- Email: [email protected]
International investment in financial services
Journal of Banking & Finance 25 (2001) 317±337 www.elsevier.com/locate/econbase
International investment in ®nancial services Fariborz Moshirian
*
School of Banking and Finance, The University of New South Wales, Sydney NSW 2052, Australia Received 27 May 1997; accepted 4 October 1999
Abstract This paper analyzes and models the signi®cant components of international trade in ®nancial services, namely, foreign direct investment in banking for the US, the UK and Germany. It distinguishes between banks' activities abroad and FDI in banking by banks and non-banks. A model for FDI in banking is proposed which contains certain explanatory variables peculiar to FDI in banking as compared to FDI in manufacturing. The components of the model of FDI in banking is dierent from those models designed to explain banks activities abroad. The empirical results of this study of FDI in banking indicate that bilateral trade, banks' foreign assets, the cost of capital, relative economic growth, exchange rates and FDI in non-®nance industries are the major determinants of foreign investment in banking. Ó 2001 Elsevier Science B.V. All rights reserved. JEL classi®cation: G15; G24 Keywords: Financial services; International banking; Foreign investment; Banks' foreign assets
1. Introduction According to the UN World Investment Report (1994), world-wide ¯ows of foreign direct investment (FDI) have grown at unprecedented rates, to reach a total out¯ow of $225 billion in 1990 (from an outward stock of $1.7 trillion). *
Tel.: +61-2-9385-5858; fax: +61-2-9385-6347. E-mail address: [email protected] (F. Moshirian).
0378-4266/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 8 - 4 2 6 6 ( 9 9 ) 0 0 1 2 5 - 9
318
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
The average annual growth rate of FDI has been high between 1987 and 1997 and this growth far exceeded that of merchandise exports and nominal GDP. One of the major categories of US FDI abroad is FDI in banking. US FDI in banking has increased ®vefold over the period 1983±1997. Statistical data also show that German FDI in banking has increased over ®vefold during the period 1983±1997, while UK FDI in banking has more than tripled during the same period of time. 1 It is worth noting that for all three countries Europe remains a primary host for foreign banking expansion, whilst Japan remains relatively closed. In fact, apart from the continued growth in the number of foreign banks in Hong Kong and Singapore, Asia still lags behind the developing countries of Latin America as host countries for FDI in banking. This can be attributed to the delayed and incomplete approach many of the Asian nations are taking toward ®nancial deregulation. The growth in FDI in banking has been encouraged by a global trend towards ®nancial deregulation, information and telecommunication advances, and globalisation of the capital market. Indeed, growth in FDI in banking has not only provided new opportunities for banks to expand their international businesses from the home country, but has also allowed them to use FDI in banking to expand their international activities. As more countries deregulated their ®nancial markets during the 1980s allowing foreign banks and non-banks to invest in their banking industry, the study of FDI in banking is more signi®cant than ever before. Despite the continued increase in international investment in banking during the 1980s and 1990s, there have been few major studies of US banks' activities abroad, partly due to the lack of quarterly data on FDI in banking. Researchers have had to use indirect data related to the assets of foreign branches of US banks, or the number of foreign branches of US banks to study FDI in banking. Goldberg and Johnson (1990) who examined those factors which determine the number of foreign branches of US banks and those factors aecting the assets of foreign branches of US banks over the period 1972±1985, remains one of the few and indeed the last major study of US banks' activities abroad. However, the recent unpublished quarterly data on US FDI in banking has provided an opportunity for researchers to measure both bank and non-bank US foreign investment activities in banking for the ®rst time. Thus, this study is the ®rst study
1 FDI in banking of the United Kingdom and Germany are in the following countries: the UK (Germany only), Belgium, Denmark, Germany (for the UK only), France, Greece, Ireland, Italy, Netherlands, Portugal, Spain, Austria, Sweden, Switzerland, Africa, Hong Kong, Brazil, Japan, Australia, New Zealand, South Africa. The United States not only has FDI in the above countries but also in the following countries: Norway, Turkey, Argentina, Chile, Columbia, Mexico, Romania, Nigeria, Israel, India, Indonesia, Malaysia, Philippines, Singapore, South Korea, Taiwan, Thailand. Note that over 70% of the total FDI in banking from these three countries (i.e., the US, the UK and Germany) are to OECD countries.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
319
in US banking which uses data on FDI in banking, in an attempt to measure US banks' activities abroad over the period 1983±1995. Furthermore, this study is also the ®rst study to measure FDI in banking for Germany and the UK for which data on FDI in banking could be obtained. As this study, unlike previous studies on US banks' activities abroad, applies the most sophisticated econometric techniques to measure the determinants of US FDI in banking, its empirical ®ndings will be statistically more reliable for interpretation. The remaining part of the paper is structured as follows: Section 2 discusses trade in ®nancial services and de®nes banks' foreign assets and FDI in banking, Section 3 proposes various factors which determine FDI in banking; Section 4 models FDI in banking; Section 5 describes the sources of data and the methodology used in this paper; Section 6 reports the empirical ®ndings; and Section 7 makes some concluding remarks. 2. Foreign direct investment in banking versus the assets of US banks foreign branches There have been a few empirical studies about the determinants of the number of US banks' branches and oces abroad. Sagari (1986) analysed the determinants of US foreign investment in banking. Goldberg and Johnson (1990) examined those factors which determine the number of foreign branches of US banks and those factors aecting the assets of foreign branches of US banks over the period 1972±1985. 2 These studies, as mentioned by Moshirian and Pham (1999), did not measure US FDI in banking abroad, but rather the assets of US banks foreign branches and the number of branches of US banks abroad. However, there are at least three major categories of investors who invest in banking abroad: 3 (i) Banks that establish their branches, agencies or subsidiaries in a foreign country. 2 There have also been some studies of foreign banks in the US. Amongst these studies are the works of Goldberg and Saunders (1981), Hultman and McGee (1989) and Grosse and Goldberg (1991). Goldberg and Saunders (1981) consider foreign banks' shares of total US commercial bank assets, and the number of oces in the US owned by foreign banks as the major dependent variables. As the purpose of Grosse and Goldberg's study (1991) was to analyze foreign bank activities in the US with respect to countries of origin, they used the assets of the banks of each of the foreign countries operating in the US as the dependent variable. Their study also has a good survey of literature in the area of US banks activities abroad and foreign banks activities in the US. 3 Foreign direct investment (cf. portfolio investment) implies that a person in one country has a lasting interest in, and a degree of in¯uence over the management of a business enterprise in another country. According to the US Commerce Department, US foreign direct investment abroad is ownership or control, directly or indirectly, by a single person (or associated group of persons) from the US of 10% or more of the voting securities of an incorporated foreign business enterprise or an equivalent interest in an unincorporated foreign business enterprise.
320
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
(ii) Banks that do not have any physical oce in the foreign country, but are shareholders of foreign banks (i.e., the name of the acquired banks do not change). While they may own more than 10% of some foreign banks, they do not have their own business in the foreign market. (iii) Non-bank corporations, institutions and individuals who invest in foreign banks, and who own 10 or more percent of these foreign banks. Using the example of US FDI abroad, the number of foreign oces owned by US banks may encompass some of the foreign business de®ned above, however neither the number of oces owned by US banks nor the assets of US banks' foreign branches fully re¯ect the amount of US FDI in banking. Indeed, according to the US Direct Investment Abroad: 1989 Benchmark Survey, bank aliates of non-bank US parents (i.e., category (iii)) accounted for 8% of the total US FDI abroad in banking in 1989. This ®gure for 1994 was 9.6%. In the case of category (ii), while the US Department of Commerce acknowledges the existence of such a speci®c category, it does not distinguish direct acquisitions by banks and partial ownership of foreign banks in its statistics. Table 2 in Appendix A to this paper reports the signi®cance of the geographical breakdown of the assets of US banks foreign branches, the number of foreign branches of US banks as well as US FDI in banking abroad. As can be seen, there is not a strong correlation between these three activities. For instance, while only about 4% of the assets of US banks foreign branches and 23% of the number of foreign branches of US banks are located in Latin America, over 36% of US FDI in banking is made in this region. Similarly, in Europe (excluding member countries of the EU), while less than 1% of the assets of US foreign branches are located here, over 10% of US FDI is occurring in this region. The same discrepancy can be seen in the EU and other regions for which data are available. These ®gures may reinforce the above three categories where not all FDI in banking is made by banks which establish their branches and oces abroad (i.e., category (i), as de®ned above) and hence the assets of US banks foreign branches abroad may not necessarily re¯ect the full magnitude of FDI in banking. Thus, given the above discussion and data reported in Table 2, one can see that data on FDI in banking are quite dierent from data on the assets of foreign branches of banks or that of the number of foreign branches of banks. 3. Factors contributing to FDI in banking A number of researchers have studied the determinants of US FDI in manufacturing abroad and foreign FDI in manufacturing in the US. 4 The
