United States. Congress. House. Committee on Small.

The effects of bank consolidation on small business lending : joint hearing before the Subcommittee on Taxation and Finance and the Subcommittee on Government Programs of the Committee on Small Business, House of Representatives, One Hundred Fourth Congress, second session, Boston, MA, March 4, 1996 online

. (page 13 of 21)
Online LibraryUnited States. Congress. House. Committee on SmallThe effects of bank consolidation on small business lending : joint hearing before the Subcommittee on Taxation and Finance and the Subcommittee on Government Programs of the Committee on Small Business, House of Representatives, One Hundred Fourth Congress, second session, Boston, MA, March 4, 1996 → online text (page 13 of 21)
Font size
QR-code for this ebook

1994. The share of small business loans (less than $1 million)
relative to total business loans is similar to that of the United
States, as shown in Table 1. However, the shares for smaller
loan sizes, particularly at the smaller banks, do show larger

23-344 96-5


differences compared Co the nationwide sample. In the bottom
panel, the shares of small business loans held by New England
banks compared to the national sample again differ more the
smaller the bank size class and Che smaller the size of loan.
However, for loans of less than $1 million, the primary
difference is that New England banks with less than $100 million
in assets hold a share that is substantially lower than that for
banks nationwide, 17 percenc nacionwide buc only 7 percenc in New
England, with the offsec appearing in the $300 million to $5
billion bank asset classes.

Ameliorating the Consequences of Regional Capital Shocks

One of the major benefits of having large, well-diversified
lenders is that bank capital shocks will be ameliorated. The
costs of having geographically specialized lenders have been made
particularly clear by the recent experience in New England.
Figure 1 shows aggregate capital - to-assec ratios for commercial
banks in New England and the United States from i960: IV through
1994:11, with shading indicating recession periods. 2 From 1960
to Che mid 1970s, bank capital ratios declined steadily, showing
relatively little sensitivity to recession periods. Both series
temporarily rebounded in 1975 and 1976. The national series then
rose gradually throughout the 1980s and at an accelerated pace in
the early 1990s. In New England, however, the decline continued
until 1983, when the ratio began to rise. This rise was then
interrupted by a dramatic, buC Cemporary, decline in 1989 and


1990, followed by a sharp rise in the early 1990s.

Figure 1 also shows that the overall decline in bank capital
ratios was greater in New England than in the rest of the
country. And, unlike the national trend, a final sharp decline
preceded the most recent recession. Not only was the overall
decline sharper in New England, but the subsequent recovery
appears more dramatic. In fact, the particularly sharp initial
rise in the capital ratio in New England was accomplished in
large part through reductions in assets. Had large nationwide
banks been active in New England, the repercussions of the loss
in bank capital in the region would have been significantly
ameliorated. Thus, one of the major features of large nationwide
lenders, diversification across regions and products, should act
to ensure that regional recessions no longer affect such
institutions to the same degree, significantly softening the
regional effects on loan supply of a shock to bank capital in
that particular region.

The extent of the recent reduction in credit availability as
a result of loan supply shocks has been documented in two studies
by Peek and Rosengren (1995a, 1995b) . The first study (1995a)
finds that large New England banks shrank as a result of formal
regulatory actions, with the shrinkage generally occurring in
loans rather than securities holdings. The second study expands
the analysis to include an explicit test that is able to identify
a regulatory- induced constraint on bank loan supply, unlike
recent credit crunch studies that rely on bank capital-to-asset


ratios to proxy for supply constraints, which have difficulty
disentangling the loan supply and loan demand components of the
observed reduction in loans.

The evidence surrounding this regulatory- induced event is
inconsistent with the hypothesis that regulatory actions serve as
a proxy for loan demand shocks. In the quarters immediately
prior to a regulatory action, loans were increasing. Then, in
the precise quarter of the regulatory action, loans decreased
abruptly and then continued to decrease. These regulatory
actions occurred at individual institutions at different times
over the entire sample period. Their effects were found to be
significant, even though the estimated equation included a
variety of other variables to proxy for loan demand shocks,
including variables to capture portfolio concentrations of the
individual banks and over 100 time and location dummy variables.
In addition, these loan supply constraints were found to be
particularly important at small banks and in lending categories
likely to be dominated by borrowers dependent on bank financing.

