Temi di discussione

del Servizio Studi

The Effects of Bank Consolidation and Market Entry

on Small Business Lending

by Emilia Bonaccorsi di Patti and Giorgio Gobbi

Number 404 - June 2001

The purpose of the “Temi di discussione” series is to promote the circulation of working
papers prepared within the Bank of Italy or presented in Bank seminars by outside
economists with the aim of stimulating comments and suggestions.
The views expressed in the articles are those of the authors and do not involve the
responsibility of the Bank.

Editorial Board:
ANDREA BRANDOLINI, FABRIZIO BALASSONE, MATTEO BUGAMELLI, FABIO BUSETTI, RICCARDO
CRISTADORO, LUCA DEDOLA, FABIO FORNARI, PATRIZIO PAGANO; RAFFAELA BISCEGLIA
(Editorial Assistant).
 THE EFFECTS OF BANK CONSOLIDATION AND
MARKET ENTRY ON SMALL BUSINESS LENDING

by Emilia Bonaccorsi di Patti* and Giorgio Gobbi*

Abstract

Consolidation in the banking industry of many countries has reduced the number of
small banks and led to significant shifts in market shares; deregulation has fostered entry in
local credit markets and branch expansion, which in turn have increased competition. Small
businesses are believed to be more vulnerable to these changes since they are more
dependent on credit from local banks. In this paper we investigate the consequences of
consolidation and entry for these borrowers compared with those for large firms. We employ
a data set for Italy, which provides information on volumes of loans and bad loans by size of
borrower with a detailed geographical partition. We find that mergers are followed by a
temporary reduction in outstanding credit to all sizes of borrowers and by an increase in bad
loans, most likely due to the reassessment of banks portfolios. Entry has a relatively
persistent negative impact on credit supply to small and medium-sized firms. Our results
also show that concentration, branch density and the share of branches of small banks affect
the volumes of credit and bad loans of small borrowers.

JEL classification: G21, G34.

Keywords: commercial banks, mergers, entry, small business lending.

Contents
1. Introduction........................................................................................................................... 7
2. Previous evidence ................................................................................................................. 9
3. The empirical analysis ........................................................................................................ 13
4. Data and variables............................................................................................................... 18
5. Results................................................................................................................................. 21
6. Conclusion.......................................................................................................................... 24
Tables...................................................................................................................................... 26
References............................................................................................................................... 35

____________________________
* Banca d’Italia, Research Department.
1. Introduction1

According to the consensus view emerging from theoretical investigation and

supported by numerous empirical studies, small businesses as opposed to large firms face
specific constraints in raising external finance (Berger and Udell, 1998). Information of
value to investors about small businesses is more costly to gather and most of it is not
publicly verifiable. Since financial intermediaries can exploit economies in information
gathering and processing, bank credit accounts for a large share of small business financing
(Table 1).

Because of their special nature, small opaque firms may be adversely affected by
changes in the structure of the banking industry. The issue of availability of credit to these
borrowers is relevant not only from a theoretical point of view but also for policy purposes.
In many countries deregulation and financial and technological innovation have stimulated
extensive restructuring in the financial sector. Commercial banks have engaged in mergers
and acquisitions, leading to the disappearance of many small credit institutions and the
emergence of complex financial conglomerates. The lifting of barriers to geographic
expansion has allowed entry in previously isolated local markets, reducing segmentation.
Structural changes can influence credit flows to small firms through three channels.
First, changes that permanently modify the size distribution of banks and their geographic
reach are likely to have effects on the allocation of credit across classes of borrowers with
different degrees of opaqueness. Small banks tend to specialize in small business lending
both because they have a limited lending capacity and because their role in local
communities gives them a comparative advantage in acquiring information on borrowers.
The reduction in their ranks, mainly but not exclusively due to consolidation, and the move
of the industry towards a smaller number of large institutions can adversely affect the
availability of credit to small, relationship intensive borrowers.
Second, a negative direct effect can result from changes that imply the loss of soft
information or the interruption of previously established relationships, when adverse

1
We thank Tyler Cowen, Andrea Generale, Tim Hannan, Fabio Panetta and Massimo Roccas for
comments, and Roberto Felici for excellent research assistance. The opinions expressed do not necessarily
reflect those of the Banca d’Italia or its staff. Email: bonaccorsidipatti.emilia@insedia.interbusiness.it.
 8

selection problems are severe. This loss of information could be a consequence of the
restructuring that typically occurs after mergers and acquisitions (Berger, Demsetz and
Strahan, 1999).
Third, changes in the number and composition of suppliers are likely to be reflected in
changes in the competitiveness of credit markets. Their effect on small business lending
cannot be determined unambiguously ex ante. On one hand, opaque borrowers have been
found to benefit from stable credit relationships because market power reduces hold-up
problems (Petersen and Rajan, 1994, 1995). On the other, small businesses are more
vulnerable than large firms to local monopoly power because the information that they
provide to their intermediaries cannot be transferred, and this increases switching costs
(Hannan, 1991).
Consolidation can reduce the competitiveness of the banking industry in local markets
if it increases concentration, but it can also increase industry rivalry if the advent of more
efficient managers or strategy changes of some institutions induce similar behavior in other
banks. By contrast, the lifting of barriers to entry is most certainly conducive to an increase
in competition. According to the standard mark-up model, entry increases the availability of
credit because it shifts the supply curve, leading to lower interest rates and greater quantity.
On the other hand, incentives to invest in collecting information may diminish, to the
detriment of opaque borrowers (Broecker, 1990; Von Thadden, 1998). Entry may therefore
either increase or reduce small business lending, particularly its relationship-intensive
component (Boot, 2000).
The empirical relevance of all these effects is likely to depend on the information
structure of the market, including banks’ ability to acquire information about the credit-
worthiness of borrowers, the appropriability of that information, the heterogeneity of
borrowers and their ability to signal their quality.
Advances in information technology have enormously reduced the cost of gathering
and processing information. In some countries, such as the United States, the dissemination
of valuable credit information about individual borrowers is rapidly expanding owing to
private credit bureaus (Miller, 2000). The ability of small firms to disclose their credit
histories and other relevant information through credible agencies extends the range of their
possible lenders beyond local credit markets by transforming some of the previous soft
information into hard information (Petersen and Rajan, 2000). But, if a significant amount of
 9

intrinsically soft information remains, advances in information technology will not eliminate
the potential adverse effects of structural changes in local credit markets for small, opaque
borrowers.
In this paper we develop an empirical analysis to assess the effects of consolidation
and entry in local credit markets on the availability of credit to small businesses. We
examine two aspects of credit, volumes and quality, proxied by the proportion of bad loans.
We apply our analysis to Italian data at the local market level because our focus is the
evaluation of net effects beyond the behavior of individual institutions, e.g. consolidating
banks or new entrants.
Our data set has many advantages. First, detailed information on the stock of loans in
each local market by size of borrower is available whereas in other countries, including the
US, such data has to be estimated. Second, data on bad loans by local market and by size of
the borrower are also available. Most of the existing literature on consolidation or entry has
not studied the effects on the quality of credit. Although we recognize that our measure of ex
post quality is imperfect, it can provide insights into changes in the behavior of banks in
screening and monitoring. Third, the level of information-sharing in credit markets in our
data set is relatively high and is stable over the period examined, anticipating what may
occur in countries where private and centralized credit bureaus are being set up. Since the
early 1960s data on the credit records of virtually all Italian firms have been collected and
provided to the banking system by the Central Credit Registry, a database managed by the
Bank of Italy. This fact reduces the potential bias from omitted variables related to
technological change, such as the size of geographical markets and the availability of
increasing information flows within the banking system.
The remainder of the paper is as follows. Section 2 briefly reviews the relevant
literature in small business finance. Section 3 discusses the empirical methodology and
section 4 describes the data and the variables. Section 5 illustrates the results and section 6
concludes.

