Blasko 2006

The Quarterly Review of Economics and Finance
46 (2006) 5381
Bank asset structure, real-estate lending, and

risk-taking
Matej Blasko a, , Joseph F. Sinkey Jr. b,1
a
Department of Accounting and Finance, School of Business Administration,

Oakland University, Rochester, MI 48309, USA
Department of Banking and Finance, Terry College of Business, The University of Georgia,
Athens, GA 30602-6253, USA
Received 26 November 2003; received in revised form 9 December 2003; accepted 11 November 2004
Available online 6 July 2005
Abstract
The banking industry changed substantially in the 1990s as the number of banks declined rapidly,
and as we document, commercial banks dramatically shifted their assets to real-estate loans. The
portfolio restructuring seems to be followed mainly by capital-constrained banks as real-estate banks
have lower risk-based-capital ratios relative to those of our benchmark group. Trading off credit
risk for interest-rate risk is only one of the ways to arbitrage regulatory capital. We also show that
real-estate banks keep higher ratios of fixed-rate loans to total assets and face higher probabilities of
insolvency. The increasing proportion of banks specializing in real-estate lending, the incentives of
regulatory discipline, and the weaknesses of risk-management strategies could stress the condition of
the banking system during periods of large unexpected increases in interest-rates and are important
issues for regulators and bank managers.
2005 Board of Trustees of the University of Illinois. All rights reserved.
JEL classication: G21; G28; G32
Keywords: Commercial banks; Real-estate banks; Financial risk management; Interest-rate risk; Regulatory capital
arbitrage; Risk shifting
Corresponding author. Tel.: +1 248 370 3277; fax: +1 248 370 4275.
E-mail addresses: blasko@oakland.edu (M. Blasko), jsinkey@terry.uga.edu (J.F. Sinkey Jr.).
Tel.: +1 706 546 0331.
1062-9769/$ see front matter 2005 Board of Trustees of the University of Illinois. All rights reserved.
doi:10.1016/j.qref.2004.11.002
54
M. Blasko, J.F. Sinkey Jr. / The Quarterly Review of Economics and Finance 46 (2006) 5381
1. Introduction
The banking industry and housing finance have changed substantially since the savingsand-loan crisis of the 1980s. Prior to this, traditional banking markets were fragmented,
containing a large number of small and specialized firms. Structural changes (e.g., technology and deregulation) slowly but surely transformed segmented banking firms into one of
consolidated, multiple-product financial institutions and marked the birth of the financialservices industry. During this transition, commercial banks increasingly shifted their product
and asset portfolios into consumer lending, traditionally a specialty of thrifts and credit
unions. This paper investigates real-estate lending, asset structure, and risk-taking of all
insured commercial banks during 19891996. We focus primarily on banks specializing in
real-estate lending (REBs), defined as commercial banks holding more than 40% of their
total assets in loans secured by real estate. All other commercial banks in the economy provide a benchmark used to compare and contrast the behavior of REBs. Given the changes
in bank capital regulation as well as the greater interest-rate risk of fixed-rate mortgages,
we also explore how bankers manage risk.
We present evidence that commercial banks have increasingly shifted their portfolios to
real-estate loans. For example, the number of REBs has increased from 1724 at year-end
1989 to 2835 at year-end 1996. This bank-level shift to real-estate lending has been paralleled by an equivalent increase in real-estate loans at the macro-industry level. This finding
is especially noteworthy given the substantial consolidation in the banking industry, which
experienced a drop from 12,702 banks (1989) to 9529 banks (1996).2 This shift in portfolio
strategy seems to be followed mainly by banks constrained by regulatory requirements. We
find that REBs have lower risk-based-capital (RBC) ratios relative to ratios of our benchmark group. This result can be explained, in part, by the existence of RBC standards that
allowed banks to shift their portfolios to presumably safer assets, such as mortgages, with
the benefit of lower capital requirements but without accountability for interest-rate risk.
Trading off credit risk for interest-rate risk is only one of the ways to manipulate regulatory
discipline.3
Based on our risk measures, we show that REBs, relative to less-specialized banks,
face higher probabilities of insolvency. While the overall riskiness of the banking sector
declined during our sample period, the higher riskiness of REBs is observed in most of
the sample years and confirmed by a multivariate linear regression model. Given the expost construction of our risk measure, the higher riskiness of real-estate banks can be
attributable either to their higher natural economic risk exposures (bank equity risk), or
to desired risk-level targets. It is not the goal of this paper to test which explanation is
correct.
2 The consolidation in the banking industry accompanied the contraction of the thrift industry. Although some
thrifts converted to bank charters, the number of conversions was dwarfed by the increase in the number of
REBs.
3 Merton (1995) provides additional examples of loopholes that allow bank managers to circumvent regulatorycapital rules for risk-weighted assets. In general, banks can create the economic equivalent of the same desired
portfolio that is treated differently under the current rules governing regulatory capital requirements.
55
Banks use two main channels to manage interest-rate risk: on-balance sheet operational
activities and off-balance sheet techniques. Although real-estate banks have significantly
better maturity GAPs than less-specialized banks, their use of off-balance sheet activities in
the form of derivatives is not different. For example, only about 5% of REBs use interest-rate
derivatives, the same proportion observed for all other banks. In addition, cross-sectional
differences in bank asset structures are noteworthy. Although REBs, relative to other banks,
use floating-rate loans to a greater degree, they still keep higher ratios of fixed-rate loans
to total assets. Also, the evidence does not support the hypothesis that banks benefit from
hedging at the holding-company level. Jointly, this indicates a higher economic riskiness of
REBs, and therefore higher values of the put options written by deposit insurers. Since the
optimal bank risk-level depends on shareholder preferences, we cannot and do not claim that
REBs should hedge more. However, observing the steady increase in REBs and their lower
capital ratios, the evidence indicates that bank shareholders do prefer higher risk-levels
relative to the alternative strategies or they are capital constrained and shift their portfolios
in response to regulatory incentives or both.
Several motivations drive our research. Scant evidence exists on the risk-taking behavior
of commercial banks, especially community organizations, and the relationship between
bank risk and real-estate lending in the 1990s.4 The S&L crisis of the 1980s provides an
important historical precedence of how costly neglected bank risk-taking preferences can
be. Also, regulatory capital standards introduced in the late 1980s may have provided banks
with new incentives and may explain their risk-taking and risk-management strategies. The
regulatory environment provides some clues to the observed bank behavior. Our evidence is
consistent with the hypothesis of regulatory capital arbitrage (Jones, 2000, among others).
This paper addresses the systemic risk posed by a growing subset of the U.S. banking
system in the form of REBs banks for the years 19891996. Since smaller banks dominate
the population of REBs, our research primarily applies to them. If large banks engage in
risk-shifting behavior and exploit deposit-insurance subsidies (e.g., Hovakimian & Kane,
2000, among others), it is more than likely that REBs also respond strongly to regulatory
incentives. Moreover, although Eisenbeis & Kwast (1991) document the viability of REBs
for the years 1978 through 1988, they do not assess the hedging and risk-management
practices of REBs, especially in light of S&L crisis of the late 1980s. And finally, our study
extends the research on risk management at financial institutions of Esty (1997, 1998),
Esty, Tufano, & Headley (1994), and Froot, Scharfstein, & Stein (1993), Froot & Stein
(1998).
The paper proceeds as follows. The next section briefly reviews the banking environment
and the theories related to corporate hedging, capital regulation, and deposit insurance. The
third section describes our sample and methodology. The fourth section introduces the
stratification procedures for constructing our bank groups and describes the resulting data.
4 Banks manage several risks including default or credit risk, interest-rate risk and liquidity risk. Most of
the previous studies ignore the interest-rate risk, which is the most important risk facing banks (e.g., REBs) that
concentrate in long-term, fixed-rate assets such as mortgages. Gilbert & Sierra (2003) assess the financial condition
of community banks. Interestingly and proving our point, their work ignores interest-rate risk. Even the newly
proposed Basel II Accord rules require less, not more regulatory capital backing mortgage lending relative to the
current Accord.
56
The fifth section presents our empirical results. The last section summarizes and concludes
the paper and draws some policy implications.
2. Theoretical, empirical, and banking-industry background

