Effects of capital regulation and information asymmetries on bank lending

MISSING IMAGE

Material Information

Title:
Effects of capital regulation and information asymmetries on bank lending
Physical Description:
v, 87 leaves : ; 29 cm.
Language:
English
Creator:
Marcus, David Frederic, 1968-
Publication Date:

Subjects

Subjects / Keywords:
Finance, Insurance, and Real Estate thesis, Ph. D
Dissertations, Academic -- Finance, Insurance, and Real Estate -- UF
Genre:
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1996.
Bibliography:
Includes bibliographical references (leaves 84-86).
Statement of Responsibility:
by David Frederic Marcus.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 023821059
oclc - 35771954
System ID:
AA00018866:00001


This item is only available as the following downloads:


Full Text









EFFECTS OF CAPITAL REGULATION AND INFORMATION
ASYMMETRIES ON BANK LENDING













By

DAVID FREDERIC MARCUS


A DISSERTATION PRESENTED TO THE GRADUATE
SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL
FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

1996





UNIVERSITY OF FL.C?.IlA IJBRARES













ACKNOWLEDGMENTS



I owe special thanks to Professors Chris James, Joel Houston, Mike Ryngaert, and Mark

Flannery for their excellent guidance and many patient discussions. I have also benefited

greatly from conversations with Professors Jon Hamilton, Carolyn Takeda, Charles

Hadlock, and Jon Garfinkel. Most of all, I owe an enormous debt to my family, and

especially my wife Deborah, for their dedication to helping me realize my goal.














TABLE OF CONTENTS


ACKNOWLEDGMENTS -------------------------------------

ABSTRACT---------------------------

CHAPTERS


1

2


INTRODUCTION-------------------------------

BACKGROUND DISCUSSION -------------

Literature on Capital Regulation and Bank Growth-----------
Internal Additions to Capital and Bank Holding Companies------


3 DATA ---------------------

4 BANK HOLDING COMPANY ANALYSIS--------------

5 BANK SUBSIDIARY ANALYSIS---------------------

6 EXTERNAL CAPITAL ISSUANCE------------

7 SUMMARY AND CONCLUSIONS -....------------------

APPENDIX ESTIMATION OF RISK-WEIGHTED ASSETS------------

REFERENCES ---------------------------

BIOGRAPHICAL SKETCH- ------ ----------------------------------


pag&

ii

iv



1

8

8
15

22

27

44

57

78

80

84

87













Abstract of Dissertation Presented to the Graduate School of the University of Florida in
Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

EFFECTS OF CAPITAL REGULATION AND INFORMATION ASYMMETRIES ON
BANK LENDING

By

David Frederic Marcus

August 1996

Chairman: Christopher M. James
Major Department: Finance, Insurance, and Real Estate

This paper provides evidence that asymmetric information problems increase the

costs of external finance for banking firms. Specifically, I find a positive and significant

relation between bank loan growth and internally generated additions to capital.

Consistent with the hypothesis that capital requirements limit bank financing choices, this

cash flow sensitivity of investment is positively related to the extent that capital

requirements are binding. I also find that the formulation of the capital ratio itself is

important in determining bank loan growth. Specifically, with regulators enforcing

leverage-based capital standards, banks can rely on a buffer stock of securities to fund

investment in a liquidity crisis. However, the use of risk-based standards substantially

reduces the effectiveness of securities as financial slack. My results also suggest that bank

holding companies establish internal capital markets to allocate scarce capital among their

subsidiaries in response to costly external finance. I find that investment by bank







subsidiaries is more sensitive to the cash flows and capitalization of its holding company

than its own cash flows and capitalization. Finally, I find that the severity of information

asymmetries affects both the likelihood of an external capital issuance and the expected

costs of issuance. I find a negative relation between the cash flow sensitivity of investment

and the probability that banks issue external capital. I also find that firms that anticipate

larger external finance costs exhibit significantly higher cash flow sensitivities of

investment.













CHAPTER 1
INTRODUCTION

Over the last fifteen years, banks operating in the United States have faced

increasingly stringent capital requirements. Concurrently, the growth rate of bank lending

has been sharply declining. Recent empirical work has questioned whether the more rigid

capital standards induced the slowdown in loan growth.' These studies rely on the critical

assumption that banks face information asymmetries which create a wedge between

internal and external financing costs. This results in capital market frictions that tie loan

growth to internally generated funds.2 Previously, scholars believed that the federal safety

net covering deposits removed financial firms from this problem. However, capital

requirements restrict the amount of (insured) debt funds that banks can utilize. As a

result, banks may benefit from holding capital greater than regulatory requirements

(surplus capital) as financial slack. This suggests that banks with more surplus capital

should invest more, and their investment should be less constrained by internally generated

funds than banks with less surplus capital. Moreover, increases in capital requirements

can prompt a decline in lending as banks replenish their surplus capital to its (new) optimal

level.



1 For an excellent review see Berger and Udell (1994) and Sharpe (1995).

2 See Fazzari, Hubbard, and Petersen (1988); Hoshi, Kashyap, and Scharfstein
(1991), and Bernanke, Gertler, and Gilchrist (1994).

1








2
A second effect of capital regulation is that the formulation of regulatory capital

ratios affects bank liquidity stocks. During the 1980s, regulators relied on leverage-based

requirements, mandating minimum capital levels per total assets. Under leverage-based

requirements, banks facing a funding shortage can support new loans through the

liquidation of other assets, namely securities, without changing required capital levels. In

this vein, security holdings may substitute for surplus capital by decreasing the dependency

on internally generated funds. However, beginning in 1990 regulators adopted risk-based

standards which require minimum capital per risk-weighted assets. Risk-weighted assets

are determined by assigning a weight based on credit risk to every bank asset. Since

securities receive a lower risk weight than loans, asset substitution could substantially

influence capital adequacy. In particular, an increase in loans increases the amount of

required bank capital even if securities are liquidated. Therefore, a change from leverage

to risk based standards may affect loan growth by reducing the role of securities as

liquidity and effectively increasing banks' desired surplus capital.

Given the existence of capital market frictions which tie investment to earnings, the

magnitude of the effect of internally generated funds may proxy for the size of the wedge

in financing costs. As a result, banks that are more constrained by internally generated

funds may be more concerned about facing a funding shortage. This suggests that these

banks may have a larger incentive to hedge their risks and therefore may rely more heavily

on derivatives and other off-balance sheet assets. Moreover, these banks may be less

likely to issue external capital and may anticipate larger costs when bringing capital issues

to the market (either in the form of underwriter fees or abnormal stock returns).








3

Most recent studies on capital regulation and loan growth focus on the adoption of

risk-based requirements and the simultaneous "credit crunch" of the early 1990s (in which

banks allegedly decreased lending in response to the new capital standards). In general,

these studies document a positive correlation between loan growth and capitalization.

Nonetheless, whether the decline in growth results from more stringent capital

requirements is not obvious. In particular, the observed relationship between loan growth

and capitalization may arise from either capital market imperfections or simply because

earnings and capitalization proxy for the profitability of lending opportunities.

An additional problem with prior studies of the relationship between capital

regulation and loan growth is that most concentrate on individual bank data. However,

the majority of banks are subsidiaries of multiple bank holding companies. If holding

companies manage capital on a consolidated basis, one would expect at best a weak link

between investment and earnings at the subsidiary level. Moreover, regulators have

recently attempted to require holding companies to inject capital into undercapitalized

subsidiaries.3 This 'source of strength' doctrine mandates that bank holding companies

must downstream capital to its subsidiary banks if they fail capital standards, as long as

this act does not cause the holding company itself to fail capital requirements. Thus, it

follows that the primary determinant of loan growth should be the capitalization and

earnings of the holding company and not the subsidiary bank.

In this paper I investigate the relationship between loan growth and internally

generated funds for a sample of 289 publicly traded bank holding companies from 1982-


3 See Keeton (1990), Fallon (1991), and Wall and Petersen (1995)










1994. Following the approach of Fazzari, Hubbard, and Petersen (1988), I assume that

capital market imperfections create a wedge between internal and external financing costs.

As a result, banks prefer to grow with internally generated funds. To control for growth

opportunities I include a measure ofTobin's Q and the bank's previous loan growth as

explanatory variables. I test the hypothesis that capital requirements increase banks' costs

of finance, and therefore I expect a negative relation between the overall sensitivity to

internally generated funds and surplus capital. In addition, I test the hypothesis that the

formulation of capital standards affects the role of securities holdings as liquidity. In

particular, I expect a negative relation between the sensitivity to internally generated funds

and securities holdings to exist while regulators impose leverage-based standards, but not

risk-based standards. I also examine the relation between the overall sensitivity to

internally generated funds and the amount of off-balance sheet assets utilized to test how

the severity of capital market imperfections affects banks' incentives to hedge risks.

To further test the existence of capital market frictions, I follow an approach

similar to Lamont (1993) and examine the relation between loan growth of individual bank

subsidiaries of multiple bank holding companies and the subsidiary's own earnings, as well

as the earnings of other subsidiaries within the same holding company.4 A finding that

subsidiary banks are more constrained by the earnings of the rest of the holding company

than its own earnings is consistent with the operation of an internal capital market. Since

holding company wide earnings may proxy for investment opportunities at the subsidiary


4 Lamont (1993) examines investment of firms with subsidiaries in both oil and
non-oil related business. His results suggest that investment of non-oil related subsidiaries
is positively related to the cash flows of the oil subsidiaries.









bank, I include both loan growth at other subsidiaries and the earnings of all non-bank

subsidiaries within the holding company as explanatory variables. As an additional test I

examine how the severity of information asymmetries at the holding company level affect

subsidiary bank dependence on both its own and holding company earnings.

An additional test on the effects of capital market frictions is to examine the

relation between the severity of information asymmetries and external capital issuances.

Following Bayless and Chaplinsky (1991, 1996), I develop a model which predicts

security issuance based on firm and market characteristics. Assuming the overall

sensitivity to internally generated funds proxies for the severity of capital market frictions,

banks that are more constrained by internally generated funds are expected to be less likely

to issue external capital. To control for a potential causality problem, I test this relation

using a lagged value of the sensitivity to internally generated funds. Because capital

deficient banks are likely to be the most constrained by internally generated funds and the

most in need of an external capital issuance, I include a dummy variable for whether banks

maintain minimum capital ratios as an explanatory variable.

Continuing along these lines, the costs associated with bringing an external capital

issue to the market should be related to the degree of information asymmetries. Following

Calomiris and Himmelberg (1995), I estimate underwriting fees associated with security

issuance after controlling for selectivity bias and examine the relation between these fees

and the sensitivity to internally generated funds. In addition, I predict the abnormal stock

returns associated with the announcement of a security issuance. A finding of a positive

relation between these costs of external finance and the sensitivity to internally generated










funds provides additional support for the hypothesis that the reliance on internal funds

proxies for the severity of capital market frictions.

Overall, I find a strong positive correlation between bank holding company loan

growth and internally generated additions to capital after controlling for differences in

growth opportunities. Consistent with the hypothesis that capital requirements increase

banks' financing costs, I find that the sensitivity of loan growth to earnings is significantly

greater among banks that are close to or below the minimum capital requirement. In

addition, I find that securities holdings provide banks with significant financial slack by

decreasing the dependency on internally generated funds. Moreover, after the

implementation of risk-based standards in 1990, loan growth is unrelated to securities

holdings, suggesting that the type of capital requirement enforced can have a significant

impact on loan growth. Also, consistent with the hypothesis that the severity of

information asymmetries influences the incentives to hedge, I find that banks that are more

constrained by internally generated funds hold more off-balance sheet assets.

In tests of the relation between loan growth and earnings at the subsidiary level, I

find that subsidiary loan growth is positively related to both its own earnings and the

earnings of other bank subsidiaries within the holding company. However, the sensitivity

of loan growth to the earnings of other subsidiaries is significantly greater than the

subsidiary's own earnings. Moreover, subsidiary loan growth is unrelated to its own

capitalization, but positively related to the holding company's capitalization. I also find

that subsidiary loan growth is negatively related to loan growth at other subsidiaries within

the holding company. This is consistent with the operation of an internal capital market in










which the overall investment capacity of the holding company is constrained. In addition,

I find that subsidiary loan growth is positively related to the earnings of non-bank

subsidiaries within the holding company. Furthermore, I find that as holding companies

become more constrained by internally generated funds, subsidiaries grow slower and

become more dependent on holding company earnings.

Finally, I examine the relation between the severity of information asymmetries, the

decision to issue external capital, and costs associated with an external capital issuance. I

find that the severity of information asymmetries affects banks' decisions to issue external

capital. In particular, my results document a negative relation between the overall

sensitivity to internally generated funds and the probability that the firm chooses to issue.

In tests of the relation between issuance costs and information asymmetries, I find that

banks that anticipate higher costs of security issuance are more constrained by internally

generated funds. Specifically, I find that the sensitivity to internally generated funds is

significantly higher for banks which expect high underwriting fees for issuing external

funds or large negative abnormal stock returns following the announcement of a security

issuance.

The remainder of the paper is organized in six chapters. Chapter 2 provides a

background discussion on the effects of capital requirements on bank investment activity.

Chapter 3 describes the data and empirical methodology. Chapter 4 documents empirical

tests for the holding company sample, while Chapter 5 documents the subsidiary bank

sample. Chapter 6 presents the external capital issuance analysis, and Chapter 7

summarizes and concludes.













CHAPTER 2
BACKGROUND DISCUSSION

Literature on Capital Regulation and Bank Growth


Bank capital regulation has changed substantially over the past fifteen years.

Before 1981, regulators relied solely on discretion when evaluating a bank's capital

adequacy. To eliminate potential bias in the examination process, regulators adopted a

minimum leverage-based capital ratio in 1981.1 A problem with leverage-based standards

is that the riskiness of bank assets is not considered when determining the minimum capital

level. Accordingly, banks had incentives to alter the riskiness of their asset portfolios. In

response, regulators created risk-based standards, which were implemented beginning in

1990.2 These new guidelines assign risk weights to all bank assets, including off-balance

sheet assets, based on credit risk.3 For example, commercial and industrial loans receive

100% risk weight, while cash and Treasury securities receive a risk weight of zero.

Some recent theoretical models have attempted to link capital regulation and bank

growth. Using an options-pricing model, Furlong and Keeley (1989) investigate how


Regulations required that banks maintain a minimum total capital ratio of 5.5%.
In 1985, regulators increased the standard to 6%.

2 Regulators imposed a minimum risk-based ratio of 7.25% beginning in 1990, and
8% beginning in 1992.
3 See Bank Holding Company Supervision Manual, section 4060, for a complete
description of risk-based capital standards.










increases in requirements affect bank risk taking behavior. Public opinion held that banks

would respond to increased requirements by increasing asset risk to offset the cost of

holding more capital. Contrary to this prediction, Furlong and Keeley show that banks

actually decrease asset risk following an increase in requirements. Although this is

excellent news for the FDIC, it has less to say about how bank growth responds, if at all,

to changes in capital requirements.

Passmore and Sharpe (1994) analyze how economic shocks and regulatory shifts

affect a profit maximizing bank. One prediction of their model is that increased capital

requirements cause a slowdown in loan growth by raising the marginal cost of funds. In

addition, they provide evidence that the resultant decline in bank lending may be most

pronounced at highly capitalized banks. This occurs because banks hold capital as a buffer

against regulatory intervention and funding shortages. One reason why banks may choose

high capital ratios is that they face high costs of capital shortages. Therefore, these firms

may react more to an increase in required capital ratios. This finding is at odds with the

popular notion that banks which are close to minimum standards would be the most

affected by an increase in capital requirements.4 This work allows for the possibility that

well capitalized institutions may be affected by increases in capital standards. In

particular, I test the hypothesis that banks rely on surplus capital (amount above

regulatory requirements) as a buffer stock against funding shortages.



4 Berger and Udell (1994) note that the commercial and industrial lending decline
in the early 1990s was not concentrated in banks with low risk-based capital ratios. In
turn, they cite this as evidence that risk-based standards were unlikely to have caused the
credit crunch.







10

Thakor (1995) develops an asymmetric information model in which banks perform

two key lending functions: pre-lending screening and post-lending monitoring. One result

of his model is that an increase in capital requirements elevates the endogenously

determined probability of a borrower being credit rationed by the entire banking system,

thereby reducing aggregate lending. This is consistent with Passmore and Sharpe and

suggests that the recent increase in capital standards may have significantly contributed to

the simultaneous decline in loan growth.

Because securities receive a lower risk-weight than loans, a change from leverage

to risk-based standards can have a significant impact on bank behavior. First, securities

require little or no capital backing, which increases their attractive relative to loans.

Consequently, banks may have gained incentive to shift assets out of loans and into

securities. Second, the role of securities as liquidity may have been diminished. With

regulators enforcing leverage-based requirements, banks could rely on a buffer stock of

securities to fund growth during a liquidity crisis. However, with risk-based standards

capital ratios decline following this type of asset substitution. Therefore, banks may have

experienced a drain on liquidity following this change in regulatory regimes. These

problems led recent empirical studies to investigate whether the regime shift induced the

"credit crunch" of the early 1990s. To date, the literature offers mixed predictions.