4 For a comprehensive survey of literature on FDI see Agarwal (1980), Grosse (1981), Lizondo (1990) and Caves (1996).
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
321
major determining factors identi®ed in these studies are: the size of the market in the host country, relative economic growth, cost of capital, exchange rates, interest rates and taxes. In addition, Williams (1997) has recently surveyed various established theories of FDI in the context of multinational banking in which he discusses major FDI theories such as the international investment theory, the eclectic theory and the industrial organisation hypothesis. In this study, similar to the study by Moshirian and Pham (1999) who analysed Australia's FDI in banking, hypotheses based mainly on the eclectic theory will be used to estimate FDI in banking. This approach is consistent with the previous studies of foreign banks' activities cited in this study. However, this study also uses a new proxy for the cost of capital and new explanatory variables such as banks' international lending activities and FDI in non-®nance industries which are dierent than those proxies and variables used by previous studies of foreign banks' activities abroad. This represents an enrichment of the traditional literature on the determinants of FDI in banking. Also, the use of superior data to measure US, UK and German investment in foreign banking ensures that the selected explanatory variables chosen to test the determinants of FDI in banking abroad re¯ect those factors aecting both non-bank and bank investors. The following factors, based predominantly on the eclectic theory of FDI have been chosen as describing a model of the determinants of foreign direct investment in banking: (i) banks' foreign assets, (ii) FDI in non-®nance industries, (iii) bilateral trade, (iv) the size of the foreign banking market, (v) the cost of capital dierential, (vi) relative economic growth, and (vii) the exchange rate. The remaining part of this section will discuss the rationale behind including the above factors. 3.1. Banks' foreign assets One of the most likely factors contributing to the expansion of FDI in banking is banksÕ foreign assets. Establishing a branch abroad makes the practice of international lending activities both easier and more eective. While banksÕ foreign assets (i.e., the banks claim against foreigners) can be generated both in domestic and foreign markets, the existence of a foreign branch makes it easier for banks to expand their international lending. Obviously, a foreign presence could also reduce transaction costs (including telecommunication costs, information gathering and processing costs) and further facilitate the expansion of international lending. Banks' foreign assets have not been used by previous researchers as an independent variable for measuring banks' activities abroad. This is because the previous studies, such as Goldberg and Johnson (1990) and Grosse and Goldberg (1991), did not distinguish between the US banks' foreign assets and the US banks' assets in their foreign branches. This distinction is important in this study,
322
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
as unlike previous studies in banks' foreign activities, the dependent variable used in this study is FDI in banking rather than US banks' assets in their foreign branches. 5 Thus, it is hypothesised that the larger a countryÕs banksÕ foreign assets, the greater will be the size of that countryÕs FDI in banking abroad. 3.2. FDI in non-®nance industries It has been argued by researchers such as Grubel (1977) and Gray and Gray (1981) that there is a positive relationship between FDI in manufacturing and the expansion of banks abroad. Researchers such as Nigh et al. (1986) and Goldberg and Johnson (1990) found a positive relationship between the US banksÕ foreign activities (including activities with US citizens abroad and foreign customers) and the size of US FDI. The rationale for this is that multinational banks will follow their multinational manufacturers abroad so that they can provide services for their customersÕ foreign operations. Unlike previous studies on banksÕ activities abroad, this study has access to both FDI in general less FDI in banking and ®nance and data on FDI in banking. Thus, as the dependent variable in this study is FDI in banking, the independent variable used to capture the so called ``gravitational pull eect'' is total FDI of country i less FDI in banking and ®nance. Thus, the hypothesis to be tested is whether FDI in banking is positively related to FDI in country i less FDI in banking and ®nance. 3.3. Bilateral trade Empirical studies such as Roemer (1975) and Agarwal (1980) on the relationship between general FDI and foreign trade indicate a positive relationship 5 Table 3 in Appendix A to this paper reports the size of the US banks' assets in their foreign branches, from 1972±1985 (the period for which Goldberg and Johnson (1990) measured US banks' activities abroad) and was compiled by Houpt (1988). These data sets are compared with the US banks' foreign assets reported by the IMF Balance of Payments Statistics. As can be seen there is a signi®cant discrepancy between these two data sets as they are representing two dierent things. According to the IMF de®nition of banks' foreign assets provided by Landell-Mills (1986), the US banks' foreign assets are de®ned as those assets held against non-US citizens, which can be generated both within the United States economy and outside the United States. However, the US banks have foreign branches in various countries, particularly in countries where US multinational corporations are operating. Some parts of the assets of US banks' branches and aliates abroad are held by US companies. In other words, the US banks' activities abroad are comprised of activities with US individuals and companies abroad as well as with foreign individuals and companies.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
323
between them. Furthermore Jain (1986) found that, in a sample of 46 countries, the USÕ share of total trade was highly correlated with its share of banking assets in these countries. Furthermore, Nigh et al. (1986) found a positive relationship between trade and foreign banksÕ activities. In this study, similar to previous studies of foreign banking activities, one of the hypotheses to be tested is a relationship between FDI in banking and bilateral trade for the three countries studied. In order to calculate the trade variable the following steps have been taken: if we take the US as the example, the relevant variable has been constructed as the weighted average of bilateral exports and imports between the US and her trading partners. The weight given to the sum of exports and imports with a given trading partner is given by that countryÕs share of US FDI in banking. 6 Given past empirical research on the eects of international trade on banking and the expectation that increased foreign trade will encourage trade ®nancing, a positive coecient is expected for the bilateral trade variable. 3.4. Size of the foreign banking market When investors (banks and non-banks) consider whether to invest in foreign banking, the size of the particular foreign banking market is likely to be one of the factors they take into account. This claim is supported by Kravis and Lipsey (1982) and O'Sullivan (1985) who argued that the size of the host country's market is one of the signi®cant determinants of FDI. The size of the foreign banking market is therefore proposed as a location-speci®c determinant of FDI in banking. In this study, similar to the study of Grosse and Goldberg (1991), the size of the foreign banking market is proxied by the weighted sum of two types of deposits held by the banks in country i: (i) time, savings and foreign currency deposits, and (ii) demand deposits, where the weights are given by that country's share of US, UK or German foreign direct investment in banking. The larger the foreign banking market, the greater the number of potential new customers for local banks. In turn, this suggests that there is a greater number of investors (banks and non-banks) willing to invest in foreign markets in order to take advantage of the market's potential. If the above were observed, a positive relationship between US, UK and German FDI in banking and the size of the foreign banking market would be expected.
6
Note that the weights used in this study are calculated from the annual data over 1983±1995 based on the US, the UK and German geographical distributions to each country in which they have FDI in banking. The weights are constant throughout the sample period.
324
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
3.5. Cost of capital dierential One of the possible determinants of FDI in banking is the cost of capital in the host and source countries. The traditional literature on cost of capital, especially in the banking area (see for example, McCauley and Zimmer, 1989) de®nes ``cost of capital'' as ``the minimum rate of return investors will require on their investment''. Furthermore, McCauley and Zimmer (1991) de®ne a banks' cost of capital for a ®nancial product as ``the spread or fee that allows the required regulatory capital to earn the rate of return demanded by the market''. Taxes are another of the most likely variables in¯uencing the level of FDI in banking in a country. The tax regime in use will determine whether or not a country is an attractive location for private investors to establish part of their foreign operations. Thus, the tax variable should be incorporated into the cost of capital so as to yield the after-tax cost of capital. This is in line with most studies of international investment, such as Lunn (1980), which does not explicitly consider the impact of tax on FDI. Even in FDI studies where the primary concern is the impact of taxation (like Hartman, 1984; Shah and Slemrod, 1991), the empirical analyses still focus on the eects of the after-tax rates of return to foreigners. As foreign investors consider the bank cost of capital in their own country as compared to that of foreign countries, the relative bank cost of capital is an important factor in determining whether a foreign investor will be willing to invest in banking abroad, as opposed to banks in her own country. Previous studies on foreign bank activities such as Goldberg and Saunders (1981) and Hultman and McGee (1989) used the price±earnings ratio for bank stocks in the US (annual average) as an indicator of principal purchases of banks' stock. However, as Poterba (1991) discussed several theoretical and empirical diculties in using the price±earnings ratio as a measure of the cost of equity, particularly across countries, in this study the cost of capital proposed by McCauley and Zimmer (1991) will be used as one of the hypotheses contributing to FDI in banking abroad. McCauley and Zimmer (1991) have shown that once the bank managers have calculated the cost of equity per bank, they should work out the spread or fee needed to be charged on individual ®nancial products to cover their equity costs. In other words, assuming that the spread is set in such a way to satisfy the required equity cost, one may use the spread (measured as lending rate minus deposit rate corrected for corporate tax) as a proxy for the cost of equity for banks. Assuming that the spread is set in such a way that it satis®es the required equity cost, one may use the spread (measured as lending rate minus deposit rate corrected for corporate tax) as a proxy for the cost of capital for banks. Thus, the larger the spread required by banks in order to cover their cost of
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
325
capital, the less competitive is their position in the international banking market. In this study, the spread between lending and deposit rates corrected for corporate tax rates of the US, the UK and Germany vis-a-vis the ``rest-ofthe-world'', will be used as a proxy for the cost of capital dierential of banks. 7 3.6. Relative economic growth Some researches have argued that one of the factors contributing to the expansion of foreign banks activities abroad is strong economic activity in the host countries. For instance, Hultman and McGee (1989) found a positive relationship between foreign banksÕ activities in the US and economic growth in the US. Furthermore, Goldberg and Saunders (1981), Nigh et al. (1986) and Sabi (1988) found that relative economic growth contributes to the expansion of foreign banking investment. In this study, similar to the above studies, a relative economic growth variable is calculated as a ratio of the growth in GDP of the US, the UK or Germany over the weighted average growth in GDP of those countries in which these three countries invest in banking. A positive relationship indicates that an increase in relative economic growth expands FDI in banking abroad more than it does domestic investment in banking. However, a negative relationship implies that an increase in relative economic growth makes investment in domestic banking more likely than FDI in banking abroad.
7 Note that when the bank cost of capital was calculated for the period after 1992, the calculations were slightly dierent than the spread minus corporate tax used in the previous quarters due to the so-called Basle agreement. The equation used for the post 1992 calculations is based on the spread, in which the overall requirement of 4% equity (the Basle Agreement risk weights, where risk weight for a corporate loan is 100%) is taken into account. As McCauley and Zimmer (1991) argued, banks are funding 4% of the loan with shareholder equity. The payment to equity is for the spread on the loan plus the real (net of in¯ation) after-tax return earned by investing the shareholder equity in a riskless asset. Thus instead of using S 1 ÿ tc1 COE RW 0:04, the following equation is used to calculate the spread for the years post 1992 when the Basle agreement was in place:
0:96S 1 ÿ tct 0:04f1 rt 1 ÿ tct g= 1 ps ÿ 1 COE RW 0:04; where S is the spread between lending and deposit rates, tct the marginal corporate income tax rate at time t, rt the riskless nominal interest rate at time t, ps the in¯ation rate at time t, COE the cost of equity, and RW is the risk weighting. As can be seen from the above two equations, the dierence between the cost of capital calculated prior to 1992 and that calculated post 1992 is very small.