2. Previous evidence
A large number of studies has addressed the issue of how changes in the structure of
the banking industry affect small business lending. The majority of these studies has focused
exclusively on the effects of consolidation, without considering entry, and has been
 10

motivated by concern that the move to a more concentrated banking industry might reduce
the availability of credit to small borrowers. Fewer empirical studies have examined the
impact of competition and changes in competitiveness on specific categories of borrowers.
Indirect evidence on consolidation and small business lending is offered by studies that
examine the static relationship between the size and complexity of banks, and the share of
small business loans held in their portfolio (Berger, Kashyap and Scalise, 1995; Levonian
and Soller, 1995; Berger and Udell, 1996; Peek and Rosengren, 1996; Strahan and Weston,
1996, 1998). Their general finding is that larger banks hold less small business loans than
small institutions.
One possible explanation is that small banks are limited by their financial capacity and
cannot lend to larger firms, whereas large banks have access to a larger pool of potential
borrowers and can supply a greater variety of products. Another argument is that small banks
may enjoy a cost advantage in supplying small business loans, some of which would not be
profitable for larger banks. If the first hypothesis holds, changes in bank size may or may not
reduce small business lending; if the loans that large banks do not want to make are
profitable for other banks, total credit to small businesses will not decrease in the aggregate.
If the second hypothesis is true, on the other hand, the reduction in the number of small
banks may lead to a permanent decrease in small business credit, because loans that had
positive net present value for small banks are no longer profitable for larger ones.
Organizational complexity may be another determinant of the propensity of banks to
provide credit to small borrowers. Theory suggests that small business lending is
characterized by soft information and that monitoring and control by loan officers can be
more difficult in larger and complex organizations. In addition, Williamson-type
diseconomies might make joint production of transaction and relationship loans less
efficient. Larger banks can employ more quantitative methods for screening rather than
relying on the discretion of the loan officer.
Empirical evidence at the market level is controversial. Jayaratne and Wolken (1999)
test if the relative weight of small banks in local credit markets influences the probability
that a small firm has a line of credit from a bank. They find that this probability is unaffected
by the number of small banks in the local market. Nonetheless, structural changes such as
consolidation may cause short-run disruptions in the availability of credit to small
businesses.
 11

Berger and Udell (1996) find that larger banks tend to charge lower rates and are less
likely to require collateral from small business borrowers, but they also tend to issue far
fewer loans to these borrowers. Banks of greater organizational complexity, as measured by
proxies of the holding company structure, generally provide less credit to small borrowers,
but the effect of these variables on the lending rates charged is mixed. By contrast, Strahan
and Weston (1998) find that neither the size of the bank holding company nor the
complexity of its structure affects its share of lending to small businesses.
Mergers and acquisitions are likely to be associated with substantial restructuring,
which is only partly related to the size and the complexity of the emerging firms. A strand of
literature addresses the dynamic effects of consolidation by examining small business
lending by consolidating banks before and after mergers (Keeton 1995, 1997; Peek and
Rosengren, 1998; Strahan and Weston, 1998; Craig and Santos, 1997; Walraven 1997;
Berger, Saunders, Scalise and Udell, 1998; Focarelli, Panetta and Salleo, 1999).
Generally speaking, the results differ by type of operation, size of institutions involved,
econometric technique and number of years examined after the M&As. The most common
finding is that consolidation involving large banking organizations tends to reduce small
business lending. When mergers are distinguished by the relative size of acquirer and
acquired bank and by their shares of small business credit, the conclusion is that acquirers
tend to drive the new institution’s share to converge to their pre-merger portfolio share in
small business loans (Peek and Rosengren, 1998; Walraven, 1997).
Evidence for countries other than the US is scarce. A recent study of the effects of
bank mergers and acquisitions in Italy (Focarelli, Panetta and Salleo, 1999) finds that banks
involved in mergers and acquired banks reduce the share of credit to small businesses. The
decline seems to be at least partly motivated by risk concerns since the quality of the loan
portfolio tends to improve after a temporary increase in bad loans.
Similar results were found by a previous study (Sapienza, 1998), suggesting that
relationship-intensive borrowers may be adversely affected by bank consolidation. The
effects differ depending on whether the borrower had a relationship with the acquiring only,
with the acquired only, or with both banks. The interruption of the relationships tends to be
most frequent if the borrower was only a customer of the acquired bank and results in a
higher cost paid for financing after the merger. Relationships that remain with the acquiring
bank benefit from a decrease in rates paid.
 12

A common finding of these studies is that restructuring of the loan portfolios is a rather
common practice and small business credit is likely to be adversely affected at the bank
level. The overall effect is likely to depend on the risk profile of cut off borrowers, on their
informational opaqueness and on the behavior of the other suppliers in the market.
Berger, Saunders, Scalise and Udell (1998) decompose the total effect of consolidation
on the share of small business lending by banks into various components. Consistently with
other studies, the static aggregation of banks is associated with a reduction in small business
lending; however, other banks in the same market tend to increase their small business
lending. This “external effect” of M&As is empirically relevant and its magnitude may be
sufficient to counteract the negative effect on small business lending directly attributable to
consolidating banks.
Other studies have found that de novo banks tend to lend more to small businesses
than other banks of similar size (De Young, Goldberg and White, 1999), providing an
alternative source of credit to these borrowers. Berger, Bonime, Goldberg and White (1999)
investigate the effects of mergers and acquisitions on de novo entry and small business
lending at the local market level. Their results suggest that while M&As are associated with
subsequent increases in the probability of entry into the local market where the consolidation
has taken place, they do not support the view that M&As generate an increase in lending to
small businesses by recent de novo entrants.
Avery and Samolyk (1999) estimate how changes in aggregate small business lending
are affected by total consolidation activity. They specify a reduced form for growth rates of
small business loans as a function of consolidation activity, other bank structure variables
and demand variables. According to their evidence, the relationship between consolidation
and small business lending growth differs across types of market. In urban markets the
growth rate of small business lending by consolidating banks is lower but other institutions
tend to partly compensate for the reduction. No effect is found in rural markets.
There is almost no direct evidence on the effects of entry on small business lending.
Indirect evidence can be found in the literature on the relationship between competition,
mainly measured by structural proxies, and the provision of credit to small, opaque
borrowers. Some studies find that concentration is associated with higher interest rates on
small loans, consistent with the classical mark-up model (Hannan, 1991). Others find that it
is beneficial to small borrowers (Petersen and Rajan, 1995), suggesting that the relationship
 13

is non-monotonic and varies depending on characteristics of the borrowers such as age, size,
opaqueness. De Young, Goldberg and White (1998) find that concentration affects small
business lending positively in urban markets (characterized by low concentration), but
negatively in rural markets (which have higher concentration).
Further evidence can be inferred from the literature that investigates the relationship
between bank competition and the rate of creation of firms. Jackson and Thomas (1995) find
a negative effect of bank size and a positive effect of bank concentration on the rate of birth
of manufacturing firms. On the opposite side, Black and Strahan (2000) find that a greater
presence of large banks increases lending and is associated with a higher level of business
starts, possibly because large banks can diversify more than small ones. Merger activity, by
contrast, does not affect the start of businesses. Bonaccorsi di Patti and Dell’Ariccia (2001),
based on Italian data, provide evidence that competition in the banking system has a negative
differential effect on the rate of creation of firms in industries with increasing degrees of
opaqueness but has a positive effect overall.

3. The empirical analysis

Our testing strategy consists in specifying a simple model where consolidation and
entry variables affect the volume and the quality of credit at the local market level. Since we
focus on potential differences between small and large firms, we estimate separately one
equation for large non-financial firms and one for small and medium-sized non-financial
firms. Before examining the details of the specifications, we restate the potential effects of
consolidation in terms of testable hypotheses. We then outline a set of predictions about the
effects of entry.
According to the large banks hypothesis, consolidation permanently modifies the size
distribution of banks in a given local market. If large banks have a lower propensity than
small banks to lend to small businesses, total credit to small businesses in that market will
diminish, everything else being equal. This hypothesis does not imply any effect on bad
loans.
The second hypothesis we test – the efficiency improvement hypothesis – follows from
the view that consolidation is a way for a more efficient bank to expand by taking control of
a less efficient one: the new management adopts more efficient lending policies and cuts
credit to negative net present value borrowers who previously received credit, possibly
 14

prompting emulation by its competitors. In the aggregate, consolidation activity would lead
to a temporary increase in bad loans of all size groups and a temporary reduction in
outstanding credit. This effect could stem from the behavior both of the consolidating banks
and of other institutions in the market, because of competitive pressure not induced by
changes in market structure.
We distinguish the conjecture that consolidation disrupts credit relationships into two
separate testable hypotheses, in both cases assuming that the borrowers who are no longer
served are positive net present value borrowers (if they are negative NPV borrowers we are
under the efficiency improvement hypothesis).
The first case occurs when adverse selection problems are so severe that borrowers
who are no longer served by merging banks cannot get credit from other banks. We would
then expect a reduction in small business credit in markets where there has been significant
consolidation activity. We call this hypothesis the disruption of relations I (with adverse
selection). The reduction in lending should be limited to small and medium borrowers; it
should also be temporary, because the loss of information is relevant for existing
relationships but not for future ones. Hence, this hypothesis differs from the large banks
hypothesis, which predicts a permanent modification in banks’ propensity to lend to small
firms in the market. There should be no effect on bad loans in the short term.

