The Effect of Foreign Entry on Bank Performance

(1)

Faculty of Economics & Business

MSc International Economics and Business

The Effect of Foreign Entry on Bank Performance

Case of the SADC countries, 1998-2005

Author:

Stephanie van Rijckevorsel

Supervisors:

M. Koetter A. Meesters

(2)

The Effect of Foreign Entry on Bank Performance: Case of the SADC countries, 1998-2005

The Effect of Foreign Entry on Bank Performance

Case of the SADC countries, 1998-2005

ABSTRACT

In this research I investigate whether foreign ownership enhances bank performance, from 1998-2005 in the Southern African Development Community (SADC). I will do this by testing hypotheses, drawing upon Claessens, Demirguc-Kunt and Huizinga (2001) and the existing literature on the role of foreign banks in developing countries. Moreover, I test whether a host country’s level of control on corruption affects the relationship between foreign ownership and bank performance. Using a sample of 344 observations in 14 countries the results show that foreign banks are indeed significantly more profitable than their domestic counterparts. Unfortunately, no significant results are found on the relationship between control of corruption and bank performance of foreign banks in the SADC.

Stephanie van Rijckevorsel ID: S1323288 stephanie@rijckevorsel.net

(3)

TABLE OF CONTENT

1. INTRODUCTION 4

2. LITERATURE REVIEW 7

2.1 Foreign bank ownership

& the effects of foreign entry on domestic banking 8

2.2 The Southern African Development Community 11

2.3 Corruption in the SADC 14

3. EMPIRICAL FRAMEWORK 15

3.1 Bank performance 15

3.2 Independent variables 16

3.2.1 Bank speficic characteristics 16

3.2.2 Country specific characrteristcs 17

3.2.3 Foreign ownership 18

3.2.4 Year variable 18

3.3 Regression model 19

4. DATA 20

4.1 Data sources and sample construction 20

(4)

The Effect of Foreign Entry on Bank Performance

Case of the SADC countries, 1998-2005

Stephanie van Rijckevorsel

University of Groningen

1. INTRODUCTION

International banking is booming since the nineties. The increase in international trade in goods and financial services, the increase in foreign investment activities, the globalization of capital markets, and the liberalisation of domestic financial

markets during the past few decades have stimulated international banking activities1.

International banking activities may involve cross-border activities and activities of banks outside their home-country. Banks expand internationally by establishing foreign subsidiaries and branches or by taking over established banks. Both developed and developing countries now increasingly allow banks to be foreign owned and allow foreign bank entry on a national treatment basis (Claessens et al., 2001, p. 892). Recent years have seen an increased interest of policy makers and academics in the activities of these foreign banks, particularly because in many emerging market economies the presence of foreign banks has increased dramatically, especially during the late 1990s.

This paper aims to provide an empirical study of how foreign banks presence has affected bank performance in the 14 Southern African Development Community (SADC)

1_{For an early survey of the literature on the internationalization of banking the reader is referred to Aliber}

(5)

countries2. I will use bank-level accounting data and macroeconomic data for the

1998-2005 period. Drawing upon Claessens et al. (2001), I first examine the scale of foreign bank operations in each of the 14 countries. I define a bank to be foreign owned if at least 50% of its shares are foreign owned. Then I extend the work on the accounting decomposition of interest margins by Hanson and Rocha (1986), and use the data to examine how foreign banks differ from domestic banks in terms of interest margins. Claessens et al. (2001) analyse the effect of foreign ownership on interest margins, taxes paid, overhead expenses, loan loss provisioning, and profitability, and find that foreign banks in developing countries tend to be more profitable than domestic banks are, while in developed countries the

domestic banks are more profitable than foreign banks are3_{. Using net interest revenues as}

financial indicator for the research on foreign bank profitability, I make the proposition that a rise in foreign ownership increases net interest revenues in the SADC countries.

Next, I continue the study by examining a condition on a higher abstraction - control of corruption - that may especially explain bank performance differences in African countries. To my knowledge there is little research done that provides some evidence on the institutional effects of corruption in Sub-Saharan Africa on the relationship between foreign ownership and banks’ performance. However, based on the existing literature (Naaborg, 2007; Figuiera et al., 2005, Berger et al., 2000) I propose that foreign institutions in particular are subject to corrupt legal or public officers. I suggest this, because domestic banks may have a comparative advantage due to the home field advantage hypothesis. The home field advantage hypothesis predicts that domestic owned banks have a comparative advantage to foreign banks. Foreign banks are assumed to perform less well than domestic banks due to higher costs of providing the same financial services or due to lower revenues.

2_{The fourteen member states of the SADC are Angola, Botswana, the Democratic Republic of Congo,}

Lesotho, Madagascar, Malawi, Mauritius, Mozambique, Namibia, South Africa, Swaziland, United Republic of Tanzania, Zambia and Zimbabwe.

3_{According to Naaborg (2007), these differences can be explained by informational (dis)advantages,}

(6)

Factors leading this home field advantage are the advantages domestic banks have in local language, culture, economy, legal conditions, and politics. This leads to the assumption that foreign banks suffer more from a bad institutional framework in the host country than domestic banks (Naaborg, 2007). However, if public action is taken to control corruption, foreign banks may be less affected and thus increase bank performance. Especially in the 14 African countries of interest, where the lack of regulatory structures gives way for corruption practices. Therefore, I make the proposition that the higher the level of control of corruption in the host country - of a foreign owned bank - the higher perfomance results can be achieved in the SADC countries. To test this I use a worldwide governance indicator

developed by Kaufmann et al. (2007) measuring control of corruption4_{. The dimension of}

control of corruption measures the extent to which public power is exercised for private gain, including both petty and grand forms of corruption, as well as “capture” of the state by elites and private interests (Kaufmann et al., 2007).

Not many studies have been published that focus on the relationaship between foreign bank entry and bank performance in developing countries only. Recent literature mainly covers foreign bank entry in industrialized countries and transition economies (e.g. European transition economies, CEE). There are few studies on developing countries in Latin America and East Asia, and some literature that addresses the efficiency effects on developing economies in Sub-Saharan Africa (Clarke et al., 2000; Figueira et al. 2005). Because the largest part of the performance literature relates to developed countries, and initial indications are that many of the results for developed countries do not carry over, further examination is desirable in developing countries in Africa.

