International trade in banking services

UTTERWQRTH I N E M A

N N

Journal of International Monzv and Finance, Vol. 14, No. 1, pp. 47-64, 1995 Copyright © 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0261-5606(94)00006-9 0261-5606/95 $10.00 + 0.00

International trade in banking services J A N TER W E N G E L *

Economic and Social Institute, Free University, 1081 H V Amsterdam, The Netherlands This paper investigates which of the theories of international trade best explains trade in international banking in its various forms: branches, subsidiaries and representatives. Strong evidence is found to support the newer economies of scale theories and it is proposed that the foreign exchange and capital markets exhibit declining costs to production. It is also discovered that the relaxation of exchange and capital controls by potential host countries diminishes the incentives of banks to seek direct representation. (JEL F33).

International banking is hardly a new invention. Nevertheless, the rapid growth of this activity together with the multiplication of banking services has motivated an increased interest in the evaluation of the costs and benefits of the establishment of foreign banks. This interest is not only academic. In the recent Uruguay Round of Trade Negotiations no agreement was reached on the issue of trade in financial services. Many policy makers in both developing and developed nations are wary of not getting market access or losing market share in the delivery of these services. The profession has recently responded to this augmented interest in foreign banking with a number of articles designed to empirically investigate the determinants of international banking. A good example of the recent literature is provided by the article in this journal by Heinkel and Levi (1992) who study foreign banking in the USA to examine the choice of form of representation by foreign banks--branches, agencies, representatives or subsidiaries--on the basis of the following country-specific factors: exports to the US; claims against the specific country held by US customers; and the size of the capital market of the bank exporting country. In a somewhat similar effort Hultman and McGee (1989) examined the variation over time in the total foreign banking presence in the USA. In another article in this journal, Goldberg and Johnson (1990) seek to explain the determinants of US bank presence abroad on the basis of: trade, foreign direct investment, population, income per capita, domestic banking activity, * The project on which the present paper is based was funded by the Indo-Dutch program on Alternatives in Development (IDPAD) in collaboration with the Indian Council of Social Science Research (ICSSR) and the Institute for Social Science Research in Developing Countries (IMWOO). The help of Marc Schramm in the compilation of the data is gratefully acknowledged.

Trade in bankin# services: J ter Wenoel

regulation of foreign banks, and exchange rate stability. Since in the world context, the USA represents a polar case with respect to most economic variables, it is felt that many of the conclusions drawn from the above mentioned studies cannot be extended to intermediate sized countries, developed or developing. The purpose of this study is to search for the determinants of international trade in banking services in a global context. To accomplish this objective the bilateral pattern of international trade in banking among 141 countries 1 is analyzed in the light of the different theories of international trade for three forms of international bank representation: branches, representatives and subsidiaries. 2 Attention was focused on the different theories of international trade because their informational content might be suggestive of the future developments of this activity. This paper also makes a modest contribution to the literature on international financial centers by proposing an hypothesis with respect to the development of banking centers. Unlike most of the literature on financial centers--for example Reed (1981) or Goldberg et al. (1991)--we do not deal with the determinants of international financial centers at an aggregate level but with the bilateral and disaggregated services of banks to identify the characteristics of countries that export different banking services through branches, subsidiaries or representatives on the realistic assumption that an investment in these forms of representation indicates the disposition to provide a service. We avoid the unresolved question of whether the provision of banking services constitutes an international trade or a foreign direct investment issue because the analytical question to be answered in both instances is the same: What is the source of comparative advantage accruing to a bank of one nation in the provision of services to another in competition with domestic banks familiar with local customers, capital markets and regulations? The paper is organized as follows. In Section I a model is specified to infer the factors that provide some nations with a comparative advantage in the banking industry. This model is applied empirically in Section II. In Section II.A the sources, the collection and the quality of the data employed to estimate the model are discussed. Section II.B deals with the empirical results. A number of implications of the study are presented in a brief concluding section.

I. A model to infer the determinants of international banking

The purpose of this section is to specify a model to investigate which of the principal trade theories best explains the actual pattern of trade in banking services. The trade theories considered are the classical, the Heckscher-Ohlin, the Linder, 3 the product-cycle and the economies of scale or monopolistic competition theories, that deal with intra-industry trade and product differentiation. Although the above mentioned theories have been carefully specified and provide clear policy recommendations, there is no generally agreed upon method to verify the relevance of the different theories in situations with many countries, many goods and many factors as made clear by Deardorff (1984). Nevertheless, although the theoretical justification may not be rigorous, a large body of 48