326
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
3.7. Exchange rate Given that foreign private investors' operations may involve substantial ¯ows of diverse foreign currencies, exchange rates are expected to have an impact on their FDI decisions in banking. One possible explanation for a positive correlation between the exchange rate and FDI in banking abroad may be found in the valuation by investors of their foreign banking assets in the host countries' currencies. Hultman and McGee (1989) argued that given a standard accounting practice is to adjust the book value of nondomestic assets to re¯ect changes in the exchange rate, foreign investors may expect that any investment they make in the host countries will appreciate as the local currencies appreciate. On the other hand, it could be argued that such behavior is more likely for short term liquid assets. The book value of long term assets should be less prone to such ¯uctuations. Extending this argument, one would expect a negative correlation between the value of the host countries' currencies and FDI in banking by the US, the UK and Germany. When the host countries' currencies appreciate with respect to foreign investorsÕ currencies, FDI in the US is expected to decrease as it becomes more expensive for foreign investors to invest in those countries, and vice versa. In other words, a depreciated domestic currency will give foreigners an edge in acquiring the control of domestic productive assets. Such a negative correlation is reported by authors such as Goldberg and Saunders (1981), Cushman (1988) and Froot and Stein (1991). Note also that, as the host countries' currencies depreciate, private investors may reduce their repatriated income and increase their reinvestment in the host countries, as they may want to avoid exchange rate losses. Thus, the hypothesis to be tested is whether ¯uctuations in the value of the host countries' currencies aect the level of FDI in banking in those countries. 8 A few researchers such as Goldberg and Johnson (1990) have considered risk as one of the factors aecting FDI. At the same time, the recent study by Santomero and Babbel (1997) indicates some of the shortcomings of using proxies to measure ®nancial risk. As mentioned in footnote 1, over 96% and 90% of UK and German FDI in banking, respectively, are in major developed countries and in the case of the US this ®gure is 80%. Thus, given that the US, the UK and Germany have such a high proportion of FDI in banking in major developed countries, country risk would not be a major factor for investment
8
There are several indices available to account for exchange rates. The most appropriate one is the IMFÕs Multilateral Exchange Rate Model (MERM) index, which is superior to the other indices available for the purposes of this study. The way the index is constructed implies that an increase in the US$Õs MERM index indicates an appreciation of the US$ with respect to a weighted basket of the other currencies.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
327
and hence the comparative risk for investors investing in the form of FDI in banking will not be considered in this study. 4. A model of foreign direct investment in banking Based on the previous discussion, the proposed model for US, UK and German (country i) FDI in banking is as follows: Xi f BAi ; FDi ; BTi ; CDi ; CCi ; EGi ; EXi :
1
Xi stands for the stock of country iÕs FDI in banking. BAi is the banks' foreign assets of country i. FDi stands for the stock of country iÕs FDI in non-®nancial industries, BTi is the weighted average of country iÕs bilateral exports and imports with her major trading partners. The weight given to each trading partner is based on country iÕs total FDI in banking in that country. CDi stands for the weighted average domestic commercial bank deposits for those countries facing country i's FDI in banking. CCi is the cost of the equity dierential for banks of country i and those countries facing country i's FDI in banking. The weight given to each country is based on that country's share of FDI in banking from country i. EGi stands for the relative economic growth of country i. The economic growth of countries facing country i's FDI in banking is calculated as a weighted average of these countries' economic growth. The weight given to each country is based on country iÕs FDI in banking in that country. EXi stands for country iÕs MERM index (a measure of the value of currency i). The period covered is from 1983:1 to 1995:4, the only period for which data are available. All variables are expressed in current US$. In summary, the model for country iÕs FDI in banking can be expressed as lnXi a0 a1 lnBAi a2 lnFDi a3 lnBTi a4 lnCDi a5 CCi a6 EGi a7 lnEXi
2
with the following expected signs: a1 > 0; a2 > 0; a3 > 0; a4 > 0; a5 < 0; a6 ÿ ?; a7 ÿ ?:
5. Data and methodology The model employed uses time-series quarterly data for the period 1983:1 to 1995:4. Eq. (2) was estimated for the US, the UK and Germany using the data base IMF International Financial Statistics, various years, from which banks' foreign assets, exchange rates, bankÕs lending and deposit rates, national in-
328
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
comes, and banksÕ domestic deposits were obtained for all relevant countries. Tax rates are from publications of the Brookings Institution (edited by Pechman, 1988), the National Bureau of Economic Research (edited by Razin and Slemrod, 1990), and the OECD Observer Supplement (1996). US, UK and German bilateral exports and imports were taken from the IMF Directions of Trade, various issues. Data on foreign direct investment in banking and non-banking were obtained from the International Division of the US department of Commerce, the Bank of England Central Statistics Oce, and the Deutsche Bundesbank Statistics Oce (Note that Germany's annual data on various FDI have been used to construct their quarterly data.) Disaggregated data on foreign direct investment in banking have been obtained, for the US, from the ``Survey of Current Business'', for the UK, from the ``Bank of England Statistical Abstract'', and for Germany, from the ``Deutsche Bundesbank Zahlungsbilanzstatistics''. Data on general FDI for the US come from the US Department of Commerce ``Survey of Current Business'', for the UK, from the Central Statistics Oce ``Economic Trends'', and for Germany, from the ``Deutsche Bundesbank Zahlungsbilanzstatistics''. The Survey of Current Business has published the total FDI abroad for the US, and total FDI in the US at the market value and the current cost for the last few years. For instance, the current value of total FDI in the US for 1995 comprises total FDI in the US in 1994 corrected for capital in¯ows, ``price changes'', ``exchange rate changes'' and ``other changes'' 9 in 1995. However, unlike the UK and Germany, detailed estimates of FDI in the US by industry and by country are on a historical-cost basis. As Bargas and Lowe (1994) indicate ``estimates on a historical-cost basis largely re¯ect prices at the time of investment rather than prices at the current or any other period. Historical cost is the basis used for the evaluation of company accounting records in the United States and is the only basis on which companies can report data in the direct investment surveys conducted by the Bureau of Economic Analysis''. 10 The Survey of Current Business has also published annual data on FDI in banking in the US since 1980. These data which consist both of book values of
9 ``Other changes'' re¯ects changes in the value of the ocial gold stock due to ¯uctuations in the market price of gold. However, as this item is very small compared to price and exchange rate changes, this study will not take it into account when converting FDI in banking from book value to nominal or constant value. 10 Helkie and Stekler (1988) were two of the ®rst researchers to argue that the FDI book values of US companies abroad and foreign companies in the US are not good economic measures of the stock of FDI on the ground, and that these values are expressed neither in current nor in constant dollars, but rather they are historical stock values and hence should be corrected for exchange rate variations and in¯ation. This study will follow their approach for the correction of the book values of FDI in banking.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
329
FDI and capital in¯ows of FDI in banking are on a historical-cost basis. Indeed, many studies in FDI used either the book value of FDI or the dierence between consecutive quarters in FDI as a measure of capital ¯ow (see, for instance, Cushman, 1988). In this study, the data on the stock of FDI in banking in the US is converted from being on a historical-cost basis to being in current values. This ensures that both dependent and independent variables are compatible with each other. For instance, FDI in banking in the US for 1985:1 is constructed on the basis of the book value of FDI in banking in 1984:4 which has been corrected for the exchange rate variations and in¯ation of 1985:1. The result is then added to quarterly capital in¯ows in banking in the quarter 1985:1. 6. Empirical results The ®rst step in estimating Eq. (2) was to establish whether all variables are stationary. For this purpose, the Augmented Dicky±Fuller (ADF) and Phillips±Perron unit root tests were employed. The results of these tests showed that most of the variables for the US, the UK and Germany are non-stationary and that some of these variables have one unit root. This suggests that the corresponding ¯ow variables will be stationary. The Augmented Dicky±Fuller unit root tests applied on the ¯ow variables suggest that all variables are stationary, as can be seen from Table 4 in Appendix A to this paper. Thus, the ¯ow variables in Eq. (2) were used for estimation purposes for the US, the UK and Germany. Furthermore, as the ¯ow data used in Eq. (2) had random variations, a moving average with four lags was used to smooth out the volatility of the quarterly data. In addition, Table 5 in Appendix A to this paper reports the correlation matrix between all the variables in Eq. (2) using the ¯ow data. As can be seen, the correlation matrix between all variables is low. As can be seen from previous studies of FDI, almost all these studies used OLS regression as a means of statistical analysis. However, the empirical results of past studies such as Goldberg and Johnson (1990) and Grosse and Goldberg (1991) have been subject to statistical problems such as serial correlation, heteroskedasticity and simultaneity bias and hence their results should have been treated with caution. In order to avoid these statistical problems, the Generalised Methods of Moments (GMM) is used in this study. The empirical results show that US, UK and German banksÕ foreign assets (BAi ) are positively related to FDI in banking. This result indicates that, as the international lending activities of banks increase, there is an accompanying increase in their FDI in banking. This result is consistent with the ®ndings of Moshirian and Pham (1999) for Australia. Thus, while one can argue that US, UK and German banks' international lending can account for the expansion of these three countries' FDI in banking, the coecient of this variable is larger
330
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
for Germany than the US and the UK indicating that this variable has greater in¯uence on FDI in banking in Germany than in the two other countries. The positive sign between FDI in banking and FDI in the non-®nance industries (FDi ) indicate that FDI in banking and FDI in other industries are complementary. This empirical result is consistent with the ®ndings of Hultman and McGee (1989) and Goldberg and Johnson (1990) who found a positive relationship between foreign banks activities in the US and the total of FDI minus FDI in ®nance, insurance and real estate, as an explanatory variable. The t-test for the statistical dierences between the coecients indicated that the German coecient is greater than that of the UK implying that FDI in manufacturing has more impacting than on that of the UK. Bilateral trade (BTi ) variable is found to be positively related to FDI in banking. This result con®rms the theoretical argument that trade is one of the contributors of FDI in banking. This result is also consistent with the ®ndings of Nigh et al. (1986) and Goldberg and Johnson (1990). The test for the statistical dierences between each pair of coecients demonstrates a dierence between the coecients of the US and Germany, indicating that the trade variable has a stronger impact on German FDI in banking. The banks' commercial deposit variable (CDi ) is statistically signi®cant with a positive sign for the US, the UK and Germany. This result is consistent with the ®ndings of Grosse and Goldberg (1991) that foreign banks invest in countries whose domestic banking market is large. Given this result, one could assume that the host countries' governments have every incentive to increase domestic savings and deregulate the domestic ®nancial market in an eort to attract more foreign banks. As can be seen from Table 1, FDI in banking abroad is negatively related to the cost of equity dierential (CCi ) for banks between the US, the UK and Germany and their partner countries. This result implies that an eective and competitive cost of capital structure by banks contributes to their success when entering into the foreign banking markets. It is conceivable that banks from the US, the UK and Germany have access to more diverse markets for capital or a better credit rating than the banks from the host countries. The size of the coecient of this variable for the US is much greater than that of the other two countries. This indicates that the cost of capital dierential has more impact on the amount of the FDI in banking by the US than the UK and Germany. The relative economic growth variable (EGi ) is statistically signi®cant with a negative sign for the US, the UK, Germany. The empirical result shows that as economic growth in these three countries exceed that of their host countries, FDI in banking to these host countries declines. In other words, stronger economic activities in the source countries increase demand for domestic credit and hence more domestic investment in banking. The size of the coecient of this variable indicates that German FDI in banking is more aected by relative economic growth than that of the US or the UK.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
331
Table 1 The empirical results of the foreign direct investment in banking model (1983:1±1995:4)a US a BAi FDi BTi CDi CCi EGi EXi R2 DW JBb WHETc
UK
)0.37 (2.1) 0.04 (11.6) 0.3 (2.1) 0.40 (3.0) 0.008 (4.6) )13.71 (2.8) )1.67 (7.0) )40.5 (2.3) 0.67 1.9 0.80 28.6
Germany
)0.97 ()16.5) 0.013 (20.2) 0.03 (8.0) 0.56 (7.9) 0.005 (4.9) )0.49 ()11.2) )1.05 ()7.5) )81.9 ()4.9) 0.76 1.9 1.8 42.3
)0.44 ()7.4) 0.55 (3.9) 0.19 (5.2) 0.88 (5.4) 0.01 (3.4) )0.41 ()6.3) )21.3 ()16.6) )392 ()5.7) 0.29 1.98 3.4 34.7
a and indicate that an estimate is signi®cantly dierent from zero at the 5% level and 1% level, respectively. Student t-statistics are shown in parentheses. GMM: generalised methods of moments. b Jacque±Bera test statistic for normality; p value in parentheses. c White test for heteroscedasticity.
