THE PATTERNS OF CROSS-BORDER BANK MERGERS

AND SHAREHOLDINGS IN OECD COUNTRIES

by Dario Focarelli* and Alberto Franco Pozzolo*
Abstract
In recent years there has been a surge in mergers and acquisitions in all sectors of economic
activity. Cross-border operations have followed this trend, creating global companies
operating in all major world markets. In this paper we study the pattern of cross-border
M&As in the banking industry relative to the non-financial sector of the economy and
investigate which factors make it more likely that a bank will expand its activities abroad. We
find that cross-border M&As are rarer in banking than in other sectors, possibly owing to the
importance of information asymmetries in banking relationships and to regulatory restrictions.
Using data on almost 2,500 banks from 29 OECD countries, we also show that the most
significant features of banks with foreign equity interests relate to efficiency: banks with
cross-border shareholdings are on average larger, more profitable, and based in countries with
a more highly developed banking market.

Keywords: Foreign direct investment, International banking, Cross-border M&As.

JEL-classification: F21, F23, G21, G34.

* Bank of Italy, Research Department. Part of the analysis presented in this paper draws on a previous work
that was circulated under the title The Determinants of Cross-Border Bank Shareholdings: an Analysis with
Bank-Level Data from OECD Countries. For comments on the previous versions, we thank Jason Abrevaya,
Allen Berger, Claudia Buch, Nicola Cetorelli, Robert DeYoung, Michele Gambera, Hesna Genay, Giorgio
Gobbi, Luigi Guiso, Anil Kashyap, George Kaufman, Fabio Panetta, Anthony Saunders and seminar participants
at the Bank of Italy, at the Federal Reserve Bank of Chicago, at the Finance Brown Bag Lunch of the University
of Chicago, at the 2nd Kiel Workshop in Economics, at the 36th Annual Conference on Bank Structure and
Competition of the Federal Reserve Bank of Chicago, at the 2000 European Economic Association Congress and
at the XIII Australasian Finance and Banking Conference for comments and suggestions. All remaining errors
are of course our own responsibility. Opinions expressed are those of the authors and do not necessarily reflect
those
of
the
Bank
of
Italy.
E-Mail:
focarelli.dario@insedia.interbusiness.it;
pozzolo.albertofranco@insedia.interbusiness.it.

Introduction
During the nineties the number and the value of mergers and acquisitions (M&As)

increased rapidly in virtually all sectors of economic activity. Firms reacted to the
intensification of competition in the internal and international markets by increasing their
scale of operations. M&As have been particularly frequent in the banking sector, thanks also
to widespread deregulation, which permitted the integration of financial activities such as
banking, asset management and insurance.
Following the rapid increase in international trade and cross-border financial
transactions, in all sectors of economic activity a large share of mergers and acquisitions
involved firms operating in different countries. In the banking industry, this has determined
an increase in both the number of acquisitions of foreign banks and the scale of activity of
foreign branches (see, for example, Berger, DeYoung, Genay and Udell, 2000).1
An enterprise can profitably extend its activities to foreign countries only if it has a
competitive advantage over local rivals: otherwise, local firms would crowd it out of the
market. Cross-border expansions follow a pattern that reflects the structure of the comparative
advantages of the home-country enterprises.2
Historically, the pattern of banks international shareholdings has followed that of the
economic integration between countries: banks extended their activities abroad in order to
1

The literature on international banking has considered mainly three kinds of operations abroad: loan
provision and asset and liability management with foreign counterparts, foreign branching, and the acquisition of
shareholdings in foreign banks (subsidiaries). Goldberg and Saunders (1981) analyze the different organizational
structures available to a bank that wants to expand its activities abroad (representative offices, agencies, branches
and subsidiaries) finding that, in practice, a bank has two options: it can open a new branch or it can buy an
equity share in a bank that is already operational. Commonly, foreign branches are used by parent banks to
support the activities of home-country clients who operate abroad and to operate in the leading financial centers
(Brealey and Kaplanis, 1996). To offer retail services to local residents or to engage in activities that are not
permitted to branches (deposit-taking and lending) banks prefer to operate through subsidiaries. Our analysis
focuses on the acquisition of foreign shareholdings, which is how banks commonly choose to expand abroad in
the retail sector.
2

Two main strands of theoretical literature have developed to explain the pattern in multinational
investment: the internalization theory, which stresses that the advantages of multinational enterprises derive from
the possibility of limiting the cost of market failures by carrying out a share of their transactions within the
boundaries of the firm, and the eclectic theory, which adds ownership-specific advantages (for example, access
to the endowments of the parent company at costs below market price) and location advantages (for example,
barriers to trade or institutional arrangements) to the incentives related to internalization (see Dunning, 1988, for
a review of these theories). The standard point of view in the banking literature is the eclectic theory (see, for
example, Williams, 1997). In this paper, we simply assume that the maximization of the stream of expected
profits determines a banks decision to expand abroad.

provide services to their home-country clients in international transactions; afterwards, with a

growing understanding of the foreign market (in particular of regulatory and institutional
aspects) and a developed network of relationships with local financial institutions, some
banks were induced to increase the range of their operations and provide services to the local
population too.
Although this account is broadly accurate, we believe that today the pattern of bank
international shareholdings has some characteristics that are peculiar to the industry, and that
the choice of expanding abroad depends on a wider range of factors than just the overall
degree of economic integration between countries.
In this paper we study what characterizes cross-border M&As in the banking industry
relative to the non-financial sector of the economy and inquire into the main factors
determining a banks decision to expand its activities abroad. Our contribution is twofold.
First, we compare the pattern of cross-border M&As for banks and for non-financial firms, an
exercise that we have not found in the previous literature. Second, in studying the
determinants of a banks decision to expand abroad, we use information on nearly 2,500
banks from the 29 OECD countries, a larger sample than those of previous studies on
international banking.
Our results show that during the nineties cross-border M&As in banking followed the
same pattern as in the non-financial sector but were significantly lower in intensity. This
confirms the view that banks are special, possibly because of the importance of information
asymmetries in banking relationships and of more stringent regulatory restrictions. Second,
we show that the probability of a banks having foreign shareholdings is determined not only
by the overall degree of economic integration between countries but by a broader range of
factors, most of which are measures of profitability and efficiency.
The paper is organized as follows. The next section briefly reviews the previous
literature and sets up the background for the empirical analysis. Section 3 describes the data
set. Section 4 presents the results of the empirical analysis on the pattern of cross-border bank
M&As. Section 5 presents the results of the econometric analysis on the characteristics of
banks that have cross-border equity interests. Conclusions are set out in the final section.

Theoretical background and previous evidence

The role of banks in channeling funds from savings to productive investment is

important, in particular when information asymmetries between investors and entrepreneurs

are sizeable. Banks are in fact better equipped than other financial intermediaries to solve
informational problems, by pooling the costs of state verification (Diamond, 1984) and by
creating long-term relationships (Petersen and Rajan, 1994).
However, the value of bank loans to an outsider may be difficult to assess if the original
lender has soft information on the borrowers activity.3 Morgan (2000) provides indirect
evidence of this, showing that there is less agreement among rating agencies in assessing bank
bonds than bonds of non-financial firms, probably because of the greater uncertainty over the
market value of loans and trading assets than of the assets held by non-financial firms. In the
case of cross-border investment the problems for outsider investors who have to judge the
value of a bank are even worse, due to the incomplete understanding of foreign markets and
the presence of cultural, linguistic and regulatory barriers. Such difficulties may lead to less
frequent international M&As in banking than in other sectors. Thus, the first hypothesis under
scrutiny here is that cross-border banking M&As are less frequent than similar operations in
other sectors.4
Regulatory restrictions on the activities that banks can undertake are also likely to affect
the pattern of internationalization, although, a priori, their effect is uncertain. On the one
hand, government supervision resolves some of the problems created by information
asymmetries (e.g., safety nets can almost eliminate the risk of bank runs) and, as noted by
Mishkin (2000), is often essential to ensure the safety of the whole banking system. By
making the relationship between banks and depositors more transparent, public control may
therefore lay the basis for a safer financial system, more able to expand its activities abroad.
3

This is one of the reasons why banks have a very fragile financial structure (Diamond and Rajan, 2000) and
are vulnerable to runs (Diamond and Dybvig, 1983).
4

In fact, the hypothesis that banks are less likely than non-financial firms to do cross-border M&As is at
odds with the view that firms in the service sector, producing non-tradables, may only exploit their competitive
advantages by expanding their activities, while those in the manufacturing sector also have the option of
exporting their final products. However, in the case of tradables the standardization of products and production
processes is likely to make it easier for foreign investors to export the know-how acquired in the local market.
This distinction applies in general to foreign investment in the tradables and non-tradables sectors, but it is even
more relevant in banking, where personal relationships are fundamental.

