The Dynamics of Market Entry: The Effects of Mergers and Acquisitions on De Novo Entry and Small Business Lending in the Banking Industry

The Dynamics of Market Entry: The Effects of Mergers and Acquisitions on De Novo Entry and Small Business Lending in the Banking Industry

Allen N. Berger Board of Governors of the Federal Reserve System Washington, DC 20551 U.S.A. Wharton Financial Institutions Center Philadelphia, PA 19104 U.S.A.

Seth D. Bonime Yale School of Management New Haven, CT 06520 U.S.A.

Lawrence G. Goldberg Department of Finance, University of Miami Coral Gables, FL 33124 U.S.A.

Lawrence J. White Stern School of Business, New York University New York, NY 10012 U.S.A.

July 1999

Abstract

We study the dynamics of market entry following mergers and acquisitions (M&As), and the behavior of recent entrants in supplying output that might be withdrawn by the consolidating firms. The data, drawn from the banking industry, suggest that M&As are associated with subsequent increases in the probability of entry. The estimates suggest that M&As explain more than 20% of entry in metropolitan markets, and more than 10% of entry in rural markets. Additional results suggest that bank age has a strong negative effect on the small business lending of small banks, but that M&As have little influence on this lending.

JEL Classification Numbers: G21, G28, G34, E58, L89 Keywords: Entry, Barriers to Entry, Bank, Mergers, Small Business

The opinions expressed do not necessarily reflect those of the Board of Governors or its staff. The authors thank John Boyd for excellent discussant comments, Bob Avery, Sally Davies, Bob DeYoung, Luke Froeb, Al Gilbert, Greg Werden, and the participants at the Chicago Federal Reserve Conference on Bank Structure and Competition for helpful comments, and Kelly Bryant for outstanding research assistance.

Please address correspondence to Allen N. Berger, Mail Stop 153, Federal Reserve Board, 20th and C Sts. NW, Washington, DC 20551, call 202-452-2903, fax 202-452-5295, or email aberger @frb.gov.

I. Introduction

Entry, or the credible threat of entry, is important in many aspects of industrial organization, from theory to empirical research to institutional practice. All of these aspects share a concern that participants in concentrated markets may not behave in a competitive, efficient fashion that keeps goods and services available to consumers at prices close to marginal costs unless there is active market entry or a credible threat of entry. The theory has generated a number of factors that should be important to market entry, the empirical literature has tested some of these hypothesized factors, and antitrust authorities have weighed these factors in their approval/denial processes.

However, the literature is largely missing dynamic analyses of market entry. The theories and tests to date have primarily focused on factors such as profitability, concentration, and growth of the market, which may proxy for the prospects of future abnormal profits and invite entry. There has been little dynamic analysis of entry in response to the activities of the incumbent market participants, and in particular, there has been little analysis of the effects of mergers and acquisitions (M&As) on entry. This may be surprising, given the important role that M&As play in changing the structure and competitive conditions in a market, and the role that the prospects for future entry in response to M&As plays in antitrust analysis.

The issue of entry into local banking markets in response to bank M&As is a critical policy issue as well as an interesting research question. Whether or not there is new entry after consolidation activity has important implications for antitrust policy, for the efficiency of banking markets, and for the adequate supply of funds to small business borrowers. There are a number of studies of entry into local banking markets that use conventional explanatory variables, such as profits, concentration, growth, etc. However, only one study of which we are aware examines entry in response to M&As, and it finds the unexpected result that bank

consolidation tends to reduce the probability of entry into local banking markets (Seelig and Critchfield

1999). As will be seen, we find substantially different results.

There is also little prior evidence on the effects of entry on customer service. If M&As result in fewer customers being served or a deterioration in service quality, an important question is whether entry makes up for these losses. If new entry replaces any lost output/service quality after M&As, then vigorous antitrust policy may not be as important as would otherwise be the case. Although the entry literature has

investigated the conditions under which firms enter the market, the literature has generally not examined the

2 output of these firms to determine whether they are taking up the slack from noncompetitive behavior, or whether their output depends in a dynamic way on recent M&A activity.

In this research, we address both of these issues. First, we build a dynamic empirical model of entry into local banking markets that includes M&A activity and other market factors. Second, we model the small business lending output provided by the new entrants as a function of M&A activity and the other market factors, as well as variables measuring the size and condition of the bank itself. Thus, we not only model the entry process, but we measure the extent to which entry helps supply the output that may be withdrawn by firms engaging in consolidation activity.

The commercial banking industry provides an almost ideal laboratory for analyzing these issues. First, banks produce relatively homogeneous products in a large number of geographically segmented markets of various sizes, with different M&A experiences, and other variations in market structure and economic conditions. We examine entry activity in over 2,700 local banking markets — Metropolitan Statistical Areas (MSAs) or non-MSA counties across the U.S. — during a period of 19 years from 1980- 1998, for a total of over 52,000 market-year observations of whether entry occurs, and almost 4,000 actual market entries. We are able to trace the effects of over 10,000 bank M&As over the time period and to distinguish among types of consolidation. This contrasts well with most other studies of entry, which typically have many fewer observations, often use 4-digit SIC categories that represent much more heterogenous products and services, and generally do not have information on M&A activity.

The use of banking markets also allows us to examine the effects of differences in barriers to entry created by regulatory restrictions. Over our sample period, a number of states had unit banking laws and other regulations that restricted the ability of incumbent firms from other markets from entering local banking markets. We hypothesize that these legal restrictions on competition would increase de novo entry by creating local market power and limiting the ability of existing banks outside the local market to enter by opening branch offices. That is, barriers to one type of entry (branching by existing banks in other markets) may create incentives and opportunities for another type of entry (de novo entry by banks with new charters).

Our analysis also addresses directly several important policy issues in banking. It has been found

that some types of M&As tend to reduce small business lending by the M&A participants. This could be

3

because of organizational diseconomies from serving relationship-based small business borrowers along with the large transactions-based customers that large banks specialize in serving. The reduction in services to small customers may also result in part from the increased investment opportunities to serve large customers afforded by larger bank size, which “crowds out” small business loans in the use of increasingly costly funds. A reduction in services provided to small customers might also result in part from short-term disruptions caused by the consolidation process that gives other banks opportunities to take away customers who perceive a reduction in service quality or availability. The press often reports substantial customer runoffs after an M&A.

Consistent with these arguments and observations, some research has suggested a substantial positive external effect of M&As on the small business lending of other banks in the local market. However, there is little prior evidence on the extent to which this external effect was the result of established, mature banks increasing their small business lending versus new entry and/or additional lending by recent entrants. We address this issue by examining the probability of entry and the small business lending output of recent entrants as functions of local market M&As.

The remainder of the paper is organized as follows. Section 2 gives some facts about de novo entry that are of interest and that help guide the empirical analysis. Section 3 briefly reviews some of the relevant empirical literature on entry in general and in the banking industry. Section 4 describes the empirical literature on the effects of bank M&As that may help determine the probability of entry and small business lending by recent entrants and other small banks. Section 5 discusses the hypotheses, methodology, and variables employed in the analysis. Sections 6 and 7 summarize the empirical results for the probability of entry and small business lending, respectively. Section 8 concludes.

2. Some facts about de novo entry

This section presents some statistics about de novo entry in U.S. banking and some of the factors that are associated with this entry. These data suggest some potentially important determinants of entry that are incorporated in our multivariate empirical analysis below.

Table | reports statistics on the annual frequency of de novo entry into local U.S. banking markets

—— defined as Metropolitan Statistical Areas (MSAs) or non-MSA counties — over the period 1980-1998.

4

As shown in the bottom row of the table, there were 52,598 total market years, or 19 years times an average of 2,768 local markets in each year.’ The bottom row also shows that 50,383 times or 95.79% of the time there were no new commercial banks entering the market. Nonetheless, the remaining 4.21% of the time still represents a tremendous amount of entry, with 3,875 total entries or 7.37 entries per 100 local markets per year. Given that there is no entry the vast majority of the time, it is perhaps surprising that a number of markets had many entrants in the same year. On 124 separate occasions, there were 5 or more entries in the same market in the same year, with a maximum of 40 entries.

The data in Table | suggest that the size of the market is a strong determinant of the probability of entry and number of entrants. This is not surprising —- a large market should have more of all structural changes, including entry, than a small market. The size classes are based on total deposits in the market, adjusted to be in constant 1994 dollars using the GDP deflator.” In the smallest two market size classes shown — local markets with less than $300 million in total bank deposits — there was only about a 1% - 2% probability of entry, and there were no cases of more than 2 entries in the same market in the same year. On average, these small markets had only about | to 2 entrants per 100 markets per year and had a total of only 537 entries or about 14% of all entries, despite having about 73% of the total market-years. By contrast, in our largest market size class of over $5 billion in deposits, entry occurred 51.32% of the time, with most cases of entry involving multiple entrants. In 8.92% of these market-years, there were 5 or more entrants. On average, these large markets had 158.28 entrants per 100 markets per year, or more than 1.5 entrants per market per year. The 2,094 entries in these large markets account for more than half of all 3,875 de novo entries, despite the fact that these large markets account for less than 3% of the total market years.

Tables 2 and 3 segment the data by metropolitan (MSA) versus rural (non-MSA) markets. Not

surprisingly, almost all of the large markets are metropolitan. Of the 1,323 market years with deposits over

‘Although these market definitions are imperfect, there is considerable research evidence supporting these local geographic demarcations as the markets for retail banking products such as small business loans and retail deposits. Survey data indicate that small businesses and households almost always choose a local financial institution (Kwast, Starr-McCluer, and Wolken 1997), while corporate customers may shop for wholesale financial services in national or international markets.

*Market deposits are determined from the FDIC Summary of Deposits data, the only publicly available source of information about the location of banking offices within a state.

5

$5 billion, 1,321 are MSAs and only 2 are rural. Also not surprising, almost all of the small markets are tural. Of the 38,389 market years with deposits under $300 million, 38,372 are rural and only 17 are MSAs. In the middle size classes in which both metropolitan and rural markets are well represented, metropolitan markets appear to have substantially higher probabilities of entry and greater numbers of entrants per market. For example, in the $1 billion - $5 billion deposit class, MSAs had a probability of entry of 18.04% and an average of 23.72 entrants per 100 markets, whereas non-MSA markets had entry probability and entrants per 100 markets less than half — 7.89% and 9.65, respectively. Prior research often finds that metropolitan and rural banking markets differ in competitive conditions, and these data confirm the prior research.

Based on these findings, we conduct separate multivariate regression analyses for three subsets of the data — 1) large metropolitan markets (deposits > $5 billion); 2) small metropolitan markets (deposits < $5 billion); and 3) rural markets of all sizes. This segmentation provides a better chance of disentangling the effects of M&As and other important variables from the effects of market size and market type, which are highly correlated with both entry and M&A activity. The large metropolitan markets have the advantage for our logit equations of having a sample average probability of entry well away from the boundaries of 0 and 1, which should help us differentiate among the hypotheses about M&A activity and other factors. However, the rural markets have the advantage of having many more observations.

Table 4 shows the frequency of entry by year and by whether or not there was any M&A activity in the local market in the past 3 years — i.e., by whether or not at least one local bank survived a merger of bank charters or was acquired by a different top-tier bank holding company (BHC) in the three prior years. As discussed above, we hypothesize that recent M&As in the local market increase the probability of entry, and the data in Table 4 strongly support this hypothesis. In every year of the sample, 1) the overwhelming majority of de novo entries were in markets with recent M&A activity; 2) the frequency of entry per market was much higher in markets with M&A activity; and 3) entries per 100 markets were many times higher in M&A markets than in non-M&A markets. Moreover, the ratios of entry activity in M&A markets to non- M&A markets appear to be increasing over time. To illustrate, in 1998, 1) M&A markets had 188 entries versus 2 entries in non-M&A markets; 2) M&A markets had entry 5.77% of the time versus 0.33% of the

time in non-M&A markets, and 3) M&A markets had 8.68 entries per 100 markets versus 0.33 entries per

6 100 non-M&A markets. These results undoubtedly depend in part on market size, metropolitan verus rural location, profitability of the market, and other factors that are controlled for in our multivariate regressions. Nonetheless, the raw data are quite one-sided and fairly persuasive on their own.

Two other findings are apparent from Table 4. First, the proportion of markets with recent M&A activity has increased steadily over time, from less than 30% at the beginning of the period to over 75% at the end of the period. Second, there was substantially more entry in the U.S. banking industry in the 1980s than in the 1990s. There were over 200 entries in every year in the 1980s, with a high of 408 entries in 1984, There were also 3 other years in the 1980s (1982, 1983, and 1985) with over 300 entries. One reason for the high frequency of entry in the 1980s may be a liberalization of charter policy at that time. Beginning in 1980, the Office of the Comptroller of the Currency began to pay less attention to the community’s “capacity” to support an additional bank in granting national bank charters, and this policy of allowing market competition to determine entry and exit has continued since then (see White 1992, Office of the Comptroller of the Currency 1995, DeYoung and Hasan 1998).

In contrast with the 1980s, there were fewer than 200 entries in every year of the 1990s (through 1998), including a low of just 48 entries in 1994. However, there was a substantial upswing over the period 1995-1998, with a local maximum of 190 entrants in 1998. The percent of markets with entry and the number of entries per 100 market years are generally lower in the 1990s as well.?

In our empirical analysis below, we report separate regressions for the full 1980-1998 period and for the recent upswing subperiod of 1995-1998, allowing all the model parameters to differ. The full time period has the advantage over the recent subperiod of having more observations. In addition, the full period covers a broad range of macroeconomic, interest rate, profitability, and bank regulatory conditions, as opposed to the sustained favorable macroeconomic conditions, low interest rates, high bank profitability, and reduced

regulations of the recent subperiod. The tests of the effects of M&As on the probability of entry and other

“Interestingly, the frequency of entry over time does not seem to correspond as well as might be expected with aggregate banking industry profitability. Profitability was generally falling during the mid-1980s, while entries were rising to their sample peaks. As well, the recent run of high profitability began in the early 1990s, with record profits being reached in 1993 and 1994, the years with the fewest entries. However, entries have increased in the 1995-1998 time interval as would be expected, given that strong aggregate profitability and other favorable conditions for the banking industry have continued.

7 hypotheses over the full 1980-1998 time period may be viewed as the average effects over the long term and many environmental conditions, which may be more appropriate tests of the economic hypotheses. However, the tests over the 1995-1998 subperiod may be more revealing about the conditions for entry currently and in the near future.

Table 5 shows entry statistics classified by state banking regulations each year.’ Unit banking regulations generally allow only one full service office per bank. Banks in limited branching states can have multiple branches, but are restricted in some way (e.g., branches in only one county). Over our time period, State geographic regulations were substantially liberalized, moving some states from unit to limited or statewide branching and some states from limited branching status to statewide branching. By the end of 1991, the final unit banking regulations had been eliminated.

As discussed above, we hypothesize that the state restrictions on competition from existing firms outside the local market would create market power and increase de novo entry. In some cases, de novo entry might be newly chartered banks opened by bank holding companies (BHCs) with other banks in the same or nearby local markets as legal alternatives to branching where branching is prohibited. For example, several BHCs owned a number of individual banks in Texas in the 1980s, which was a unit banking state until 1987.

The data shown in Table 5 are only somewhat consistent with these predictions. During the 1980s, unit banking markets generally had higher percentages of entry and more entries per 100 markets than markets in limited branching states, but were not much different on average from markets in statewide branching states. Presumably, markets in statewide branching states had comparable statistics to unit banking states because statewide branching states tend to be those with relatively large markets (e.g., California). As noted above, nationwide there was more entry in every year of the 1980s than in every year of the 1990s, which also supports the favorable effect of unit banking restrictions on entry, given that there

were substantial numbers of unit banking markets only in the 1980s. However, there is no evidence in this

*To be consistent with our empirical analysis below, the state rules shown in Table 5 are those in effect as of December 31 of the prior year. This reduces endogeneity problems and assures that the explanatory variables are known to those making the entry decisions.

8 table to support any favorable effect on entry of limited branching restrictions — limited branching markets generally had less entry than statewide branching states — although this may also partially reflect differences in market sizes. We deal with this further below in the multivariate regression analyses.

Some additional examination of the markets with entry in the most years and the most entries in a single year (not shown in tables) gives some further indications as to the importance of market size and restrictions on competition. Perhaps surprisingly, no market — not even New York with its overwhelming size — had entry every year. New York had entry in 18 of the 19 years, as did Denver and St. Louis. Eight MSAs with entry in 15 or 16 out of 19 years are Chicago, Los Angeles, Dallas, Houston, Atlanta, Kansas City, Phoenix, and Seattle. For the most part, these markets can be identified with large deposit size (New York, Chicago, Los Angeles), fairly restrictive state branching laws over most of the time period (Denver, St. Louis, Dallas, Houston, Kansas City), or high market growth (Atlanta, Phoenix, Seattle).

The single market year with the most entries, 40, was not one of the largest in size, but was Houston in 1983. The remainder of the 9 market-years with 21+ entries were Houston in 1984 (32 entries) and 1985 (25), Dallas in 1983 (25), 1984 (32), and 1985 (22), Denver in 1982 (25), and Los Angeles in 1982 (21) and 1983 (23). Houston, Dallas, and Denver in the early to mid-1980s were all in unit banking states, and this presumably reflects the state restrictions in Texas and Colorado. Los Angeles is presumably represented because of its market size (third largest market), high market growth in the early 1980s, and other factors. Note that all of these high entry cases occurred in the 1982-1985 time period, whereas during the recent 1995-1998 period, the maximum number of entrants in any one market in one year was 9.

A final set of findings not shown in the tables concerns de novo entry by thrift institutions (savings and loans and savings banks). For the main part of our analysis, we include only commercial bank entry, since thrifts typically do not compete effectively in the commercial loan market. As discussed earlier, the main policy questions we address about de novo entry in banking concern the effects of M&As on small business lending and the extent to which any external effect of M&As on the small business lending of other local banks operates through de novo entry and the lending of recent entrants. Thrift institutions typically devote an even smaller percentage of their assets to small business loans than do the very largest commercial

banks, and so thrift entry generally is not a significant part of the external effect of M&As at issue here

9 (Kwan 1998). Nonetheless, we include thrift entry as a robustness check on our main results because many thrift products are substitutes for bank products and because the other study of the effects of M&As on entry included thrifts (Seelig and Critchfield 1999).