In all, my thesis aims to study foreign bank ownership in the Southern African Development Community (SADC) by examining the following research questions:

4_{The worldwide governance indicators measure six dimensions of governance in 212 countries: voice and}

(7)

How did foreign ownership affect the performance of a bank in the SADC countries in the period from 1998 to 2005, and what effect did the level of corruption control have in the specific countries on this relationship?

Drawing on the literature this paper tests two particular research hypotheses:

(H1) For developing countries in the SADC from 1998-2005, Foreign Ownership increases Bank Performance.

(H2) For developing countries in the SADC from 1998-2005, high levels of Control of Corruption enhance the positive relationship between Bank Performance and Foreign Ownership.

The paper is organized as follows. The next section discusses the related literature on foreign bank ownership and its effect on performance in the banking sector. Furthermore, it discusses the setting of the study, the Southern African Development Community, the African financial sectors and the issue of corruption that may influence the foreign ownership-bank performance relationship. The following section provides the empirical framework for the study with a description of the data and performance measures used in the study, and the method used to assess performance. This is followed by the estimation of the effect of foreign entry on individual bank performance and the presentation of the findings in the results section. Finally, the paper concludes by summarizing the main findings and provides some limitations and suggestions for further research.

(8)

This section is divided in three sub-sections. Sub-section 1 reviews the concept of foreign bank ownership and discusses the literature regarding the effects of foreign bank entry on domestic banking sectors. Sub-section 2 provides insight on the setting of this study, the Southern African Development Community (SADC). And finally, sub-section 3 reviews the institutional concept corruption in the SADC.

2.1 Foreign bank ownership & the effects of foreign entry on domestic banking sectors

(9)

Several issues on the effects of foreign bank entry on domestic financial markets and institutions have been discussed. Many support the argument of lifting restrictions on foreign bank entry and address the potential benefits of foreign bank entry for the domestic economy in terms of better resource allocation and higher efficiency. Specifically, the presence of foreign banks may improve the quality and availability of financial services in the domestic financial market by increasing competition (i.e. stimulate domestic banks to reduce costs, increase efficiency and increase diversity of financial services). Further, foreign bank entry may enable the application of more modern banking skills and technology. Also, foreign banks may enhance a country’s access to international capital. And finally, foreign banks may increase the quality of human capital in the domestic banking market, e.g. through planning, credit evaluation & marketing and recruitment (Cho, 1990; Gelb and Sagari, 1990; Levine, 1996; Buch, 1997; Berger and Hannan, 1998; Denizer, 2000; Claessens et al., 2001; Lensink and Hermes, 2004). Other authors have also discussed the potential costs accompanying foreign bank entry. For instance, costs may incur domestic banks, which have to compete with large international banks with a better reputation; And, for local businesses foreign bank entry can result in less access to financial services since foreign banks focus on multinational firms (Stiglitz, 1993).

(10)

(11)

profits. There were thus many domestic failures in the mid-1990s, but the banks that failed were not heavily concentrated in the types of lending favored by foreign banks (Clarke et al., 2000). By contrast, fewer studies have looked at bank performance in developing

countries5. A notable exception is the empirical study by Claessens, Demirguc-Kunt and

Huizinga (2001). They analyse the extent of foreign ownership in national banking markets and point out the benficial effects of foreign bank entry. Using bank-level data for 80 countries (developed and developing countries) from 1988-1995, they compare net interest margins, overhead, taxes paid, and profitability of foreign and domestic banks. They find that foreign banks are more profitable than domestic banks in developing countries. In industrialized nations, however, the opposite is found. Table 1 provides an overview of the literature discussed in this section on foreign bank ownership and bank performance.

2.2 The Southern African Development Community

The rise in foreign banking has been brought about by international liberalization and harmonization of the financial sector. Foreign banking is particularly important in Central Asia, Central and Eastern Europe, Latin America, the Caribbean and Sub-Saharan Africa (Naaborg, 2007). This paper is specifically concerned with foreign banking in Sub-Saharan Africa, namely in the Southern African Development Community (SADC). Since a decade the SADC region has been in the throes of a process of integration around critical areas of the economy. Besides infrastructure, education, knowledge and networks, investments are critical in this process, because they provide oil for the economic engine of

5_{A developing country is a country which has a relatiely low overall standard of living The country}

(12)

the region. Furthermore, the SADC provides small, but fast growing markets. And, lots of countries in the SADC have a large amount of important natural resources, that enhance the

potential of the economic markets6.

In spite of this, there are several risks for banks in Africa. Financial sectors in

Sub-Saharan African (SSA) countries are considered among the least developed in the world7.

According to the IMF (International Monetary Fund, 2000) the financial sector of several African countries lacks depth, is highly fragmented, and is exposed to significant risks. Access to financial services is among the lowest in Africa. Weaknesses in cash management practices by governments and in the systemic liquidity management framework exacerbate financial sector vulnerabilities. Furthermore, financial supervisors, particularly in the banking sector, are well trained, but the interference of the national authorities in some aspects of financial supervision, together with a shortage of staff, limits the overall effectiveness of the supervisory framework. Moreover, authorities view the risk of money laundering as high, in particular the risk of money laundering tied to corruption, embezzlement of public funds, and fraud linked to oil production, in addition to criminal activities such as smuggling, counterfeiting, and trafficking (in precious stones, weapons, and narcotics). Although organized criminal activity differs from one country to another, there is a general risk here. Western economic agents also perceive country risks of Sub-Saharan Africa that effect the banking sectors. These country risks include for one, the risk of political breakdown. Property is also endangered in Africa by the risk of damage associated with riots, civil wars, or vandalism (Terrier, 1994). The lack of a well grounded judicial system and policing system, and their enforcement, enhance these country risks.

6_{For example, Botswana, Namibia and South Africa export diamonds, Mozambique exports aluminium and}

timber, Zambia and Congo export copper and Angola exports oil and uranium.