Peek and Rosengren (1995) and Hancock and Wilcox (1993) examine how loan

growth differs for banks based on whether they pass or fail capital requirements. Peek and

Rosengren (1995) study growth of New England banks which underwent formal

regulatory enforcement actions between 1989 and 1993. After controlling for size, time,










and region, they find that banks under formal action grow at a significantly slower pace.

In addition, they find that these banks were growing at a faster pace than other banks in

the quarters leading up to the enforcement actions. Thus, their evidence is consistent with

regulatory intervention significantly decreasing loan growth. This suggests that if the

change to risk-based standards caused an increase in regulatory intervention, then this

change may have contributed to the credit crunch.

However, the authors provide no proof of an increase in the number of banks with

enforcement actions based on the change to risk-based standards. In fact, the large

number of institutions under formal action may reflect the regional economic difficulties

during their sample period and not changes in capital regulations. Also, they do not

include a control period to compare the relationship between enforcement actions and

growth. Although this does not take away from their insight that in the 1990s regulatory

intervention has substantial effects (for New England banks), it does limit their ability to

test whether the change to risk-based standards had any impact on loan growth.

Hancock and Wilcox (1993) study the cross-sectional determinants of bank loan

and security growth during 1990 and 1991 based on whether banks experienced a "capital

shortfall" during their sample period. Capital shortfall is defined as the difference between

the actual 1990 year end capital and the regulatory minimum based on the 1990 beginning

of year total assets and a 5% capital requirement.5 In short, their findings reveal that a

capital shortfall has a large negative cross-sectional effect on loan growth in 1990.


5 The authors reasoned that banks could largely anticipate the amount of capital
that would be on the books at year end, and that this is the figure that should influence
growth.










A potential problem with this study is the authors' treatment of capital shortfall.

They assume that a 5% minimum leverage-based standard (introduced in 1981) remains in

effect through 1991. However, capital standards have been increased twice since 1981.6

Hence, the authors significantly understate the number of banks experiencing a capital

shortfall, and actually focus on only the most capital deficient banks. Therefore,

ascertaining exactly how a capital shortfall affects growth is difficult given the formulation

of these tests.

Due to the possible incentive banks may have gained to shift assets out of loans

and into securities, Haubrich and Wachtel (1993) explore how risk-based standards may

influence asset portfolio choice. The authors sort banks according to risk-based ratios and

analyze subsequent changes in portfolio mix. Their results suggest that banks in lower

capital groups tend to shift toward assets with lower risk weights, and in particular away

from commercial loans and into Treasury securities. They interpret this as evidence that

risk-based standards may have partially caused the credit crunch.

Since the authors fail to provide a benchmark period for comparing bank behavior,

however, it is possible that any changes in portfolio mix observed is simply optimal

rebalancing without regard to capital requirements. Likewise, the authors do not control

for bank size, growth opportunities, loan loss provisions, or previous growth, all of which

may significantly impact asset portfolio shifts.

6 It could be argued that the change to risk-based standards did not constitute an
increase in requirements since some banks would actually find their required capital
declining due to the new standards. However, for the majority of banks, and especially the
larger banks, the change to risk-based standards can be considered an increase in
requirements.










Given the existence of capital market imperfections, loan growth may be directly

related to capitalization. In this vein, Bernanke and Lown (1991) and Berger and Udell

(1994) investigate a direct link between capitalization and growth. Bemrnanke and Lown

(1991) examine growth as a function of beginning of period capital ratios. In their

interpretation, coefficients on capitalization serve to identify short run effects over a

period during which capital might be reasonably treated as exogenous. Relying on

aggregate state level data, their findings suggest that loan growth is positively related to

capitalization. A potential problem with this study is the utilization of state level data.

This analysis is likely to drop potentially important information specific to individual

banks.

Probably the most cited study in this area, Berger and Udell (1994) use a long

panel of observations to investigate asset growth's relationship with capitalization. In

particular, they examine differences in asset expansion during the credit crunch (early

1990s) relative to earlier years, especially with regard to capitalization. They find that

loan growth during the credit crunch does not appear to be consistently more sensitive to

capitalization than it was in the 1980s. In addition, they find that the decline in lending

was not concentrated in banks with low capital ratios. The authors interpret these findings

as evidence that risk-based standards were unlikely to have been the culprit behind the

contraction in lending during the early 1990s.

A basic problem in interpreting these results is that no change in coefficient

estimates does not necessarily imply no impact from increased regulatory standards.

Specifically, if all banks respond in a similar fashion to the change in regulations, then








14
coefficient estimates may not change at all. Therefore, the finding that loan growth is not

more affected by capitalization during the 1990s is not proof that the change to risk-based

standards had no effect. Moreover, the finding of a decline in lending at banks with high

capital ratios does not imply that the change in capital standards had no effect. Recall that

Passmore and Sharpe (1994) find that reductions in lending may be more severe at banks

with high capital ratios.

In this study, I test for the effects of capital regulation on loan growth in two ways.

First, consistent with prior studies I expect a positive relation between capital and growth.

More specifically, I hypothesize that banks consider the cushion between their capital and

required capital as a type of financial slack. Thus, an external shock which depletes this

slack could induce a slowdown in growth. Second, I test for the effects of the change to

risk-based standards through the relation between securities holdings and growth. In

particular, I expect securities to be positively related to growth in the 1980s, but unrelated

to growth in the 1990s, since risk-based standards may diminish the effectiveness of

securities as a buffer stock.

To improve upon previous studies, I provide evidence of an increase in the number

of capital deficient banks following the change to risk-based standards. I employ a long

panel of observations to observe bank behavior over time. Moreover, I use a number of

control variables to alleviate problems associated with bank growth opportunities and size.

Finally, I calculate surplus capital, which is in the same spirit as Hancock and Wilcox's

capital shortfall. However, my measure requires banks to comply with capital

requirements exactly as defined by regulations.









Internal Additions to Capital and Bank Holding Companies


It is widely accepted that banks play an important role in mitigating information

problems. Given this role, assuming that at least some bank assets will be difficult for

outsiders to value seems logical. Even so, access to federally insured deposits and the

absence of capital requirements may insulate banks from any adverse selection problems in

raising external funds. However, limited deposit insurance combined with capital

requirements suggest that banks must raise at least some funds in markets in which

asymmetric information may create a wedge between internal and external financing

costs.7 This implies that the more constrained banks are by capital requirements, the more

sensitive their growth might be to internally generated additions to capital (earnings

available to augment regulatory capital).'

A number of studies (some cited above) examine the empirical relation between

bank growth and capitalization, and in particular investigate the sensitivity of growth to

capital shocks. However, except work by Baer and McElravey (1993), studies do not

explicitly examine the sensitivity of loan growth to internal additions to capital. Since an

underlying assumption for capital shocks to adversely affect loan growth is capital market

friction which creates a wedge between internal and external finance costs, loan growth is

expected to be positively related to internal additions to capital.



7 See Myers and Majluf(1984), and Fazzari, Hubbard, and Petersen (1988).

8 Due to the way in which capital requirements are calculated, internal additions to
capital differs slightly from internally generated cash flows for non-financial firms.
Chapter 3 discusses the differences in detail.








16
Baer and McElravey (1993) investigate the relation between growth and internally

generated capital, and specifically address the issue of whether banks manage their assets

as if external finance is costly. Moreover, they examine how changes in capital

requirements might influence growth. Their results indicate that banks manage assets as if

there are significant costs with issuing new equity, or in other words, internally generated

capital strongly influences growth. They also find that growth explained by regulatory

capital increased dramatically following the introduction of specific minimum capital

standards in 1981. This suggests that banks view capital requirements as important, and

that increases in standards may have significant negative effects on growth.

A problem with their methodology is that the authors do not explicitly include

market or bank level economic control variables, such as Tobin's Q, previous growth, or

loan loss provisions. In addition their measure of internally generated capital is after

deductions for dividend payments and loan loss provisions. Since these are both

endogenous choice variables management has in its control, the inclusion of these items in

regressions may lead to misleading results. The authors also do not mention liquidity,

specifically securities holdings, playing a role in their investment model. This is surprising

since similar studies for non-financial firms generally recognize firm liquidity as an

important determinant of growth. Although it is conceivable that the availability of

insured deposit financing obviates the need to worry about liquidity, the existence of

capital requirements limits the amount of deposit financing allowable. As a result, liquidity

should be an important contributor to investment.








17

Most prior studies on bank growth and capitalization rely on bank subsidiary data.

However, if asymmetric information problems create capital market frictions, for most of

banks these frictions will occur at the holding company level. This is because usually the

parent company and not the subsidiary accesses the capital market. By definition, a bank

holding company is any organization which owns or controls at least 25 percent of any

class of voting stock of a commercial bank.9 Since the 1970s, bank holding companies

have dominated bank ownership, holding more than 90 percent of all commercial bank

assets in the United States in 1993. While the formulation of holding companies allows

banks to circumvent branching restrictions and other regulations imposed on individual

banks, the operation of a holding company also provides a mechanism for consolidating

the management and funding operations of individual subsidiary banks.

In the absence of regulations restricting bank holding companies from managing

capital on a consolidated basis, loan growth would be sensitive to the internally generated

capital of the entire holding company. Moreover, subsidiary bank loan growth would be

related primarily to the capitalization and earnings of the holding company, and not its

own capitalization and earnings. However, some restrictions on inter-company transfers

may potentially weaken the relation between subsidiary growth and holding company

earnings. For example, if holding companies are restricted in their ability to upstream

capital from subsidiaries, then each subsidiary's loan growth should partially depend on its

own earnings.


9 The 1970 Amendments to the Bank Holding Company Act of 1956 provide a
definition of a bank holding company and establishes limits on the activities in which
holding companies may engage.










One restriction, in particular, is the requirement that all subsidiaries plus the

holding company must individually maintain minimum capital ratios. This "building block"

approach implies that failure of any subsidiary to meet capital standards will impede the

holding company's ability to manage capital on a consolidated basis.10 A second

restriction is the Federal Reserve policy of viewing the holding company as a "source of

strength" to individual subsidiaries. This creates an obligation for the holding company to

downstream capital to inadequately capitalized subsidiaries. As a result, holding

companies may not be able to allocate capital to subsidiaries with positive NPV projects.

Finally, sections 23A and 23B of the Federal Reserve Act place restrictions on inter-

company transfers. Specifically, dividends, fees, and intercompany asset sales are

restricted to transfers of less than 10 percent of the bank's capital." Again, these

restrictions limit the ability of the holding company to allocate capital on a consolidated

basis.

I analyze the effects of capital market imperfections on the sensitivity of bank

investment, at both the holding company and subsidiary level, to internally generated

additions to capital. I assume that loan growth (net of loan losses) is the banking

equivalent of investment by non-financial firms.12 Given the existence of capital market



0 See the Bank Holding Company Supervision Manual, sections 2010 and 4060.2.

n See section 2020.1 of the Bank Holding Company Supervision Manual.

12 Bank investment in real assets is less than 3 percent of total assets. Arguably,
investment should consider securities. However, one motive for bank investment in
securities is liquidity. I control for securities holdings as a form of bank liquidity when
analyzing loan growth.










imperfections which create a wedge between internal and external finance, one would

expect a positive relation between holding company loan growth and internally generated

funds. Moreover, since capital requirements limit a bank's ability to substitute deposits for

equity, I expect the sensitivity of loan growth to internally generated funds (investment-

cashflow sensitivity) to be greatest for firms where the capital requirement is most binding.

I also examine how the nature of enforced capital requirements affects bank

investment-cashflow sensitivities. In particular, with regulators mandating leverage-based

capital standards, securities holdings may substitute for surplus capital as financial slack.

This is because banks can fund growth through the liquidation of securities without

changing required capital levels. Therefore, I expect the investment-cashflow sensitivities

to be decreasing the in amount of securities (relative to assets) that banks hold on their

balance sheets during the 1980s. However, beginning with the introduction of risk-based

standards in 1990, securities holdings may no longer be as efficient at providing financial

slack, and as a result I expect the investment-cashflow sensitivities to be unrelated to

securities holdings in the 1990s.

A common criticism of studies of the cash flow sensitivity of investment is that

current cash flow may be correlated with the profitability of investment opportunities. As

a result, even without capital market imperfections, investment may be positively related

to cash flows. I address this issue by including in the analysis the bank's market to book

value of assets as a measure ofTobin's Q. In addition, I include the bank's previous

growth as a second proxy for growth opportunities. I expect a positive relation between

loan growth and both the market to book ratio and lagged loan growth.








20

An additional check on the existence of capital market frictions is to examine the

operation of the internal capital market within a holding company. Specifically, if a

positive correlation between cash flows and loan demand drives the relation between cash

flows and loan growth, then subsidiary loan growth should be positively related to

subsidiary cash flows. Moreover, holding company cash flows (net of the subsidiary's

cash flows) will be related to loan growth at the subsidiary level only to the extent that

they proxy for local demand characteristics. It is likely that holding company cash flows

are a poorer proxy for local demand then subsidiary cash flows. Thus, in the absence of

capital market imperfections, they would be expected to be less important than the

subsidiary's own cash flows. Hence, a finding of holding company cash flows being more

important than subsidiary cash flows would be consistent with the hypothesis of costly

external capital.

Furthermore, a finding of subsidiary loan growth being positively related to the

cash flows of non-bank subsidiaries within the holding company can be interpreted as

strong evidence that external finance is costly and holding companies operate and internal

capital market. Indeed, it seems unlikely that in absence of costly external finance,

subsidiary bank loan growth would be at all related to non-bank cash flows of the holding

company since arguably these non-bank cash flows are less likely to proxy for local loan

demand.

As a final test on the existence of capital market frictions, I examine the

information asymmetry surrounding external capital issuances. If the magnitude of the

investment-cashflow sensitivity proxies for the severity of capital market frictions that









banks face, banks with a large investment-cashflow sensitivity are expected to be less

willing to issue external capital. I develop a logit model which predicts banks' decision to

issue external capital based on firm and market characteristics. A finding of a negative

relation between the probability of issuance and the investment-cashflow sensitivity

provides evidence that the investment-cashflow sensitivity proxies for the severity of

information asymmetries.

The severity of information asymmetries may also be related to the expected costs

of bringing an external capital issue to the market. To test this hypothesis, I estimate the

relation between investment-cashflow sensitivities and costs associated with security

issuance. Following Calomiris and Himmelberg (1995), I estimate anticipated

underwriting fees based on firm characteristics, after controlling for selectivity bias. In a

likewise fashion, I estimate expected abnormal stock returns associated with the

announcement of a security issuance. A finding that banks that anticipate larger

underwriting fees or more negative abnormal stock returns have higher investment-

cashflow sensitivities would provide additional support for the hypothesis that expected

external finance costs affect banks' dependence on internally generated funds.













CHAPTER 3
DATA

I collect bank holding company data from the Federal Reserve Y-9 tapes from

1982-1994 (annual observations). Banks included in the sample are required to have a

minimum of two years of data, a non-negative book value of equity, and an available

market value of common equity. All stock price data come from the CRSP and NASD

master tapes. The final holding company sample contains 289 banks and 2229

observations.

Subsidiary bank data are collected from the Federal Reserve Reports of Income

and Condition (Call Reports). Call report data is only available from 1985-1989.

Subsidiary banks are required to have at least two year-end observations and be part of a

multiple bank holding company. I restrict the sample to multi-bank holding companies

because I am interested in examining whether holding companies act as an internal capital

market. In addition, the subsidiaries must be part of a holding company included in the

holding company sample described above. The subsidiary bank sample contains 2339

different bank subsidiaries of 215 holding companies yielding 7023 observations.

Studies of investment spending for nonfinancial firms consider investment to be a

function of internally generated funds after controlling for firm growth opportunities (see

for example Fazzari, et. al. (1988)). Typically, investment is considered changes in

property, plant, and equipment, deflated by the firm's capital stock at the beginning of the










period. Capital stock is usually proxied by property, plant, and equipment. In addition,

the existing literature generally deflates internally generated funds by the capital stock. I

consider bank investment to be the change in loans outstanding, and the capital stock to be

the beginning of period loans outstanding. Therefore, investment (loan growth) is defined

as the percentage change in loans outstanding.

The appropriate measure of internally generated funds for banking firms differs

slightly from the measure used in studies of nonfinancial firms. Specifically, studies of

nonfinancial firms generally measure internally generated cash flows as net income before

extraordinary items plus depreciation. However, banks may not be as constrained by cash

flow as nonfinancial firms because of the availability of insured deposit financing.

Nevertheless, they are constrained by the amount of debt financing they can utilize.

Regulations mandate capital requirements which limit banks' ability to borrow, and thus

banks should be concerned with the amount of regulatory capital that they generate. I

measure internally generated funds as net income before extraordinary items plus

depreciation and additions to loan loss provisions (since loan loss provisions are a non-

cash expense and are included in regulatory capital), and I scale this measure by the

beginning of period loan balance.' To control for differences in investment opportunities,

I use the holding company's market to book ratio ( a proxy for Tobin's Q) at the end of

the prior year and the bank's previous period loan growth. Furthermore, I include the log

of assets as a control for economies of scale.