There were 743 thrift entries over our full time period of 1980-1998, about 20% as many as commercial bank entries. Many of the thrift entries occurred in the same markets in the same years as commercial bank entries, so the percent of market years with entry by commercial banks or thrifts was 4.65%, not much above the 4.21% of the time that one or more commercial banks entered. Importantly, the thrift entries do not alter the finding that entries occur much more often if there was bank M&A activity in the local market in the past 3 years. Over the 1995-1998 subperiod, M&A markets had 65 thrift entries versus | thrift entry in non-M&A markets. If anything, the thrift entry data are even more one-sided than the commercial bank entry data shown above in Table 4, strongly suggesting that M&A activity is associated with increased likelihood of entry.

3. Prior Evidence on Entry

In this section, we briefly review some of the relevant empirical literature on entry. We first discuss the evidence on entry in general, and then turn to evidence on entry in the banking industry. Section 4 reviews the empirical literature on the effects of bank M&As, which may help determine the probability of entry in the local banking markets.

The General Entry Literature

As noted above, theory predicts entry when there are strong prospects for future abnormal profits. The empirical literature has concentrated primarily on factors such as profitability, concentration, and growth to serve as proxies for expected future abnormal profits from entry into a market. The dynamic effects of M&As are generally not considered in the empirical literature, despite some good reasons to expect that consolidation may affect the supply of services and profitability of firms in the market.

The results in the empirical literature on entry do not provide as consistent a picture as does the theoretical literature. Siegfried and Evans (1994) summarized recent empirical work on entry across industries. They found that entry is positively related to industry profitability and to industry growth (see

also Yip 1982, Deutsche 1984, Acs and Audretsch 1990, Chappel, Kimenyi, and Mayer 1990, Rosenbaum

10 1993). Earlier empirical research by Siegfried and Evans (1992) found that entry rates and exit rates are positively correlated. They interpreted exit, however, as firms’ responses to declining prospects, as would occur when a firm is liquidated or when it is acquired by another firm after a string of losses. However, they do not address the possibility that M&As (as a form of exit) occur in good times as well as bad.

Another review of the entry literature by Geroski (1995) reported some additional findings and yielded somewhat different conclusions (see also Dunne, Roberts, and Samuelson 1988, Audretsch 1991, Baldwin and Gorecki 1991). Geroski (1995) found that de novo entry is common across industries, but entrants appear to have a short life span. The survival rate of entrants is low, and it takes more than a decade to achieve the size of the average incumbent when entrants do survive. Entry rates vary more within industries than across industries. Even though there is substantial entry, econometric estimates indicate that entry barriers are high. Entry reacts slowly to high profits, and it is difficult to estimate entry rates using measures of profitability and entry barriers.

Further, the ex post economic effects of entry do not follow theoretical expectations as closely as would be expected. Entry has only modest effects on average price-cost margins. Incumbent response to entry is selective. Pricing is not usually used to block entry, although marketing efforts are used for this purpose. Most of the literature has examined the short-run effects of entry rather than the long-term effects. Therefore, it is not too surprising to find that entry does not greatly affect profitability. Entry does appear to be highly correlated with high rates of innovation and increases in efficiency. Geroski (1995) concluded that active rivalry among existing firms in the market is preferable to entry since entry is often too slow, too small in scale, and too erratic to matter much. Although he welcomed the current antitrust emphasis on entry barriers as an important determinant of market structure, he argued that the pro-competitive effects of entry may be exaggerated.

There are distinct research advantages to examining regulated industries. The government mandates substantial data requirements of regulated firms. Also, many of these regulated markets are local, and the detailed data allow the researcher to segment the regulated industry into geographic markets, thus permitting more extensive analysis of the industry. Entry studies have been done on two regulated industries — airlines

and banking.

1

For the airline industry, Joskow, Werden, and Johnson (1994) found that entry and exit have important competitive effects, contrary to the findings of the interindustry studies. Joskow, Werden, and Johnson (1994) found that though entry is generally not induced by above-normal price levels on a city-pair basis, entry reduces fares and increases output while exit increases fares and reduces output. In addition, in response to entry, incumbents cut price and maintain output, while in response to exit, survivors increase both price and output.

The findings from the general (nonbanking) literature as a whole cast doubt upon the ability of entry to restrain anticompetitive behavior and are consistent with a continued need for antitrust restraint on M&As. Essentially, the findings suggest that entry frequently does not occur when entry is needed to curb the exercise of market power. Moreover, although entry may “work” in terms of lowering prices when it does occur, the effect may be short-lived because of the low survival rate.

The Bank Entry Literature

The most relevant entry evidence for our study comes from research examining the banking industry. More detailed data are available for banking than for any other industry, and the results of these studies are more in line with theoretical models of market behavior. We focus on three recent studies here.>

Amel and Liang (1997) estimated a two-equation model of entry and profits for 1977-88. Their results support the existence of a competitive process in banking markets. New entrants are attracted by high profits, market size, and market growth.° Entry directly reduces market prices, especially in rural markets, and thus prevents super-competitive prices from persisting over time.

Moore and Skelton (1998) examined the paradox of the recent increase in de novo bank entry, given the decrease in the total number of banking organizations at the same time. Most of the analysis covers the year 1997. They found that de novo banks organize to take advantage of local conditions favoring entry such as high growth and deficiencies in the provision of certain services such as small business lending. The

authors concluded that new banks have increased the level of competition in markets that needed more

°See also Rhoades (1997) for an extensive discussion of the barriers to entry in banking.

°Amel and Liang defined entry by either de novo charters by new banks or new branch offices by existing banks from other markets. In contrast, we focus exclusively on de novo charters.

12 competitive behavior. Though multivariate analysis is not used by Moore and Skelton, their results support the notions that entry reacts to market forces and that entry has a competitive impact.

Finally, Seelig and Critchfield (1999) examined the determinants of entry in banking in MSAs over 1995-1997, expanding on the early work of Hanweck (1971), Rose (1977), and Rose and Savage (1983), who found that profitability and concentration are related to market entry. Seelig and Critchfield examined the hypothesis that merger activity encourages new entry because of negative market reaction to the merged larger banks. Their results not only reject the hypothesis that merger activity encourages the formation of de novo banks, but found a significant negative relationship between mergers and de novo entry. Our own empirical analysis is closest to that of Seelig and Critchfield, although we come to the opposite conclusion about the effects of M&A activity on entry.

4. Prior Evidence on the Effects of Bank M&As

Although there is very little direct empirical evidence on the effects of bank M&As on entry, there is substantial evidence on other effects of M&As that may influence entry decisions. There are at least three ways that M&As can affect the probability of de novo entry in the same local market. First, M&As may affect the exercise of market power in setting prices in the local market. The probability of entry should vary directly with the change in market power. Second, M&As may change the efficiency of the M&A participants or other local competitors. The probability of entry should vary inversely with the change in efficiency. More efficient incumbents are stronger competitors who may be expected to stay in business longer and be able to compete more effectively in pricing in the future. Third, M&As may affect the supply of services to small customers in the local market for reasons discussed below. The probability of entry should vary directly with the degree to which institutions participating in M&As reduce their supplies of services to small customers.’

The literature contains a number of static analyses of the effects of M&As, which examine the effects of changes in bank characteristics that are associated with M&As, such as the effects of increased size or

market concentration. There are also a number of dynamic analyses, which compare the data for banks after

"For a detailed review of the consequences of bank M&As, see Berger, Demsetz, and Strahan (1999).

13 M&As with pre-M&A data or with data on other banks that have not recently engaged in M&As. Dynamic analyses take into account that M&As may involve long-term changes in organizational focus or managerial behavior and/or short-term costs or disruptions from consummating the M&A in addition to the static effects of changes in size, concentration, etc. Bank M&As and Market Power

M&As between banks that have significant local market overlap ex ante may increase local market concentration and allow the consolidated bank and other banks in the local market to raise profits by setting prices less favorable to customers. This may affect rates and fees on small business loans and small deposits, as these products are typically competed for on a local basis. M&As of the market-extension type that join local banks with banks from outside the local market are less likely to increase local market power. The static literature on the effects of local market concentration generally found that banks in more concentrated markets charge higher rates on small business loans and pay lower rates on small deposits (e.g., Berger and Hannan 1989, 1997, Hannan 1991) and that their deposit rates were “sticky” or slow to respond to changes in open-market interest rates (Hannan and Berger 1991, Neumark and Sharpe 1992, Hannan 1994, Jackson 1997), consistent with market power.

Dynamic studies of the effects of bank M&As on prices, which are based on prices actually charged before and after M&As, have mixed results as to the overall importance of M&As to the prices charged by M&A participants and their local market rivals (Akhavein, Berger, and Humphrey 1997, Prager and Hannan 1998, Sapienza 1998, Simons and Stavins 1998). However, one study that focused on in-market M&As with significant market overlap found strong effects on deposit prices (Prager and Hannan 1998). Thus, M&As — at least in-market M&As with substantial local market overlap — may boost the exercise of market power in pricing and increase the probability of entry.

There are reasons to expect that potential M&As that would result in large increases in concentration

and market power do not actually occur. First, antitrust authorities often block, alter, or deter M&As that

‘It is possible that the exercise of market power in banking has declined in recent years due to the removal of geographic restrictions, the standardization of products, and expanded methods of delivery of services (see Radecki 1998, Berger, Demsetz, and Strahan 1999).

14 are expected to result in substantial increases in market power. Second, a theoretical model has been developed that shows that such M&As may also be deterred by the prospects of entry, which would raise output and/or lower prices so as to eliminate most of the private gains to these M&As (Werden and Froeb 1998). Bank M&As and Efficiency

M&As may affect scale, scope, or product mix efficiency by moving the banks to a different output vector. M&As may also change X-efficiency if the consolidated bank alters its organizational focus or managerial behavior or suffers costs or disruptions from consummating the M&A. Banks may also improve their efficiency, broadly defined, by diversifying their portfolios and improving their risk-expected return tradeoffs.

Static studies of scale, scope, and product mix efficiencies using 1980s data generally suggest little or no cost efficiency improvement from the change in the output vector (e.g., Hunter and Timme 1986, Berger, Hanweck, and Humphrey 1987, Noulas, Ray, and Miller 1990), although recent research suggests that cost scale economies may have increased in the 1990s (e.g., Berger and Mester 1997). The size and geographic spread of interstate banking organizations also appears to have improved the risk-expected return tradeoffs for many of these banks, allowing them to increase lending and conserve on costly equity capital (e.g., Hughes, Lang, Mester, and Moon 1996, 1999, Demsetz and Strahan 1997). However, one study found that banks in markets with high concentration tend to have lower X-efficiency, all else held equal, possibly because managers of weakly controlled firms may take part of the benefits of the extra revenues from market power in pricing as reduced effort to maximize efficiency, or a “quiet life” (Berger and Hannan 1998). These results suggest that the effects of M&As on efficiency may depend in an important way upon whether the M&A is of the in-market type (which may increase market power and reduce efficiency), or of the marketextension type (which has little effect on market power, but may improve efficiency through diversification).

Dynamic studies of the X-efficiency effects of bank M&As generally have found very little or no cost efficiency improvement (e.g., Berger and Humphrey 1992, DeYoung 1997, Peristiani 1997, Rhoades 1998). However, studies of profit efficiency effects found that M&As improved profit efficiency, and that

this improvement could be linked to improved diversification of risks (Akhavein, Berger, and Humphrey

15 1997, Berger 1998). After consolidation, the banks tend to shift their asset portfolios from securities to loans, to have more assets and loans per dollar of equity, and to raise additional uninsured purchased funds at reduced rates, consistent with a more diversified loan portfolio. In effect, capital market participants gave consolidated banks a benefit from diversification by allowing them to increase their lending without the penalty of more equity or higher rates on uninsured funds.

There is little information on the efficiency of de novo entrants, which is relevant to whether the entrants can compete with the firms engaging in M&As. One study found that de novo banks begin less profit efficient than mature banks of similar size, make up most of the difference within three years, and become equally efficient by about age nine (De Young and Hasan 1998). It seems unlikely that the entrants can match the geographic diversification advantages of banks that have engaged in market-extension M&As.

Bank M&As and Small Business Lending

Bank M&As may result in a reduction in the supply of credit to some small business borrowers as the consolidated banks at least partially exit from certain product niches of the market. This may occur because the larger, more organizationally complex banks created by M&As may encounter Williamson (1967, 1988) type organizational diseconomies associated with providing certain services to small business customers while engaging in their main business of providing services tailored to their large business customers. Large complex banks may withdraw from providing relationship-based small business loans because of diseconomies associated with providing these loans alongside their core business of delivering transactions-based loans, derivatives, other off-balance sheet guarantees, etc., to large business customers. That is, it may be too costly to provide relationship-based services, which often demand intimate knowledge of the small business, its owner, and its local market, together with the capital market financial services to

large businesses in which large banks specialize.? The reduction in services to small, relationship-based

°A considerable amount of recent evidence supports the notion that banks use relationships to garner information about small businesses and that small businesses benefit from these relationships. U.S. small businesses with stronger banking relationships have been found to receive loans with lower rates and fewer collateral requirements, be less dependent on trade credit, enjoy greater credit availability, and have more protection against the interest rate cycle than other small businesses (e.g., Lang and Nakamura 1989, Petersen and Rajan 1994, 1995, Berger and Udell 1995, Blackwell and Winters 1997, Berlin and Mester 1998, Cole 1998, Hubbard, Kuttner, and Palia 1998). The data also suggest that banks gather valuable private information from depositors and in some cases use this information in credit decisions (Allen, Saunders, and

16

customers may also result in part from the increased investment opportunities afforded by larger bank size, including serving large, transactions-based customers. The improved alternatives may reduce investments in small business loans if the supply curve of funds to the bank is upward sloping, so that the marginal cost of funding small business loans increases with bank size. It is also possible that short-term disruptions caused by the consolidation process itself give other banks — either local incumbents or new entrants — the opportunity to “steal” customers who perceive a reduction in service quality or availability.

Supporting these arguments, static studies have found that larger banks devote lesser proportions of their assets to small business lending than smaller institutions (e.g., Berger, Kashyap, and Scalise 1995, Keeton 1995, Levonian and Soller 1995, Berger and Udell 1996, Peek and Rosengren 1996, Strahan and Weston 1996). As banks get larger, their proportions of assets devoted to small business lending (measured by domestic commercial and industrial loans to borrowers with bank credit less than $1 million) declines sharply from about 9% of assets for small banks (assets below $100 million) to less than 2% for very large banks (assets over $10 billion).!°

A number of dynamic studies examined the effects of U.S. bank M&As on small business lending (e.g., Keeton 1996, 1997, Peek and Rosengren 1996, 1998, Strahan and Weston 1996, 1998, Craig and Santos 1997, Kolari and Zardkoohi 1997a,b, Zardkoohi and Kolari 1997, Walraven 1997, Berger, Saunders, Scalise, and Udell 1998). The most common findings are that M&As in which one or more of the banking organizations is large tend to reduce small business lending, whereas M&As between small organizations tend to increase small business lending, although there are exceptions. Since M&As involving large organizations dominate M&As in terms of assets, these studies suggest an aggregate net reduction in small

business lending by the banks participating in M&As.

Udell 1991, Nakamura 1993, Mester, Nakamura, and Renault 1998).

'°Some evidence also suggests that it is specifically relationship-dependent small borrowers that tend to receive less credit from large banks. One study found that large banks tend to charge about | percentage point less on small businesses loans and require collateral about 25% less often than small banks, other things equal (Berger and Udell 1996). These data suggest that large banks tend to issue small business loans to higher-quality transactions-based credits, rather than relationship-based loans that tend to have higher interest rates and collateral requirements. Similarly, another study found that large banks tend to base their small business loan approval decisions more on financial ratios, whereas the existence of a prior relationship with the borrowing firm mattered more to decisions by small banks (Cole, Goldberg, and White 1999).

17

Some research also measured an external effect that captures the dynamic responses to M&As of other lenders in the same local markets (Berger, Saunders, Scalise, and Udell 1998). The external effect incorporates the possibility that some loans that are dropped by the consolidating institutions may represent valuable business opportunities for other competitors or potential entrants. Although the external effect was not precisely measured, it appeared that the small business lending of other banks in the same local market tended to offset much, if not all of the negative effects on small business lending of M&As.

What is not known is the extent to which this external effect was the result of established banks’ increasing their small business lending versus expansion by de novo entrants. As noted above, only one other study examined the effects of bank M&As on entry and found a negative effect, which suggests that entry may actually reduce the external effect of M&As. Nonetheless, some other evidence is consistent with the possibility of a strong external effect through entry. Several studies found that de novo banks tend to lend more to small businesses as a percentage of assets than other small banks of comparable size, and that this difference tends to persist for as long as 20 years after entry (DeYoung 1998, Goldberg and White 1998, DeYoung, Goldberg, and White 1999). One way that this might occur is that loan officers who leave the consolidated institution take some of their relationship-based loan portfolios with them and start a de novo bank. Unfortunately, none of these studies of lending by recent entrants included bank M&As as an explanatory variable, so it is difficult to determine whether this strong lending is part of an external effect of consolidation. As described in the next section, we try to extend the research on the external effects of M&As by investigating the effects of M&As on both the probability of entry and the small business lending of recent entrants.

5. Hypotheses and Methodology

In this section, we outline our main hypotheses to be tested and our empirical methodology for conducting the tests. The hypotheses concern factors that may affect the likelihood of de novo entry ina banking market and the small business lending of recent entrants and other small banks, and we develop empirical models of both entry and lending to test these hypotheses. For convenience, all of the variables used in the main entry regressions are described in detail in Table 6, along with the sample means and

standard deviations of these variables for the large metropolitan, small metropolitan, and rural markets.

