10 1108@03074350910931762 PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

The current issue and full text archive of this journal is available at

www.emeraldinsight.com/0307-4358.htm

MF
35,3
Measuring and explaining the
impact of vertical product
differentiation on banking
246 efficiency
Ana Lozano-Vivas
Dpto. de Teorı́a e Historia Económica, Universidad de Málaga,
Málaga, Spain
Abstract
Purpose – The paper attempts to analyze vertical product differentiation as a strategy pursued by
European banks seeking greater market power and higher reputation for quality, and to examine
whether this entails losses in banking efficiency.
Design/methodology/approach – First, the empirical analysis seeks to demonstrate whether
borrowers at banks in Europe are willing to pay a premium to operate with banks that attempt to
increase their reputation for quality in the market, i.e. whether banks use quality to vertically
differentiate and so soften competition. To test such hypothesis requires us to define an empirical
model with variables that describe certain characteristics of banking quality as explanatory variables
of the loan interest to the market interest rate margin. This model is estimated by two stage least
squares. Second, the paper seeks to test whether the market power derived from vertical product
differentiation (quality reputation) prevents banks from operating efficiently. To test this hypothesis
first we estimated cost efficiency taking into account bank risk preferences and then we define an
empirical model that relates the results on efficiency with the margin of interest loan rate over the
market interest rate.
Findings – The results show that less competition, deriving from a bank’s ability to differentiate its
services from those of its rivals through quality, is positive because it helps to provide a more stable
banking system. Moreover, the banking market power generated by investing in quality does not
prevent banks from operating efficiently from a production point of view.
Research limitations/implications – The findings are consistent with the view that European
banks soften competition by being more stable, and this does not prevent cost efficiency. So it seems
that the regulatory authorities should improve their solvency measures since borrowers’ preferences
are to maintain relationships with non-fragile banks, and on the other hand banks’ risk preferences
seem to be to look for sound borrowers.
Practical implications – Frontier cost efficiency scores that account for bank’s risk preference are
able to be related with customer preferences based on the model of the industrial organization (10)
based on vertical product differentiation in banking.
Originality/value – This is the first paper that relates vertical product differentiation with the
results obtained from the literature on x-efficiency. It is also the first paper that studies the impact of
banking market power jointly with cost efficiency in social efficiency when market power comes as
result of investing in reputation for banking quality.
Keywords Banking, Product differentiation, Quality, Cost effectiveness, Europe
Paper type Research paper

The author thanks the anonymous referee, as well as Miguel Angel Durán and seminar
Managerial Finance
Vol. 35 No. 3, 2009 participants at the IV North American Productivity Workshop, and III Congreso de Eficiencia
pp. 246-259 and Productividad EFIUCO for helpful comments. Financial support from the Ministry of
# Emerald Group Publishing Limited
0307-4358
Education and Science-FEDER within the framework of research program SEJ2005-08598 is
DOI 10.1108/03074350910931762 gratefully acknowledged by the author.
1. Introduction Vertical product
The deregulation and liberalization of financial services in the European Union (EU),
the establishment of Economic and Monetary Union, a higher financial culture and
differentiation
technical progress in the European banking industry, over the last two decades, have
opened up an important debate on the economic role of market competition in the
banking industry in Europe. In this regard, the last few years have seen the
publication of a huge number of research studies analyzing the consequences of these
changes on the competitive level of the banking industry and its concentration ratio
247
(Hannan, 1991; Molyneux et al., 1994; De Bandt and Davis, 2000; Bikker and Haaf,
2002). Independently of the methods used to assess competition and concentration
levels in the banking industry, these papers have in common the goal of testing
whether the liberalization of the banking system has led to an increase in banking
competition. The relevance of this hypothesis is justified by the conventional argument
that improvement in competition should produce social welfare benefits, which are
assumed to lie at the heart of the market competition theory of standard industrial
organization (IO). Furthermore, in a competitive environment, banks need to be more
efficient if they want to survive in the long run. If we export this principle to the
banking industry it appears, mutatis mutandis, that there is also a desire for greater
competition there since banks exercising market power should obtain extraordinary
income, particularly by charging higher interest rates to their borrowers and lower
interest rates to depositors.
However, given the particular characteristics of the banking business and the
essential role of banks in the economy, it is not obvious that the standard hypothesis
set out by the IO, which is perfectly suited to other production activities, should be
exported per se to the banking industry since one requirement of the banking industry
to generate a positive impact on the economy is banking stability. Allen and Gale (2000)
prove that one of the possible effects of increased competition in the banking sector is
that banks may assume higher risks in their investments in order to offset the loss of
profits entailed by increased competition, leading to greater instability. With this
premise in mind one of the goals of this paper is to evaluate whether banks are able to
exercise market power when that power is oriented towards increasing the quality of
banking services and products, and thus benefits borrowers. Actually, borrowers
should not object to paying higher interest rates if they obtain higher-quality banking
services and products. In others words, the first key point in this paper is to attempt to
empirically test whether borrowers are willing to pay more for higher-quality banking
services and products. To deal with this question we go along with the model
developed by Kim et al. (2005), which analyzes the strategies pursued by Norwegian
banks to differentiate their services and increase the interest rates that banks can
charge their borrowers in equilibrium.
Given that services and product differentiation lead to softer competition, i.e. less
competitive pressure, the incentive of banks to offer higher quality in their banking
services and products should be expected to result in a decline in the effort of managers
to maximize economic efficiency. In the words of Hicks (1935), ‘‘the best of all monopoly
profits is a quiet life’’. That is, besides the higher prices and the lower production at
equilibrium that are characteristic of a market structure with market power, lack of
interest on the part of managers has been pointed out as an additional cost per unit of
output in this type of market (Berger and Hannan, 1998). The second goal in this paper
is to analyze that point. In particular, we are interested in obtaining an overview of the
impact generated by market power on the economic efficiency of the banking industry
MF in order to test whether lower competition, due to banking services and product
differentiation, lead to a bank industry having a quiet life. Actually, there is a major
35,3 economic reason to be interested in investigating this issue: the final upshot of market
power together with lower banking efficiency will provide higher penalties for the
payment of borrowers for banking services. We pursue this enquiry by relating the
results derived from banking market power and the ability of banks to differentiate and
provide higher-quality services with the current cost efficiency of banks. The
248 stochastic frontier approach is used to determine the cost efficiency which is needed to
establish such a relationship.
We conduct our empirical analysis on the European banking industry to illustrate the
areas in which banks may find it most profitable to differentiate and soften competition,
and to analyze whether the increase in market power has a positive or negative impact on
banking cost efficiency. Since the European banking industry has undergone a major
deregulation and liberalization process, it would seem to be an interesting subject for
study with a view to learning how European banks behave in terms of competition and
economic efficiency as they face this new environment. Even though we follow the paper
by Kim et al. (2005) in addressing our first goal, we apply our empirical exercise to analyze
vertical banking product differentiation in the European banking industry rather than
using a specific country’s banking industry, (as Kim et al., 2005, do with Norway).
Moreover, in contrast to Kim et al. (2005), we take a second step in the analysis and attempt
to examine whether or not the market power derived from vertical product differentiation
prevents banks from operating efficiently. Our efforts to prove this second hypothesis lead
us to relate the results of the theoretical model of vertical product differentiation (Kim et al.,
2005) with the results obtained from the literature on X-efficiency. To our knowledge there
are no previous papers that address this issue. However, there is research that suggests
ways in which these two literatures may be related.
The rest of the paper has the following structure. Section 2 examines the relevant
questions that we are attempting to test, paying special attention to the theoretical
model on which our idea is based. Section 3 presents the empirical model used to test
the hypothesis shown in section 2. The empirical results are presented in section 4, and
section 5 concludes.