Journal of International Money and Finance 1995 Volume 14 Number 1

Trade in banking services. J ter Wengel

literature that has contributed significantly to the understanding of the factors that explain the patterns of trade has been built by regressing trade on its proposed determinants. The majority of these studies concentrated on explaining the commodity composition of the trade of one country by regressing the trade performance of a cross section of industries on their input requirements. Relatively few studies have followed the approach pioneered by Learner (1974) in which the dependent variable, the trade of a particular good, is regressed on resource availability to 'estimate (roughly speaking) industry input requirements.'4 To examine bilateral trade in banking the dependent variable, BX~j,is regressed on the factors or characteristics suggested by the different theories of international trade. The classical or Ricardian theory explains trade on the basis of the relative endowments of natural factors, N~/Nj. The natural endowments that would be relevant to the banking industry might be political or macroeconomic stability or the abundance of capital as reflected by low interest rates or balance of payments surpluses. In fact there exists such a multitude of variables that could be utilized in this context that the relation N~/Njis only included in the theoretical model but is not employed in the empirical work of the next section. The Heckscher-Ohlin theory states that trade depends on the relative capital and labor endowments of countries. In the more sophisticated treatments of the theory attention is not limited to physical capital and human capital is given an important place. Moreover, labor is not taken to be homogeneous but is subdivided along skill categories. Although these refinements are extremely useful in the situation of one country with respect to the rest of the world, in the analysis of a large cross-section of countries the availability of data becomes a limiting factor. Therefore in many empirical studies the capital-labor ratio is proxied by income per capita. This example is followed here and the second independent variable to explain BX~jis the ratio of the incomes per capita in the two countries: Yi/Y~. Assuming banking to be capital intensive a positive coefficient is expected for this ratio. The Linder hypothesis proposes that trade takes place among countries with similar tastes and it is often assumed that these tastes are determined by their incomes per capita. An index of difference, denominated LINij, was developed to test whether banking exports take place among countries with more similar or dissimilar tastes:

(1)

LIN, j -

A

Styi - y j)

(1/2)(y, + y~)'

where A B S stands for the absolute value of the differencein incomes per capita. It is anticipated that the index of difference will reveal a negative coefficient because of the close association that is presumed to exist between the degree of development and the provision of financial services. That is, countries at different levels of development might exhibit very different needs with respect to banking services and therefore such trade might not be very important. The product-cycle theory proposes that new products move from the richer to the poorer countries over time. Although international banking is not a new phenomenon, its rapid development in the past few decades, plus the possibilities Journal o[' International Money and Finance 1995 Volume 14 Number 1

49

Trade in banking services: J ter Wengel

generated by the revolutions in telecommunications and data processing, suggest that the new international banking may be considered a new product. Thus the product-cycle theory would be suggested if the coefficient for the Heckscher-Ohlin test was positive and the Linder test was rejected by a positive coefficient for the index of difference term. A better test of the product-cycle theory would be obtained by breaking up the sample according to rich and poor countries and observing positive and negative coefficients for the income per capita ratios for the two sub-samples, respectively. The newer theories of international trade indicate that countries will tend to export those goods in which they can internalize economies of scale. Because the literature on domestic banking has found little evidence for the proposition of economies of scale to branching (Ferrufino, 1991; Nelson, 1985), the proposition of the economies of scale theory requires the formulation of an hypothesis with respect to the source of economies of scale at the international level. Such an hypothesis might be inferred from the association between specialization and economies of scale (Young, 1928) and the discovery by Heinkel and Levi that banks use different forms of representation as the bilateral relationship between the countries grows in importance from exports to investment banking to participation in the capital market. Turning this line of reasoning around, it is possible to suggest that economies of scale might be achieved in the foreign exchange and capital markets. In order to develop a simple intuitive model of the operation of economies of scale in the foreign exchange and capital markets it may be useful to imagine a world of isolated nations in which there would be no need for a foreign exchange market. In this world, a trading scheme between two nations, A and B, would induce the creation of a market between their currencies. Initially the currency exchange costs might be very high as the foreign exchange markets would be dominated by brokers that would limit themselves to the search and matching of buyers and sellers. With the growth of trade the foreign exchange market would expand and some brokers might become market makers and be willing to quote two-way prices, and to take open positions in any of the two currencies. With the expansion of trade and the growth of the foreign exchange market the costs per unit of currency converted would decline. The suggestion of the existence of declining costs to currency conversion does not provide any clues as to the location of the foreign exchange market. Even if it were assumed that country A was much larger than country B there would be no clear motivation for the foreign exchange market to concentrate in either A or B. The addition of another small country C to the trading scheme could resolve the foreign exchange market location problem. If the volume of trade between A and B and A and C were sufficient to lead to low enough conversion costs, then the small volume of trade between B and C might make use of the B - A - C currency conversion channel and lead to the concentration of the foreign exchange market in country A. The tendency for the foreign exchange market to locate in country A would be reinforced by any learning effects in the foreign exchange markets. Thus, if by chance the exchange market had already been established in A when country C joined the trading scheme, it would require less of an initial investment on the 50

Journal of International Money and Finance 1995 Volume 14 Number 1

Trade in banking services. J ter Wengel

part of A than of C to set up an A-C foreign exchange market. Like the liberalization of trade, the relaxation of restrictions on foreign investment would also promote the concentration of the foreign exchange market in the large country. The very simple model delineated above suggests a tendency for the foreign exchange and capital markets to concentrate in the larger countries due to economies of scale. Therefore country GNPs, denoted as Yi for the exporting country and as Yj for the importing country, could be selected as the size variables relevant for the examination of the existence of economies of scale. The size variables are included as absolute variables rather than a relative variable because a ratio would hide the relevance of sheer size. The hypothesis of economies of scale would be supported by a positive coefficient for the economic size of the exporting country and a null or negative coefficient for the size of the importing country. Beyond the examination of the various theories of international trade, the analysis could be extended by the inclusion of a number of other terms to test various hypotheses generally held with respect to trade in banking services and to take into account a number of characteristics particular to banking. Because banking is a service industry concerned with the money flows of counterpart real international transactions, it might be expected that bilateral exports, X~j, would also be an important determinant of trade in banking. This hypothesis would be supported by the results of Goldberg et al. (1989) who found that the foreign banking presence in different states of the USA is strongly related to state imports. Among nations the imports of country j from country i are equivalent to the bilateral exports X~j. Therefore X~j is used as an independent variable to explain banking exports, BX~j. Since it has often been hypothesized that banks follow their multinational clients overseas, the location of multinational corporations, MNCij, is also included as an independent variable to test its influence on international banking. A positive coefficient for MNC~j is anticipated. Regulation has often been suggested as an important factor for trade in banking services. Therefore a variable denoting the restrictiveness of regulation with regard to the establishment of foreign banks by the host country,j, RESTRICj, is included in the equation. The coefficient for the restrictiveness measure is expected to be negative. On the other hand certain countries encourage the immigration of foreign banks to promote their position as overseas banking centers. Although such policies might be considered under the RESTRICj variable described above, it is felt that an additional dummy variable, OBCj, would be helpful in the explanation of international banking. Finally, because of the emphasis placed on the foreign exchange and capital markets and the progress that the Organisation for Economic Co-operation and Development (OECD) has achieved in the liberalization of capital movements (OECD, 1990) a dummy variable is introduced to indicate intra-OECD trade in banking. It is anticipated that the coefficient will be positive to reflect an extensive network of banks linking the different OECD countries.