18 Similarly, the variables employed in the main small business lending regressions are shown in Table 7, along with the means and standard deviations of these variables for different age groups of small banks (GTA < $100 million). In the interest of brevity, variables used only in the robustness checks are not shown in the tables. Hypotheses about Entry

The major hypotheses about the causes of entry concern the expected future profitability of a prospective entrant. We focus first on the expected future profit opportunities afforded by M&A activity in the local market. As indicated above, M&As may affect expected profitability from entry by 1) changing the exercise of market power in setting prices in the local market; 2) changing the efficiency of the M&A participants or other local competitors; and/or 3) altering the supply of services to small relationship-based customers in the local market, possibly creating external profitable opportunities to serve these customers.

Based on the evidence summarized above, the third effect of bank M&As on entry — the external effect from a possible reduction in the supply of services to small customers by banks engaging in M&As — is likely to be the strongest. Potential M&As that would result in large increases in market power likely do not occur often because of antitrust authorities and/or fear of entry or responses of other competitors. The only pronounced efficiency gains from bank M&As appear to be gains from increases in diversification, which would typically be expected to be strong only for market-extension type M&As. In contrast, the effect of M&As involving large banks in reducing small business lending is fairly well established, and this is the type of lending that is the niche of de novo banks for a number of years after entry. However, there is very little evidence directly connecting the external effect to de novo entry or to the lending of recent entrants.

There are a number of other influences on potential entrants' expectations of future profitability and hence on entry. Because those profit expectations are known only by the potential entrants, a number of proxies must be used. As indicated above, the literature often focuses on the recent profitability of incumbent firms, recent market growth, and the level of market concentration. Generally, profits, growth, and concentration are expected to have positive effects on the probability of entry, all else held equal, although concentration is sometimes thought to have a negative effect on the probability of entry because

high concentration may be an indicator of high barriers to entry that are difficult to measure. We include

19 these variables in our empirical model of the probability of entry. We hypothesize that the effects of profitability and growth will be positive and that the effects of concentration will depend on whether it is more of a proxy for high profit opportunities or high entry barriers.*

We also include two variables which are similar in spirit to market growth. The market’s ratio of purchased funds to total assets is included as an indicator of lending opportunities that exceed locally generated deposits. Similarly, the growth of income in the state is an indicator of future profitable opportunities for new banks, and is hypothesized to have a positive effect on the likelihood of de novo entry.

The entry literature also often includes various measures of barriers to entry. As discussed above, the state banking regulations such as unit banking rules act as barriers to entry by incumbent firms from other markets, but are not barriers to de novo entry. As shown in Table 6, we include a number of variables measuring these restrictions, including indicators of unit banking restrictions, limited branching restrictions, and recent changes in these restrictions. We hypothesize that legal restrictions on competition by existing banks increase the probability of de novo entry by creating local market power and limiting the ability of incumbent banks to enter. We also include a variable for whether BHCs from out of state may enter the market. However, the expected sign of the effect of this variable is ambiguous, since state laws differ and some of them affect both barriers to acquisitions by existing institutions and barriers to de novo entry.

We also include specific average market prices for major outputs (consumer loans, non-real estate business loans, real estate loans, and securities) and inputs (purchased funds, core deposits, and labor). We generally hypothesize that the probability of entry should be increasing in the output prices and decreasing in the input prices, all else held equal, as high output prices and low input prices are associated with profit making opportunities. However, because all but one of these prices are interest rates received by or paid by a bank, another plausible hypothesis is that the prices (other than labor) may reflect differences in risk across banking markets. High average market rates charged on loans could reflect high default premiums for lending to local customers and high rates paid by banks could reflect high relatively high bankruptcy probabilities for local banks. If these risks tend to deter entry, this could reverse the hypothesized si gns on the outputs (e.g., high loan rates would reduce the probability of entry), but would reinforce the predicted

signs on the funding inputs (e.g., high purchased funds rates would reduce the probability of entry both

20 because of low market power and high bankruptcy risk).

We also include a number of other factors shown in Table 6. We hypothesize that greater market shares of large or organizationally complex banks increase the probability of entry, as large or complex banks may be competitively disadvantaged in serving the small relationship customer niche in which de novo banks tend to specialize. We include variables measuring the efficiency and financial condition of market banks, given that more efficient or financially sound incumbents should make stronger competitors, which is likely to deter entry. However, it is also possible that better performing incumbents in a market is a proxy for favorable market conditions, which may encourage entry. Finally, as discussed earlier, we control for the size of the banking market and whether it is metropolitan or rural. As shown in Table 6, we group our explanatory variables into five broad categories: Market M&As; Other Competitive Conditions; Market Demand Conditions; Market Prices; and Condition of Market Banks.

Hypotheses about Small Business Lending

For our hypotheses about small business lending, we again focus on the expected effects of M&As, but also include a number of other hypotheses suggested by the literature. As discussed above, the theory and empirical evidence suggests that in most cases, bank M&As are likely to reduce relationship-based small business lending by the participants, which may create an external effect in which other existing banks or new entrants may have profitable opportunities to increase their lending by picking up some small business customers. In our empirical analysis, we examine the effects of M&As in a local market on small business lending by recent entrants and by other small banks of different ages. This allows a closer look at the external effect and the hypothesis that M&As increase small business lending by small banks as a whole. This also allows us to test the hypothesis that recent entrants react more to market M&As than other small banks, which may be the case if the purpose of entry was to react to lending opportunities created by M&As.

As discussed above, several studies found that de novo banks tend to lend more to small businesses as a percentage of assets than other small banks for as long as 20 years after entry (DeYoung 1998, Goldberg and White 1998, DeYoung, Goldberg, and White 1999). The raw data here support this negative effect of bank age on lending for small banks. As shown in Table 7, small banks that are young (ages 1-5 years)

devote 17.9% of their assets to commercial and industrial (C&I) loans on average, compared to 14.0% for

21

adolescent small banks (ages 6-20 years), and 8.0% for mature small banks (21+ years). This is similar to the findings of the earlier studies. Some hypotheses that may explain this fact have not previously been explored. It may be the case that the external effect of M&As on small business lending is particularly strong for recent entrants, especially if the entry itself was part of the external effect of the M&As. Alternatively, recent entrants may behave about the same as other local small banks in terms of their lending reactions to M&As, but de novo entrants may be more highly represented in markets with recent M&As because M&As raise the probability of entry (consistent with our hypotheses about entry above). Our empirical analysis allows us to replicate the earlier research and test the hypothesis that recent entrants lend more than other banks of comparable size with a more comprehensive data set and more control variables than earlier research. We can also distinguish among the alternative hypotheses to explain this result by comparing the reactions to recent M&As of banks of different ages.

We also study the effects of the bank’s own M&As differentiated by the age of the bank. This allows a first look at the effects of M&As in which recent entrants are participants. To date, there has been no analysis of which we are aware of the effects of M&As on the lending of participating banks of different ages. Thus, we can test the hypothesis for the first time that M&As involving recent entrants affect their lending differently from the effects of M&As on the lending of more mature small banks.

Most of our other hypotheses about the effects of market conditions on small business lending are essentially the same as our hypotheses described about the conditions that affect the likelihood of de novo entry. That is, factors that proxy for high profitability from entering a market generally are also associated with profitable lending opportunities and so should be associated with more small business lending by market participants. For example, we hypothesize that high values of market profits, growth, and other measures of market demand are likely to be associated with more small business lending, all else equal. In our empirical analysis, we include modified versions of all of the explanatory variables from the entry equations to allow all of these market conditions to affect lending.

We also test a number of other hypotheses related to the condition of the bank itself. The most important of these concern the age of the bank, given the previous empirical results and hypotheses about

younger banks having more small business loans than other banks of comparable size. Based on the prior

22 research, we also test the hypotheses that banks that are larger or more organizationally complex devote smaller proportions of their assets to small business loans and that banks in worse financial condition (lower profits, lower equity capital, more nonperforming loans) also make fewer of these loans. We also include the bank’s average market share (as well as the bank’s average market concentration) to control for the effects of market power, which could result in either less or more small business lending.'' As shown in Table 7, we group our explanatory variables into ten broad categories. The first five replicate the market variables used in the entry equations and described in Table 6, except that they are weighted averages over all the markets in which the bank operates. The last five categories are Bank’s Age, Bank’s Size; Bank's Ownership Structure; Bank’s Financial Condition; and Bank’s Market Share. Methodology We first model the probability of entry based on M&As in the market and other market variables hypothesized to affect the prospects for profitable entry. We estimate the model a number of different ways for several subsets of the data based on sample period, market size, and market type, but the main model is a simple logit equation of the probability of entry in market m in year t, which takes the form: In(P(ENTRY),,/(1-P(ENTRY),..)) = f (MARKET M&A, 44913, OTHER COMPETITIVE

CONDITIONS,,,.., MARKET DEMAND CONDITIONS,,,.,, (1)

MARKET PRICES,, ..,, CONDITION OF MARKET

BANKS at) + Emi The data are annual, so that entry is measured as having occurred or not occurred during year t in market m. All of the right-hand-side variables are measured as of the end of year t-1 or earlier to reduce endogenous feedback effects. For expositional clarity, most of these explanatory variables shown in Table 6 have the prefix “MKT”, indicating that these are calculated for the market. In most cases, these are averages over the banks which have deposits in the market weighted by each bank’s share of these deposits. Other right-hand-

side variables, such as the banking regulations, are for the state in which the market is located.

"High market power may be associated with less small business lending because of higher prices charged for these loans. However, market power could alternatively result in more relationship lending by allowing the bank to enforce long-term implicit contracts in which the borrower receives a subsidized interest rate in the short term, and then compensates the bank by paying a higher-than-competitive rate in a later period (Sharpe 1990, Petersen and Rajan 1995).

23

Because the prior research suggests that different types of M&As appear likely to have different market effects, we distinguish among the effects of different types of consolidation. As shown in Table 6, the MARKET M&A variables included in our main model measure the proportion of market deposits involved in recent bank mergers (MKT-MERGB), in which two or more bank charters are consolidated, and the proportion involved in BHC acquisitions (MKT-ACQUIS), in which the banks retain their separate charters but their BHC ownership is changed.'* In our robustness checks, we also distinguish between inmarket and market-extension mergers, mergers in which the pro forma consolidated bank is small versus large, and mergers between holding company affiliates versus mergers of previously unaffiliated banks.

We include 3 past years of M&As because prior analyses of bank M&As and reports by bank consultants suggest that at least 3 years are needed as a gestation period to complete any restructuring or refocusing of a bank after an M&A (e.g., Cornett and Tehranian 1992, Toevs 1992, Berger, Saunders, Scalise, and Udell 1998). The external effect on other banks in the local market and on new entry is likely to take at least as long as the effects on the institutions involved in the M&As, implying that a minimum of 3 years is necessary for the analysis. As well, the press reports that it often takes one to three years after an M&A for the departing loan officers to start a rival de novo bank (e.g., Epstein 1996). The main regressions we report specify the average over years t-1, t-2, and t-3 of the proportions of local market deposits in the offices of banks that were involved in mergers or acquisitions. As shown below, the results are robust with respect to an alternative specification in which the three individual annual lags are included.

In the interest of brevity, we will not discuss the other exogenous variables in the equation further here. The rationales for including these variables were given in the “Hypotheses about Entry” subsection above, and the individual descriptions and summary statistics on these variables are given in Table 6.

We next model the small business lending of de novo banks after entry based on the market variables relevant to the bank plus variables measuring the bank’s own age, M&A experience, size, organizational

complexity, financial condition, and market share. The dependent variable is in the form of a log-odds ratio

"Note that our use of the proportions of market deposits in the offices of banks involved in M&A activity differs from the specification of Seelig and Critchfield (1999), who use unweighted numbers of M&As of banks in the market without respect to the market shares of banks involved.

24 of the proportion of assets devoted to commercial and industrial (C&I) lending. In effect, we are estimating the probability that a dollar of assets is devoted to small business lending, rather than other investments. For robustness, we also run the model with the log of the C&I lending ratio as the dependent variable and the results are materially unchanged. To be reasonably certain that the loans are to small businesses, we follow past research and include in these regressions only banks with gross total assets (GTA) below $100 million, as these banks typically are constrained by legal lending limits and problems of diversification from making large loans. The regression equation for the lending of bank i in market m in year t is of the form:!3 In(C&V/GTA),; ./U(C& UGTA), 9.) = g (MARKET M&A, 4.1 ¢2¢3» OTHER COMPETITIVE CONDITIONS, ,,..;, MARKET DEMAND CONDITIONS , ,,...; MARKET PRICES, ,, ,.,, CONDITION OF MARKET BANKS, a1 (2) BANK AGE;,.,, BANK M&A, ,. 4.543, BANK SIZE, ,,, BANK ORGANIZATIONAL COMPLEXITY,,,, BANK CONDITION, .,, BANK MARKET SHARE,, ,) + 7;,- Equation (2) is estimated by weighted least squares in order to avoid heteroskedasticity problems and the adjusted R’s are corrected.'*

The dependent variable is based on the allocation of assets at the end of year t and all of the righthand-side variables are measured as of the end of year t-1 or earlier to reduce endogenous feedback effects. As discussed above, we include modified versions of all of the right-hand-side variables from the entry equations, including market M&As, since variables that proxy for high profitability from entering a market also proxy for profitable lending opportunities and should be associated with more small business lending in the market. Since the lending equations refers to a bank rather than a market, we use weighted averages of the market variables in the markets m in which bank i has deposits, where the weights are the proportions

of the bank’s deposits that are in each market. The means of the weighted average market variables as shown

"In the few cases where C&I/GTA were zero or close to it, we substituted the value .0001 for C&I/GTA to avoid having the dependent variable be at or close to --. There were no observations of C&V/GTA equal to or near 1, so no adjustments were necessary on the upper end of the distribution.

“Each observation is divided by a number proportional to the estimated standard error of its error term [{C/C&YGTA), 4.) + [I/I-C&VGTA), 1} / GTA,”

25 in Table 7, which bear the prefix “WAM” differ from the corresponding variables with the prefix “MKT” shown in Table 6 because each observation of the variables summarized in Table 7 is for a bank, rather than for a market. The explanatory variables for the individual bank’s age, M&A history, size, ownership structure, financial condition, and market share begin with the prefix “BANK,” and are based on data for bank i as a whole.

Equation (2) is run two different ways. First, we include small banks (GTA < $100 million) of all ages, except for de novo entrants in their first year, since these banks do not have lagged values for the “BANK” variables. In this regression, we control for bank age with both a continuous variable (BANK- LNAGE) and with dummies for young banks (ages 1-5 years), adolescent banks (ages 6-20 years), and mature banks (21+ years). In the second form of equation (2), we run separate regressions for young, adolescent, and mature small banks. This allows us to compare the reactions to recent M&As of banks in different age groups and test whether the prior research results that young and adolescent banks (i.e., banks up to 20 years after entry) lend more than mature small banks might be explained by young and adolescent banks having larger reactions to market M&As.

6. Probability of Entry Results

Tables 8a and 8b show the results of the simple logit model of the probability of entry in equation (1). Table 8a applies the model to large metropolitan markets, small metropolitan markets, and rural markets for the full 1980-1998 period, and Table 8b repeats the exercise for the 1995-1998 subperiod.’* The full time period has the advantage of more observations and coverage of a broad range of macroeconomic, interest rate, and bank regulatory conditions, whereas the more recent subperiod may be closer to the conditions for entry currently and in the near future. As discussed above, the effects over the full period may be viewed as the average effects over the long term and multiple environmental conditions, which may be more appropriate tests of the economic hypotheses.

The regressions shown in these tables exclude a small percentage of the observations, mostly because

of missing or unreliable values of some of the variables. Importantly, we also eliminate large or unusual

'°A small percentage of the observations, mostly from rural markets, are missing from the regressions due to missing or unreliable values of some of the variables.

26 types of de novo entries that could otherwise affect the results. We eliminate all entrants with GTA exceeding $1 billion by the end of the year, which may not fit our profile of small operators that may enter to provide services to small business customers. We also exclude commercial banks with bank types coded as subject to special analysis, such as bankers’ banks, credit card banks with commercial bank charters, depository trust companies, and bridge entities. The means and standard deviations in Table 6 also reflect these exclusions. The Effects of M&As on Entry

The data are strongly consistent with the hypothesis that M&As are associated with subsequent increases in the probability of entry into the markets in which the M&As occur. For the full 1980-1998 data set shown in Table 8a, the coefficients on the proportions of market deposits involved in recent bank mergers (MKT-MERGE) and in BHC acquisitions (MKT-ACQUIS) are positive and statistically significant at the 1% level in all three types of markets (large metropolitan, small metropolitan, and rural). For the recent subperiod 1995-1998 shown in Table 8b, the coefficients of MKT-MERGE and MKT-ACQUIS remain positive in all cases, but the degree of statistical significance is less. Four of the six parameters are significant at the 5% level, one is significant at the 10% level, and one coefficient (MKT-ACQUIS in the small metropolitan markets) is not statistically significant.

The drop in significance from the full sample to the 1995-1998 subsample is not surprising, given that most of the observations are excluded from the estimation. In addition, there is less variance in most of the economic conditions specified in the exogenous variable list during the subperiod, which makes accurate estimation more difficult.

The results of these logit equations also suggest that M&As have an economically significant impact as well as a statistically significant impact on the probability of de novo entry. One way to summarize the economic impact is to evaluate the predicted increase in entry probability from doubling the proportions of market deposits involved in recent M&As from their sample mean values. of the data used in the logit equations. Doubling the mean values for the full 1980-1998 data set of .181 for MKT-MERGE and .035 for MKT-ACQUIS for the large metropolitan markets and multiplying by the corresponding coefficients from

the first column of Table 8a of 1.6196 and 5.1520, respectively, implies an increase in the predicted

27

probability of de novo entry from its mean of 51.7% to 63.2%.'° This is an increase of 11.5 percentage points or 22.2%.'’ An alternative summary statistic for the effects of M&As is the decrease in predicted entry probability from the sample mean from eliminating M&A activity. Again using the sample means and parameter estimates for the large MSAs implies a decrease in the predicted probability of entry from its mean of 51.7% to 40.0%, a decline of 11.7 percentage points or 22.6%.'* Put another way, M&As are estimated to be responsible for over one-fifth of observed de novo entry in large metropolitan markets. For small metropolitan markets, doubling the mean values of the M&A variables increases the predicted entry probability from its mean of 15.0% to 19.5%, an increase of 4.5 percentage points or 30.0%. Eliminating M&A activity in small metropolitan markets reduces the mean predicted probability from 15.0% to 11.4%, a decline of 3.6 percentage points or 24.0%. For rural markets, doubling M&As increases predicted entry probability from 1.8% to 2.0% (an 11.1% increase), and eliminating M&As reduces the predicted probability to 1.6% (an 11.1% decrease). These results suggest that at least part of the external effect of M&As found in prior research represents increased de novo entry, i.e., new lending by banks that previously did not exist.