SResults are similar if I do not include additions to loan losses as part of internally
generated funds. The results are also similar if I deduct dividend payments from internally
generated funds. See Chapter 4 for more detail.








24

To determine the effect of capitalization on loan growth, I estimate surplus capital

for all banks, both holding companies and subsidiaries. Surplus capital is defined as the

bank's beginning of year capital ratio minus the end of year required ratio.2 I choose the

end of year required ratio because this requirement is always at least as strict as the

current requirement. This assumes that banks have perfect foresight regarding short term

capital requirements, earnings and growth. Tests rely on the total or Tier II capital ratio.3

The Tier II ratio is chosen because regulations currently allow banks to pass the Tier I

ratio and yet fail the Tier II ratio, but not the reverse. Specifically, the secondary portion

of Tier II capital (loan losses and subordinated debt) is restricted to be no greater than the

primary portion of capital. In addition, current regulations require 4 percent Tier I and 8

percent Tier II capital ratios. Hence it is obvious that banks which pass the Tier II

requirement must by definition have passed the Tier I requirement.

Surplus capital provides an indication of financial slack, i.e., the cushion banks

have in their capital ratios (similar to the cash and liquid assets measure used in studies of

nonfinancial firms). I also include a dummy variable, BIND, which equals one if capital

surplus is non-positive, and zero otherwise. This variable indicates whether a bank failed

to meet the minimum capital requirement in any given year.



2 Data for risk-based capital ratios are not available. Therefore, I rely on a
methodology presented by Takeda (1994) to estimate risk-based ratios. See Appendix for
a description of this methodology.
3 Tier II capital is defined as total equity plus subordinated notes plus the
allowance for loan losses all scaled by assets (either total or risk-weighted). Total equity
includes both common and preferred equity. Tests were also performed using the primary
or Tier I ratio (simply total equity over assets), with similar results.









Required capital ratios have varied over time. From 1981-1989, regulators

enforced leverage-based capital ratios which they define as total equity plus subordinated

notes plus the allowance for loan losses, all divided by total assets. Required ratios were

5.5 percent from 1981-1984 and 6 percent from 1985-1989. In the 1990s, risk-based

capital ratios became enforced, with the only change in the calculation of the capital ratio

being the substitution of risk-weighted assets for total assets.4 Required ratios were 7.25

percent from 1990-1991, and 8 percent from 1992-1994.

In addition, I study announcements of all external security offerings which

augment regulatory capital (except for initial public offerings) by bank holding companies

that were publicly traded in the United States from 1982-1994. These offerings consist of

common stock, preferred stock, or subordinated notes. The initial sample of issuances

was collected from the Investment Dealer's Digest (IDD). I searched Dow Jones News

Retrieval for the announcement of these issuances and used these dates as the initial

announcement date. If no mention of the offering was found on Dow Jones News

Retrieval, I used the registration date listed in the IDD. The final sample contains 461

security offerings by 157 different bank holding companies.

To calculate abnormal returns following the announcement of a security offering, I

use the standard event study methodology (see Asquith and Mullins (1986)). All stock

return data are collected from either the CRSP or NASD data tapes. Abnormal returns for

security i on event date t are defined as:




4 See Appendix for estimation of risk-weighted assets.










AR = R,, (a,+Pfn,)


where R1t and Rnm, are the rate of return on security i and the return on the CRSP equally

weighted index on event day t respectively. The coefficients a and P3 are ordinary least

squares estimates of the intercept and slope of the market model regression. The

estimation period used for the market model comes from the period t= -100 through t= -

20 (where t=0 equals the event day).

The average abnormal return for a portfolio of N securities is:


AAR E AR1
N
AARt_ AR,t
N i=1



The test statistic, Z, for AAR, is based on the standardized abnormal return SAR, 5, has a

unit-normal distribution, and is calculated as:

N
Z--E SAR,, I N
i=1




5 Where SAR. = AR,, /Si,,


[S2S[1+I+ (R Mt--R)2 1/2
80 80
E (R.k-R)2
k=l

and Si is the residual standardized error from the market model regression, Rl the return
on the market portfolio for the kth day of the estimation period, and Rl the average return
of the market portfolio for the estimation period.













CHAPTER 4
BANK HOLDING COMPANY ANALYSIS

Table 1 provides descriptive statistics for the bank holding company sample. The

holding companies in my sample are relatively large, with median assets of over $2.6

billion during the entire sample period. As expected, loans make up the majority of bank

assets, with more than 62 percent of aggregate bank assets allocated to loans for the full

sample. In addition, securities holdings make up a large portion of bank assets,

comprising more than 14 percent of aggregate bank assets. For the full sample, the

median bank holding company's Tier II capital ratios exceeded the regulatory minimum by

approximately two percentage points. Furthermore, only about 6 percent of banks failed

to meet minimum capital standards.

Loan growth at the holding company level averaged about 6 percent a year.

Internal additions to capital for the average and median bank was approximately 1.5

percent of loans per year. Given capital requirements less than 8 percent, internal

additions to capital appear, on average, to be sufficient to support the observed asset and

loan growth.

One purpose of this paper is to examine possible effects of changes in capital

regulation on loan growth. From 1982-1994, capital regulation can be classified into three

regimes. The first, from 1982-1984, marks the introduction of minimum leverage-based

capital standards. Coincidentally, this period also corresponds with the announcement of











Table 1
Descriptive statistics (means, with medians in parentheses) for 289 publicly traded bank
holding companies.'


variable full sample 1982-1984 1985-1989 1990-1994

Total Assets (millions) 10,300 7,911 9,260 12,700
______________ (2,621) (2,074) (2,545) (3,619)

Loan Growth b 0.062 0.109 0.078 0.020 *
_______(0.076) (0.121) (0.090) (0.029) *

Internal Additions to 0.014 0.017 0.014 0.013
Capital/ Loans.,' (0.016) (0.017) (0.016) (0.016)

Market /Book Assetsd 1.005 0.986 1.010 1.010
(0.999) (0.985) (1.004) (1.003)
Book Capital in Excess 0.024 0.019 0.020 0.031 *
of Requirement / Assets (0.020) (0.016) (0.018) (0.028) *

Percentage with Capital 5.65% 8.12% 3.83%* 6.41%*
less than requirement.

Aggregate Industry 62.28% 60.14% 64.42% 61.25%
Loans / Assets

Aggregate Industry 14.62% 11.29% 13.12% 16.87%
Securities / Assets

Number of Observations 2229 431 940 858


a. Data are from the Federal Reserve Y-9 tape.
b. Loan growth equals change in total loans outstanding divided by loans outstanding at time t- 1.
c. Internal additions to capital equals net income plus changes in loan loss provisions (up to regulatory
maximum).
d. Market to book value of assets equals (Total Assets Book Equity + Market Equity) / Total Assets. Market
Equity equals the market value of common equity from CRSP. The ratio is calculated at year end for the prior
year.
e. Book capital in excess of requirement equals the bank's book capital for regulatory minimum Tier II capital
ratio. Tier II capital equals common stock, preferred stock, plus eligible subordinated debt and loan loss
reserves. For the period 1982-1984 the requirement is 5.5% of total assets. For 1985-1989 the requirement is
6%. Beginning in 1990, the requirement is based on risk-weighted assets. For 1990-1991, the minimum is
7.25% of risk-weighted assets, while from 1992-1994 the minimum is 8%.
* mean or median significantly different from previous time period at better than the 5% level.









the 'too big to fail' policy in which certain banks were deemed too important to be

allowed to fail.1 O'Hara and Shaw (1990) document that following this announcement,

large banks experienced an increase in stock price, which they attribute to be due to the

expanded conjectural guarantees. The second regime lasts from 1985-1989, when

regulators increased the minimum capital requirement from 5.5 to 6 percent. In addition,

regulators attempted to remove the expanded conjectural guarantees implied by the 'too

big to fail' policy. Finally, the third regime starts in 1990 and begins the era of risk-based

capital standards.

Table 1 provides the descriptive statistics by the three regulatory regimes. Notice

that with each regime, loan growth has declined (both average and median). If bank

growth opportunities have also declined, this could explain the decrease in loan growth.

Market to book ratios (a proxy for Tobin's Q) have increased since the early 1980s,

indicating that overall growth opportunities have improved. However, a potential

explanation for why overall growth opportunities improved while loan growth suffered is

that off-balance sheet growth drives the increase in market to book ratios.

A second possibility for why loan growth has suffered could be a simultaneous

decline in bank internal additions to capital. While it's true that internal additions to

capital declined since the first regime, the amount of the decline hardly matches the

substantial pace of the decline in lending.



This policy implies that regulators will attempt to bail out any large institution
which is insolvent. Furthermore, since the FDIC effectively insured all bank debt (not just
small deposits) in the Continental Illinois case, management of large banks may have
reasonably assumed that the federal safety net had been expanded.








30

Changes in capital regulation could induce a slowdown in lending if banks become

inclined to increase surplus capital. First, if penalties from being undercapitalize increase,

then banks will desire a larger buffer from regulatory intervention. Second, if risk-based

capital standards reduce the effectiveness of securities as financial slack, banks will require

more surplus capital as compensation for their lost liquidity. While average and median

surplus capital do not appear to be any different in the first or second regimes, since 1990

banks have increased surplus capital by more than one percentage point. The number of

capital deficient banks increased from just less than 4 percent in the mid 1980s to more

than 6 percent in the 1990s. Moreover, from 1990 to 1992, almost 10 percent of banks

failed capital standards (not reported).

If risk-based standards increase the attractiveness of securities relative to loans,

banks may increase the proportion of securities in their portfolio. Consistent with this

hypothesis, securities holdings increased from about 13 to almost 17 percent of aggregate

bank assets, while loans fell from 64 to 61 percent of aggregate assets following the

introduction of risk-based standards in 1990. Thus the change to risk-based capital may

have caused many banks to fail capital standards, induced banks to desire a larger surplus

capital, and given banks incentives to shift assets out of loans and into securities, all of

which may have contributed to the concurrent decline in loan growth.

If capital market imperfections create a wedge between internal and external

finance costs for banks, then inadequately capitalized banks may be more likely to pass up

profitable new lending opportunities than adequately capitalized banks. To test this

hypothesis, I examine whether loan growth is related to capitalization for the banks in my








31
sample. However, loan growth and capitalization may be correlated for other reasons. In

particular, loan losses are likely to be correlated with loan demand, causing a positive

correlation between loan growth and capitalization. Therefore, I also examine the relation

between internal additions to capital and bank capitalization.

Table 2 presents the results of this analysis. The top portion of the table analyzes

differences in holding company loan growth in three different capitalization categories:

failure to meet capital requirements, capital in excess of requirements by 2 percentage

points or less, and capital in excess of requirements by greater than 2 percentage points.

Loan growth at inadequately capitalized banks was significantly less than loan growth at

either of the other two categories. In addition, there is no difference between loan growth

at either of the two adequately capitalized categories. Moreover, as shown in the bottom

portion of the table, the amount of internal capital generation increases significantly with

capitalization, suggesting that performance strongly influences capital adequacy.

That loan growth is correlated with capitalization is consistent with capital

requirements and costly external finance constraining loan growth. However, as

mentioned above the finding that internal additions to capital is correlated with

capitalization may drive this result. Since loan losses and poor performance are likely to be

correlated with weak loan demand, the positive relation between loan growth and

capitalization may reflect demand as opposed to supply characteristics. To address these

concerns, I examine the relation between loan growth and internal additions to capital.

In Table 3, I present the results of a fixed-effects regression relating loan growth

to internal additions to capital, the market to book value of assets, previous loan growth,











Table 2
Differences in loan growth and internal additions to capital based upon whether minimum
capital requirements are binding for a sample of 289 bank holding companies from 1982-
1994.


Loan Growth

Mean Median
1. Capital less than or equal to regulatory -0.032% -0.034
minimum, N=126

2. Capital greater than regulatory minimum 0.066 0.083
by 2% or less, N=968

3. Capital greater than regulatory minimum 0.069 0.077
by more than 2%, N=1143

Test statistic of difference between 1 and 2. t=7.64 z=7.75

Test statistic of difference between 1 and 3. t=7.97 z=8.01

Test statistic of difference between 2 and 3. t=0.72 z=0.08

Internal Additions to Capital

Mean Median
1. Capital less than or equal to regulatory 0.002 0.005
minimum, N= 107

2. Capital greater than regulatory minimum 0.015 0.013
by 2% or less, N=908

3. Capital greater than regulatory minimum 0.017 0.019
by more than 2%, N=931

Test statistic of difference between 1 and 2. t=7.23 z=8.95

Test statistic of difference between 1 and 3. t=9.83 z=l 1.10

Test statistic of difference between 2 and 3. t=6.92 z=10.28











Table 3
Fixed effects regressions relating loan growth to internal additions to capital, capital
requirements, and firm financial characteristics. The sample consists of 289 bank holding
companies from 1982-1994 (standard errors in parentheses).


Dependent Variable = (Loans, Loanst., ) / Loans,.,

Coefficient (1) (2)

Additions to Capital /Loans,.,a 4.771 ** 3.906 **
(0.269) (0.198)
Surplus Capital / Assetst_,b 0.824 **
(0.155)
Bind' -0.061 **
(0.009)
Surplus Capital *Additions to -27.80 **
Capital / Loanst., (4.471)

Bind Additions to Capital / Loans,., 1.006 *
~_______________~_____ ~(0.512)
Securities / Assets,., 0.138** 0.133 **
~____~____________(0.039) (0.039)

Market / Book Assets,., 0.270** 0.260**
__(0.049) (0.049)

log (Assets,.,) -0.063 ** -0.064 **
_________________ (0.005) (0.005)

Lag loan growth 0.078 ** 0.078 **
_________________ (0.009) (0.009)

R2 0.383 0.385

N (categories) 1987(289) 1987(289)

F statistic, Bank dummies 2.278 ** 2.571 **

a. Additions to capital equals net income plus changes in loan loss provisions (up to regulatory maximum).
b. Surplus capital equals actual capital less capital required to meet minimum regulatory standards.
c. Bind=l if surplus capital is less than or equal to zero, =0 otherwise.
*, ** denote significance at the 5% and 1% levels respectively.










bank size, asset composition, and two variables designed to measure the extent to which

the bank faces binding capital requirements. These two variables are surplus capital,

defined as the difference between bank capital ratios and the required ratio, and a dummy

variable that takes on the value of one if surplus capital is non-positive and zero otherwise.

Furthermore, I interact these two variables with internal additions to capital to investigate

whether the cashflow sensitivity of loan growth varies depending on whether the holding

company faces a binding capital constraint. I assume that the lower the surplus capital, the

greater the likelihood that capital requirements are binding. These are the banks that are

most likely to constrain loan growth because of high external finance costs.

As shown in Table 3, loan growth is positively related to internal additions to

capital even after controlling for differences in growth opportunities with the market to

book ratio and previous loan growth. This is consistent with external finance being costly

relative to internal finance. In addition, loan growth is positively related to surplus capital,

and is significantly lower for holding companies which fail to meet capital requirements.

Furthermore, the sensitivity of loan growth to internal additions to capital decreases as

surplus capital increases. Notice a similar result that the sensitivity is significantly higher

for capital deficient banks. These results are consistent with the hypothesis that the loan

growth of capital constrained banks is significantly more sensitive to internally generated

funds than it is for banks that maintain adequate capital. This also suggests that increases

in capital standards may influence loan growth by diminishing surplus capital for all banks.

Following an increase in capital standards the percentage change in the predicted

sensitivity of loan growth to internal additions to capital will be largest for banks that had










the largest surplus capital (lowest sensitivity) before the change. This is consistent with

Passmore and Sharpe's hypothesis that the decline in lending may be most severe at well

capitalized institutions.

Since banks may be able to fund loan growth by selling securities holdings, the

sensitivity of loan growth to internal additions to capital may also be negatively related to

securities holdings.2 On the other hand, securities may only provide this type of financial

slack to banks that are not constrained by capital requirements. That is, capital

constrained banks need to increase capital, and therefore selling securities to fund growth

may not be an option. To investigate these possibilities, Table 4 presents regression

results relating loan growth to the same measures used in Table 3, plus six additional

interaction variables. These interaction variables are designed to explain the role of

securities as financial slack. In particular, I expect that the sensitivity of loan growth to

internal additions to capital to be decreasing in the amount of securities holdings, and this

relationship to be less important for banks which fail to maintain minimum capital ratios.

Like the results presented in Table 3, Table 4 provides evidence that banks view

external finance as more expensive than internal finance as indicated by the positive

coefficient on internal additions to capital. Moreover, these results support the claim that

surplus capital is positively related to loan growth, and negatively related to the sensitivity

of loan growth to internal additions to capital. In three of the four regressions presented,

the coefficient on the interaction between internal additions to capital and securities is

2 Even with risk-based standards, this may be true since banks can sell securities
with a non-zero risk weight or that have a capital gain to fund loan growth without
penalizing capital ratios.