Journal of International Money and Finance 1995 Volume 14 Number 1

51

Trade in bankin9 services: J ter Wenoel

The equation to be estimated is the following:

BX~j=~o(Nl~l( yi~]J2(tlNij)[~3Y~i4Yffj5

(2)

\Nj] \yjJ X~6MNC~rRESTRIC~O BC~"OECD~°.

The model will be applied to the various forms of bank representationlbranches, representatives and subsidiaries--to investigate whether banks choose different forms of bank representation according to the bilateral economic relationships and the regulatory restrictions.

II. Empirical analysis of the pattern of international banking

II.A. The data requirements The purpose of this section is to estimate the model presented in the last section on the basis of bilateral data on trade flows in banking services among 141 countries in order to establish the determinants of international banking exports. The data set employed to achieve this task is described in this subsection. In the analysis of trade in banking services among 141 countries attention is limited to the number of branches, subsidiaries and representatives from a given country i in another country j. Although data on numbers, just like data on assets, represent stocks, they are used as a proxy for banking service flows because there is no adequate body of data on trade in banking services, partly because 'banks do not charge their customers directly but recoup themselves in the form of factor income. '5 Data on assets is not used as an alternative because such data are also unavailable. A branch is defined as an integral part of the banking organization and constitutes the highest level of foreign representation a bank may attain since foreign branches may accept deposits, make loans and provide a full line of banking services in the host market on an equal footing with domestic banks. At the other extreme, although representatives are also an integral part of the parent bank, representatives are barred from taking deposits or making loans and in theory have little more than a public relations function. In contrast to branches, subsidiaries are incorporated in their respective countries and form no integral part of the parent bank. The availability of data from The Bankers' Almanac (1990) determined the examination of trade in banking among 141 countries, listed in Table 1, that have or host foreign bank representatives. 6 The Bankers" Almanac provides information on more than 3600 major banks which operate internationally. Banks that do not operate internationally to any great extent or do not offer a complete line of banking services are generally not included. Given the data on the locations of branches, subsidiaries, and representatives, three 141 by 141 bilateral trade matrices were constructed. It must be emphasized that these matrices describe both the direct and indirect connections between the parent banks and their

52

Journal of International Money and Finance 1995 Volume 14 Number 1

Trade in bankinO services." J ter Wenoel

TABLE 1. List of the countries studied. ~ Afghanistan Algeria Andorra Angola Argentina Australia Austria Bahamas Bahrain Bangladesh Belgium Belize Bermuda Bhutan Bolivia Botswana Brazil Brunei Burkina Faso Burundi Windward Leew Isl. Cameroun Canada Cayman Islands Central African Rep.

Channel Islands Chile Colombia Comores Archipelago Congo Costa Rica Cote d'Ivoire Cyprus Denmark Djibouti Dominican Republic Ecuador Egypt El Salvador Ethiopia Falklands Finland France Gabon Gambia Germany Ghana Gibraltar Greece Guatemala

Guinea Equatorial Guinea Guinea-Bissau Guyana Haiti Honduras Hong Kong India Indonesia Iran Iraq Ireland Israel Italy Jamaica Japan Jordan Kenya Korea Kuwait Lebanon Lesotho Liberia Libya Liechtenstein

Luxembourg Macao Malaysia Maldives Mali Malta Morocco Mauritania Mauritius Mexico Monaco Namibia Nepal Netherlands Netherlands Antilles New Zealand Nicaragua Niger Nigeria Norway Oman Pakistan Panama Papua New Guinea Paraguay

Peru Philippines Portugal Puerto Rico Qatar Reunion Rwanda Saudi-Arabia Senegal Seychelles Sierra Leone Singapore Somalia South Africa Spain Sri Lanka Pacific Ocean Isl. Sudan Surinam Swaziland Sweden Switzerland Syria Taiwan Tanzania

Thailand Togo Trinidad and Tobago Tsjaad Tunisia Turkey United Arab Republic Uganda UK Uruguay USA Venezuela Yemen Zaire Zambia Zimbabwe

Please refer to note number 7 for further details.