Notably, the estimated parameters for the M&A variables in the logit equations using the 1995-1998 data subsample shown in Table 8b are generally larger than those for the full data set, and there is more M&A activity in this subperiod. Thus, the data from the shorter time period actually imply greater estimated economic significance, although with reduced statistical significance. Overall, the logit regressions suggest a strong, statistically significant and economically significant effect of M&As on entry, especially in metropolitan markets, where M&As are estimated to explain more than 20% of all de novo entry.

Robustness of the Findings on the Effects of M&As on Entry

To check robustness and assure confidence in our main findings about the effects of M&As on the

'°We base the economic significance evaluations on the sample mean value of the data used in the logit equations of 51.7%, as shown in Table 6, rather than the slightly different mean of 51.32% shown earlier. As noted above, the data used in the logit equations exclude some observations with missing or unreliable values of some of the variables and large or unusual types of de novo entries.

"Beginning at the mean probability of entry of .517, the effect of increasing MKT-MERGE by .181 and MKT-ACQUIS by .035 is calculated as raising the probability of entry to P, in the formula: In(.517/(1-.517)) + (1.6196 © .181) + (5.1520 © .035) = In(P,/(1-P,)).

'’The formula for this probability is: In(.517/(1-.517)) - (1.6196 © .181) - (5.1520 « .035) = In(P,/(1-P,)).

28

likelihood of entry, we also run three series of robustness checks. The first series keeps the exogenous variable specification and data samples the same as in the main runs, but uses alternative econometric methods besides the logit model of entry versus no entry in equation (1). The second series uses the same econometric specifications and data samples as the main runs, but alters the specifications of the exogenous variables in various ways. The third series uses the main econometric specification and exogenous variables, but alters the data samples. To save space, the robustness results are summarized here, but are not shown in any tables.

The first alternative econometric method is a linear probability model of entry versus no entry. This yields similar results to the main model. For the full 1980-1998 sample, all six coefficients on the M&A variables (MKT-MERGE and MKT-ACQUIS) are positive, and 5 of them are statistically significant at the 1% level (MKT-MERGE is not significant in the rural markets regression). For the 1995-1998 subperiod, all six coefficients are again positive and five are significant, although the significance level is often lower, similar to the main results shown in Table 8b.

A Tobit model of entry and number of entries similarly supports the robustness of the main results. For the full sample, all six M&A coefficients are positive and statistically significant at the 10% level or better. For the 1995-1998 subperiod, five of the six parameters are positive and three of them significant at the 1% level. The one negative coefficient for this subperiod (MKT-MERGE in large metropolitan markets) is small and not statistically significant. Even in this case, the combined economic significance effect of doubling or eliminating all M&As still nets out to a strong positive effect of consolidation activity on the probability of entry.

We also run OLS on the number of entries two ways, once using the level (i.e., 0, 1, 2, ...) and once using the natural log of one plus the level (i-e., Int, In2, In3, ...). In both cases, all M&A coefficients are positive for both the full sample and the 1995-1998 subperiod. For the full sample regressions, five of the six parameters are statistically significant, while three of six are significant for the recent subperiod.

We also run some ordered logit equations. These are similar to the main logit model of entry versus no entry, except that we use multiple categories of positive entry and force the slope coefficients to be

identical. Using the information on multiple entries in Tables 1-3, for the rural and small metropolitan

29 markets, we specify 0, 1, and 2+ entries as the outcome categories and for the large metropolitan markets, we specify 0, 1, 2, 3, 4+ and alternatively, 0, 1, 2, 3, 4, 5, 6-10, 11-20, 21+ entries. The results are very similar to the main logit results in Tables 8a and 8b — all M&A coefficients are positive, all are Statistically significant at the 1% level in the full 1980-1998 period, five of six are significant at some level in the 1995- 1998 subperiod.

Turning next to our robustness checks that use alternative specifications of the exogenous variables, we first try distinguishing among more types of M&As. As noted above, we try distinguishing between inmarket versus market-extension mergers, mergers in which the pro forma consolidated bank is small versus large, and mergers between holding company affiliates versus mergers of previously unaffiliated banks. To do so, we simply add variables to our main specification measuring the average over years t-1, t-2, and t-3 of the proportions of market deposits in the offices of banks that were involved in 1) in-market mergers, 2) mergers in which the pro forma bank had GTA < $100 million, and 3) mergers of banks in the same BHC. The original M&A variables MKT-MERGE and MKT-ACQUIS are also included, so the coefficients of these additional variables measure potential additional effects. The results are consistent with those in the main regressions — all six coefficients ofp MKT-MERGE and MKT-ACQUIS are positive and Statistically significant for the full sample, all six are positive and five are significant for the 1995-1998 subsample. Only one of eighteen coefficients of the other M&A variables were statistically significant, suggesting that the addition of these variables does not alter our main results.

Next we try specifying the individual t-1, t-2, and t-3 lags of the main M&A variables instead of the three year averages (the additional M&A variables are not included). For the full sample, the coefficients of all eighteen lagged M&A variables are positive, and nine of them are statistically significant. For the recent subsample, fifteen of eighteen coefficients are positive, with seven of them significant. The three negatives are small and insignificant, and the averages of the lags of every M&A variable in every regression iS positive.

We also rerun the main logit model adding in second-order terms on the continuous exogenous variables to allow for nonlinearities (e.g., % MKT-HERF’). The result is no material change in the main

results. The coefficients of all six M&A variables are positive and statistically significant for both the full

30 period and the 1995-1998 subperiod.

We also try rerunning the main model adding in time dummies for every year but one to allow for the possibility that our main results are driven by other factors that vary in an important way over time. Nonetheless, the main results are not materially changed. For the full sample, all six coefficients on the M&A variables are positive and statistically significant, and for the 1995-1998 subperiod, all six are also positive and five of the six are statistically significant.

We finally turn to our robustness checks that alter the data samples from our main specifications. First, we include thrift entry as well as commercial bank entry in the dependent variable, similar to Seelig and Critchfield (1999). As in the main results, we find that for the full sample, all six coefficients of the M&A variables are positive and statistically significant at the 1% level and for the 1995-1998 subsample, all six are positive and five of the six are statistically significant at some level. The coefficients are also of the same order of magnitude as in the main results.

We also try rerunning the main model using an alternative definitions of large, metropolitan markets that excludes the very largest markets with deposits over $50 billion, which may be quantitatively different from the banks with $5 billion to $50 billion.'? No material difference is found — the coefficients of the M&A variables remain positive and statistically significant.

In sum, our statistically and economically significant positive relationship between M&As and the probability of de novo entry is quite robust. Three categories of robustness checks — alternative econometric methods, alternative specifications of the exogenous variables, and alternative data samples — all support our main results.

The Effects of Variables other than M&As on Entry

The estimated coefficients on the variables other than M&As in the probability-of-entry regressions in Tables 8a and 8b are mostly, but not always, consistent with the hypotheses given above. About half of the coefficients of these variables are statistically significant in the full 1980-1998 sample, but only about

one-quarter are significant in the 1995-1998 subperiod, likely due to fewer observations and less variance

"The markets with over $50 billion are the New York, Chicago, and Los Angeles MSAs in every year and the San Francisco and Philadelphia MSAs in various years.

31 in the exogenous variables.

We hypothesized above that market profits, market growth in deposits, and related variables like the purchased funds ratio and the growth of state income should have positive effects on expected future profitability and the likelihood of de novo entry. The coefficients of the profitability variable MKT-ROE in Tables 8a and 8b are mixed — positive for the metropolitan markets, but negative for the rural markets — giving no clear effect of profitability on the likelihood of entry, once M&As and all the other factors are controlled for in the regressions. The coefficients of MKT-GROW are mostly negative and not statistically significant, but the coefficients of MKT-PFRAT and STGROW are all positive and are mostly Statistically significant, giving support for the importance of some of the factors that are similar to growth, but not supporting the growth of the market itself as predicting more entry after controlling for these other factors.

We also predicted that concentration may have a positive effect on the probability of entry if it proxies mostly for profit opportunities from market power and a negative impact if it proxies mostly for high entry barriers. The coefficients of MKT-HERF are of mixed signs and are not statistically significant, except for a significant positive effect in rural markets for the full sample. Thus, neither effect appears to dominate overall.

We hypothesized that state banking regulations that create barriers to entry by incumbent firms from other markets may increase the probability of entry. The coefficients of UNITB are positive, statistically significant, and larger than the LIMITB coefficients for the full time period, consistent with the hypothesis (UNITB is excluded from the 1995-1998 subperiod, see notes to Table 8b). The coefficients of LIMITB are positive and significant for rural markets in both time periods, also consistent with the hypothesis. However, the LIMITB coefficients have mixed signs and are not significant for the metropolitan markets, presumably because limited branching laws typically allow unlimited branching within counties and mostly keep large banks based in metropolitan counties from branching elsewhere. States with newly liberalized branching rules (NEWLIB) have positive coefficients, suggesting that part or all of the effect of moving from unit or limited to statewide branching status or from limited to statewide branching is offset in the initial year of liberalization. Interstate BHC access (INTST), which has an ambiguous predicted effect as discussed above,

has mixed coefficients and one significant negative coefficient for the full sample (excluded or unreliable

32 for the 1995-1998, see Table 8b notes).

According to our hypotheses, the probability of entry may be positively or negatively related to the output prices, depending on whether high rates proxy more for profitable opportunities (positive) versus high default risks (negative). The results are somewhat mixed, with the prices of consumer loans (P1) and nonreal estate business loans (P2) generally being positively related to the probability of entry and the price of real estate loans (P3) being negatively and statistically significantly related to the probability of entry for the full sample period. These results suggest that while market power in pricing loans may encourage entry in some cases, a high price of real estate loans more likely signals high default probabilities in the market that may deter future bank entry. The input prices (W1, W2, W3) generally have either positive or insignificant coefficients, rather than the negative hypothesized signs.

We hypothesized that greater market shares of large or organizationally complex banks increase the probability of entry, but the data are only half consistent with this. The regressions in Tables 8a and 8b suggest that a greater presence of large banks discourages entry, contrary to expectations, while a greater presence of complexly organized banks (owned by out-of-state or multi-layered BHCs) encourages entry, consistent with expectations.

We hypothesized that more efficient or financially sound incumbents may make stronger competitors and deter entry. Again, the data are mixed. High equity capital (MKT-EQRAT) is associated with less entry in small metropolitan and rural markets as predicted. However, the coefficients of efficiency (MKT-EFFIC) are usually positive and not significant and the coefficients of nonperforming loans (MKT-NPL) are generally negative and significant in the later time period, contrary to the prediction. As discussed above, it is also possible that better performing incumbents in a market is a proxy for favorable market conditions which encourage entry, which may explain these last results.

Finally, we hypothesized that larger markets and metropolitan markets had more entry. Our main controls for these are in setting up the separate regressions for the large metropolitan, small metropolitan, and rural data subsets. Within these regressions, the log of market bank deposits (MKT-LNDEP) always has a positive, statistically significant coefficient, consistent with expectations. The additional size class

dummies have negative, insignificant coefficients, suggesting that our combination of dividing the data into

33

subsets plus including MKT-LNDEP was sufficient to account for most of the important effects of market size on the probability of entry. Overall, the logit regression results shown in Tables 8a and 8b are strongly consistent with positive effects of M&As and market size, but most of the other hypotheses receive only mixed support. 7. Small Business Lending Results

Tables 9a and 9b show the results of the small business lending model in equation (2) for the full 1980-1998 period and the 1995-1998 subperiod, respectively. As a reminder, our small business lending analysis includes only small banks (GTA < $100 million) to be reasonably certain that the loans are to small businesses. The first columns of each table show regressions which include all small banks ages | and over (banks younger than 1 year do not have the required lagged values of the BANK variables and are excluded from the lending analysis). The different ages of banks are accounted for by the inclusion of the BANK- LNAGE, BANK-YOUNG, and BANK-ADOLES variables. These first regressions test for the effects of M&As, bank age, and other variables on smal] business lending, assuming that the effects of M&As and other variables are the same for banks of all ages. The remaining columns show regressions which allow the effects of M&As and all of the other slope coefficients to differ by age groups. We run separate regressions for young small banks (ages 1-5 years), adolescent small banks (ages 6-20 years), and mature small banks (21+ years) to test whether banks of different ages respond to M&As and other factors differently.”

The Effects of Market M&As on Small Business Lending

In contrast to the probability-of-entry results, the small business lending results do not consistently support an external effect of M&As that operates through any increased lending of recent de novo entrants. For the first regression with banks of all ages shown in Table 9a, the coefficient on WAM-MERGE is positive and statistically significant. However, this appears to reflect primarily a positive external effect of mergers on the small business lending of mature small banks, which constitute the majority of the data

sample. In the separate equations by age group, the coefficient of WAM-MERGE is positive and significant

*°We do not run the lending regressions separately along metropolitan versus rural market lines or on the basis of market size, since individual banks typically lend in multiple types of markets, and we do not have separate data on the location of their loans. However, as shown below, we run a robustness check based on whether the bank is headquartered in an MSA or non-MSA county.

34

for mature small banks, but is negative and significant for young small banks, and negative and not significant for adolescent small banks. The coefficients of WAM-ACQUIS are negative in all cases and statistically significant except in the young small banks regression, contrary to expectations. However, the economic significance of these effects is relatively small. For example, doubling the proportions of market deposits involved in recent M&As from their sample mean values decreases the predicted proportion of small business lending by young small banks by about 2.5%, and eliminating M&A activity implies a predicted increase in the C&I lending proportion by these banks of about 2.5%.

Thus, the full sample results for the effects of market M&As do not support the hypothesis that there is a positive effect of consolidation on the small business lending of small banks in the market and in particular, do not support a positive effect for the most recent entrants. These data also suggest that exposure to market M&As does not explain the observed greater small business lending of recent entrants, although we present more direct evidence on this point below.

Interestingly, the results appear to turn around somewhat for the recent 1995-1998 subperiod. As shown in Table 9b, the coefficients of WAM-MERGE and WAM-ACQUIS are positive all eight cases, and are statistically significant positive in four of these cases. It may be the case that the hypothesized external effect has appeared more in recent years or that some other factor about the recent economic boom is related to M&As and small business lending. Nonetheless, the estimated economic magnitudes again suggest that the impact of M&As on small business lending by recent entrants is likely to be relatively small, on the order of about a 4% change in small business lending by young small banks from doubling or eliminating market M&A activity.

The Effects of Bank Age on Small Business Lending

We hypothesized above that market M&As may be responsible in whole or in part for the observed higher small business lending ratios for young and adolescent small banks versus mature small banks. This may occur because 1) recent entrants may have stronger external reactions to M&As or because 2) recent entrants are more often in markets with high M&A activity. The results already reviewed suggest that young small banks do not have an estimated positive external effect in the full sample and only a small, statistically

insignificant effect in the 1995-1998 subsample, so the first potential explanation is not supported by the

35 data. Here we look at the measured effects of bank age on small business lending after controlling for M&As in the regression equations shown in Tables 9a and 9b. If either or both potential explanations are correct and M&As tend to explain the higher lending of young and adolescent small banks, then there should be little or no effects of the age variables BANK-LNAGE, BANK-YOUNG, and BANK-ADOLES remaining in these regressions that control for M&As.

The results shown in the first regression in Table 9a strongly suggest that as banks age, their small business lending ratio decreases even after controlling for the effects of market M&As and all of the other market- and bank-specific variables in the equation. The coefficients of BANK-YOUNG and BANK- ADOLES are both positive and statistically significant at the 1% level, suggesting that banks aged 20 or under have higher C&I/GTA ratios than other banks, all else equal. In addition, the coefficient of BANK- LNAGE is negative and significant at the 1% level, suggesting that even within age categories, younger banks have higher small business lending ratios after controlling for M&As and the other exogenous variables. Similarly, the coefficients of BANK-LNAGE are negative and significant at the 1% level in all three of the regressions by age group in Table 9a. In Table 9b, the coefficients of BANK-YOUNG, BANK- ADOLES, and BANK-LNAGE have the same signs in all six cases as in Table 9a, suggesting negative effects of bank age on lending, although the statistical significance is generally reduced or eliminated. These results strongly suggest that controlling for the presence of market M&As does not eliminate the observed relationship between age and small business lending.

To evaluate the economic significance of age differences by group, we use the coefficients in the first regressions in Tables 9a and 9b that include banks of all ages. To compare young small banks with mature small banks, we begin at the mean C&I/GTA ratio for young small banks of .179, remove the estimated effect of the BANK-YOUNG dummy, and add in the estimated effect of moving BANK-LNAGE from its mean value for young small banks of 1.176 (3.241 years old) to its mean for mature small banks of 4.294 (73.259 years old). Using the full sample estimated coefficients of .1514 for the BANK-YOUNG

dummy and -.1791 for the LNAGE variable implies a decrease in the predicted proportion of small business

36 lending from 17.9% to 9.7%, a decrease of 8.2 percentage points or 45.9%.”! Thus, even after controlling for M&As and all the other exogenous variables, the age variables alone bring the predicted C&I/GTA ratio most of the way down to the 8.0% mean value for the mature small banks shown in Table 7.

To compare adolescent small banks with mature small banks, we use the mean C&I/GTA ratio for adolescent small banks of .140, the estimated coefficient of the BANK-ADOLES dummy of .0588, and move BANK-LNAGE from its mean value for adolescents of 2.410 (11.134 years old) to the mature small banks mean. This implies a decrease in the predicted proportion of small business lending from 14.0% to 9.9%, a decrease of 4.1 percentage points or 29.5%. Again, the age variables alone bring the predicted lending ratio most of the way down to the 8.0% mean value for the mature small banks.