Box A: The effects of consolidation

Total Credit Total Credit Bad Loans Bad Loans

Hypothesis:
Small Large Small Large
Large banks − 0 or + 0 0
Efficiency improvement − temporary − temporary + temporary + temporary
Disruption of relations I
− temporary 0 or + 0 0
(adverse selection)
Disruption of relations II
0 0 0 0
(no adverse selection)

In the second case, adverse selection problems are marginal at most and borrowers
dropped by consolidating banks are able to find credit from other banks. There should be no
effect at the market level on credit and bad loans. We call this conjecture the disruption of
 15

relations II hypothesis (without adverse selection). The predicted signs of the effects
indicated by each of the hypotheses described are summarized in Box A.
The theoretical literature yields opposite predictions on the effects of entry on the
volume of credit. A positive effect is indicated by the standard mark-up channel; a negative
one by several information-based models. The estimated coefficient of Entry is the net of
these effects.
We suggest two potential interpretations for a negative coefficient of Entry. The first
posits that competition increases as a consequence of entry of new participants in the market,
worsening hold-up problems, moral hazard and adverse selection, thereby reducing the
incentive for banks to lend to opaque borrowers (information asymmetries hypothesis). The
empirical prediction in this case is a negative sign for small business lending but not for
credit to large firms. Nothing can really be said about bad loans without making additional
assumptions on the degree of risk of the pool of small versus large borrowers.
The second interpretation is that entry causes an increase in competition not only in the
loans market but also in the deposit market, thereby shrinking interest margins. Banks facing
higher deposit interest rates tend to curtail lending to lower quality borrowers. This
hypothesis has been studied in the context of the effects of a monetary policy restriction
(Asea and Blomberg, 1998; Lang and Nakamura, 1995) and we will refer to it as the flight to
quality hypothesis. Entry would determine a reallocation of credit away from marginal
borrowers in all size classes. No direct effect on bad loans can be predicted in the short run
but a reduction in the medium term is likely. The hypotheses tested about entry are
summarized in Box B.

Box B: The effects of entry

Total Credit Total Credit Bad Loans Bad loans

Hypothesis: Small Large Small Large
Mark-up + + 0 0
Information asymmetries − 0 or + ? ?
Flight to quality − − −* −*

*In the medium run.

 16

We specify a simple model where the volume of credit in local market i at time t is a
function of consolidation, entry and a set of controls as follows:

ln(yk )it = α +β1 (L)Sh_mergedi +β2 (L)Sh_acquiredi + β3 (L)Entryi +

Β1 Controlsi + Β2 xi + Β3 zt + eit (1)

where xi and zt are market and time dummy variables. The subscript k for the dependent
variable indicates that the regression is estimated separately for k = large non-financial
firms, small and medium-sized non financial firms. We define local markets as provinces and
employ yearly data.
Consolidation activity is measured by the share of credit in the local market that has
been involved in a consolidation. We construct separate variables for mergers (Sh_merged)
and acquisitions (Sh_acquired) according to the definitions discussed in the next section.
The variable Entry is computed as the proportion of branches at time t that belongs to banks
that were not present in the local market i at time t-1. The three variables enter the regression
with a lag structure in order to capture time patterns that contribute to discriminate among
different hypotheses.
The set of control variables includes measures of economic activity and subsidized
credit. Local economic conditions are captured by the natural log of GDP in each province.
Subsidized credit is included because several government schemes for subsidized lending
were phased out during the sample period, potentially affecting small borrowers in a
different way than large ones (Gobbi, 1996).
Since the structure of the local banking system may change for reasons other than
consolidation and entry, as a robustness check we estimate the effect of consolidation on
credit including bank structure variables such as the level of concentration in local markets,
the geographic expansion of banks within the market captured by branch density, and the
relative weight of small banks.
This specification allows us to separate the direct effects of consolidation and entry
from those induced through changes in concentration, the presence of small banks and
branch density. This second equation is of the form:
 17

ln(yk)it = α + +γ1 (L)Sh_mergedi +γ2 (L)Sh_acquiredi + γ3 (L)Entryi + γ4Concentrationit-

1 + γ5 Branch Densityit-1 + γ6 Small Banks it-1 + Γ7 Controlsit + Γ8 xi + Γ9 zt + v it.. (2)

The specification is, again, estimated separately for credit to large and to small and
medium-sized non-financial firms. To determine whether changes in bank structure affect
the quality of credit we specify the log-odds ratio of bad loans to total loans as a function of
the same regressors included in equation (1). The dependent variable is defined as ln[(bad
loansk/total loansk)/(1–(bad loansk/total loansk))], which is a formulation of the bad loans
ratio. Symmetrically, we estimate the log-odds model with the same set of explanatory
variables included in equation (2). The specification described by equation (2) not only
provides a means to control for changes in credit or loan quality not induced by
consolidation and entry, but also a way to focus on the hypotheses about the disruption of
relations, thereby improving identification.
Specifically, the effect of the large banks hypothesis is removed when we include the
share of small banks directly in the regression because the coefficient of the consolidation
variables is conditioned on the size distribution of banks. Also the effect of mergers on
competition caused by changes in market shares is absorbed by the concentration variables.
The coefficient of the merger variables would capture strategy changes of market
participants and any effect on competition not produced by concentration. What remains in
the estimated coefficients is the net effect of the efficiency improvement hypothesis and of
the disruption of relationships I and II hypotheses.
Similarly, the effect of entry that remains once we control for the number of branches
and for concentration is the composition effect due to the presence of outside competitors,
which is more closely related to asymmetric information issues because outsiders do not
have specific knowledge of the local market.
It is crucial to underline that we are only able to assess market level effects and we are
not attempting to draw any conclusion about the behavior of individual banks, which has
been extensively analyzed by other studies. In addition, we are identifying mainly first order
effects, particularly on the quality of loans. Consolidation can lead to different risk-taking in
the merging banks, because size allows more diversification within each size class. If this
effect is large enough, it may show up as an increase in the bad loans ratio of each size class
of borrowers, although most likely not immediately after the merger.
 18

4. Data and variables

Our data refer to lending in local credit markets, defined as provinces, during the
period 1990-98. Aggregate volumes of loans and bad loans 2 are constructed from the Italian
Central Credit Register (CCR), a public credit reporting system managed by the Bank of
Italy that collects from banks information on individual borrowers with bank debt of at least
ITL 150 million. 3 The CCR releases aggregate credit statistics by sector, location and size of
borrowers. These statistics include information on outstanding credit and bad loans. Bad
loans are defined on a customer basis and therefore include all the outstanding credit to
borrowers considered insolvent.
We define as small and medium-sized firms those with total bank debt of between ITL
250 million (around USD 130,000) and ITL 5 billion (USD 2,6 million). 4 Firms include non-
financial corporations, individual enterprises and sole proprietorships.
A potential bias in the data could be generated by the exclusion of a large number of
micro enterprises. Using the information available from banks’ supervisory reports we have
estimated that about 70 per cent of total credit to firms with less than ITL 5 billion of bank
debt is included in our data. 5

2
According to the Italian supervisory guidelines a loan is classified as a “bad loan” if the borrower is facing
serious economic and financial problems that may threaten his ability to meet his obligations (repay principal,
interests or both) or if legal proceedings have been initiated. The assessment of the borrower’s ability to repay
has to be made independently of loan guarantees and collateral.
3
Before 1995 the reporting threshold was ITL 80 billion. Since 1997 non-bank financial institutions (e.g.
leasing, factoring and consumer credit companies) also report to the CCR. A brief comparative description of
the Italian CCR is contained in Miller (2000).
4
Data in the lowest class (borrowers with outstanding bank debt of between ITL 150 and 250 million) are
rather noisy and may introduce a bias because of multiple lending relationships even among relatively small
firms (Detragiache, Garella and Guiso, 2000). Since each bank reports only borrowers to whom it has lent more
than ITL 150 million, after a merger many previously unreported customers may appear in the CCR statistics if
the sum of their previous loans surpasses the threshold.
5
The majority of the firms excluded belongs to agriculture and retail trade. As a robustness check we have
employed supervisory report data on outstanding credit to sole proprietorships. These data do not distinguish by
size of borrower but have the advantage of no lower bound for reporting. Consistent information on credit to
sole proprietorships with a breakdown by local markets is available only since 1995. Estimates for this period
employing sole proprietorship data are consistent with the results obtained with the CCR data for small and
medium-sized firms.
 19