7_{To some extent this can be blamed on misguided policies of the past that encouraged substantial political}

(13)

Steps have been taken to strengthen the banking sector by regional commissions

(e.g. the COBAC8 and the SADC Banking Association) to implement the regional

framework for Anti-Money Laudering- Combating the Financing of Terrorism (AML-CFT), but according to the IMF, the division of labour between and accountability of the regional and national authorities needs to be clarified. In particular, the national authorities should confirm their commitment to AML-CFT by promptly establishing the national Financial Intelligence Units and building up their operational capacities. More emphasis needs to be placed on the benefits of an effective AML-CFT framework, in terms of good governance, the fight against corruption and the trafficking of natural resources.

However, opportunities do exist in potential high profits, that may outweigh the costs accompanying potential risks. Opportunities exist in the demographic growth, economic growth and especially in the urban demography in Sub-Saharan Africa. The rate of urbanisation in Africa is twice that in Asia and is expected to rise from 30 percent in 1965-1970 to 65 percent in 2015. The combination of rapid urbanisation and regional integration could provide an attractive market (Terrier, 1994). According to the IMF as well, Africa is enjoying its best period of sustained economic expansion since independence. Real GDP growth has been growing and is expected to rise from 5,7% in 2006 to 6,8% in this year. This good performance is partly driven by high commodity and oil prices. Foreign aid has also helped. But this is according to several economics due to better economic management, more openness and more stable politics. Such policies mean banks have to work harder to make a profit, but also help them grow (World Bank, 2007). Furthermore, international organizations such as the World Bank and IMF have been encouraging privatization and foreign involvement in the African economies the last decade (IMF, 2000). Their public policy agenda stimulates foreign entry along with the opportunities in the in the African markets.

(14)

In all, the risks are high for banking in Africa, but the possibilities for profits exist. Therefore, the interst in banking in Africa has increased and provides an incentive for further development and research. The developing economies and the financial sectors in the Southern African Development Community seem to be a suitable setting for testing the validity of extant theoretical arguments on foreign bank entry in a new and under-explored volatile environment.

2.3 Corruption in the SADC

(15)

that foreign institutions in particular are subject to corrupt legal or public officers. Therefore, country differences on the level of control of corruption in the 14 African countries may influence banks’ performance, and are of interest in this study. However, it should be noted that even with the publications indices, the case could be that too many actors have an interest in keeping the corruption habit going. Foreign investors, especially in Africa’s oil industry, have long paid massive bribes to politicians to win contracts or the rights to pump. (The Economist Newspaper and The Economist Group, 2008).

3. EMPIRICAL FRAMEWORK

This section presents the empirical framework that I use to test the two hypotheses of this study. Firstly, sub-section 1 describes the measure I have chosen for bank performance. Next sub-section 2 describes the independent variables chosen. And finally, sub-section 3 presents the regression model with all the variables of this study.

3.1 Bank performance

The dependent variable this study examines is bank performance. Drawing upon the literature (Claessens et al., 2001; Naaborg, 2007), I have chosen one financial indicator to measure bank performance and on which to analyse the effect of foreign ownership: net interest revenues (NETINTREV). Claessens, Demirguc-Kunt, and Huizinga (2001) use three performance measures: (1) net interest revenues, (2) overhead costs and (3) profit before taxes. These financial indicators are found from the following accounting identity that is driven from the bank’s income statement: net margin/ta + non-interest income/ta =

before tax profits/ta + overhead/ta + loan loss provisioning/ta9_{. Numerous authors construct}

9_{The first two ratios are the accounting value of a bank’s net interest income over total assets, or net}

(16)

dependent variables from this accounting equation in their analyses on foreign ownership and bank performance (e.g. Demirguc-Kunt and Huizinga, 2000; Clarke et al., 2000; Denizer, 2000; Naaborg, 2007). There are several reasons why studies focus on accounting measures of income and profitability rather than rates of return on stocks. These include firstly, that financial returns on bank stocks are equalized by investors in the absence of prohibitive international investment barriers or other transaction costs. Comparing these returns across different banking systems would thus not allow for investigation on the effects of different degrees of competition and the effects of foreign bank entry. Also, financial returns data are not available for a similarly large set of banks and countries (Claessens et al., 2001, p. 895). For the 14 African countries in my sample, data problems as well incurred in terms of rates of return on stocks. I therefore chose to use an accounting measure. And because the accounting measure net interest revenues resulted in the largest data sub-set, I have finally chosen to simply use this financial indicator for bank performance in my sample. In the regression model net interest revenues is scaled over total assets.

3.2 Independent variables

A large number of variables may influence the level of bank performance, measured by net interest revenues. Besides the main independent variable foreign ownership, variables related to bank specific characteristics and country specific characterictics influence bank performance. These are included in the regression analysis to control for their influences. In addition, year variables are constructed to control for time fixed effects. The following sections discuss all independent variables chosen for the regression.

(17)

3.2.1 Bank specific characteristics

For the bank specific variables I have drawn upon the bank specific control variables used in the related literature (Demirguc-Kunt and Huizinga, 2000; Claessens et al., 2001; Naaborg, 2007). Firstly, there are several measures that could proxy for bank size, of which the number of employees and total assets are most commonly used. I have chosen total assets (TA) as a proxy of bank size, because of this measure the largest amount of observations could be obtained, compared to the number of employees of each bank used in the sample. Secondly, I have chosen equity scaled over total assets (EQ) (leverage ratio) to control for a scale bias making large banks more efficient than small banks by virtue of the equity they have built up over time (Berger and Mester, 1997). Thirdly, loan loss provisions (LLP) are included. This measures the new provisioning made during the accounting year for any previously contracted credits. Differences between domestic and foreign banks here may reflect a difference in customer mix, with foreign banks focusing on large corporations (Claessens et al., 2001 p. 903). And finally, I include net loans (NETLOANS) as a proxy for bank risk10. Financial risk is the additional risk a shareholder bears when a bank uses debt in addition to equity financing. Banks that issue more debt instruments would have higher financial risk than banks that finance more by equity. Differences in this bank risk are controlled for in the regression by net loans.