Using the full sample means, but applying the coefficients from the 1995-1998 subperiod yields somewhat smaller predicted changes, but again, the age variables alone explain most of the difference in mean lending ratios. For young small banks, the C&I/GTA ratio is predicted to move from 17.9% to 11.4% as banks become fully mature, a decrease of 35.8%, and for adolescent small banks, the ratio is predicted to move from 14.0% to 10.8%, a decrease of 22.9%.

Overall, these results suggest that M&A activity was not the driving force behind the finding in earlier studies and here that recent de novo entrants and other young banks tend to do more small business lending than other small banks. The data do not support a strong positive external effect of consolidation activity on the small business lending of young banks — the estimated effect is either small negative (full sample) or small positive (1995-1998 subperiod). In addition, the effects of age on lending remain statistically and economically significant after controlling for M&As and the other exogenous variables. The estimated magnitudes suggest that market M&As are not responsible for the observed high ratios of small business lending ratios of young and adolescent small banks.

Robustness of the Findings on the Effects of Market M&As and Bank Age on Small Business Lending

We also run several robustness checks to verify and confirm our findings about the effects of market

M&As and bank age on small business lending by small banks. Most of these checks are analogous to the

*'The formula for the predicted small business lending proportion from increasing the age of the bank P, is given by: In(.179/(1-.179)) + .1514 + ((4.294 - 1.176) * (-.1791)) = In(P,/(1-P,)).

37 checks on the probability-of-entry regressions that alter the specifications of the exogenous variables in various ways.

As above, we try alternative specifications of the M&A variables. We again distinguish between in-market versus market-extension mergers, mergers in which the pro forma consolidated bank is small versus large, and mergers between holding company affiliates versus mergers of previously unaffiliated banks. The main results concerning the effects of market M&As and bank age are consistent with those in the main regressions shown in Tables 9a and 9b. The measured effects of M&As on lending are mixed for the banks of all ages together (positive effects of WAM-MERGE and negative effect of WAM-ACQUIS), are either negative or small for young small banks, and are generally more positive during the 1995-1998 subperiod. The effects of bank age on small business lending are strongly negative and Statistically significant, even after controlling for M&As and the other exogenous variables. The coefficients of the additional M&A variables are mostly statistically insignificant.

We again try specifying the individual t-1, t-2, and t-3 lags of the main M&A variables in place of the three year averages. Consistent with the main results, the measured positive effect of WAM-MERGE and negative effect of WAM-ACQUIS tend to offset each other for banks of all ages together, the effects of market M&As are either negative or small for young small banks, the external effects of M&As are generally more positive during the 1995-1998 subperiod, and the measured effects of bank age on small business lending are strongly negative and statistically significant.

We again rerun the main logit model adding in second-order terms on the continuous exogenous variables to allow for nonlinearities. The results are generally consistent with the main findings, except that the measured positive effect of WAM-MERGE in the regression for banks of all ages together is much smaller and not statistically significant.

We also rerun the lending model adding in the time dummies for individual years. The only important change is that the external effect actually becomes weaker. In the full sample with the banks of all ages together, the measured effect of WAM-MERGE becomes negative, although Statistically insignificant, while the coefficient of WAM-ACQUIS continues to be negative and statistically significant.

Again, the effects of market M&As are either negative or small for young small banks, the external effects

38 of M&As are generally more positive during the 1995-1998 subperiod, and the measured effects of bank age on small business lending are strongly negative and significant.

As a final robustness check, we run the lending model separately for banks headquartered in MSAs versus those headquartered in non-MSA counties, although this does not imply that all of the lending is either in metropolitan or rural markets. All of the findings with regard to market M&As and bank age are consistent with the main results, except that the external effect of M&As is even weaker in some cases (similar to our findings summarized just above for the time-dummies robustness check). In the full sample with all bank ages included for MSA-headquartered banks, the coefficients of both WAM-MERGE and WAM-ACQUIS are negative and statistically significant.

The Effects of Variables other than Market M&As and Bank Age on Small Business Lending

The estimated coefficients on the variables other than market M&As and bank age in the small business lending results of Tables 9a and 9b are only partly supportive of the hypotheses discussed above. As was true for the entry results of Tables 8a and 8b, more coefficients tend to be significant for the 1980- 1998 sample than is true for the 1995-1998 subperiod.

We hypothesized that favorable market conditions for future bank profitability — such as high market profits, growth, purchased funds ratio, and state income growth — should also positively affect a bank's small business lending. The coefficients of WAM-ROE, WAM-GROW, WAM-PEFRAT, and STINCOME are generally positive and often statistically significant, as expected, although the coefficients of WAM-PFRAT have reversed signs in some cases for the 1995-1998 subperiod.

The predicted consequences of market concentration and bank market share were uncertain, and their results are somewhat mixed. The coefficients of WAM-HERF are consistently positive and often significant, whereas the coefficients of BANK-SHARE are generally negative and often significant. Further, the marketlevel shares of large and of complex banks (WAM-SHAREL and WAM-SHAREC, respectively) appear to have mixed consequences for small business lending, depending on sample period and age of the banks.

With respect to some of the other market-level variables, the coefficients of WAM-MDEP1 and WAM-MDEP?2 were generally positive and the coefficients of WAM-RDEP1I and WAM-RDEP? were

generally negative, indicating that small business lending tended to be greater in smaller metropolitan areas

39 and larger rural areas. Once these variables were controlled for, the presence of a bank in an MSA (the variable BANK-INMSA) had a mixed effect on small business lending. With respect to market output prices, the price of consumer loans (P1) is positively related to small business lending, while the prices of the other outputs were negatively related. The prices of inputs (W1, W2, W3) tend to have a positive effect on small business lending, rather than the negative effect that would be expected, although the price of core deposits (W2) has a negative effect for the 1995-1998 subperiod.

The effects of efficiency and financial soundness are somewhat mixed, similar to our earlier results. The coefficients of WAM-NPL are generally negative, consistent with unfavorable lending conditions deterring small business lending. However, the effects of WAM-EQRAT and WAM-EFFIC are also generally negative, suggesting that better capitalized and more efficient banks lend less, contrary to expectations.

With respect to bank-level variables, the coefficients of BANK-LNGTA are generally positive and often significant and the coefficients of the size class dummies BANK-SZU10, BAN K-SZ1025, and BANK- SZ2550 are generally negative, suggesting that within the set of small banks, larger small banks tend to lend more. Related to this, the bank’s own acquisitions (BANK-ACQUIS) tend to have a positive effect on small business lending (with significant coefficients for the full 1980-1998 period), but the bank’s own mergers (BANK-MERGE) tends to be negative and significant for the full period or insignificant (for the 1995-1998 subperiod).

As for the bank’s own complexity, we hypothesized that more complex banks should make fewer small business loans. This prediction is only partially supported. The coefficient of BANK-ONEBHC is consistently positive and significant (except for the young banks), apparently indicating that some modest complexity (or sophistication) may encourage small business lending. However, coefficients of BANK- MBEC are generally negative for the young and adolescent banks (which is consistent with the findings of DeYoung, Goldberg, and White 1999), the coefficients of BANK-OUTST and are consistently negative, while the coefficients of BANK-MUL_LAY are mixed.

With respect to the bank’s profitability and financial strength, we hypothesized that a more profitable

and financially strong bank would be expected to make more loans to small businesses. The results generally

40

do not support this position. The coefficients of BANK-ROE and BANK-EQRAT were mixed, while the coefficients of BANK-NPL were generally positive. Overall, the results shown in Tables 9a and 9b are only weakly supportive of the hypotheses discussed above, except for the strong results related to bank age. 8. Conclusions

We study the dynamics of market entry following mergers and acquisitions (M&As) by market incumbents and the behavior of recent entrants in supplying output that might be withdrawn by the firms engaging in consolidation activity. Our empirical application focuses on the commercial banking industry, which provides a superior laboratory for analyzing these dynamic effects of M&As because banks produce relatively homogeneous products in a large number of geographically segmented markets under different conditions. We are able to trace the effects of over 10,000 M&As in over 2,700 local banking markets during a period of 19 years, with over 52,000 market-year observations, and almost 4,000 actual market entries.

We are motivated in part by the lack of dynamic analysis in the general industrial organization, where M&As and other actions of incumbent firms are rarely considered in studies of market entry. We are also motivated by some significant policy and research questions relevant to the banking industry. The sensitivity of entry to M&A activity in local banking markets has important implications for antitrust policy, for banking market efficiency, and for the supply of credit to small business borrowers. Prior research found that some types of bank M&As tend to reduce the supply of small business credit by the consolidating banks, but that this reduction in supply may be offset in part or in full by an external effect in which other banks respond with increases in the supply of small business funds. However, there is little prior evidence suggesting the extent to which this external effect operates through increased de novo entry or additional lending by recent entrants. Additional research has found that recent entrants tend to lend much more to small businesses as a percent of assets than other small banks of comparable size for as long as 20 years after entry. But there is little prior evidence about the extent to which this additional lending represents reactions to M&As in the local markets of the recent entrants.

To address these issues, we estimate the probability of entry into local banking markets and the small business lending of recent de novo entrants and other small banks as functions of M&A activity in the local

market and a number of other factors. Additional variables in these models include market concentration,

41 regulatory barriers to entry, and other competitive conditions; market growth and other demand conditions; market prices; and market profitability and other indicators of bank conditions. In the small business lending equations, we also take into account the bank’s age, its own M&A activity, size, ownership structure, financial condition, and market share. Our main focus is on the effects of market M&As on the probability of entry and the small business lending of recent entrants, but we also try to explain the effects of bank age on small business lending and test a number of other hypotheses about factors that affect entry and lending.

The findings are strongly consistent with the hypothesis that M&As are associated with subsequent increases in the probability of entry into the markets in which the M&As occur. The coefficients of the M&A variables in the logit equations for the probability of entry into local banking markets are consistently positive and highly statistically significant across all types of markets for the full 1980-1998 data set, and are consistently positive and almost always statistically significant for the recent 1995-1998 subperiod. The results of these logit equations also suggest that M&As have an economically significant impact. The estimates suggest that M&As explain more than 20% of all de novo entry in metropolitan markets, and more than 10% of all de novo entry in rural markets. These results suggest that at least part of the external effect of M&As found in prior research — in which other local banks respond with increases in small business lending — operates through increased de novo entry, or lending by banks that did not exist prior to the M&As. We also run three series of robustness checks that 1) use alternative econometric methods; 2) change the specifications of the exogenous variables; or 3) alter the data samples. In all cases, the findings support our main result that M&As tend to increase the probability of de novo entry.

In contrast to the probability-of-entry results, the small business lending results do not consistently support an external effect of M&As that operates through any increased lending of recent de novo entrants. The measured effect of M&As on the lending of “young” small banks (ages 1-5 years) over the full 1980- 1998 time period is negative, contrary to the hypothesized sign, although the magnitude is small, only about 2.5% of the lending of these firms. This result is reversed in the recent 1995-1998 subperiod, but again the magnitude is small.

Consistent with prior research, we find that bank age has a strong negative effect on the small

business lending of small banks. The data strongly suggest that as banks age, their small business lending

42 ratio decreases even after controlling for the effects of market M&As and all of the other variables in the model. This result is both statistically and economically significant. For young small banks (ages 1-5), the model predicts a drop of about 45% in the ratio of small business loans to assets as the banks become mature (21+ years). The model similarly predicts a lending ratio drop of about 30% for “adolescent” small banks (ages 6-20) as they move to maturity. The findings suggest that market M&As are not responsible for the observed high ratios of small business lending ratios of young and adolescent small banks.

Most of the hypotheses concerning variables other than market M&As and bank age receive only mixed support. An exception is that market size is strongly positively related to the probability of de novo entry.

Finally, these findings have implications for policy and future research. With respect to policy, the result that the probability of entry increases substantially after the occurrence of local market M&As tends to support a continued role for prospective entry in antitrust analysis of M&A cases. The probability-of-entry result — combined with the finding that recent entrants tend to have higher ratios of small business loans to assets than do other banks — is consistent with an efficient response to market needs by prospective and recent entrants. In addition, these results imply an external effect of consolidation in which the small business lending that may be reduced by M&As may be partially replaced by new entrants. However, the finding that the lending of existing recent entrants does not appear to respond positively to M&As suggests that there is no additional “kick” to the external effect from a change in lending behavior by these young banks.

With respect to future research, the findings suggest that M&As and possibly other actions of incumbent firms should be included in future models of entry in both industrial organization and banking research. The results also suggest that effort be invested in evaluating a number of different external effects of M&As on the behavior of other market participants beyond what has been considered here. In addition to entry and output, the responses of other firms in terms of their prices, locations, and product mix could be included in future industrial organization research. In banking research in particular, the external effects of M&As on deposit and loan interest rates, quantities of other types of loans, geographical distributions of

deposits, and quantity and location of bank branches could be further studied.

REFERENCES Acs, Z. J., Audretsch, D.B., 1990, Innovation and Small Firms, MIT Press, Cambridge, MA.

Akhavein, J.D., Berger, A.N., Humphrey, D.B., 1997, The effects of bank megamergers on efficiency and prices: Evidence from the profit function, Review of Industrial Organization 12, 95-139.

Allen, L., Saunders, A., Udell, G.F., 1991, The pricing of retail deposits: concentration and information. Journal of Financial Intermediation 1, 335-361.

Amel, D.F., Liang, J.N., 1997, Determinants of entry and profits in local banking markets, Review of Industrial Organization 12, 59-78.

Audretsch, D.B., 1991, New firm survival and the technological regime, Review of Economics and Statistics 68, 520-526.

Baldwin, J., Gorecki, P., 1991, Firm entry and exit in Canadian manufacturing sector, Canadian Journal of Economics 24, 300-323.

Berger, A.N., 1998, The efficiency effects of bank mergers and acquisitions: A preliminary look at the 1990s data. In: Amihud, Y., Miller, G. (Eds.), Bank Mergers and Acquisitions, Kluwer Academic, Boston, MA, 79-111.

Berger, A.N., Demsetz, R.S., Strahan, P.E., 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., Hannan, T.H., 1989, The price-concentration relationship in banking, Review of Economics and Statistics 71, 291-299.

Berger, A.N., Hannan, T.H., 1997, Using measures of firm efficiency to distinguish among alternative explanations of the structure-performance relationship, Managerial Finance 23, 6-31.

Berger, A.N., Hannan, T.H., 1998, The efficiency cost of market power in the banking industry: A test of the ‘quiet life’ and related hypotheses, Review of Economics and Statistics 80, 454-465.

Berger, A.N., Hanweck, G.A., Humphrey, D.B., 1987, Competitive viability in banking: Scale, scope, and product mix economies, Journal of Monetary Economics 20, 501-520.

Berger, A.N., Humphrey, D.B., 1992, Megamergers in banking and the use of cost efficiency as an antitrust defense, Antitrust Bulletin 37, 541-600.

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

Berger, A.N., Mester, L.J., 1997, Inside the black box: What explains differences in the efficiencies of financial institutions?, Journal of Banking and Finance 21, 895-947.

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

Berger, A.N., Udell, G.F., 1995, Relationship lending and lines of credit in small firm finance, Journal of

Business 68, 351-382.

Berger, A.N., Udell, G.F., 1996, Universal banking and the future of small business lending. In: Saunders, A., Walter, I. (Eds.), Financial system design: The case for universal banking, Irwin, Burr Ridge, IL, $59-627.

Berlin, M., Mester, L.J., 1998, On the profitability and cost of relationship lending, Journal of Banking and Finance 22, 873-897.

Blackwell, D., Winters, D.B., 1997, Banking relationships and the effect of monitoring on loan pricing, Journal of Financial Research 20, 275-289.

Chappel, W. F., Kimenyi, M.S., Mayer, W.J., 1990, A Poisson probability model of entry and market structure with an application to U.S. industries during 1972-77, Southern Economic Journal 56, 918-

927.

Cole, R.A., 1998, The importance of relationships to the availability of credit, Journal of Banking and Finance 22, 959-977.

Cole, R.A., Goldberg, L.G., White, L.J., 1999, Cookie-cutter versus character: The micro structure of small business lending by large and small banks, unpublished manuscript.

Cornett, M., Tehranian, H., 1992, Changes in corporate performance associated with bank acquisitions, Journal of Financial Economics 31, 211-234.

Craig, B.R., dos Santos, J.C., 1997, Banking consolidation: Impact on small business lending, working paper, Federal Reserve Bank of Cleveland.

Demsetz, R.S., Strahan, P.E., 1997, Diversification, size, and risk at bank holding companies, Journal of Money, Credit, and Banking 29, 300-313.

Deutsch, L.L., 1984, Entry and the extent of multiplant operations, The Journal of Industrial Economics 32, 477-488.

DeYoung, R., 1997, Bank mergers, X-efficiency, and the market for corporate control, Managerial Finance 23, 32-47.

DeYoung, R., 1998, Comment on Goldberg and White, Journal of Banking and Finance 22, 868-872. DeYoung, R., Goldberg, L.G., White, L.J., 1998, 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.

DeYoung, R., Hasan, I., 1998, The performance of de novo commercial banks: A profit efficiency approach, Journal of Banking and Finance 22, 565-587.

Dunne, T., Roberts, M., Samuelson, L., 1988, Patterns of firm entry and exit in US manufacturing industries, Rand Journal of Economics 19, 495-515.

Epstein, J.D., 1996, Years of bank mergers trigger a rise in start-ups looking for their niche, American

Banker, December 27, 1, 6.

Geroski, P.A., 1995, What do we know about entry?, International Journal of Industrial Organization 13 (4), 421-440.

Goldberg, L.G., White, L.J., 1998, De novo banks and lending to small businesses, Journal of Banking and Finance 22, 851-867.

Hannan, T.H., 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.

Hannan, T.H., 1994, Asymmetric price rigidity and the responsiveness of customers to price changes: The case of deposit interest rates, Journal of Financial Services Research 8, 257-67.