Journal o f International Money and Finance 1995 Volume 14 Number 1

53

Trade in bankin9 services: J ter Wenoel

ultimate forms of foreign representation. The Appendix provides details on the collection of data. The data on G N P were derived from World Bank, Worm Tables, and International Monetary Fund, International Financial Statistics. The data on G N P refer to G N P in current prices and valued in US dollars at the current exchange rate. The year 1989 was chosen for data reasons and to allow a lag between the attainment of a particular threshold level of G N P and the decision to establish any form of bank representation. The same World Bank and IMF sources were used to obtain 1989 data on population. Bilateral trade data for 1989 were obtained from the International Monetary Fund, Direction of Trade Statistics: Yearbook 1990. The data on foreign subsidiaries of multinational corporations (MNCs) were derived from John M. Stopford, The Worm Directory of Multinational Enterprises: 1982-1983. Unfortunately the data refer to the 500 largest MNCs only and therefore leave out many bilateral connections that might be important. However, it proved impossible to find a more extensive and up-to-date data source. The data on bilateral regulation posed another almost insurmountable problem. The measure of regulatory restrictiveness for the foreign establishment of US banks developed by Tshoegl (1981) was used as a proxy for bilateral restrictiveness. 7 The offshore banking centers (OBCs) were taken to be the following: Bahamas, Bahrain, Windward and Leeward Islands, Hong Kong, Cayman Islands, Channel Islands, Netherlands Antilles, Panama, Singapore and the Pacific Ocean Islands. A general evaluation of the data indicates that there is much room for improvement. It would be especially desirable to get more up-to-date data on regulatory restrictiveness and the bilateral establishment pattern of MNCs.

ll.B. The determinants of trade in bankin9 services A review of the matrices constructed to depict the international flows of banking services show that these services are not universally traded but that trade in these services is limited to a modest number of trade links for the majority of countries. If all countries traded with each other we would find 19 740 bilateral trade relationships among the 141 countries considered. An examination of the trade matrix constructed for branches indicates that of the possible 19 740 links only 740 indicate positive trade flows: the remaining 19 000 cells are empty. Representatives exhibit 752 positive trade flows out of the same 19740 possible bilateral connections. Subsidiaries show 451 positive trade flows. The analysis of the connections between countries and the great number of zeroes in the bilateral trade matrix indicate the need to use a logit or a probit regression to estimate equation ( 2 ) rather than ordinary least squares. The case for the utilization of a logit or a probit procedure is further supported by the fact that the majority of bilateral connections represent the activities of one or two branches, subsidiaries or representatives. As can be seen from Table 2, among branches, 274 of the 740 connections are established with a single branch; one subsidiary forms the link in 42.8 percent of the cases; and in the case of representatives, 48.7 percent of the entries refer to just one representative. 54

Journal of International Money and Finance 1995 Volume 14 Number 1

Trade in bankin9 services." J ter Wen9el

TABLE2. The reduced number of bank officesthat establish many country connections. Branches 1 2 3 4 5

Frequency

Percent

Cumulative frequency

Cumulative percent

274 128 78 59 23

37.0 17.3 10.5 8.0 3.1

274 402 480 539 562

37.0 54.3 64.9 72.8 75.9

193 253 290 321 331

42.8 56.1 64.3 71.2 73.4

366 504 582 624 649

48.7 67.0 77.4 83.0 86.3

Subsidiaries 1 2 3 4 5

193 60 37 31 10

42.8 13.3 8.2 6.9 2.2

Representatives I 2 3 4 5

366 138 78 42 25

48.7 18.4 10.4 5.6 3.3

Concentrating on the explanation of bank connections among countries, Table 3 summarizes the results of the logit estimation of equation (2). A first impression that might be glimpsed from the more significant coefficients of Table 3 is that the coefficients for all types of bank offices are fairly similar, suggesting that banks may consider branches, subsidiaries and representatives as substitutes. Although this might be true to a certain extent, a closer look at the coefficients presented in Table 3 also reveals some striking differences. However, before analyzing the differences as well as the common elements it might be useful to dwell on the fit of the model and the significance of the results. In Table 3 the standard errors are presented below the corresponding coefficients. Although not shown in the table in addition to the standard error the Wald chi-square statistic was computed for each parameter estimate. The level of significance of this statistic is presented next to each parameter estimate by three, two and one asterisks corresponding to the 0.01, 0.1 and 1 percent levels of significance, respectively. At the bottom of Table 3 the sensitivity, specificity and correct rate for each test are presented. The sensitivity measure reflects the number of times that the model correctly predicts the occurrence of an event and is given by the ratio of predicted events to total events. In the three tests presented in Table 3, the sensitivity statistic indicates the ratio of the predicted bank connections to the actual bank connections. The specificity rate presents the ratio of accurately Journal of International Money and Finance 1995 Volume 14 Number 1

55

Trade in banking services: J ter Wengel TABLE 3. Logit estimation of the factors that determine the export of banking services. Branches

Subsidiaries

Coeff. Level (std. error) of sign. 1

Coeff. Level (std. error) of sign. 1

INTERCPT

-7.78 0.55

***

-6.74 0.64

Yl/Yj

-0.10 0.03

**

***

0.02 0.03

0.25 0.03

***

0.22 0.08

-0.14 0.05

0.07 0.06

Yi

0.40 0.03

***

0.26 0.03

Yj

0.08 0.03

*

0.06 0.03

X0

0.27 0.01

***

0.11 0.01

MNCij

0.30 0.05

***

- 1.29 0.13 2.76 0.19

RESTRICj OBCj OECD SENSIT. SPECIF. CORR-RT. # OF 1 # OF 0

-0.03 0.22 27.6 99.5 96.8 740 19,000

Coeff. Level (std. error) of sign. 1 - 13.56 0.68

LINi~

***

Representatives

0.33 0.03

***

0.59 0.03

***

***

0.23 0.02

***

0.47 0.06

***

0.41 0.06

***

***

- 1.10 0.16

***

-0.19 0.10

***

2.81 0.22

***

2.93 0.26

2.39 0.24

***

0.45 0.20

23.1 99.7 98.0 451 19,289

***

***

38.8 99.5 97.2 752 18,988

1Level of significance based on a chi-square distribution: *** significant at the 0.01% level; **, significant at the 0.1% level; and * significant at the 1.0% level.