3.2.2 Country specific characteristics

In the literature on foreign ownership and bank performance the most used country control variables are GDP growth rates, real interest rates, nominal interest rates, GDP per capita, inflation rates and GDP. I have chosen to draw upon the related literature and use the following country variables: GDP per capita, annual GDP growth rate and, real interest rate. To control for differences in the level of economic development of the host countries, I have firstly chosen GDP per capita (GDPPC), because it can be used as an

(18)

indicator of the average standard of living in an economy, and it is measured frequently, widely and consistently. Secondly, economic growth (GDPGROWTH) is widely measured as the percent rate of increase in real GDP. And as GDP per capita of an economy is often used as an indicator of the standard of living, economic growth is often seen as indicating an increase in the average standard of living. And finally, real interest rate (IR) can explain international capital movements (capital flight), which is of interest for the research on foreign banking. International capital moves to markets that offer higher real interest rates from markets that offer low or negative real interest rates, which can trigger speculation in

equities, estates and exchange rates (World Bank, 2005)11_.

In addition, to examine whether control of corruption (CC) of the host country affects the relationship between foreign ownership and bank performance, I will interact foreign ownership with the governance indicator developed by Kaufmann et al. (2007), control of corruption (FOS*CC). Assuming foreign banks are more subject to corrupt practices, I expect higher levels of control of corruption interacted with the foreign ownership variable to explain higher levels of net interest revenues.

3.2.3 Foreign ownership

The independent variable foreign ownership (FOS) is of primary interest in this analysis. Most studies on foreign banking use a foreign ownership dummy, which, to ensure foreign control of operations, defines a bank as foreign if at least 50 percent of its shares is foreign owned (Claessens et al., 2001; Naaborg, 2007). I will also measure presence of foreign banks by a dummy variable. The constructed dummy variable is coded 1 if the bank is defined as foreign, and 0 otherwise.

3.2.4 Year variable

(19)

Furthermore, I will control for time fixed effects in the sample, by including a year variable in the regression. Year 1998 will be taken as base year and therefore not be included in the regression. I construct year dummies that account for the remaining years the sample covers (1999-2005). Table 2 gives an overview of all the variables discussed in this section and used in the research.

3.3 Regression model

Following Claessens et al. (2001), to examine whether the performance of a bank is affected by foreign ownership and what effect control of corruption has on this relationship in the Southern African Development Community (SADC) from 1998-2005 the dependent financial variable is regressed. To examine hypothesis (1), I estimate the folowing basic equation:

(1) NETINTREVijt = ! + "1FOSijt + "2CCjt + "3Bijt + "4Cjt + "51999 + "62000 + "72001 +

"82002 + "92003 + "102004 + "112005 + #ijt

NETINTREVijt is the performance measure net interest revenues of bank i in country j.

FOSijt is the dummy variable, indicating 1 if bank i in country j is in foreign hands, and 0

otherwise. Bijt is a vector consisting bank specific variables for bank i in country j. These

include total assets, equity, loans and loan loss provisions. Cjt is a vector containing control

variables for a country j. These include GDP growth, GDP per capita and the real interest

rate. CCjt indicates the governance indicator control of corruption, from Kaufmann et al.

(2007). The years 1999-2005 indicate the seven binary year dummy variables. And #ijt is an

error term.

Next to test whether the level of control of corruption affects the relationship between ownership and performance, the estimation will be done including an interaction term. This to provide results for hypothesis (2). The equation will then be:

(20)

The Effect of Foreign Entry on Bank Performance: Case of the SADC countries, 1998-2005 (2) NETINTREVijt = ! + "1FOSijt + "2CCjt + "3Bijt + "4Cjt + "51999 + "62000 + "72001 +

"82002 + "92003 + "102004 + "112005 + "12FOSijtCCjt + #ijt

Where FOSijtCCjt is the interaction variable of the foreign ownership variable and the

governance variable, control of corruption.

4. DATA

This section presents the data sample that I use in the empirical estimation. Sub-section 1 describes the data sources and the way I have constructed the sample. And sub-section 2 presents the descriptive statistics of the variables chosen for the sample.

4.1 Data sources and sample construction

(21)

due to missing data, implying that it is not clear which percentage of a bank is owned by

foreign or domestic investors. I eliminate all incomplete ownership data12. When matching

the foreign ownership data subset with the other banking variables constructed from BankScope, numerous banks are eliminated due to missing information. The final dataset results in a balanced, irregular sample of a 338 observations, corresponding to a number of

banks that ranges between 64 (in 1998) and 87 (in 2005). Table 3 provides an overview of

the number of banks in the final sample in each country of the SADC in 1998 and 2005. It also shows the amount of domestic banks and foreign banks in each year, for each country.

The specific country control variables are obtained from the World Bank Development Indicators 2007 for 1998-2005. The World Bank provides data on GDP per

capita, GDP growth and real interest rate for nearly each year and all countries. All three

macroeconomic variables are measured in US dollars and are inflation adjusted.

And finally, data for control of corruption are obtained from Kaufmann et al. (2007). Kaufmann et al. (2007) have developed six governance indicators reflecting the statistical compilation of responses on the quality of governance given by a large number of enterprise, citizen and expert survey respondents in industrial and developing countries, as reported by a number of survey institutes, think tanks, non-governmental organizations, and international organizations. The aggregate governance indicator control of corruption from Kaufmann et al. (2007) covers 215 countries and the years 1998, 2000, 2002, 2003, 2004 and 2005. For the years 1999 and 2001 I take the average of the years before and after. All values determined by Kaufmann et al. (2007) range on a scale from -2.5 to 2.5.

4.2 Descriptive statistics

This sub-section provides an overview of descriptive statistics of the data sample. The

12_{One of my supervisors pointed out that since foreign banks are probably more transparent than domestic}

(22)

statistics are based on a sample of 338 observations of banks in the SADC countries for the period 1998-2005. Table 4 presents an overview of descriptive statistics of the dependent variable and of the independent variables in the regression.

