North-Holland
Philip BOURKE
University College Dublin, Dublin 4, Ireland
This study reviews the performance of banks in twelve countries or territories in Europe, North
America and Aus~ra!ia and examines the internal and external determination of profitability. To
circumvent some of the difficulties in making comparisons between banks in different countries,
the concept of 'value added' is introduced. Results parallel those in domestic U.S. studies and
provide some support for the Edwards-Heggestad-Mingo hypothesis of risk avoidance by banks
with a high degree of market power.
1. Introduction
Many studies of the determinants of bank profitability in the United States
have been undertaken~ including those which have focussed on the relation-
ship between concentration and profitability and those which have examined
the possibility of expense preference behavior existing in regulated and
concentrated industries such as banking (see section 3). H_owever, the.re have
been only three major studies of international bank profitability [Revell
(1980), Short (!977, t979)'1. These works showed that it was possible to
conduct a meaningful analysis in spite of the substantial differences in
accounting practices and legal form between banks in various parts of the
world. Of relevance also is the increased interest internationally in the effects
of augmented competition and deregulation on banking systems and finan-
cial markets.
This paper has the objective of further examining the determinants of
international bank profitability and particularly of reviewing the relevance
of expense preference behavior theories [Edwards (1977)] in this context.
The Edwards-Heggestad-Mingo theory [Edwards and Heggestad (1973);
Heggestad and Mingo (1976)] that higher concentration in banking markets
encourages banks to hold less risky assets and to modify their behavior in
*The generous financial support of Allied Irish Banks for the research leading to this study is
warmly acknowledged, as are the helpful comments of J.R.S. Revell, Brock K. Short, W.K.
O'Riordan and R.P. Kinsella and an anonymous referee.
2. Data collection
The data is based on the financial statements of 90 banks each year in the
ten years from 1972 to 1981 in twelve countries or territories - Australia,
California, Massachusetts, New York, Canada, Ireland, England and Wales,
Belgium, Holland, Denmark, Norway and Spain. The banks included in the
sample were every bank in these countries which fell within the top 500
banks in the world in June 1980, ranked by total assets.
In so far as possible data were standardised to remove differences in local
accounting practices particularly in relation to the treatment of reserves.
Contrary to common practice, total assets are defined to include acceptances
on both sides of the balance sheet. This is largely a matter of convenience as
Spanish banks conduct a large part of their business through the acceptance
mechanism; my approach and the conventional approach are both subject to
disadvantages.
than by being reflected in higher profits. They suggest that a lower level of
service as measured in a variety of ways is detectable.
4.1. Regulation
In relation to regulation, for instance, there is an implied assumption that
within the U.S. the intensit/ of regulation is a constant; there may be a
prcsump~lt, n that regulation out~.~de the U.S. varies from country to country.
Accordingly, the literature has never adequately examined the consequences
of changes in the intensity of regulation. The empirical problems of doing so
are, however, daunting as regulation is not readily susceptible of comparative
measurement. I considered several approaches and the most promising was a
Delphi/Jury of Expert Opinion ranking of the intensivity of regulation on a
!imited scale. While it may be possible to locate experts who are familiar
enough with regulation in each of the sample countries to provide a
comparative ranking for a pa:ticular year, it proved impossible to do this in
a time series study extending over ten years.
Seminar participants have proposed that the severity of regulation may be
examined by constructing a matrix to investigate the presence or otherwise of
various restrictions in different countries - entry barriers, interest rate
restrictions, credit ceilings, et cetera. However, the direction of any of these
individual effects is unclear. For instance contestable market theory and
indeed regulation theory in general point to the importance of barriers to
entry in enhancing profitability while some of the other regulatory interven-
tions may depress profitability.
The issues of regulation, competitive behaviour and concentration are
related because as a referee of this journal has pointed out the source of the
correlation between profitability and concentration may be a correlation
between concentration ~.nd regulatory prntp.etic~n on banks and regulatory
protection and profitability. ! will return to the question of entry barriers
and related issues under the heading of Industry Structure but in the absence
of a clear theoretical model, a matrix or tabulation of various regulatory
interventions will assist us little in the attempt to measure the differential
effects of regulation across countries.
It has also been suggested that differences in the capital ratio could be
68 P. Bourke, ~a~k profitability
used as a proxy for regulation on ~he basis that the r~mrket would equalise
capital ratios for banks of the m~,gnitude and state, re of the banks in this
sample. However, this requires tMt market-derived cost of capital figures be
first obtained and excess costs of capital be computed for banks in each
country. Aliber (1984, p. 670) suggests the use of Q ratios in examining
changes in cost of capital for banks in different countries. (Q ratios relate the
market value of the firm to its bock value.) Goodman (1984, p. 685)
questions the conceptual appropriateness of the use of book value as a proxy
for replacement cost but, in any case, many of the banks included in this
survey are not publicly traded, usually because they are state owned, and no
market capitalisation figures exist.