The empirical results also show that the exchange rate variable for the US, the UK and Germany (EXi ) has a negative relationship across these countries with respect to FDI in banking. One could argue that the higher the value of the currencies of these three countries relative to the currencies of their host countries, the cheaper it becomes for investors from these three countries to invest in foreign banking. The size of the coecient of this variable indicates that UK investors are more responsive in their investment abroad in the wake of depreciation of the host countries currencies, than the other two countries studied. 7. Conclusion This paper has distinguished between banks' activities abroad and FDI in banking by banks and non-banks. It has argued that the number of oces owned by banks abroad may correlate to the amount of FDI in banking, however, it does not include investment by banks who invest in incorporated foreign banks. Furthermore, there are non-bank investors who own over 10% of some banks in foreign countries, and yet neither the number of oces owned by banks nor the total assets of banks' branches in foreign countries fully re¯ect the amount of FDI in banking. A model for FDI in banking has been proposed which comprises certain explanatory variables peculiar to FDI in banking as compared to FDI in general and/or FDI in manufacturing for the US, the UK and Germany. The components of the model of FDI in banking is dierent from those models designed to explain
332
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
banks activities abroad. The empirical results of this study of FDI in banking indicate that banks' foreign assets contribute to the expansion of FDI in banking by both banks and non-bank investors. This result indicates that while banks could be engaged in international lending activities from their home countries, access to foreign markets is a very eective way of expanding their international activities. The empirical results also show a close link between FDI in banking and FDI in non-®nance sector as well as between bi-lateral trade. The empirical results support the theoretical argument that those countries with better cost of capital structures, which have stronger ®nancial positions when competing with the host countries' banks. Furthermore, this paper has shown that if the economic growth in the US, the UK and Germany increases faster than that of their host countries, investors of these three countries tend to invest more in their own countries rather than in the banking sector of the foreign countries. The exchange rate appears to be one of the decisive factors in expanding FDI in banking. As currencies appreciate, there is less incentive for investment in the foreign banking market. This result implies that both bank and nonbank investors have a long term view of investment in banking and hence they are not motivated for short term pro®ts. Given some of the above factors which can be generalized as those factors that the US, German and British private investors consider as universally important factors in determining their amount of these three countries' FDI in banking, one would expect that, with better foreign market access and ``right of establishment'' for trade in ®nancial services in the ``Post Uruguay Era'' and successful bilateral trade agreements between the US and Japan, private investors (banks and non-banks) may increase their investment in some nonOECD countries where there is strong economic growth, a high level of FDI in manufacturing, low cost of capital and rapid expansion of the domestic banking market.
Acknowledgements The author would like to thank the two anonymous referees for their valuable comments and suggestions. The assistance of Alexander Van der Laan is also gratefully appreciated. However, all errors are mine.
Appendix A See Tables 2±5.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
333
Table 2 Comparison of geographical distribution of the assets of US banks foreign branches, the number of foreign branches of US banks and US FDI in banking in percentagesa Region
1979
1980
1981
1982
1983
1984
1985
European Union
Ab Bc Fd
48 19.3 35
48.5 19 35
47.1 19 30.4
45.9 18.2 39.8
43 17.3 28.5
44.9 17.5 28
46 18.6 31.6
Other Europe
A B F
1.11 0.02 10.2
Japan
A B F
n.a. n.a. n.a.
n.a. n.a. n.a.
n.a. n.a. n.a.
5.1 )0.3 1.6
Other Asia
A B F
12 15.8 16
7.9 17.8 9
8.9 17.8 10.2
10.3 19.5 12.4
11.3 0.2 12
12.4 20 11.3
12.6 20 8.5
Latin America
A B F
4.3 20 34
4.3 20.4 34.1
4.5 21.5 37.3
4.3 22.5 33
3.3 22.8 34.4
3.3 23.8 37.7
3.4 23.7 3.4
Africa
A B F
a
0.004 0.02 1.3
1 0.024 9.7
0.002 0.02 1.7
1.2 0.0221 8.2
0.002 0.02 1.8
0.008 0.02 8
0.002 0.02 1.6
0.009 0.021 10.8
0.059 0.025 9.5
0.007 0.024 11.3
5.5 0.03 1.4
5.6 0.03 1.3
5.4 0.03 1.2
0.002 0.025 1.6
Sources: The Survey of Current Business, various issues and Houpt (1988). Percentage of the assets of US banks foreign branches. c Percentage of the number of foreign branches of US banks. d Percentage of US FDI in banking abroad. b
0.002 0.02 1.6
0.003 0.02 0.15
a
77.4 118.0 18.0 23.9
140.5 42.5
1974 162.7 54.7
1975 193.8 72.6
1976 227.9 88.0
1977 257.6 130.0
1978 312.9 156.57
1979
Source: Houpt (1988) and The IMF, Balance of Payment Statistics Yearbook, various annual issues.
The Assets US banks foreign branches The US banks' foreign assets
1972 1973 343.5 203.9
1980
391.0 292.8
1981
Table 3 Comparison between the assets of US banks foreign branches and US banks' foreign assets (in billions of dollars)a 388.5 401.3
1982
386.1 443.1
1983
337.4 443.3
1984
392.2 446.7
1985
334 F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
335
Table 4 Augmented Dicky±Fuller statistics for the ¯ow model of Eq. (2) s^
s^T
s^u
Variables
US
UK
Germany US
UK
Germany US
UK
Germany
Dx DBA DFD DBT DCD DCC DEG DEX
)4.6 )4.0 )6.3 )2.3 )3.7 )5.3 )3.6 )4.2
)4.6 )4.1 )9.6 )5.6 )4.1 )5.4 )4.5 )4.5
)4.6 )3.4 )9.6 )4.8 )4.7 )5.5 )3.9 )4.9
)6.9 )4.5 )9.5 )5.8 )3.9 )5.5 )4.9 )3.4
)6.9 )4.4 )9.5 )4.7 )4.8 )5.5 )5.1 )5.6
)7.1 )4.7 )9.4 )5.7 )4.1 )5.4 )4.0 )3.4
)7.1 )4.8 )9.4 )4.8 )4.9 )5.5 )5.1 )5.5
)6.9 )4.4 )6.2 )2.9 )3.9 )5.5 )5.1 )5.8
)7.1 )5.0 )6.1 )2.9 )4.4 )5.2 )5.0 )6.0
Table 5 The correlation matrix for the US/Germany/UK using data in Eq. (2) X
BA
FD
BT
CD
CC
1 0.17 0.37 0.16 0.52 0.03 )0.14 )0.53
1 0.05 )0.22 )0.21 )0.08 )0.03 )0.15
0.1 0.01 )0.35 )0.02 )0.09 )0.28
1 )0.19 )0.17 0.22 )0.08
1 )0.05 )0.09 )0.75
1 )0.22 0.05
1 0.12
1
For Germany X 1 BA 0.40 FD 0.44 BT 0.17 CD 0.49 CC 0.02 EG 0.09 EX 0.35
1 )0.50 0.29 0.48 )0.2 0.02 0.42
1 0.33 0.72 0.06 0.08 0.56
1 0.36 )0.08 )0.07 0.56
1 )0.04 0.32 0.35
1 )0.13 )0.00
1 )0.09
1
For UK X BA FD BT CD CC EG EX
1 )0.14 )0.26 0.13 0.15 )0.44 0.28
1 )0.05 )0.02 )0.28 )0.32 )0.09
1 )0.02 )0.10 )0.02 0.11
1 0.20 )0.35 )0.02
1 0.02 )0.15
1 0.39
1
For US X BA FD BT CD CC EG EX
1 0.33 )0.11 0.14 )0.02 0.03 0.51 0.43
EG
EX
336
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
References Agarwal, J., 1980. Determinants of foreign direct investment: A survey. Weltwirtschaftliches Archiv 116, 739±773. Bargas, S., Lowe, J., 1994. Direct investment positions on a historical basis. Survey of Current Business 62, 72±85. Caves, R., 1996. Economic Analysis and Multinational Enterprise, second ed. Cambridge University Press, Cambridge. Cushman, D., 1988. Exchange-rate uncertainty and foreign direct investment in the US. Weltwirtschaftliches Archiv 124 (2), 322±336. Froot, K., Stein, J., 1991. Exchange rates and foreign direct investment: An imperfect capital markets approach. The Quarterly Journal of Economics 106 (4), 1191±1217. Goldberg, L., Johnson, D., 1990. The determinants of US banking activity abroad. Journal of International Money and Finance 9, 123±137. Goldberg, L.G., Saunders, A., 1981. The determinants of foreign banking activity in the US. Journal of Banking and Finance 5, 17±32. Gray, J., Gray, P., 1981. The multinational bank: A ®nancial MNC? Journal of Banking and Finance 5, 33±63. Grosse, R., 1981. The theory of foreign direct investment. University of South Carolina Essays in International Business 3, 1±15. Grosse, R., Goldberg, L.G., 1991. Foreign bank activity in the US: An analysis by country of origin. Journal of Banking and Finance 15, 1093±1112. Grubel, H., 1977. A theory of multinational banking. Banca Nazional del Lavoro Quarterly Review 30, 349±364. Hartman, D., 1984. Tax policy and foreign direct investment in the US. National Tax Journal 37, 475±488. Helkie, W., Stekler, L., 1988. Modelling investment income and other services in the US international transactions accounts. International Finance Discussion Papers 319. Houpt, J., 1988. International Trends for US Banks and Banking Markets. Washington, DC. Hultman, C.W., McGee, L.R., 1989. Factors aecting the foreign banking presence in the US. Journal of Banking and Finance 13, 383±396. Jain, A.K., 1986. International lending patterns of US commercial banks. Journal of International Business Studies Fall 73±88. Kravis, I.B., Lipsey, R.E., 1982. The location of overseas production and production for export by US multinational ®rms. Journal of International Economics 12, 201±223. Landell-Mills, J., 1986. The Fund's International Banking Statistics. IMF, Washington, DC. Lizondo, S.J., 1990. Foreign direct investment. International Monetary Fund Working Paper: WP/ 90/63, International Monetary Fund, Washington, DC. Lunn, J., 1980. Determinants of US direct investment in the EEC: Further evidence. European Economic Review 24, 93±101. McCauley, R.N., Zimmer, S.A., 1989. Bank cost of capital and international competition. Federal Reserve Bank of New York Quarterly Review 15, 33±59. McCauley, R.N., Zimmer, S.A., 1991. Cost of capital for industry and banks. Business Economics 26, 14±18. Moshirian, F., Pham, T., 1999. Cost of capital and Australia's banking investment abroad. Applied Financial Economics 9, 295±303. Nigh, D., Cho, R., Krishnan, S., 1986. The role of location-related factors in US banking investment abroad: An empirical examination. Journal of International Business Studies, Fall, 59±72. O'Sullivan, P., 1985. Determinants and impact of private foreign direct investment in host countries. Management International Review 25, 28±35.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
337
Poterba, J., 1991. Comparing the cost of capital in the US and Japan: A survey of methods. Federal Reserve Bank of New York Quarterly Review 15, 20±32. Roemer, J.E., 1975. US-Japanese competition in international markets: A study of the trade± investment cycle in modern capitalism. University of California, Institute of International Studies, Research Series 22, Berkeley. Sabi, M., 1988. An application of the theory of foreign direct investment to multinational banking in LDCs. Journal of International Business Studies 19, 433±447. Sagari, S., 1986. The Financial Services Industry: An International Perspective (unpublished doctoral dissertation). New York University. Santamero, A., Babbel, D., 1997. Financial risk management by insurers: An analysis of the process. Journal of Risk and Insurance 64, 231±270. Shah, A., Slemrod, J., 1991. Do taxes matter for foreign direct investment? The World Bank Economic Review 5, 473±491. UN World Investment Report, 1994. United Nations, New York. Williams, B., 1997. Positive theories of multinational banking: Eclectic theory versus international theory. Journal of Economic Survey 11, 71±100.