The mean of equity scaled over total assets in the sample is 13.46 percent. This is fairly high. This can be the result of outliers in the data set. The high outliers mostly incur to domestic banks, which have a higher ratio of equity over total assets, and can explain the somewhat high mean of equity scaled over total assets. However, these outliers are correct, and therefore I decided to keep them in the model. Majnoni et al. (2003) also find a fairly high ratio of 12 percent for their sample in Hungary. And I thus do not find this ratio concerning for this study. In addition, Table 4 shows a high upper bound and lower bound for total assets. This is also the result of outliers. There are a few foreign banks present in the data set with a great deal more total assets than the average domestic banks, and other

foreign banks13.

Furthermore, we observe a high ratio for the foreign ownership dummy. This is 0.61, which means that 61 percent of the banks in the sample are foreign owned. Potentially, this fairly high share can be explained by the composition of the dataset. As discussed in section 4.1, while constructing the complete sample several banks in the SADC countries were excluded due to incomplete information on ownership and/or banking variables. In this process domestic banks were most likely excluded more frequent than foreign banks, because foreign banks are assumed to be more transparent than domestic banks and provide more information.

4.3 Correlation

Strong collinear relationships among explanatory variables in an econometric model will cause multi-collinearity. In this case, the least squares estimator is not defined. The

13_{It must be taken into account that, however, I have chosen to include these outliers, because they are correct}

(23)

estimates of the !’s using the least squares principle cannot be obtained. To detect collinear relationships sample correlation coefficients between pairs of explanatory variables are studied. It is considered that a correlation coefficient between two explanatory variables greater than 0.8 or 0.9 in absolute value indicates a strong linear association and a potentially harmful collinear relationship (Hill et al, 2001). The correlation matrix is presented in Table 5.

The matrix shows that only control of corruption has an excessive correlation with the GDP per capita level in the countries. The coefficient is 0.85. A manner to control for this multi-collinearity is to use country fixed effects. The model will then capture country differences in behaviour. The country control variables (GDP per capita, GDP growth and real interest rate) are excluded from the regression equation and replaced by country dummies to control for country differences. This results in 13 binary country dummy variables, one for each country, except Angola, which is considered the base country and for which no dummy variable is specified. Doing this I include new information in the analysis, but still control for the differences of the 14 countries included in the sample. Replacing the country variables from the first equations by country dummy variables changes the regression equations, and the final sample results into a balanced, irregular

sample of 344 observations. The equations (1) and (2) are rewritten to the following14:

(1) NETINTREVijt = ! + "1FOSijt + "2CCjt + "3Bijt + "4BW + "5CD + "6LS + "7MG +

"8MU + "9MW + "10MZ + "11NA + "12SZ + "13TZ + "14ZA + "15ZM + "16ZW+

"171999 + "182000 + "192001 + "202002 + "212003 + "222004 + "232005 + #ijt

(2) NETINTREVijt = ! + "1FOSijt + "2CCjt + "3Bijt + "4BW + "5CD + "6LS + "7MG +

"8MU + "9MW + "10MZ + "11NA + "12SZ + "13TZ + "14ZA + "15ZM + "16ZW+

"171999 + "182000 + "192001 + "202002 + "212003 + "222004 + "232005 +

14_{The abbreviations of the country dummy variables used in the regression equations are mentioned in the}

(24)

The Effect of Foreign Entry on Bank Performance: Case of the SADC countries, 1998-2005 "24FOSijtCCjt + #ijt

The abbreviations of the countries indicate the thirteen binary country dummy variables. Table 6 provides a new overview of the variables used in the regression analysis.

Table 7 provides the correlation matrix where the country dummy variables are included. This matrix shows no highly correlated variables. However, it should be noted that when studying sample correlation coefficients between pairs of explanatory variables less collinear relationships would be detected than when more variables are combined. In this study though collinear relationships are studied pair wise.

Therefore, Table 7 is used in the regression and the country control variables, GDPPC, GDPGROWTH, IR, are not further included in the regression, since this would wrongly inflate the predictive value of the model. Because the rest of the correlations between the other variables are not very high, I conclude that multicollinearity does not play a significant role in my regression. Thus, I argue that all the independent variables can be included in the OLS regressions.

5. RESULTS

This section aims to provide evidence for the influence of foreign bank ownership on the level of performance in the SADC countries in the period from 1998 to 2005, and for the influence of the level of corruption control in the specific countries on this relationship. Sub-section 1 describes the results from the regression analysis that is performed. In sub-section 2 a discussion of the results is presented.

5.1 Regression analysis

(25)

ownership and control of corruption, to answer hypothesis (2). Table 8 shows the results of this analysis.

As can be seen from Table 8 the main independent variable, foreign ownership is positively related to the level of bank performance in both regressions, and is statistically significant in both regressions at a 1 percent significance level. This result is in line with my expectations and the findings by Claessens et al. (2001), that foreign banks increase bank performance in developing countries.

With regard to the relationship between control of corruption and bank performance, unexpected results are found in both regressions. Firstly, the sign of the coefficient of the governance variable is negative, whereas I expected a higher level of control of corruption to be positively related to performance. This relationship is also significant on a 1 percent confidence level. What's more, the interaction variable of control of corruption and foreign ownership in the second regression provides unsatisfactorily results; the relationship is insignificant, thus no inferences may be drawn upon the interaction variable.

Of the banking control variables used in the regression models, the leverage ratio (equity scaled over total assets), the provisioning measure (loans loss provisions scaled over total assets), and the bank risk proxy (net loans scaled over total assets) show positive relationships with bank performance in both regressions; The first on a 1 percent significance level, and the latter two on a 10 percent significance level.

Because country dummies are used in this regression model, a glance can be taken to the differences per country. As can be seen from Table 8 several countries indicate a significant positive relationship towards net interest revenues on 1 percent significance levels; e.g. Botswana, Lesotho, and South Africa. And in both equation models the Democratic Republic of Congo shows a significant negative relationship.

(26)

means that the null hypotheses (1) and (2), that no relationship exists between the dependent variable and all of the independent variables, can be rejected for these regressions. The R-squared values confirm this. The values are fairly high; the R-squared values are 0.57 in both estimations. This implies that the independent variables chosen in the model explain a modest part of the variation in the dependent variable.