Table 4
Fixed effects regressions relating loan growth to internal additions to capital, capital
requirements, and firm financial characteristics. The sample consists of 289 bank holding
companies from 1982-1994 (standard errors in parentheses).


Dependent Variable = (Loans, Loans,. ) / Loans,.,
coefficient (1) (2) (3) (4)
Additions to Capital / Loans,,a 4.737** 4.446** 4.763** 4.623**
(0.316) (0.345) (0.419) (0.351)
Surplus Capital / Assets,1b 0.678 ** 0.661 *
(0.153) (0.298)
Bind -0.061 ** -0.012
(0.009) (0.019)
Surplus Capital *Additions to -18.89 ** -19.47 *
Capital /Loans,., (4.417) (8.964)
Bind Additions to Capital / 0.916 -0.388
Loans,, (0.514) (1.172)
Securities/Assets,., 0.206** 0.182** 0.204** 0.212**
(0.044) (0.046) (0.059) (0.475)
Securities Additions to -2.763* -2.619** -2.876 -3.441**
Capital / Loans,., (1.288) (1.375) (1.906) (1.401)
Surplus Capital Securities / 0.077
Assets,-, (1.329)
Bind Securities / Assets,., -0.318 **
______________________________ (0.110)
Surplus Capital Securities 1.749
Additions to Capital / Loans,., (34.88)
Bind Securities Additions 8.313
to Capital / Loans%., (6.750)
Market/Book Assets,., 0.287** 0.256** 0.287** 0.260**
________________ (0.049) (0.049) (0.049) (0.049)
log (Assets,.) -0.066** -0.064** -0.066** -0.065**
________________ (0.005) (0.005) (0.005) (0.005)
Lag loan growth 0.080 ** 0.077 ** 0.080 ** 0.076 **
________________ (0.009) (0.009) (0.009) (0.009)
R 20.373 0.386 0.373 0.389
N (categories) 1992(289) 1987(289) 1992(289) 1987(289)
F statistic, Bank dummies 2.304** 2.489** 2.301** 2.502 **

a. Additions to capital equals net income plus changes in loan loss provisions (up to regulatory maximum).
b. Surplus capital equals actual capital less capital required to meet minimum regulatory standards.
c. Bind=l if surplus capital is less than or equal to zero, =0 otherwise.
*, ** denote significance at the 5% and 1% levels respectively.










negative and significantly different from zero at the one percent level. This is consistent

with the hypothesis that securities holdings serve as a buffer stock against funding

shortages. The negative coefficient on the interaction between BIND and securities

indicates that the role of securities as financial slack is less important for capital deficient

banks. In particular, the estimated coefficient on securities holdings for capital deficient

banks (by combining the two coefficients) is not significantly different from zero. These

results suggest that both surplus capital and securities holdings serve as liquidity by

decreasing banks' dependency on internally generated additions to capital.

This paper is interested in examining the effect of changes in capital regulations on

loan growth, specifically with regard to risk-based capital standards. Specifically, with

regulators enforcing risk-based standards, the role of securities as financial slack may be

significantly diminished. To investigate further, Table 5 presents results of fixed-effects

regression models of loan growth to the same measures used in Table 3, while allowing

coefficients on key explanatory variables to change with each regulatory regime. This will

enable me to investigate if changes in regulation altered the determinants of the loan

growth equation.

Surprisingly, results indicate that in the first regime (1982-1984), capital surplus

did not influence loan growth. This is consistent with banks believing that capital

requirements were not stringently enforced. A possible explanation for this is the

expanded conjectural guarantees implied by 'too big to fail.' In regimes two (1985-1989)

and three (1990-1994), surplus capital is positively related to loan growth and negatively

related to the sensitivity of loan growth to internal additions to capital. On the other hand,











Table 5
Fixed effects regressions relating loan growth to internal additions to capital, capital
requirements, and firm financial characteristics. The sample consists of 289 bank holding
companies from 1982-1994 (standard errors in parentheses).

Dependent Variable = (Loanst Loans,1 ) /Loanst.,
coefficient (1) (2)
Additions to Capital / Loans,, 5.348 ** (0.351) 4.783 ** (0.346)
Surplus Capital /Assetst., time 0.238 (0.549)
Surplus Capital / Assets,, time 1.987 ** (0.290)
Surplus Capital / Assets,. time 0.767 ** (0.179)
bind time 1 0.026 (0.028)
bind time -0.055** (0.015)
bind time -0.084** (0.014)
Surplus Capital timel Additions to Capital/Loans -4.385 (28.87)
Surplus Capital time Additions to Capital/Loans -61.31** (13.10)
Surplus Capital time Additions to Capital/Loans -28.16 ** (5.028)
bind time Additions to Capital -0.974 (2.009)
bind time Additions to Capital 0.513 (0.782)
bind time Additions to Capital -0.149 (0.740)
Securities / Assetst., time 0.553 ** (0.875) 0.449 ** (0.079)
Securities / Assetst., time 0.187** (0.053) 0.256** (0.050)
Securities / Assets,.,1 time 0.013 (0.049) 0.036 (0.048)
Securities* time *Additionsto Capital/Loans -11.30** (4.196) -8.119** (3.235)
Securities time Additions to Capital/Loans -7.718** (1.808) -3.275** (1.620)
Securities time Additions to Capital/Loans 2.360 (0.048) -0.038 (1.451)
Market /Book Assets,, 0.231 ** (0.048) 0.223** (0.049)
log (Assets.1) -0.024** (0.007) -0.024** (0.007)
Lag loan growth 0.058** (0.010) 0.058** (0.010)
R2 ______________________________0.439 0.431
N (categories) 1990(289) 1990(289)
F statistic, Bank dummies 2.022** 2.152**

*, ** denote significance at the 5% and 1% levels respectively.









securities holdings appear to be significant contributors of financial slack up until the

1990s. Specifically, the regime three coefficients on securities holdings (and the

interaction variables) are not significantly different from zero. Moreover, these

coefficients are significantly different from previous period coefficients suggesting a real

change in the value added of securities holdings. This suggests that the change to risk-

based standards significantly altered the role of securities as liquidity. As a result, this

change in regulatory regimes may have induced a slowdown in bank growth.

The results presented in Tables 3-5 indicate a negative relation between surplus

capital and the sensitivity of loan growth to internal additions to capital. In addition, the

results demonstrate that securities holdings are also negatively related to this sensitivity, at

least before the introduction of risk-based capital standards. However, whether a firm's

overall investment-cashflow sensitivity is negatively related to surplus capital, after

controlling for the effect of securities remains a question. Table 6 presents results of the

predicted sensitivity of loan growth to internal additions to capital from regression (1) in

Table 5.3 To estimate this overall sensitivity, I simply calculate the total effect of internal

additions to capital by multiplying the coefficient estimates on the interaction terms by the

appropriate variables, and adding these products to the coefficient of additions to capital.

This allows the estimate of the investment-cashflow sensitivity to vary by bank and year.

Notice that the mean and median investment-cashflow sensitivity is largest for banks that

fail to meet minimum capital standards (significantly different from the sensitivity for the



3 Results are similar if the investment-cashflow sensitivities are estimated from
regressions in Table 4.











Table 6
Characteristics of predicted investment-cashflow sensitivity


Variable=Predicted Investment-Cashflow Sensitivity estimated from coefficients on Additions to Capital
and interaction terms from regression (1), Table 5.

Descriptive Statistics, by Capitalization mean median

full sample, N=2229 3.706 3.774

1. Capital less than or equal to regulatory minimum, N=126 4.534 4.559

2. Capital greater than regulatory minimum by 2% or less, N=960 4.003 4.019

3. Capital greater than regulatory minimum by more than 2%, N= 143 3.365 3.469

Test statistic of difference between 1 and 2. t=19.94 z=15.44

Test statistic of difference between 1 and 3. t=-39.37 z=18.20

Test statistic of difference between 2 and 3. t=36.66 z=32.05


Table 7
Fixed effects regressions relating the predicted investment-cashflow sensitivity to off-
balance sheet assets and other firm financial characteristics. Off-balance sheet assets data
are available beginning in 1991. The sample consists of 230 bank holding companies from
1991-1994 (standard errors in parentheses).

Dependent Variable=Predicted Investment-Cashlflow Sensitivity estimated from coefficients on Additions to
Capital and interaction terms from regression (1), Table 5.

Variable (1) (2)

Constant 3.261 ** 8.795 **
(0.052) (1.325)
Off Balance Sheet Assetst./ Assets,.- 0.875** 0.661**
(0.227) (0.207)
Bind 0.548 **
_____________________________ (0.059)
Log (Assetst..) -0.364 **
(0.087)
R20.097 0.220
N 651 651
F statistic, Bank dummies 13.93** 13.62**


*, ** denote significance at the 5% and 1% levels respectively.









full sample, and from adequately capitalized banks at better that the one percent level).

Moreover, the sensitivity decreases in groups based on capitalization. Therefore, even

after controlling for the effect of securities, the investment-cashflow sensitivities are

negatively related to surplus capital.

This overall investment-cashflow sensitivity may proxy for the severity of capital

market frictions. If a larger sensitivity indicates more severe information asymmetries,

banks with higher investment-cashflow sensitivities may face higher costs of external

finance. As a result these banks may have increased incentives to hedge the risks inherent

in their asset portfolios. This implies that the amount of off-balance sheet assets as a

proportion of total assets may be positively related to the investment-cashflow sensitivity.

Table 7 presents fixed-effects regression results relating the investment-cashflow

sensitivity to off-balance sheet assets. Because of reporting requirements, complete off-

balance sheet data are only available beginning in 1991. Results indicate that banks with

larger investment-cashflow sensitivities use more off-balance sheet assets, presumably to

hedge against funding shortages. These results also provide evidence that larger banks are

less constrained by internally generated funds. All together, these results support the

hypothesis that the investment-cashflow sensitivity proxies for the severity of information

asymmetries.

As a check on the robustness of the results presented above, I perform a series of

tests intended to examine the stability of the coefficient estimates. Results for these tests

are not reported, however, tables are available upon request. To control for the possibility

that results are being driven by merger and acquisition activity, I perform all above tests








42

scaling observations by year end data. Results are qualitatively similar. I also perform the

regressions limiting bank loan growth to be less than 25 percent. In addition, I choose

other various limits for loan growth, none of which make a substantial impact on the

results. As a control for possible bias due to firms not surviving the sample period, I

estimate the basic regressions presented in Table 3 for the 97 banks which survive the

entire period. Results are consistent with previous results.

One possible difference for the relation between investment and cash flow for

banking firms relative to nonfinancial firms is that a large portion of the cash flow from a

new investment may be received up front. In particular, new loan originations usually

generate immediate fee income for banks, unlike investment for industrial firms which may

take years after the initial investment before income is realized. To investigate this further,

I examine the degree of autocorrelation in additions to capital. I find that there is a strong

one period correlation of more than 50 percent. Additionally, the second and third lagged

observations are correlated at more than 10 percent, although the partial correlation

coefficient is very small after the first period. Moreover, the results of an ARMA model

using four lagged observations are consistent with a strong one period correlation, with

subsequent periods significantly related, although to a much smaller extent.

A possible interpretation of these results is that the positive relation between loan

growth and internal additions to capital is not due to firms investing when they generate

income, rather they generate income when they invest. To address this concern, I estimate

the basic regressions replacing additions to capital with a lagged observation of additions

to capital. Results are consistent with previous results.








43
Furthermore, the formulation of internal additions to capital assumes that all bank

income which augments capital is available to fund loan growth. This implies that banks

may alter dividend policy to meet investment needs. However, it is possible that banks

find dividend cuts prohibitively expensive, and therefore do not vary dividend policy. In

addition, the inclusion of loan losses may bias estimates if firms manipulate these

provisions to their advantage. Therefore, I repeat the basic regression using three

variations of internal additions to capital. Variation one is simply the original definition

minus dividend payments. Variation two is defined as net income minus dividends (no

treatment for loan loss provisions). Finally, variation three is simply net income. Results

are qualitatively similar.

Loan growth may measure investment with error. Specifically, banks can invest in

loan commitments, which generate fee income and now require capital backing. Thus, I

estimate the regressions replacing loan growth with growth in loans and loan

commitments. Because of reporting requirements, growth in loan commitments is only

available beginning in 1990. Results from a heteroskedastic consistent model (I choose not

to use fixed-effects due to the small number of observations per firm) conform with

previously reported results, and support the hypothesis that external finance is costly and

that the sensitivity of investment to internal additions to capital is negatively related to

surplus capital. Moreover, securities holdings are not related to growth in this particular

model. This provides additional support that with risk-based capital standards, the role of

securities as financial slack has been diminished (since data is only available for these tests

in the 1990s).













CHAPTER 5
BANK SUBSIDIARY ANALYSIS

The above results are consistent with capital regulation significantly affecting bank

loan growth. Moreover, these results lend credence to the argument that banks find it

more costly to raise capital externally than through internal funds. To further examine this

issue, this chapter analyzes bank subsidiary loan growth as it relates to holding company

and subsidiary characteristics. Table 8 documents the descriptive statistics for bank

subsidiaries in my sample. Loan growth averaged about 9.5 percent a year, while internal

additions to capital averaged around 1.5 percent of total loans. Furthermore, the median

subsidiary and holding company's Tier II capital ratio exceeded the regulatory minimum

by slightly less than 2 percentage points.

The bottom section of Table 8 describes differences in subsidiary loan growth

based on capitalization of both holding companies and subsidiaries. The median loan

growth of subsidiaries whose holding company was inadequately capitalized was just over

1 percent. However, median growth of subsidiaries whose holding company was

adequately capitalized was significantly greater, at over 7.5 percent. In addition, there

appears to be no difference in subsidiary loan growth based on whether the subsidiary

itself maintains adequate capital ratios. These results are consistent with bank holding

companies operating capital on a consolidated basis.











Table 8
Descriptive statistics of subsidiaries of 215 publicly traded multiple bank holding
companies from 1985-1989'


variable mean median

Subsidiary Total Assets (millions) 250 99

Internal Additions to Capital / Loans, 0.018 0.018

Internal Additions to Capital / LoansH.W 0.014 0.016

Internal Additions to Capital /LoansNo..ac 0.004 0.004

Securities / Assets 0.219 0.205

Lead Bank Assets / Holding Company Assets 0.354 0.244

(Book Capital in Excess of Requirement / Assets), 0.024 0.018

(Book Capital in Excess of Requirement / Assets)H 0.019 0.017

Subsidiary Bank Loan Growth 0.096 0.075

Subsidiary Bank Loan Growth if holding company 0.028 0.012
capital is less than regulatory minimum, N=240

Subsidiary Bank Loan Growth if holding company 0.098 0.076
capital is greater than regulatory minimum, N=7037

Test Statistic of Difference in Subsidiary Loan Growth t=5.34 z=5.48
based on holding company capitalization

Subsidiary Bank Loan Growth if its capital is less than 0.095 0.073
regulatory minimum, N=275

Subsidiary Bank Loan Growth if its capital greater 0.111 0.108
than regulatory minimum, N=7002

Test Statistic of Difference in Subsidiary Loan Growth t=0.84 z=l.6
based on its own capitalization

a. Data are from the Federal Reserve Reports of Income and Condition (Call Reports).
b. Internal Additions to Capital H-Net equals holding company additions to capital less the bank's additions to
capital divided by holding company loans less loans of the subsidiary bank.
c. Internal Additions to Capital non-bank equals holding company additions to capital net of the aggregate
additions to capital of all bank subsidiaries divided by holding company loans less loans of the subsidiary bank.








46

As described in the previous chapter, loan growth may be related to capitalization

for reasons other than costly external finance. In particular, this relationship may arise due

to bank performance or loan losses. Recall that results presented are consistent with bank

holding companies operating as if external capital is more expensive than internally

generated capital. To further examine this issue, I look at the relationship between loan

growth at individual bank subsidiaries and internal additions to capital at both subsidiary

and holding company levels. To the extent that either subsidiary or holding company

additions to capital are positively related to loan growth (after controlling for the

profitability of the holding company's lending opportunities), this provides additional

evidence that bank holding companies are liquidity constrained.

Additionally, examining this relationship may alleviate concerns that the observed

correlation between internally generated capital and loan growth at the holding company

level arises because internally generated capital proxies for the profitability of lending

opportunities not captured by Tobin's Q. Indeed, a finding that subsidiary loan growth is

related to internally generated capital at the holding company's other subsidiaries and in

particular its non-bank subsidiaries would weaken the validity of this argument. In this

regard, my tests are similar to Lamont (1993) who studied how a subsidiary's cash flows

affected investment by an unrelated subsidiary within the same firm.

Furthermore, in light of the evidence suggesting that bank holding companies are

liquidity constrained, an interesting issue is how holding companies allocate capital among

their subsidiaries. Stein (1995), following Williamson (1975), argues that capital market

imperfections may give firms incentives to establish internal capital markets to allocate









resources more efficiently. Moreover, he speculates that firms with a narrow focus and

hard to value assets may have greater incentives to create internal capital markets. A

finding that subsidiary loan growth is more strongly related to holding company additions

to capital than their own additions to capital would be consistent with bank holding

companies operating internal capital markets and that banks are liquidity constrained.