To determine the magnitude of the effect of formal actions
on this study's sample of all New England banks, the following
regression taken from Peek and Rosengren (1995b) was reestimated:

±Hhl 3 ■ 1 *{« 1 *«,^ijM i , e *« 4 ^£i{l-M i>e )*px i , M +€ i . e (1)

"i.e-i ^i.t-l "l,c-i

The dependent variable is the change in total loans of bank i
scaled by total assets of bank i. The equation includes a dummy
variable for formal actions (FA) with a value of one for any


quarter the bank is under a formal regulacory action and zero
otherwise .

Because formal actions specify a leverage ratio, usually 6
percent, that the bank is legally required to achieve, the most
poorly capitalized banks have the greatest incentive to shrink.
Thus, the magnitude of the effect of formal actions on the change
in loans may differ across banks, in particular because it is
related to a bank's beginning-of -period (end-of -previous-period)
leverage ratio. Consequently, the coefficient on FA has been
specified to be a function of the leverage ratio, with a 3
predicted to be positive. We also have included the leverage
ratio for banks not under a formal action as an argument in the
equation to allow for the possibility that a bank would respond
by voluntarily rebuilding its capital ratio even in the absence
of a formal action. That is, this specification allows one to
distinguish between bank responses that are voluntary and those »
that are imposed by regulators. We anticipate that being below
minimum capital requirements may not in itself generate a bank
response to restore its capital position in the absence of formal
regulatory actions, implying that a 3 > c* k .

While many of the differences across banks in the demand for
loans will be ameliorated by concentrating on banks in one
geographic region, this study also includes a series of
classification variables intended to control for any remaining
differences in loan demand shocks arising from a bank's size, its
specialization in particular types of lending activities, volume


of troubled loans, and bank charter type, as well as a set of
dummy variables for each of the six New England states interacted
with a set of quarterly time dummy variables, one for each
quarter in the sample. The estimation technique is a variance
components model. For a more detailed description of the
estimation technique and variables, see Peek and Rosengren
(1995b) .

Using estimates of equation l for total loans on the sample
of FDIC- insured New England commercial and savings banks, it is
possible to calculate the total effect of the formal actions on
bank lending. Because leverage ratios with and without formal
actions have different estimated impacts, the effect of the
leverage ratio also must be incorporated in order to calculate
the net impact of formal actions on loan shrinkage. That is, it
is necessary to calculate the magnitude of the effect over and
above what would have occurred because of low leverage ratios in
the absence of formal actions. The total effect of formal
actions is thus calculated as a z + (a 3 - or,) *K/A (here, the
calculation is -1.33 + (.16 - .03)*K/A) summed over all banks
under formal actions.

Figure 2 shows the path of actual bank loans in New England
during the 1989:11 to 1994:11 period, compared to this study's
estimate of the magnitude of bank loans in the absence of formal
actions. The latter path is derived by adding to actual loans
this study's measure of the reduction in bank loans attributable
to formal actions. The figure shows that from the peak in


1989:111 to Che trough in 1993:1, loans held by New England banks
dropped by 30 percent. Of that $55 billion decline in bank
loans, 18 percent can be attributed to formal actions. The
magnitude of the decline that can be attributed to formal actions
indicates that these regulatory actions may have been an
important contributor to the credit crunch that occurred in New
England during this period.

If New England had had banking institutions that were
diversified across regions and products, the credit availability
problems would have been significantly diminished. Capital for
banks in the region would not have been as dependent on the local
economy, and many banks would have been in a position to fill the
voids created as some of the less diversified banks were forced
to shrink.

III. Potential Costs of Universal Banking

While many large banking institutions have chosen to focus
on small business lending, it remains an open question whether
large banks will find small business lending profitable in the
long run. Evidence from acquisitions in New England indicates
that small business lending portfolios frequently are not
retained by large banks that acquire smaller banks. Large banks
have been less aggressive in attracting small business loans, and
when they acquire email banks with large shares of small business
loans, these loans generally decline.