We define local markets as the 95 provinces. 6 Provinces are the finest geographical
partition for which a rich set of economic statistics exists. On average, about 80 per cent of
credit to small and medium-sized businesses, under our definition, is granted by bank
branches operating in the same province.
We construct our consolidation variables as the share of the loan market that has been
shifted in market i at time t by consolidation activity. We classify the operations of
consolidation in two types: “mergers” and “acquisitions”. The class “mergers” comprises (i)
actual mergers, where two or more banks form a new entity, and (ii) acquisitions of a bank
by another, followed by a merger (the acquired bank disappears). The class “acquisitions”
comprises (iii) acquisitions where a bank acquires control over another bank, and (iv) the
creation of a banking group from existing banks or existing bank holding companies. Our
classification is motivated by the intuition that in the first two cases the degree of integration
is more extensive than in (iii) and (iv), usually implying a reorganization within the new
bank created. Plans to shift the business focus of the bank towards financial services, as in
the case of mergers, are likely to be associated with changes in credit standards and a
reduction in lending activity.
The notion of control in (iii) is that adopted by the Bank of Italy’s Banking
Supervision Department and implies the power to influence the business strategies of the
entity rather than the ownership of a given shareholding. In the 1990s there were almost 300
mergers and more than 100 acquisitions where the acquired bank continued to operate as a
separate legal entity (Table 3).
In calculating market shares shifted by consolidation activity, our treatment of both
types depends on whether the consolidation originates a new institution and whether a leader
and a target can be unambiguously identified.
The Sh_merged variable is constructed as follows. In the case of actual mergers
“between equals” (i) we compute the share summing those of the banks involved. In the case
of an acquisition followed by a merger (ii) we only count the share of the target (or passive)
bank.

6
We use the partition into 95 provinces. The 8 additional provinces that were created in 1995 have been
aggregated backwards, according to the old boundaries.
 20

The Sh_acquired variable is calculated by a similar reasoning: in the case of standard

acquisitions of control (iii) we only consider the market share of the acquired (passive) bank.
Finally, when existing banks or holding companies form a banking group or another holding
company (iv) we count the share of all entities involved if a leader cannot be identified;
otherwise we count only the share of the target. The rationale of this criterion is to isolate the
share of the credit market that will most likely undergo a change in lending strategies after
the deal; the effects induced by changes in competition are captured indirectly through
changes in concentration.
We construct the variable Entry as the share of branches in each province held by
banks that were not present in that province in the previous year. The share is adjusted for
the effect of mergers and acquisitions.
The variables that describe the structure of the banking system in the provinces are
calculated as follows. Concentration is measured by the Herfindahl Index of loans, based on
the residence of the borrower (Herfindahl). The variable Branch_density is the ratio of the
number of branches to population, multiplied by 10,000.
The presence of small banks (Sh_smallbanks) in the local banking market is measured
by the share of their branches. Small banks are defined according to the official classification
currently used by the Bank of Italy. 7 Average total assets for these institutions were about
ITL 700 billion in 1995 (Table 2).
The share of subsidized credit (Sh_subscredit) is computed as total credit issued under
some central or local government subsidy scheme divided by total credit (see Gobbi, 1996
for a description of subsidies). The level of economic activity is measured by the natural log
of nominal value added (lnGDP). The data on value added at the province level are released
by the Union of the Italian Chambers of Commerce and are available up to 1997. Figures for
1998 have been estimated using data on employment growth.
Tables 2 and 3 report some statistical information on the Italian banking system and its
evolution in the last decade. In table 2 the breakdown of outstanding credit by size of
borrower and size of bank is shown. Between 1989 and 1998 the share of total bank lending

7
A bank is classified as small if its gross total assets are less than ITL 5,500 billion (around USD 2.8
billion). In 1995 there were 987 banks in Italy, 926 of which were classified as small. Among the latter 827
institutions were defined as minor, with gross total assets of less than ITL 1,500 billion (around USD 780
million).
 21

to small and medium-sized firms declined by about 3.5 percentage points. Table 3 reports the
number of mergers and acquisitions, classified by the size of target banks. Small banks
represented the vast majority of target banks in both types of operations. The second section
of Table 3 shows the distribution of local markets by the share of credit involved in
consolidation, based on our methodology.
The quantity and log-odds equations have been estimated employing data referring to
two periods: the full period available (1990-98), and a shorter period (1995-98), to exclude
the effects of the 1992-93 recession, which probably cannot be captured by the time
dummies because of a significant differential impact across sectors and provinces. A large
number of firms failed, potentially reducing the pool of borrowers, again in a uneven way
across provinces. Descriptive statistics are reported for both periods in table 4.

5. Results

Tables 5-A and 5-B report the results of the estimation of equation (1) on 1990-98 and
1995-98, respectively, based on the within panel estimator obtained estimating on deviations
from means. 8 The log-odds model has been estimated instead with weighted least squares to
improve efficiency (see Greene, 1993). Dummy variables for each local market have been
included directly in the set of regressors. The weights are computed as (BLi*PLi)/(BLi+PLi)
where BLi is the value of bad loans and PLi is the value of performing loans in market i, as
in standard estimation of proportion data. The variables of structural change (Sh_merged,
Sh_acquired and Entry) enter the regression with three lags. This lag structure has been
motivated by previous evidence on the time span of the adjustment period that follows
M&As and entry (Berger, Saunders, Scalise and Udell, 1998; Berger, Bonime, Goldberg and
White, 1999).
The overall pattern of coefficients is consistent across the two periods, although the lag
structure is estimated more precisely in 1995-98 because consolidation activity and entry
were more extensive after 1993, generating more cross-sectional variation in the data
especially for acquisitions. We will briefly discuss only the result for this period (Table 5-B).

8
We report results for the within estimator. We have also performed FGLS to account for potential
autocorrelation in residuals. Comparison of results shows that the fixed effects remove most of the
autocorrelation, hence the LSDV model is well specified without the need to include the lagged dependent
variable.
 22

A first finding is that mergers have an opposite effect on the volume of credit with
respect to acquisitions. The effect of Sh_merged is negative and tends to vanish over time,
whereas that of Sh_acquired is positive and persistent. More importantly, these results hold
for both large firms and small and medium-sized firms, and are robust to the inclusion of the
variables describing the structure of local banking system (concentration, branch density and
presence of small banks). The effect of Entry on the volume of credit is negative for both
categories of firms, but it is statistically significant and quite persistent only for small and
medium-sized ones.
Tables 6-A and 6-B show the results of the log-odds ratio estimates for the periods
1990-98 and 1995-98 respectively. Again, the coefficient estimates for consolidation and
entry are stable over the two periods, so we will briefly comment on the shorter period
(Table 6-B).
Merger activity (Sh_merged) has a positive and statistically significant effect on the
probability of one currency unit falling in the category of bad loans. This effect is true
whatever the size class of the borrower. Acquisition activity tends to reduce this probability.
The effect of entry is negative but not statistically significant with the exception of the
specification for small and medium-sized firms with the controls for market structure
(column 4, Table 6-B). This result is likely to be a consequence of the interaction between
entry and concentration which may generate strong multicollinearity.
A comparative reading of the estimates of the effects of consolidation on credit flows
and on the share of bad loans (Tables 5-A and 5-B, 6-A and 6-B) is possible referring to the
hypotheses previously described in Box A. As merger activity has different effects from
acquisition activity, we discuss them separately. The large banks hypothesis appears to be
inconsistent with our findings for two reasons. First, we find a temporary negative impact of
mergers on both large and small borrowers; second, we observe an increase in the share of
bad loans not predicted by this hypothesis.
The predictions of the disruption of relations II hypothesis are also not consistent with
the pattern of our estimated coefficients, as we do observe a reduction in credit volumes. The
disruption of relations I hypothesis is consistent with the temporary negative effect on small
business lending but not with the effect on loans to large firms. The signs of the effect on
bad loans would appear inconsistent with this hypothesis as well. Finally, our results seem to
 23