Further, banks operating in more highly regulated markets may have greater incentives to
expand their activities abroad to bypass restrictions. On the other hand, regulatory restrictions
may reduce competition, hence the efficiency and the international competitiveness of the
domestic banking system. Which effect prevails is therefore a matter for empirical analysis, a
second hypothesis to investigate.
The research on banking consolidation at domestic level has shown how in a substantial
share of mergers and acquisitions, larger and more efficient institutions tend to take over
smaller, less efficient ones, presumably to spread their expertise and operating procedures
over additional resources (see, for example, Berger, Demsetz and Strahan, 1999; Focarelli,
Panetta and Salleo, 1999). Consolidation may also enhance efficiency if diversification
improves the expected risk-return tradeoff.
These results provide the motivation for a third hypothesis: that banks capable of
profitably exporting their know-how to foreign countries will be among the largest and most
efficient in their home market, and that they will come from the most highly developed
financial systems.
Grosse and Goldberg (1991) provide some evidence consistent with this hypothesis,
showing a positive correlation between the number of foreign banks in the United States from
a given country and the development of that countrys financial sector. Moreover, Berger,
DeYoung, Genay and Udell (2000) show that it is not always the case that foreign banks are
less efficient than domestically owned institutions, as most previous studies had found. In
fact, they show that in a number of countries there is a subset of foreign banks that are more
efficient than local institutions. Most empirical studies on foreign banking have also found a
positive correlation between the size of banks and their degree of internationalization.5 This
may be due to increasing returns in the activities that characterize international banking. In
particular, banks with a large home-market share may be more likely to search for risk
diversification and new profit opportunities abroad. Moreover, the corporate customers of
5

Ball and Tschoegel (1982) show that foreign banks that are more committed to doing business in
California (i.e., those having a subsidiary) are larger than those that only have a branch. Tschoegel (1983)
analyzes the worlds 100 largest international banks in 1976, defined as those with at least one office abroad,
finding a positive correlation between the size of the bank and the degree of trans-nationality (measured by the
number of countries where it is present). Ursacki and Vertinsky (1992), studying a sample of Japanese banks
with investment in South Korea, find a positive relation between a banks asset size and the number of branches.
Williams (1996) finds similar results for a sample of Japanese banks with shareholdings in Australia. Williams
(1998) again finds similar results for a larger sample of foreign banks in Australia.

larger banks are generally larger and more internationally-diversified themselves, so such
banks have grater incentive to follow them abroad.
A related issue is non-interest income. It is an empirical regularity that the most
efficient banks in developed countries earn a smaller percentage of profits from traditional
activities and a larger share from off-balance-sheet operations (see, for example, Generale and
Gobbi, 1999). These banks have reduced the relative size of their traditional activities, not
because of lower productivity, but because of lower growth potential. In fact, these are also
likely to be the more efficient banks. However, an opposite force might offset this effect: offbalance-sheet activities normally require less frequent face-to-face meetings with clients, are
more standardized and, in general, imply less intense personal relationships; as a result they
can often be exported directly. So if banks with a larger share of revenues coming from offbalance-sheet operations were not interested in expanding their traditional banking business
abroad, but only their most advanced activities, we should find a negative correlation between
the share of revenues coming from non-traditional activities and the degree of
internationalization of the bank. Whether these two forces offset each other or whether one
prevails is the fourth question to be analyzed here.
Finally, as we mentioned before, it has been argued that banks extend their activities
abroad in order to provide services there to their home clients. It is therefore worth testing
whether the pattern of banks foreign direct investment is positively correlated with measures
of openness of the economy, such as the volume of trade flows and of foreign direct
investment.6
3

Data and summary statistics

3.1

Data on mergers
The data on mergers and acquisitions used in this work are from Thomson Financial-

Security Data Company Platinum, an international data set on corporate finance operations

This hypothesis has been confirmed by a large number of studies. In particular, the volume of bilateral
trade flows has been used by Goldberg and Saunders (1980 and 1981), Goldberg and Johnson (1990), Grosse
and Goldberg (1991), Brealey and Kaplanis (1996), Yamori (1998); the value of bilateral foreign direct
investment by Nigh, Cho and Krishnan (1986), Goldberg and Johnson (1990), Grosse and Goldberg (1991),
Sagari (1992), Brealey and Kaplanis (1996), Williams (1998), Yamori (1998), Miller and Parkhe (1998), Buch
(2000).

for a very large number of companies and countries. We have considered the total number of
mergers and acquisitions, including acquisitions of minority stakes, among the 29 OECD
countries. Country and sector refer to the bidder.
Tables 1 and 2 show that the number of mergers and acquisitions in the OECD countries
increased rapidly during the nineties in all sectors of economic activity. Relative to total
M&As, the number of cross-border operations was higher at the beginning of the decade,
when the consolidation process was slower; the proportion then decreased, to recover only in
1998-99. The share of cross-border in total M&As varies significantly across sectors. In the
period 1990-1999, it was 12.9 per cent in the banking industry. In the entire non-financial
economy it was 29.6 per cent, with a peak of 35.3 per cent in manufacturing, the most
internationalized sector.

3.2

Data on banks
All bank-specific data are taken from Bankscope, an international data set of balance

sheet items on individual banks, where all the main information on assets, liabilities and
revenues is reported according to a common, comparable standard. The analysis is conducted
on a sub-sample of 2,449 banks with assets of more than $1 billion and with headquarters in
one of the 29 OECD countries. Foreign subsidiaries are included in the sample as autonomous
banks, while foreign branch activity is included in that of the parent bank. As is common in
the literature, we consider only the first level of foreign shareholdings.
In order to minimize the effects of particular events, all data on banks assets, liabilities
and revenues are averages of annual values from 1994 to 1997. Information on foreign
shareholders refers to the end of 1998. Some clear outliers are identified,7 so that in the end
only 2,148 banks are included. Further, we also consider a sub-sample of the 260 banks with
assets of more than $25 billion.
The distribution of the banks in the sample by size and by country is reported in table 3.
The United States has the largest number (488); Germany has 472; Italy, the United Kingdom,

In particular, 301 banks are excluded because they have negative interest margins, negative operating
income, net return on assets lower than 1 per cent or higher than 4 per cent, or overheads higher than total
operating income.

Japan and France, between 119 and 178 each; all the other countries have fewer then 100.
Japan has the highest number of large banks (56), followed by the United States (42),
Germany (33), the United Kingdom (22) and France (20). Of the 2,148 banks in the sample,
146 have cross-border shareholdings (6.8 per cent of the sample), while 276 have foreign
shareholders (12.8 per cent).
Panel A in table 4 reports some summary statistics for the 2,148 banks surveyed. Panel
B reports on the sub-sample of 260 large banks (assets of more than $25 billion). In both
samples, banks with foreign shareholdings are on average larger than other banks and have a
higher share of revenues from non-traditional activities. The returns on equity are similar, but
the return on assets is lower for larger banks, owing to lower net interest income.
3.3

Data on countries
Data on GDP, population, bank credit and inflation are from IMF, International

Financial Statistics (1998). Export data are from IMF, Direction of Trade Statistics (1998).
Stock market capitalization is from IFC, Emerging Stock Markets Factbook (1998) and
Datastream. Data on the efficiency of the judicial system are taken from La Porta, Lopezde-Silanes, Shleifer and Vishny (1998). Data on the level of regulatory restrictions on
domestic banking activity and on the relevance of state-owned banks are taken from Barth,
Caprio and Levine (2000). Finally, the indices on the level of restrictions on inward and
outward shareholdings in the banking sector have been built from IMF, Exchange
Arrangements and Exchange Restrictions (1997). All the variables considered have a high
degree of cross-country variability (table 5).
4

Cross-border M&As in the banking industry

The evidence on bank M&As in the OECD countries between 1990 and 1999, presented

in tables 1 and 2, is broadly consistent with the hypotheses put forward in section 2. In
particular, cross-border operations in banking have been less frequent than in the nonfinancial sector.
In order to test this hypothesis formally, we run a panel data regression for the OECD
countries between 1990 and 1999. We controlled for common shocks and idiosyncratic

factors by introducing time and country dummies.8 The following regression has been
estimated:

LRBit = ci + dt + LRNit + Zit + it,

(1)

where RBit is the ratio of cross-border M&As to total acquisitions in country i at time t in the
banking sector, LRBit = log[RBit /(1-RBit )]; LRNit is defined in the same way as LRBit and
refers to the non-financial sector; ci is the country i dummy; dt is the time effect for period t;
Zit is a vector of control variables (when included); it is a white noise residual. Data on the
ratios of cross-border to total M&As have been transformed to account for the fact that they
vary only between 0 and 1.9
The results (table 6, column A) strengthen the evidence of tables 1 and 2: the coefficient
of the ratio of the number cross-border to total M&As in the non-banking sector is 0.39. The
marginal effect, dRB/dRN, evaluated at the median levels, is equal to 0.23 and it is
significantly different from zero and 1 at the 5 per cent level.
Next, we tested the hypothesis that regulations affect the pattern of cross border M&As,
using the index of restrictions on banking activity calculated by Barth, Caprio and Levine
(2000). Because these data lack of time variability, we adopted the methodology suggested by
Blanchard and Wolfers (2000), testing whether common shocks affecting the share of
international M&As in the banking sector have an impact on each countrys operations that
depends on the prevailing level of regulatory restrictions to banks. In practice, we estimated
the following regression:

LRBit = ci + dt (1 + RESTi ) + LRNit + Zit + it,

(2)

Greene (1997) and Pesaran, Shin and Smith (1999), among others, argue that a fixed effect formulation is
to be preferred when the whole population of countries in a particular category is considered.
9

As is customary (see, e.g., Greene, 1997, p. 896), we have added (subtracted) a small constant to
observations equal to 0 (1).