Hannan, T.H., Berger, A.N., 1991, The rigidity of prices: Evidence from the banking industry, American Economic Review 81, 938-945.

Hanweck, G., 1971, Bank entry into local markets: An empirical assessment of the degree of potential competition via new bank formation, Proceedings of a conference on bank structure and competition, 161-172.

Hubbard, R., Kuttner, K., Palia, D., 1998, Are there bank effects in borrowers’ costs of funds? Evidence from a matched sample of borrowers and banks. Unpublished working paper, Columbia University.

Hughes, J.P., Lang, W., Mester, L.J., Moon, C.-G., 1996, Efficient banking under interstate branching, Journal of Money, Credit, and Banking, 28, 1043-1071.

Hughes, J.P., Lang, W., Mester, L.J., Moon, C.-G., 1999, The dollars and sense of bank consolidation, Journal of Banking and Finance 23, 291-324.

Hunter, W.C., Timme, S.G., 1986, Technical change, organizational form, and the structure of bank productivity, Journal of Money, Credit and Banking 18, 152-166.

Jackson, W.E. III, 1997, Market structure and the speed of price adjustments: Evidence of non-monotonicity, Review of Industrial Organization 12, 37-57.

Joskow, A.S., Werden, G.J., Johnson, R.L., 1994, Entry, exit, and performance in airline markets, International Journal of Industrial Organization 12, 457-471.

Keeton, W.R., 1995, Multi-office bank lending to small businesses: Some new evidence, Federal Reserve Bank of Kansas City Economic Review 80 (2), 45-57.

Keeton, W.R., 1996, Do bank mergers reduce lending to businesses and farmers? New evidence from Tenth District States, Federal Reserve Bank of Kansas City Economic Review 81 (3), 63-75.

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 Bank of Kansas City.

Kolari, J., Zardkoohi, A., 1997a, The impact of structural change in the banking industry on small business lending, Report to the Small Business Administration.

R-4

Kolari, J., Zardkoohi, A., 1997b, Bank acquisitions and small business lending, working paper, Texas A&M University.

Kwan, S., 1998, Bank charters vs. thrift charters, Federal Reserve Bank of San Francisco Weekly Letter, No. 98-13 (April 24).

Kwast, M.L., Starr-McCluer, M., Wolken, J., 1997, Market definition and the analysis of antitrust in banking, Antitrust Bulletin, 42, 973-995.

Lang, W.W., Nakamura, L.I., 1989, Information losses in a dynamic model of credit, Journal of Finance 44, 731-746.

Levonian, M., Soller, J., 1995, Small banks, small loans, small business, Federal Reserve Bank of San Francisco working paper.

Mester, L.J., Nakamura, L.I., Renault, M., 1998, Checking accounts and bank monitoring, Federal Reserve Bank of Philadelphia working paper.

Moore, R.R., Skelton, E.C., 1998, New banks: Why enter when others exit?, Financial Industry, Federal Reserve Bank of Dallas, First Quarter, 1-6.

Nakamura, L.I., 1993, Commercial bank information: Implications for the structure of banking. In: Klausner, M, White, L.J. (eds.), Structural change in banking, Irwin, Homewood, IL, 131-160.

Neumark, D., Sharpe, S.A., 1992, Market structure and the nature of price rigidity: Evidence from the market for consumer deposits, Quarterly Journal of Economics 107, 657-680.

Noulas, A.G., Ray, S.C., Miller, S.M., 1990, Returns to scale and input substitution for large U.S. banks, Journal of Money, Credit, and Banking, 22, 94-108.

Office of the Comptroller of the Currency, 1995, Comptroller’s Corporate Manual. Peek, J., Rosengren, E.S., 1996, Small business credit availability: How important is size of lender? In: Saunders, A., Walter, I. (Eds.), Financial system design: The case for universal banking, Irwin, Burr

Ridge, IL, 628-655.

Peek, J, and E.S. Rosengren, 1998, Bank consolidation and small business lending: It’s not Just bank size that matters, Journal of Banking and Finance 22, 799-819.

Peristiani, S., 1997, Do mergers improve the X-efficiency and scale efficiency of U.S. banks? Evidence from the 1980s, Journal of Money, Credit, and Banking 29, 326-337.

Petersen, M.A., Rajan, R.G., 1994, The benefits of firm-creditor relationships: Evidence from small business data, Journal of Finance 49, 3-37.

Petersen, M.A., Rajan, R.G., 1995, The effect of credit market competition on lending relationships, Quarterly Journal of Economics 110, 407-443.

Prager, R.A., Hannan, T.H., 1998, Do substantial horizontal mergers generate significant price effects?, Evidence from the Banking Industry, Journal of Industrial Economics, 46, 433-452.

R-5

Rhoades, S.A., 1997, Have barriers to entry in retail commercial banking disappeared?, Antitrust Bulletin 42, 997-1013.

Rhoades, S.A., 1998, The efficiency effects of bank mergers: An overview of case studies of nine mergers, Journal of Banking and Finance 22, 273-291.

Rose, J.T., 1977, The attractiveness of banking markets for de novo entry: The evidence from Texas, Journal of Bank Research 7, 284-293.

Rose, J.T., Savage, D.T., 1983, Bank holding company de novo entry, bank performance, and holding company size, Quarterly Review of Economics and Business 23, 54-62.

Radecki, L.J., 1998, The expanding geographic reach of retail banking markets, Federal Reserve Bank of New York Economic Policy Review 4, 15-34.

Rosenbaum, D. I., 1993, Profit, entry and changes in concentration, International Journal of Industrial Organization 11, 185-203.

Sapienza, P., 1998, The effects of banking mergers on loan contracts, Northwestern University working paper.

Seelig, S.A., Critchfield, T., 1999, Determinants of de novo entry in banking, Federal Deposit Insurance Corporation Working Paper 99-1.

Sharpe, S.A., 1990, Asymmetric information, bank lending, and implicit contracts: A stylized model of customer relationships, Journal of Finance 45, 1069-1087.

Siegfried, J.J., Evans, L.B., 1992, Entry and exit in United States manufacturing industries from 1977 to 1982. In: Audretsch, D.W., Siegfried, J.J. (Eds.), Empirical Studies in Industrial Organization: Essays in Honor of Leonard W. Weiss, Kluwer Academic Publishers, Dordrecht.

Siegfried, J.J., Evans, L.B., 1994, Empirical studies of entry and exit: A survey of the evidence, Review of Industrial Organization 9, 121-155.

Simons, K., Stavins, J., 1998, Has antitrust policy in banking become obsolete? New England Economic Review, Federal Reserve Bank of Boston, 13-26.

Strahan, P.E., Weston, J., 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.

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

Toevs, A., 1992, Under what circumstances do bank mergers improve efficiency? Proceedings of a conference on Bank Structure and Competition, Federal Reserve Bank of Chicago, Chicago, IL, 602- 628.

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.

R-6

Werden, G.J., Froeb, L.M., 1998, The entry-inducing effects of horizontal mergers: An explanatory analysis, Journal of Industrial Economics, 46, 525-543.

White, E.N., 1992, The Comptroller and the Transformation of American Banking, 1960-1990, Washington, DC: Office of the Comptroller of the Currency.

Williamson, O.E., 1967, The economics of defense contracting: Incentives and performance. In: McKean, R. (Ed.), Issues in Defense Economics, Columbia University Press, New York.

Williamson, O.E., 1988, Corporate finance and corporate governance, Journal of Finance 43, 567-91. Yip, G., 1982, Barriers to Entry: A Corporate Strategy Perspective, D.C. Heath, Lexington, MA.

Zardkoohi, A., Kolari, J., 1997, The effect of structural changes in the U.S. banking industry on small business lending, working paper, Texas A&M University.

%00'001 86S¢S

%00°004 Eel

%00°001 86EE

%00°001 88r6

%00°001 SL66L

%00°001 pdyst

sieed py Jeo}

SZ8E

y60e

ble

OLb

62E

8st

syuesjua # |2}0}

Ze°2

82°8Sl

aloe

Sé't

06'4

98°0

sleek JHW OOL /s\uesjUue

"sanuUNOD YSW-UOU JO (SYSW) Sealy jeoNSsHEIS UENOdoROy Se PoUap ale sjeyseW BurUeg :se]ON

%c0'0 %S0'0 %el'0 %90°0 %60'0 %0Z'0 %eS'0 %LLE ELSE Ov 6 Se 19 62 6p 901 LLé S9SL £8e0S %89'0 PIB L WEY PbO? SLVE %S0'9 %86'6 YWOSES WISH Ov 6 ve 8S de ov 08 cel Zoe ro %EO'O %60'0 %90'0 %81'O %89'0 WIGS WLLEL %b9'SB bb b € 2 9 €¢ £8 89r 8082 %10°0 %E0'0 %ev'O WLEE %9S' SE v 4 € Ov ZLe 2906 %S0°0 Yo L Seb\'86 é 6 LOE SPpSél %c0'0 %z8'0 %9L'66 @ & eS 6LE8l SyUBIUa # +142 Og-bk O1-9 S v € z b 0

wnuixew

(8661-0861 Bleq Jenuuy) Jeyey jo azig Aq Aug OAON 9g YURg felOIeWWOD Jo Aouenbel4

} e1geL

SjuelUg JO JEqUINY

WLOL

as$<

as-l$

al$-wooes

WOOE-001$

Wwoo1$> susodap ul aZis JoueW

SLAYMYVA TIV

%00'001 6€09

%00°001 Leel

%00'001 OZLLE

%00'001 vest

%00°001 as

seed yyw J2}0}

800€

£602

eSZ

LOL

sjueUue # J2}0}

L8'6r

br est

el €e

Os'OL

62°r1

00°0

sieah PW OOL /sjuesjue

"sanunod YSW-uoU JO (syS) Sealy jeoNsHe]S UeOdoNey se peuyap ose sjoyre BupuRg ‘selon

%Sb'0 *bv'O %IO'L %8v'0 %6L'0 7 a WEE HE vl WeOLL

Ov 6 SZ 19 62 8p SOL 92 S28 Sor WLOL %89'0 %c8' | WEY %b0'S BLE %90'9 %66'6 WIVES R*WBISH Or 6 ve 8S Ze ov 08 cel 90E evo ass$< %EO'O %60'0 %90'0 %6 10 %EL'O %29'S? WO VL %96'18 tL 1 € 2 9 £2 €8 Sp 86S2 gs-1$ %EL'O MLE L LE L WELLE € é Le elk S6EL at$-Wooes 6S VI %IL°S8 b re As WOO0E-001$ %00°001 0 € WOOl$> susodep ul aZIg JOUeW sjuelue # +12 Og-tt _—Ol-9 S id € z L 0 wunuuIxelw sjuedUg JO JaquNnn

SLSEMYVW NVLIIOdOYLAW (8661-0861 B}ed Jenuuy) jaysey jo azis Aq Aug OAON 8q YURG [eloIe@UWOD Jo Aouenbel4

2 gel

%00°001 9SS9b

%00°001 i

%00°001 820

%00°001 vS6L

%00°001 LO66L

%00°001 LZpel

sieed yyw

JB}O}

998

ce

60€

2L€

8Sl

syuesjue # J2}0}

98°L

00°0S

S9'6

88'e

68'L

98°0

sJeeA HW OOL /sjuesjue

sjueue # +12

*S@HUNOD YSW-UOU JO (SYSW) Sealy jeoNsHeIS UeWOdoNay] se peuyep ose sjayseW Bunjueg selon

£ 9iqeL

(8661-0861 e}eq fenuuy) Jayey jo azig Aq Anuy OAON eg YURg jeOUeWIWO4 yo Aouenba:4

%00'0 %00'0 %00'0 %00°0 %80'0 ROLL %ee' BE 0 0 b b SE 062 6cLSP %00'0S += %00'0S L b WSL Rbel'9 Sol b'e6 v as Ole %10'0 %10°0 Yve'0 WEEE Wer’ 96 b b 6h 9c 6992 %S0'0 %08" | %SL'8E 6 6SE eeS6l %c0'0 %28'0 %9L'66 € eSh SLEst 01-9 S v € é b 0

S]UBI]UQ JO JOQUINN

WLOL

as$<

as-l$

gl$-Wooes$

WOO0E-001$

WOOl$> syisodep ul

ezis Joe,

SLAWYVW Wend

*(diyssaumo Auedwoo Buiproy yueq yo eaBueyo jnq paurejas sayeYo) UONIsinboe Ue JO (payeptjosuoo B1OM SIBYWeYD YU BJO JO OM)) JoBsewW & U! PEAJOAU! SEM Joye OY) UL YUeG AUR ‘sIBAaA BalU) SNOIAeld euy jo Aue ul j! AAnoe Wey JoNd aaey 0} pesapisuod s! joysew

:SOJON €Lle S39 %8S'b O61 209 ee'0 ECO @ 9912 898 LLG 8st 8661 9222 9s°9 %O06'P cet ZL9 69°0 %6S'0 v 6602 8r'38 %6C'9 BZl 2661 LLle ees %96'E Shi ep. ¥S'0 %byS'O v yeo7d 802 %le's brl 9661 LLLE SLE %88'?e pol brs €8°0 %EBO Z ee6l 20'S BLE 26 S661 LLLE elt %BS | 8b SZOL ZE'0 WLE'O id cOLt 69°¢ SES by y66l p9le 82% %8B' | e€9 vELL SEO %SEO v O€9L coe v6? 6S e661 tLZe Sec %MLL'S 62 cOlh 2s'0 %eS'O 9 6091 Sy WIE'E 7 2661 LZZe LL'b SSE vit 9oLL 68°0 %68'0 Ol Sb9OL ce'9 %98'v pol L661 LLZ2e toa ee) %lO'v OZL Lyel 830 %8B'0 bh veSh er ol %9S'9 6SL O661 elle ced %80'v £02 OSeL c2'0 %eEL'0 6 coSh SLel NES’O v6l 6861 9Sl2 798 RLY te 9SelL sel %WSEL Zi oost eZ vl LOL Loo 8s6l €9l2 Sel Eby Lle evel ell ELL bl O¢SIL SEEL vel £02 2861 olle ve'é RESP 6Sc brel e6'l EE b 92 8erk ce'ol WIL eee 986 2922 pie?el %EC'S 9ee rach ag ove %80'E os SOE} cole %el'6 982 S861 9922 SZ'Pk %B1'9 80b OSI co %S6'E cB 9614 9c Le %ib'6 9ce yeél p9le Orel %v9O'S ele ZELb ery %Sv'E LL Zc0l oLl'8e %SE'6 S6e €86i L920 eS be %EB’S bee 6rs8l £87 Yobb bv 06 clé eeSe CEE Le? c86l o9le 092 wer'y Ole ZE6L SIE EIS 19 S28 90°81 %L9'S 6h L86L 8sl2 CLL %98'v Ele S202 ope SLE OZ eel LS'6t %69'6 erl og6! sjeyew jo sjeywewoo, Ajuajo sjuequs jo — sJayJeW jo = sJayeW QO] = Anju jo SJUeqUajO = SJayewW yO —s sjayre QOL = Aqua Jo sjueque jo YyVvaA Jequinu /s\ueue Aouenbey = sequunu Jequinu /s\uesque Aouenbay = saquinu Jequinu /sjuesjua Aouenbey = saquunu SSLAYVW TIV NI AWLN3A ALIALLOV VW YOldd LNOHLIM SLAYYVW NI AYLNS = ALIAILOV VW HOIdd HLIM SLAYHVW NI AHLNA

(8661-0861 eIeq fenuuy) AlAnoY WRN InoUIM PUR WIM SJOUeW UI AU OAON aq yURg |eIQJaWIWIOD Jo sWaNeg

pagel

cbsk ZSZt LZSLL ZSZL 8SZl 9991 OZ9L Z9S1 Q8Lt ZSOL Ov9 96P cé6y bev vip v0r c0r LLE 9LE sjeyiew jo Jequuinu

198 Ord SZg Lop 66'L Aci Eve 86°F 922 Olt 168 82 OL se6 plch 09°81 S002 bebe 90°bL 6e'rl s}eywew OOL /S\URJUS

%EI'S %9I'S Sle v SIP’ %c8 | W?IA'S Ob'2 %IE'E OP'S %ES'O %IL'S %I9'L %ES'S WIE L %O0L'S %EB'O %WEL9 %EI'D %S9'9 Ajjua jo Aouenbady

9h oel LOL 8 SE cy bs gL cé ESL Zs LS <4 $S Zé 18 107 €S 9S s}UedUa JO Jequunu

SALVLS ONIHONVYE A0IMALVLS

“(Ayunod duo Ajuo ul sayouelg “6"8) Aem wos UI! pe}o\sei ase ese) jnq ‘seyouesg ajdiyjnw aaey ued Sejye}s Bulyouesg pay} ul syuReg “yuRg Jed adyJo oles IN} auO AjUO More Ajjessueb suoNe;nBas Buryueq yun

298 926 226 £6 £26 lol PLOL 0901 €8ck EbLbl esl 0091 9091 e29h cegl vest 96E1 ecrk 2vel SJEHIBW JO Jequinu

suoneinBey Bunjueg ayels Aq Anjug oAoN aq yueg jelouaWWOD Jo swayed

cee 29'S 20'S gr? Opt L0°¢ Lye oe 09°F ory v6'8 Be2 sed 28 8e'8 28'S S69 80°F 09'r S}OHIBW OOL /Sjues}Uua

%BE'?S eA %BB'E ILS NES | %BS' | LE t Sv? %96'S %Bv'S %99'v BSE %98'E %Z0'S %eS'Y %LBE RELY %99'E %EB'E Auyue jo Aouanbesy

gs aqey

ve es Lv & el Le Se ve 6S €9 e9b 9¢1 SLt Or trl 06 246 8S AC] SJUBIUd JO Jequinu

S3LVLS ONIHONWYS GALINIT

2S

Si¢ Sle ple L6S €6S €6S c6S 8rZ see 988 8s6 s}@4J2WW JO Jequinu

SE ves bed bys 229 c0°9L S62 60°2E 28°92 Ov Lt Zibb c6'6 S}OHJEW OOL /S}uejUe

SALVLS ONIANVE LINN

VISE HEL v GLY YveL€E %I0'v GOL %9IC'B %L6'6 %60°6 %BIL %80°S %SE'S Ajua jo Aouenbey

6L

db

gL

Ov

S6

cr O61 LO¢ pSl

66

S6 SJUBIUS JO Jequinu

8661 2661 9661 S661 661 e661 e661 L661 0661 6861 8861 £861 9861 S86l P86El e861 c86l 1861 0861

HVSA

:SO]ON

Symbol

ENTRY

MKT-MERGE

MKT-ACQUIS

MKT-HERF

MKT-SHAREL

MKT-SHAREC

UNITB

LIMITB

NEWLIB

INTST

MKT-GROW

STINCOME

MKT-PFRAT

TABLE 6

Variables Employed in Entry Regressions

Definition

Dependent Variables

Dummy for positive entry.