2. Product differentiation and production efficiency: hypothesis


As indicated in the previous section, the first hypothesis to be tested is whether
borrowers are willing to pay more for higher-quality banking services and products,
which would lead banks to increase their market power. However, such market power
has to be understood as an incentive to banks to offer better information to their
customers and higher-quality banking services and products. To give consistency to
this hypothesis we base it on the theoretical model of Kim et al. (2005), who analyze
endogenous vertical product differentiation in banking. In particular, using a stylized
two-stage model where in the first stage banks choose quality variables
simultaneously, and in the second stage they choose interest rates simultaneously, and
borrowers accept an offer from one of the banks, Kim et al. (2005) analyze whether a
bank’s ability to avoid losses, its capital ratio or its size can be used as strategic
variables to make banks different and increase the interest rates that they can charge
their borrowers in equilibrium. This model allows us to analyze what quality
characteristics (loan loss provisions, equity ratio, size, etc.) banks choose in order to
differentiate themselves from competing banks. In that sense, it is possible to say that
if banks differentiate themselves in such a way it is because they expect to find
borrowers that value high-quality services and are willing to pay for them. Economic Vertical product
banking theory proposes two reasons why borrowers would be willing to pay for
higher quality:
differentiation
(1) Because banks are able to give signals as to the quality of their loans. For
instance, in the bank-borrower loan relationship, when the bank agrees to
extend a loan to a borrower, the latter is required to give proprietary
information (within the monitoring and screening process). Once the loan is 249
approved borrowers receive not just the funding but also additional value from
the loan, given the positive influence that the approval of the loan has on the
market value of the borrower’s claims. This fact creates a lender certification
for investors since they broadcast the fact that the lender knows that the
borrower can be trusted to pay back the loan. Cook et al. (2003) call this ‘‘the
lender certification’’, and it is akin to a credit rating. This certification benefit
leads us to assume that just as lenders look for sound borrowers, borrowers
look for sound lenders. That is, borrowers will seek a loan from a reputable
lender (lenders capable of favourably influencing market pricing). However, the
lender should attempt to benefit from borrowers by charging borrowers a
certification premium embedded in the loan rate as a bank monitoring method.
In that case, borrowers will be willing to pay a premium loan rate to lenders
matching the overall value of the certification. Jonshon (1997) uses loan loss
provisions to proxy reputation in bank monitoring abilities, arguing that a
change in loan loss provisions indicates a change in managerial evaluation of
loan portfolio quality and a change in monitoring and screening abilities. That
is, if banks have low loan loss provisions then they have a high-quality loan
portfolio, in which case borrowers should be willing to pay a higher interest
rate to those banks. This is known in banking literature as the ‘‘certification
effect’’ (Cook et al., 2003; Lummer and McConnell, 1989; among others).
(2) Because banks have enough potential to generate loans in the future. Since a
bank which actually lends to a customer learns more about that borrower’s
characteristics than other banks do, when customers wish to refinance their
loans they stay with the same bank, not simply because the bank treats them
particularly well, but because high-quality borrowers are informationally
captured or locked-in (see Sharpe, 1990 and Detragiache et al., 2000). Actually, a
borrower who switches to another bank for refinancing faces an adverse
selection problem, as the new bank, with which he/she has no prior
relationship, will suspect that the application is a lemon (Detragiache et al.,
2000). Given the costs for borrowers entailed by switching to another bank
(Kim et al., 2003) they value those characteristics of banks that produce
information about their ability to generate loans in the future and, therefore, to
maintain stable relationships with them. That is, they evaluate bank fragility,
namely the propensity of banks to experience liquidity problems that force
them to cut back their loan portfolio. The ability to prevent fragility depends on
the solvency and the diversification capacity of banks: well-diversified, well-
capitalized banks are less likely to incur in future losses. This is known in
banking literature as the ‘‘refinancing effect’’ (Chenmanur and Fulghieri, 1994;
Detragiache et al., 2000; Kim et al., 2005; among others).
On the basis of the fundamentals of the banking economy, and using the results given
by the theoretical model developed by Kim et al. (2005), we attempt to test whether