f o r e c a s t z e r o e s to t h e o b s e r v e d l a c k o f b a n k i n g c o n n e c t i o n s . F i n a l l y , t h e c o r r e c t r a t e s t a t i s t i c is t h e r a t i o o f t h e n u m b e r o f c o r r e c t p r e d i c t i o n s to t h e t o t a l n u m b e r of observations. U n f o r t u n a t e l y t h e r e is n o single s t a t i s t i c a n a l o g o u s t o t h e coefficient o f d e t e r m i n a t i o n , t o j u d g e t h e fit o f a l o g i t r e g r e s s i o n . T h e r e f o r e t h e fit h a s t o b e j u d g e d o n a m o r e q u a l i t a t i v e basis, d e p e n d i n g o n t h e o b j e c t i v e s o f t h e e s t i m a t i o n . Since t h e p u r p o s e o f t h e i n v e s t i g a t i o n is t o d e t e r m i n e t h e f a c t o r s t h a t e x p l a i n t r a d e in b a n k i n g services a m o n g n a t i o n s , it is a p p a r e n t t h a t t h e c o r r e c t e x p l a n a t i o n of trade and no trade should be taken to be the objective of the estimation and t h e h i g h c o r r e c t r a t e s o b t a i n e d w o u l d i n d i c a t e a g o o d fit. 56

Journal of lnternational Money and Finance 1995 Volume 14 Number 1

Trade in bankiny services." J ter Wengel

Nevertheless, it should be recognized that the model presents rather low sensitivity rates: 27.6, 23.1 and 38.8 percent for branches, subsidiaries and representatives, respectively. Although the ideal would be numbers close to 100 percent, I do not consider the results discouraging. Taking branches as an example, the model correctly pinpoints 205 of the 740 bank branches in a field that presents 19 740 possibilities. If the 740 branches had been randomly assigned, only 28 of the actual connections would be identified. Having analyzed model fit, it also seems appropriate to dwell on the question of any possible biases that might have been introduced by the utilization of banking presence as a proxy for the delivery of banking services. There are two sources of possible bias: the number of foreign banking offices--branches, subsidiaries and representatives--may not be a good proxy for the flow of banking services; and, the bias resulting from the simplification of the analysis of presence to a one-zero exercise. The interaction of the two sources of bias is such that unless a relationship between flow of services and number of establishments is considered completely untenable, my working with a dichotomous variable for banking representation makes unnecessary any further concern over the relationship, or possible bias between service flows and banking presence. With regard to the second source of bias, resulting from our aggregation of one or more banking offices into a single category, it is possible to test for this bias by estimating an ordered response model. For branches, subsidiaries and representatives ordered response logits were estimated with five categories: no offices, one office, two or three offices, four or five offices, between five and ten offices and more than ten offices. The results are presented in Table 6 in the Appendix and indicate no substantial change in the value of the estimated coefficients. This result is not surprising given the significant number of connections that are established with only one branch, subsidiary or representative. It may also be feared that some of the dummies or the restrictiveness variable, employed as explanatory variables, might bias the estimated coefficients or might otherwise account for the adequate fit of the model. A look at the variance-covariance matrix (not reproduced here) does not indicate serious multicollinearity problems that might lead to biased coefficients. Moreover, for the case of branches, runs without the explanatory dummies and the restrictiveness variable, presented in the Appendix in Table 7, show only moderate changes that could only lend further support to the arguments put forth in the analysis of Tables 3 and 5. Table 3 indicates that the following explanatory factors turned out to have the expected sign at a high level of significance (0.01 percent): 1. Within the group of variables used to test for the relevance of the different theories of international trade, the coefficient for the economic size of the exporting country was positive and highly significant in the three cases. 2. With respect to the variables associated with the functions of banks, servicing exports and providing financial services for compatriot multinational corporations, the coefficients again turned out to be positive and highly significant. Journal of lnternational Money and Finance 1995 Volume 14 Number 1

57

Trade in bankino services: J ter Wenoel

3. With respect to the group of regulatory measures, the positive and significant coefficient for overseas banking centers points to the attractiveness of these centers. The negative and significant coefficient for the restrictiveness of regulation of the host country shows the deterrent effect of this regulation except in the case of representatives. These results are important because they indicate that: regulation has an important restraining impact on the trade of international banking services; exports and the presence of compatriot multinational corporations foster the establishment of bank offices abroad; and economic size and economies of scale are an important factor in the explanation of international trade in banking services. With regard to the last conclusion, it might be argued that the positive coefficients regarding country size could indicate that large countries have a greater number of independent banks that might want foreign offices. However, this argument does not appear to be supported by the data. Data on foreign representation indicate that only a limited number of banks per country control the majority of foreign offices as is illustrated on the basis of branches in Table 4. The majority of country links are established with one or two offices as can be seen from Table 2. As noted above, there are also important differences in the estimated coefficients for the three forms of banking represented in Table 3. Because branches, subsidiaries and representatives have different functions and because the tenure of these different forms of bank representation implies different degrees of involvement by the parent banks, the comparison of the differences in the coefficients might reveal new insights into the pattern of international banking and might provide additional evidence to support or reject the conclusions drawn

TABLE 4. The principal bank branch exporting countries, the number of countries in which they are represented and the concentration of branches held by banks.