International investment in ®nancial services Fariborz Moshirian
*
School of Banking and Finance, The University of New South Wales, Sydney NSW 2052, Australia Received 27 May 1997; accepted 4 October 1999
Abstract This paper analyzes and models the signi®cant components of international trade in ®nancial services, namely, foreign direct investment in banking for the US, the UK and Germany. It distinguishes between banks' activities abroad and FDI in banking by banks and non-banks. A model for FDI in banking is proposed which contains certain explanatory variables peculiar to FDI in banking as compared to FDI in manufacturing. The components of the model of FDI in banking is dierent from those models designed to explain banks activities abroad. The empirical results of this study of FDI in banking indicate that bilateral trade, banks' foreign assets, the cost of capital, relative economic growth, exchange rates and FDI in non-®nance industries are the major determinants of foreign investment in banking. Ó 2001 Elsevier Science B.V. All rights reserved. JEL classi®cation: G15; G24 Keywords: Financial services; International banking; Foreign investment; Banks' foreign assets
1. Introduction According to the UN World Investment Report (1994), world-wide ¯ows of foreign direct investment (FDI) have grown at unprecedented rates, to reach a total out¯ow of $225 billion in 1990 (from an outward stock of $1.7 trillion). *
Tel.: +61-2-9385-5858; fax: +61-2-9385-6347. E-mail address: [email protected] (F. Moshirian).
0378-4266/01/$ - see front matter Ó 2001 Elsevier Science B.V. All rights reserved. PII: S 0 3 7 8 - 4 2 6 6 ( 9 9 ) 0 0 1 2 5 - 9
318
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
The average annual growth rate of FDI has been high between 1987 and 1997 and this growth far exceeded that of merchandise exports and nominal GDP. One of the major categories of US FDI abroad is FDI in banking. US FDI in banking has increased ®vefold over the period 1983±1997. Statistical data also show that German FDI in banking has increased over ®vefold during the period 1983±1997, while UK FDI in banking has more than tripled during the same period of time. 1 It is worth noting that for all three countries Europe remains a primary host for foreign banking expansion, whilst Japan remains relatively closed. In fact, apart from the continued growth in the number of foreign banks in Hong Kong and Singapore, Asia still lags behind the developing countries of Latin America as host countries for FDI in banking. This can be attributed to the delayed and incomplete approach many of the Asian nations are taking toward ®nancial deregulation. The growth in FDI in banking has been encouraged by a global trend towards ®nancial deregulation, information and telecommunication advances, and globalisation of the capital market. Indeed, growth in FDI in banking has not only provided new opportunities for banks to expand their international businesses from the home country, but has also allowed them to use FDI in banking to expand their international activities. As more countries deregulated their ®nancial markets during the 1980s allowing foreign banks and non-banks to invest in their banking industry, the study of FDI in banking is more signi®cant than ever before. Despite the continued increase in international investment in banking during the 1980s and 1990s, there have been few major studies of US banks' activities abroad, partly due to the lack of quarterly data on FDI in banking. Researchers have had to use indirect data related to the assets of foreign branches of US banks, or the number of foreign branches of US banks to study FDI in banking. Goldberg and Johnson (1990) who examined those factors which determine the number of foreign branches of US banks and those factors aecting the assets of foreign branches of US banks over the period 1972±1985, remains one of the few and indeed the last major study of US banks' activities abroad. However, the recent unpublished quarterly data on US FDI in banking has provided an opportunity for researchers to measure both bank and non-bank US foreign investment activities in banking for the ®rst time. Thus, this study is the ®rst study
1 FDI in banking of the United Kingdom and Germany are in the following countries: the UK (Germany only), Belgium, Denmark, Germany (for the UK only), France, Greece, Ireland, Italy, Netherlands, Portugal, Spain, Austria, Sweden, Switzerland, Africa, Hong Kong, Brazil, Japan, Australia, New Zealand, South Africa. The United States not only has FDI in the above countries but also in the following countries: Norway, Turkey, Argentina, Chile, Columbia, Mexico, Romania, Nigeria, Israel, India, Indonesia, Malaysia, Philippines, Singapore, South Korea, Taiwan, Thailand. Note that over 70% of the total FDI in banking from these three countries (i.e., the US, the UK and Germany) are to OECD countries.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
319
in US banking which uses data on FDI in banking, in an attempt to measure US banks' activities abroad over the period 1983±1995. Furthermore, this study is also the ®rst study to measure FDI in banking for Germany and the UK for which data on FDI in banking could be obtained. As this study, unlike previous studies on US banks' activities abroad, applies the most sophisticated econometric techniques to measure the determinants of US FDI in banking, its empirical ®ndings will be statistically more reliable for interpretation. The remaining part of the paper is structured as follows: Section 2 discusses trade in ®nancial services and de®nes banks' foreign assets and FDI in banking, Section 3 proposes various factors which determine FDI in banking; Section 4 models FDI in banking; Section 5 describes the sources of data and the methodology used in this paper; Section 6 reports the empirical ®ndings; and Section 7 makes some concluding remarks. 2. Foreign direct investment in banking versus the assets of US banks foreign branches There have been a few empirical studies about the determinants of the number of US banks' branches and oces abroad. Sagari (1986) analysed the determinants of US foreign investment in banking. Goldberg and Johnson (1990) examined those factors which determine the number of foreign branches of US banks and those factors aecting the assets of foreign branches of US banks over the period 1972±1985. 2 These studies, as mentioned by Moshirian and Pham (1999), did not measure US FDI in banking abroad, but rather the assets of US banks foreign branches and the number of branches of US banks abroad. However, there are at least three major categories of investors who invest in banking abroad: 3 (i) Banks that establish their branches, agencies or subsidiaries in a foreign country. 2 There have also been some studies of foreign banks in the US. Amongst these studies are the works of Goldberg and Saunders (1981), Hultman and McGee (1989) and Grosse and Goldberg (1991). Goldberg and Saunders (1981) consider foreign banks' shares of total US commercial bank assets, and the number of oces in the US owned by foreign banks as the major dependent variables. As the purpose of Grosse and Goldberg's study (1991) was to analyze foreign bank activities in the US with respect to countries of origin, they used the assets of the banks of each of the foreign countries operating in the US as the dependent variable. Their study also has a good survey of literature in the area of US banks activities abroad and foreign banks activities in the US. 3 Foreign direct investment (cf. portfolio investment) implies that a person in one country has a lasting interest in, and a degree of in¯uence over the management of a business enterprise in another country. According to the US Commerce Department, US foreign direct investment abroad is ownership or control, directly or indirectly, by a single person (or associated group of persons) from the US of 10% or more of the voting securities of an incorporated foreign business enterprise or an equivalent interest in an unincorporated foreign business enterprise.
320
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
(ii) Banks that do not have any physical oce in the foreign country, but are shareholders of foreign banks (i.e., the name of the acquired banks do not change). While they may own more than 10% of some foreign banks, they do not have their own business in the foreign market. (iii) Non-bank corporations, institutions and individuals who invest in foreign banks, and who own 10 or more percent of these foreign banks. Using the example of US FDI abroad, the number of foreign oces owned by US banks may encompass some of the foreign business de®ned above, however neither the number of oces owned by US banks nor the assets of US banks' foreign branches fully re¯ect the amount of US FDI in banking. Indeed, according to the US Direct Investment Abroad: 1989 Benchmark Survey, bank aliates of non-bank US parents (i.e., category (iii)) accounted for 8% of the total US FDI abroad in banking in 1989. This ®gure for 1994 was 9.6%. In the case of category (ii), while the US Department of Commerce acknowledges the existence of such a speci®c category, it does not distinguish direct acquisitions by banks and partial ownership of foreign banks in its statistics. Table 2 in Appendix A to this paper reports the signi®cance of the geographical breakdown of the assets of US banks foreign branches, the number of foreign branches of US banks as well as US FDI in banking abroad. As can be seen, there is not a strong correlation between these three activities. For instance, while only about 4% of the assets of US banks foreign branches and 23% of the number of foreign branches of US banks are located in Latin America, over 36% of US FDI in banking is made in this region. Similarly, in Europe (excluding member countries of the EU), while less than 1% of the assets of US foreign branches are located here, over 10% of US FDI is occurring in this region. The same discrepancy can be seen in the EU and other regions for which data are available. These ®gures may reinforce the above three categories where not all FDI in banking is made by banks which establish their branches and oces abroad (i.e., category (i), as de®ned above) and hence the assets of US banks foreign branches abroad may not necessarily re¯ect the full magnitude of FDI in banking. Thus, given the above discussion and data reported in Table 2, one can see that data on FDI in banking are quite dierent from data on the assets of foreign branches of banks or that of the number of foreign branches of banks. 3. Factors contributing to FDI in banking A number of researchers have studied the determinants of US FDI in manufacturing abroad and foreign FDI in manufacturing in the US. 4 The