MF banks are interested in investing to obtain a higher reputation for quality so that they
can differentiate themselves from competing banks and therefore reduce competition.
35,3 In others words, the first hypothesis to be tested can be formulated as follows:
H1. Borrowers pay a premium to borrow from banks with a higher reputation for
quality. That is, from those banks that gives a certification (low loan-loss
provisions) or banks that give a guarantee of their capacity to offer future
250 refinancing if needed. If such is the case, then banks face a market discipline
induced by borrowers that gives them market power.
To accept this hypothesis is to suggest that the banking industry is characterized by
the presence of firms with market power. In contrast with the negative appraisal that is
provided by the standard IO, but in consonance with recent trends in banking theory,
such a result encompasses social welfare gains. At the cost of increasing market power,
the differentiation allows banks to charge higher interest rates on loans, i.e. it permits a
rise in prices which, however, borrowers are willing to pay in order to get higher
quality. However, on the positive side, the strategy of differentiation discussed
increases stability in the banking industry because it gives banks the opportunity to be
more transparent and to incur less risk. Overall, this asset side market discipline made
by borrowers is beneficial for social welfare.
The second phase of this research entails an in-depth analysis of the impact that the
exercise of market power may have on an additional dimension that affects costs in
society: bank production inefficiency[1]. Notice that in this regard the acceptance of H1
means that vertical product differentiation implies greater market power, with the
inherent disadvantage but also with greater market discipline, which should give rise
to social efficiency gains. However, economic theory suggests that market power
generates additional social costs as a result of higher prices and lower production
levels. Moreover, following Hicks (1935), the lower competition pressure derived from
market power may affect the efforts of managers, given inefficiency in the production
process. Related to this additional effect, the second hypothesis that we will attempt to
test in this paper is as follows:
H2. The market power derived from vertical product differentiation (quality
reputation) prevents banks from operating efficiently.
Seeking to prove this second hypothesis leads us to relate the results of the theoretical
model on vertical product differentiation (Kim et al., 2005) with the results obtained
from the literature on X-efficiency. To our knowledge there are no previous papers that
address this issue. However, there is research that suggests ways in which these two
literatures may be related.
Without abandoning the fundamental reason that leads to H2, i.e. exploring the
additional social costs, if any, that may be generated if banks operate with market
power, observe that an empirical test in favour of H1 could be seen as follows: if banks
try to reduce their loan-loss provisions and/or improve their solvency and diversify
their assets, then they can find borrowers willing to pay a higher interest rate for loans
to obtain some level of guarantee as to the quality of banking services and products.
The efforts made by banks to achieve this objective may induce them to incur higher
costs but, at the same time, banks may see current profits become more stable, since
with this practice they are assuming less risk.
The fact that risk forms an intrinsic part of banking activity turns it into an essential
element of production decisions (Modigliani and Miller, 1958). On that basis, the mid-1990s
saw a huge number of research studies that introduced risk into the analysis of the Vertical product
banking production process (for instance, Hughes, 1999; Hughes et al., 1995, 1996; Hughes
and Mester, 1993; among others). The fundamental contribution of these studies is to
differentiation
incorporate the risk preferences of banks into to their production technology and thus
obtain unbiased measures of frontier efficiency that avoid classifying as inefficient those
banks that are risk-adverse and have a different input mix. In particular, these studies
suggest the introduction of solvency and service quality indicators into the cost function –
particularly proxied by equity and loan-loss provisions, respectively – to reflect risk 251
preferences and to avoid measuring cost increases in the production process due to higher
risk equal than inefficiency and, on the other hand, to avoid labelling as efficient banks that
are not monitoring their loans. As can be seen in the following section, and as pointed out
above, the indicators used to control for risk preferences in measuring the efficiency of
banking production process are the same as those that should be used to evaluate the
incentives that banks have to invest in obtaining a higher reputation for quality, and the
market power that this practice entails.