1

2 3 4 5 6 7 8 9 10

58

Exporting country

No. of countries with representation

No. of banks that control 50% of foreign branches

USA UK France Netherlands Canada West Germany India Pakistan Hong Kong Brazil

72 54 46 35 30 28 26 26 23 22

4 2 3 1 1 3 2 1 1 2

Journal of International Money and Finance 1995 Volume 14 Number 1

Trade in bankin9 services: J ter Wenoel

above. The review of Table 3 presents the following salient results: 1. The coefficient for the income term of the destination country is positive and significant for representatives. This result indirectly lends support to the economies of scale argument: if the destination country is large, as measured by a large Y~, then it probably enjoys a comparative advantage that would discourage other countries from seeking any form of representation beyond the minimum afforded by representatives. 2. The relative incomes per capita variable, Yi/Yj, shows a negative coefficient for branches and a positive one for representatives. Since branches and representatives depend directly from the parent company this means that, all other things being equal, banks tend to send branches to the higher income per capita countries and representatives to the poorer countries. 3. A final result is the importance of the OECD market to subsidiaries in contrast to its lack of relevance to branches and representatives. This result is surprising in view of the importance of the trade and financial links among the O E C D countries. This last and apparently paradoxical result can probably be explained by the continued efforts of the O E C D to facilitate capital movements among the member countries. The progress achieved in the pursuit of this goal already permits a great deal of freedom for the provision of services across national boundaries. This freedom implies that an enterprise in a given country may have access to the capital market of another through the banks in the latter country. The significance of this freedom for banks is that it is no longer necessary to seek direct representation in other OECD countries in the form of branches or representatives. The importance of OECD membership for the location of subsidiaries may be explained as an investment alternative as suggested by Heinkel and Levi. The above inferences may be analyzed in greater detail by an econometric comparison of the presence of branches with respect to subsidiaries and representatives and that of subsidiaries with respect to representatives. By means of a number of logit regressions that have a dependent variable representing one form of representation or another, the comparative influence of the different factors on the choice of bank offices is examined. The results are presented in Table 5. In the statistical analyses presented in Table 5 those cases where both types of bank offices were present were excluded. The examination of the logit of branches with respect to subsidiaries confirms that: the O E C D area is a very important market for the location of subsidiaries with respect to branches; exports are very important to the establishment of branches with respect to subsidiaries; subsidiaries rather than branches would play an important role in the service of compatriot multinational corporations, perhaps because the financing of an MNC by a subsidiary might not attract as much criticism; the size of the exporting country and by implication economies of scale are more important for the establishment of foreign branches than subsidiaries. Small countries might resort to subsidiaries more than the large countries as suggested by Heinkel and Levi. The comparison of branches and representatives highlights the fact that they Journal o/' International Money and Finance 1995 Volume 14 Number 1

59

Trade in banking services: J ter Wengel TABLE 5. Comparison of the factors that influence the location of branches, subsidiaries and representatives. Branch/subsidJ Coeff. Level (std. error) of sign. 2 INTERCPT

-2.97 0.96

yJy~

*

Branch/repres. Coeff. Level (std. error) of sign. 2

Subsid./repres. t Coeff. Level (std. error) of sign. 2

5.79 0.97

***

8.68 1.18

***

-0.09 0.05

-0.28 0.05

***

-0.27 0.07

**

LIN~

0.03 0.12

0.17 0.09

0.17 0.13

Y~

0.22 0.06

**

0.01 0.05

-0.14 0.07

Yj

0.14 0.05

*

-0.43 0.05

X/~

0.12 0.02

***

-0.64 0.06

***

0.04 0.02

-0.10 0.03

**

- 0.01 0.09

0.31 0.10

*

-0.90 0.25

**

MNCi)

- 0.34 0.10

RESTRICj

-0.22 0.23

- 1.33 0.22

0.78 0.42

0.27 0.39

-0.25 0.43

-0.74 0.35

1.79 0.40

OBC~ OECD

-2.46 0.39

SENSIT. SPECIF. CORR-RT. # OF 1 # OF 0

90.3 33.3 73.2 505 216

*

***

***

65.9 73.9 69.9 428 440

***

***

93.1 57.7 82.1 246 547

1The bank office in the numerator is assigned the value of 1 and the other 0. 2Level of significance on the basis of a chi-square distribution: *** significant at the 0.01% level; ** significant at the 0.1% level; and * significant at the 0.1% level.

a r e b o t h d i r e c t l y d e p e n d e n t f r o m a p a r e n t b a n k a n d t h a t t h e difference lies in the greater importance of branches with respect to representatives. Thus, neither t h e size o f t h e e x p o r t i n g c o u n t r y , b i l a t e r a l e x p o r t s , t h e p r e s e n c e o f c o m p a t r i o t m u l t i n a t i o n a l c o r p o r a t i o n s o r O E C D m e m b e r s h i p e x p l a i n t h e differences in t h e location patterns of branches and representatives. The results obtained show that: b a n k s find it e a s i e r t o e s t a b l i s h r e p r e s e n t a t i v e s r a t h e r t h a n b r a n c h e s a b r o a d ; banks tend to send branches to the richer countries and representatives to the p o o r e r ones; b a n k s s e n d l o w c o s t r e p r e s e n t a t i v e s r a t h e r t h a n h i g h c o s t b r a n c h e s t o l a r g e c o u n t r i e s w i t h e c o n o m i e s o f scale in b a n k i n g . 60