Of greater immediate relevance are the expense preference theories of
Franklin Edwards (1974) who postulated that excess or supernormal profits
of regulated industries may be diverted away from net profit into sub-
optimal expenditure patterns related to management as oppposed to share-
holder preferenccs. There is also the possibility, relevant particularly in
countries where the banking industry is unionised, that the super-normal
earnings of firms in a regulated industry may be appropriated in the form of
payroll expenditure.
Two questions arise. What is the extent of regulatory barriers to entry? Does
the existence of concentration in itself constitute an effective barrier, particu-
larly in retail markets?
Looking firstly at the question of the existence of regulatory barriers,
paradoxically these barriers when they exist may well be more apparent than
real. For instance, while many countries in this survey maintained formal
entry barriers, particularly in relation to foreign banks, almost all countries
had ~xperienced a growth of potential competitors to traditionally organised
banks - b~,2-:,,~ ~ocleties in Britain, Ireland and Australia, Trustee Savings
Banks in i'.'ew Zealand, thrifts in the. United States, et cetera. Additionally,
there are very few examples of completely new entrants to banking markets
attempting to compete in the retail sector and in fact examples are scarce of
any form of retail activity transferring successfully across international
frontiers in recent years. [See Tschoegl (1987) for a comprehensive discussion
of the factors affecting foreign direct investment in retail banking. He
suggests that regulatory constraints are only one of many factors.]
Baer and Mote (1985) present evidence that concentration is higher
internationally than in the United States and within the United States is
higher in branching than in non-branching states. They also make the
important point that studies of market structure must adequately define the
relevant market which in the banking context includes non-bank financial
intermediaries. Their approach to market definition is similar to that adopted
in my work.
International evidence in the literature is also scant in relation to the
question of whether concentrated m~rkets themselves constitute a barrier to
entry. Ball and Tschoegl (1982, p. 417) suggest that concentrated markets
may offer a price umbrella under which new, fringe entrants may shelter, but
offer no empirical evidence.
Spiller and Favaro (1984) -'resent evidence from Uruguay of changes in
competitive behaviour among dominant banks when barriers to entry were
lowered and local non-bank financial intermediaries were allowed to enter
the market.
In summary, the present state of the literature offers little clearcut
guidance in relation tt~ the role of concentration in bank profitability
determination - in some ins~,,,nces data unavailability makes the testing of
some promising concepts impossible for international markets.
subsidiaries are both included in the data, the unconsolidated results from
the parent are used.
I tested the relationship betw~n profitability as defined in the several ways
described below and the following independent variables:
- Staff expenses, both in their own right and as a proxy for overhead
expenses generally, whose reliability in the data was not felt to be adequate.
- Capital ratios, i.e. capital including reserves as a percentage of total
assets.
Liquidity ratio, i.e. the ratio of liquid assets to total assets. The
-
reciprocal of this ratio can also be used as a proxy for the loan/deposit ratio.
Concentration ratio. The three bank concentration ratio is used, the
-
capital scarcity although it may have been more appropriate to use real
interest rates. It is possible that interest rates have a direct influence on
profitability and this will be tested in the context of return on assets as the
dependent variable.
- Market growth. The use of market growth as a variable is not
suggested extensively in the literature. However, Short found that asset
growth in individual banks was not significant - he had used it to control for
banks' managements who were growth, as opposed to profit, maximisers. It
is suggested that growth in total market may be considered as a potential
variable in the sense that an expanding market, particularly if associated with
entry barriers, should produce the capability of earning increased profits.
Accordingly, annual growth in money supply in each country is suggested as
an independent variable.
- Inflation. Revell (1979) has suggested that inflation may be a factor in
the causation of variations in baek profitability although this is not widely
discussed in the literature elsewhere. Its effect depends on the assumption
that wages and other non-interest costs are growing faster than the rate of
inflation, which is not unusual and, accordingly, the annual growth in the
consumer price index in each country is used as an independent variable.
(a) Return on capital. Net income before and after tax as a ratio of total
capital including all reserves. The ratio of after-tax profits to capital was
used by Short (1979). A further variable is defined as the ratio of net
income before tax to capital+total borrowings (as a proxy for sub-
ordinated loan stock which is used in substitution for equity capital).
(b) Return on assets. Net income before tax as a ratio of total assets.
(c) Value added return on total assets.
(0 The ratio of net income before taxes+staff expenses to total assets.
(i0The ratio of net income before taxes+staff expenses + loan losses to
total assets.
y = c + a t x t +azx2 +aaxa,...,anxn,
Table 1
Estimates of the relation between return on capital a n d selected independent
variables."
5. Resuhs
Findings are reported in tables 1 and 2.
Table 1 replicates Short's work to a greater or lesser extent depending on
the particular equation examined•
However, eq. (4) in table 1 is the exact equivalent of Short's sixth equation
(1979) which he expressed as follows (using the variable names in table 1):
ATCR=2.04-2.36GOVT~+O.O3CONC'+O.6INT" R 2 =0.52,
aSignificant at 5% level•
The almost total lack of correspondence between the present results and
those of Short are surprising and are difficult to explain. However, the
following comments may contribute to an understanding of the differences.