where RESTi is the index of restrictions to banking activities in country i and the other
variables have already been defined.
The coefficient of the index of restrictions is negative and significantly different from
zero at the 5 per cent level (table 6, column B), confirming that part of the difference between
the share of international M&As in banking and in the other sectors of the economy is
explained by regulation in the banking industry. As a robustness check of the results obtained
with the specification in equation (2) we run a regression where the dependent variable is the
vector of fixed effects estimated from equation (1) and the regressor is the index of
restrictions to banking activities: as before we obtained a negative and significant coefficient
(table 6, column C).
This evidence is consistent with the view that more stringent regulatory restrictions, by
reducing the level of competition, lower the international competitiveness of the domestic
banking industry.
We also included a number of additional variables consistent with the hypotheses
discussed in section 2. The inclusion does not alter the estimates of the coefficients on the
share of cross-border M&As in the non-financial sector and on the level of regulatory
restrictions. The results, reported in table 7, show that the additional variables are not
significant, although in most cases they have the expected sign. In particular, only the
coefficient of the level of banks return on assets is close to significance, with a p-value of 13
per cent.
The evidence presented so far clearly shows that cross-border mergers and acquisitions
are less frequent in the banking sector than in the rest of the economy, and that the difference
depends partly on the level of regulatory restrictions. However, these results may suffer from
the fact that they are obtained from regressions on aggregated data. For this reason, the
empirical analysis on bank foreign equity interests presented in the next section is based on
individual bank data. This allows us to control for a number of idiosyncratic factors that are
likely to reduce the significance of estimates based on aggregated data. Moreover, in this way
it is possible to identify both bank-specific and country-specific characteristics affecting the
probability of a banks having a foreign equity interest. Unfortunately, these data only have

cross-section variability: information on the stock of foreign shareholdings for a large number
of banks from different countries is very difficult to obtain for a sufficiently long period.10
5

The pattern of cross-border bank shareholdings

5.1

Choice of explanatory variables

Following the discussion in section 2, we specify an empirical model hypothesizing that

banks that have foreign equity interests are larger, more efficient and have their headquarters
in countries with a more efficient banking system and a more open economy.
The measure of banks efficiency is a complicated issue, and it is crucial to correctly
identify the effects of banking consolidation (Berger, Demsetz and Strahan, 1999). In the
following, we consider profitability as a proxy for efficiency. This is justified by the fact that
in the econometric analysis we control for other market characteristics that may influence the
return on assets, such as market concentration, the banks average size, non-interest income,
or the availability of free cash flow.
We subsequently try to identify the sources of profit (again conditional on market
characteristics): the net interest income after charge-offs, which measures retail banking
profitability, and the cost-income ratio (overheads to total income), which is our proxy for
cost efficiency.
5.2

Econometric setup
A banks decision on whether to expand abroad is modeled as a binary choice, based on

all the available information used to forecast the expected profitability of the investment. We
look for an answer to the following question: which characteristics of a bank, and its country
of origin, make it more probable that it will hold an equity interest in a foreign bank?
Accordingly, we estimate the following binary choice model:

Pr (Yij = 1) = f (Xi,Zj),

(3)

10

One advantage of using cross-section information on stocks of foreign shareholdings rather than panel
data on flows of acquisitions during a specific period is that the latter could give a picture of the phenomenon
strongly biased by the initial conditions.

10

where: Yij equals one when the bank i of country j has a foreign subsidiary and zero otherwise,
Xi is a vector of bank-specific variables and Zj is a vector of country-specific variables. We
adopt a discrete choice model because in general it is difficult to infer the effective degree of
involvement in the operations of an invested bank from the share of capital controlled: in
some cases shareholdings of 10 per cent of capital may be sufficient to exert substantial
control of a bank; in others, an absolute majority is necessary.
We use a probit specification to estimate equation (3). This choice is the result of the
following specification search process. We first estimated a fixed effect and a random effect
logit (where the individual effect was calculated with respect to the country of origin) on the
same covariates, in order to test for independence of the random effect from the exogenous
variables. We prefer the random effect model, as it allows us to introduce country-specific
exogenous variables. The Hausman test does not reject the hypothesis that there is no
systematic difference between the coefficients obtained with the random and with the fixed
effect estimator. Then we estimated a random effect probit, including a number of countryspecific variables. However, the null hypothesis of absence of random effect was not rejected,
suggesting that the presence of a significant random effect in the logit regression was due to
the absence of country-specific variables. We therefore chose a probit specification for our
estimation.
5.3

Results from basic regressions

Table 8 reports the estimates of three specifications of equation (3). Of the 2,148

observations originally used, 56 are lost in the estimation presented in Panel A, because we do
not have data on total credit, stock market capitalization and exports for Luxembourg. In
panel B, 33 more observations, relative to the Czech Republic, Hungary and Poland are also
excluded, because we lack information on regulatory restrictions on banking activity in those
countries. Panel C considers the sub-sample of 257 banks with more than $25 billion of total
assets (excluding the 3 from Luxembourg). In all three specifications, both bank-specific and
country-specific characteristics are included as explanatory variables.11

11

Banks in the sample are classed as commercial banks, cooperative banks, medium and long-term banks,
real estate-mortgage banks, savings banks, specialized government credit institutions, and investment banks and

11

In the first regression (panel A) the coefficients of five country-specific variables are
significantly different from zero. As expected, the degree of trade openness (measured by the
ratio of exports to GDP) is positively correlated with the probability of equity interests
abroad, consistent with the hypothesis that one of the main determinants of banking
internationalization is that banks follow their clients. The size of the banking sector
(measured by the ratio of total credit to the economy to GDP) and the average national return
on assets in banking are also positively correlated with the degree of internationalization.
These results confirm that banks in countries with a more profitable and developed banking
sector are more likely to have a competitive advantage over their competitors in the
destination market. More surprisingly, we find that the size of the stock market (measured by
the ratio of total stock market capitalization to GDP) is negatively correlated with banks
internationalization. We interpret this as evidence that the choice to expand abroad is driven
by the banks search for profit opportunities beyond those offered by traditional banking
activity at home: when the financial sector is sufficiently developed so that additional profit
opportunities can be exploited in the home country simply by offering more innovative
financial services, there is less incentive to expand abroad. Finally, we find that banks from
countries with a higher average rate of inflation are less internationalized: a higher rate of
inflation weakens the currency, making foreign acquisitions more expensive.
Three coefficients of bank-specific variables are significant in the regression reported in
panel A. The return on assets is positively correlated with the probability of having an equity
interest abroad; this is consistent with the hypothesis that efficiency, here proxied by
profitability, positively affects the level of internationalization. Banks with a larger share of
non-interest income are also more likely to have foreign equity interests. Apparently, this
result conflicts with the hypothesis that banks that are able to shift their activity toward
innovative financial services in their home market have less incentive to expand abroad.
However, conditional on the opportunities of diversification offered by the home market
(measured by the size of the stock market) it is not surprising to find that more innovative
banks look for new profit opportunities and, therefore, have both a larger share of revenues
from non-traditional activities and a greater propensity to expand abroad.

securities houses. Although they are not reported in the tables, dummies for each category of banks are included
in all the regressions.

12

Finally, consistent with previous literature, our results show that the size of the bank
(measured by the logarithm of its total assets) is positively correlated

with

internationalization.
In our view, the variables included in the regression reported in panel A offer a very
plausible characterization of the pattern of bank internationalization. The explanatory power
is also fairly good, with a pseudo R2 of 0.56.
In theory, this specification might be problematic since it assumes that balance-sheet
data affect the decisions of banks to expand abroad but some of the variables could actually
be the consequence of the foreign shareholding (i.e., there may be problems of reverse
causality: in particular, banks may be larger because of internationalization). Thus, we
checked for this possibility using a regression with individual bank fixed effect and found a
negative and insignificant effect of foreign shareholdings on the rate of growth of total assets
between 1994 and 1997.12
Table 8 also reports the magnitude of the effect of each explanatory variable, measured
as the change in the probability that a bank will have a foreign shareholding (expressed in
percentage points) associated with a change in the covariate from the 25th to the 75th
percentile of the sample distribution, leaving all other variables at their sample values; we call
this the marginal effect. The results show that the factors most strongly affecting the
probability of a foreign shareholding are the measures of bank profitability and efficiency.
The marginal effects of changes in the average country level of return on assets and in the size
of the banking sector are particularly strong (respectively 7.00 and 5.31 percentage points).
The impact of the measure of trade openness is weaker (1.53). The most important individual
bank characteristics is size (4.71); return on assets and non-interest income have weaker
effects (1.98 and 1.12, respectively). These results confirm our contention that the overall
degree of economic integration between countries is only one, and not the most important, of
the determinants of the pattern of banks international shareholdings.
Panel B reports the estimates of a specification that includes two variables describing
the regulatory restrictions on domestic banking activity and on banks outward foreign direct
investment. The coefficients of both variables are significant and with the expected sign.

12

The regression is not reported here. Results are available form the authors on request.

13

Restrictions on outward foreign direct investment obviously reduce the propensity of banks to
acquire foreign equity interests. We also find that banks in more strictly regulated markets are
less likely to hold foreign equity interests, a result providing further evidence that the
restrictions reduce the international competitiveness of the domestic banking sector.
Including the variables on regulatory restrictions results in only two changes with
respect to the first regression: the coefficient on inflation becomes insignificant, while that on
average country level bank non-interest income becomes significant at the 10 per cent level.
However, the change in the inflation coefficient is due to the elimination of the observations
from the Czech Republic, Hungary and Poland (where there has been high inflation) and not
to the inclusion of the measures of regulatory restrictions.
The result concerning the average level of non-interest income might depend on the fact
that when no measure of regulatory restrictions is included, this variable may capture two
opposite effects. In fact, higher non-interest income implies: a) that banks can search for new
profit opportunities within their home country; b) that there are fewer institutional restrictions
and a higher degree of internationalization. After including a direct measure of regulatory
restrictions, however, only the first effect, which reduces the probability of international
shareholdings, remains. Aside from these changes, the results reported in Panel B are
substantially identical to those of Panel A, including the magnitude of the marginal effects.
Panel C presents the results of a regression on the sub-sample of 257 banks with assets
of more than $25 billion. Analysis of this sub-sample is suggested by the large impact of bank
size in the first two regressions. Surprisingly, narrowing the sample to larger banks does not
reduce the level of significance of size coefficient; on the contrary, it increases the magnitude
of its marginal effect relative to the other explanatory variables. This seems to suggest that
increasing returns to scale from internationalization do not take the form of a minimum scale
that banks must reach in order to find it profitable to expand abroad.
There are other differences between the results of panel C and those of the first two
regressions. The coefficient of the degree of openness of the home country becomes
insignificant when we consider only larger banks. This may be explained by the fact that large
banks all have the same incentive to follow their clients abroad, since all have international
firms among their clients. The coefficient of the measure of the restrictions on domestic
banking activity is insignificant, possibly because this sub-sample includes only relatively