Market Mergers and Acquisitions (M&A)

Share of local market deposits in banks involved in mergers in which two or more bank charters are consolidated, averaged over the previous three years.

Share of local market deposits in banks involved in acquisitions in which the banks retain their separate charters but change their bank holding company ownership, averaged over the previous three years.

Other Competitive Conditions

Local market Herfindahl index.

Share of market deposits held by large banks (GTA > $100M).

Share of market deposits held by complex banks (owned by out-of-state or multilayered (BHC).

Dummy variable indicating unit banking state.

Dummy variable indicating limited branching state (statewide branching dummy is excluded from the regressions as the base case).

Dummy, equals | if the state moved to a more liberal branching rule this year.

Dummy variable indicating interstate bank holding company expansion is allowed.

Market Demand Conditions

Growth rate of market deposits.

Real state income growth.

Market average purchased funds/GTA ratio.

Large MSAs

Mean (Std. Dev.)

0.517 (0.500

0.181 (0.152)

0.035 (0.049)

0.153 (0.072)

0.658 (0.037)

0.223 (0.187)

0.062 (0.242)

0.362 (0.481)

0.042 (0.200)

0.69 (0.464)

0.035 (0.195)

0.033 (0.024)

0.302 (0.111)

Small MSAs

Mean

Rural Markets

Mean

(Std. Dev.) (Std. Dev.)

0.150 (0.358)

0.136 (0.137)

0.04] (0.062)

0.201 (0.082)

0.852 0.142

0.240 (0.272)

0.092 (0.290)

0.431 (0.495)

0.046 (0.211)

0.599 (0.490)

0.022 (0.187)

0.031 (0.027)

0.214 (0.088)

0.018 (0.135)

0.068 (0.126)

0.026 (0.064)

0.422 (0.231)

0.397 (0.402)

0.136 (0.247)

0.134 (0.341)

0.488 (0.500)

0.045 (0.207)

0.578 (0.494)

0.035 (0.973)

0.030 (0.028)

0.147 (0.081)

Symbol

MKT-LNDEP

MKT-MDEPI1

MKT-MDEP2

MKT-RDEP2

MKT-RDEP3

MKT-P1

MKT-P2

MKT-P3

MKT-P4

MKT-WI1

MKT-W2

MKT-W3

MKT-ROE

MKT-NPL

MKT-EQRAT

MKT-EFFIC

Definition

Market size (log of market deposits).

Size dummy, indicates metropolitan market < $1B in deposits (This variable is excluded as the base case in the 1995-1998 subperiod regressions).

Size dummy for metropolitan markets from $1B - $5B in deposits (This variable is excluded as the base case in the 1980-1998 full sample period regressions).

Size dummy, indicates rural market $100M-$300M in deposits. (Dummy for rural markets less than $100M is excluded as the base case in rural market regressions).

Size dummy for rural markets greater than $300M in deposits.

Market Prices

Market average price of consumer loans (installment, credit cards, and related plans).

Market average price of non-real estate business loans (commercial and industrial loans, agricultural loans, loans to depository institutions, etc.).

Market average price of real estate loans.

Market average price of securities (all non-loan financial assets).

Market average price of purchased funds (jumbo CDs, foreign deposits, federal funds purchased, all other liabilities except core deposits).

Market average price of core deposits (domestic transactions accounts, time and savings).

Market average price of labor (1000's of constant 1994 dollars per employee).

Condition of Market Banks

Return on equity average in market.

Market average nonperforming loan ratio.

Market average equity/GTA ratio.

Market average cost efficiency measure (the negative of the market average residual from Fourier flexible cost function).

Large MSAs

Mean (Std. Dev.)

16.401 (0.732)

0.092 (0.036)

0.132 (0.087)

0.076 (.028)

0.051 (0.012)

0.063 (0.024)

0.034 (0.015)

35.341 (6.879)

0.106 (0.105)

0.030 (0.025)

0.067 (0.012)

0.004 (0.070)

Small MSAs

Mean (Std. Dev.)

14.155 (0.637)

0.329 (0.470)

0.671 (0.470)

0.083 (0.048)

0.157 (0.098)

0.068 (0.037)

0.057 (0.014)

0.064 (0.025)

0.040 (0.016)

30.694 (9.987)

0.115 (0.092)

0.025 (0.023)

0.074 (0.012)

-0.005 (0.075)

Notes: All financial values are measured in real 1994 dollars using the GDP deflator.

Rural Markets

Mean (Std. Dev.)

11.736 (0.976)

0.445 (0.497)

0.183 (0.387)

0.070 (0.061)

0.161 (0.087)

0.061 (0.052)

0.063 (0.016)

0.058 (0.024)

0.046 (0.019)

30.147 (5.414)

0.112 (0.113)

0.029 (0.029)

0.087 (0.020)

-0.012 (0.090)

Symbol

C&V/GTA

WAM-MERGE

WAM-ACQUIS

WAM-HERF

WAM-SHAREL

WAM-SHAREC

BANK-INMSA

UNITB

LIMITB

NEWLIB

INTST

Young Adolescent Mature Small Small Small Banks Banks Banks (1-5 years) (6-20 years) (21+ years) Definition Mean Mean Mean ; (Std. Dev.) - (Std. Dev.) (Std. Dev.) SSS SS NSS SSS Dependent Variable Commercial and industrial loans as a proportion of the 0.179 0.140 0.080 bank’s assets. (0.109) (0.093) (0.055) Mergers and Acquisitions (M&As) in the Bank’s Market(s) Weighted average of the shares of market deposits in 0.101 0.103 0.113 banks involved in mergers (in which two or more bank (0.129) (0.137) (0.129) charters are consolidated, averaged over the previous three years), where the weights are the proportions of this bank’s deposits in each market. Weighted average of the shares of market deposits in 0.038 0.035 0.029 banks involved in acquisitions (in which the banks (0.056) (0.055) (0.050) retain their separate charters but change their bank holding company ownership, averaged over the previous three years), where the weights are the proportions of this bank’s deposits in each market. Other Competitive Conditions in Bank’s Market(s) Weighted average of local market Herfindahl index. 0.185 0.200 0.265 (0.135) (0.146) (0.164) Weighted average of shares of market deposits held by 0.747 0.673 0.418 large banks (GTA > $100M). (0.289) (0.312) (0.346) Weighted average of shares of market deposits held by 0.229 0.207 0.229 complex banks (owned by out-of-state or multilayered (0.243) (0.253) (0.247) (BHC). Dummy, indicating that bank is headquartered in an 0.734 0.643 0.233 MSA. (0.442) (0.479) (0.423) Dummy variable indicating unit banking state. 0.272 0.224 0 (0.445) (0.417) (0) Dummy variable indicating limited branching state 0.354 0.434 0.411 (State-wide branching dummy is excluded from the (0.478) (0.496) (0.492) regressions as the base case). Dummy, equals 1 if the state moved to a more liberal 0.082 0.060 0.004 branching rule this year. (0.278) (0.243) (0.066) Dummy variable indicating interstate bank holding 0.462 0.509 1.000 company expansion is allowed. (0.499) (0.500) (0.012)

TABLE 7 Variables Employed in Lending Regressions

For Small Banks (GTA < $100 Million)

Young Adolescent Mature

Small Small Small

Banks Banks Banks (1-5 years) (6-20 years) (21+ years)

Symbol Definition Mean Mean Mean

(Std. Dev.) (Std. Dev.) (Std. Dev.) SSS SSS SS

Demand Conditions in the Bank’s Market(s)

STINCOME Real state income growth 0.029 0.301 0.045 (0.027) (0.027) (0.027) -- WAM-GROW Weighted average of growth rate of market deposits. 0.030 0.021 0.021 (0.169) (0.171) (0.142) WAM-PFRAT Weighted average of market average purchased 0.278 0.236 0.162 funds/GTA ratio. (0.129) (0.121) (0.072) STINCOME Real state income growth 0.029 0.301 0.045 (0.027) (0.027) (0.027) WAM-LNDEP Weighted average of market size (log of market 15.058 14.566 12.805 deposits). (2.085) (2.146) (1.662) WAM-MDEP1 Proportion of bank’s deposits in metropolitan markets 0.046 0.055 0.019 < $IB. (0.208) (0.226) (0.131) WAM-MDEP2 Proportion of bank’s deposits in metropolitan markets 0.187 0.197 0.100 from $1B - $5B. (0.389) (0.394) (0.292) WAM-RDEP1 Proportion of bank’s deposits in rural markets < 0.053 0.073 0.184 $100M. (0.223) (0.256) (0.379) WAM-RDEP2 Proportion of bank’s deposits in rural markets $100M - 0.112 0.164 0.421 $300M. (0.314) (0.366) (0.482) WAM-RDEP3 Proportion of bank’s deposits in rural markets greater 0.099 0.118 0.179 than $300M. (0.297) (0.318) (0.375)

Prices in the Bank’s Market(s)

WAM-P1 Weighted average of market average price of consumer 0.087 0.085 0.089 loans (installment and credit cards and related plans). (0.035) (0.036) (0.023)

WAM-P2 Weighted average of market average price of non-real 0.156 0.169 0.104 estate business loans (commercial and industrial loans, (0.092) (0.099) (0.032) agricultural loans, loans to depository institutions, etc.).

WAM-P3 Weighted average of market average price of real estate 0.0743 0.072 0.077 loans. , (0.026) (0.025) (0.012)

WAM-P4 Weighted average of market average price of securities 0.057 0.058 0.047 (all non-loan financial assets). (0.012) (0.014) (0.006)

WAM-WI1 Weighted average of market average price of purchased 0.070 0.066 0.040 funds Gumbo CDs, foreign deposits, federal funds (0.022) (0.025) (0.008) purchased, all other liabilities except core deposits).

WAM-W2 Weighted average of market average price of core 0.042 0.040 0.027 deposits (domestic transactions accounts, time and (0.015) (0.017) (0.007)

savings).

Young Adolescent Mature Small Small Small Banks Banks Banks (1-5 years) (6-20 years) (21+ years) Symbol Definition Mean Mean Mean (Std. Dev.) (Std. Dev.) (Std. Dev.) SSSSSSSSSSSSSSSSSSSSM9MaMS er WAM-W3 Weighted average of market average price of labor 32.797 32.026 33.506 (1000's of constant 1994 dollars per employee). (5.973) (6.342) (5.191) Condition of Banks in the Bank’s Market(s) WAM-ROE Weighted average of average ROE in markets. 0.069 0.101 0.133 (0.146) (0.111) (0.039) WAM-NPL Market nonperforming loan ratio. 0.036 0.028 0.024 (0.033) (0.028) (0.012) WAM-EQRAT Market average equity/GTA ratio. 0.069 0.074 0.094 (0.015) (0.016) (0.018) WAM-EFFIC Weighted average of market efficiency measure (the -0.005 -0.007 -0.001 negative of the market average residual from Fourier (0.062) (0.066) (0.078) flexible cost function). Bank Age BANK-LNAGE Log of bank’s age, in years. 1.176 2.410 4.294 (0.345) (0.371) (0.421) BANK-YOUNG Dummy variable, indicating a young bank, age 1-5 years 1.000 - - (banks younger than | year do not have the required (0.000) - lagged values of the BANK variables and are excluded from the lending analysis). BANK-ADOLES — Dummy variable, indicating an adolescent bank, age 6- - 1.000 - 20 years. - (0.000) - BANK-MATURE Dummy variable, indicating a mature bank, age 21+ - - 1.000 years. - - (0.000) Bank Mergers and Acquisitions (M&As) BANK-MERGE Dummy variable, indicating that bank survived a 0.010 0.011 0.023 merger. (0.097) (0.106) (0.151) BANK-ACQUIS Dummy variable indicating that bank changed top-tier 0.045 0.082 0.061 BHC (0.207) (0.274) (0.239) Bank Size BANK-LNGTA Log of gross total assets. 10.241 10.507 10.491 (0.647) (0.611) (0.662) BANK-SZU10 Dummy variable indicating GTA < $10M. 0.065 0.027 0.039 (0.246) (0.162) (0.193) BANK-SZ1025 Dummy variable indicating GTA $10M - $25M. 0.355 0.237 0.247 (0.479) (0.425) (0.431) BANK-SZ2550 Dummy variable indicating GTA $25M - $50M. 0.373 0.389 0.355 (0.484) (0.488) (0.478) $Z50100 Dummy variable indicating GTA $50M - $100M. 0.253 0.355 0.351 (0.435) (0.479) (0.477)

Young Adolescent Mature

Small Small Small

Banks Banks Banks (1-5 years) (6-20 years) (21+ years)

Symbol Definition Mean Mean Mean

(Std. Dev.) (Std. Dev.) (Std. Dev.) SaSaNaesenanaawjwj[q»oqNCSM WS SsSsSsSsSsSsSsSmmnmnMSmaS ee Sa

Bank’s Ownership Structure

BANK-ONEBHC = Dummy variable indicating that bank is member of a 0.192 0.298 0.521 BHC. (0.394) (0.457) (0.500) -. BANK-MBHC Dummy variable indicating that bank is member of a 0.216 - 0.235 0.233 multi-bank BHC. (0.412) (0.424) (0.423) BANK- Dummy variable indicating that bank is member of a 0.026 0.430 0.079 MUL_LAY multi-layered BHC. (0.159) (0.203) (0.270) BANK-OUTST Dummy variable indicating that bank is member of an 0.0290 0.025 0.033 out-of-state BHC. (0.168) (0.156) (0.179)

Bank’s Financial Condition

BANK-ROE Bank’s return on equity. -0.010 0.072 0.111 (0.349) (0.251) (0.064)

BANK-EQRAT Bank’s equity/GTA ratio. 0.107 0.081 0.105 (0.053) (0.031) (0.034)

BANK-NPL Bank’s nonperforming loan ratio. 0.027 0.029 0.028 (0.040) (0.038) (0.024 )

Bank’s Market Share BANK-SHARE Weighted average of bank’s shares of local market 0.034 0.065 0.174

deposits. (0.080) (0.121) (0.206)

Table 8a Logit Estimation of Probability of Commercial Bank De Novo Entry Full Sample Period 1980-1998

Large Metropolitan Markets Small Metropolitan Markets Rural Markets

Parameter Standard Parameter Standard Parameter Standard Variable Estimate Error Estimate Error Estimate Error INTERCEPT -10.6576 ~ 2.2223 -15.0134 ~~ 1.6273 -17.8439 ~ 1.2874 MKT-MERGE 1.6196 ~ 0.4754 1.6939 0.3492 = 1.0037" 0.3132 MKT-ACQUIS 5.1520 ~ 1.5103 2.1145 0.7448 = 1.8296 0.5175 MKT-HERF -0.2346 0.8058 -0.1050 0.4821 1.4353 ~ 0.2591 MKT-SHAREL -6.7606 ~ 1.3424 “1.3741” 0.4420 -0.2148 0.1535 MKT-SHAREC 0.6421 * 0.3297 0.4817 ~ 0.2007 0.7585 ~ 0.1526 UNITB 1.5537 ~ 0.5452 4.1217" 0.1855 1.3767 ~ 0.1347 UMITB -0.2565 0.1612 0.0452 0.1175 0.4542 ~ 0.1014 NEWLIB 1.0286 ~~ 0.3451 0.1452 0.2369 0.1562 0.2079 INTST 0.1822 0.2319 -0.1496 0.1585 -0.35 ~ 0.1199 MKT-GROW 0.2522 0.3493 -0.1014 0.2301 _-0.2447 0.1990 STINCOME 23.5777 3.1415 13.1544" 1.7283 7.7188 1.2105 MKT-PFRAT 3.1862 ~ 0.9502 2.5616 ~ 0.6252 = 2.8581 0.4973 MKT-LNDEP 0.8842 ~ 0.1194 0.9784 ~~ 0.1104 = 1.0433 ~ 0.1057 MKT-MDEP1 -0.0874 0.1565 MKT-RDEP2 -0.104 0.1490 MKT-RDEP3 -0.2352 0.2299 MKT-P1 3.7846 3.7339 7.0249 ~ 2.2164 1.3733 1.2954 MKT-P2 1.4196 1.6419 2.6202 ~ 0.9119 1.2439 ~ 0.5986 MKT-P3 -8.1778 4.5493 -7.9600 ~~ 2.6969 -3.126~ 1.4916 MKT-P4 2.4507 8.8961 -7.0684 5.1620 5.1978 4.0226 MKT-W1 2.2188 5.6042 3.0477 3.3577 1.0824 1.9667 MKT-W2 17.2020 ~ 8.4107 -1.8333 4.9403 -4.9823 3.6202 MKT-W3 -0.0156 0.0163 -0.0139 0.0127 0.00186 0.0081 MKT-ROE 0.1729 0.9036 1.4109 ~ 0.5632 -0.6341 ~ 0.1813 MKT-NPL -5.2705 4.8372 0.8775 3.2750 0.3613 1.7320 MKT-EQRAT 0.2618 7.3512 -12.7580 ~~ 4.6397 -7.3865 ~ 2.6819 MKT-EFFIC -0.9546 1.0558 -0.9081 0.6112 -0.3864 0.4748 Num. Obs. 1,316 4713 44,688 Adjusted R? 0.1775 0.1145 0.0938

” Significant at the 10% level ” Significant at the 5% level ™ Significant at the 1% level

The adjusted R? for the logit equations measures the proportion of the log-likelihood value explained by

the model's non-intercept independent variables, i.e., 1 - (log L act/(log Lint), where Lai. denotes the likelihood value of estimation with all the independent variables and L jy; denotes the likelihood value of estimation with only the intercept.

