I q(1−µ) ,X ¡ ¢ and prob RFA = X = 1 − FFA (X). Proof. See the Appendix. 2.4 Comparison The aim of our study is to determine and compare what credit contracts look like if a foreign bank enters either through greenfield investment or through acquisition. The result is summarized in the following proposition. Proposition 4 The average repayment a foreign bank demands is higher in the case of acquisition than in the case of greenfield investment. Proof. See the Appendix. In order to study the repayments, we compare the cumulative distribution functions. When we compare the expected repayment demanded by the domestic bank with the repayment asked by the foreign bank, the cumulative distribution function of the domestic bank is always a fraction s of the foreign bank’s distribution ¡ cumulative ¢ ¡ ¢function. We can show that both FFA (R) < FFG (R) and prob RFA = X > prob RFG = X hold. Thus, the value of the cumulative distribution function for each R is lower in the case of acquisition than in the case of greenfield. Thus, higher repayments are assigned a higher probability in the case of acquisition. This is confirmed if we look at probability for R = X. The probability with which the bank demands the highest repayment is higher in the case of acquisition. As a consequence, the average repayment is higher in the case of acquisition because domestic as well as foreign banks demand a higher repayment compared to entry via greenfield investment. The explanation for this result is as follows. The market entry strategy influences how information about old firms is distributed between the two competing banks. In the case of greenfield investment, the domestic bank has information about all old firms whereas the foreign bank does not have any information about old firms. In contrast, in the case of acquisition, each bank knows the quality of its old customers. The mode of 11 entry determines the distribution of information, and, consequently, the position of the weaker bank. Independent of the mode of entry, the domestic bank is always the weaker bank. However, its information disadvantage of the domestic bank with respect to the foreign bank depends on the entry mode. The degree to which the domestic bank has a disadvantage in terms of information determines the scope of the foreign bank to extract higher rents from the firms, which influences the minimal repayment required. In the case of greenfield investment, the relative position of the domestic bank is determined by the amount of information it has about its old customers compared to the screening skills of the foreign bank. The relative position of the domestic bank is weaker in the case of acquisition compared to the case of greenfield. Now the foreign bank not only possesses a better screening technology but has information about old firms as well. Consequently, the foreign bank is better able to exclude bad old firms from receiving credit than in the case of greenfield investment. This also implies that the domestic bank receives more applications from bad old firms. Due to the lack of screening techniques, it finances the bad old firms coming from the foreign bank. This means that the domestic bank needs a higher repayment on average compared to the foreign bank in order to make zero expected profits. Along the same line of arguments, the domestic bank rations credit with a higher probability since it wants to avoid making losses. 3 3.1 Empirical Analysis Data We use a sample of 10 transition countries, namely Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Slovakia and Slovenia for the period 19952001. For these countries, we use bank-specific ownership data taken from de Haas and van Lelyveld (2002) which specifies whether a bank is domestically owned, whether and when it has been acquired or whether it is a greenfield investment. We match all banks for which de Haas and van Lelyveld (2002) provide ownership data with data on balance sheet and income and loss accounts taken from the BankScope database maintained by Fitch/IBCA/Bureau Van Dijk. Consolidated statements were preferred but unconsolidated statements were used when the consolidated one was not available. The macroeconomic data on GDP growth and inflation was taken from various EBRD reports (2001, 2003, 2004 update). Data on lending an deposit rates was taken from the IMF International Financial Statistics. Table 2 (panel a) reports the number and market shares of domestic and foreign banks included for each year and country. In Hungary, foreign banks already outnumbered domestic banks in 1995 and represent almost 80 percent of the market. Due to the Hungarian privatization strategy that started in the early 80ies, the share of foreign banks has gradually risen. In the beginning of the 90ies, the Polish government sold 12 its weak domestic banks to foreign owners. These privatizations are marked as a set of greenfield investments that ocurred before 1995. The cumulative market share for greenfield banks is, however, relatively small (between 2.8 and 8.6 percent) compared to the asset share in other countries. Since 1999 the government started to sell majority shares to foreign investors. This lead the number of foreign banks in Poland to exceed the number of domestic banks and dominate the market in terms of assets since 2001. Table 2 shows that there is a considerable amount of foreign entry ocurring in most countries included in our sample. Furthermore, the market shares of foreign banks have gradually risen and are starting to dominate the market (see table 2b that reports average market shares). Estonia, for example, has only two foreign owned banks (Hansabank and Eesti Ühispank), but these account for over 90 percent of assets since 2000. 