14

competitive markets and because larger banks can more easily find ways to sidestep
regulations. The coefficient of the average non-interest income in the home country is also
insignificant, although it remains negative as in the first two regressions.
Finally, somewhat surprisingly, the coefficient of individual bank return on assets is
negative and largely insignificant, whereas it was positive and highly significant in the first
two regressions. Probably, this depends on the reduced variability of return on assets within
the same country when only large banks are considered. In fact, in an unreported regression
we find that if only return on assets at the individual level is included, the coefficient is
positive and significant at the 13 per cent level.
The pseudo R2 of the regression in panel C is 0.40, lower than in the previous two
specifications but still reasonably high.
Table 9 shows the results of three regressions similar to those reported in table 8, where
return on assets is split into its two major components: net interest income less charge-offs
over total assets (a measure of profitability of traditional activities) and the cost-income ratio.
Overall, the results are consistent with the view that banks with higher returns from traditional
activity and lower costs are more internationalized, but the estimates of the cost-income
coefficients are not precise.
In panel A, the coefficients have the expected sign, positive for net interest income and
negative for costs (although only the national average level of costs is significant). The
coefficients of the other variables are essentially unchanged with respect to the correspondent
specification in table 8, except for national non-interest income, which is now positive and
significant. This upward bias can be explained by the presence of banks that have a high
return on assets because of their income from off-balance-sheet operations, an effect that is
accounted for in table 8 but not here. Panel B reports the results obtained including the
measure of regulatory restrictions: in this case the cost-income variables are not significant,
although they still have the expected sign. Finally, panel C reports the results of estimates on
the sample of larger banks. The findings in this case are less neat. On the revenue side, only
the coefficient of the average level of interest income is significant, probably due to the low
variability of individual observations within each country when only large banks are
considered. On the cost side, the coefficient of the national average is negative and
significant, as expected, but surprisingly that of individual bank overheads is positive and

15

significant. This last result may perhaps recall the anecdotal evidence that in countries with
the most efficient banking systems, more innovative banks invest heavily in technology and
human capital, and so have higher costs.
5.4

Robustness checks
In order to verify the robustness of our preferred specification (table 8, panel B), we test

for the significance of other explanatory variables that are consistent with explanations of the
patterns of international shareholdings suggested in the literature. The results, reported in
table 10, show that none of the additional covariates is significant even at the 10 per cent level
and that their inclusion does not affect the coefficients of the variables in the basic
specification.
The national average level of banks assets verifies the hypothesis that banking systems
with fewer and larger banks are more likely to expand abroad than more fragmented systems.
The coefficient of this variable is positive but insignificant, showing that size matters only at
individual level. The degree of market concentration, measured by the Herfindahl index,
controls for the effects of less competition on return on assets and checks whether greater
concentration generates a greater need for cross-border diversification. The coefficient is
negative (consistent with the hypothesis that higher concentration is associated with lower
efficiency) but not significantly different from zero. Similarly, the hypothesis that banking
systems largely controlled by the government are less efficient is checked by including the
share of assets controlled by state-owned banks. As before, the coefficient has the expected
sign but is not significant. The national average and the individual level of bank cash flow can
be interpreted as a measure of efficient use of internal resources (if the cash flow is too high,
resources could be used more profitably) but also of the availability of liquid assets for
acquisitions abroad. The coefficients of these variables are not significant, perhaps because
the two opposing effects offset one another. Per capita GDP tests the hypothesis that more
developed countries, with generally more efficient banking and financial markets, are more
likely to be the home of international banks. The negative and insignificant coefficient shows
that the other variables included in the basic regression already measure the effects of the
average efficiency of the banking system. The size of the population checks the possibility of
a correlation between country size and the degree of internationalization of the banking
system (say, because smaller countries have more open economies). However, the estimated

16

coefficient is insignificant, possibly because this effect is already captured by trade openness.
Countries with more efficient judicial systems might have a less developed banking system,
because they have more efficient markets. As a result, local banks may have a lower degree of
internationalization. The coefficient of this variable is insignificant, however, although it is
negative as expected. Finally, average stock market turnover has been included as an
additional measure of the efficiency of the non-banking financial sector.13 Its coefficient,
however, is insignificant.
6

Conclusions
The number of cross-border mergers and acquisitions in the banking industry has risen

rapidly in recent years, following a trend common to other sectors, but the process has been
less intense than in the non-financial sector: the share of cross-border M&As among banks is
about half that of non-financial firms. This may be for two reasons. First, the non-tradability
of a large part of banking services and the importance of relationships characterized by strong
information asymmetries make it more difficult to judge from outside the value of a bank than
of a manufacturing firm. The impact of characteristics that increase the opaqueness of bank
value is amplified in the case of cross-border transactions, making it harder to develop the
consolidation process internationally. Second, as our empirical evidence shows, more
stringent regulatory restrictions complicate cross-border mergers by banks compared with
non-financial firms.
It being recognized that the frequency of cross-border M&As in banking is significantly
below average, more direct study of the factors determining a banks decision to acquire a
foreign equity holding is of particular interest. The empirical evidence on this point presented
in this paper shows the importance of the factors associated with the efficiency of the single
bank and of the overall banking market in the home country. Banks in countries where the
banking sector is larger and more profitable, after controlling for market characteristics,
should be able to export a superior skill and therefore be more likely to expand abroad. In the
sample restricted to the largest banks, we also find evidence of an important role for costefficiency. The previous literature already provides some evidence that banks from countries
with a more highly developed banking sector are more likely to have international equity
13

In fact, the value of stock market capitalization is more comparable with the level of development of the
banking sector since they are both stock measures.

17

interests. Our results strengthen this thesis, suggesting that efficiency is more important than
the overall degree of economic integration between countries in explaining the pattern of
internationalization.
The size of the bank is a key determinant of the decision to expand abroad. Larger
banks are more likely to have foreign shareholdings, and this remains true even considering a
sub-sample of banks with assets of more than $25 billion. This result is consistent with the
view that larger banks have stronger incentives to search for additional profit opportunities
abroad. Moreover, it appears that banks with a larger share of non-interest income are more
likely to have foreign shareholdings, perhaps because they have more innovative and
aggressive strategies both at home and abroad.
The importance of the degree of openness of the home country is confirmed by the
positive and significant coefficient of the share of exports in GDP, but the marginal effect of
this variable is quite small, suggesting that the need to follow home-country clients operating
abroad is not the major factor in banks choices. Further, when only large banks are
considered, its significance vanishes altogether.
These results can shed light on the future evolution of international banking. Our two
main findings in this regard are that M&As are less frequent in banking than in the other
sectors of economic activity, partly because of regulatory restrictions, and that the banks that
are more likely to expand abroad tend to be the most efficient. The first regularity is likely to
remain true in the years ahead. Although the recent development of technologies that make
the acquisition and transfer of information much easier than in the past will probably foster
international integration, there is no reason to believe that this can overcome the effect of
information asymmetries in banking relationships.
The fact that banks with foreign shareholdings are on average more efficient might lead
one to think that they will eventually gain large market shares abroad. This radical view is
disproved by a number of empirical studies showing that foreign banks are generally less
efficient than their local competitors.14 However, Berger, DeYoung, Genay and Udell (2000)
show that although only US banks are more efficient than local competitors in France,
Germany, Spain and the United Kingdom, in a number of countries foreign banks can operate

14

See Berger, DeYoung, Genay and Udell (2000) and the literature cited therein.

18

as efficienctly as domestic banks. Therefore, some foreign banks are likely to continue
expanding their activities abroad, in particular toward the less developed countries.15
The changing institutional environment is also likely to affect the degree of bank
internationalization. The introduction of the single European currency, for example, has
removed a large number of factors that previously segmented the banking markets in the euro
area. More in general, our evidence suggests that regulatory restrictions have a negative effect
on the degree of banking internationalization. The progressive deregulation under way in the
OECD countries (see, for example, Mishkin, 2000) is thus likely to accelerate the
internationalization of banks.

15

This evidence is consistent with some of the findings reported in the previous version of the paper
(Focarelli and Pozzolo, 2000), showing that banks expand mostly toward countries with a less efficient banking
sector.

19

Table 1

Cross-Border M&As Distribution by Bidders Sector and Year

Source: Thomson Financial. Number of M&As is the number of operations with both bidder and target companies in an OECD country.
Cross-border is the ratio of cross-border to total M&As. The Sectors are the SIC sector classification (1 digit). Transportation and Public
Utilities (SIC=4) is divided into two groups. Financial Sector (SIC=7) is divided into banks, insurance and other company.