Small Business Lending in New England

Table 4 shows the changes in small business loans for FDIC-
insured banks in New England that were neither an acquirer nor a
target of an acquisition from 1993:11 to 1994:11 and neither
purchased nor sold branches during this period. For each bank
asset size category, we include the mean share of small business
loans relative to total assets in 1993:11 (Share93) , the mean
share of small business loans relative to total assets in 1994:11
(Share94) , the change over the year in the mean share of small
business loans relative to total assets (CHShare) , the mean of
the change in small business loans divided by assets
(CHSmallbus/Assets) , and the percentage change in total bank
assets (CHAssets/Assets) .

For this sample of banks unaffected by acquisitions during
this one-year window, small business loans increased at the
smallest institutions and decreased at the largest institutions.
While small business loans increased at banks with assets under
$300 million, the mean share relative to assets at these banks
declined somewhat, reflecting even faster growth of total assets.

However, in each of the three largest size categories, small
business loans decreased, both absolutely and as a share of total
assets. In the case of the five banks with assets above $3
billion, holdings of small business loans decreased despite a 15
percent increase in total assets. Thus, the smallest banks that
focus on small business lending showed increases in small
business loans, while the larger banks, some of which expanded


quite rapidly, contracted their lending to small business

Acquisitions and Small Business Lending

Table 5 provides information on the set of FDIC- insured New
England banks that acquired other FDIC- insured institutions
during the 1993:11-1994:11 period, ordered by their beginning-of -
period assets. 3 To compare shares of small business loans at the
beginning and end of the period, it is necessary to adjust the
surviving institutions for acquisitions that occurred between
1993:11 and 1994:11. To make the data comparable, we force
merged the institutions so that any bank acquired during the
window is treated as if the acquisition were consummated in
1993:11. In this way, we are able to compare data for the same
"institution" across time. For example, to compare the small
business loan shares at a given bank, we would compare Merged
Share94 with Merged Share93, not with the premerger Share93.
Thus, all changes are calculated with the forced-merged data.
However, Share93 does provide useful information by serving as a
premerger benchmark for the small business loan share of assets
at the acquiring bank. This can be interpreted as an indicator
of the longer- run desired share to which the bank may return
after digesting new acquisitions.

We also highlight several factors that would influence the
merged shares. Three institutions acquired branches or non-bank
subsidiaries during this period. However, because of data


limitations these asset acquisitions could not be incorporated
into the forced-merger procedure. Formal regulatory actions are
also highlighted Cor the reason described earlier: Formal
actions tend to retard the overall growth of institutions. The
termination of a formal action is noted as well, because its
removal may eliminate a supply constraint on that bank's lending.
We also note acquisitions that occurred in 1992 or the first half
of 1993, because the bank may still be digesting the earlier
acquisition (s) , affecting the ability of the (premerger) Share93
variable to serve as an accurate benchmark. For example, if an
acquiring bank had not yet fully readjusted its loan portfolio
from a recent acquisition of a bank with a high share of small
business loans, the premerger Share93 would overstate the bank's
longer- run desired small loan share.

Only the two smallest institutions increased their (merged)
shares of small business loans during this period. These two
banks increased their shares of small business loans both
relative to their initial shares and relative to their shares
after force merging. In each instance, the change in small
business loans was dramatic, 2.95 and 4.67 percent of assets,
respectively. And, in the case of the smallest institution, this
increase occurred even though the bank was shrinking its assets,
presumably because it was under a formal regulatory action.

Among the 11 larger acquirers, each decreased its (merged)
share of small business loans, with only three banks showing
increased holdings of small business loans. Eight of the 11


banks initially increased their share of small business loans as
a result of the acquisition, consistent with acquiring smaller
banks with a higher share of small business loans. However, of
the eight banks, by 1994:11 five had reduced their shares of
small business loans to or below their premerger 1993:11 values.
And of the three acquirers that made acquisitions that reduced
their small business loan shares, each had a share of small
business loans in 1994:11 below that prior to the acquisition.
Thus, while the smallest acquirers were aggressively increasing
their shares of email business loans, the larger acquirers did
not retain the acquired small business loan portfolios. Of the
three larger acquirers that actually increased their small
business loan holdings, each had large increases in assets, but
only a small increase in small business loans.