3.2 Empirical predictions First, we do not formally model the comparison between a closed economy with only domestic banks and the situation after market entry. The impact of market entry on the average interest rate spread is therefore theoretically not derived. One can argue that market entry by foreign banks changes the market structure and increases competition which consequently leads to a lower average spread. However, in our model, foreign banks always undercut the domestic bank, but the domestic bank might well be asking higher rates because of the increase in asymmetric information. In this case, the foreign bank also asks higher rates (only marginally below the domestic bank), which might generate an increase in the average spread. We analyze how the relative presence of foreign to domestic banks impacts the banking market spread to test the impact of foreign entry. Consistent with previous empirical work, we expect that the higher the share of foreign to domestic assets, the lower the average spread will be. Second, when entry occurred via a greenfield investment, the average interest rate will be lower than in the case of acquisition. This is because the acquired bank has a larger comparative advantage in information distribution compared to the case of a greenfield investment. So the higher the share of greenfield investments compared to acquired banks, the lower will be the average interest rate offered to borrowers. This can be tested by looking at the differential impact of acquired versus greenfield banks on the average interest rate spread. Third, the domestic bank will ration credit (for new applicant firms) with a higher probability when the foreign bank entered via acquisition compared to a greenfield investment. We test this by analyzing how the mode of entry affects average credit growth of the domestic banks. We excpect that average domestic credit growth will be lower in a market where acquired banks are predominantly present relative to greenfield banks. 13 3.3 Estimation procedure [to be completed] 4 Conclusions 5 References Berger, Allen N., DeYoung, Robert, Genay, Hesna and Udell, Gregory F., 2000, Globalization of financial institutions: Evidence from cross-border banking performance, in: Brookings-Wharton Papers on Financial Activity, 3, 23-158. Besanko, David, and Thakor, Anjan V., 1987, Collateral and Rationing: Sorting Equilibria in Monopolistic and Competitive Credit Markets, International Economic Review 28 (3), 671-689. Bonin, John P., Miszei, K., Székely, István P. and Wachtel, Paul, 1998, Banking in Transition Economies: Developing Market Oriented Banking Sectors in Eastern Europe, Edward Elgar Publishing: Cheltenham and Northampton. Bonin, John P., Hasan, Iftekhar and Wachtel, Paul, 2004, Privatization Matters: Bank Efficiency in Transition Countries, Journal of Banking and Finance X (X), 2005, 31-53.. Boukaert, Jan and Degryse, Hans, 2002, Entry and Strategic Information Display in Credit Markets, mimeo, Katholieke Universiteit Leuven, January 2002. Claessens, Stijn, Demirgüç-Kunt, Asli, Huizinga, Harry, 2001, How Does Foreign Entry Affect Domestic Banking Markets, Journal of Banking and Finance, 25, 891-911. Claessens, Stijn, Demirgüç-Kunt, Asli, Huizinga, Harry, 2000, The Role of Foreign Banks in Domestic Banking Systems, in: Claessens, Stijn and Jansen, Marion (Eds.), The Internationalization of Financial Services, Kluwer Law International, The Hague, November 2000, 117-137. Clark, George, Cull, Robert, Peria, Maria Soledad M. and Sánchez, Susana M., Foreign Bank Entry: Experience, Implications for Developing Counties, and Agenda for Further Research, World Bank Research Observer 18 (1), Spring 2003, 25-59. de Haas, Ralph T. A. and van Lelyveld, Iman P. P., 2002, Foreign Bank Penetration and Private Sector Credit in Central and Eastern Europe, De Nederlandsche Bank Staff Reports, No. 91/2002. Dell’Ariccia, Giovanni, Friedman, Ezra and Marquez, Robert, 1999, Adverse selection as a Barrier to Entry in the Banking Industry, RAND Journal of Economics 30 (3), 515-534. 14 Dell’Arricia, Giovanni and Marquez, Robert, 2004, Information and Bank Credit Allocation, Journal of Financial Economics, Forthcoming. DeYoung, Robert and Nolle, Daniel, 1996, Foreign-owned banks in the US: Earning market share or buying it? Journal of Money, Credit, and Banking, 28 (4), 622-636. Dittus, Peter, and Prowse, Stephen, 1996, Corporate Control in Central Europe and Russia. Should Banks Own Shares?, in: Roman Frydman, Cheryl W. Gray, and Andrzej Rapaczynski (Eds.): Corporate Governance in Central and Eastern Europe and Russia, Vol. 2., Insiders and the State, Budapest et al.: Central European University Press, 1996, 20-67. Gehrig, Thomas, 1998, Screening, Cross-Border Banking, and the Allocation of Credit, mimeo, University of Freiburg, August 1998. International Monetary Fund International Financial Statistics, Yearbook 2003. Laeven, Luc and Majnoni, Giovanni, 2003, Does Judicial Efficiency Lower the Cost of Credit?, World Bank Policy Research Working Papier 3519, Washington D.C.: Worldbank, October 2003. Majnoni, Giovanni, Shankar, Rashmi and Vàrhegyi, Eva, 2003, The Dynamics of Foreign Bank Ownership: Evidence from Hungary, World Bank Policy Research Working Paper 3114, August 2003. Müller, Thomas, 2002, Modes of Foreign Entry under Asymmetric Information about Potential Technology Spillover, mimeo, University of Munich. Naaborg, Ilko, Scholtens, Bert, Bol, Hanneke, De Haan, Jakob and de Haas, Ralph, 2003, How Important are Foreign Banks in the Financial Development of European Transition Countries? CESifo Working Paper Series, December 2003, No. 1100. Papi, Luca and Revoltella, Debora, 2000, Foreign Direct Investment in the Banking Sector: A Transitional Economy Perspective, in: Claessens, Stijn and Jansen, Marion (Eds.): The Internationalization of Financial Services – Issues and Lessons for Developing Countries, Kluwer Law International, The Hague, November 2000, 437-457. Peria, Maria Soledad M. and Mody, Ashoka, 2004, How Foreign Participation and Market Concentration Impact Bank Spreads: Evidence from Latin America, Journal of Money Credit and Banking, 36 (3, 2)), 539-542. 15