Sector

Year
1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

Total

Agriculture,
Forestry,
Fishing

# of M&As
% of Cross-Border

5
20.0

11
45.5

11
9.1

7
28.6

10
30.0

16
12.5

21
28.6

16
25.0

24
33.3

20
5.0

141
23.4

Mining

# of M&As
% of Cross-Border

167
27.5

232
25.0

211
28.0

299
22.1

252
27.0

296
27.4

316
23.7

249
24.5

241
20.7

262
21.8

2525
24.6

# of M&As
% of Cross-Border

17
29.4

37
27.0

34
35.3

28
32.1

40
25.0

39
25.6

56
23.2

46
15.2

62
14.5

75
20.0

434
23.0

Manufactur- # of M&As
ing
% of Cross-Border

507
51.1

599
37.9

698
36.4

778
34.1

832
33.8

1077
34.4

1099
31.9

1061
35.1

1100
34.4

1447
33.8

9198
35.3

Transportation

# of M&As
% of Cross-Border

31
58.1

52
23.1

58
22.4

67
23.9

73
23.3

76
15.8

86
37.2

70
32.9

70
28.6

100
38.0

683
29.4

Public Utilities

# of M&As
% of Cross-Border

87
21.8

116
18.1

117
18.8

216
25.0

176
23.9

269
23.8

233
24.5

281
21.4

303
29.4

476
31.5

2274
25.4

Wholesale
Trade

# of M&As
% of Cross-Border

63
73.0

80
45.0

95
26.3

109
34.9

145
32.4

145
40.0

148
31.8

146
28.8

148
26.4

190
26.3

1269
33.7

Retail Trade

# of M&As
% of Cross-Border

33
24.2

64
25.0

62
27.4

90
15.6

100
12.0

98
18.4

113
17.7

109
20.2

116
16.4

158
22.2

943
19.2

Banks

# of M&As
% of Cross-Border

193
21.2

245
15.5

312
9.3

428
9.8

489
10.0

496
12.7

460
10.9

532
10.2

547
14.1

530
19.6

4232
12.9

Insurance

# of M&As
% of Cross-Border

126
30.2

102
29.4

116
25.9

118
24.6

121
23.1

131
30.5

135
25.9

127
28.3

123
37.4

110
40.9

1209
29.5

Other Finan- # of M&As

cial
% of Cross-Border

671
15.1

785
13.5

745
17.3

799
11.4

985
12.3

1276
12.2

1376
11.3

1030
13.6

900
19.2

1228
21.9

9795
14.7

Services

# of M&As
% of Cross-Border

113
30.1

156
30.1

203
19.7

274
13.9

312
19.9

367
22.6

467
23.8

501
20.6

528
26.3

656
27.4

3577
23.4

Public Administration

# of M&As
% of Cross-Border

16
25.0

31
35.5

20
15.0

19
5.3

13
15.4

12
16.7

25
20.0

21
9.5

26
34.6

38
26.3

221
22.2

Total

# of M&As
% of Cross-Border

2029
30.6

2510
24.6

2682
23.6

3232
20.6

3548
20.9

4298
22.3

4535
21.1

4189
22.1

4188
25.2

5290
27.3

36501
23.6

Construction

Table 2

Ratio of Cross-Border M&As Distribution by Bidders Sector and Country

Source: Thomson Financial. Number of M&As is the number of operations with both bidder and target companies in an OECD country.
Cross-border is the ratio of cross-border to total M&As. Non-Financial Sector are firms of all SIC sectors with the exception of
Financial and Public Administration.

Country

Banks
# of M&As % of Cross-border

Australia
Austria
Belgium
Canada
South Korea
Denmark
Finland
France
Germany
Japan
Greece
Ireland
Iceland
Italy
Luxembourg
Mexico
Norway
New Zealand
Netherlands
Poland
Portugal
United Kingdom
Czech Republic
Spain
United States
Sweden
Switzerland
Turkey
Hungary

93
24
42
129
9
20
23
202
82
57
12
26

29.0
25.0
73.8
26.4
11.1
30.0
30.4
25.2
58.5
24.6
0.0
50.0

213
6
15
37
11
34
27
41
329
6
177
2497
59
50
8
3

Total

4232

Non-Financial Sector
# of M&As % of Cross-border

14.6
100.0
13.3
2.7
45.5
61.8
3.7
19.5
19.1
66.7
18.6
4.2
22.0
34.0
0.0
0.0

803
61
120
2068
120
96
213
748
382
842
47
174
1
356
21
86
255
137
249
92
63
3590
22
311
9654
317
156
26
34

26.0
57.4
63.3
28.8
42.5
65.6
29.1
49.1
72.8
45.8
27.7
69.0
0.0
38.5
90.5
33.7
36.9
40.1
75.5
13.0
30.2
36.0
22.7
28.6
18.0
53.0
77.6
11.5
20.6

12.9

21044

29.6

Table 3

Banks Distribution by Size and Country

Source: Bankscope. Panel A includes the 2,148 banks in the regression sample. From the 2,449 banks with headquarters in an OECD
country and total assets of more than $1 billion at December 1997 we exclude 301 banks with either negative net interest income, or
negative non-interest income, or net return on assets lower than 1.0 per cent or higher than 4.0 per cent, or overheads higher than total
operating income. Panel B refers to the subset of banks with total assets of more than $ 25 billions.
Panel A:
All banks

Country
Total

Australia
Austria
Belgium
Canada
South Korea
Denmark
Finland
France
Germany
Japan
Greece
Ireland
Iceland
Italy
Luxembourg
Mexico
Norway
New Zealand
Netherlands
Poland
Portugal
United Kingdom
Czech Republic
Spain
United States
Sweden
Switzerland
Turkey
Hungary
Total

42
41
27
23
36
14
7
178
472
159
12
23
2
138
56
12
21
9
31
16
26
119
10
94
488
14
62
9
7
2148

With
shareholdings
abroad
3
5
8
4
0
2
0
15
32
14
1
2
0
10
3
0
1
0
4
0
3
11
0
5
18
2
3
0
0
146

With
foreign
shareholders
10
8
7
9
10
2
1
11
19
6
1
9
0
13
46
1
10
6
8
10
8
29
6
12
18
1
11
0
4
276

Panel B:
Banks with total assets of more than $ 25
billions
Total
With
With
shareForeign
holdings
shareabroad
holders
4
3
0
3
3
2
6
5
1
7
4
0
9
0
2
4
2
1
3
0
0
20
12
1
33
26
3
56
14
5
1
1
0
2
2
0
0
0
0
15
7
5
3
0
2
3
0
0
1
0
1
0
0
0
5
3
0
0
0
0
3
2
1
22
8
2
0
0
0
8
4
3
42
14
1
5
2
0
5
2
0
0
0
0
0
0
0
260

114

30

Table 4

Bank Summary Statistics

Source: Bankscope. Panel A refers to the 2,148 banks included in the regression sample. In particular, from the
2,449 banks with headquarters in an OECD country and assets of more than $1 billion at December 1997 we
exclude 301 banks with negative net interest income or negative non-interest income or with net return on assets
lower than 1.0 per cent or higher than 4.0 per cent or with overheads higher than total operating income; panel B
refers to all the banks of the sample with assets of more than $25 billion. Total assets are expressed in billions of
US dollars. ROE is defined as income after tax in proportion to equity. ROA (defined as income after tax), net
interest income, cash-flow (defined as equity less fixed assets and loan loss reserves), off-balance-sheet
operations, and net charge-offs are expressed as percentage ratios to total assets. Non-interest income is
expressed as a percentage ratio to the sum of net interest income and non-interest income. All data are calculated
as the average value of 1994-1997 figures.

Variables

Obs.

Median
Mean
Panel A: All Banks

Std. Dev.

Min.

Max.

Total Assets
Net ROE
Net ROA
Net Interest Income
Non-Interest Income
Overheads
Cash Flow
Off-Balance
Net Charge-offs

2148
3.29
17.15
2148
7.84
8.87
2148
0.39
0.58
2147
2.56
2.55
2148
22.11
26.30
2148
63.38
62.31
2148
3.87
4.69
1911
9.30
23.45
2148
0.27
0.37
Banks with no shareholding abroad

51.50
8.19
0.61
1.46
17.98
14.93
4.74
73.71
0.56

0.57
-34.01
-0.87
0.02
0.00
0.00
-10.23
0.00
-1.87

715.45
122.06
3.90
14.51
100.00
100.00
88.13
1131.28
7.03

Total Assets
Net ROE
Net ROA
Net Interest Income
Non-Interest Income
Overheads
Cash Flow
Off-Balance
Net Charge-offs

2002
2.99
9.13
2002
7.82
8.86
2002
0.39
0.59
2001
2.64
2.60
2002
20.97
25.44
2002
63.18
62.19
2002
4.01
4.74
1776
8.82
22.58
2002
0.27
0.37
Banks with shareholding abroad

23.20
8.20
0.61
1.46
17.62
14.98
4.66
72.99
0.57

0.57
-34.01
-0.87
0.02
0.00
0.00
-10.23
0.00
-1.87

432.00
122.06
3.90
14.51
100.00
100.00
88.13
1131.28
7.03

137.04
8.08
0.60
1.14
18.86
14.26
5.79
81.99
0.42

1.17
-15.19
-0.85
0.13
4.12
18.68
-0.68
0.75
-0.94

715.45
34.19
3.72
8.12
91.84
98.61
64.81
873.94
2.84

Total Assets
Net ROE
Net ROA
Net Interest Income
Non-Interest Income
Overheads
Cash Flow
Off-Balance
Net Charge-offs

146
146
146
146
146
146
146
135
146

77.58
8.47
0.36
1.50
34.43
65.27
2.83
19.32
0.22

127.16
9.03
0.49
1.78
38.00
64.03
4.02
34.99
0.31

Table 4 (continued)
Variables

Obs.

Median

Mean

Std. Dev

Min.

Max.