Table 6 shows the three FDIC- insured New England banks that
remained independent after being acquired and the 12 FDIC-insured
New England banks that had mergers announced but not consummated
during the 1993:11-1994:11 period. Of the three banks that
remained independent, only the smallest increased its small
business lending and all three banks decreased their share of
small business loans relative to assets. For acquisitions that
were announced but not consummated, three of the four smallest
institutions increased their small business loans, while the
eight largest institutions decreased their small business loans.
Only two of the 12 institutions actually increased their shares
of small business loans, with one a case of a substantial


decrease in assets overwhelming a alight decline in small
business loans.

This case study of New England banks raises issues of
whether larger institutions will retain small business loans,
unfortunately, it is impossible to distinguish whether these
decreases in small business loan shares reflect the choices of
lenders or of borrowers. On the one hand, a target bank may be
adjusting its portfolio composition to conform to the wishes of
its eventual acquirer. On the other hand, small firms value
lending relationships and perhaps the flexibility and personal
attention often less available at larger banks, and thus tend to
have, and benefit from, banking relationships at small banks
(see, for example, Petersen and Raj an 1994; Berger and Udell
1994; Elliehausen and Wolken 1990) . Consequently, the reductions
in small business loans at these banks may reflect a desire among
small business borrowers to terminate their relationship with the
target bank in order to establish a lending relationship with
another small bank. In this way, a small firm would avoid having
its primary banking relationship switched to the larger acquiring
bank that may be less able to satisfy its needs and, at the same
time, avoid the uncertainty associated with having its primary
lending relationship with a bank faced with the upheaval of an

By absorbing small bank loan portfolios into larger banks,
the borrower concentration constraint associated with the
acquired portfolio is relaxed, freeing the acquiring bank to


reallocate portfolio shares away from small loans if it so
chooses. Small banks, on the other hand, have little choice.
Because of their small capital base, their business is, for the
most part, limited to small loans. To be successful, they must
be able to exploit their information advantages to offset any
loss in economies from their small size. Given that numerous
studies have documented that small banks are generally more
profitable than large banks, apparently many small banks are able
to do this well (for example, Boyd and Graham 1991) . If the
larger acquiring bank retains the acquired small loan portfolio,
it must be because the larger bank believes it can maintain the
business line's profitability. However, to keep a small loan
portfolio viable, a large bank must hold down the fixed costs
associated with making these loans, while at the same time being
more flexible in its loan underwriting procedures than may have
previously been the case.

This case study of small business loans in New England
indicates that most acquisitions by larger banks actually result
in a shrinkage of small business loans. Thus, it appears that
most of these acquisitions are driven by reasons other than
acquiring the small business loan portfolios of smaller banks.
It is possible that acquiring core deposits, greater geographic
diversity, or potential cost savings, rather than the small
business loan portfolios and the associated business expertise
and private information about loan customers, drive these
acquisitions. Consistent with this view, Whalen (1994) argues


that mergers are, in part, driven by gains in operational
efficiencies, since he finds that intracompany mergers of
subsidiary banks by multibank holding companies result in
significant, positive abnormal stock returns. Similarly, Cornett
and Tehranian (1992) find that merged banks tend to have better
Chan average performance, perhaps associated with being better
able to attract loan and deposit customers as well as attaining
gains in efficiency.

If larger institutions choose not to maintain a significant
presence in small business lending, it becomes critical for small
business borrowers to find other small banks interested in having
small business loan customers. As consolidation continues,
borrowers in some areas may have few viable alternatives for
satisfying their email business borrowing needs. This will
require existing small business lenders to increase their market
share or de novo entry to fill the gap.

Can De Novo Banks Fill the Void?