3. Method
In this section, we present the empirical models that we use to empirically test the
hypotheses in section 2.
The first empirical analysis seeks to demonstrate whether borrowers at banks in
Europe are willing to pay a premium to operate with banks that attempt to increase their
reputation for quality in the market (H1). This premium can be justified in economic
terms because banks that opt for quality in this way will incur higher costs and hence, at
the same time, will invest in less risky loan portfolios, thus reducing their probability of
obtaining higher profits, but they will be more stable. Following Kim et al. (2005), testing
H1 requires us to define an empirical model with variables that describe certain
characteristics of banking quality. In particular, with the goal of introducing into the
empirical model the precise characteristics of the theoretical model, variables that enable
the certification and refinancing effects to be tested for must be defined. The empirical
model to be estimated is the following:

X
3 X
3
si;c;t ¼ 1 si;c;t1 þ j vðqj Þi;c;t1 þ j gðqj Þi;c;t1 þ xi;c;t þ c þ t þ i þ "i;c;t
j¼1 j¼1

ð1Þ

This function relates loan interest to the market interest rate margin, s, of bank i in
country c at time t, with variables that describe banking quality characteristics, v(qj). In
the framework of the theoretical model, and in consonance with the normal practice
in banking theory of using loan-loss provisions as a proxy to reveal a bank’s ability to
avoid losses, the amount of that provision will be used as a proxy of quality. The other
variables used as indicators of quality are the equity ratio (as a proxy of solvency) and
total assets (which give information about the diversification level of banks). Since the
quality variables are those that borrowers choose as signals when deciding to operate
with a particular bank, we introduce those variables as their relative value with respect
the market’s median value (i.e. vðqj Þi;c;t ¼ qji;t =mediani2c qji;t ), following Kim et al. (2005).
Furthermore, the lag of the dependent variable is used jointly with country
dummies, c, time,  i, and bank dummies, vi, to control for characteristics of the interest
rate margin that could be due to risk preferences of borrowers (Kim et al., 2005). As an
MF additional control variable the operating cost, x, is used to evaluate banks’ ability to
pass on their higher costs to borrowers through prices in imperfect competition. Where
35,3 , ,  and  are the parameters to be estimated.
The theoretical model allows us to measure the impact of quality on the interest rate
margin and the impact of the degree of competition to be monitored. As indicated
above, product and service differentiation triggers smooth competition. To measure
this last factor we introduce into the empirical model not only the quality control
252 variables shown above but also the Gini index of those variables, g(qj)[2]. According to
the nature of this index as a measurement of dispersion, its interpretation in this
particular case should be that a higher dispersion of banking quality indicators
reduces the increase in competition due to market liberalization. Thus, it allows banks
to increase interest rates on their loans.
Observe that the quality variables and the Gini index of those variables are taken
with lags. This means that the decisions of borrowers about whether to contract
services with one bank or another are based on the information published in the banks’
yearly statistics (Kim et al., 2005). The variables introduced into Equation (1) and their
expected sign are shown in Table I.