4 For a comprehensive survey of literature on FDI see Agarwal (1980), Grosse (1981), Lizondo (1990) and Caves (1996).
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
321
major determining factors identi®ed in these studies are: the size of the market in the host country, relative economic growth, cost of capital, exchange rates, interest rates and taxes. In addition, Williams (1997) has recently surveyed various established theories of FDI in the context of multinational banking in which he discusses major FDI theories such as the international investment theory, the eclectic theory and the industrial organisation hypothesis. In this study, similar to the study by Moshirian and Pham (1999) who analysed Australia's FDI in banking, hypotheses based mainly on the eclectic theory will be used to estimate FDI in banking. This approach is consistent with the previous studies of foreign banks' activities cited in this study. However, this study also uses a new proxy for the cost of capital and new explanatory variables such as banks' international lending activities and FDI in non-®nance industries which are dierent than those proxies and variables used by previous studies of foreign banks' activities abroad. This represents an enrichment of the traditional literature on the determinants of FDI in banking. Also, the use of superior data to measure US, UK and German investment in foreign banking ensures that the selected explanatory variables chosen to test the determinants of FDI in banking abroad re¯ect those factors aecting both non-bank and bank investors. The following factors, based predominantly on the eclectic theory of FDI have been chosen as describing a model of the determinants of foreign direct investment in banking: (i) banks' foreign assets, (ii) FDI in non-®nance industries, (iii) bilateral trade, (iv) the size of the foreign banking market, (v) the cost of capital dierential, (vi) relative economic growth, and (vii) the exchange rate. The remaining part of this section will discuss the rationale behind including the above factors. 3.1. Banks' foreign assets One of the most likely factors contributing to the expansion of FDI in banking is banksÕ foreign assets. Establishing a branch abroad makes the practice of international lending activities both easier and more eective. While banksÕ foreign assets (i.e., the banks claim against foreigners) can be generated both in domestic and foreign markets, the existence of a foreign branch makes it easier for banks to expand their international lending. Obviously, a foreign presence could also reduce transaction costs (including telecommunication costs, information gathering and processing costs) and further facilitate the expansion of international lending. Banks' foreign assets have not been used by previous researchers as an independent variable for measuring banks' activities abroad. This is because the previous studies, such as Goldberg and Johnson (1990) and Grosse and Goldberg (1991), did not distinguish between the US banks' foreign assets and the US banks' assets in their foreign branches. This distinction is important in this study,
322
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
as unlike previous studies in banks' foreign activities, the dependent variable used in this study is FDI in banking rather than US banks' assets in their foreign branches. 5 Thus, it is hypothesised that the larger a countryÕs banksÕ foreign assets, the greater will be the size of that countryÕs FDI in banking abroad. 3.2. FDI in non-®nance industries It has been argued by researchers such as Grubel (1977) and Gray and Gray (1981) that there is a positive relationship between FDI in manufacturing and the expansion of banks abroad. Researchers such as Nigh et al. (1986) and Goldberg and Johnson (1990) found a positive relationship between the US banksÕ foreign activities (including activities with US citizens abroad and foreign customers) and the size of US FDI. The rationale for this is that multinational banks will follow their multinational manufacturers abroad so that they can provide services for their customersÕ foreign operations. Unlike previous studies on banksÕ activities abroad, this study has access to both FDI in general less FDI in banking and ®nance and data on FDI in banking. Thus, as the dependent variable in this study is FDI in banking, the independent variable used to capture the so called ``gravitational pull eect'' is total FDI of country i less FDI in banking and ®nance. Thus, the hypothesis to be tested is whether FDI in banking is positively related to FDI in country i less FDI in banking and ®nance. 3.3. Bilateral trade Empirical studies such as Roemer (1975) and Agarwal (1980) on the relationship between general FDI and foreign trade indicate a positive relationship 5 Table 3 in Appendix A to this paper reports the size of the US banks' assets in their foreign branches, from 1972±1985 (the period for which Goldberg and Johnson (1990) measured US banks' activities abroad) and was compiled by Houpt (1988). These data sets are compared with the US banks' foreign assets reported by the IMF Balance of Payments Statistics. As can be seen there is a signi®cant discrepancy between these two data sets as they are representing two dierent things. According to the IMF de®nition of banks' foreign assets provided by Landell-Mills (1986), the US banks' foreign assets are de®ned as those assets held against non-US citizens, which can be generated both within the United States economy and outside the United States. However, the US banks have foreign branches in various countries, particularly in countries where US multinational corporations are operating. Some parts of the assets of US banks' branches and aliates abroad are held by US companies. In other words, the US banks' activities abroad are comprised of activities with US individuals and companies abroad as well as with foreign individuals and companies.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
323
between them. Furthermore Jain (1986) found that, in a sample of 46 countries, the USÕ share of total trade was highly correlated with its share of banking assets in these countries. Furthermore, Nigh et al. (1986) found a positive relationship between trade and foreign banksÕ activities. In this study, similar to previous studies of foreign banking activities, one of the hypotheses to be tested is a relationship between FDI in banking and bilateral trade for the three countries studied. In order to calculate the trade variable the following steps have been taken: if we take the US as the example, the relevant variable has been constructed as the weighted average of bilateral exports and imports between the US and her trading partners. The weight given to the sum of exports and imports with a given trading partner is given by that countryÕs share of US FDI in banking. 6 Given past empirical research on the eects of international trade on banking and the expectation that increased foreign trade will encourage trade ®nancing, a positive coecient is expected for the bilateral trade variable. 3.4. Size of the foreign banking market When investors (banks and non-banks) consider whether to invest in foreign banking, the size of the particular foreign banking market is likely to be one of the factors they take into account. This claim is supported by Kravis and Lipsey (1982) and O'Sullivan (1985) who argued that the size of the host country's market is one of the signi®cant determinants of FDI. The size of the foreign banking market is therefore proposed as a location-speci®c determinant of FDI in banking. In this study, similar to the study of Grosse and Goldberg (1991), the size of the foreign banking market is proxied by the weighted sum of two types of deposits held by the banks in country i: (i) time, savings and foreign currency deposits, and (ii) demand deposits, where the weights are given by that country's share of US, UK or German foreign direct investment in banking. The larger the foreign banking market, the greater the number of potential new customers for local banks. In turn, this suggests that there is a greater number of investors (banks and non-banks) willing to invest in foreign markets in order to take advantage of the market's potential. If the above were observed, a positive relationship between US, UK and German FDI in banking and the size of the foreign banking market would be expected.
6
Note that the weights used in this study are calculated from the annual data over 1983±1995 based on the US, the UK and German geographical distributions to each country in which they have FDI in banking. The weights are constant throughout the sample period.
324
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
3.5. Cost of capital dierential One of the possible determinants of FDI in banking is the cost of capital in the host and source countries. The traditional literature on cost of capital, especially in the banking area (see for example, McCauley and Zimmer, 1989) de®nes ``cost of capital'' as ``the minimum rate of return investors will require on their investment''. Furthermore, McCauley and Zimmer (1991) de®ne a banks' cost of capital for a ®nancial product as ``the spread or fee that allows the required regulatory capital to earn the rate of return demanded by the market''. Taxes are another of the most likely variables in¯uencing the level of FDI in banking in a country. The tax regime in use will determine whether or not a country is an attractive location for private investors to establish part of their foreign operations. Thus, the tax variable should be incorporated into the cost of capital so as to yield the after-tax cost of capital. This is in line with most studies of international investment, such as Lunn (1980), which does not explicitly consider the impact of tax on FDI. Even in FDI studies where the primary concern is the impact of taxation (like Hartman, 1984; Shah and Slemrod, 1991), the empirical analyses still focus on the eects of the after-tax rates of return to foreigners. As foreign investors consider the bank cost of capital in their own country as compared to that of foreign countries, the relative bank cost of capital is an important factor in determining whether a foreign investor will be willing to invest in banking abroad, as opposed to banks in her own country. Previous studies on foreign bank activities such as Goldberg and Saunders (1981) and Hultman and McGee (1989) used the price±earnings ratio for bank stocks in the US (annual average) as an indicator of principal purchases of banks' stock. However, as Poterba (1991) discussed several theoretical and empirical diculties in using the price±earnings ratio as a measure of the cost of equity, particularly across countries, in this study the cost of capital proposed by McCauley and Zimmer (1991) will be used as one of the hypotheses contributing to FDI in banking abroad. McCauley and Zimmer (1991) have shown that once the bank managers have calculated the cost of equity per bank, they should work out the spread or fee needed to be charged on individual ®nancial products to cover their equity costs. In other words, assuming that the spread is set in such a way to satisfy the required equity cost, one may use the spread (measured as lending rate minus deposit rate corrected for corporate tax) as a proxy for the cost of equity for banks. Assuming that the spread is set in such a way that it satis®es the required equity cost, one may use the spread (measured as lending rate minus deposit rate corrected for corporate tax) as a proxy for the cost of capital for banks. Thus, the larger the spread required by banks in order to cover their cost of
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
325
capital, the less competitive is their position in the international banking market. In this study, the spread between lending and deposit rates corrected for corporate tax rates of the US, the UK and Germany vis-a-vis the ``rest-ofthe-world'', will be used as a proxy for the cost of capital dierential of banks. 7 3.6. Relative economic growth Some researches have argued that one of the factors contributing to the expansion of foreign banks activities abroad is strong economic activity in the host countries. For instance, Hultman and McGee (1989) found a positive relationship between foreign banksÕ activities in the US and economic growth in the US. Furthermore, Goldberg and Saunders (1981), Nigh et al. (1986) and Sabi (1988) found that relative economic growth contributes to the expansion of foreign banking investment. In this study, similar to the above studies, a relative economic growth variable is calculated as a ratio of the growth in GDP of the US, the UK or Germany over the weighted average growth in GDP of those countries in which these three countries invest in banking. A positive relationship indicates that an increase in relative economic growth expands FDI in banking abroad more than it does domestic investment in banking. However, a negative relationship implies that an increase in relative economic growth makes investment in domestic banking more likely than FDI in banking abroad.
7 Note that when the bank cost of capital was calculated for the period after 1992, the calculations were slightly dierent than the spread minus corporate tax used in the previous quarters due to the so-called Basle agreement. The equation used for the post 1992 calculations is based on the spread, in which the overall requirement of 4% equity (the Basle Agreement risk weights, where risk weight for a corporate loan is 100%) is taken into account. As McCauley and Zimmer (1991) argued, banks are funding 4% of the loan with shareholder equity. The payment to equity is for the spread on the loan plus the real (net of in¯ation) after-tax return earned by investing the shareholder equity in a riskless asset. Thus instead of using S 1 ÿ tc1 COE RW 0:04, the following equation is used to calculate the spread for the years post 1992 when the Basle agreement was in place:
0:96S 1 ÿ tct 0:04f1 rt 1 ÿ tct g= 1 ps ÿ 1 COE RW 0:04; where S is the spread between lending and deposit rates, tct the marginal corporate income tax rate at time t, rt the riskless nominal interest rate at time t, ps the in¯ation rate at time t, COE the cost of equity, and RW is the risk weighting. As can be seen from the above two equations, the dierence between the cost of capital calculated prior to 1992 and that calculated post 1992 is very small.