Secondly, I use Ordinary-Least-Squares (OLS) regressions for both equations, and

this model assumes that there is a linear association between the dependent variables and the independent variables. Related studies using similar variables as I use, do not transform their functional forms to back up this assumption. Nevertheless, I analyzed the scatter-plots of the relationships between the various variables (these results are not included in this paper) and I also concluded that it was not necessary to transform my data.

Thirdly, when using OLS an assumption is made that the error term has a constant variance. When the variances for all observations are not the same, it is said that hetreoskrdasticity exists. The existence of different variances, or heteroskedasticity, is often encountered when using cross-sectional data. Heteroskedasticity does not cause OLS coefficients to be biased, but the variance – and thus the standard error – of the coefficients tends to be underestimated, inflating the results from t-tests. This can cause insignificant variables appearing to be statistically significant (Hill et al., 2001). Therefore, to test for the presence of heteroskedasticity I use White’s approximate estimator in the regressions.

(27)

The Effect of Foreign Entry on Bank Performance: Case of the SADC countries, 1998-2005 analysis.

5.2 Discussion of results

The results presented in the previous sub-section show that foreign owned banks located in the Southern African Development Community (SADC) indeed increase bank performance in 1998-2005. I find that the average foreign bank in my sample has a significant positive relationship on a 1 percent confidence level with bank performance, net interest revenues. The average foreign bank has a level of net interest revenues that is 3.43% points higher than that of a domestic bank. This result confirms previous findings that foreign banks tend to be better performing than domestic banks in developing countries (Claessens, Demirguc-Kunt and Huizinga, 2001).

Does this outcome imply that national- and international organizations, such as the World Bank and IMF, should further encourage African countries to open their domestic financial markets? International banking activities have been stimulated the past few decades, and evidence of increasing financial performance may further stimulate international capital flight. Thus, this positive outcome may provide support for policy implications to encourage foreign involvement in the African economies. However, it must be taken into account that the dataset used in this model has it’s limitations. It may fundamentally be missing information on domestic banks that was not included. This can explain the higher portion of foreign banks present in the sample compared to domestic banks. And therefore, influence the outcome as well. Thus, if policy makers are to draw inference on this positive, significant outcome, I suggest further research with a larger and more complete dataset.

(28)

Southern African Development Community. Foreign banks would only be exploiting emerging markets, and not contributing to the development of financial markets in the SADC. However, if improved revenues are spent in the host countries, potential benefits may incur, in the form of e.g. improved financial services. This can also result in more competition on the domestic financial markets, of which domestic consumers would benefit. To answer if improved bank performance has a positive effect on the financial sectors of the SADC, more information and research are required.

(29)

rejected, due to inferences made on too little related literature. More research could be done on this relationship, or other forms to insure governance should be considered. Kaufmann et al. (2007) provide more governance indicators for an institutional framework, and other indicators have been developed as well, e.g. Corruptions Perceptions Index by Transparency International and Indices on the Freedoms of economies. Finally, it is difficult to discuss the strength of the interaction variable, because the measures for foreign ownership and control of corruption differ. Foreign ownership is indicated by a binary dummy variable and control of corruption by a value ranging on a scale from -2.5 to 2.5. This makes the comparison of the separate coefficients complicated, and therefore the interpretation of the interaction variable.

6. CONCLUSION

Throughout this study I find empirical evidence for a proposed relationship between foreign bank ownership and bank performance in developing countries. Related literature confirms this proposition, and I extend the literature by studying a less studied environment, the Southern African Development Community. Moreover, I study the role of corruption in these Southern African countries on the relationship between foreign ownership and bank performance, but I am not able to provide significant results on this relationship.

(30)

the literature, that foreign banks perform better than domestic banks in developing countries. As a note of caution, it is important to mention that these results do not necessarily imply that domestic banks play no role in development. Furthermore, this study finds no significant results on which conclusions can be drawn on the effect of corruption on this relationship. This gives rise to further research.

There are also some limitations to this study that need to be considered. Firstly, I have chosen to study foreign bank ownership and bank performance by using a financial accounting measure for bank performance, net interest revenues. This is in accordance with the related literature. However, other measurements, like rates of return on bank stocks, may provide different results than this study. Secondly, I rely on BankScope for the largest part of my data sample. From BankScope I construct a final data sample of 344 observations in the SADC. BankScope however, is maintained for commercial reasons and one of its main limitations is the almost total omission of rural and very small banks, which could be found particularly in developing economies (Micco et al., 2007). Furthermore, it only consists of data from banks that publish independent financial reports. Thus, it may omit some branches and subsidiaries of foreign banks (Bhattacharya, 2003). And although the greatest precaution has been taken in collecting data, coding banks, and constructing the final data set, possible inadequacies are hard to avoid.

(31)

7. REFERENCES

Aliber, R.Z. (1984). International banking, a survey. Journal of Money, Credit and Banking, 16, 661-712.

Berger, A.N. and Mester, L.J. (1997). Inside the black box: what explains differences in the efficiencies of financial institutions?, Journal of Banking & Finance, 21, 895-947.

Berger, A.N. and Hannan, T.H. (1998). The efficiency cost of market power in the banking industry; a test of the “quiet life” and related hypotheses. Review of Economics and Statistics, 80, 454-465.

Berger, A.N., De Young, R., Genay, H. and Udell, G. (2000). Globalisation of financial institutions: evidence from cross-border banking performance, Brookings-Wharton Papers on Financial Services, 3, 23-158.

Bhattacharya, K. (2003). How good is the BankScope database? A cross-validation exercise with correction factors for market concentration measures. Working Paper no. 133, Bank for International Settlements.

Bonin, J.P., Hasan, I., Wachtel, P. (2005). Bank performance, efficiency and ownership in transition countries, Journal of Banking & Finance, 29, 31-53.

Buch, C.M (1997). Opening up for foreign banks: How Central and Eastern Europe can benefit. Economics of Transition, 5(2), 339-366.

(32)

The Effect of Foreign Entry on Bank Performance: Case of the SADC countries, 1998-2005 Cho, K.R. (1990). Foreign Banking Presence and Banking Market Concentration: The Case of Indonesia, The Journal of Development Studies, 27(1), 98-110.