Table 9 presents regression results relating subsidiary loan growth to the same

measures use to explain holding company loan growth in Table 3. One difference in this

table is the inclusion of separate measures for the cash flows produced by the subsidiary

bank and the rest of the holding company. I also use separate measures indicating

capitalization of the subsidiary and the holding company. Regressions are performed

using a between-effects regression which pools the observations for each subsidiary, using

mean values of all variables. This controls for the autocorrelation in the residuals across

the various years for each bank. I choose between-effects rather than fixed-effects due to

the relatively short time period in which I have data (1985-1989). The sample contains

over 2000 subsidiaries with at most 5 years of data per bank. Tests were also performed

with OLS estimates (not reported) with qualitatively similar results.

Results are presented for the full sample, the sample of banks whose assets

represent less than 15 percent of their holding company's assets (bottom three size

quartiles), and the sample of lead banks in each holding company (the largest subsidiary

within the holding company). For the full sample and the small bank sample, loan growth

at subsidiary banks is positively related to the holding company's market to book ratio (a

proxy for loan opportunities). In addition, loan growth is positively related to the











Table 9
Between effects regressions relating subsidiary loan growth to internal additions to capital,
capital requirements, and subsidiary and bank holding company financial characteristics.
The sample consists of 2339 subsidiaries of 215 multiple bank holding companies from
1985-1989 (standard errors in parentheses)

Dependent Variable = (Loans, Loans,. ) / Loans,.

Overall Sample Small Bank Lead Bank
Coefficient (1) (2) (1) (2) (1) (2)
(IACH-IACQ/(Loan,-LoanB)* 1.69** 1.65** 1.94** 1.90** 0.05 0.03
__________(0.184) (0.182) (0.219) (0.218) (0.270) (0.260)
IAC / Loan, 0.29** 0.25** 0.26** 0.23** 1.86** 1.99**
__________(0.071) (0.069) (0.074) (0.071) (0.531) (0.509)
(Surplus Capital /Assets), -0.16 -0.35 0.58
(0.294) ______ (0.33) (0.774)
(Surplus Capital / Assets), -0.16 -0.17 1.25**
(0.108) (0.11) (0.430)
Bindcb -0.18** -0.18** -0.18**
___________ _____ (0.025) ______ (0.026) (0.064)
Binds 0.03 0.03 -0.16*
________________ (0.019) (0.020) (0.082)
Market / Book8 0.50** 0.47** 0.52** 0.48** 0.30 0.27
______ ___(0.077) (0.077) (0.091) (0.091) (0.181) (0.178)
(Securities / Assets), 0.03 0.02 0.02 0.01 0.09 -0.02
______ ___(0.03) (0.03) (0.03) (0.027) (0.101) (0.093)

R20.097 0.116 0.096 0.114 0.125 0.139
N (Categories) 6999 6999 6328 6328 762 762
_________(2339) (2339) (2162) (2162) (278) (278)


a. IACi = internal additions to capital (I=H for holding company and B for subsidiary)
b. Bind, = 1 if bank capital ratio is less than or equal to the regulatory minimum.
*, ** denote significance at the 5% and 1% levels, respectively








49

subsidiary's own internal additions to capital. However, loan growth is also significantly

related to the additions to capital produced by all other firms within the holding company

(measured by IACH IACB). For these samples the coefficient estimates on other

subsidiaries' cash flow are nearly eight times that of the coefficient estimate on the bank's

own cash flow. Furthermore, although it does not seem as if there is a strong link between

surplus capital and loan growth, evidence suggests that subsidiaries are less likely to lend

if their holding company (and not the subsidiary itself) is inadequately capitalized.

Results for the lead bank sample differ substantially from the other samples.

Specifically, the lead bank's own additions to capital are much more strongly correlated

with loan growth than the holding company's additions to capital. Moreover, loan growth

appears to be positively related to capitalization at the lead bank as indicated by the

positive coefficient on surplus capital and negative coefficient on BIND for the lead bank.

These results should not be surprising, given that the lead bank generates the majority of

the loans, cash flows, and capital at the disposal of the holding company. In particular, the

correlation between the cash flows of subsidiaries and the cash flows of the entire holding

company is only 0.10 for the full sample, yet 0.62 for the lead bank sample.

To further investigate the relation between capitalization and loan growth, Table

10 presents regressions for the full sample and the small bank sample relating loan growth

to additions to capital and three measures for capital adequacy. Each measure is a dummy

variable indicating whether the bank (holding company, subsidiary, or both) fails capital

standards. Results are consistent with those presented in Table 9, and suggest that the

capitalization of the holding company and not the subsidiary bank constrains loan growth.










Table 10
Between effects regressions relating subsidiary loan growth to internal additions to capital,
capital requirements, and subsidiary and holding company financial characteristics. The
sample consists of 2339 subsidiaries of 215 multiple bank holding companies from 1985-
1989 (standard errors in parentheses).

Dependent Variable = (Loans; Loans,., ) /Loans,.,

Overall Sample Small Bank
Coefficient (1) (2) (3) (1) (2) (3)
(IACH-IACB)/(LoanH-LoanB)' 1.69** 1.63** 1.71** 1.94** 1.87** 1.95**
(0.182) (0.181) (0.184) (0.217) (0.216) (0.220)
IACB /Loan, 0.29** 0.26** 0.26** 0.24** 0.23** 0.24**
(0.069) (0.069) (0.069) (0.071) (0.071) (0.072)
BindH=l and BindB=lb -0.70** -0.77**
(0.115) (0.125)
BindH -0.18** -0.18**
(0.025) (0.026)
Bind, 0.03 0.03
(0.019) (0.020)
Market / BookH 0.50** 0.47** 0.50** 0.52** 0.49** 0.52**
__________(0.077) (0.077) (0.078) (0.090) (0.090) (0.091)
(Securities / Assets)B 0.02 0.01 0.03 0.003 0.001 0.02
(0.025) (0.025) (0.026) (0.027) (0.027) (0.027)

R 0.11 0.12 0.10 0.11 0.12 0.09
N (Categories) 7023 7023 7023 6328 6328 6328
_________________ (2339) (2339) (2339) (2162) (2162) (2162)


a. IAC1 = internal additions to capital (I=H for holding company and B for subsidiary)
b. Bind, = 1 if bank capital ratio is less than or equal to the regulatory minimum.
*, ** denote significance at the 5% and 1% levels, respectively









The results presented in Tables 9 and 10 provide support for the hypothesis that

bank holding companies find external finance more costly than internally generated

finance, and in response they establish an internal capital market. This interpretation

follows from the observation that subsidiary loan growth is strongly related to holding

company additions to capital. A competing explanation is that holding company cash

flows proxy for investment opportunities at the subsidiary bank that are not captured by

the holding company's market to book ratio or in the subsidiary's own additions to capital.

To address this concern, I include loan growth at other subsidiaries, and cash flows of

non-bank subsidiaries within the same holding company as explanatory variables.

Table 11 documents results including the loan growth of the rest of the holding

company as an additional variable to explain subsidiary loan growth. I still find that loan

growth is positively related to additions to capital at both the holding company and

subsidiary levels, and that the holding company effect remains considerably larger.

Likewise, I still find that binding capital requirements matter at the holding company level,

but not the subsidiary level. Surprisingly, the estimated coefficient on the holding

company's loan growth is negatively related to subsidiary growth. Conceiving why a

negative coefficient except in the context of an internal capital market in which holding

companies allocate capital across competing uses is difficult. Overall, these results

strongly affirm the conclusion that bank holding companies establish internal capital

markets to allocate capital on a consolidated basis.'


1 The negative coefficient on other subsidiaries' growth is somewhat sensitive to
sample and specification. If I exclude banks from Texas and Oklahoma (arguably the most
constrained banks), the coefficient is positive, though not significantly different from zero.











Table 11
Between effects regressions relating subsidiary loan growth to internal additions to capital,
loan growth in related subsidiaries, capital requirements, and subsidiary and holding
company financial characteristics. The sample consists of 2339 subsidiaries of 215
multiple bank holding companies from 1985-1989 (standard errors in parentheses).


Dependent Variable = (Loans, Loans,, ) / Loans._,

Coefficient Overall Sample Small Bank Lead Banks
Subsidiaries
(IACH-IACB)/(LoanH-LoanB)' 2.37** 2.46** 0.85
(0.296) (0.335) (0.520)
IACa / LoanB 0.33** 0.32** 2.07**
(0.069) (0.071) (0.544)
BindHb -0.19** -0.019** -0.19**
_____________________ (0.026) (0.028) (0.068)
Bind, 0.03 0.03 -0.16*
(0.020) (0.021) (0.086)
Loan GrowthHc -0.05* -0.06* -0.04
(0.026) (0.029) (0.036)
Market /BookH 0.52** 0.059** 0.34
(0.082) (0.097) (0.192)
(Securities / Assets)B -0.03 -0.04 -0.04
____________________ (0.027) (0.029) (0.099)

R' 1 0.12 0.12 0.15
N (Categories) 6999 (2339) 6328(2162) 762 (278)

a. IACi = internal additions to capital (I=H for holding company and B for subsidiary)
b. Bind = 1 if bank capital ratio is less than or equal to the regulatory minimum.
c. Loan GrowthH equals loan growth of other subsidiaries in the holding company divided by their beginning of
period loans outstanding.
*, ** denote significance at the 5% and 1% levels, respectively








53
Table 12 presents results from the second robustness check, in which I replace the

holding company's additions to capital with non-bank subsidiary's additions to capital.

The results indicate that bank loan growth is positively related to the non-bank cash flows

of the holding company, which provides further evidence that external capital is costly and

banks operate internal capital markets. Moreover, the magnitude of this effect is similar to

the magnitude of the holding company cash flows. Arguably, concluding that these results

are spurious is harder, since it is less likely that non-bank subsidiary cash flows are

positively related to lending opportunities of bank subsidiaries. In this regard, these tests

provide the closest parallel to the experiment provided by Lamont (1993).

As an additional test of the operation of an internal capital market, I examine the

relation between the overall investment-cashflow sensitivity of the holding company and

the subsidiary's dependence on both the holding company's and its own cash flows. The

investment-cashflow sensitivity may estimate the magnitude of information asymmetries,

although it is measured with error since it is an approximation using regression coefficients

(see Table 6). To alleviate an errors in variables bias, I use the investment-cashflow

sensitivity lagged one period as an instrumental variable.

The severity of information asymmetries should be directly related to the growth of

individual banks. Therefore, I expect subsidiary loan growth to be negatively related to

the investment-cashflow sensitivity of the holding company. A finding that the importance

of holding company cash flows is positively related to the holding company's investment-

cashflow sensitivity would provide further evidence of the operation of an internal capital

market. This indicates that the severity of the constraint faced by the holding company











Table 12
Between effects regressions relating subsidiary loan growth to internal additions to capital,
capital requirements, and subsidiary and holding company financial characteristics. The
sample consists of 2339 subsidiaries of 215 multiple bank holding companies from 1985-
1989 (standard errors in parentheses).

Dependent Variable = (Loanst Loans., ) / Loans,._
Coefficient Overall Sample Small Banks Lead Bank
(Non-Bank IAC)/(Loan,-LoanB)* 0.56** 1.28** 0.09
(0.173) (0.279) (0.203)
IACB / LoanB 0.33** 0.28** 1.98**
(0.069) (0.072) (0.506)
Bind Hb -0.19** -0.19** -0.18**
(0.025) (0.026) (0.064)
Bind, 0.01 0.02 -0.16*
_________________ (0.019) (0.020) (0.082)
Market / BookH 0.66** 0.69** 0.28
___________________ (0.075) (0.087) (0.177)
(Securities / Assets), 0.05* 0.05 -0.02
___________________ (0.026) (0.027) (0.093)
R2 0.09 0.09 0.14
N (Categories) 6999 (2339) 6328 (2162) 762 (278)


a. IAC, = internal additions to capital (I=H for holding company and B for subsidiary)
b. Bind, = 1 if bank capital ratio is less than or equal to the regulatory minimum.
*, ** denote significance at the 5% and 1% levels, respectively









directly influences the subsidiary's dependence on holding company earnings. Table 13

presents between effects regressions including the holding company's investment-cashflow

sensitivity and two interaction variables designed to test the relation between this

sensitivity and subsidiary reliance on both its own and holding company cashflows. These

variables are the interactions between holding company investment-cashflow sensitivity (as

estimated from equation (1) Table 5) and the two cash flow measures used above.2

At first glance, results in Table 13 might appear to be confusing. In particular, the

negative and significant coefficient on the holding company's cash flows for both the

overall sample and the lead banks may initially cause some concern. However, consider

that this variable appears twice in the regression, the second time interacted with the

holding company's investment-cashflow sensitivity. This sensitivity averages just under 4

(recall from Table 6), and for the majority of banks lies in a range from 3 to 5. After

combining these two estimates, it becomes clear that the total effect of holding company

cash flows will be positive for virtually all banks. Moreover, the positive coefficient on

the interaction term indicates that the severity of the holding company's constraint directly

affects the subsidiary's dependence on holding company earnings. Specifically, as holding

companies become more cash flow constrained, subsidiaries become more reliant on

holding company earnings. Likewise, the negative and significant coefficient on the

holding company's sensitivity (for the overall sample and small banks) indicates that as

holding company's become more constrained, their subsidiaries invest less. These results

provide additional evidence that holding companies operate internal capital markets.


2 Results are similar if the sensitivities are estimated from regressions in Table 4.











Table 13
Between effects regressions relating subsidiary loan growth to internal additions to capital,
estimated investment-cashflow sensitivity, capital requirements, and subsidiary and holding
company financial characteristics. The sample consists of 2339 subsidiaries of 215
multiple bank holding companies from 1985-1989 (standard errors in parentheses).

Dependent Variable = (Loans, Loans,. ) / Loans,.1,

Coefficient Overall Sample Small Banks Lead Bank
(IACH-IACB)/(LoanH-LoanB)' -1.984 ** -0.837 -10.60 *
(2.052) (2.571) (4.276)
IACB / LoanB 0.616 0.169 11.87*
(1.328) (1.513) (6.56)
BindHb -0.145** -0.159** -0.193**
__________________ (0.022) (0.023) (0.069)
Bind, 0.026 0.031 0.177*
(0.017) (0.019) (0.087)
Market / BookH 0.344** 0.468** 0.221
(0.068) (0.089) (0.197)
(Securities / Assets), 0.032 0.007 -0.063
____________(0.026) (0.029) (0.115)
Holding Company Investment- -0.038 ** -0.033 0.002
Cashflow Sensitivity ,., (0.023) (0.013) (0.044)

(IACH-IACB)/(LoanH-LoanB)d 1.059 0.742 2.997 *
Holding Company Investment- (0.532) (0.658) (1.214)
Cashflow Sensitivity,-
IAC / Loan Holding Company -0.103 0.012 -2.448
Investment-Cashflow (0.332) (0.378) (1.652)
Sensitivity-

R' 0.155 0.136 0.169
N (Categories) 6785 (2305) 6186 (2143) 751 (274)

a. IAC1 = internal additions to capital (I=H for holding company and B for subsidiary)
b. Bind4 = 1 if bank capital ratio is less than or equal to the regulatory minimum.
c. Loan GrowthH equals loan growth of other subsidiaries in the holding company divided by their beginning of
period loans outstanding.
d. Holding Company Investment-Cashflow Sensitivity is estimated from coefficients on internal additions to
capital and interaction variables from regression (1) in Table 5. To alleviate errors-in variables bias, I use the
lag investment-cashflow sensitivity as an instrumental variable.
*, ** denote significance at the 5% and 1% levels, respectively













CHAPTER 6
EXTERNAL CAPITAL ISSUANCE

The above analysis is consistent with the hypothesis that capital market frictions tie

investment to earnings for banks. To further study this issue, this chapter examines the

severity of information asymmetries surrounding external capital issuances. Specifically, if

the magnitude of the investment-cashflow sensitivity proxies for the severity of

information asymmetries, then ceteris paribus the likelihood that banks issue external

capital should be negatively related to this sensitivity. In addition, this sensitivity may

also be related to the expected costs associated with a security issuance. Banks that face

larger information asymmetries may expect higher underwriting fees and more negative

abnormal stock returns associated with an external capital issuance than banks that face

smaller information asymmetries.

Table 14 describes a summary of 461 security issuances by 157 different bank

holding companies from 1982-1994. The timing of security issuances suggests that

increases in capital standards may have induced banks to raise external funds in order to

replenish surplus capital. Notice that the most active issuance years (1985-1987, and

1991-1993) follow increases in capital standards.