match more closely the predictions of the efficiency improvement hypothesis not only for the
signs of the derivatives but also for their time patterns.
The effect of acquisitions does not fit with any of the above hypotheses, because
acquisition activity is followed by an increase in credit to borrowers of all sizes. One
possible explanation is that acquisitions do not necessarily imply, at least in the short run, a
loss of information as long as local management stays in office. In addition, if the acquiring
bank is large, the acquired may benefit from economies of organization, diversification and
funding, with a consequent expansion in lending. At the local market level we may observe
an increase in credit supply to all borrowers. In the case of Italy, there is evidence that
acquirers are significantly larger than acquired banks, whereas in the case of mergers the size
difference is smaller (Focarelli, Panetta and Salleo, 1999). The reduction in the log-odds
ratio may be explained both by write-offs and by an improvement in the quality of new
loans.
The negative impact of Entry on credit appears to be inconsistent with the standard
mark-up model in which entry increases competition in a given market, resulting in lower
interest rates and a greater supply of credit to all borrowers. It is not fully consistent even
with information-based models because the negative sign also appears in the regression for
large borrowers. Consistently with our interpretation of the effects of consolidation, the
results may be explained by the flight to quality hypothesis discussed in the literature
because we observe a reduction both for large and small firms and a tendency towards a
reduction in bad loans after the second period. However, it is difficult to disentangle
potentially coexisting effects.
The estimates reported provide some insights into the effect of the structure of the
banking system on the volume and quality of credit. In particular, we find that concentration
has a positive and significant coefficient in the equation for credit to small and medium-sized
firms whereas it has a negative and significant coefficient in the case of large ones (Table 5-
A). Branch density has a positive significant effect on credit to small firms, consistent with
the intuition that the geographic expansion of banks is beneficial to local borrowers.
The share of branches of small banks has a positive effect on credit to both size
classes of firms for a relatively low value; the effect becomes negative as the share of small
banks increases. The derivative at the sample mean is positive for both classes of firms,
consistent with the view that small banks engage relatively more in traditional lending than
 24

large banks but that banking markets characterized by a disproportionate number of very
small banks reflect underdeveloped financial systems.
Results in Table 6-A show that the structure of the local banking system has a strong
effect on the probability of one monetary unit of credit falling in the category of bad loans.
The share of subsidized credit has a highly significant positive coefficient, corroborating the
intuition that government credit facilities can create moral hazard and distortions in fund
allocation. Branch density has a negative coefficient, which may suggest that the closeness
of banks to firms improves their screening and monitoring ability. The share of small banks
has a non-monotonic effect on the quality of credit; at the sample mean the coefficients for
both large and small firms are negative.
The Herfindahl index has a positive and significant coefficient, showing that higher
concentration is associated with poorer loan quality for all categories of borrowers. In the
case of small and medium-sized firms this result, combined with the finding that
concentration increases the volume of credit, is consistent with lower efficiency of banks in
concentrated markets (Table 6-B).
As a final check of the stability of the relationship between bank structure and credit
volumes, we have removed the consolidation and entry variables. As shown in table 7, the
results are consistent with those of table 5-A and 5-B. The effect of concentration is positive
for small and medium-sized firms and negative for large ones. By contrast, branch density
has a positive effect on credit availability for all firms. Concentration is associated with a
higher incidence of bad loans, measures of bank development are associated with a lower
one.

6. Conclusion
We have investigated the effects of structural change in the banking industry on the
availability of credit to small businesses, considering two sources of structural change:
consolidation and entry. We find that consolidation and entry influence the volume of credit
to both large firms and small and medium-sized ones. Specifically, mergers are followed by
a temporary reduction in credit and by an increase in the share of bad loans, which is
consistent with the view that they lead to efficiency improvements in lending policies and a
cut in credit to negative present value borrowers. However, based on our findings, we cannot
exclude some temporary disruption of relationships at the expense of non-negative net
 25

present value small borrowers. We do not find support for the hypothesis of a permanent
reduction in credit to small firms due to changes in bank size.
Our results show that entry has a negative impact on lending to all size categories,
significant for small ones, and no effect on bad loans. While this effect seems inconsistent
with the standard model of competition in which a larger number of suppliers in a market
yields lower interest rates and more credit, it may be explained by a “flight to quality” effect
driven by increased competition in lending and deposit markets. Finally, the estimates
suggest that the structure of local banking markets affects the quantity and quality of credit
in important ways. Concentration is associated with a larger volume of credit to small
borrowers, but also with a lower quality of the loan portfolio for all size categories. Branch
density has positive effect on credit flows. The share of small banks has a non-monotonic
effect on the quality of credit, possibly indicating that bank size is capturing other factors
such as the degree of development of the financial system.
Tables
Table 1
Size and debt composition of firms in selected countries
Data on the share of employment by firm size are from European Commission (1998). Small and
medium-sized enterprises are those with less than 250 employees. For the European countries figures are
of 1995, for Canada 1994, for the US and Japan 1993. Data on firms’ debt composition are computed for a
sample of firms contained in the European Commission Bank for the Accounts of Companies Harmonized
(BACH). Data refer to 1998 for Belgium, France and Italy, to 1997 for the other European countries, and
to 1995 for the US and Japan. Small and medium-sized firms are those with annual turnover below ECU
40 million. For the US, “Small and medium-sized firms” refers to firms with turnover up to ECU 7
million and “Large firms” refers to firms with turnover between ECU 7 million and ECU 40 million.

Share of employment accounted by Bank debt as a

small and medium-sized enterprises Percentage of total debt
Trade,
Industry and Cons- All Small and Large firms
hotels,
energy truction sectors medium-
restaurants
sized firms

Belgium 55.9 93.7 90.4 72.6 46.5 50.1

France 51.2 84.9 81.3 65.9 48,8 21.3
Germany 37.6 87.1 73.3 57.7 57.4 29.9
Italy 74.1 95.2 94.8 79.9 66.4 27.3
Netherlands 49.9 70.9 77.9 60.6 54.9 35.9
Spain 70.2 90.9 89.1 79.4 66.5 50.4
UK 47.9 88.2 58.1 56.9 n.a. n.a.
US 37.5 88.9 58.5 52.9 40.9 7.9
 Table 2
Banks by size and composition of loans
Banks are grouped according to the Bank of Italy’s classification introduced in 1995. In this paper we
have collapsed the official “major banks” and “large banks” classes into a single group referred to as
“large banks”. In the same way we consider “small banks” and “minor banks” jointly as “small banks”.
All data are end-of-year values. Banks chartered but with no assets at the reporting date are excluded.
Special credit divisions have been consolidated into the parent bank. Data on branches and total assets
come from the supervisory reports. Data on loans come from the aggregate statistics of the Central Credit
Register and refer to total outstanding loans (inclusive of bad loans) to customers whose total bank debt is
greater than ITL 250 million. Medium-sized firms are defined as all the non-financial enterprises with
bank debt between ITL 1 million and ITL 5 billion. Small firms are defined as non-financial firms with
bank debt between ITL 0.25 billion to ITL 1 billion.
1989 1992 1995 1998

Large banks:
Number 26 25 24 24
Branches 6,426 91,32 11,464 12,566
Average total assets (billions of lire) 35,095 49,995 63,684 72,730
National loan market share 57.0 58.4 59.8 59.2
Loan composition:
Large non-financial firms 42.9 42.9 43.2 41.6
Small and medium-sized non-financial firms 26.5 24.1 25.9 23.1
Other borrowers 30.7 33.0 31.5 35.3

Medium-sized banks:
Number 40 39 36 32
Branches 2,742 3,356 3,964 4,484
Average total assets (trillions of lire) 7,408 10,533 13,963 16,564
National loan market share 21.1 20.7 21.6 19.8
Loan composition:
Large non- financial firms 42.4 44.9 44.8 47.9
Small and medium-sized non-financial firms 32.6 30.1 30.4 28.7
Other borrowers 25.0 25.1 24.8 23.4

Small banks:
Number 1,057 1,000 912 862
Branches 6,401 7,523 8,210 9,577
Average total assets (trillions of lire) 408 562 718 979
National loan market share 21.9 20.9 18.6 20.9
Loan composition:
Large non-financial firms 31.2 32.9 32.1 34.5
Small and medium-sized non- financial firms 45.5 43.9 46.9 43.5
Other borrowers 23.2 23.2 21.0 22.0

Total banks
Number 1,176 1,073 970 921
Branches 15,569 20,011 23,638 26,627
Average total assets (trillions of lire) 1,462 2,086 2,802 3,365
Loan composition:
Large non-financial firms 40.6 41.6 41.7 41.8
Small and medium-sized non-financial firms 31.3 28.8 29.7 27.6
Other borrowers 28.1 29.6 28.5 30.5

Sources: Italian Central Credit Register and Bank Supervisory Reports.

 Table 3
Banks by size and distribution of local markets by number of M&As
Banks are grouped according to the Bank of Italy’s classifications introduced in 1995. If the acquired
bank loses its charter and the acquiring bank maintains it, we count only the target bank; in the case of
a merger that originates a new institution we count all the banks involved. These cases are all included
in the category “Mergers”. Within the category “Acquisitions” the creation of bank groups are treated
in the same way: whenever there is a clear leader we consider only the target banks, otherwise all the
institutions involved are included. Provincial market shares are expressed in percentage points. The
number of provinces indicates the number of cases in which the market share shifted by mergers and
acquisitions falls in the percentage class specified.
1990-92 1993-95 1996-98
Mergers Acquisitions Mergers Acquisitions Mergers Acquisitions

Number of Banks
Size group:

Small banks 123 9 150 14 80 62

Other banks 5 0 10 10 2 11
Total banks 128 9 160 24 82 73

Mergers Acquisitions Mergers Acquisitions Mergers Acquisitions

Number of provinces
Percentage of loans
shifted by M&As:
From 0 to 1 46 34 38 40 49 30
From 1 to 2 16 2 16 12 9 11
From 2 to 5 16 2 20 18 9 23
5 and more 16 2 16 24 6 29

Sources: Italian Central Credit Register and Bank Supervisory Reports.