The thrift-industry or S&L crisis in the U.S. during the 1980s and early 1990s created
a dilemma: the perceived obsolescence of the industry versus the need for a continued
flow of funds for housing finance (Kane, 1989). When failures and consolidation led to
the demise of the thrift industry, the commercial-banking industry took up the subsequent
slack in real-estate lending and kept the flow of funds to housing finance steady.5 However, concentration in mortgage lending challenges the willingness and ability of banks to
manage increased interest-rate risk. Moreover, without proper regulatory incentives, REBs,
like large banks, can shift risk onto the federal safety net. As previously documented,
shareholders of limited liability banks have incentives to take risk above the levels optimal for an all-equity bank (see John, John, & Senbet, 1991). Risk-shifting behavior by
banks was seen as a major reason for the recent S&L crisis. Nevertheless, securitization,
the origination and selling of loans without the need for funding them, presents banks and
thrifts with a technique for managing credit, interest-rate, and liquidity risks but not prepayment risk. Securitization is a risk-removal technique as opposed to on- or off-balance sheet
hedging.
Substantial changes in the bank regulatory environment have occurred during the 1990s,
namely, the passage of the Financial Institutions Reform, Recovery, and Enforcement Act
of 1991; the FDIC Improvement Act (FDICIA) of 1991; the implementation risk-based
capital requirements by 1992; and the National Depositor Preference Act of 1993. Riskbased capital requirements were partially implemented in 1990 and fully effective by 1992.
Besides other motives for real-estate lending strategies, risk-based capital and the promptcorrective-action feature of FDICIA created important motivations for banks to either raise
their capital ratios or revise their asset-allocation strategies to avoid supervisory sanctions.
The regulatory discipline embodied in these acts was intended, among other things, to
better monitor and price bank risk-taking. For a sample of 123 large bank holding companies for the years 19851994, Hovakimian & Kane (2000) find that capital regulation
has failed to control risk-shifting incentives. Their study captures the focus of research on
systemic risk centered on large banks and based on market data in an option-pricing framework. However, if many small banks encounter financial difficulties, the banking system
also can become stressed and local economic activity can be affected adversely. For example, smaller banks account for a disproportionate share of bank loans to small businesses
and also losses to the deposit insurance funds (Gilbert & Sierra, 2003). In addition, loss
rates on bank failures have been inversely related to bank size. When it comes to interestrate risk, our results suggest a similar conclusion for REBs. Thus, banks constrained by
regulatory capital requirements may shift to real-estate lending, without a correspond-
An important pre-condition for further bank entry into real-estate lending was the development of mortgagebacked securities and the roles of government agencies such as Ginnie Mae, Fannie Mae, and Freddie Mac.
57
ing decrease in economic risk, which presumably is the intention of capital-adequacy