Claessens, S., Demirgüç-Kunt, A. and Huizinga, H. (2001). How does foreign entry affect domestic banking markets?, Journal of Banking & Finance, 25, 891-911.

Clarke, G., Cull, R., D’Amato, L. and Molinari, A. (2000). The effect of foreign entry on Argentina’s domestic banking sector, in Claessens, S. and Jansen, M. (Eds.), The Internationalization of Financial Services –Issues and Lessons dor Developing Countries, Kluwer Academic Press, London/The Hague/Boston.

Clarke, G., Cull, R., Peria, M.S.M. and Sánchez, S.M. (2003). Foreign bank entry: experience, implications for developing economies, and agenda for further research, The World Bank Research Observer, 18(1), 25-59.

Corruption Perceptions Index (2007). Corruption perceptions Index Regional Highlights: Africa, 1998 and 2005, www.transparency.org.

Demirguc-Kunt, A. and Huizinga, H. (2000). Determinants of commercial bank interest margins and profitability: Some international evidence, World Bank Economic Review, 13(2), 379-408.

(33)

De Young, R. and Nolle, D.E. (1996). Foreign owned banks in the US: Earning market share or buying it?, Journal of Money, Credit, and Banking, 28, 622-636.

The Economist Newspaper and The Economist Group (2008). Getting to grips with graft, The Economist Global Agenda, May 18th 2003.

The Economist Newspaper and The Economist Group (2008). Africa’s unending war on

corruption, The Economist Global Agenda, Feb 3rd 2006.

Figueira, C., Nellis, J., Parker, D. (2005). Does ownership affect the efficiency of African banks?, Cranfield School of Management, 2005.

Focarelli, D. and Pozzolo, A. (2000). The detrminants of cross-border bank shareholdings: an analysis with bank-level data from OECD countries. Federal Reserve Bank of Chicago, Proceedings, 199-232.

Gelb, A. and Sagari, S. (1990). Banking, in Messerlin, P., Sanvant, K. (Eds.), The Uruguay Round: Services in the World Economy. The World Bank and UN Centre on Transnational Corporations, Washington D.C.

Green, J.C., Murinde, V. and Nikolov, I. (2004). Are foreign banks in Central and Eastern European countries more efficient than domestic banks?, Journal of Emerging Market Finance, 3, 175-205.

(34)

Hanson, J.A. and Rocha, R. (1986). High interest rates, spreads, and the cost of intermediation, two studies. Industry and Finance Series 18, World Bank.

Hill, R.C., Griffiths, W.E. and Judge, G.G. (2001). Undergraduate Econometrics 2nd ed.,

United States of America, John Wiley & Sons, Inc.

IMF (International Monetary Fund) (2000). International Capital Markets: Developments, Prospects and Key Policy Issues, Washington D.C.

Kaufmann, D., Kraay, A. and Mastruzzi, M. (2007). Governance Matters VI: Governance Indicators for 1996-2006, World Bank Policy Research September 2006, World Bank, Washington D.C.

Lensink, B.W. and Hermes, N. (2004). The short-term effects of foreign bank entry on domestic bank behaviour: Does economic development matter?, Journal of Banking & Finance, 28, 553-568.

Levine, R. (1996). Foreign banks, financial development, and economic growth, in Cheng, H. (Ed.), International Financial Markets. AEI Press, Washington D.C.

Majnoni, G., Shankar, R. and Várhegyi, E. (2003). The dynamics of foreign bank ownership: Evidence from Hungary, World Bank Policy Research Working Paper no. 3114, World Bank, Washington D.C.

(35)

Mercan, M., Reisman, A., Yolalan, R., Emel, A.B. (2003). The effect of scale and mode of ownership on the financial performance of the Turkish banking sector: results of a DEA-based analysis, Socio-Economic Planning Sciences, 37, 185-202.

Micco, A., Panizza, U., Yanez, M. (2007). Bank ownership and performance. Does politics matter? Journal of Banking & Finance, 31, 219-241.

Naaborg, I.J. (2007). Foreign Bank Entry and Performance with a focus on Central and Eastern Europe, Eburon.

Papi, L. and Revoltella, D. (2000). Foreign direct investment in the banking sector: A transitional economy perspective, in Claessens, S. and Jansen, M. (Eds.), The Internationalization of Financial Services –Issues and Lessons dor Developing Countries, Kluwer Academic Press, London/The Hague/Boston.

Stiglitz, J.E. (1993). The role of the state in financial markets, in Proceedings of the World Bank Annual Conference on Development Economics, 19-52.

Sturm, J.E. and Williams, B. (2004). Foreign bank entry, deregulation and bank efficiency: Lessons from the Australian experience, Journal of Banking & Finance, 28, 1775-1799. Terrier, G. (1994). Economic Trends in Africa: The Economic Performance of Sub-Saharan African Countries, Working Paper no. 94/109, International Monetary Fund.

(36)

Weill, L. (2003). Banking efficiency in transition economies: the role of foreign ownership, The Economics of Transition, 11, 569-592.

www.worldbank.org/data/

www.worldbank.org/wbi/governance/govdata.

Word Bank (2005). World Development Indicators 2005.

(37)

8. TABLES

TABLE 1

Overview of Literature on Foreign Ownership and Bank Performance

Author(s) Setting Main Contributions

De Young and Nolle (1996) Industrialized economies: United States

Suggest that foreign owned U.S. banks are less efficient, perform more poorly than U.S. owned banks, as foreign banks may have focused more on growth than on profitability.

Sturm and Williams (2004) Industrialized economies: Australia

Their results show that foreign banks are more cost efficient than are domestic banks, but that there is no evidence that they are more profitable Berger et al.

(2000)

Industrialized economies: Spain, France, Germany, U.K., U.S.

They find that foreign banks have both lower cost efficiency and lower profit efficiency than domestic banks. Grogorian and manole (2002) 17 Transition economies

Find that foreign ownership with controlling power and enterprise restructuring enhances commercial bank efficiency.

Weill (2003) Transition economies: Czech Republic, Poland

Finds that foreign banks are more efficient than domestic banks

Bonin et al. (2005)

11 Transition economies

They find that banking sectors in these countries became more efficient and more competitive toward the end of the 1990s.