If a void in small business lending were to occur as a
result of acquisitions by large lenders, would de novo entry be
able to fill the gap? Some guidance may be provided by examining
the experience in New England over the past decade, a period of
unusually active de novo entry. During that time, 80 new FDIC-
insured banks were created in New England. As can be seen in
Table 7, most new entries occurred during a boom period in New
England, with 56 banks started from 1985 through 1988. With the


collapse in real escace prices and Che accompanying increase in
nonper forming loans at most banks, de novo entry dramatically
declined during the subsequent years of widespread banking
problems, with only seven new entrants over the past four years.
Thus, while research has documented diminished credit
availability in New England in the early 1990s, it resulted in
relatively little entry. Instead, entry occurred during a period
characterized by rapidly expanding loan portfolios at most banks
in New England.

Although spawned by a boom period, fewer than half of the
New England banks started from 1985 through 1988 are still in
existence as independent banks, and not one has yet grown to be a
large bank. Of the 56 bank9, 20 had failed by 1994 and another
11 were acquired in non-assisted transactions (five of which were
mergers within the same holding company) , leaving only 25 of the
de novo banks started between 1985 and 1988. The average size of
these survivors remains quite small, despite their being in
existence for more than five years. One might suspect chat this
is a consequence of selection bias, perhaps with the faster-
growing de novo banks being more successful and thus becoming
attractive acquisitions for larger banks, or perhaps just the
opposite, with the faster-growing banks accomplishing their
growth chrough excessive risk-taking and consequently failing.
However, this is not apparent from the data. The assets of both
failed and acquired banks from this set of de novo banks averaged
well under $100 million.


Similarly, the average size of the more recent de novo banks
remains well below $100 million. Thus, de novo banks could fill
any void created by a credit crunch or bank consolidation only to
a limited extent, given the small number of recent new entrants
and their small average size. And the timing appears such that
entry occurs in the good times, rather than in periods when banks
are under severe pressure and, thus, more likely to be
contracting lending than expanding it.

The second panel of Table 7 provides an indication of the
areas of specialization of the surviving new entrants. The
average size of the small business loan portfolios (defined as
loans of less than $1 million) of de novo entrants operating for
at least two years averages well above the 22.6 percent of assets
for small commercial banks and significantly above the 13.9
percent for all small banks.' Thus, the de novo banks have been
targeting small business lending, although perhaps not totally by
choice, given the borrower concentration limits. In general,
these banks have also had a smaller exposure in residential
mortgages and securities than that for small commercial banks. 5

The capital-to-asset ratios show the typical pattern for de
novo bankB, high capital ratios at the outset and much lower
capital ratios as the new banks lend sufficiently to reach their
target capital ratios. Surprisingly, de novo banks created prior
to 1989 all have average capital ratios below the average for
existing small commercial banks in general. Thus, significant
further expansion of these de novo banks is unlikely in the


absence of an infusion of additional capital. Consistent with
the pattern of capital ratios, the nonperf orming loan ratios
shown in the final column indicate that the older de novo banks
have more loan problems and, again, are less likely to be in a
position to step in and increase lending to fill any void in
small business lending created by bank consolidation.

The third panel of Table 7 shows the growth in capital and
in asset categories from 1993:11 to 1994 :ll. The de novo banks
created after 1988 have grown much faster than the average for
all small commercial banks, and this growth has been particularly
rapid in the loan categories. While the growth in total assets
has been typical for those de novc banks created during the 19 87-
88 period, their growth rates of total loans and small business
loans were more than twice the all -banks average, although their
growth in total business loans was only slightly larger. Por
those banks created in the 1985-86 period, the growth in capital
was much slower than average and the growth in total assets and
small business loans somewhat slower than average. This likely

1 2 3 4 5 6 7 8 9 10 11 13 15 16 17 18 19 20 21

Online LibraryUnited States. Congress. House. Committee on SmallThe effects of bank consolidation on small business lending : joint hearing before the Subcommittee on Taxation and Finance and the Subcommittee on Government Programs of the Committee on Small Business, House of Representatives, One Hundred Fourth Congress, second session, Boston, MA, March 4, 1996 → online text (page 13 of 21)