Variables Definition

si;c;t Loan interest rate margin


si;c;t1 Lag of loan interest rate margin
Quality variables, vðqj Þi;c;t1
vðq1 Þi;c;t1 Total assets of bank i in country c at the end of year t1
vðq2 Þi;c;t1 Equity ratio of bank i in country c at the end of year t1
vðq3 Þi;c;t1 Loan-loss provisions over total loan-loss of bank i in country
c at the end of year t1
Gini index of the quality variables, gðqj Þc;t1
gðq1 Þc;t1
gðq2 Þc;t1
gðq3 Þc;t1
Control variables (xi,t, dummies)
xi;c;t Operating cost over total loans of bank i in country c for
year t
c ; t ; i Country, year and bank dummies
Expected sign of the variables (certification effect vs refinancing effect)
Variables Certification effect Refinancing effect
Quality variables, vðqj Þi;c;th
vðq1 Þi;c;th Not significant þSignificant
vðq2 Þi;c;th Not significant þSignificant
vðq3 Þi;c;th Significant Significant
Gini index of the quality variables, gðqj Þc;th
gðq1 Þc;th þSignificant
gðq2 Þc;th þSignificant
gðq3 Þc;th þSignificant þSignificant
Table I.
Variables used in the Control variables (xi,t)
estimation of Equation (1) xi;t þSignificant þSignificant
The model should show whether borrowers are willing to pay a premium over the loan Vertical product
interest rate to obtain higher quality, i.e. whether banks use quality to vertically differentiation
differentiate and so soften competition. That is, in the framework of H1, which includes
the certification and refinancing effects, as loan loss provision seems to be a quality
proxy for a bank’s ability to screen and monitor, i.e. to avoid losses (or in the opposite
case the bank’s willingness to take on risk in its loan portfolio), if banks have low loan
loss provisions then they have a high-quality loan portfolio, in which case borrowers
should be willing to pay a higher interest rate to those banks. If the estimated
253
parameter of loan loss provision has a negative sign and is statistically significant and
the estimated parameter of the rest of the quality variables is not statistically
significant then the certification effect is accepted within H1.
However, loan loss-provision has as additional effect on the solvency of the banks
and hence on their capacity to refinance borrowers in the future. Considering this fact,
along with the fact that the ability to prevent fragility and to guarantee the generation
of loans in the future depends on the solvency and the diversification capacity of banks,
well-capitalized (higher equity) and well-diversified (bigger) banks and those with
lower loan loss-provisions will comprise a reputation brand for borrowers in terms of
their ability to generate loans in the future and, therefore, to maintain stable
relationships with borrowers, if borrowers are willing to pay. Thus, a statistically
significant positive sign of the estimated parameter of equity and asset jointly with a
statistically significant negative sign of the estimated parameter of loan-loss provision
will imply the acceptance of H1 in terms of the refinancing effect. Once we test this first
hypothesis it should be possible to evaluate banking efficiency taking into account the
bank’s risk preferences using the stochastic frontier approach.
As pointed out above, taking the efficiency assessment given risk preferences as a
preference pattern allows us to evaluate precisely H2, which seeks to test the type of
relationship between banking efficiency and market power of banks obtained as a result of
investing in high-quality banking services and outputs. In the end, the reputation obtained
by providing quality is linked to the same variables defined to monitor the risk level of
banks. So to establish a strong relationship the efficiency with which banks operate must
be measured taking into account their risk aversion. Following Mester (1996), this
requirement can be met by introducing into banking technology the characteristics that
define quality – which are related, as pointed out above, with risk. Then cost efficiency is
measured using the stochastic frontier approach proposed by Aigner et al. (1977). Cost
efficiency provides a measure of how far a bank’s observed cost is from what a best-
practice bank’s cost would be for producing an identical output bundle under comparable
conditions. The cost function to be estimated can be written as follows:

ln Cit ¼ ln C ð yit ; wit ; qit ; kit ; C ; t ; i ; BÞ þ uit þ vit ð2Þ

where y is the banking output vector, which comprises loan and other earning assets;
and w is the input price vector, which comprises labour, physical capital and deposits.
The price of inputs is defined as the expenses arising from labour, other operating costs
less labour expenses and financial cost divided by total assets, fixed assets and total
deposits, respectively. Moreover, q and k are indicators that reveal banking risk
preferences, q defines loan-loss provisions and k defines equity. We estimate a
common frontier for the whole country sample and introduce country dummies, c,
time,  i and bank dummies, vi. The country dummies in a common frontier capture the
environment of each country where the banks are operating (e.g. economic, regulatory
MF and/or institutional characteristics), while the time trend captures technical change.
35,3 Finally,  is a vector of the parameters to be estimated. The total cost, which is the
dependent variable, is defined as the sum of financial plus operating costs. The error
term is treated as a composite error term where v represents standard statistical noise
and u captures inefficiency. Following Greene (1990) we assume a gamma distribution.
The individual values of inefficiency for each bank are calculated using the
distribution of the inefficiency term conditioned to the estimation of the composite
254 error term, as per Jondrow et al. (1982). As usual in efficiency literature, the translog
cost function is used to specify the cost of banking technology, with the standard
assumption of linear homogeneity and symmetry. The stochastic cost frontier is
estimated using the LIMDEP econometric program developed by Greene.
Once the efficiency level is estimated, we proceed to test H2. The examination of this
hypothesis requires us to define an empirical model that relates the results on efficiency
with the margin of interest loan rate over the market interest rate.
To that end, we first specify and estimate a regression to establish the relationship
between cost efficiency and the margin of the loan interest rate over the market interest
rate (a higher margin implies greater market power). So the regression to be estimated
is the following:

eu
i;c;t
¼
si;c;t þ c þ t þ "0i;c;t ð3Þ

If the sign of the estimated parameter of the loan interest rate margin over the market
interest rate is negative (positive) and significant then we will accept (reject) the quiet life
hypothesis. An interesting empirical exercise would be to analyze the relationship of the
cost efficiency with all the variables used as explanatory variables in Equation (1) in order
to check whether those variables hold the same significance and sign that obtained when
the dependent variable was the margin of the loan interest rate over the market interest
rate. In this way it would be possible to evaluate the impact of market power due to vertical
product differentiation on banking efficiency. The regression to be estimated is as follows:

X
3 X
3
eu
i;c;t
¼ 01 si;c;t1 þ j0 vðqÞji;c;t1 þ j0 gðqÞj;c;t1 þ 0 xi;c;t þ c þ t þ "00i;c;t ð4Þ
j¼1 j¼1

If the significance and sign of the explanatory variables in Equation (4) are the same as
in Equation (1) then it should be concluded that the preferences of the borrowers are in
line with the decision of the banks about their risk preferences.
Since the dependent variable in Equations (3) and (4), by construction, takes a value
between 0 and 1, the regression model estimated by least squares (LS) generates biased
results. We use the Tobit procedure to estimate this equation.
The empirical analysis is applied to the European banking industry. In particular, the
sample used covers nine EU countries: Austria, Belgium, Denmark, France, Germany,
Italy, Portugal, Spain and the UK, over the period 1998-2003. The sample comprises 2,110
commercial banks. The statistical information needed for the empirical analysis comes
from the balance sheet and cost and revenue information of the BankScope database.