In addition to the inherent data collection problems of the present work
which have been described, Short faced further difficulties:
(a) Short's profit data related to a three year average rather thaa to the ten
year time span of the present work. However, eqs. (!)--(5) were re-
estimated on a year by year basis for years 1981, 1979, 1976 and 1973
with results in line with those of the time series with equally poor
explanatory power. Additionally, the effects of Short's averaging of profit
returns would be to reduce the variability in report profit figures a~Jd
thereby presumably provide more si~,~cai, t regression result: than in
mine.
76 P. Bourke, Bank profitability
(b) The data sources used were secondary, e.g almanacs as opposed to bank
financiai statements.
(c) While a greater number of countries are included, the number of banks
from each country in the sample was small. For instance, Belgium is
represented by two banks as opposed to eight banks in the present work.
More importantly, only five banking markets are common to both
(Belgium, Canada, Denmark, England and Wales and Holland) and
some important markets are present in ShoWs sample but not in mine
(France, Germany, Italy, Japan and Switzerland) and vice versa
(California, Massachusetts and New York). I understand from Dr. Short
that removing the above five countries (France, Germany, et cetera) frcm
,his sample substantially increases the mean and reduces the standard
deviation of his H values which may help to explain my lower R 2 adj.
figures in table !~
(d) ~veral countries are included in ShoWs sample where financial state-
ment information is notoriously unreliable, e.g. Germany, Italy and
Switzerland, while several of the U.K. banks enjoyed and practised the
privilege of hidden reserves. Of ShoWs sample of 60 banks, 21 banks are
subject to these problems.
The results sho~n in table 2 all relate to a~set (as opposed to capital)
based returns and, in general, show capital ratios, liquidity ratios and interest
rates as being positively related to profitability. The finding in relation to
capital ratios ,~s to be expected as, in accounting terms, capital represents a
'free' resource and Revell (1980) had noted an inverse relationship between
capital ratios and costs of intermediation. It is also possible to speculate that
well capitalised banks enjoy access to cheaper (because less risky) sources of
funds or that the prudence implied by high capital ratios is maintained in the
loan portfolio with consequent improvement in profit rates. The results in
relation to liquidity ratios are less expected as conventional wisdom is that
liquidity holdings (particularly if imposed by government) represent an
~xpense to banks.
As was expected, staff exp.~nses show an inverse if weak correlation with
pre-tax return on z~sets.
The results in relation to concentration require some consideration. In line
with findings in other parts of the ~iterature, concentration is shown to be
moderateiy and positively related to pre-tax return on assets. However, when
used in equat_ions having one of the measures of value added as dependent
variable, the sign of the relationship change~ to an inverse relationship. It
had been postulated that if, for instance, support were to be shown for the
expense preference theories, the sign of the relationship would remain
positive and the relationsh,~p strengthen in the case of the dependent variable
BTSETA (net income before ~ax + staff expenses as ratio of total assets). The
P. Bourke, Bank profitability 77
D~ml,al,I I I
I I I I I I I I
I I
,ID ,ID JD J~ ~L
0
.ID
S
I I I I I i I I
i 8 i
'i.@
~-
]-
78 P. Bourke, Baak peof~a~l~tF
6. C ~ ~
The findings of Short's study are not confirmed except in the most general
sense. The results are, however, in agreement with concentration and bank
profitability studies for the domestic U.S. market and support is found for
Me Edwards-Heggestad-Mingo hypothesis. No support is found for expense
preference expenditure theories.
References
Aliber, Robe.rt Z., 1984, International banking - A survey, Journal of Money, Credil, and
Banking 16, no. 4, part 2, Nov.
Baer. Herbert and Larry R. Mote, 1985, The effects of nationwide banking on concentration:
Evidence from abroad, Economic perspectives, Federal Reserve Bank of Chicago 9, no. 1,
Jan./Feb.
Ball, Clifford A. and Adrian Tscboegl, 1982, The decision io establish a foreign bank branch or
subsidiary: An application of binary classification procedures, Journal of Financial and
Qualitative Analysis XVII, no. 3, Sept.
Baumol, William J., John C. Panzar and Robert D. Willir,, 1982, Contestable markets and the
theory of industrial structure (Harcourt Brace Jovanovich, New York).
Bell Frederick W. and Nell Mushy, 1969, Impact of market structure on the price of a
commercial banking service, The Review of Economics and Statistics, May.
Benston, George J., Gerald A. Hanweck and David B. Humphrey, 1982, Scale economies in
banking, Journal of Money, Credit, and Banking XIV, no. 4, part 1, Nov.
Chappell, William F. and Rex L. Cottle, 1984, Sources of concentration - Related profits,
Southern Economic Journa| 51, no. 4.
P° Bc~rke, ~ ~,~tab~ity