Journal of International Money and Finance 1995 Volume 14 Number 1

Trade in banking services: J ter Wengel

The analysis of the location patterns of subsidiaries with respect to representatives indicates: banks tend to employ representatives rather than subsidiaries in large destination countries; banks tend to have subsidiaries rather than representatives in the OECD area; banks tend to use subsidiaries in the richer countries and representatives in the poorer countries.

III. Conclusions

The analysis of the preceding section points to the fact that the size of the country of origin is important in the determination of the export of bank services, thus substantiating the hypothesis that the newer theories of international trade, that stress economies of scale, would be the most useful in the explanation of international trade in banking. None of the other theories tested appears to have significant explanatory power. To establish the economies of scale hypothesis test it was supposed that economies of scale could be derived from the operation of foreign exchange markets. This supposition is supported by the empirical findings of the last section that indicate that destination country size is an important factor in the determination of the form of representation and that branches are established primarily by the large countries. Thus, banks from big countries would tend to establish branches abroad unless: the destination country is also large and similarly competitive; the destination country has regulations that hinder the establishment of branches; the freedom of capital movements and the absence of foreign exchange regulations erode the large countries' comparative advantage. Subsidiaries are employed as an investment alternative in countries with liberal capital regulations and high incomes per capita. Representatives are used in situations where: the destination country appears to have a size determined comparative advantage; regulations do not permit higher forms of representation; or, profit opportunities may be limited because of a low income per capita. These results fit a pattern that is similar to that found by Heinkel and Levi (1992) suggesting that a model like theirs could also be utilized in a broader global framework. The coincidence of results indicates progress in the understanding of the determinants of international banking and suggests that further analysis along the lines sketched by both papers might prove highly useful.

Appendix: Definitions adopted for the collection of data on banks The following definitions were adopted to assign banking offices to the different forms of bank representation--branches, subsidiaries and representatives--and to the ultimate parent countries. The term 'branches' comprises branches, sub-branches, agencies and sub-agencies and only applies to the direct foreign branches of an ultimate parent bank. The term 'subsidiary' applies to banks in which a foreign majority shareholding exists and is concentrated in the hands of Journal of International Money and Finance 1995 Volume 14 Number 1

61

Trade in bankin9 services: J ter Wengel TABLE 6. Ordered response logit estimation of the factors that determine the export of banking services. Branches

Subsidiaries

Coeff. Level (std. error) of sign?

INTERCPT Yi/Yj

-

Coeff. Level (std. error) of sign. 1

10.89 0.57

***

-8.68 0.64

-0.11 0.03

**

0.02 0.03

L1Ni~

0.09 0.06

Y/

0.42 0.03

Yj

Representatives

***

0.22 0.08

*

***

0.26 0.03

***

0.11 0.03

**

0.04 0.03

X/j

0.28 0.01

***

0.12 0.02

MNCij

0.32 0.05

***

-1.33 0.13 2.82 0.19

RESTRICj OBCj OECD INT-LVL2 INT-LVL3 INT-LVL4 INT-LVL5

- 18.40 0.70

***

0.27 0.03

***

- 0.12 0.05 0.36 0.03

***

0.63 0.03

***

***

0.24 0.02

***

0.43 0.05

***

0.38 0.05

***

***

-1.11 0.16

***

-0.15 0.10

***

2.70 0.22

***

3.04 0.25

2.43 0.24

***

0.32 0.18

0.51 0.86 1.46 2.22

0.07 0.09 0.10 0.11

1.02 1.72 2.97 4.07

-0.06 0.20 0.77 1.28 2.09 2.78

Coeff. Level (std. error) of sign?

0.08 0.09 0.I0 0.11

***

0.13 0.15 0.16 0.17

t Level of significance based on a chi-square distribution: *** significant at the 0.01% level; ** significant at the 0.1% level; and * significant at the 1.0% level.

one or more institutions from the same country. If the foreign subsidiary bank has foreign offices these are regarded as subsidiaries of the ultimate owner if the ultimate owner has a majority shareholding in the subsidiary of the subsidiary. The term 'representatives' comprises representatives and advisors. A representative of a foreign subsidiary is regarded as a representative of the ultimate parent. If a group of banks from one country have a joint representative then the country is assigned only one representative. If the group involves parent banks with different nationalities then each of the origin countries is assigned one representative. The set of choices delineated above obviously influences the character of the data. Different definitions would have led to different data sets that might have performed better or worse empirically. The special feature of the data set constructed is that it captures the ownership and control linkages of international banking networks. 62

Journal of International Money and Finance 1995 Volume 14 Number 1

Trade in banking services." J ter Wengel

TABLE 7. Analysis of the contribution of the O E C D and OBC dummies and the restrictiveness variable in the Iogit for branches. Coeff. Level (std. error) of sign)

Coeff. Level (std. error) of sign)

Coeff. Level (std. error) of sign?