Naaborg (2007)

European Transition economies

Find that foreign ownership is negatively related to bank profitability Majnoni, Shankar and Varhegyi (2003) Transition economies: Hungary

They find that foreign banks are able to achieve a consistently higher profitability and is strictly related to the duration of their presence in Hungary and to the nature of the initial investment. Clarke et al. (2000) Big Emerging Market: Argentina

Find that domestic banks with loan portfolios concentrated in manufacturing tended to have lower net margins and lower profits than other domestic banks. On the other hand they find that banks with greater consumer lending had higher net interest margins and higher profits.

Claessens et al. (2001)

80 Developed and Developing economies

(38)

TABLE 2 Overview of Variables

Variable Description

Dependent Variable

NETINTREV Net interest revenues in thousands US dollars.

Independent Variables

FOS Dummy variable indicating if a bank is foreign

owned (at least 50% of its shares is foreign owned), 1 = yes, 0 = no.

CC Control of corruption, a governance indicator

developed by Kaufmann et al. (2007). The values determined by Kaufmann et al. (2007) range on a scale from -2.5 to 2.5.

TA Total assets in millions US dollars.

EQ Equity scaled over total assets in thousands US

dollars.

LLP Loan loss provisions scaled over total assets in

thousands, in US dollars.

NETLOANS Net loans scaled over total assets in thousands US

dollars.

GDPGROWTH Annual GDP growth in US dollars, in percentages

(%).

GDPPC GDP per capita in US dollars.

IR Real interest rate in US dollars, in percentages (%).

Year Dummies 7 binary dummy variables indicate if an observation

(39)

TABLE 3

Number of banks in the SADC, 1998 and 2005

!"#$% &''(% )**+% ,-./0$1% 2-3"4056%7#/84% 9-$"5:/%7#/84% 2-3"4056%7#/84% 9-$"5:/%7#/84% !"#$%&' (' )' (' *' +$,-.&"&' (' )' )' *' /010'$2'3$"#$' )' (' )' (' 45-$,6$' "0&0' (' "0&0' (' 7&8&#&-9&:' (' *' )' ;' 7&<:=,=<-' (' >' )' ?' 7&%&.=' *' (' (' @' 7$A&BC=D<5' )' *' )' ;'

(40)

(41)

TABLE 5

Correlation Matrixa

a_{N = 338}

Means s.d. 1 2 3 4 5 6 7 8 9 10

1 _{Net Interest Revenues} _6.07 _5.37 _1.00

2 _{Foreign Ownership} _0.61 _0.49 _0.09 _1.00

3 _{Total Assets} _20.27 _82.36 _-0.09 _0.08 _1.00

4 _Equity _13.46 _12.88 _0.30 _-0.18 _-0.11 _1.00

5 _{Loan Loss Provisions} _1.38 _3.57 _0.01 _-0.08 _-0.04 _-0.15 _1.00

6 _{Net Loans} _41.27 _22.63 _0.10 _0.04 _-0.01 _0.29 _-0.03 _1.00

7 _{GDP Growth} _3.97 _4.71 _-0.16 _0.12 _0.12 _0.07 _0.00 _-0.02 _1.00

8 _{GDP per Capita} _1.40 _1.53 _-0.20 _-0.01 _-0.10 _0.26 _-0.11 _0.42 _0.07 _1.00

9 _{Interest Rate} _7.49 _16.02 _-0.20 _0.20 _0.03 _0.03 _-0.09 _0.04 _0.37 _0.12 _1.00

(42)

TABLE 6

New Overview of Variables

Variable Description

Dependent Variable

NETINTREV Net interest revenues in thousands US dollars.

Independent Variables

FOS Dummy variable indicating if a bank is foreign

owned (at least 50% of its shares is foreign owned), 1 = yes, 0 = no.

CC Control of corruption, a governance indicator

developed by Kaufmann et al. (2007). The values determined by Kaufmann et al. (2007) range on a scale from -2.5 to 2.5.

TA Total assets in millions US dollars.

EQ Equity scaled over total assets in thousands US

dollars.

LLP Loan loss provisions scaled over total assets in

thousands, in US dollars.

NETLOANS Net loans scaled over total assets in thousands US

dollars.

Country dummies 13 binary dummy variables indicate if an

observation belongs to a certain country. The countries defined are all SADC countries, except Angola.

Year Dummies 7 binary dummy variables indicate if an observation

(43)

TABLE 7

Correlation Matrix with country dummiesa

(44)

TABLE 8

OLS Regression Estimatesa

(White Heteroskedasticity-Consistent Standard Errors & Covariance)

1 2

Variables ! s.e. ! s.e.

Intercept -12.20*** 3.28 -11.76*** 3.42 Independent Variables Foreign Ownership 3.43*** 0.59 2.95*** 0.52 Control of Corruption -7.93*** 2.19 -7.43*** 2.32 CC*FOS -1.04 0.82 Control Variables Total Assets 0.002 0.002 0.002 0.002 Equity 0.17*** 0.03 0.17*** 0.03 Loan Loss Provisions 0.28* 0.16 0.28* 0.16 Net Loans 0.02* 0.01 0.02* 0.01 Country Dummies Botswana 16.26*** 5.04 16.10*** 5.12 D.R. of Congo -2.39** 1.14 -2.67** 1.19 Lesotho 9.05*** 2.76 8.94*** 2.80 Madagascar 7.25*** 2.72 7.16*** 2.77 Mauritius 7.81* 4.06 8.08** 4.06 Malawi 2.87 2.12 2.69 2.16 Mozambique 4.02** 1.81 3.81** 1.87 Namibia 10.01** 4.20 10.29** 4.21 Swaziland 6.94** 2.87 6.88** 2.92 Tanzania 1.97 1.61 1.81 1.66 South Africa 12.08*** 4.31 11.89*** 4.40 Zambia 6.78*** 1.49 6.62*** 1.53 Zimbabwe 13.45*** 2.28 13.42*** 2.30 Observations 344 344 R-squared 0.57 0.57 Adjusted R2 0.53 0.53 F-statistic 16.05*** 15.56***

a _{Year dummies were included in the regressions, but are not presented in the table.}