4. Empirical results
In this section, we report our empirical results. The presentation follows the same order
as the exposition of the hypothesis to be tested presented in section 2. Thus, first we
show the results obtained for H1, where we attempt to test whether borrowers are Vertical product
willing to pay for reputation for quality when European banks use that reputation as a
strategy to differentiate their services and output from those of their rivals. That requires
differentiation
us to estimate Equation (1). As pointed out in section 3, testing H1 implies accepting
(rejecting) the certification or refinancing effects, and whether or not those hypothesis of
the certification and refinancing effects is accepted depends on the sign and significance
of the coefficients estimated. Accepting/rejecting one effect or the other gives information
about the preferences of borrowers that reveals whether they prefer to pay a higher
255
interest rate due to the certification effect or to the refinancing effect. That is, empirical
results in favour of either of the two effects imply approval of H1.
The model presented in Equation (1) is estimated by two-stage least squares. The
results of the estimation are in Table II. They show that all the quality variables and
their corresponding Gini indexes are significant. Considering the expected sign and
significance that the estimated coefficient of each explanatory variable would need in
order to accept the certification or refinancing effects shown in Table I, these results
suggest that borrowers are willing to pay a higher loan interest rate if they borrow
from banks that provide guarantees of high-quality loans. In particular, related to
Table I, the results are consistent with the refinancing effect. That is, the preferences of
banking customers on quality are related more to the capacity of banks to refinance the
needs of borrowers in the future than with the quality of the loan portfolio of the banks.
The results obtained are in line with those obtained by Kim et al. (2005) in terms of
acceptance of H1. However, they find results that support the certification effect, rather
than the refinancing effect, for the case of Norwegian banks.
The empirical results suggest that banks in Europe attempt to use the vertical
product differentiation strategy to obtain a good reputation for quality in regard to
their banking services and output, and thus seek to soften competition from rival
banks. That is, they acquire a market power that turns out to be great enough for
borrowers to be willing to pay a premium over the loan interest rate in exchange for
assurances that they are working with banks with adequate levels of solvency and
diversification. So in the European banking industry there seems to be a trade-off

Variables Coefficients effect Refinancing

ln si;c;th 0.0687* (0.0142)


Quality variables, vðqj Þi;c;th
ln vðq1 Þi;c;th 0.1004* (0.0176) þSignificant
ln vðq2 Þi;c;th 0.0628* (0.0157) þSignificant
ln vðq3 Þi;c;th 0.0164* (0.0080) Significant
Gini index of the quality variables, gðqj Þc;th
ln gðq1 Þc;th 0.1286* (0.0282) þSignificant
ln gðq2 Þc;th 0.1835 (0.1033)** þSignificant
ln gðq3 Þc;th 0.1058* (0.0421) þSignificant
Control variables (xi, t, dummies)
ln xi;t 0.10535* (0.0130)
i ; c ; t In Table II.
R2 adj. 0.8868 Estimated results of the
certification effect vs. the
Note: (*) Standard error that allows the significance of the coefficients estimated to be calculated refinancing effect
MF between competition and banking stability. Particularly, the results show that
borrowers’ appreciation of banks which invest in taking care to avoid fragility (higher
35,3 solvency, more diversified and lower losses) works as an important disciplinary
instrument, inducing banks to be more stable. These results should, on average, affect
social welfare positively given that a stable banking system stimulates the economy
(Allen and Gale, 1997).
Having analyzed the preferences of bank customers on quality and the impact of
256 their preferences on market power, we now assess what the banks’ preferences on
quality are. As pointed out in the previous section, the banks’ preferences are revealed
through the cost that banks support in their production process, and thus through
frontier efficiency. To address this analysis it is necessary to introduce into the
production technology variables that define the risk preferences of banks – such
variables are closely linked with banks’ quality preferences.
Table III shows the results obtained, on average, when the cost efficiency of the total
sample of banks is estimated, with and without the risk preferences of the banks. It can
be observed that the results are more statistically robust when quality variables are
introduced into banking technology.
Once we have evaluated cost efficiency taking into account banks’ risk preferences,
and given the information obtained in relation to the incentive for banks to invest in
quality to gain market power, these results can be used to test H2. That is, given banks’
risk preferences we can ask whether it is possible for banks to operate efficiently if they
have market power as well as if they operate efficiently as a result of differentiating
through high-quality banking and services.
To answer this question, first we estimate Equation (3), which relates banking
efficiency to market power. Table IV shows the results of this empirical analysis. The
first column presents the results when the estimation of inefficiency levels is used as a
function of the loan and market interest margin, and the country and time dummies.
Given that the estimated coefficient is positive and significant, the results suggest that
market power is associated with higher levels of cost efficiency, so the ‘‘quiet life’’
hypothesis is rejected. It seems that managements achieve cost efficiency even when
banks have market power. Since market power allows banks to enjoy greater profits,
the results suggest that this situation might create incentives for banks to behave
prudently (enhancing bank stability). This more prudent behaviour is conducive to the
selection of less risky activities with costs of monitoring (Petersen and Rajan, 1995),
therefore increasing cost efficiency leads to improved social welfare since those lower
costs should entail lower interest rate for borrowers.
Given the above results, it would be interesting to perform a second exercise to
check whether the cost efficiency of banks is related with all the explanatory variables
used to test whether market power exists and to determine bank customers’
preferences in terms of quality. That requires us to estimate Equation (4). This exercise