standards.
On balance, REBs face few, if any, regulatory penalties for taking on extra interest-rate
risk. Moreover, risk-based capital requirements encourage them to engage in regulatory
capital arbitrage, which Jones (2000, p. 36), among others, describes as opportunities for
banks to reduce substantially their regulatory measures of risk with little or no corresponding
reduction in their overall economic risk. Several potential explanations exist for banks to
change their risk-management strategies.
The first argument focuses on the lack of regulatory discipline with respect to interestrate risk. Merton (1977) established the academic foundations for risk-based supervision
of bank capital requirements.6 A deposit guarantee can be viewed as a put option written by the regulator. The value of deposit insurance to bank stockholders will increase
with increases in bank equity risk, a function of asset risk and leverage. Bank regulators
can measure and control their loss exposure by monitoring these variables. Regulators,
similar to uninsured creditors, demand adequate capital to protect themselves against the
costs of financial distress, agency problems, and the reduction in market discipline caused
by the federal safety net. Differences in risk-based pricing of deposit insurance between
the best- and worst-rated banks may be insufficient relative to the differential charged in
capital markets. Kane (1995), among others, argues that regulatory capital requirements
may be inefficient tools for managing the risk, as private entities would limit risk exposure
more rigorously. He also discusses the role of capital requirements in insured financial
institutions.
Another explanation builds on Hogan & Malmquists (1999) study that attributes the
absence of hedging by small financial institutions to their limited access to OTC derivatives.
Transaction costs and access restrictions may impair community banks ability to costefficiently increase their equity capital or hedge their underlying asset risks.
Esty (1997, 1998) and Schrand & Unal (1998) find that stock-owned savings institutions
engage in more risk-taking than mutually-owned thrifts. Whidbee & Wohar (1999), investigating hedging behavior by BHCs, find that the extent of hedging is negatively related
to the level of managerial shareholdings. Managers whose interests are more aligned with
those of shareholders (reflected by insider shareholdings exceeding 10%) are less likely
to hedge, suggesting risk-shifting and potential wealth transfers due to mispriced deposit
insurance. The universe of U.S. commercial banks, as used in our study, consists primarily
of community banks. As ownership tends to be more concentrated in these firms, our results
are consistent with the findings of Whidbee & Wohar (1999).
While we do not argue that bankers should hedge more, as the optimal level of bank
risk depends on shareholder preferences, bank safety-net managers clearly care about the
total firm risk. Therefore, the bank asset-allocation strategies should be of interest to bank
regulators. This research provides a partial look at bank risk-taking and strategies that reflect
and may, indeed, be driven by bank regulations.
It also is instructive to explore the benefits of corporate hedging policies. Even though
almost three-fourths of large corporations have adopted at least some financial-engineering
6
Merton (1995) also presents a general discussion of financial innovation, regulation, and risk management of
financial institutions.
58
techniques (Smithson, 1996a, 1996b; Group of Thirty, 1993) to control their exposures
to interest rates, foreign-exchange rates, and commodity prices, academics seem to know
little about corporate risk management (Tufano, 1996, among others). In a Modigliani &
Miller (1958, M&M) framework, corporate hedging and other financial decisions cannot
create value for a firm unless they affect either the firms ability to operate its business or
its incentives to invest in the future. Thus, from the perspective of modern finance theory,
corporate hedging practice may seem puzzling, as shareholders may diversify and hedge
risks on their own. However, violations of assumptions of M&M theorems have led to
several hypotheses of how risk-management can enhance firm value. The primary venues
for value creation through corporate hedging are linked to decisions changing tax liabilities,
changing contracting costs, or changing investment incentives.
The finance literature identifies several benefits of corporate-hedging policies. Most
importantly, hedging can reduce the costs of financial distress, reduce expected taxes (Smith
& Stulz, 1985), and prevent a shortfall in funds and under-investment problems as firms
can avoid costs associated with external financing (Froot et al., 1993).7 Also, since financial institutions have to conform to risk-based-capital requirements imposed by regulators,
reduction in risk may be a less costly alternative to raising additional capital after a negative
return shock (Froot & Stein, 1998).
The empirical evidence on hedging practices for non-banking firms is somewhat mixed.
Nance, Smith, & Smithson (1993) found weak evidence that firms with more leveraged
capital structures hedge more.8 Geczy, Minton, & Schrand (1997) find no relation between
leverage and the use of derivatives for non-financial corporations. Our results for commercial
banks are consistent with the results related to leverage. Tufano (1996) finds that managerial
incentives have an important effect on the risk-management practices in the gold-mining
industry.
The same deviations from the frictionless world of M&M that determine capital structure
of non-financial firms also affect the capital structure of banks. Diamond & Ragans (2000)
theory of bank capital contends that the optimal capital structure for banks involves tradeoffs with respect to liquidity creation, costs of bank distress, and ability to force borrower
repayment.
3. Sample selection and methodological issues