Panel B: Banks with total assets of more than $25 billion

Total Assets
Net ROE
Net ROA
Net Interest Income
Non-Interest Income
Overheads
Cash Flow
Off-Balance
Net Charge-offs

260
57.93
105.93
260
7.61
7.57
260
0.28
0.41
259
1.63
1.75
260
32.08
33.71
260
65.11
63.55
260
2.74
3.36
218
13.22
22.56
260
0.22
0.31
Banks with no shareholding abroad

Total Assets
Net ROE
Net ROA
Net Interest Income
Non-Interest Income
Overheads
Cash Flow
Off-Balance
Net Charge-offs

146
146
146
145
146
146
146
113
146

42.89
6.81
0.27
1.75
28.75
64.05
3.04
6.80
0.21

64.15
7.16
0.43
1.86
30.22
62.07
3.50
18.04
0.33

113.12
8.40
0.50
1.02
20.98
15.20
2.95
28.85
0.37

25.08
-20.95
-0.85
0.09
0.47
14.91
-1.21
0.00
-0.94

715.45
31.92
2.27
5.07
100.00
98.61
31.10
257.69
2.15

61.79
8.67
0.52
1.12
21.66
16.36
3.39
23.65
0.41

25.08
-20.95
-0.66
0.09
0.47
14.91
-1.21
0.00
-0.42

432.00
31.92
2.27
5.07
100.00
95.33
31.10
112.11
2.15

138.88
8.04
0.46
0.86
19.26
13.40
2.25
32.99
0.32

28.25
-15.19
-0.85
0.13
4.12
26.64
-0.68
0.75
-0.94

715.45
29.73
1.77
3.92
91.84
98.61
14.09
257.69
1.33

Banks with shareholding abroad

Total Assets
Net ROE
Net ROA
Net Interest Income
Non-Interest Income
Overheads
Cash Flow
Off Balance
Net Charge-offs

114
114
114
114
114
114
114
105
114

110.91
8.01
0.30
1.45
35.26
65.95
2.68
19.38
0.23

159.44
8.08
0.38
1.62
38.18
65.46
3.18
27.43
0.30

Table 5

Country Summary Statistics

The summary statistics in Panel A are country averages of the 2,168 banks included in the sample. ROA is defined as income after tax as a percentage of total assets. Net interest income (NII) less net
charge-offs and cash flow (defined as equity less fixed assets and loan loss reserves) are expressed as percentages of total assets. Overheads and non-interest income are expressed as percentages of the
sum of net interest income and non-interest income. Total assets are expressed in billions of dollars. Market concentration is computed by the Herfindhal index. In panel B, data on GDP, population,
bank domestic credit and inflation are from IMF, International Financial Statistics (1998); stock market capitalization is from IFC (1998), Emerging Stock Markets Factbook. Exports are only toward
other OECD countries and are from IMF, Direction of Trade Statistics (1998). GDP is expressed in billions of dollars. Per capita GDP is expressed in dollars. Bank domestic credit, stock market
capitalization and exports are expressed as percentages of GDP. All data in Panels A and B are calculated as the average value of 1994-1997 figures. In panel C, Efficiency of judicial system is taken
from La Porta et al. (1998); Restrictions on Banking and State-Owned Banks are from Barth et al. (2000); Restrictions on inward and outward Foreign Direct Investment are our computation
based on IMF, Exchange Arrangements and Exchange Restrictions (1997).
Panel A
Panel B
Panel C
Country
Per
Return
NonNII
OverCash AverMarket GDP
Bank
Stock
InflaExRestric- StateEffi- Restric- Restriccapita Domes- Market
on As- Interest
Net
heads
flow
age To- Contion
ports tions on owned ciency of tions on tions on
GDP
sets
Income Charge
tal As- centratic
banking banks Judicial outward inward
Offs
sets
tion
Credit
System
FDI
FDI
0.84
38.4
59.6
5.25
12.8
362.3
19918
84.9
114.2
2.4
8.9
10.00
Australia
1.86
0.108
2.00
0.00
0
3
0.28
28.8
64.0
2.96
11.3
213.4
26503
128.9
15.5
2.1
22.1
9.50
Austria
1.29
0.118
1.25
0.04
0
0
0.33
28.0
70.7
2.88
28.6
253.2
24992
151.0
44.0
1.9
54.0
9.50
Belgium
1.48
0.119
2.50
0.00
0
1
0.43
33.3
67.4
4.85
31.9
579.2
19445
76.2
74.9
1.4
31.0
9.25
Canada
1.69
0.142
2.25
0.00
0
1
0.45
36.0
64.8
3.39
15.8
388.2
8566
64.4
35.7
5.0
15.2
6.00
South Korea
1.40
0.049
2.25
0.00
2
2
0.74
19.0
63.0
6.57
16.3
171.1
32635
55.1
40.3
2.1
23.1
10.00
Denmark
2.54
0.183
1.75
0.00
0
0
0.30
37.7
62.2
2.59
33.7
118.1
23079
65.0
46.3
1.0
23.6
10.00
Finland
1.24
0.286
1.75
0.41
0
0
0.41
30.9
72.4
4.67
21.4
24915
101.4
38.6
1.7
14.9
8.00
France
2.17
0.043 1451.4
2.00
0.15
0
1
0.28
17.9
60.7
3.28
12.0
27194
133.2
28.6
2.0
18.1
9.00
Germany
2.10
0.027 2220.6
1.75
0.43
0
0
-0.03
14.9
67.2
1.79
56.7
35347
134.6
71.6
0.6
5.3
10.00
Japan
1.31
0.029 4429.3
3.25
0.00
0
2
0.96
41.0
70.6
3.14
9.5
112.4
10751
59.2
20.1
8.4
5.9
7.00
Greece
2.63
0.190
2.25
0.63
0
0
0.79
18.4
44.6
7.06
5.5
64.8
18159
44.6
27.2
2.0
60.4
8.75
Ireland
1.90
0.134
1.75
0.00
0
0
0.46
37.0
69.5
2.71
1.3
7.0
25757
54.8
11.6
1.8
24.2
.
Iceland
2.67
0.519
2.75
0.64
0
2
0.39
28.5
67.1
6.33
12.4
19422
96.5
22.3
3.8
15.8
6.75
Italy
2.68
0.038 1113.8
2.25
0.25
0
0
0.40
37.7
47.3
2.62
7.2
16.3
39873
.
192.0
1.7
.
9.50
Luxembourg
0.69
0.034
.
.
0
0
0.46
25.4
73.7
3.29
11.9
304.3
3234
29.4
39.8
24.2
26.4
6.00
Mexico
3.29
0.095
3.25
0.42
1
2
0.99
21.6
60.4
5.28
5.5
145.5
33304
57.7
35.2
1.9
27.4
10.00
Norway
2.65
0.120
2.00
0.38
0
0
0.75
40.6
67.5
2.09
9.4
59.9
16673
89.1
78.6
2.2
15.5
10.00
New Zealand
2.07
0.150
1.25
0.00
0
1
0.53
29.0
56.9
4.12
30.4
368.7
23804
121.7
100.8
2.2
41.6
10.00
Netherlands
1.28
0.203
1.50
0.00
0
0
2.49
27.2
45.3
6.66
3.0
113.9
2950
36.4
6.2
24.0
15.4
.
Poland
5.30
0.117
.
.
0
1
0.57
33.5
62.8
4.15
8.9
100.7
10199
93.1
24.4
3.6
19.8
5.50
Portugal
1.74
0.087
2.00
0.17
0
0
0.76
35.8
58.7
6.25
24.7
20043
125.9
135.3
2.9
16.1
10.00
United Kingdom
1.80
0.044 1174.4
1.25
0.00
0
0
0.85
41.1
53.7
3.98
4.6
48.8
4731
79.2
26.9
9.1
26.9
.
Czech Republic
1.54
0.174
.
.
.
.
0.83
21.2
62.0
3.97
10.5
535.5
13648
105.4
41.4
3.7
14.3
6.25
Spain
2.83
0.057
1.75
0.02
1
1
1.01
27.2
60.0
6.89
10.6
28368
78.3
105.8
2.7
5.7
10.00
United States
2.80
0.011 7496.2
3.00
0.00
0
1
0.52
16.9
49.7
3.28
27.6
229.8
26045
70.0
90.2
1.5
28.0
10.00
Sweden
1.41
0.113
3.00
0.00
0
0
0.64
48.7
69.2
4.60
20.5
278.2
39478
182.8
152.3
1.0
22.4
10.00
Switzerland
1.07
0.259
1.50
0.15
0
0
1.30
18.5
62.3
4.42
4.2
127.9
2058
31.1
26.1
90.1
10.7
4.00
Turkey
4.81
0.082
3.00
0.37
1
3
1.78
40.0
61.7
3.80
2.7
40.3
3945
55.8
15.1
22.2
26.5
.
Hungary
4.00
0.242
.
.
0
0

Table 6

The Determinants of Bank Cross-border M&As

The empirical models are: equation (1) for column (a); equation (2) for column (b); an OLS regression of the country dummies from
the regression of column (a) for column (c). The share of cross-border M&As in total M&As in the banking sector is from Thomson
Financial. Restrictions on Banking is taken from Barth et al. (2000). The regressions in columns (a) and (b) also include countryspecific and time-specific dummy variables (unreported). Standard errors are reported in italics. The symbol *** indicates a
significance level of 1 per cent or less; ** between 1 and 5 per cent; * between 5 and 10 per cent.
(a)
Dependent variable:
cross-border over total
M&As in banking
Coeff.
Signif.
(Std.
Err.)

VARIABLES

Cross-border over total M&As

in the non-financial sector

(b)
Dependent variable:
cross-border over total
M&As in banking
Coeff.
Signif.
(Std.
Err.)

0.389 ***
0.134

0.438 ***
0.136
-0.446 **
0.195

Restrictions on banking
226
0.21

No. Of Observations
Adjusted R-Square

(c)
Dependent variable:
Country dummies from
regression in column (a)
Coeff.
Signif.
(Std.
Err.)