Table 14 also documents the average abnormal returns for the security issuances in

my sample. I find that the average abnormal return following the announcement of a

common stock issuance is just less than -1%. Moreover, capital deficient banks











Table 14
Summary of bank holding company security issuances from 1982-1994


year common stock preferred stock subordinated total
notes
1982 1 10 4 15

1983 4 19 4 27

1984 9 8 17 34

1985 19 6 26 51

1986 28 4 16 48

1987 7 8 35 50

1988 1 4 9 14

1989 4 6 15 25

1990 2 4 4 10

1991 16 16 18 50

1992 22 18 36 76

1993 9 6 25 40

1994 0 5 16 31

total 122 114 225 461

Mean Offer Size (millions) 95,451 123,128 135,393 124,810

Mean (Offer Size/Assets) 1.0% 0.6% 1.1% 0.9%

Mean Underwriter Fees 4.2% 2.8% 1.4% 2.7%

Mean Abnormal Return -1.32% 0.15% -0.19% -0.42%
full sample (z=-5.6, N=105) (z--0.3, N=98) (z=-0.7, N=158) (z=-3.5, N=361)

Mean Abnormal Return -1.49% -0.02% -0.16% -0.48%
if bind=0 (z=-6.4, N=85) (z-0.1, N=74) (z--0.5,N=144) (z=-3.5, N=303)

Mean Abnormal Return 0.37% 0.05% -0.05% -0.12%
if bind=l1 (z=0.1, N=20) (z=0.7, N=24) (z=-0.7, N=14) (z=-0.8. N=58)

Test Statistic of Difference
in Mean Abnormal z=1.50 z=0.39 z=1.15 z=0.86
Returns, by bind I


Security issuances collected from the Investment Dealers Digest.
Announcement dates collected from Dow Jones News Retrieval










experience approximately zero abnormal return, while adequately capitalized banks lose

approximately 1.5% in value. This is consistent with Cornett and Tehranian (1994), and

provides evidence that the market anticipates external equity issues by inadequately

capitalized banks. Also consistent with previous studies, I find that preferred stock and

subordinated note issues do not invoke significant abnormal returns. This may occur

because these securities are less informationally intensive than common stock, so an

issuance of these securities may not elicit as severe of a lemon's problem.1

A bank's decision of whether to issue is likely to be related to the information

asymmetries it faces. Therefore, I develop a logit model which predicts the choice to issue

based on firm and market characteristics. Following Bayless and Chaplinsky (1991)

certain factors can be identified which are likely to influence the decision to issue. It can

be shown that an increase in the firm's stock price increases the share of returns to an

investment project retained by old stockholders and reduces the loss of existing firm value

to new stockholders. Thus, I include the ratio of the last three months' average stock

price to the prior thirty-six months' average. In addition, prior studies find that the

aggregate market conditions at the time of issuance have a significant influence on offer

choice. Essentially, firms are more likely to issue equity following strong equity market

performance. To control for market conditions, I include the ratio of the three-month

average market price (CRSP equal weighted index) to the thirty-six month average. The

variable Risk, the standard deviation of the firm's common stock returns, controls for

stock volatility.


See Mikkelson and Partch (1986)








60
In previous chapters, I argue that capital requirements affect bank decision making

by placing a binding constraint on the utilization of debt funds. This suggests that banks

which are capital deficient may be more likely to issue external funds. To control for this

possibility, I include a dummy variable for whether banks fail capital requirements, Bind,

as an explanatory variable. A second implication of this paper is that securities holdings

may provide banks with significant financial slack, especially if regulators enforce

leverage-based capital standards. In particular, firms with sufficient resources allocated to

securities holdings can fund growth through the liquidation of these holdings. As a result,

firms with a large buffer stock of securities may be less likely to issue external funds.

Thus, I include the ratio of securities to assets as an explanatory variable.

In the Myers and Majluf (1984) model, an increase in financial slack increases the

costs of adverse selection. Therefore, and increase in slack reduces the likelihood of an

equity offering. As an additional proxy for slack, I include a variable called free cash flow

(see Bayless and Chaplinsky (1996)). Free cash flow is designed to measure the flow of

funds constraint which motivates firms to issue securities when positive new present value

projects cannot be financed internally. Free cash flow is defined as current net income

less dividends and loan growth (investment). Banks are expected to be more likely to

issue when they have less free cash flow, hence I expect a negative coefficient.

This paper maintains that banks are concerned with the amount of regulatory

capital that they generate internally. Presumably, banks would fund all growth through

internally generated additions to capital if possible. However, that a bank chooses to issue

external funds does not necessarily mean that it requires a capital injection from a








61

regulatory point of view. It could be that sufficient profitable lending opportunities exist

that the bank could not take advantage of without raising external capital. Either way, the

amount of capital generated internally is likely to be an important factor in the decision to

issue external funds. To capture this relationship, I include the additions to capital,

lagged one observation, as an explanatory variable. This variable indicates the amount of

regulatory capital generated in the previous year. A negative coefficient on additions to

capital indicates that banks are more likely to issue if they have not been generating

sufficient capital internally. On the other hand, a positive coefficient indicates that banks

are more likely to issue when they have profitable growth opportunities.

Optimal capital structure theory maintains that firms have target debt ratios.

Because the costs of debt exceed the benefits for debt ratios above the target, firms are

predicted to be more likely to issue the further the firm's current debt ratio is above the

target ratio. Banks' optimal capital structure is likely to depend on the current capital

requirements. Indeed, the previous chapters provide evidence that surplus capital is an

important determinant of bank growth. I approximate each bank's optimal capital

structure as its average surplus capital over the entire sample period (1982-1994).2 The

deviation from the optimal, designated Target, is then calculated as the difference between

this average and the bank's current period surplus capital. A positive value for Target

indicates that the bank has less surplus capital than optimal. Since all security issues in my

sample augment regulatory capital, I expect a positive coefficient on Target.

2 This definition may not be completely accurate, since the optimal surplus capital
may change as capital standards change. As a separate test, I calculate Target over each
regime. Results are similar.








62

In a recent study, Bayless and Chaplinsky (1996) find that firms are more likely to

issue equity in a 'hot' issue market. Drawing on this research, I identify hot issuance

markets, and create an appropriate dummy variable (Hot) to include as an explanatory

variable. To identify hot markets, I use aggregate equity issue volume data from the

Federal Reserve System's Annual Statistical Digest. I classify the periods using scaled

issue volume, which is monthly issue volume divided by the month end value of

outstanding equity from the CRSP and NASD tapes.3 I rank three-month moving

averages of equity issue volume into quartiles. Hot markets are identified as at least three

contiguous months where equity volume exceeds the upper quartile.

The main purpose of this analysis is to examine the relation between information

asymmetries and the likelihood of issuance. If the predicted investment-cashflow

sensitivity proxies for the severity of capital market frictions, it should be negatively

related to the probability of issuance. The investment-cashflow sensitivity is estimated

using regression (1) from Table 5.4 Specifically, the coefficients on additions to capital

and the interaction variables are combined to predict the sensitivity for each bank in every

year (see Table 6).

Two potential problems arise when including this sensitivity as an explanatory

variable. First, since it is estimated using regression coefficients, it is measured with error.

To alleviate this problem, I instrument for this sensitivity using a lagged observation.


3 Like Bayless and Chaplinsky (1996), results are not sensitive to the use of scaled
volume. If I classify based on nominal dollar or real dollar volume, results are similar.

4 Results are similar if the investment-cashflow sensitivities are estimated from
regressions in Table 4.










Second, there may be a causality problem. Banks that issue securities may be less cash

flow constrained not because of information costs but because they raised external capital.

However, it is less likely that banks that issue were also less constrained by cash flow in

the previous year because of the decision to raise funds. Again this can be alleviated by

using the investment-cashflow sensitivity lagged one year.

Table 15 presents descriptive statistics based on whether the bank issues in a given

year, and if so which type of security it issues. Recall from Table 7 that the investment-

cashflow sensitivities are negatively related to bank size. This is consistent with large

banks facing less severe information asymmetries than small banks. Thus it should not be

surprising that banks which issue are significantly larger than banks which do not issue.

The average size for non-issuers is just over six billion, while stock issuers average over

twenty billion. Also, issuers are increasing loans faster and generating more internal

capital (except for preferred stock issuers) than non-issuers. However, issuers also have

significantly less free cash flow. These results are consistent with firms issuing when they

need funds to implement positive net present value projects.

This chapter is interested in examining the relation between the investment-

cashflow sensitivities and the likelihood of security issuance. One implication of capital

market frictions is that firms with more severe information asymmetries will be less likely

to issue external funds. Perhaps surprisingly, non-issuers have significantly lower

investment-cashflow sensitivities than issuers. However, this is likely influenced by the

fact that issuers have significantly less financial slack as measured by surplus capital and

securities holdings, and are more likely to be inadequately capitalized. To further address











Table 15
Descriptive statistics (means, with medians in parentheses) for 289 bank holding
companies classified by type of security issuance.'


Variable no issuance common stock preferred stock subordinated
issuance issuance note issuance
Total Assets (millions) 6,533 20,100 51,300* 38,600 *
(2,276) (5,597) (32,300)* (22,800)*
Loan Growth' 0.061 0.191* 0.087 0.152*
(0.076) (0.151) (0.055) (0.116)*
Internal Additions to 0.014 0.019* 0.015 0.019*
Capital/Loans.,t (0.017) (0.019)* (0.014) (0018)

Market /Book Assetsd 1.005 1.004 0.991 1.006
(0.999) (0.996) (0.989)* (1.000)
Book Capital in Excess 0.025 0.017 0.017* 0.023
of Requirement /Assetse (0.021) (0.014)* (0.010)* (0.017) *

Securities / Assets 0.215 0.193* 0.134* 0.165 *
(0.208) (0.195) (0.119)* (0.159)*
Risk' 0.023 0.022 0.023 0.019
(0.018) (0.017 (0.018) (0.016)
Predicted Investment- 3.688 3.915* 4.072 3.903*
Cashflow Sensitivity., (3.758) (3.915)* (4.174)* (4.006)*

Free Cash Flow/Loans ih -0.055 -0.185* -0.082* -0.144*
(-0.066) (-0.145)* (-0.048) (-0.110)*
Percentage with Capital 4.55% 9.02%* 21.9%* 9.78%*
less than requirement.

Number of Observations 1954 122 114 225

a. Data are from the Federal Reserve Y-9 tape.
b. Loan growth equals change in total loans outstanding divided by loans outstanding at time t- 1.
c. Internal additions to capital equals net income plus changes in loan loss provisions (up to regulatory
maximum).
d. Market to book value of assets equals (Total Assets Book Equity + Market Equity) / Total Assets. Market
Equity equals the market value of common equity from CRSP. The ratio is calculated at year end for the prior
year.
e. Book capital in excess of requirement equals the bank's book capital for regulatory minimum Tier II capital
ratio. Tier II capital equals common stock, preferred stock, plus eligible subordinated debt and loan loss
reserves. For the period 1982-1984 the requirement is 5.5% of total assets. For 1985-1989 the requirement is
6%. Beginning in 1990, the requirement is based on risk-weighted assets. For 1990-1991, the minimum is
7.25% of risk-weighted assets, while from 1992-1994 the minimum is 8%.
f Risk equals the standard deviation of the firm's daily stock return, after adjusting for bid-ask bounce.
g. Investment-Cashflow Sensitivity is estimated using coefficients on internal additions to capital and
interaction variables from regression (1) in Table 5.
h. Free Cash Flow is net income less dividends and loan growth.
* denotes significantly different from the no investment sample at better than the 5% level.








65

this issue, I estimate probit regression models of security choice while controlling for the

investment-cashflow sensitivity, surplus capital, and securities holdings.5

The sample contains a pooled time-series cross section of data. As a result,

observations may not be independent, especially for the common stock issuance model. In

particular, if a bank issues common stock in any given year, it is likely to not issue again in

the next year. In fact, for only twenty out of one hundred and twenty-two issues did a

bank issue common stock in consecutive years. Likewise, forty-two out of one hundred

and fourteen preferred stock, and sixty-eight out of two hundred and twenty-five

subordinated note issues occurred from a bank in consecutive years. To address this

problem, I include a dummy variable indicating whether the bank issued in the previous

year. Specifically, I create four dummy variables (any security, common stock, preferred

stock, and subordinated note) to include in the appropriate regressions, and expect the

coefficients to be negative.

Table 16 presents the results from probit models which estimate the decision to

issue based on the aforementioned firm and market characteristics. Because the

investment-cashflow sensitivity is related to capitalization and securities holdings, I

estimate the regressions both with and without this variable. Regardless of security type,



5 Results are similar using logit regressions.

6 As a separate test, I estimate the probit models for each year independently. In
these regressions, the coefficient estimates were relatively stable for the any issuance,
preferred stock, and subordinated note models. However, for the common stock model,
coefficients were not as stable. Moreover, in all models, coefficient estimates were not as
significant as those presented in Table 15, perhaps due to the use of many fewer
observations.






Table 16
Probit regressions relating a security issuance to firm financial characteristics. Year dummy variables included (results not reported).
The sample consists of 289 bank holding companies from 1982-1994 (standard errors in parentheses).

Dependent Variable=ISS =l if issuance, =0 otherwise______
coefficient ISSany ISSmy ISS, ISS, ISS, ISSP ISS. ISS.
Investment-Cashflow Sensitivityt.,b -0.5 (0.1) -0.6 (0.2) -0.9 (0.2) -0.1 (0.2)
Issued Last Year =1 ifyes, 0 ifno. -0.4 (0.1)* -0.4 (0,1)* -0.5 (0.2)* -0.5 (0.2)* -0.7 (0.2)* -0.6 (0.2)* -0.5 (0.1)* -0.5 (0.1)*
Bind 0.5 (0.2) 0.6 (0.2) 0.5 (0.2) 0.7 (0.2) 0.5 (0.2) 0.6 (0.2) 0.3 (0.2) 0.3 (0.2)
Prices 1.0 (0.5)* 1.1 (0.5)* 0.9 (0.6) 1.1 (0.6)* 1.2 (0.7) 1.4 (0.7)* 0.4 (0.9) 0.4 (0.8)
Marketd 16 (3.8)* 17 (3.8)* 14 (4.9)* 15 (5.0)* 14 (5.9)* 17 (6.3)* 10 (4.8) 10 (4.8)*
Risk' 1.4 (3.7) 1.3 (3.8) 11 (9.2) 12 (9.8) 1.3 (4.3) 1.3 (4.6) 0.9 (4.4) 0.9 (4.4)
Risk2 -3.4 (9.2) -3.3 (10) -59 (68) -70 (76) -3.1 (6.6) -3.0 (7.7) -2.4 (5.6) -2.4 (5.6)
log (Total Assets1,) 0.4 (0.03)* 0.4 (0.04)* 0.2 (0.04)* 0.3 (0.04)* 0.4 (0.05)* 0.5 (0.1)* 0.5 (0.1)* 0.5 (0.1)*
Hot Issuance Market 0.04 (0.1) 0.03 (0.1) -0.04 (0.1) -0.04(0.1) 0.04 (0.1) 0.04 (0.1) 0.03 (0.1) 0.03 (0.1)
Free Cash Flow / Loans,1 -1.3 (0.5)* -0.7 (0.5) -0.3 (0.6)* -0.7 (0.6) -1.2 (0.6)* -0.7 (0.6) -1.2 (0.6)* -1.2 (0.6)*
Lag (Additions to Capital/ Loans,.,) 6.9 (3.5)* 4.4 (3.6) 8.5 (4.0)* 5.9 (4.1) -4.3 (4.5) -4.3 (4.6) 5.9 (4.6) 5.8 (4.8)
Target' 2.2 (2.8) 1.3 (2.8) -1.3 (3.2) -1.4 (3.1) -5.8 (3.4) -5.8 (3.4) 2.5 (3.4) 2.4 (3.4)
Securities/Assets,; -1.2 (0.5)* -1.5 (0.6)* -0.9 (0.5)* -0.7 (0.7) -1.3 (0.6)* -1.3 (0.6)* -0.3 (0.7) -0.3 (0.9)
N (Pseudo R squared) 1879 (0.22) 1879 (0.23) 1879 (0.09) 1879(0.11) 1879 (0.27) 1879 (0.30) 1879 (0.29) 1879(0.30)

a. ISS, equals a dummy variable for security issuance in a given year (any=any security, c=common stock, p=preferred stock, sn=subordinated notes).
b. Investment-Cashflow Sensitivity is estimated using coefficients on internal additions to capital and interaction variables from regression (1) in Table 5.
c. Price=3 month average stock price / 36 month average price
d. Market=3 month average value of CRSP value-weighted index / 36 month average
e. Risk= standard deviation of daily stock return
f Target = the bank's average surplus capital over the entire sample period minus it's actual surplus capital.
*denote significance at the 5% level.









inclusion of the investment-cashflow sensitivity increases the pseudo R squared of the

model (and log likelihood ratio). Moreover, in three of four models after controlling for

capitalization and securities holdings, I find a negative and significant relation between the

choice of issuance and the investment-cashflow sensitivity. This provides evidence that

investment-cashflow sensitivities proxy for the severity of information asymmetries.

Consistent with the hypothesis that capital requirements impose binding

constraints, I find that the probability of issuing external funds is greater for banks that fail

capital requirements. Specifically, for the any security, common stock, and preferred

stock models, coefficient estimates on Bind, are positive and significant after controlling

for the investment-cashflow sensitivity. Moreover, the negative coefficients on securities

indicate that banks with more securities holdings are less likely to raise external capital.