We employ the population of all insured U.S. commercial banks in our analysis. During
our test period (1989 through 1996), the number of banks declined from 12,702 to 9529
(Table 1). In total, we have almost 90,000 bank-year observations. Section four explains
how we stratify these banks to identify those specializing in real-estate lending.
7 The other reasons for hedging include a potential decrease in agency costs of managerial discretion (Stulz,
1990), improvements in executive-compensation contracts, performance evaluation, and the ability of central
headquarters to better allocate capital to internal business units due to the reduction of gains and losses due to
market exposure (DeMarzo & Duffie, 1995; Grinblatt & Titman, 1998, p. 723).
8 These results may reflect the tendency of firms facing high costs of distress to have low leverage ratios.
Table 1
Sample
Asset category ($ millions)
REB1-3
REB4+
All-total
Number of REBs
Ratio of REBs (%)
1989
1990
1991
1992
1993
1994
1995
1996
731
2,737
2,837
2,290
1,864
181
338
594
2,451
2,712
2,230
1,851
178
344
484
2,158
2,615
2,165
1,819
187
326
395
1,911
2,454
2,106
1,801
184
328
325
1,673
2,279
2,026
1,734
176
319
281
1,498
2,042
1,818
1,617
172
317
245
1,319
1,909
1,758
1,572
182
329
227
1,171
1,717
1,589
1,521
167
302
Total
10,978
10,360
9,754
9,179
8,532
7,745
7,314
6,694
010
1025
2550
50100
100500
5001000
>1000
19
76
98
123
139
18
23
11
74
101
118
152
22
13
12
64
106
82
129
15
16
9
47
75
72
95
12
15
3
47
88
90
93
7
15
7
83
150
188
177
18
19
6
72
142
222
222
13
32
9
100
214
339
326
31
34
Total
496
491
424
325
343
642
709
1,053
010
1025
2550
50100
100500
5001000
>1000
18
155
303
333
372
31
16
16
164
350
407
466
43
16
18
186
399
516
548
57
24
16
182
418
609
644
55
36
16
168
427
659
720
59
47
13
140
405
637
750
65
55
12
103
318
557
800
72
59
9
75
274
485
802
76
61
Total
1,228
1,462
1,748
1,960
2,096
2,065
1,921
1,782
12,702
1,724
13.6
12,313
1,953
15.9
11,926
2,172
18.2
11,464
2,285
19.9
10,971
2,439
22.2
10,452
2,707
25.9
9,944
2,630
26.4
9,529
2,835
29.7
Panel A: size distribution of REBs and non-REBs