Translog cost function With quality variables (risk preferences) Without quality variables

Log likelihood 1,389.09 1,137.47



Mean E ðui ="i Þ 0.2614 0.2976

Table III. Min E ðui ="i Þ 0.0159 0.0198

Cost efficiency Max E ðui ="i Þ 0.8308 0.7249
Variables Margin quality Equation (3) Reputation for Equation (4) Vertical product
differentiation
ln si;c 0.1256* (0.0292)
ln si;c;th 0.1445* (0.0297)
Quality variables, vðqj Þi;c;th
ln vðq1 Þi;c;th 0.0019 (0.0021)
ln vðq2 Þi;c;th 0.9390* (0.1669) 257
ln vðq3 Þi;c;th 0.0891* (0.011)
Gini index of the quality variables, gðqj Þc;th
ln gðq1 Þc;th 0.2611 (0.2313)
ln gðq2 Þc;th 0.0413* (0.00546)
ln gðq3 Þc;th 0.0133* (0.00048)
Control variables (dummies) Table IV.
c ; t in in Relationship between
Log-likelihood 146.33 135.02 cost efficiency (eu) and
loan interest margin and
Note: (*) Standard error that allows the significance of the coefficients estimated to be calculated reputation for quality

will reveal whether the more efficient banks, taking into account their risk aversion, are
those with higher solvency and diversification and lower loan losses levels, or only
those with lower loan losses levels (refinancing versus certification effect). If we find
exactly the same relationship between the explanatory variables and cost efficiency as
obtained with the margin of the loan interest rate over the market interest rate, then it
should be concluded that the more efficient banks are those that are more stable, which
is in line with the borrowers’ preferences found for the refinancing effect.
The estimation of Equation (4) is shown in the second column of Table IV.
The results reveal that the signs of all variables are the same as those obtained when
the interest rate margin is used instead of cost efficiency as the dependent variable. The
main conclusion that can be obtained from these results is that given the willingness of
bank customers to pay, those banks that invest in quality are more efficient and, at the
same time, have greater market power.

5. Conclusions
The main goal of this paper is to analyze in depth the benefits and costs of having
market power in the European banking industry. In particular, we address a problem
that has not been covered in previous studies of banks: whether greater market power
derived from the strategic behaviour of banks due to vertical product differentiation
decreases the cost efficiency of those banks. Moreover, using risk preferences on quality
of customers and banks as a link, it has proved possible to associate this problem of
vertical product differentiation in banking with the literature on production efficiency.
The most significant conclusion is that reduced competition derived from banks’
ability to differentiate their services from their rivals through quality is positive
because it helps to provide a more stable banking system. Banks seems to be aware of
this issue and the most solvent, most diversified banks are the most efficient.
Therefore, in general terms, the standard competitive paradigm defended by the IO
does not fit the banking industry. The market power generated by investing in quality
does not prevent banks from operating efficiently from a production point of view.
MF Moreover, the results suggest that at the same time as they strive to improve their
35,3 cost efficiency, banks in Europe adopt strategic behaviour, the quality of their services
improves and products and competition from rival banks is reduced. Our results
suggest that the banks that enjoy greater market power are subject to pressure to
increase the quality of their services but since that quality provides them with stability
it then leads to lower monitoring costs and increases their cost efficiency, which is a
positive effect for social welfare.
258 The policy conclusion is that European banks soften competition by being more
stable, and this does not prevent cost efficiency. So it seems that the regulatory
authorities should improve their solvency measures since borrowers’ preferences are to
maintain relationships with non-fragile banks, and on the other hand banks’ risk
preferences seem to be look for sound borrowers.

Notes
1. Just as the existence of market power, in general, implies a loss of welfare, the
inefficiency of banks is also a cost for society since banks establish their margins as a
function of the operating costs that they have to bear, so borrowers in banking products
and services have to pay the costs of the operational inefficiency of banks.
2. The Gini index of each quality variable has been calculated as follows:

1 2 X
gðqj Þc;t ¼ 1 þ  2 rqi;t
nc nc qc;t i2c

where nc denotes the number of banks in each country c, and r is a rank number
assigned to the value of each bank’s quality variable over time, taken in decreasing
order of size.