 Table 4
Descriptive statistics
The statistics reported for the explanatory variables with the exception of ln(Value Added) are calculated for
end-of-previous-year values, as employed in the estimation.
Mean St. Dev. Minimum Maximum
1995-98 1990-98 1995-98 1990-98 1995-98 1990-98 1995-98 1990-98
Dependent Variables

Ln(loans) 14.423 14.237 1.227 1.250 10.237 10.058 18.242 18.242

Large Firms
Ln(loans) Small & 14.530 14.406 0.882 0.882 12.367 12.001 17.252 17.252
Medium-sized Firms
Ln(Odds ratio) -2.503 -3.007 2.445 2.948 -22.659 -23.391 0.449 0.449
Large Firms
Ln(Odds ratio) Small & -1.659 -1.933 0.796 0.737 -3.566 -4.078 0.053 0.053
Medium-sized Firms

Explanatory Variables

Sh_merged 0.026 0.026 0.045 0.045 0 0 0.361 0.361

Sh_acquired 0.032 0.021 0.059 0.049 0 0 0.431 0.431

Entry 0.014 0.014 0.040 0.029 0 0 0.407 0.407

Herfindahl 0.092 0.092 0.041 0.044 0.035 0.028 0.244 0.269

Branch_density 4.421 3.954 1.589 1.750 1.459 0.398 9.702 17.189

Sh_smallbanks 0.398 0.422 0.201 0.210 0.040 0.040 0.948 0.959

LnGDP 16.336 16.195 0.790 0.792 14.342 14.200 19091 19.091

Sh_subscredit 0.110 0.115 0.066 0.061 0.013 0.013 0.380 0.380

Sources: Italian Central Credit Register and Bank Supervisory Reports. GDP and population data are provided
by the Union of the Italian Chambers of Commerce.
 Table 5-A
Structural change and market loans by type of borrower (1990-98)
The dependent variable is defined as ln(loansi), where i indicates each type of borrower and is equal to:
large non-financial firms, small and medium-sized non-financial firms. The number of observations
employed in the estimation is 855; province fixed effects and year dummy variables are included in each
regression (coefficients for these variables are not shown). Standard errors are reported below
coefficients.
Credit to: Large Firms Large Firms Small and Medium- Small and Medium-
sized Firms sized Firms

Constant 15.379 *** 14.434 *** 14.796 *** 14.104 ***

0.679 0.790 0.247 0.280
Sh_merged_1 -0.234 -0.235 -0.197 *** -0.189 ***
0.163 0.163 0.059 0.058
Sh_merged_2 -0.060 -0.069 -0.026 -0.050
0.177 0.179 0.064 0.063
Sh_merged_3 0.142 0.133 0.063 -0.029
0.177 0.179 0.064 0.063
Sh_acquired_1 0.204 0.188 0.141 ** 0.114 **
0.154 0.154 0.056 0.054
Sh_acquired_2 0.167 0.178 0.102 * 0.091
0.162 0.162 0.059 0.057
Sh_acquired_3 0.355 ** 0.362 ** 0.060 0.036
0.167 0.169 0.061 0.060
Entry_1 -0.435 * -0.426 * -0.569 *** -0.557 ***
0.243 0.241 0.088 0.086
Entry_2 -0.190 -0.239 -0.240 -0.272 *
0.418 0.416 0.151 0.148
Entry_3 1.110 *** 1.052 ** -0.294 * -0.335 **
0.411 0.408 0.149 0.145
Herfindahl_1 - -1.283 ** - 0.636 ***
- 0.550 - 0.195
Branch_density_1 - 0.014 * - 0.011 ***
- 0.008 - 0.003
Sh_smallbanks_1 - 0.970 ** - 0.674 ***
- 0.446 - 0.158
Sh_smallbanks squared_1 - -1.023 ** - -0.749 ***
- 0.438 - 0.155
LnGDP -0.055 -0.005 -0.011 0.015
0.041 0.046 0.015 0.016
Sh_subscredit_1 0.291 0.125 0.375 *** -0.177
0.310 0.330 0.112 0.117

R-squared (within) 0.602 0.611 0.844 0.853

F(23[19],737[741]) 59.20 50.33 211.23 187.27
F(94,737[741]) that all ui=0 61.54 51.10 155.57 126.07

Note: Statistically different from zero, respectively, at: *** 99 percent, **95 percent and *90 percent
significance level.
 Table 5-B
Structural change and market loans by type of borrower (1995-98)
The dependent variable is defined as ln(loansi), where i indicates each type of borrower and is equal to:
large non financial firms, small and medium-sized non-financial firms. The number of observations
employed in the estimation is 380; province fixed effects and year dummy variables are included in each
regression (coefficients for these variables are not shown). Standard errors are reported below
coefficients.
Credit to: Large Firms Large Firms Small and Medium- Small and Medium-
sized Firms sized Firms

Constant 22.015 *** 20.602 *** 16.833 *** 15.297 ***

4.001 4.128 1.399 1.253
Sh_merged_1 -0.697 *** -0.539 ** -0.420 *** -0.150 **
0.220 0.245 0.077 0.074
Sh_merged_2 -0.527 ** -0.460 ** -0.252 *** -0.196 ***
0.214 0.217 0.074 0.066
Sh_merged_3 -0.365 * -0.343 * -0.084 -0.119 *
0.204 0.205 0.071 0.062
Sh_acquired_1 0.172 0.151 0.123 ** 0.091 *
0.176 0.176 0.061 0.053
Sh_acquired_2 0.293 * 0.244 0.161 *** 0.085 *
0.152 0.155 0.053 0.047
Sh_acquired_3 0.403 ** 0.350 ** 0.113 * 0.002
0.167 0.172 0.058 0.052
Entry_1 -0.275 -0.278 -0.486 *** -0.429 ***
0.228 0.231 0.079 0.070
Entry_2 -1.167 ** -1.116 * -0.628 *** -0.762 ***
0.531 0.567 0.185 0.172
Entry_3 -0.226 -0.220 -0.400 ** -0.401 **
0.545 0.553 0.190 0.168
Herfindahl_1 - -1.103 - 0.572 **
- 0.745 - 0.226
Branch_density_1 - 0.081 - 0.160 ***
- 0.064 - 0.019
Sh_smallbanks_1 - 0.974 - 0.386
- 1.563 - 0.475
Sh_smallbanks squared_1 - -1.015 - -0.826 *
- 1.630 - 0.494
LnGDP -0.456 * -0.397 -0.135 -0.092
0.244 0.248 0.085 0.075
Sh_subscredit_1 0.155 0.009 -0.281 * -0.016
0.463 0.489 0.161 0.148

R-squared (within) 0.325 0.335 0.328 0.502

F(18[14],267[271]) 9.33 7.50 9.47 14.95
F(94,267[271]) that all ui= 0 41.65 37.07 99.37 96.76

Note: Statistically different from zero, respectively, at: *** 99 percent, **95 percent and *90 percent
significance level.
 Table 6-A
Market odds ratios of bad loans by type of borrower (1990-98)
The dependent variable is defined as ln[(Bad Loansi /Loansi )/(1 - (Bad Loansi /Loansi))], where i
indicates each type of borrower and is equal to: large non-financial firms, small and medium-sized non-
financial firms. The number of observation employed in the estimation is 855; market fixed effects and
year dummy variables are included in each regression (coefficients for these variables are not shown).
The regression is estimated with weighted least squares. Values of zero for the volume of bad loans have
been replaced with 1.
Odds ratio of bad loans for: Large Firms Large Firms Small and Medium- Small and Medium-
sized Firms sized Firms