INTERCPT

-7.77 0.55

***

-5.85 0.51

***

-6.68 0.50

***

5'i/.vj

-0.10 0.03

**

-0.18 0.02

***

-0.23 0.02

***

LINi~

0.07 0.06

L

0.40 0.03

***

Vj

0.08 0.03

*

0.27 0.02

***

0.30 0.02

***

0.32 0.02

***

0.30 0.05 1.29 0.13

***

0.33 0.05 - 1.45 0.13

***

0.33 0.05

***

2.76 0.19

***

MNC~j

RESrRICj OBQ

-

0.04 0.06

***

0.39 0.03

0.14 0.06 ***

- 0.04 0.03

0.35 0.03 - 0.07 0.03

*** *

***

OECD

SENSIT. SPECIF. CORR-RT.

27.4 99.5 96.8

23.4 99.6 96.7

18.8 99.6 96.6

1Level of significance based on a chi-square distribution: *** significant at the 0.01% level; ** significant at the 0.1% level; and * significant at the 1.0% level.

Notes

1.

2. 3. 4. 5. 6.

7.

This n u m b e r includes all countries that have or host any form of foreign banking offices with the exception of the formerly Communist countries. In Section II.A a more detailed description of the data is presented. Because of data restrictions it was not possible to employ the forms of representation suggested by Heinkel and Levi (1992). I refer to the Linder Hypothesis as the Linder theory to facilitate the exposition. Learner (1974), p. 374. Arndt (1988), p. 63. The following islands were aggregated as the Windward and Leeward Islands: Anguilla, Bequia, Carriacou, Nevis, Montserrat, Antigua and Barbuda, Dominica, St. Kitts, St. Lucia, Grenada, St. Vincent, St. Pierre and Miquelon, Martinique, Guadeloupe, Barbados, Turks and Caicos and the Virgin Islands. Similarly, Kiribati, Fiji, Solomon Islands, Vanuatu, Tonga, American and Western Samoa, Guam, French Polynesia, Nauru, New Caledonia, Mariana Islands and Tuvalu were aggregated into the Pacific Ocean Islands. The inverse of Tschoegl's ease of regulation index was used.

Journal of International Money and Finance 1995 Volume 14 Number 1

63

Trade in banking services: J ter Wengel

References ARNDT, H.W., 'Comparative Advantage in Trade in Financial Services,' Banca Nazionale del Lavoro Quarterly Review, March 1988, 51: 61-78. DEARDORF, ALAN V., 'Testing Trade Theories and Predicting Trade Flows,' in Ronald W. Jones and Peter B. Kenen, eds, Handbook of International Economics, North Holland Publishing Co.: Amsterdam, 1984. FERRUFINO, ADRIANA,'Reestimaci6n y Ampliaci6n de la Evidencia sobre las Econom}as de Escala en el Sistema Fianciero Colombiano,' Ensayos Sobre Politica Economica, Colombia: Banco de la Repflblica, June 1991, 19: 69-98. GOLDBERG,LAWRENCEG. AND DENISEJOHNSON,'The Determinants of US Banking Activity Abroad,' Journal of International Money and Finance, June 1990, 9: 123-127. GOLDBERG, MICHAEL A., ROBERT W. HELSLEY AND MAURICE O. LEVI, 'The Location of International Financial Activity: An Interregional Analysis,' Regional Studies, February 1989, 23: 1-7. GOLDBERG, MICHAEL A., ROBERT W. HELSLEY AND MAURICE D. LEVI, 'The Growth of International Financial Services and the Evolution of International Financial Centres: A Regional and Urban Economic Approach,' in P.W. Daniels, ed, Services and Metropolitan Development, Routledge: London, 1991. HEINKEL,ROBERTL. AND MAURICED. LEVI,'The Structure of International Banking,' Journal of International Money and Finance, June 1992, ll: 251-272. HULTMAN, CHARLESW. AND L. RANDOLPH MCGEE, 'Factors Affecting the Foreign Banking Presence in the U.S.,' Journal of Banking and Finance, July 1989, 13: 383-396. INTERNATIONALMONETARYFUND, International Financial Statistics (various issues). INTERNATIONALMONETARYFUND, Direction of Trade Statistics: Yearbook 1990. LEAMER, EDWARD E., 'The Commodity Composition of International Trade in Manufactures: An Empirical Analysis,' Oxford Economic Papers, November 1974, 26: 350-474. NELSON, RICHARD W. 'Branching, Scale Economies, and Banking Costs,' Journal of Banking and Finance, June 1985, 9: 177-191. ORGANISATIONFOR ECONOMIC CO-OPERATIONAND DEVELOPMENT,Liberalisation of Capital Movements and Financial Services in the OECD Area, OECD: Paris, 1990. REED, HOWARD C., The Preeminence of International Financial Centers, Praeger Publishers: New York, 1981. REED INFORMATION SERVICES, The Bankers' Almanac, Reed Information Services: East Grinstead, England, 1990. STOPFORD, JOHN M., The World Directory of Multinational Enterprises: 1982-1983, Macmillan, London, 1982. TSCHOEGL, ADRIAN E., The Regulation of Foreign Banks: Policy Formation in Countries Outside the United States, New York University Graduate School of Business Administration: The Monograph Series in Finance and Economics, New York, 1981. WORLD BANK, Worm Tables (various issues). YOUNG, ALLYN A., 'Increasing Returns and Economic Progress,' The Economic Journal, December 1928, 38: 527-542.

64

Journal of International Money and Finance 1995 I/olume 14 Number 1