BAD LOANS AND ENTRY IN LOCAL CREDIT MARKETS
In
the case of banking services, more competition should lead to higher efficiency, higher
deposit rates and lower loan rates, therefore increasing borrowers’ and depositors’ welfare
(Besanko and Thakor, 1992).
However, the existence of financial intermediaries has been
since long traced back to market imperfections. In credit markets asymmetries in the
distribution of information and externalities influence banks’ risk-taking behavior, their
ability and their incentives to price risk correctly.
In a number of circumstances the
heightening competition can exacerbate market distortions making the welfare effects less
straightforward.
Economic theory also suggests that asymmetric information can work as a barrier to
entry in credit markets and that incumbents’ market power can be to some extent shielded
from outside potential competition.
This view is supported by two main arguments. The
first one is related to the possibility that once entry occurred, previously rejected
applicants can apply for loans at additional banks.
As far as borrowers’ credit-worthiness
is assessed through screening procedures which are not fully revealing and are unperfectly
correlated across banks, a larger number of banks rises the probability that a bad risk is
considered as credit worthy by at least one of them (Broecker, 1990).
Adverse selection is
greater for new entrants because the pool of their applicants is likely to include those
potential borrowers previously rejected by mature banks in the market (Shaffer, 1998).
The second argument relates on the informational advantages of the incumbents on
market characteristics.
A relevant amount of information used by banks for screening loan
applicants and for monitoring borrowers is generated through repeated interaction with
their customers.
Many studies, both of theoretical and empirical nature, have documented that long term relationships established between lenders and borrowers are an important
feature of most bilateral credit markets (i.e. Sharpe,1990; Rajan 1992; Boot, 2000).
A
considerable amount of valuable information can be acquired only on a market specific
“learning by doing” basis, thus implying that incumbents’ credit-worthiness tests may well
be more precise than those of the entrants.
Empirical evidence on the link between entry and default rates of the loans extended
by the entrants is scarce. Several episodes of banking crisis around the world have been
directly or indirectly related to the lifting, or relaxation, of the constraints imposed by
regulation.
Demirguc-Kunt and Detragiache (1998), analyzing a sample of 53 countries,
show that financial liberalization rises the probability of a bank crisis. Caprio and
Klingebiel (2000) document a recent increase in the frequency of bank crises and argue
that it is at least partially due to the lifting of structural controls.
Pesola (2001) finds that
market liberalization had a substantial role in the bank crisis experienced by the
Scandinavian countries during the early 1990s.
To our knowledge, Shaffer (1998), which
is very close in spirit to our paper, is the only study that empirically tests the winner’s
curse hypothesis, i. e. the idea that entrants in a credit market are forced to assume too
much risk because their pool of borrowers is adversely selected.
This paper contributes to the subject through an empirical study of the Italian
outburst of bad loans in the early 1990s. The 1992-93 recession caused an unprecedented
upsurge in non-performing loans followed by severe losses for a large number of banks.
Thanks to a unique database on loan default rates made available by the Italian Central
Credit Register, we investigate individual banks’ loan default rates in each local market.
The regulatory reforms introduced in the late 1980s allowed a substantial increase in the
number of banks operating in each local market (here defined as a province).
We
distinguish between two different ways in which a bank enters a local market. The first
one consists of extending loans from branches or the headquarters outside the local
market. The second one is by directly opening of branch.
We argue the two types of entry
differ substantially with respect to the information gap vìs a vìs the incumbents. Having on
site branches allows banks a more rigorous monitoring of the borrowers and a better understanding of the local economy.
Moreover entry with branches is usually anticipated
by entry with loans extended from outside.
Consistently with the predictions of the theory we find evidence that entrants are
more exposed to bad loans than the incumbents, since they have to deal with the backlog
of previously rejected applicants.
The consequences of entry on the loan default rate vary
with the level of information. Those banks entered with relatively more information, i.e.
by opening a branch, experienced a lower default rate than those entered without branches.
The negative consequences for the entrants are less severe when markets are characterized
by a high level of banks’ customers turnover. According to our data there is also a positive
relation between the loan default rate and the number of banks operating in a market, as
predicted by the winner’s curse hypothesis.
As a general result, we find that the suboptimal effects of entry on loans quality are mitigated when entrant banks belong to the
top performers of the industry. Borrowers of well capitalized, efficient, and aboveaverage profitable banks are characterized by substantially lower default rates.
The remaining of this paper is organized as follows. In Section 2 some theoretical
and empirical contributions in this field are surveyed. In section 3 we illustrate our
empirical specifications, while in Section 4 we discuss our data and explain how the
relevant variables have been constructed. Results are presented in Section 5 and section 6
draws the conclusions.
Related literature
The possibility that an increase in competition in the banking industry may have
sub-optimal allocative effects has been recognized since long, but only recently the
argument has been cast in formal models.
The backbone of most of them is an application
of the theory of common value first-bid auctions (Milgrom and Weber, 1982).
When
banks compete in prices (i.e. interest rates) and have an imperfect knowledge of the
would-be borrowers’ ability to repay their debts, they face an externality caused by the decisions of the other banks.