Constant -6.852 * 2.625 -3.200 ** 2.546

3.503 4.022 1.587 1.747
Sh_merged_1 1.584 *** 1.697 *** 0.784 *** 0.863 ***
0.369 0.373 0.216 0.210
Sh_merged_2 0.965 ** 1.031 *** 0.259 0.335
0.402 0.398 0.229 0.223
Sh_merged_3 0.541 0.732 ** -0.049 0.026
0.377 0.372 0.220 0.212
Sh_acquired_1 -1.030 *** -0.876 ** -0.629 *** -0.539 ***
0.382 0.376 0.199 0.191
Sh_acquired_2 -0.945 ** -0.883 ** -0.444 ** -0.431 **
-0.387 0.379 0.203 0.194
Sh_acquired_3 -1.051 *** -0.898 ** -0.519 ** -0.465 **
0.401 0.396 0.210 0.203
Entry_1 0.169 0.352 0.122 0.213
0.619 0.611 0.306 0.294
Entry_2 -1.786 -1.293 -0.524 -0.266
1.358 1.334 0.650 0.625
Entry_3 -0.325 0.042 -0.259 -0.047
1.322 1.297 0.642 0.617
Herfindahl_1 - 3.525 *** - 2.737 ***
- 1.092 - 0.620
Branch_density_1 - -0.130 *** - -0.084 ***
- 0.030 - 0.013
Sh_smallbanks_1 - -2.720 ** - -1.591 ***
- 1.224 - 0.601
Sh_smallbanks squared_1 - 2.997 ** - 1.645 ***
- 1.333 - 0.618
LnGDP 0.163 -0.328 0.046 -0.252 ***
0.191 0.216 0.086 0.094
Sh_subscredit_1 1.573 2.230 ** 1.949 *** 2.432 ***
0.962 0.970 0.466 0.468

Adj. R-squared: 0.837 0.844 0.912 0.919

F(117[113], 738[741]) 40.03 40.66 80.02 84.75

Note: Statistically different from zero, respectively, at: *** 99 percent, **95 percent and *90 percent
significance level.
 Table 6-B
Market odds ratios of bad loans by type of borrower (1995-98)
The dependent variable is defined as ln[(Bad Loansi /Loansi )/(1 - (Bad Loansi /Loansi))], where i
indicates each type of borrower and is equal to: large non-financial firms, small and medium-sized non-
financial firms. The number of observation employed in the estimation is 380; market fixed effects and
year dummy variables are included in each regression (coefficients for these variables are not shown).
The regression is estimated with weighted least squares. Values of zero for the volume of bad loans have
been replaced with 1.

Odds ratio of bad loans for: Large Firms Large Firms Small and Medium- Small and Medium-
sized Firms sized Firms

Constant -43.008 *** -33.915 *** -23.675 *** -16.885 ***

9.777 10.452 3.804 3.651
Sh_merged_1 2.440 *** 1.941 *** 0.868 *** 0.341 *
0.402 0.435 0.179 0.178
Sh_merged_2 1.797 *** 1.645 *** 0.570 *** 0.403 **
0.410 0.413 0.171 0.158
Sh_merged_3 1.343 *** 1.402 *** 0.262 0.246 *
0.364 0.378 0.160 0.148
Sh_acquired_1 -0.676 -0.587 -0.213 -0.180
0.417 0.420 0.161 0.148
Sh_acquired_2 -0.870 ** -0.591 -0.284 ** -0.074
0.354 0.364 0.132 0.123
Sh_acquired_3 -0.589 -0.311 -0.299 ** -0.070
0.388 0.401 0.146 0.137
Entry_1 0.088 -0.044 -0.294 -0.414 **
0.543 0.549 0.191 0.177
Entry_2 -0.402 0.478 -0.695 -1.015 **
1.529 1.576 0.528 0.499
Entry_3 0.414 0.096 -0.583 -0.833 *
1.493 1.511 0.527 0.489
Herfindahl_1 - -0.919 - 0.132
- 1.048 - 0.435
Branch_density_1 - -0.401 *** - -0.378 ***
- 0.138 - 0.050
Sh_smallbanks_1 - 0.403 - 2.437 **
- 2.634 - 1.041
Sh_smallbanks squared_1 - 0.327 - -2.696 **
- 3.136 - 1.132
LnGDP 2.134 *** 1.726 *** 1.172 *** 0.865 ***
0.535 0.562 0.208 0.197
Sh_subscredit_1 -0.491 -0.685 0.256 0.242
1.166 1.193 0.441 0.419

Adj. R-squared: 0.935 0.937 0.982 0.985

F(112[108], 267[271]) 52.19 51.31 198.08 232.34

Note: Statistically different from zero, respectively, at: *** 99 percent, **95 percent and *90 percent
significance level.
 Table 7
Market structure, credit volume and quality by type of borrower (1990-98)
In columns 1 and 2 the dependent variable is defined as ln(Loansi), where i indicates each type of
borrower and is equal to: large non-financial firms, small and medium-sized non-financial firms. In
columns 3 and 4 the dependent variable is defined as ln[(Bad Loansi /Loansi)/(1 - (Bad Loansi /Loansi ))],
where i indicates each type of borrower and is equal to: large non-financial firms, small and medium-
sized non-financial firms. The number of observation employed in the estimation is 855; market fixed
effects and year dummy variables are included in each regression (coefficients for these variables are not
shown). The regressions in column 3 and 4 are estimated with weighted least squares. Values of zero for
the volume of bad loans have been replaced with 1.
Credit Log-odds
Credit to: Large Firms Small and Large Firms Small and
Medium-sized Medium-sized
Firms Firms

Constant 14.354 *** 13.969 *** 3.786 2.862

0.791 0.291 4.059 1.765
Herfindahl_1 -1.112 ** 0.701 *** 3.455 *** 2.573 ***
0.551 0.203 1.099 0.625
Branch_density_1 0.015 * 0.011 *** -0.128 *** -0.083 ***
0.008 0.003 0.030 0.013
Sh_smallbanks_1 1.161 *** 0.718 *** -2.486 ** -1.729 ***
0.443 0.163 1.235 0.604
Sh_smallbanks squared_1 -1.191 ** -0.786 *** 3.095 ** 1.871 ***
0.438 0.161 1.351 0.624
LnGDP -0.001 -0.021 -0.388 * -0.268 ***
0.046 0.017 0.218 0.095
Sh_subscredit_1 0.066 -0.193 1.818 * 2.349 ***
0.330 0.121 0.982 0.472

R-squared: Within 0.600 0.839 - -

F(94, 746) that all ui=0 50.64 116.21 - -
F(14,746) 80.15 287.07 - -
Adj. R-squared: - - 0.837 0.917
F(108, 746) - - 41.79 88.52

Note: Statistically different from zero, respectively, at: *** 99 percent, **95 percent and *90 percent
significance level.
 References

Asea, P., and B. Blomberg (1998), “Lending Cycles,” Journal of Econometrics 83: 89-128.

Avery, R.A., and K. Samolyk (1999), “Bank consolidation and small business lending
patterns,” mimeo, Federal Reserve Board and FDIC.

Berger, A.N., S.D. Bonime, L.G. Goldberg, and L.J. White (1999), “The Dynamics of
Market Entry: The Effects of Mergers and Acquisitions on De Novo Entry and Small
Business Lending in the Banking Industry,” Finance and Economics Discussion Series,
Board of Governors of the Federal Reserve System.

Berger, A.N., R. Demsetz, and P.E. Strahan (1999), “The Consolidation of the Financial
Services Industry: Causes, Consequences, and Implications for the Future,” Journal of
Banking and Finance 23: 135-194.

Berger, A.N., and R. DeYoung (1997), “Problem Loans and Cost Efficiency in Commercial
banks,” Journal of Banking and Finance 21: 219-236.

Berger, A.N., A.K. Kashyap, and J.M. Scalise, (1995), “The transformation of the US
banking industry: What a long, strange trip it’s been,” Brookings Papers on Economic
Activity 2: 55-218.

Berger, A.N., A. Saunders, J.M. Scalise, and G.F. Udell (1998), “The effects of bank
mergers and acquisitions on small business lending,” Journal of Financial Economics 50:
187-229.

Berger, A.N., and G.F. Udell (1996), “Universal Banking and the future of small business
lending,” in: A. Saunders and I. Walter, (eds.), Financial system design: The case for
universal banking, Burr Ridge, IL, Irwin.

Berger, A.N., and G.F. Udell (1998), “The Economics of Small Business Finance: The Role
of Private Equity and Debt Markets in the Financial Growth Cycle,” Journal of Banking
and Finance 22: 613-673.

Black, S.E., and P.E. Strahan (2000), “Entrepreneurship and Bank Credit Availability,”
mimeo, Federal Reserve Bank of New York.

Bonaccorsi di Patti, E., and G. Dell’Ariccia (2001), “Bank Competition and Firm Creation,”
IMF Working Paper 01/21.

Boot, A. (2000), “Relationship Banking: What Do We Know?,” Journal of Financial

Intermediation 9:7-25.

Broecker, T. (1990), “Credit-worthiness Tests and Interbank Competition,” Econometrica

58: 429-452.
 36

Craig, B.R., and J.A.C. Santos (1997), “Banking Consolidation Impact on Small Business
Lending,” Working Paper, Federal Reserve Bank of Cleveland.