- 0.67 **
0.24

210
0.35

24
0.18

Table 7

The Determinants of Bank Cross-border M&As

(robustness check)
The empirical model is equation (2). The share of cross-border M&As in total M&As in the banking sector is from Thomson Financial.
ROA is defined as income after tax as a percentage of total assets, non-interest income is expressed as a percentage of the sum of net
interest income and non-interest income; they are calculated at the country level as yearly averages of individual bank values obtained
up to 1998 from Bankscope. Data on GDP, inflation, and bank domestic credit are from IMF, International Financial Statistics (1998).
Stock market capitalization is from Datastream. Exports are only toward other OECD countries and are from IMF, Direction of Trade
Statistics (1998). Restrictions on Banking is taken from Barth et al. (2000).
Last row: Included variables
VARIABLES

Cross-border over total M&As

in the non-financial sector
Restrictions on banking

No. Of Observations
Adjusted R-Square

Return on
Assets
Coeff. Signif.
(Std.
Err.)

Non-Interest
Income
Coeff. Signif.
(Std.
Err.)

Bank credit /
GDP
Coeff. Signif.
(Std.
Err.)

Coeff.
(Std.
Err.)

0.443 ***
0.141
-0.539 ***
0.198
0.747
0.494

0.479 ***
0.139
-0.489 ***
0.198
0.017
0.022

0.405 ***
0.140
-0.430 **
0.201
0.710
1.473

0.418 ***
0.136
-0.454 **
0.194
-0.064
2.480

209
0.35

209
0.43

191
0.33

Exports / GDP
Signif.

196
0.36

Stock market
cap. / GDP
Coeff. Signif.
(Std.
Err.)
0.442 ***
0.164
-0.443 **
0.224
0.349
0.576
171
0.36

Table 8

The Determinants of a Banks Decision to Expand Abroad

The empirical model in equation (3) has been estimated using a probit specification, where the dependent variable equals one if the
bank has a foreign subsidiary and zero otherwise. ROA is defined as income after tax as a percentage of total assets. Non-interest
income is expressed as a percentage of the sum of net interest income and non-interest income. Total assets are expressed in billions of
dollars. Data on GDP, inflation, and bank domestic credit are from IMF, International Financial Statistics (1998). Stock market
capitalization is from IFC, Emerging Stock Markets Factbook (1998). Exports are from IMF, Direction of Trade Statistics (1998). All
data are calculated as the average value of 1994-1997 figures. Restrictions on Banking is taken from Barth et al. (2000);
Restrictions on Outward Foreign Direct Investment is our computation based on IMF, Exchange Arrangements and Exchange
Restrictions (1997). The regression also includes dummy variables (unreported) to account for the nature of the main activity of the
bank (commercial banks, cooperative banks, medium and long term banks, real estate-mortgage banks, savings banks, specialized
government credit institutions, and investment banks and securities houses). Standard errors are reported in italics. The symbol ***
indicates a significance level of 1 per cent or less; ** between 1 and 5 per cent; * between 5 and 10 per cent. Marginal effect is the
change in the average of the individual predicted probability (expressed in percentage points) due to an increase from the 25th percentile
to the 75th percentile of the variable of interest.

VARIABLES

Exports / GDP
Bank Credit / GDP
Stock Market
Capitalization / GDP
Inflation
Restrictions on
banking
Restrictions on
outward FDI
Return on Assets
(Country)
Non-Interest
Income (Country)
Return on Assets
(Individual)
Non-Interest Income
(Individual)
Total Assets
(log value)
No. Of Observations
Pseudo R-Square
Observed Prob.

Panel A:
All Banks except those from
Luxembourg
Coeff.
(Std.
Err.)

Signif.

2.151
0.582
1.536
0.338
-1.579
0.348
-0.141
0.041

***

1.53

***

5.31

***

-3.76

***

-0.64

1.421
0.387
-0.001
0.010
0.489
0.164
0.010
0.004
0.881
0.068

***

Marginal
Effect

7.00
-0.06

***

1.98

**

1.12

***

4.71

2,092
0.56
6.84

Panel B:
All Banks except those from
Luxembourg, Czech Republic,
Hungary and Poland
Coeff. Signif. Marginal
(Std.
Effect
Err.)
1.801
0.602
1.419
0.416
-1.374
0.330
-0.006
0.013
-0.323
0.153
-0.851
0.343
1.349
0.427
-0.020
0.012
0.486
0.167
0.011
0.004
0.897
0.066

***

1.26

***

4.69

***

-3.35

Panel C:
Banks with Total Assets of
more than $25 billion
Coeff.
(Std.
Err.)

Signif.

0.498
1.116
1.651 **
0.755
-2.136 ***
0.523

Marginal
Effect
1.45
21.41
-24.10

-0.02
**

-1.75

**

0.00

***

6.54

-1.61

***

1.90

***

1.18

***

5.28

2,059
0.58
6.95

-0.273
0.187
-1.160
0.456
2.540
0.836
-0.026
0.018
-0.196
0.326
0.021
0.008
0.962
0.143

-6.73
**

0.00

***

29.08
-11.08
-2.87

***

10.64

***

31.89

257
0.40
44.36

Table 9

The Determinants of a Banks Decision to Expand Abroad

The empirical model in equation (3) has been estimated using a probit specification, where the dependent variable equals one if the
bank has a foreign subsidiary and zero otherwise. Net interest income (NII) less net charge-offs is expressed as a percentage of total
assets. Non-interest income and overheads are expressed as a percentage of the sum of net interest income and non-interest income.
Total assets are expressed in billions of dollars. Data on GDP, inflation, and bank domestic credit are from IMF, International
Financial Statistics (1998). Stock market capitalization is from IFC, Emerging Stock Markets Factbook (1998). Exports are from IMF,
Direction of Trade Statistics (1998). All data are calculated as the average value of 1994-1997 figures. Restrictions on Banking and
State-Owned Banks are taken from Barth et al. (2000); Restrictions on Outward Foreign Direct Investment is our computation on
IMF, Exchange Arrangements and Exchange Restrictions (1997). The regression also includes dummy variables (unreported) to
account for the nature of the main activity of the bank (commercial banks, cooperative banks, medium and long term banks, real estatemortgage banks, savings banks, specialized government credit institutions, and investment banks and securities houses). Standard errors
are reported in italics. The symbol *** indicates a significance level of 1 per cent or less; ** between 1 and 5 per cent; * between 5 and
10 per cent. Marginal effect is the change in the average of the individual predicted probability (expressed in percentage points) due to
an increase from the 25th percentile to the 75th percentile of the variable of interest.

VARIABLES

Exports / GDP
Bank credit / GDP
Stock Market
Capitalization / GDP
Inflation
Restrictions on
banking
Restrictions on
outward FDI
Overheads
(Country)
NII - Net Charge-Offs
(Country)
Non-Interest Income
(Country)
Overheads
(Individual)
NII - Net Charge-Offs
(Individual)
Non-Interest Income
(Individual)
Total Assets
(log value)
No. Of Observations
Pseudo R-Square
Observed Prob.

Panel A:
All Banks
Except those from
Luxembourg
Coeff. Signif. Marginal
(Std.
Effect
Err.)
2.643
0.686
1.568
0.345
-1.261
0.314
-0.164
0.054

-0.054
0.020
0.671
0.166
0.032
0.014
-0.008
0.006
0.172
0.096
0.019
0.005
0.873
0.066

***

1.83

***

5.41

***

-3.10

***

-0.76

***

-2.28

***

4.60

**

2.45
-0.78

1.83

***

1.98

***

5.01

2,092
.566
6.84

Panel B:
All Banks Except those from
Luxembourg, Czech Republic,
Hungary and Poland
Coeff.
(Std.
Err.)

Signif.

2.621
0.696
1.138
0.427
-0.846
0.367
-0.030
0.042
-0.299
0.191
-0.561
0.297
-0.023
0.023
0.524
0.178
0.001
0.019
-0.008
0.006
0.185
0.093
0.020
0.005
0.872
0.065

***

1.83

***

3.81

**

-2.25

Marginal
Effect

Panel C:
Banks with Total Assets of
more than $25 billion
Coeff.
(Std.
Err.)

Signif.

1.468
1.024
1.873 **
0.745
-2.325 ***
0.518

Marginal
Effect
4.26
23.63
-24.64

-0.13
-1.76
*

0.00
-0.98

***

3.45
0.10
-0.76

**

1.96

***

2.08

***

5.31

2,058
.570
6.95

-0.155
0.196
-0.800
0.419
-0.093
0.030
1.024
0.363
0.033
0.022
0.021
0.010
0.006
0.143
0.017
0.009
0.996
0.166

-3.55
*

0.00

***

-15.67

***

24.84
13.08

**

7.45
0.19

*
***

256
.413
44.53

8.30
32.49

Table 10

The Determinants of a Banks Decision to Expand Abroad

(robustness check)
The last row of each column refers to the coefficient of the variable in the column title. The empirical model in equation (3) has been estimated using a probit specification, where the dependent variable equals one if
the bank has a foreign subsidiary and zero otherwise. ROA (defined as income after tax) and cash flow (defined as equity less fixed assets and loan loss reserves) are expressed as a percentage of total assets. Noninterest income is expressed as a percentage of the sum of net interest income and non-interest income. Total assets are expressed in billions of dollars. Market concentration is computed by the Herfindhal index. Data
on GDP, population and bank domestic credit are from IMF, International Financial Statistics (1998). Stock market capitalization and liquidity are from IFC, Emerging Stock Markets Factbook (1998). Exports are
from IMF, Direction of Trade Statistics (1998). All data are calculated as the average value of 1994-1997 figures. Restrictions on Banking and State-Owned Banks are taken from Barth et al. (2000); Restrictions
on Outward Foreign Direct Investment is our computation based on IMF, Exchange Arrangements and Exchange Restrictions (1997); Efficiency of Judicial System is taken from La Porta et al. (1998). The
regression also includes dummy variables (unreported) to account for the nature of the main activity of the bank.. Standard errors are reported in italics. The symbol *** indicates a significance level of 1 per cent or
less; ** between 1 and 5 per cent; * between 5 and 10 per cent.
All Banks Except those from Luxembourg, Czech Republic, Hungary and Poland
Variables