Before granting a loan a bank needs to assess the creditworthiness of the applicant. The screening procedure may be thought as a not fully
revealing test of the quality of the applicant. Conditional on the result of the test the bank
offers an interest rate or denies credit.
The borrower chooses to sign the credit contract
with the bank that offers the lowest interest rate. If the tests run by different banks are not
perfectly correlated, there is a positive probability that an applicant’s credit-worthiness is
assessed differently by different lenders.
This implies that probability that a high risk
borrower is assessed as credit-worthy is positively correlated with the number of tests that
are run.
It follows that the average quality of the pool of borrowers that obtain a loan (and
consequently the expected losses from bad loans) declines (increases) as the number of
banks in the market increases.
This idea has been formalized by Broeker (1990) using credit-scoring tests with
binary outcome.
Competition is modeled in two different ways - as a one-stage game and
as a two-stage game - obtaining different results concerning the existence and the
characterization of the equilibrium solutions.
In both cases the intuition that an increase in
the number of banks may have negative effects on the average ability to repay the loan of
those who are granted credit holds true. A similar result was obtained by Riordan (1993)
in a model where banks are able to run credit-worthiness tests delivering continuous
signals.
An increase in the number of banks rises the threshold value of the signal above
which the loan is granted, but this effect can be offset by the increase in the probability
that at least one bank observes a “high quality” signal screening a “low quality” applicant.
In Riordan’s model the entry of new banks into a credit market is associated with a more
restrictive supply stance. For a non trivial set of parameters also the equilibrium default
rate in the market is positively correlated with the number of active banks.
This basic framework has been extended by several papers. Gehrig (1998) allows the
banks to choose the precision of their binary credit-worthiness test and models the
integration of two previously separated credit markets as a sequential entry game and as simultaneous duopoly.
In the first case entry, if it ever occurs, doesn’t have any effect on
the incumbent’s screening intensity choice, while the entrant, facing an adversely selected
pool of applicants, substitutes a less accurate screening with higher interests rates.
As auction theory has recognized since long, additional insights may be gained when
information about the value of the object being sold is asymmetrically distributed among
participants (Wilson 1967, 1977).
Dell’Ariccia et al. (1999) analyze an entry model where
the incumbent has an informational advantage on the entrant since he has a long-term
relationship with part of his customers. This advantage is greater the lower is the
customers turnover in the market.
They characterize the equilibrium under Bertrand
duopoly and show that the adverse selection of the borrowers’ pool causes entry to be
blockaded. Marquez (2002) proposes a two periods model assuming that borrowers’
characteristics are observable by banks only once a loan has been granted and that there is
some turnover among borrowers.
In the second period banks will refuse to continue
financing borrowers revealed to be bad. Since information is proprietary, these borrowers
remain as part of the pool of customers unknown to all other banks.
In this framework an
increase in the number of banks disperses borrower-specific information reducing banks’
screening ability. Incumbents’ informational advantage may also act as a barrier to entry if
borrower turnover is low.
The role of relationship lending in magnifying entrants’ vulnerability has been
extensively studied. Relationship lending generates informational rents accruing to the
banks from which “captured” firms can try to escape searching for better deals in the
credit market.
Sharpe (1990) has shown that if an uninformed outsider bank offers a
competitive interest rate (e.g. reflecting the average credit quality), only bad borrowers
would prefer to switch.
The information asymmetries on the same side of the market (the
supply side) induced by relationship lending is therefore most likely to add to the adverse
selection problems faced by a new entrant banks (Nakamura, 1993).
Furthermore, when
assessing the credit worthiness of a loan applicant, banks usually refer to their past experience with similar borrowers in similar markets. This may imply that when a bank
expands in a new market or sector the negative effects of a lack of expertise may
overcome the benefits from risk diversification (Winton 1997).
Comparing with the abundance of theoretical papers on the subject, the empirical
work has been rather limited. Shaffer (1998) tests empirically the prediction that de novo
banks should suffer higher loan losses due to the adverse selection effect.
He considers all
U.S. commercial banks during the period 1986-1995 and regress the net chargeoff ratio
versus annual age dummies for each of a bank’s first 10 years, controlling for business
cycle and other macroeconomic effects including quarterly calendar time dummies.
He
finds that the net chargeoff rates are strongly and significantly higher from year 3 on.
After discussing alternative explanations of this pattern (such as the seasoning of a new
portfolio or the presence of a learning process of an inexperienced lender), he concludes
that adverse selection appears the main cause of the observed phenomenon.
Shaffer also
estimates a cross sectional model using data from mature banks, each operating in a single
geographic market (MSA). He finds a strong positive linkage between gross chargeoff
rates and the total number of banks in each MSA.
Hedricks and Porter (1988) investigate the links between the winners’ curse and
asymmetric information among bidders in auctions. They show that in equilibrium the
uninformed buyer makes zero profits, while the informed buyer makes positive profits
thanks to its superior information.
They also test these theoretical conclusions examining
data from the federal offshore oil and gas drainage lease sales finding that both types of
buyers actually behave consistently with the Bayesian-Nash equilibrium.
Posting Komentar untuk "BAD LOANS AND ENTRY IN LOCAL CREDIT MARKETS"