This provides further evidence that banks rely on buffer stocks of capital and securities to

fund growth during a liquidity crisis.

The coefficient estimates on the dummy variable for issuance in the previous year

are negative and statistically significant for all four issuance models. This supports the

notion that banks are unlikely to issue external capital securities in consecutive years.

Also, the negative coefficients on free cash flow indicate that banks are more likely to

issue capital when they need funds to support growth.

As a second test of the relation between information costs and external capital

issuances, I examine the relation between investment-cashflow sensitivities and banks'

anticipated costs of security issuance. Following Calomiris and Himmelberg (1995), I

estimate expected underwriting costs based on firm characteristics and sort firms into high










and low cost categories. Using these categories, I determine whether high cost firms

display greater investment-cashflow sensitivities than low cost firms. Calomiris and

Himmelberg identify certain characteristics which are important in determining

underwriting fees. I use similar bank characteristics to estimate the fee model.

In Calomiris and Himmelberg, financial working capital, leverage, and sales are the

most important characteristics. I estimate financial working capital as securities holdings,

and control for leverage with capitalization. Translating sales to a banking firm

characteristic, I choose internally generated additions to capital. I pick additions to capital

over income because capital requirements cause banks to be concerned with the amount of

regulatory capital that they generate. I also include free cash flow to capture the flow of

funds constraint that motivates banks to issue. In addition, I include size, the deviation

from optimal capital structure (Target), and the volatility of stock returns (Risk).7

As in Calomiris and Himmelberg, this is not intended to be a structural model. The

process of experimentation that yields the model was an search for a model that

maximized adjusted R squared. Because of this problem, I choose not to focus on

individual coefficient estimates, rather I am more concerned with the accuracy of

predictions based on this model (the correlation between actual fees and predicted fees for

common stock issuances is more than 0.8). Moreover, I will sort firms into categories

based on this prediction so as not to rely on the prediction itself, but only a dummy

variable which indicates whether the prediction is over or under the median prediction.



7 I experiment with a number of alternative specifications, but report only the one
which yielded the largest adjusted R squared.








69

Table 17 presents results from a heteroskedastic consistent regression relating the

underwriting fees to firm characteristics. The most important firm characteristics in

determining fees are size and the amount of internal capital generation. Also, the adjusted

R squared for these models are similar to those from Calomiris and Himmelberg.

Because I am interested in using information about issuers to estimate the external

financing costs of non-issuers, I correct for potential selectivity bias before applying the

model to construct expected underwriting costs for issuers and non-issuers. That is, the

decision to issue is not random, and that decision is likely to be correlated with some or all

of the regressors in the underwriting fee model. To correct for this problem, I use a two-

step Heckman procedure in which the decision to issue is modeled as a probit model. The

probability of issuing derived from this probit enters as an explanatory variable in the

underwriting fee model. Since a probit model has been specified in Table 16, I rely on this

model for the Heckman procedure. Because my goal is to use predictors from the

Heckman procedure to explain differences in investment-cashflow sensitivities, I use the

probit model without the investment-cashflow sensitivity as an explanatory variable.

The results from the underwriting fee model and the probit model after correcting

for selectivity bias are presented in Table 18. The coefficients from the underwriting fee

model are used to construct predicted values of the cost of issuing each type of security

for all firms. I then create a dummy variable (High Fees) which equals one if the predicted

fees are above the median predicted fees, or zero otherwise for each security type. This

dummy variable is employed to estimate the relation between expected issue costs and the

sensitivity to internally generated funds.











Table 17
Heteroskedastic consistent regressions relating total underwriting expense to firm
characteristics. Year dummy variables included (results not reported). The sample
consists of 289 bank holding companies from 1982-1994 (standard errors in parentheses).

Dependent Variable= percentage fees associated with:
Variable Common Stock Preferred Stock Subordinated Notes

Bind" -0.001 0.002 -0.001
(0.002) (0.002) (0.003)
Log (Total Assets,.1) -0.007 ** -0.006 ** -0.006 **
(0.001) (0.001) (0.002)
Free Cash Flow / Loans,. b -0.001 0.002 -0.005
(0.005) (0.003) (0.014)

Lag (Additions to Capital/ -0.147 -0.189 -0.389 *
Loans,.,) (0.084) (0.085) (0.177)

Securities / Assetst., 0.025 -0.016 0.001
(0.013) (0.018) (0.024)

Target -0.085 -0.069 0.092
(0.076) (0.042) (0.095)
Riskd 0.748 ** 0.426 ** 0.432
(0.247) (0.152) (0.523)
Risk2 -4.635 ** -2.075 -3.697
(1.777) (0.763) (8.207)
N (Adjusted R2) 109 (0.661) 95 (0.419) 116 (0.417)

a. Bind =1 if surplus capital is less than or equal to zero, =--0 otherwise
b. Free Cash Flow= net income less dividends and investment.
c. Target = the bank's average surplus capital over the entire sample period minus it's actual surplus capital.
d. Risk=- standard deviation of daily stock return
*, ** denote significance at the 5% and 1% levels, respectively











Table 18
Regression models relating underwriting fees to firm characteristics, correcting for selectivity bias
using Heckman's two step procedure. (standard errors in parentheses)


common stock preferred stock subordinated notes
Variable Probit Model % fee Probit Model % fee Probit Model % fee
_____________ (1 =issuer) (1 =issuer) (1=issuer)

Binda 0.03 0.01 -0.06 0.002 0.28 0.002
______(0.067) (0.01) (0.143) (0.003) (0.138) (0.003)

Log (Total Assets,.,) 0.11 ** -0.01 ** 0.33 ** -0.01 ** 0.79 ** -0.01 **
_____________ (0.018) (0.001) (0.021) (0.001) (0.022) (0.001)

Free CashFlow/ -2.08** -0.002 -1.19** 0.002 -1.21 ** -0.01
Loans,.,' (0.117) (0.005) (0.114) (0.008) (0.116) (0.006)

Lag (Additions to 9.63** -0.17 -5.27** -0.19 37.2** -0.28
Capital /Loans,.) (2.062) (0.106) (2.167) (0.109) (2.148) (0.176)

Securities I/Assets,.- -9.14 ** 0.02 -2.50 ** -0.01 -10.8 ** -0.03
(0.332) (0.017) (0.319) (0.024) (0.336) (0.024)
Target' 54.3 ** -0.08 11.7 ** -0.08 21.2 ** 0.19
(2.126) (0.083) (1.945) (0.122) (2.084) (0.113)
Risk" -3.22 0.87 ** -8.73 ** 0.42 -0.71 0.20
(1.415) (0.271) (2.062) (0.25) (1.475) (0.504)
3 month avg Stock 5.68** 6.25** 15.1**
Price / 36 month avg (0.289) (0.287) (0.797)

3 month avg Market -35.3 ** 0.71 -22.9 **
Price / 36 mo avg (1.979) (2.089) (2.072)

Hot 0.28 ** 0.25 ** -0.59 **
_____________ (0.055) (0.059) (0.057)
Issued in the Previous -0.42 ** -0.35 ** 0.01
Year =1 if yes, 0 if no. (0.103) (0.112) (0.095)

Inverse Mills Ratio 0.003** -0.002 0.01**
from Probit' ________ (0.0002) ________ (0.0135) (0.002)

a. Bind =1 if surplus capital is less than or equal to zero, =0 otherwise
b. Free Cash Flow= net income less dividends and investment.
c. Target = the bank's average surplus capital over the entire sample period minus it's actual surplus capital.
d. Risk= standard deviation of daily stock return
e. Inverse Mills Ratio = the probability of issuance derived from the Probit model
*, ** denote significance at the 5% and 1% levels, respectively










Table 19 presents fixed-effects regressions of loan growth identical to the model from

Table 3 with one important difference. I include the interaction of the variable High Fees with

additions to capital. The three categories of regressions indicate the security type for which the

underwriting fees are predicted. Because of a potential simultaneity bias induced by estimating

fees and then using the estimation as an explanatory variable, I instrument for this variable with a

lagged observation of High Fees. This interaction term is designed to test the hypothesis that

firms which anticipate larger external finance costs are more constrained by internally generated

funds. Regardless of security type, banks which anticipate higher than median underwriting fees

are more sensitive to internally generated funds.8 In other words, expected external finance costs

affect a bank's reliance on internally generated funds. This finding provides evidence that the

magnitude of the investment-cashflow sensitivity proxies for the size of the wedge between

internal and external financing costs.

Continuing along these lines, the expected abnormal returns following the announcement

of a security issuance may proxy for expected costs of issuance, and therefore may be related to

the investment-cashflow sensitivities. Using the same specifications from the underwriting fees

model, I estimate the abnormal stock returns based on firm characteristics using weighted least

squares, weighing each observation by the standard error from a market model regression. Table

20 presents results that indicate that abnormal returns are difficult to predict with observable

characteristics, as the adjusted R squared from these models is much lower than from the

underwriting fee models. However, this problem should simply introduce more noise in the


8 In additional tests, I divided firms into quartiles based on expected financing
costs. In these tests I find a significant difference between the highest and lowest quartile.
I also find a significant difference between the top two quartiles.











Table 19
Fixed effects regressions relating loan growth to internal additions to capital, expected
underwriting fees, and firm financial characteristics. The sample consists of 289 bank
holding companies from 1982-1994 (standard errors in parentheses).


Dependent Variable = (Loans, Loanst. ) / Loans,.,

Variable common stock preferred stock subord. notes

Additions to Capital/ 4.67 ** 3.59** 4.66 ** 3.76** 4.45 ** 3.65**
Loanst., (0.231) (0.188) (0.234) (0.201) (0.242) (0.202)

High Fees,., Additions to 0.38 0.39 0.34 0.39 0.82 ** 0.93 **
Capital /Loansta (0.186) (0.184) (0.179) (0.177) (0.165) (0.216)

Surplus Capital /Assetst., 0.93 ** 0.91 ** 0.82 **
(0.162) (0.164) (0.165)
Surplus Capital *Additions -30.7 ** -30.4 ** -26.6 **
to Capital /Loans., (4.588) (4.615) (4.761)

Bindb -0.05 -0.05 ** -0.05 **
______________ (0.009) (0.009) (0.009)
Bind Additions to Capital 0.46 0.99* 0.71
/ Loans,., (0.459) (0.512) (0.514)

Securities / Assets,., 0.13** 0.15** 0.13** 0.13** 0.13** 0.13**
______________ (0.039) (0.039) (0.040) (0.040) (0.040) (0.039)
Market/Book Assets,, 0.26 ** 0.27 ** 0.27 ** 0.26 ** 0.25 ** 0.24 **
_______________ (0.048) (0.049) (0.048) (0.049) (0.048) (0.048)
log (Assets,.,) -0.07 ** -0.07 ** -0.07 ** -0.07 ** -0.07 ** -0.07 **
_______________ (0.005) (0.005) (0.006) (0.005) (0.005) (0.005)
Lag loan growth 0.07 ** 0.07 ** 0.07 ** 0.07 ** 0.05 ** 0.04 **
(0.010) (0.010) (0.010) (0.010) (0.011) (0.012)
R2 0.379 0.377 0.379 0.375 0.383 0.379
N (categories) 1986 1986 1986 1986 1986 1986
_______________ (289) (289) (289) (289) (289) (289)
F statistic, Bank dummies 2.33 ** 2.62 ** 2.33 ** 2.60 ** 2.37 ** 2.65 **

a. High Fees =1 if the firms predicted fees from the Underwriting Fees model in Table 18 (corrected for
selectivity bias) are greater than the median predicted fees, =0 otherwise.
b. Bind=l if surplus capital<=0, 0 otherwise.
*, ** denote significance at the 5% and 1% levels, respectively











Table 20
Weighted least squares regressions relating abnormal stock returns following the
announcement of a security issuance to firm characteristics. Observations are weighted
by the standard errors from a market model regression. Year dummy variables included
(results not reported). The sample consists of 289 bank holding companies from 1982-
1994 (standard errors in parentheses).

Dependent Variable= abnormal stock returns associated with:
Variable Common Stock Preferred Stock Subordinated Notes

Bind" 0.03 ** -0.01 -0.01
(0.012) (0.011) (0.001)
Log (Total Assets,.,) 0.001 0.004 0.002
(0.002) (0.004) (0.002)
Free Cash Flow / Loanst.,b -0.03 0.03 0.02
(0.023) (0.013) (0.011)
Lag (Additions to Capital / -0.31 -0.17 0.32
Loans,.1) (0.457) (0.366) (0.233)
Securities /Assets,, -0.01 0.02 0.01
(0.065) (0.077) (0.031)
Target -0.08 -0.17 0.08
(0.377) (0.366) (0.193)
Risk' -0.76 0.06 -1.15
(0.802) (0.637) (0.872)
Risk2 6.18 1.06 12.9
(5.972) (3.121) (12.07)
N (Adjusted R2) 107(0.166) 90(0.339) 155(0.158)

a. Bind =1 if surplus capital is less than or equal to zero, =0 otherwise
b. Free Cash Flow= net income less dividends and investment.
c. Target = the bank's average surplus capital over the entire sample period minus it's actual surplus capital.
d. Risk=- standard deviation of daily stock return


*, ** denote significance at the 5% and 1% levels, respectively










estimates and bias against finding any relation between expected costs and investment-

cashflow sensitivities.

Table 21 presents results from the probit and abnormal return models using the

Heckman procedure correcting for selectivity bias. As before, the coefficients from the

abnormal return model are used to construct predicted values of the cost of issuing each

type of security for all firms. I then create a dummy variable (Large Abnormal Returns)

which equals one if the predicted abnormal returns are more negative than the median

predicted returns, or zero otherwise for each security type. This dummy variable is

employed to estimate the relation between expected costs and the sensitivity to internally

generated funds.

Table 22 presents results from a fixed effects regression model of loan growth,

including the interaction between Large Abnormal Returns and additions to capital.

Again, because of a potential simultaneity bias resulting from estimating costs, I

instrument for this variable with a lagged observation of Large Abnormal Returns. As in

the underwriting fee models, regardless of security type, banks that expect abnormal

returns more negative than the median expected abnormal return are significantly more

constrained by internally generated additions to capital. This suggests that expected

abnormal stock returns influence the dependence on internally generated additions to

capital, and that banks that anticipate larger external finance costs are more constrained by

additions to capital than banks that anticipate smaller costs. Overall, these results provide

evidence that investment-cashflow sensitivities proxy for the severity of information

asymmetries.











Table 21
Regression models relating abnormal stock returns to firm characteristics, correcting for
selectivity bias using Heckman's two step procedure. (standard errors in parentheses).


common stock preferred stock subordinated notes

Variable Probit Model Abnormal Probit Model Abnormal Probit Model Abnormal
(1 =issuer) Return (1 =issuer) Return (l=issuer) Return

Bind' 0.46 ** 0.02 0.05 -0.01 0.40 ** -0.01
(0.064) (0.008) (0.145) (0.010) (0.140) (0.003)

Log (Total Assetst1) 0.04 0.002 0.40 ** 0.01 0.04 0.002 *
(0.019) (0.003) (0.021) (0.004) (0.019) (0.001)

Free Cash Flow/ 0.23 -0.03 -1.31 ** 0.03 0.15 0.02 **
Loanst.,b (0.119) (0.019) (0.116) (0.023) (0.123) (0.006)

Lag (Additions to -17.5 ** -0.02 -8.72 ** -0.17 4.69 0.25 *
Capital /Loans.t) (2.089) (0.440) (2.273) (0.358) (2.117) (0.129)
Securities/Assets,., -5.73 ** 0.001 -3.44 ** -0.01 -1.85 ** 0.03 *
(0.316) (0.072) (0.345) (0.070) (0.305) (0.013)

Targetc 18.7 ** -0.25 19.8 ** -0.02 6.53 ** 0.04
(1.858) (0.362) (2.169) (0.373) (1.862) (0.109)
Riskd -27.31 ** -0.48 -5.23 ** 0.86 -13.1 ** -1.48 *
(2.112) (0.884) (2.026) (0.819) (2.121) (0.628)
3 month avg Stock 3.82 ** 7.60 ** -5.48 **
Price / 36 month avg (0.531) (0.288) (0.703)

3 month avg Market 30.8** 18.6** 15.0**
Price /36 mo avg (2.057) (2.267) (2.075)

Hot 0.13* -0.32** -0.12*
_____________ (0.057) (0.061) (0.058)

Issued in the Previous -0.41 ** -0.37 ** -0.33 **
Year =1 if yes, 0 if no. (0.102) (0.118) (0.077)

Inverse Mills Ratio 0.01 0.02 -0.02**
from Probitr ______ (0.018) ______(0.012) (0.001)

a. Bind =1 if surplus capital is less than or equal to zero, =0 otherwise
b. Free Cash Flow= net income less dividends and investment.
c. Target = the bank's average surplus capital over the entire sample period minus it's actual surplus capital.
d. Risk=-- standard deviation of daily stock return
e. Inverse Mills Ratio = the probability of issuance derived from the Probit model
*, ** denote significance at the 5% and 1% levels, respectively











Table 22
Fixed effects regressions relating loan growth to internal additions to capital, expected
abnormal stock returns, and firm financial characteristics. The sample consists of 289
bank holding companies from 1982-1994 (standard errors in parentheses).