Non-REB
010
1025
2550
50100
100500
5001000
>1000
Year
59
60
Table 1 (Continued )
Year
Number of banks
REB1-3
Panel B: number of REBs and non-REBs and their average assets

1989
10,978
496
1990
10,360
491
1991
9,754
424
1992
9,179
325
1993
8,532
343
1994
7,745
642
1995
7,314
709
1996
6,694
1,053
Year
REB4+
Total
Non-REB
REB1-3
REB4+
1,228
1,462
1,748
1,960
2,096
2,065
1,921
1,782
12,702
12,313
11,926
11,464
10,971
10,452
9,944
9,529
238.76
253.77
271.91
283.36
330.44
373.95
401.94
477.44
224.14
295.47
274.60
626.28
297.46
221.70**
481.24
264.72**
171.34
185.48
180.95**
188.15**
187.42**
228.31**
243.11**
267.25**
Aggregate total domestic assets (millions of dollars)

Non-REB
REB1-3
REB4+
Aggregate total real estate loans (millions of dollars)

Total
Panel C: aggregate assets and real estate loans of the banking industry and real-estate banks
1989
2,621,155
111,173
210,406
2,942,734
1990
2,629,105
145,076
271,175
3,045,373
1991
2,652,247
116,430
316,306
3,084,984
1992
2,600,964
203,541
368,783
3,173,288
1993
2,819,286
102,030
392,832
3,314,149
1994
2,896,249
142,332
471,456
3,510,037
1995
2,939,756
341,202
467,009
3,747,967
1996
3,195,965
278,751
476,246
3,950,962
Non-REB
REB1-3
REB4+
Total
584,521
599,051
613,985
580,469
657,545
672,010
669,154
745,042
51,988
67,441
51,962
85,276
44,877
62,527
147,037
126,960
102,246
135,728
158,890
181,306
198,205
236,137
236,656
239,918
738,756
802,221
824,837
847,052
900,627
970,675
1,052,848
1,111,920
Note: the sample includes all insured commercial banks that appear on NTIS file of Report of Condition and Report of Income call reports, except for banks reporting negative or
zero assets. The actual numbers of banks for subsequent tables may be slightly lower due to the data errors.
Non-REB (non-real-estate-bank) group consists of all FDIC insured commercial banks whose loans secured by real estate did not exceed 40% of their total assets for a particular year.
REB (real-estate-bank) groups consist of all FDIC insured commercial banks whose loans secured by real estate exceeded 40% of their total assets for a particular year, where the
bank is classified as a below.
REB1-3 group bank, if the bank meets the above criterion in one, two, or three other years during our sample period 19891990.
REB4+ group bank, if the bank meets the above criterion in four or more other years during our sample period 19891990.
**
Statistically different from non-REB at the 5% level or better.
Non-REB
Mean total domestic assets (millions of dollars)
61
3.1. A banks risk index

The bank safety index (or risk-index, ZRISK) is due to Hannan & Hanweck (1988).9
The empirical form of the index is:
ZRISK =
[E(ROA) + CAP]
ROA
(1)
where E(ROA) is the expected return on assets, CAP the ratio of equity capital to total
assets, and ROA the standard deviation of ROA.
Using this relationship and Chebychevs inequality, Hannan & Hanweck (1988) derive
the upper-bound probability of book-value insolvency:
Pup =
0.5 2
.
[E(ROA) + CAP]2
(2)
Although a further assumption about banks return-on-assets distribution would be needed