References
Aigner, D., Lovell, C.A.K. and Schmidt, P. (1977), ‘‘Formulation and estimation of stochastic
frontier production function models’’, Journal of Econometrics, Vol. 6, pp. 21-37.
Allen, F. and Gale, D. (1997), ‘‘Financial markets, intermediaries, and intertemporal smoothing’’,
Journal of Political Economic, Vol. 105, pp. 523-46.
Allen, F. and Gale, D. (2000), ‘‘Bubbles and crisis’’, Economic Journal, Vol. 110 No. 460, pp. 236-55.
Berger, A.N. and Hannan, T.H. (1998), ‘‘The efficiency cost of market power in the banking
industry: a test of the ‘quite life’ and related hypotheses’’, The Review of Economics and
Statistics, Vol. 80 No. 3, pp. 454-65.
Bikker, J. and Haaf, K. (2002), ‘‘Competition, concentration and their relationship: an empirical
analysis of the banking industry’’, Journal of Banking and Finance, Vol. 26, pp. 2191-214.
Cook, D.O., Schellorn, C.D. and Spellman, L.J. (2003), ‘‘Lender certification premiums’’, Journal of
Banking and Finance, Vol. 27, pp. 1561-79.
Chenmanur, T.J. and Fulghieri, P. (1994), ‘‘Reputation, renegotiation and the choice between bank
loans and publicly traded debt’’, Review of Financial Studies, Vol. 7, pp. 475-506.
De Bandt, O. and Davis, P. (2000), ‘‘Competition contestability and market structure in European
banking sectors on the eve of EMU’’, Journal of Banking and Finance, Vol. 24, pp. 1045-66.
Detragiache, E., Garella, P. and Guiso, L. (2000), ‘‘Multiple versus single banking relationships:
theory and evidence’’, Journal of Finance, pp. 1133-61.
Greene, W.H. (1990), ‘‘A gamma-distributed stochastic frontier model’’, Journal of Econometrics,
Vol. 46, pp. 141-63.
Hannan, T. (1991), ‘‘Bank commercial loan markets and the role of market structure: evidence Vertical product
from surveys of commercial lending’’, Journal of Banking and Finance, Vol. 15, pp. 133-49.
Hicks, J. (1935), ‘‘Annual survey of economic theory: the theory of monopoly’’, Econometrica, Vol. 3
differentiation
No. 1, pp. 1-20.
Hughes, J.P. (1999), ‘‘Incorporating risk into the analysis of production’’, Atlantic Economic
Journal, Vol. 27, pp. 1-23.
Hughes, J.P. and Mester, L.J. (1993), ‘‘A quality and risk-adjusted cost function for banks: evidence 259
on the ‘too-big-to-fail’ doctrine’’, Journal of Productivity Analysis, Vol. 4, pp. 292-315.
Hughes, J.P., Lang, W., Mester, L.J. and Moon, Ch-G. (1995), ‘‘Recovering technologies that account
for generalized managerial preferences: an application to non-risk-neutral banks’’, working
paper, WP 95-16 Wharton Financial Institutions Centre, Philadelphia, PA.
Hughes, J.P., Lang, W., Mester, L.J. and Moon, Ch-G. (1996), ‘‘Efficient banking under interstate
branching’’, Journal of Money, Credit, and Banking, Vol. 28, pp. 1045-71.
Jondrow, J., Lovell, C.A.K., Materov, I. and Schmidt, P. (1982), ‘‘On the estimation of technical
inefficiency in the stochastic frontier production function model’’, Journal of Econometrics,
Vol. 19, pp. 233-8.
Jonshon, S. (1997), ‘‘The effect of bank reputation on the value of bank loan agreements’’, Journal
of Accounting, Auditing and Finance, Vol. 24, pp. 83-100.
Kim, M., Kristiansen, E.G. and Vale, B. (2005), ‘‘Endogenous product differentiation in credit
markets: what do borrowers pay for?’’, Journal of Banking and Finance, Vol. 29 No. 3,
pp. 681-99.
Kim, M., Kliger, D. and Vale, B. (2003), ‘‘Estimating switching costs: the case of banking’’, Journal
of Financial Intermediation, Vol. 12, pp. 25-56.
Lummer, S. and McConnell, J. (1989), ‘‘Further evidence on the bank lending process and the
capital market response to bank loan agreements’’, Journal of Financial Economics, Vol. 25,
pp. 99-122.
Mester, L. (1996), ‘‘A study of bank efficiency taking into account risk-preferences’’, Journal of
Banking and Finance, Vol. 20, pp. 1025-45.
Modigliani, F. and Miller, M.H. (1958), ‘‘The cost of capital, corporation finance, and the theory of
investment’’, American Economic Review, Vol. 48, pp. 261-97.
Molyneux, P., Lloyd-Williams, D.M. and Thornton, J. (1994), ‘‘Competitive conditions in European
banking’’, Journal of Banking and Finance, Vol. 18, pp. 445-59.
Petersen, M.A. and Rajan, R.G. (1995), ‘‘The effect of credit market competition on lending
relationship’’, Quarterly Journal of Economics, Vol. 110, pp. 407-43.
Sharpe, S.A. (1990), ‘‘Asymmetric information, bank lending, and implicit contracts: a stylized
model of customer relationships’’, Journal of Finance, Vol. 45, pp. 1069-87.

Corresponding author
Ana Lozano-Vivas can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected]


Or visit our web site for further details: www.emeraldinsight.com/reprints

You might also like