Dependent Variable = (Loanst Loans,.1 ) / Loans,.,

Variable common stock preferred stock subord. notes

Additions to Capital / Loans,, 4.52** 3.77** 4.51** 3.66** 4.68** 3.76**
(0.242) (0.207) (0.233) (0.204) (0.227) (0.199)
Large Abnormal Returns,, 0.32 0.37 0.85 ** 0.84 ** 0.72 ** 0.76 **
Additions to Capital / Loans,.," (0.177) (0.178) (0.210) (0.209) (0.189) (0.189)

Surplus Capital/Assets,., 0.80 ** 0.87 ** 0.95 **
________________ (0.163) ______ (0.155) (0.159)
Surplus Capital *Additions to -26.5 ** -27.7 ** -30.9 **
Capital /Loans.1 (4.722) (4.458) (4.551)

Bindb -0.05** -0.06** -0.06**
______ (0.009) (0.009) (0.009)
Bind Additions to Capital / 0.21 1.02 0.97 *
Loans,., (0.533) ______ (0.511) (0.501)

Securities / Assets,., 0.12** 0.12** 0.13** 0.13** 0.11** 0.11**
(0.039) (0.039) (0.039) (0.038) (0.040) (0.040)
Market/Book Assets,., 0.26 ** 0.25 ** 0.26 ** 0.24 ** 0.24 ** 0.22 **
(0.047) (0.048) (0.049) (0.049) (0.049) (0.049)
log (Assets,.,) -0.07 ** -0.07 ** -0.06 ** -0.07 ** -0.07 ** -0.07 **
________________ (0.005) (0.005) (0.005) (0.005) (0.005) (0.005)
Lag loan growth 0.07 ** 0.07 ** 0.08 ** 0.08 ** 0.06 ** 0.06 **
________(0.009) (0.009) (0.009) (0.009) (0.010) (0.010)
S0.355 0.351 0.385 0.384 0.383 0.379
N (categories) 1968 1968 1968 1968 1968 1968
________________ (289) (289) (289) (289) (289) (289)
F statistic, Bank dummies 2.35 ** 2.62 ** 2.34 ** 2.63** 2.36 ** 2.63 **

a. Large Abnormal Returns =1 if the firms predicted abnormal stock returns from the Abnormal Stock Returns
model in Table 21 (corrected for selectivity bias) are more negative than the median predicted abnormal return,
=0 otherwise.
b. Bind=l if surplus capital<=0, 0 otherwise.

*, ** denote significance at the 5% and 1% levels, respectively













CHAPTER 7
SUMMARY AND CONCLUSIONS

Overall, my results suggest that asymmetric information problems increase the

costs of external finance for banking firms. In particular, I find a positive and significant

relation between bank loan growth and internally generated additions to capital.

Moreover, consistent with the hypothesis that capital requirements limit bank financing

alternatives, this cashflow sensitivity of investment is positively related to the extent that

capital requirements are binding. This relationship implies that increases in capital

standards, which increase the likelihood that capital requirements are binding, could

induce a slowdown in loan growth.

I also find that the formulation of the capital ratio itself is important in determining

bank loan growth. Specifically, with regulators enforcing leverage-based capital

standards, banks can rely on a buffer stock of securities to fund investment in a liquidity

crisis. This results in a negative relation between the cashflow sensitivity of investment

and securities holdings. However, the use of risk-based standards substantially reduces

the effectiveness of securities as financial slack. As a result, the change from leverage-

based to risk-based standards may have caused banks to desire more surplus capital, and

therefore may have had a negative impact on loan growth.

My results also suggest that bank holding companies allocate capital in a way

consistent with the operation of an internal capital market by holding companies that find









equity expensive to raise externally. Specifically, I find that investment by bank

subsidiaries is more sensitive to the cash flows and capitalization of its holding company

than its own cash flows and capitalization. I also find that subsidiary investment is

significantly related to the non-bank earnings of its holding company. These results are

consistent with the hypothesis of costly external finance, and suggest that empirical studies

of the effects of changes in capital requirements should be considered on the holding

company level and not the individual bank level.

Finally, I find that the severity of information asymmetries affects both the

likelihood and the costs of an external capital issuance. In particular, I find a negative

relation between the cash flow sensitivity of investment and the probability that banks

issue external capital. I also find that firms which anticipate underwriting fees larger than

the median expected fee for an external capital issuance exhibit significantly higher cash

flow sensitivities of investment. Moreover, banks expecting abnormal stock returns more

negative than the median expected abnormal return following the announcement of an

external capital issue are significantly more constrained by internally generated funds.

These results provide evidence that investment-cashflow sensitivities proxy for the severity

of capital market imperfections.













APPENDIX
ESTIMATION OF RISK-WEIGHTED ASSETS


A data set with complete risk-weighted assets data is available to me for 98 bank

holding companies on December 31, 1995.1 These data disaggregate total risk-weighted

assets into three major categories: balance sheet assets, loan commitments, and

derivatives. Table 23 presents descriptive statistics. The sample contains both very large

and small banks, as total assets range from a low of $105 million to Citicorp's $216

billion. In addition, risk-weighted assets range from $51 million to $230 billion. By far,

balance sheet assets contribute the most to risk-weighted assets. For many banks, mostly

small ones, off-balance sheet assets hardly add to risk-weighted assets at all. Therefore,

for the most accurate approximation of total risk-weighted assets, I estimate each major

component separately. That is, I decompose risk-weighted assets into its three broad

categories. A regression for each component is specified, and parameter estimates

obtained. The regressions are specified as follows:

RWBS = aILOANS + a2SECS
RWLC = PJISLC + p2LC + P2LOC
RWD = YIFEXPURCH +y2FEXPOPT +Y3FEXWOPT +Y4FEXSWP+Y5FUTURES
+Y6WOPT + y7POPT + y8SWAPS




1 Data set supplied by Carolyn Takeda. Before December 1991, the breakdown of
risk-weighted assets into its components is not available.

80











Table 23
Descriptive Statistics of 98 bank holding companies at year-end 1991. Data were
collected from the Federal Reserve Y-9 Data Tapes. This sample consists of banks for
which complete risk-based capital data was available.*


Variable Mean Median Min Max


Total Assets (millions) 17,474 5,317 105 216,920


Risk-Weighted Assets (millions) 14,700 3,787 51 238,000


Risk-Weighted Assets, from on- 9,738 2,229 18 165,000
balance sheet items (millions)

Risk-Weighted Assets, from loan 2,435 216 20 54,800
commitments (millions)

Risk-Weighted Assets, from derivative 424 0 0 108,000
assets (millions)

Loans / TA 0.6008 0.6358 0.1266 0.8057


Securities / TA 0.2399 0.2233 0.0348 0.6152


(Loans Loans,-,) / Loans.-, 0.0054 -0.0044 -0.3558 0.3206


Internal Additions to Capital / Loanst., 0.0096 0.0120 -0.0213 0.0219


Market /Book Assets 1.0182 1.0149 0.9554 1.2055


Book Capital in Excess of Requirement 0.0228 0.0226 -0.0072 0.2046
/ Assets

* Special thanks to Carolyn Takeda for use of this data set.










RWBS = risk-weighted assets attributable to the balance sheet
LOANS = loans outstanding
SECS = securities held
RWLC = risk-weighted assets attributable to loan commitments
SLC = stand-by letters of credit
LC = loan commitments
LOC = lines of credit
RWD = risk-weighted assets attributable to derivative assets
FEXPURCH = foreign exchange purchase commitments
FEXPOPT = foreign exchange purchase options
FEXWOPT = foreign exchange written options
FEXSWP = foreign exchange swaps
FUTURES = futures and forwards
WOPT = written options
POPT = purchase options
SWAPS = non-foreign exchange swaps.

Actual risk-based capital data are used as the dependent variables, and broad

classifications of on and off-balance sheet components are used as the independent

variables. Coefficient estimates from these equations are then used to approximate the

risk-weights assigned to each component for purposes of computing total risk-weighted

assets. Consequently, this procedure approximates risk-based capital ratios for all banks

from 1990 to present.2

Table 24 presents the regression results. As expected, the coefficient on loans is

very significant and close to unity. Moreover, the results estimate the risk-weight on

stand-by letters of credit to be approximately one.3 The next largest risk-weight comes

from securities, at 0.34. No variable has a coefficient significantly greater than one, which

is consistent with risk-based capital standards.


2 Reporting requirements from 1990 to present ensure sufficient off-balance sheet
data to estimate risk-weighted assets for the majority of firms in the sample.

3 Coefficient not significantly different from one.











Table 24
Least squares estimation of risk-weighted assets (RWA). Data were collected from the
Federal Reserve Y-9 Data Tapes. A sample of 98 firms contains complete RWA data for
year-end 1991. These observations are used to estimate components of RWA. RWA is
decomposed into three categories, balance sheet assets, off-balance sheet loan
commitments, and derivative sheet assets. Regressions use the actual amount of RWA
attributable to a category as the dependent variable. Broad classes of asset items are used
as independent variables. Regressions are performed using White's correction for
heteroskedasticity. Standard errors are in parenthesis.

Variable RWA: Balance RWA: Loan RWA: Derivative
Sheet Assets Commitments Assets
Loans *** 1.0197
(0.0403)
Securities 0.3393
____~_________ (0.2396)
Stand-by Letters of *** 1.3174
Credit (0.3134)

Loan Commitments *** 0.1483
______~_________~___(0.0424)
Lines of Credit -0.5543
_____________ ~ (1.5209)
Foreign Exchange (FEX) *** 0.0069
Purchase Comm. (0.0008)

Foreign Exchange *** 0.3266
Purchase Options (0.0528)

Foreign Exchange *** -0.2819
Written Options (0.0482)

Foreign Exchange Swaps *** 0.0435
~____~_______~_____ ____~______(0.0012)
Futures and Forwards *** -0.0069
~____~__________ ______________ (0.0006)
Written Options (non -0.0039
FEX) _____ ___ (0.0075)

Purchase Options (non *** 0.0201
FEX) ______________ ______________ (0.0058)

SWAPS (non FEX) *** 0.0134
____________~ ~ (0.0001)
N 98 98 98
Adjusted R Squared 0.9940 0.9793 0.9999

*, **, *** denote significant at the 10%, 5%, and 1% levels respectively.













REFERENCES


Asquith, Paul and D. Mullins (1986). "Equity Issues and Offering Dilutions," Journal of
Financial Economics 15, 61-89.

Baer, Herbert L. and John N. McElravey (1993). "Capital Shocks and Bank Growth,
1973-1991," Federal Reserve Bank of Chicago Economics Perspectives 17, 2-21.

Bayless Mark, and Susan Chaplinsky (1991). "Expectations of Security Type and the
Information Content of Debt and Equity Offers," Journal of Financial
Intermediation 1,195-214.

______ (1996). "Is There a 'Window of Opportunity' for Seasoned Equity
Issuance?" Journal of Finance 51, 253-278.

Berger, Alan N. and Gregory F. Udell (1994). "Did Risk-Based Capital Allocate Bank
Credit and Cause a 'Credit Crunch' in the U.S.?" Journal ofMoney, Credit, and
Banking 26, 585-634.

Bernanke, Ben S., M Gertler and S. Gilchrist (1994). "The Financial Accelerator and the
Flight to Quality," National Bureau of Economic Research Working Paper.

Bernanke, Ben S. and Cara S. Lown (1991). "The Credit Crunch," Brookings Papers on
Economic Activity 2, 205-248.

Calomiris, Charles, and Charles Himmelberg (1995). "Investment Banking Costs as a
Measure of the Cost of Access to External Finance," National Bureau of
Economic Research Working Paper.

Comett, Marcia M. and Hassan Tehranian (1994). "An Examination of Voluntary Versus
Involuntary Security Issuances by Commercial Banks," Journal of Financial
Economics 35, 99-122.

Fallon, Kieran (1991). "Source of Strength or Source of Weakness? A Critique of the
'Source of Strength' Doctrine in Banking Reform," New York University Law
Review 66, 1344-1403.










Fazzari, Steven M., R. Glenn Hubbard and Bruce C. Petersen (1988), "Financing
Constraints and Corporate Investment," Brookings Papers on Economic Activity
1, 141-195.

Federal Reserve Board of Governors (1996). Bank Holding Company Supervision
Manual, Federal Reserve, Washington D.C.

Furlong, Frederick T. and Michael C. Keeley (1989). "Capital Regulation and Bank Risk-
Taking: A Note," Journal of Banking and Finance 13, 883-892.

Hancock, D and J. Wilcox (1993). "Has There Been a 'Capital Crunch' in Banking? The
Effects on Bank Lending of Real Estate Market Conditions and Bank Capital
Shortfalls," Journal of Housing Economics 3, 31-50.

Haubrich, Joseph G. and Paul Wachtel, (1993). "Capital Requirements and Shifts in
Commercial Bank Portfolios," Federal Reserve Bank of Cleveland Economic
Review 29, 2-15.

Hoshi, T., A. Kashyap and D. Scharfstein (1991). "Corporate Structure, Liquidity, and
Investment," Quarterly Journal of Economics 106, 33-59.

Keeton, William (1990). "Bank Holding Companies, Cross-Bank Guarantees, and Source
of Strength," Federal Reserve Bank of Kansas City Economic Review 75, 54-67.

Lamont, Owen (1993). "Cashflow and Investment: Evidence From Internal Capital
Markets," Massachusetts Institute of Technology Working Paper.

Mikkelson, Wayne H. and Megan Partch (1986). "Valuation Effects Of Security
Offerings and the Issuance Process," Journal of Financial Economics 15, 31-60.

Myers, Stewart C. and Nicholas S. Majluf(1984). "Corporate Financing and Investment
Decisions when Firms Have Information that Investors Do Not Have," Journal of
Financial Economics 13, 187-221.

O'Hara, Maureen, and Wayne Shaw (1990). "Deposit Insurance and Wealth Effects: The
Value of Being Too Big to Fail,"' Journal of Finance 45, 1587-1600.

Passmore, Wayne and Steven A. Sharpe, (1994). "Optimal Bank Portfolios and the Credit
Crunch," Finance and Economic Discussion Series 94-19, Federal Reserve Board.

Peek, Joe and Eric S. Rosengren (1995), "Bank Regulation and the Credit Crunch,"
Journal of Banking and Finance 19, 679-693.










Sharpe, Steven A. (1995). "Bank Capitalization, Regulation, and the Credit Crunch: A
Critical Review of the Research Findings," Federal Reserve Board of Governors
Working Paper.

Stein, Jeremy C. (1995). "Internal Capital Markets and the Competition for Corporate
Resources," National Bureau of Economic Research Working Paper.

Takeda, Carolyn T. (1994). "Economic Determinants of Contingent Commitment
Activity and the Effect of Risk-Based Capital," University of Florida Working
Paper.

Thakor, Anjan V (1996). "Capital Requirements, Monetary Policy, and Aggregate Bank
Lending: Theory and Empirical Evidence," Journal of Finance 51, 279-324.

Wall, Larry and D. Peterson (1995). "Bank Holding Company Capital Targets in the
Early 1990s: The Regulators Versus the Markets," Journal of Banking and
Finance 19, 563-575.

Williamson, Oliver E. (1975). Markets and Hierarchies: Analysis and Antitrust
Implications, Collier Macmillan Publishers, Inc., New York.













BIOGRAPHICAL SKETCH

David Frederic Marcus was born the third child of Jeffery Neal and Ellen Janet

Marcus on September 4, 1968, in Eau Claire, Wisconsin. He attended the University of

Colorado at Boulder from 1986-1990. David graduated Magna Cum Laude with a

Bachelor of Science. Following graduation, David worked for two years before enrolling

at the University of Florida as a doctoral student.







I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for
the degree of Doctor of Philosophy.


"topheZa m es, chairman
SunBan Crofessor of Finance,
Insurance, and Real Estate

I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for
the degree of Doctor of Philosophy. /


Dr. oel F. Houston
Associate Professor of Finance
Insurance, and Real Estate


I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for
the degree of Doctor of Philosophy.


Dr. Michael D. Ryngaert
Associate Professor of Finance
Insurance, and Real Estate

I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for
the degree of Doctor of Philosophy. /


Sr. Jonathan H. Hamilton
Associate Professor of Economics

This dissertation was submitted to the Graduate Faculty of the Department of Finance,
Insurance, and Real Estate in the College of Business Administration and to the Graduate School
and was accepted as partial fulfillment of the requirements for the degree of Doctor of
Philosophy.


Dean, Graduate School


August, 1996

















LD
1780
199_q









UNIVERSITY OF FLORIDA
11111I 11 1 111111 i 11111111 I III I 1111111111
3 1262 08555 0852




Full Text
xml version 1.0 standalone yes
Volume_Errors
Errors
PageID P212
ErrorID 4