326
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
3.7. Exchange rate Given that foreign private investors' operations may involve substantial ¯ows of diverse foreign currencies, exchange rates are expected to have an impact on their FDI decisions in banking. One possible explanation for a positive correlation between the exchange rate and FDI in banking abroad may be found in the valuation by investors of their foreign banking assets in the host countries' currencies. Hultman and McGee (1989) argued that given a standard accounting practice is to adjust the book value of nondomestic assets to re¯ect changes in the exchange rate, foreign investors may expect that any investment they make in the host countries will appreciate as the local currencies appreciate. On the other hand, it could be argued that such behavior is more likely for short term liquid assets. The book value of long term assets should be less prone to such ¯uctuations. Extending this argument, one would expect a negative correlation between the value of the host countries' currencies and FDI in banking by the US, the UK and Germany. When the host countries' currencies appreciate with respect to foreign investorsÕ currencies, FDI in the US is expected to decrease as it becomes more expensive for foreign investors to invest in those countries, and vice versa. In other words, a depreciated domestic currency will give foreigners an edge in acquiring the control of domestic productive assets. Such a negative correlation is reported by authors such as Goldberg and Saunders (1981), Cushman (1988) and Froot and Stein (1991). Note also that, as the host countries' currencies depreciate, private investors may reduce their repatriated income and increase their reinvestment in the host countries, as they may want to avoid exchange rate losses. Thus, the hypothesis to be tested is whether ¯uctuations in the value of the host countries' currencies aect the level of FDI in banking in those countries. 8 A few researchers such as Goldberg and Johnson (1990) have considered risk as one of the factors aecting FDI. At the same time, the recent study by Santomero and Babbel (1997) indicates some of the shortcomings of using proxies to measure ®nancial risk. As mentioned in footnote 1, over 96% and 90% of UK and German FDI in banking, respectively, are in major developed countries and in the case of the US this ®gure is 80%. Thus, given that the US, the UK and Germany have such a high proportion of FDI in banking in major developed countries, country risk would not be a major factor for investment
8
There are several indices available to account for exchange rates. The most appropriate one is the IMFÕs Multilateral Exchange Rate Model (MERM) index, which is superior to the other indices available for the purposes of this study. The way the index is constructed implies that an increase in the US$Õs MERM index indicates an appreciation of the US$ with respect to a weighted basket of the other currencies.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
327
and hence the comparative risk for investors investing in the form of FDI in banking will not be considered in this study. 4. A model of foreign direct investment in banking Based on the previous discussion, the proposed model for US, UK and German (country i) FDI in banking is as follows: Xi f BAi ; FDi ; BTi ; CDi ; CCi ; EGi ; EXi :
1
Xi stands for the stock of country iÕs FDI in banking. BAi is the banks' foreign assets of country i. FDi stands for the stock of country iÕs FDI in non-®nancial industries, BTi is the weighted average of country iÕs bilateral exports and imports with her major trading partners. The weight given to each trading partner is based on country iÕs total FDI in banking in that country. CDi stands for the weighted average domestic commercial bank deposits for those countries facing country i's FDI in banking. CCi is the cost of the equity dierential for banks of country i and those countries facing country i's FDI in banking. The weight given to each country is based on that country's share of FDI in banking from country i. EGi stands for the relative economic growth of country i. The economic growth of countries facing country i's FDI in banking is calculated as a weighted average of these countries' economic growth. The weight given to each country is based on country iÕs FDI in banking in that country. EXi stands for country iÕs MERM index (a measure of the value of currency i). The period covered is from 1983:1 to 1995:4, the only period for which data are available. All variables are expressed in current US$. In summary, the model for country iÕs FDI in banking can be expressed as lnXi a0 a1 lnBAi a2 lnFDi a3 lnBTi a4 lnCDi a5 CCi a6 EGi a7 lnEXi
2
with the following expected signs: a1 > 0; a2 > 0; a3 > 0; a4 > 0; a5 < 0; a6 ÿ ?; a7 ÿ ?:
5. Data and methodology The model employed uses time-series quarterly data for the period 1983:1 to 1995:4. Eq. (2) was estimated for the US, the UK and Germany using the data base IMF International Financial Statistics, various years, from which banks' foreign assets, exchange rates, bankÕs lending and deposit rates, national in-
328
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
comes, and banksÕ domestic deposits were obtained for all relevant countries. Tax rates are from publications of the Brookings Institution (edited by Pechman, 1988), the National Bureau of Economic Research (edited by Razin and Slemrod, 1990), and the OECD Observer Supplement (1996). US, UK and German bilateral exports and imports were taken from the IMF Directions of Trade, various issues. Data on foreign direct investment in banking and non-banking were obtained from the International Division of the US department of Commerce, the Bank of England Central Statistics Oce, and the Deutsche Bundesbank Statistics Oce (Note that Germany's annual data on various FDI have been used to construct their quarterly data.) Disaggregated data on foreign direct investment in banking have been obtained, for the US, from the ``Survey of Current Business'', for the UK, from the ``Bank of England Statistical Abstract'', and for Germany, from the ``Deutsche Bundesbank Zahlungsbilanzstatistics''. Data on general FDI for the US come from the US Department of Commerce ``Survey of Current Business'', for the UK, from the Central Statistics Oce ``Economic Trends'', and for Germany, from the ``Deutsche Bundesbank Zahlungsbilanzstatistics''. The Survey of Current Business has published the total FDI abroad for the US, and total FDI in the US at the market value and the current cost for the last few years. For instance, the current value of total FDI in the US for 1995 comprises total FDI in the US in 1994 corrected for capital in¯ows, ``price changes'', ``exchange rate changes'' and ``other changes'' 9 in 1995. However, unlike the UK and Germany, detailed estimates of FDI in the US by industry and by country are on a historical-cost basis. As Bargas and Lowe (1994) indicate ``estimates on a historical-cost basis largely re¯ect prices at the time of investment rather than prices at the current or any other period. Historical cost is the basis used for the evaluation of company accounting records in the United States and is the only basis on which companies can report data in the direct investment surveys conducted by the Bureau of Economic Analysis''. 10 The Survey of Current Business has also published annual data on FDI in banking in the US since 1980. These data which consist both of book values of
9 ``Other changes'' re¯ects changes in the value of the ocial gold stock due to ¯uctuations in the market price of gold. However, as this item is very small compared to price and exchange rate changes, this study will not take it into account when converting FDI in banking from book value to nominal or constant value. 10 Helkie and Stekler (1988) were two of the ®rst researchers to argue that the FDI book values of US companies abroad and foreign companies in the US are not good economic measures of the stock of FDI on the ground, and that these values are expressed neither in current nor in constant dollars, but rather they are historical stock values and hence should be corrected for exchange rate variations and in¯ation. This study will follow their approach for the correction of the book values of FDI in banking.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
329
FDI and capital in¯ows of FDI in banking are on a historical-cost basis. Indeed, many studies in FDI used either the book value of FDI or the dierence between consecutive quarters in FDI as a measure of capital ¯ow (see, for instance, Cushman, 1988). In this study, the data on the stock of FDI in banking in the US is converted from being on a historical-cost basis to being in current values. This ensures that both dependent and independent variables are compatible with each other. For instance, FDI in banking in the US for 1985:1 is constructed on the basis of the book value of FDI in banking in 1984:4 which has been corrected for the exchange rate variations and in¯ation of 1985:1. The result is then added to quarterly capital in¯ows in banking in the quarter 1985:1. 6. Empirical results The ®rst step in estimating Eq. (2) was to establish whether all variables are stationary. For this purpose, the Augmented Dicky±Fuller (ADF) and Phillips±Perron unit root tests were employed. The results of these tests showed that most of the variables for the US, the UK and Germany are non-stationary and that some of these variables have one unit root. This suggests that the corresponding ¯ow variables will be stationary. The Augmented Dicky±Fuller unit root tests applied on the ¯ow variables suggest that all variables are stationary, as can be seen from Table 4 in Appendix A to this paper. Thus, the ¯ow variables in Eq. (2) were used for estimation purposes for the US, the UK and Germany. Furthermore, as the ¯ow data used in Eq. (2) had random variations, a moving average with four lags was used to smooth out the volatility of the quarterly data. In addition, Table 5 in Appendix A to this paper reports the correlation matrix between all the variables in Eq. (2) using the ¯ow data. As can be seen, the correlation matrix between all variables is low. As can be seen from previous studies of FDI, almost all these studies used OLS regression as a means of statistical analysis. However, the empirical results of past studies such as Goldberg and Johnson (1990) and Grosse and Goldberg (1991) have been subject to statistical problems such as serial correlation, heteroskedasticity and simultaneity bias and hence their results should have been treated with caution. In order to avoid these statistical problems, the Generalised Methods of Moments (GMM) is used in this study. The empirical results show that US, UK and German banksÕ foreign assets (BAi ) are positively related to FDI in banking. This result indicates that, as the international lending activities of banks increase, there is an accompanying increase in their FDI in banking. This result is consistent with the ®ndings of Moshirian and Pham (1999) for Australia. Thus, while one can argue that US, UK and German banks' international lending can account for the expansion of these three countries' FDI in banking, the coecient of this variable is larger
330
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
for Germany than the US and the UK indicating that this variable has greater in¯uence on FDI in banking in Germany than in the two other countries. The positive sign between FDI in banking and FDI in the non-®nance industries (FDi ) indicate that FDI in banking and FDI in other industries are complementary. This empirical result is consistent with the ®ndings of Hultman and McGee (1989) and Goldberg and Johnson (1990) who found a positive relationship between foreign banks activities in the US and the total of FDI minus FDI in ®nance, insurance and real estate, as an explanatory variable. The t-test for the statistical dierences between the coecients indicated that the German coecient is greater than that of the UK implying that FDI in manufacturing has more impacting than on that of the UK. Bilateral trade (BTi ) variable is found to be positively related to FDI in banking. This result con®rms the theoretical argument that trade is one of the contributors of FDI in banking. This result is also consistent with the ®ndings of Nigh et al. (1986) and Goldberg and Johnson (1990). The test for the statistical dierences between each pair of coecients demonstrates a dierence between the coecients of the US and Germany, indicating that the trade variable has a stronger impact on German FDI in banking. The banks' commercial deposit variable (CDi ) is statistically signi®cant with a positive sign for the US, the UK and Germany. This result is consistent with the ®ndings of Grosse and Goldberg (1991) that foreign banks invest in countries whose domestic banking market is large. Given this result, one could assume that the host countries' governments have every incentive to increase domestic savings and deregulate the domestic ®nancial market in an eort to attract more foreign banks. As can be seen from Table 1, FDI in banking abroad is negatively related to the cost of equity dierential (CCi ) for banks between the US, the UK and Germany and their partner countries. This result implies that an eective and competitive cost of capital structure by banks contributes to their success when entering into the foreign banking markets. It is conceivable that banks from the US, the UK and Germany have access to more diverse markets for capital or a better credit rating than the banks from the host countries. The size of the coecient of this variable for the US is much greater than that of the other two countries. This indicates that the cost of capital dierential has more impact on the amount of the FDI in banking by the US than the UK and Germany. The relative economic growth variable (EGi ) is statistically signi®cant with a negative sign for the US, the UK, Germany. The empirical result shows that as economic growth in these three countries exceed that of their host countries, FDI in banking to these host countries declines. In other words, stronger economic activities in the source countries increase demand for domestic credit and hence more domestic investment in banking. The size of the coecient of this variable indicates that German FDI in banking is more aected by relative economic growth than that of the US or the UK.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
331
Table 1 The empirical results of the foreign direct investment in banking model (1983:1±1995:4)a US a BAi FDi BTi CDi CCi EGi EXi R2 DW JBb WHETc
UK
)0.37 (2.1) 0.04 (11.6) 0.3 (2.1) 0.40 (3.0) 0.008 (4.6) )13.71 (2.8) )1.67 (7.0) )40.5 (2.3) 0.67 1.9 0.80 28.6
Germany
)0.97 ()16.5) 0.013 (20.2) 0.03 (8.0) 0.56 (7.9) 0.005 (4.9) )0.49 ()11.2) )1.05 ()7.5) )81.9 ()4.9) 0.76 1.9 1.8 42.3
)0.44 ()7.4) 0.55 (3.9) 0.19 (5.2) 0.88 (5.4) 0.01 (3.4) )0.41 ()6.3) )21.3 ()16.6) )392 ()5.7) 0.29 1.98 3.4 34.7
a and indicate that an estimate is signi®cantly dierent from zero at the 5% level and 1% level, respectively. Student t-statistics are shown in parentheses. GMM: generalised methods of moments. b Jacque±Bera test statistic for normality; p value in parentheses. c White test for heteroscedasticity.