Exports / GDP
Bank credit / GDP
Stock Market
Capitalization / GDP
Restrictions on
banking
Restrictions on
outward FDI
Return on Assets
(Country)
Non-Interest Income
(Country)
Return on Assets
(Individual)
Non-Interest Income
(Individual)
Total Assets
(log value)

No. Of Observations

Average Total Assets Market Concentration

(Country; log value)
Coeff.
Signif.
Coeff.
Signif.
1.769 ***
2.251 ***
0.612
0.722
1.454 ***
1.360 ***
0.400
0.435
-1.437 ***
-1.310 ***
0.428
0.332
-0.332 **
-0.329 **
0.163
0.153
-0.896 **
-0.801 ***
0.346
0.308
1.469 **
1.221 ***
0.710
0.436
-0.021
-0.014
0.013
0.013
0.488 ***
0.487 ***
0.166
0.167
0.011 ***
0.011 ***
0.004
0.004
0.896 ***
0.896 ***
0.068
0.066
0.058
-1.516
0.295
1.807

2,059

2,059

Per capita GDP

Coeff.
1.791
0.599
1.496
0.416
-1.328
0.340
-0.308
0.148
-0.975
0.416
1.353
0.425
-0.023
0.014
0.487
0.166
0.011
0.004
0.897
0.066
-0.170
0.313

Signif.
***
***
***
**
**
***

***
***
***

2,059

Last rows: Included variables

State-owned Bank Efficiency of judicial
system
Coeff.
Signif.
Coeff.
Signif.
1.649 **
1.977 ***
0.664
0.608
1.504 ***
1.441 ***
0.423
0.424
-1.492 ***
-1.081 **
0.370
0.438
-0.354 **
-0.342 **
0.157
0.157
-0.967 ***
-1.125 ***
0.364
0.430
1.416 ***
1.339 ***
0.442
0.442
-0.022 *
-0.030 **
0.012
0.015
0.490 ***
0.492 ***
0.167
0.165
0.011 ***
0.012 ***
0.004
0.004
0.896 ***
0.898 ***
0.067
0.066
-0.350
-0.101
0.553
0.091

2,059

2,057

Population
(log value)
Coeff.
Signif.
2.202 ***
0.778
1.352 ***
0.437
-1.373 ***
0.324
-0.351 **
0.162
-0.817 **
0.315
1.253 ***
0.435
-0.016
0.013
0.486 ***
0.167
0.011 ***
0.004
0.898 ***
0.066
0.068
0.099

Stock Market
Liquidity / GDP
Coeff.
Signif.
1.811 ***
0.615
1.417 ***
0.429
-1.366 ***
0.400
-0.325 **
0.153
-0.862 **
0.333
1.331 ***
0.507
-0.020 *
0.012
0.487 ***
0.166
0.011 ***
0.004
0.897 ***
0.067
0.008
0.546

2,059

2,059

Cash Flow (Country

and Individual)
Coeff.
Signif.
1.906 ***
0.594
1.482 ***
0.427
-1.354 ***
0.327
-0.318 **
0.153
-0.775 **
0.344
1.288 ***
0.478
-0.022 *
0.012
0.384 **
0.181
0.011 ***
0.004
0.907 ***
0.067
0.044
0.067
0.017
0.012
2,059

References
Ball, C. A. and A. E. Tschoegel (1982), The Decision to Establish a Foreign Branch or
Subsidiary: An Application of Binary Classification Procedures, Journal of Financial
and Quantitative Analysis, 17, pp. 411-24.
Barth, J. R., G. Caprio Jr., R. Levine (2000), Banking Systems Around the Globe: Do
Regulation and Ownership Affect Performance and Stability?, mimeo.
Berger A. N., R. S. Demsetz, P. E. Strahan (1999), The Consolidation of the Financial
Services Industry: Causes, Consequences, and Implications for the Future Journal of
Banking & Finance, 23, pp.135-94.
___________, R. DeYoung, H. Genay, and G. F. Udell (2000), The Globalization of
Financial Institutions: Evidence from a Cross-Border Banking Performance, mimeo,
Federal Reserve Bank of Chicago.
Blanchard, O. and J. Wolfers (2000), The Role of shocks and Institutions in the Rise of
European Unemployment: The Aggregate Evidence, The Economic Journal, 110, pp.
C1-33.
Brealey, R. A. and E. C. Kaplanis (1996), The Determination of Foreign Banking Location,
Journal of International Money and Finance, 15, pp. 577-97.
Buch , C. M. (2000), Why Do Banks Go Abroad? Evidence from German Data, Journal of
Financial Markets, Instruments and institutions 9(1).
Diamond, D. (1984), Financial Intermediation and Delegated Monitoring, Review of
Economic Studies, 51, pp. 393-414.
___________ and P. Dybvig (1983), Bank Runs, Deposit Insurance and Liquidity. Journal
of Political Economy, 91, pp. 401-19.
___________ and R. Rajan (2000), A Theory of Bank Capital, Journal of Finance, 55,
pp.2431-65.
Dunning, J. (1988), The Eclectic Paradigm of International Production: a Restatement and
Some Possible Extension, Journal of International Business Studies, 19, pp. 1-32.
Focarelli, D., Panetta, F. and Salleo, C. (1999), Why do Banks Merge? Banca dItalia, Temi
di Discussione del Servizio Studi n. 361.
_________ and A. F. Pozzolo (2000), The Determinants of Cross-Border Bank
Shareholdings: An Analysis with Bank-Level Data from OECD Countries Banca
dItalia, Temi di Discussione del Servizio Studi n. 382.
Generale, A. and G. Gobbi (1999), Corporate Governance and Bank Profitability: Empirical
Evidence from the Italian Experience, in The Monetary and Regulatory Implications of
Changes in the Banking Industry, BIS Conference Papers, 7, pp.34-54.
Goldberg, L.G. and D. Johnson (1990), The Determinants of U.S. Banking Activity
Abroad, Journal of International Money and Finance, 9, pp. 123-37.
and A. Saunders (1980), The Causes of U.S. Bank Expansion Overseas: The
Case of Great Britain, Journal of Money Credit and Banking, 12, pp. 630-43.
(1981), The Growth and Organizational Form of Foreign
Banks in the U. S., Journal of Money Credit and Banking, 13, pp. 365-74.
Greene, W. H. (1997), Econometric Analysis, Basingstoke, MacMillan.

Grosse, R. and L. G. Goldberg (1991), Foreign Bank Activity in the United States: an
Analysis by Country of Origin, Journal of Banking and Finance, 15, pp. 1092-112.
IFC (1998), Emerging Stock Markets Factbook.
IMF (1997), Exchange Arrangements and Exchange Restrictions.
(1998), International Financial Statistics Yearbook.
(1998), Direction of Trade Statistics.
La Porta, R., F. Lopez-de-Silanes, A. Shleifer and R. W. Vishny (1998), Law and Finance,
Journal of Political Economy, 106.
Miller, S.R. and A. Parkhe (1998), Patterns in the Expansion of U.S. Banks Foreign
Operations, Journal of International Business Studies, 29, pp. 359390.
Mishkin, F. S. (2000), Prudential Supervision: Why Is It Important and What Are the
Issues?, NBER Working Paper n. 7926, Cambridge, Massachussets.
Morgan, D. P. (2000), Rating Banks: Risk and Uncertainty in an Opaque Industry, mimeo,
Federal Reserve Bank of New York.
Nigh, D., K.R. Cho and S. Krishnan (1986), The Role of Location-Related Factors in U.S.
Banking Involvement Abroad: An Empirical Analysis, Journal of International
Business Studies, pp. 5972.
OECD (1997), International Direct Investment Statistics Yearbook.
Pesaran, M. H., Y. Shin and R. P. Smith (1999), Pooled Mean Group Estimation of Dynamic
Heterogeneous Panels, Journal of the American Statistical Association, 94, pp. 621-34.
Petersen, M. A. and R. Rajan (1994), The Benefits of Lending Relationships: Evidence
From Small Business Data, Journal of Finance, 49, pp. 3-37.
Sagari, S. B. (1992), United States Foreign Direct Investment in the Banking Industry,
Transnational Corporations, pp. 93123.
Tschoegel, A. E. (1983), Size, Growth and Transnationality among the Worlds Largest
Banks, Journal of Business, pp.187-201.
Ursacki, T. and I. Vertinsky (1992), Choice of Entry Timing and Scale by Foreign Banks in
Japan and Korea, Journal of Banking and Finance, 16, pp. 405-21.
Williams, B. (1996), Determinants of the Performance of Japanese Financial Institutions in
Australia 1987-1992, Applied Economics, 28, pp. 1153- 65.
(1997), Positive Theories of Multinational Banking: Eclectic Theory versus
Internalisation Theory, Journal of Economic Surveys, 11, pp. 71-100.
(1998), Factors Affecting the Performance of Foreign-Owned Banks in
Australia: a Cross-Sectional Study, Journal of Banking and Finance, 22, pp. 197-219.
Yamori, N. (1998), A Note on the Location Choice of Multinational Banks: the Case of
Japanese Financial Institutions, Journal of Banking and Finance, 22, pp. 109-20.