Table 8b Logit Estimation of Probability of Commercial Bank De Novo Entry 1995-1998 Subperiod

Large Metropolitan Markets § Small Metropolitan Markets Rural Markets

Parameter Standard Parameter Standard Parameter Standard Variable Estimate Error Estimate Error Estimate Error INTERCEPT -27.1742 999.5000 -14.9825 ~ 3.6086 -35.7704 2464.5000 MKT-MERGE 2.1665 ~ 1.0184 1.6795 ~ 0.7561 1.5958 ~ 0.6438 MKT-ACQUIS 9.1977 ~ 4.3766 2.9420 1.8269 2.3612° 1,3222 MKT-HERF 0.8971 1.6935 -0.5789 0.9095 0.4789 0.6414 MKT-SHAREL -5.1351 4.2075 -0.7172 1.2665 0.5727 0.3870 _ MKT-SHAREC 0.6793 0.6325 0.9567 ~ 0.4081 0.24114 0.3318 LIMITB -0.2540 0.4653 0.0718 0.2569 0.7905” 0.1805 NEWLIB 14.4321 2001.5000 1.2016 0.9877 0.4913 1.0736 INTST 12,8330 999.5000 14.1510 2464.5000 MKT-GROW -1,3544 0.8530 -0.3977 0.4702 = -0.7529 0.4984 STINCOME 10.3202 5.4646 0.0086 3.5894 4.3263 2.8703 MKT-PFRAT 1.3830 2.7699 1.8347 1.5800 -2.3646 1.5326 MKT-LNDEP 0.8804 ~ 0.2381 0.9951 7” 0.2249 «1.2026 ~ 0.2314 MKT-MDEP2 -0.4031 0.3403 MKT-RDEP2 -0.3093 0.3953 MKT-RDEP3 -0.3543 0.5654 MKT-P14 5.1105 7.8357 -0.7150 5.1883 4.5292 3.2088 MKT-P2 7.2283 8.6653 2.5399 3.9784 2.9588 2.4401 MKT-P3 -16.9345 20.6689 2.5208 13.5672 -1.2502 7.9498 MKT-P4 -2.7475 32.4448 -32.0857° 15.5412 4.9674 14.6602 MKT-W1 5.8840 27.3222 -10.9362 17.5987 23.2010" 12.4094 MKT-W2 23.4911 37.1263 69.3324 22.0750 —_-1.2586 16.2992 MKT-W3 0.0333 0.0391 0.0208 ~~ 0.0252 0.0185 0.0196 MKT-ROE 12.5356 ~ 5.5629 0.7141 1.0483 -1.6152 2.5831 MKT-NPL -89.6090 ~~ 26.4133 -39.8364° 17.0324 2.6316 8.3430 MKT-EQRAT 14.1192 20.5332 -15.5751 10.8050 -2.7056 6.4260 MKT-EFFIC -1.3163 2.0023 -2.5587 ~ 1.0687 -2.0235 ~ 0.9941 Num. Obs. 303 969 9,407 Adjusted R? 0.1831 0.0847 0.1210

” Significant at the 10% level ” Significant at the 5% level ™ Significant at the 1% level

The adjusted R? for the logit equations measures the proportion of the iog-liketihood value explained by the model's non-intercept independent variables, i.e., 1 - (log L ai.)/(log Liyr), where Lau. denotes the likelinood value

of estimation with alt the independent variables and L jy; denotes the likelihood value of estimation with only the intercept.

Some missing or unusual coefficients and standard errors appear for some of the state banking regulation variables for the 1995-1998 subperiod because most of the deregulation occurred prior to 1995.

UNITB is excluded from these regressions because there were no unit banking states over this subperiod.

For large metropolitan markets, the variable NEWLIB takes the value 1 for only one of 303 market years, yielding a very large standard error.

The variable INTST takes the value 1 in almost all of the markets including all the small metropolitan markets, yielding an exclusion of this variable from one regression and large standard errors in the other two regressions.

Table 9a Grouped Logit Estimation of C&i Lending Full Sample Period 1980-1998

Smalt Banks of Young Smai! Banks Adolescent Smali Banks Mature Small Banks All Ages (1-5 years) (6-20 years) (21+ years)

Parameter Standard Parameter Standard Parameter Standard Parameter Standard Variable Estimate Error Estimate Error Estimate Error Estimate Error INTERCEP -2.5502 0.1107 -2,2263 0.3897 “1.4334 0.2605 -2.8746 0.1305 WAM-MERGE 0.0532 0.0175 -0.2943 0.0600 0.0511 0.0365 0.1630 0.0225 WAM-ACQUIS -0.2173 0.0350 -0.0215 0.1287 -0.2060 ~ 0.0848 -0.1668 0.0404 WAM-HERF 0.1566 ~ 0.0232 0.4549” 0.1020 0.1549 ~ 0.0609 0.0996 ~ 0.0257 WAM-SHAREL -0.0116 0.0112 0.1137" 0.0580 0.2261 ~ 0.0332 -0.0546 ~~ 0.0119 WAM-SHAREC 0.0922 0.0094 -0.0860 ” 0.0338 0.0774 0.0207 0.0912 0.0116 BANK-INMSA 0.0081 0.0114 0.0888 0.0540 0.1299” 0.0286 -0.0229 ° 0.0124 UNITB 0.1162 ~ 0.0060 -0.0177 0.0225 0.0767 ~ 0.0154 0.2166 ~ 0.0069 LIMITB 0.0000 0.0044 -0.0043 0.0162 0.0287 ~ 0.0113 0.0452 ~ 0:0051 NEWLIB 0.0209 0.0072 0.0944 0.0258 -0.0375 ” 0.0181 0.0576 0.0083 INTST -0.1600 0.0055 -0.1931 0.0215 0.2127" 0.0146 -0.1400 0.0061 STINCOME 1,0925 0.0637 0.3271 0.2798 0.4603 ~ 0.1656 1.2562 ~ 0.0700 WAM-GROW 0.0495 ~ 0.0122 -0.0272 0.0433 0.0039 0.0248 0.1038 0.0152 WAM-PFRAT 0.5325 ~ 0.0251 0.0018 0.0914 0.1858 ~ 0.0617 0.7562 ~~ 0.0301 WAM-LNDEP -0.0021 0.0037 0.0285 ~ 0.0112 -0.0530 0.0079 -0.0030 0.0049 WAM-MDEP1 0.0671 0.0139 0.1657 0.0450 -0.0353 0.0286 0.0842 0.0182 WAM-MDEP2 0.0304 0.0090 0.0157 0.0263 -0.0762 ~ 0.0181 0.0838 0.0122 WAM-RDEP1 -0.0393 ° 0.0227 -0.0938 0.1007 -0.0707 0.0578 0.0369 0.0277 WAM-RDEP2 -0.0227 0.0194 -0.0297 0.0761 -0.0875 ° 0.0452 0.0167 0.0242 WAM-RDEP3 0.0086 0.0173 0.0178 0.0652 -0.0429 0.0390 0.0502 ~ 0.0215 WAM-P4 0.5126 ~ 0.0753 0.1481 0.3630 0.0865 0.1994 0.6628 ~ 0.0817 WAM-P2 -0.3154 0.0339 -0.7459 0.1567 “1.10461 0.0890 -0.0350 0.0369 WAN-P3 -2.4046 0.1099 “1.3486 ~~ 0.5146 -3.0338 0.3009 -2.1288 ~ 0.1190 WAM-P4 -4.3826 ~~ 0.2078 -6.6556 0.9076 -7.1054 0.5387 -2.7556 0.2278 WAN-W1 2.8275 0.1031 5.3371 0.5119 4.8564 ~~ 0.2881 1.8916 0.1098 WAM-W2 1.8556 ~~ 0.1866 2.7246 ~~ 0.8186 4.4199 ~~ 0.4938 0.1595 0.2049 WAM-W3 0.0034 0.0004 0.0028 0.0017 0.0084 ~ 0.0011 0.0009 ~ 0.0004 WAM-ROE 0.1361” 0.0165 -0.0635 0.0677 0.4894 ~ 0.0521 0.0834 0.0178 WAN-NPL -0.4086 ~~ 0.0994 “1.2912 7 0.4199 -0.0761 0.2905 -0.6591 0.1076 WAM-EQRAT -0.6287 ~~ 0.1344 -2.1885 ~ 0.7038 -2.9073 0.3939 0.6803 0.1413 WAN-EFFIG -0.1584 "" 0.0287 -0.1732 0.1259 -0.3015 0.0705 -0.1524 0.0322 BANK-LNAGE -0.1791 0.0039 -0.1920 0.0200 0.1415" 0.0112 -0.1613 ~ 0.0043 BANK-YOUNG 0.1514” 0.0128 BANK-ADOLES 0.0588 0.0078 BANK-MERGE -0.0689 ~~ 0.0125 -0.1155 ” 0.0579 -0.2052 ~~ 0.0337 -0.0414 ~~ 0.0134 BANK-ACQUIS 0.0665 0.0078 0.1820” 0.0343 0.0956 0.0173 0.0473 0.0091 BANK-LNGTA 0.1245 ~ 0.0084 0.0779 ~ 0.0305 0.0770 ~ 0.0207 0.1464“ 0.0096 BANK-SZU10 -0.0015 0.0265 -0.1502 * 0.0834 -0,0545 0.0764 0.0439 0.0299 BANK-SZ1025 -0.0298 ~ 0.0124 -0.0967 ~ 0.0426 -0.0046 0.0302 -0.0428 ~ 0.0143 BANK-SZ2550 -0.0183 0.0065 -0.0671 0.0229 -0.0106 0.0158 -0.0227 ” 0.0074 BANK-ONEBHG 0.1245 0.0040 0.0203 0.0154 0.0858 ~~ 0.0098 0.1646 0.0045 BANK-MBHC 0.0793 0.0051 -0.1449 ~~ 0.0183 -0.0395 0.0123 0.1624 ~ 0.0059 BANK-MUL_LAY 0.0070 0.0083 0.0406 0.0437 -0.0838 “ 0.0209 0.0316 0.0091 BANK-OUTST -0.0567 ~~ 0.0101 -0.1562 0.0371 -0.0367 0.0251 -0.0467 0.0116 BANK-ROE -0.0889 “ 0.0095 -0.1237 ~ 0.0275 -0.1159 0.0193 -0.0207 0.0126 BANK-EQRAT “1.4864 0.0645 0.1189 0.1729 0.9662 ~ 0.1661 -3.4795 0.0829 BANK-NPL 0.6601 ~ 0.0624 0.8037 0.2073 0.8723 ~ 0.1521 0.6745 0.0725 BANK-SHARE -0.2218 0.0218 -0.6740 0.1793 -0.4583 0.0806 -0.2823 0.0237 Num. Obs. 164,047 11,136 27,619 125,292 Adjusted R? 0.2523 0.1602 0.1303 0.1501

* Significant at the 10% level ” Significant at the 5% level Significant at the 1% level

Table 9b Grouped Logit Estimation of C&l Lending 1995-1998 Subperiod

Small Banks of Young Small Banks Adolescent Small Banks Mature Small Banks All Ages (1-5 years) {6-20 years) (21+ years)

Parameter Standard Parameter Standard Parameter Standard Parameter Standard Variable Estimate Error Estimate Error Estimate Error Estimate Error INTERCEP -1.3959 0.3728 1.0718 1.7284 -1,7799 ~ 0.7021 -1.4507 0.3882 WAM-MERGE 0.2001 0.0403 0.1456 0.2118 0.0085 0.0861 0.3770 ~ 0.0497 WAM-ACQUIS 0.1971” 0.0959 0.8665 0.6235 0.5172" 0.2416 0.1421 0.1076 WAN-HERF 0.2068 0.0547 0.2891 0.4175 0.0928 0.1426 0.2123" 0.0592 WAM-SHAREL -0.0907 ~ 0.0277 -0.3512 0.2877 0.2755“ 0.0903 -0.1429 0.0286 WAM-SHAREG -0.0518 0.0199 -0.0354 0.1196 -0.1282 ~ 0.0458 0.0224 _ 0.0238 BANK-INMSA -0.0734 ~ 0.0237 -0.0312 0.1600 0.0138 0.0667 -0.0790 0.0252 UNITB LIMITB -0.0326 0.0096 0.0105 0.0640 -0.1353 ~ 0.0284 -0.0047 0.0102 NEWLIB -0.0913 0.0572 -0.6182 ° 0.3485 -0.1632 0.1362 -0.0344 0.0638 INTST -0.1303 0.2413 -0.3109 0.2230 STINCOME 0.4988 0.1590 0.4661 1.0952 0.4946 0.4204 0.4521 0.1715 WAM-GROW 0.0358 0.0278 0.0357 0.1635 -0.0383 0.0594 0.0842 ” 0.0329 WAM-PFRAT -0.1600 ~ 0.0773 0.0531 0.4647 0.1156 0.1976 -0.2010 ~ 0.0887 WAN-LNDEP 0.0341 0.0092 0.0070 0.0457 0.0337 * 0.0201 0.0373” 0.0114 WAM-MDEP1 0.0742 ° 0.0393 0.0005 0.1848 -0.0514 0.0830 0.1938 ~ 0.0495 WAM-MDEP2 0.0028 0.0237 -0.0519 0.1082 -0.0233 0.0480 0.0626 ~ 0.0311 WAM-RDEP1 -0.0927 ° 0.0542 -0.7307 © 0.4013 0.0198 0.1391 -0.0022 0.0652 WAM-RDEP2 -0.1006 “ 0.0460 -0.5207 ~ 0.2628 -0.0883 0.1064 -0.0239 0.0568 WAM-RODEP3 -0.0801 ~ 0.0400 -0.1626 0.2030 -0.0391 0.0864 -0.0111 0.0501 WAN-P1 0.4798 ~ 0.1877 0.3695 1.2103 -0.7676 0.5108 0.7297 0.2054 WAN-P2 -0.6048 ~~ 0.1505 -0.5142 ~ 1.1995 -0.1721 0.4321 -0.5017 ~ 0.1594 WAN-P3 -1.4607 0.4378 4.8367 4.2902 “1.7360 1.3121 -1.6996 ~~ 0.4507 WAN-P4 -6.4207 0.7807 -7.0938 5.3843 -7.7848 1.9225 -5.4039 ~~ 0.8770 WAM-W1 5.6432 0.6398 5.9737 4.6438 8.3347 ~ 1.8319 5.1396 0.6780 WAM-W2 -4.7940 7 0.7488 -8.6704 ~ 4.1808 “5.7527 2.1385 -4.1865 0.8272 WAM-W3 0.0028 ~ 0.0010 0.0015 0.0071 -0.0061 ~ 0.0026 0.0036 “” 0.0011 WAM-ROE 0.6455 0.1247 1.4315 0.9754 0.5390” 0.2885 0.7538 0.1410 WAM-NPL -0.0163 0.4327 4.8471 4.1617 1.8573 1.4789 -0.1997 0.4396 WAM-EQRAT -0.2330 0.3253 -2.9131 3.3719 -2.7489 " 0.9818 0.4486 0.3395 WAM-EFFIC -0.0321 0.0609 -0.2194 0.4287 -0.0672 0.1491 -0.0705 0.0690 BANK-LNAGE -0.1465 0.0099 -0.0919 0.0874 -0.1553 7 0.0321 -0.1337 ~~ 0.0104 BANK-YOUNG 0.0611" 0.0352 BANK-ADOLES 0.0202 0.0202 BANK-MERGE 0.0013 0.0244 0.2596 © 0.1480 -0.0248 0.0755 -0.0211 0.0255 BANK-AQUIS 0.0019 0.0206 0.0556 0.1762 0.0701 0.0534 0.0036 0.0226 BANK-LNGTA -0.0144 0.0216 -0.2088 0.1326 0.0190 0.0545 -0.0152 0.0236 BANK-SZU10 -0.2028 ” 0.0812 -0.4181 0.5113 0.0448 0.2928 -0.2432 ~~ 0.0834 BANK-SZ1025 -0.0751 ” 0.0326 -0.2426 0.1833 0.1099 0.0860 -0.1132 7 0.0354 BANK-SZ2550 -0.0228 0.0168 -0.0849 0.0951 0.0591 0.0415 -0.0527 0.0186 BANK-ONEBHC 0.1408 0.0111 0.0266 0.0593 0.1109 0.0247 0.1847 ~ 0.0133 BANK-MBHC 0.1308 0.0145 -0.1746 0.0900 0.0805 ~ 0.0378 0.1653 ~ 0.0165 BANK-MUL_LAY -0.0391 ” 0.0162 -0.1049 0.1426 -0.1683 ~~ 0.0440 -0.0102 0.0173 BANK-OUTST -0.1246 ~~ 0.0224 -0.0844 0.1301 -0.1506 0.0524 -0.1306 0.0259 BANK-ROE 0.2787 0.0535 -0.1207 0.3319 0.1376 0.0924 0.4849 ~ 0.0820 BANK-EQRAT “1.6278 0.1420 -0.6657 0.6440 0.6570 ~ 0.3274 -3.0044 0.1762 BANK-NPL 0.0184 0.1904 -1.7402 1.3419 -0.0642 0.4273 0.4883 ” 0.2184 BANK-SHARE -0.2757 0.0493 -0.1258 0.8668 0.0361 0.1845 0.3131 ~ 0.0512 Num. Obs. 19,497 765 4,020 19,497 Adjusted R? 0.1734 0.0676 0.0860 0.1068

"Significant at the 10% level “Significant at the 5% level

***Significant at the 1% level

Some missing or unusual coefficients and standard errors appear for some of the state banking regulation variables for the 1995-1998 subperiod because most of the deregulation occurred prior to 1995.

UNITB is excluded from these regressions because there were no unit banking states over this subperiod.

The variabie INTST takes the value 1 for all young and adolescent banks yielding exclusions of this variable from these regressions.