Detragiache, E., L. Guiso, and P.G. Garella (2000), “Multiple versus Single Banking
Relationships: Theory and Evidence,” Journal of Finance 55: 1133-1161.

DeYoung, R., L. Goldberg, and L. White (1999), “Youth, Adolescence, and Maturity of
Banks: Credit Availability to Small Business in an Era of Banking Consolidation,” Journal
of Banking and Finance 23: 463-492.

Focarelli, D., F. Panetta, and C. Salleo (1999), “Why do banks merge?,” Temi di
Discussione, Banca d’Italia, No. 361.

Gobbi, G. (1996), “Government Credit Facilities and Interbank Market Developments in

Italy,” in H. Mehran, B. Laurens and M. Quintyn (eds.), Interest Rate Liberalization and
Money Market Development, IMF.

Goldberg, L, and L. White (1998), “De novo banks and lending to small businesses: An
empirical analysis,” Journal of Banking and Finance 22: 851-867.

Hannan, T. (1991), “Bank Commercial Loan Markets and the Role of Market Structure:
Evidence from Surveys of Commercial Lending,” Journal of Banking and Finance 15:
133-149.

Jackson, J., and A. Thomas (1995), “Bank Structure and New Business Creation Lessons
from an Earlier Time,” Regional Science and Urban Economics, Vol. 25, pp. 323-353.

Keeton, W.R. (1995), “Multi-office bank lending to small businesses: some new evidence,”
Economic Review, Federal Reserve Bank of Kansas City 80: 45-57.

Keeton, W.R. (1997), “The effects of mergers on farm and business lending at small banks:
New evidence from tenth district states,” Working Paper, Federal Reserve of Kansas City.

Kwast M., M. Starr-McCluer, and J. Wolken (1997), “Market Definition and the Analysis of
Antitrust in Banking,” The Antitrust Bulletin, Winter.

Jayaratne, J., and J. Wolken (1999), “How important are small banks to small businesses
lending? New evidence from a survey of small firms,” Journal of Banking and Finance 23:
427-458.

Lang, W., and L. Nakamura (1995), “Flight to quality in banking and economic activity,”
Journal of Monetary Economics 36: 145-164.

Levonian, M., and J. Soller (1995), “Small Banks, Small Loans, Small Businesses,” Business
Review, Federal Reserve Bank of Philadelphia, September/October: 3-16.
 37

Miller, M. (2000), “Credit Reporting Systems Around the Glob: The State of The Art in
Public and Private Credit Registries,” mimeo, World Bank.

Peek, J. and E.S. Rosengren (1996), “Small business credit availability: How important is
size of lender?” in: A. Saunders and I. Walter (eds.), Financial system design: The case for
universal banking, Burr Ridge, IL, Irwin.

Peek, J., and E.S. Rosengren (1998), “Bank Consolidation and small business lending: it’s
not just bank size that matter,” Journal of Banking and Finance 22: 799-819.

Petersen, M.A., and R. Rajan (1994), “The benefits of lending relationships: Evidence from
small business data,” Journal of Finance 49: 3-37.

Petersen, M.A., and R. Rajan (1995), “The effect of credit market competition on lending
relationships,” Quarterly Journal of Economics 110: 407-443.

Petersen, M.A., and R. Rajan (2000), “Does Distance Still Matter? The Information
Revolution in Small Business Lending,” mimeo, Northwestern University and University
of Chicago.

Riordan, M. (1993), “Competition and bank performance: a theoretical perspective,” in C.

Mayer and X. Vives (eds.), Capital markets and financial intermediation, CUP,
Cambridge UK.

Sapienza, P. (1998), “The effects of bank mergers on loan contracts,” Northwestern

University working paper.

Strahan P.E., and J.P. Weston (1996), “Small business lending and bank consolidation: Is
there cause for concern?,” Current Issues in Economics and Finance 2, Federal Reserve
Bank of New York: 1-6.

Strahan P.E., and J.P. Weston (1998), “Small business lending and the changing structure of
the banking industry,” Journal of Banking and Finance 22: 821-845.

Thakor, A.V., and R. Callaway (1983), “Costly information, bank lending, and implicit
contracts: a stylized model of customer relationship,” Journal of Financial and
Quantitative Analysis 2: 229-256.

Von Thadden, E. L. (1998), “Asymmetric Information, Bank lending and Implicit Contracts:
The Winner’s Curse,” mimeo, Université de Lausanne.

Whalen, G. (1995), “Out-of-state holding company affiliation and small business lending”,
Economic and Policy Analysis Working Paper 95-4, Office of the Comptroller of the
Currency.

Walraven, N. (1997), “Small business lending by banks involved in mergers,” Finance and
Economics Discussion Series 97-25, Board of Governors of the Federal Reserve System.
 RECENTLY PUBLISHED “TEMI” (*)

No. 379 — Promise and Pitfalls in the Use of “Secondary” Data-Sets: Income Inequality in
OECD Countries, by A. B. ATKINSON and A. BRANDOLINI (October 2000).
No. 380 — Bank Competition and Regulatory Reform: The Case of the Italian Banking
Industry, by P. ANGELINI and N. CETORELLI (October 2000).
No. 381 — The Determinants of Cross-Border Bank Shareholdings: an Analysis with
Bank-Level Data from OECD Countries, by D. FOCARELLI and A. F. POZZOLO
(October 2000).
No. 382 — Endogenous Growth with Intertemporally Dependent Preferences, by G.
FERRAGUTO and P. PAGANO (October 2000).
No. 383 — (Fractional) Beta Convergence, by C. MICHELACCI and P. ZAFFARONI (October
2000).
No. 384 — Will a Common European Monetary Policy Have Asymmetric Effects?, by L.
GUISO, A. K. KASHYAP, F. PANETTA and D. TERLIZZESE (October 2000).
No. 385 — Testing for Stochastic Trends in Series with Structural Breaks, by F. BUSETTI
(October 2000).
No. 386 — Revisiting the Case for a Populist Central Banker, by F. LIPPI (October 2000).
No. 387 — The multimarket contacts theory: an application to italian banks, by R. DE BONIS
and A. FERRANDO (December 2000).
No. 388 — La “credit view” in economia aperta: un’applicazione al caso italiano, by P.
CHIADES and L. GAMBACORTA (December 2000).
No. 389 — The monetary trasmission mechanism: evidence from the industries of five OECD
countries, by L. DEDOLA and F. LIPPI (December 2000).
No. 390 — Disuguaglianza dei redditi individuali e ruolo della famiglia in Italia, by
G. D’ALESSIO and L. F. SIGNORINI (December 2000).
No. 391 — Expectations and information in second generation currency crises models, by M.
SBRACIA and A. ZAGHINI (December 2000).
No. 392 — Unobserved Factor Utilization, Technology Shocks and Business Cycles, by D. J.
MARCHETTI and F. NUCCI (February 2001).
No. 393 — The Stability of the Relation between the Stock Market and Macroeconomic Forces,
by F. PANETTA (February 2001).
No. 394 — Firm Size Distribution and Growth, by P. PAGANO and F. SCHIVARDI (February
2001).
No. 395 — Macroeconomic Forecasting: Debunking a Few Old Wives’ Tales, by S. SIVIERO
and D. TERLIZZESE (February 2001).
No. 396 — Recovering the Probability Density Function of Asset Prices Using GARCH as
Diffusion Approximations, by F. FORNARI and A. MELE (February 2001).
No. 397 — A Simple Approach to the Estimation of Continuous Time CEV Stochastic Volatility
Models of the Short-Term Rate, by F. FORNARI and A. MELE (February 2001).
No. 398 — La convergenza dei salari manifatturieri in Europa, by P. CIPOLLONE (February
2001).
No. 399 — Labor Income and Risky Assets under Market Incompleteness: Evidence from
Italian Data, by G. GRANDE and L. VENTURA (March 2001).
No. 400 — Is the Italian Labour Market Segmented?, by P. CIPOLLONE (March 2001).
No. 401 — Optimal Debt Maturity under EMU, by R. GIORDANO (March 2001).
No. 402 — Il modello di specializzazione internazionale dell’area dell’euro e dei principali
paesi europei: omogeneità e convergenza, by M. BUGANELLI (March 2001).

No. 403 — Seasonality and Capacity: an Application to Italy, by G. DE BLASIO and F. MINI
(June 2001).

(*) Requests for copies should be sent to:

Banca d’Italia -- Servizio Studi -- Divisione Biblioteca e pubblicazioni -- Via Nazionale, 91 -- 00184 Rome
(fax 0039 06 47922059). They are available on the Internet at www.bancaditalia.it