The empirical results also show that the exchange rate variable for the US, the UK and Germany (EXi ) has a negative relationship across these countries with respect to FDI in banking. One could argue that the higher the value of the currencies of these three countries relative to the currencies of their host countries, the cheaper it becomes for investors from these three countries to invest in foreign banking. The size of the coecient of this variable indicates that UK investors are more responsive in their investment abroad in the wake of depreciation of the host countries currencies, than the other two countries studied. 7. Conclusion This paper has distinguished between banks' activities abroad and FDI in banking by banks and non-banks. It has argued that the number of oces owned by banks abroad may correlate to the amount of FDI in banking, however, it does not include investment by banks who invest in incorporated foreign banks. Furthermore, there are non-bank investors who own over 10% of some banks in foreign countries, and yet neither the number of oces owned by banks nor the total assets of banks' branches in foreign countries fully re¯ect the amount of FDI in banking. A model for FDI in banking has been proposed which comprises certain explanatory variables peculiar to FDI in banking as compared to FDI in general and/or FDI in manufacturing for the US, the UK and Germany. The components of the model of FDI in banking is dierent from those models designed to explain
332
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
banks activities abroad. The empirical results of this study of FDI in banking indicate that banks' foreign assets contribute to the expansion of FDI in banking by both banks and non-bank investors. This result indicates that while banks could be engaged in international lending activities from their home countries, access to foreign markets is a very eective way of expanding their international activities. The empirical results also show a close link between FDI in banking and FDI in non-®nance sector as well as between bi-lateral trade. The empirical results support the theoretical argument that those countries with better cost of capital structures, which have stronger ®nancial positions when competing with the host countries' banks. Furthermore, this paper has shown that if the economic growth in the US, the UK and Germany increases faster than that of their host countries, investors of these three countries tend to invest more in their own countries rather than in the banking sector of the foreign countries. The exchange rate appears to be one of the decisive factors in expanding FDI in banking. As currencies appreciate, there is less incentive for investment in the foreign banking market. This result implies that both bank and nonbank investors have a long term view of investment in banking and hence they are not motivated for short term pro®ts. Given some of the above factors which can be generalized as those factors that the US, German and British private investors consider as universally important factors in determining their amount of these three countries' FDI in banking, one would expect that, with better foreign market access and ``right of establishment'' for trade in ®nancial services in the ``Post Uruguay Era'' and successful bilateral trade agreements between the US and Japan, private investors (banks and non-banks) may increase their investment in some nonOECD countries where there is strong economic growth, a high level of FDI in manufacturing, low cost of capital and rapid expansion of the domestic banking market.
Acknowledgements The author would like to thank the two anonymous referees for their valuable comments and suggestions. The assistance of Alexander Van der Laan is also gratefully appreciated. However, all errors are mine.
Appendix A See Tables 2±5.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
333
Table 2 Comparison of geographical distribution of the assets of US banks foreign branches, the number of foreign branches of US banks and US FDI in banking in percentagesa Region
1979
1980
1981
1982
1983
1984
1985
European Union
Ab Bc Fd
48 19.3 35
48.5 19 35
47.1 19 30.4
45.9 18.2 39.8
43 17.3 28.5
44.9 17.5 28
46 18.6 31.6
Other Europe
A B F
1.11 0.02 10.2
Japan
A B F
n.a. n.a. n.a.
n.a. n.a. n.a.
n.a. n.a. n.a.
5.1 )0.3 1.6
Other Asia
A B F
12 15.8 16
7.9 17.8 9
8.9 17.8 10.2
10.3 19.5 12.4
11.3 0.2 12
12.4 20 11.3
12.6 20 8.5
Latin America
A B F
4.3 20 34
4.3 20.4 34.1
4.5 21.5 37.3
4.3 22.5 33
3.3 22.8 34.4
3.3 23.8 37.7
3.4 23.7 3.4
Africa
A B F
a
0.004 0.02 1.3
1 0.024 9.7
0.002 0.02 1.7
1.2 0.0221 8.2
0.002 0.02 1.8
0.008 0.02 8
0.002 0.02 1.6
0.009 0.021 10.8
0.059 0.025 9.5
0.007 0.024 11.3
5.5 0.03 1.4
5.6 0.03 1.3
5.4 0.03 1.2
0.002 0.025 1.6
Sources: The Survey of Current Business, various issues and Houpt (1988). Percentage of the assets of US banks foreign branches. c Percentage of the number of foreign branches of US banks. d Percentage of US FDI in banking abroad. b
0.002 0.02 1.6
0.003 0.02 0.15
a
77.4 118.0 18.0 23.9
140.5 42.5
1974 162.7 54.7
1975 193.8 72.6
1976 227.9 88.0
1977 257.6 130.0
1978 312.9 156.57
1979
Source: Houpt (1988) and The IMF, Balance of Payment Statistics Yearbook, various annual issues.
The Assets US banks foreign branches The US banks' foreign assets
1972 1973 343.5 203.9
1980
391.0 292.8
1981
Table 3 Comparison between the assets of US banks foreign branches and US banks' foreign assets (in billions of dollars)a 388.5 401.3
1982
386.1 443.1
1983
337.4 443.3
1984
392.2 446.7
1985
334 F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
335
Table 4 Augmented Dicky±Fuller statistics for the ¯ow model of Eq. (2) s^
s^T
s^u
Variables
US
UK
Germany US
UK
Germany US
UK
Germany
Dx DBA DFD DBT DCD DCC DEG DEX
)4.6 )4.0 )6.3 )2.3 )3.7 )5.3 )3.6 )4.2
)4.6 )4.1 )9.6 )5.6 )4.1 )5.4 )4.5 )4.5
)4.6 )3.4 )9.6 )4.8 )4.7 )5.5 )3.9 )4.9
)6.9 )4.5 )9.5 )5.8 )3.9 )5.5 )4.9 )3.4
)6.9 )4.4 )9.5 )4.7 )4.8 )5.5 )5.1 )5.6
)7.1 )4.7 )9.4 )5.7 )4.1 )5.4 )4.0 )3.4
)7.1 )4.8 )9.4 )4.8 )4.9 )5.5 )5.1 )5.5
)6.9 )4.4 )6.2 )2.9 )3.9 )5.5 )5.1 )5.8
)7.1 )5.0 )6.1 )2.9 )4.4 )5.2 )5.0 )6.0
Table 5 The correlation matrix for the US/Germany/UK using data in Eq. (2) X
BA
FD
BT
CD
CC
1 0.17 0.37 0.16 0.52 0.03 )0.14 )0.53
1 0.05 )0.22 )0.21 )0.08 )0.03 )0.15
0.1 0.01 )0.35 )0.02 )0.09 )0.28
1 )0.19 )0.17 0.22 )0.08
1 )0.05 )0.09 )0.75
1 )0.22 0.05
1 0.12
1
For Germany X 1 BA 0.40 FD 0.44 BT 0.17 CD 0.49 CC 0.02 EG 0.09 EX 0.35
1 )0.50 0.29 0.48 )0.2 0.02 0.42
1 0.33 0.72 0.06 0.08 0.56
1 0.36 )0.08 )0.07 0.56
1 )0.04 0.32 0.35
1 )0.13 )0.00
1 )0.09
1
For UK X BA FD BT CD CC EG EX
1 )0.14 )0.26 0.13 0.15 )0.44 0.28
1 )0.05 )0.02 )0.28 )0.32 )0.09
1 )0.02 )0.10 )0.02 0.11
1 0.20 )0.35 )0.02
1 0.02 )0.15
1 0.39
1
For US X BA FD BT CD CC EG EX
1 0.33 )0.11 0.14 )0.02 0.03 0.51 0.43
EG
EX
336
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
References Agarwal, J., 1980. Determinants of foreign direct investment: A survey. Weltwirtschaftliches Archiv 116, 739±773. Bargas, S., Lowe, J., 1994. Direct investment positions on a historical basis. Survey of Current Business 62, 72±85. Caves, R., 1996. Economic Analysis and Multinational Enterprise, second ed. Cambridge University Press, Cambridge. Cushman, D., 1988. Exchange-rate uncertainty and foreign direct investment in the US. Weltwirtschaftliches Archiv 124 (2), 322±336. Froot, K., Stein, J., 1991. Exchange rates and foreign direct investment: An imperfect capital markets approach. The Quarterly Journal of Economics 106 (4), 1191±1217. Goldberg, L., Johnson, D., 1990. The determinants of US banking activity abroad. Journal of International Money and Finance 9, 123±137. Goldberg, L.G., Saunders, A., 1981. The determinants of foreign banking activity in the US. Journal of Banking and Finance 5, 17±32. Gray, J., Gray, P., 1981. The multinational bank: A ®nancial MNC? Journal of Banking and Finance 5, 33±63. Grosse, R., 1981. The theory of foreign direct investment. University of South Carolina Essays in International Business 3, 1±15. Grosse, R., Goldberg, L.G., 1991. Foreign bank activity in the US: An analysis by country of origin. Journal of Banking and Finance 15, 1093±1112. Grubel, H., 1977. A theory of multinational banking. Banca Nazional del Lavoro Quarterly Review 30, 349±364. Hartman, D., 1984. Tax policy and foreign direct investment in the US. National Tax Journal 37, 475±488. Helkie, W., Stekler, L., 1988. Modelling investment income and other services in the US international transactions accounts. International Finance Discussion Papers 319. Houpt, J., 1988. International Trends for US Banks and Banking Markets. Washington, DC. Hultman, C.W., McGee, L.R., 1989. Factors aecting the foreign banking presence in the US. Journal of Banking and Finance 13, 383±396. Jain, A.K., 1986. International lending patterns of US commercial banks. Journal of International Business Studies Fall 73±88. Kravis, I.B., Lipsey, R.E., 1982. The location of overseas production and production for export by US multinational ®rms. Journal of International Economics 12, 201±223. Landell-Mills, J., 1986. The Fund's International Banking Statistics. IMF, Washington, DC. Lizondo, S.J., 1990. Foreign direct investment. International Monetary Fund Working Paper: WP/ 90/63, International Monetary Fund, Washington, DC. Lunn, J., 1980. Determinants of US direct investment in the EEC: Further evidence. European Economic Review 24, 93±101. McCauley, R.N., Zimmer, S.A., 1989. Bank cost of capital and international competition. Federal Reserve Bank of New York Quarterly Review 15, 33±59. McCauley, R.N., Zimmer, S.A., 1991. Cost of capital for industry and banks. Business Economics 26, 14±18. Moshirian, F., Pham, T., 1999. Cost of capital and Australia's banking investment abroad. Applied Financial Economics 9, 295±303. Nigh, D., Cho, R., Krishnan, S., 1986. The role of location-related factors in US banking investment abroad: An empirical examination. Journal of International Business Studies, Fall, 59±72. O'Sullivan, P., 1985. Determinants and impact of private foreign direct investment in host countries. Management International Review 25, 28±35.
F. Moshirian / Journal of Banking & Finance 25 (2001) 317±337
337
Poterba, J., 1991. Comparing the cost of capital in the US and Japan: A survey of methods. Federal Reserve Bank of New York Quarterly Review 15, 20±32. Roemer, J.E., 1975. US-Japanese competition in international markets: A study of the trade± investment cycle in modern capitalism. University of California, Institute of International Studies, Research Series 22, Berkeley. Sabi, M., 1988. An application of the theory of foreign direct investment to multinational banking in LDCs. Journal of International Business Studies 19, 433±447. Sagari, S., 1986. The Financial Services Industry: An International Perspective (unpublished doctoral dissertation). New York University. Santamero, A., Babbel, D., 1997. Financial risk management by insurers: An analysis of the process. Journal of Risk and Insurance 64, 231±270. Shah, A., Slemrod, J., 1991. Do taxes matter for foreign direct investment? The World Bank Economic Review 5, 473±491. UN World Investment Report, 1994. United Nations, New York. Williams, B., 1997. Positive theories of multinational banking: Eclectic theory versus international theory. Journal of Economic Survey 11, 71±100.