to calculate the exact probabilities of insolvency, this measure is consistent and informative
across different bank groups. In terms of the risk index, the (upper bound) probability of
book-value insolvency can be expressed as 1/2(ZRISK)2 . We calculate the risk index for
each individual bank using a time-series of six prior ROAs for each particular bank and
report average values for each bank group in all years. While there are alternative index
calculations, such as Eisenbeis & Kwast (1991), we believe that our construction of the risk
index is more informative and easier to interpret given the original purpose and the intuition
behind the risk index, such as the calculation of a firms probability of insolvency.10
We hypothesize that REBs will trade-off credit risk for interest-rate risk in such a way
that their risk indices will not differ from non-REBs. If risk-based capital requirements and
deposit insurance misprice interest-rate risk, the alternative hypothesis is that REBs exploit
this advantage vis-`a-vis non-REBs, resulting in greater risk-taking and higher probabilities
of book-value insolvency as captured by the risk index.
3.2. Other performance measures
To complement the risk index described above, we analyze both the on- and off-balance
sheet activities of REBs, including the use of interest-rate derivatives to determine the
9 As documented by Hannan & Hanweck (1988), this index has been widely used by various researchers. Since
then, Liang & Savage (1990), Eisenbeis & Kwast (1991), Sinkey & Nash (1993), and Nash & Sinkey (1997) have
employed it. In the banking arena, most of the papers examining bank behavior and risk-taking use a variant of
the risk index. Although a market-return measure is preferable, it is not readily available in large cross-sectional
studies, as most of our sample banks are small and not publicly traded.
10 The probability of a book-value insolvency prob(BVE < 0), of course, can differ from probability of a marketvalue insolvency. Technically, the risk index is a measure, expressed in units of standard deviations of ROA, of
how much banks accounting earnings can decline until it has a negative book value. We also can think of it as a
cushion for regulatory losses. Eisenbeis & Kwast (1991) construct risk indices for their bank groups and denote
them as typical because they are computed not as an average of individual banks indices, but rather utilizing
the cross-sectional averages of the component variables of the indices.
62
extent of their off-balance-sheet hedging behavior. On-balance-sheet substitutes for derivatives hedging include floating-rate loans and long-term borrowing. We measure exposure
to interest-rate risk using banks maturity GAPs.11 Of course, if off-balance sheet items are
used for hedging, then GAP measures do not capture true exposure to interest-rate risk as
both measures are required to draw any conclusions. The analysis is further complicated
by the fact that report-of-condition data do not permit quantitative comparisons of how
off-balance sheet activities are used for hedging since the maturities of the contracts are
highly variable. Our analysis of interest-rate risk and bank hedging behavior and patterns
captures one of our main contributions. We also update and expand various other performance measures for our sample banks beyond those presented by Eisenbeis & Kwast (1991)
and Eisenbeis, Horvitz, & Cole (1996).
4. Stratication procedures and data
We construct three groups of banks. We define a real-estate-lending bank (REB) as a
commercial bank with 40% or more of its assets in loans secured by real estate in a particular
year.12 This criterion is the same one used by Eisenbeis & Kwast (1991). Our test period
covers the years 1989 through 1996. Banks that pass the 40% test in four or more of the
eight years comprise our REB4+ portfolio.13 Banks with less than 40% of their assets in
real-estate loans comprise our benchmark or non-REB group. The banks that move above
the 40% threshold in 1, 2, or 3 years of our test period, we define as REB1-3. We focus
on and our conclusions are based on the long-term real-estate lenders, the REB4+ group.
The main advantage of this methodology is that it selects banks (the REB4+ group) with a
clearly defined asset-allocation strategy based on real-estate lending.14
We run our screens over year-end balance sheets and income-expense statements prepared by banks for their federal regulators (Call Reports or Reports of Condition and Income
database) for the population of FDIC-insured commercial banks.15 The data tapes are
supplied by the National Technical Information Services (NTIS). To analyze interest-rate
contracts held by bank holding companies, we employ Sheshunoffs database.16
11 Duration measures are, of course, preferred but they are not available from banks reports of condition. Actual
durations are probably different and not independent of interest rates, even for fixed-rate mortgages, due to
prepayment risk. Nevertheless, maturity GAP measures are still informative about the qualitative extent of bank
risk-taking.
12 Data item (RCON 1410) of the banks report of condition.
13 The criterion for our REB classification is a dynamic one. Specifically, a bank with more than 40% of its total
assets in real-estate loans in four of our 8 sample years is classified as REB4+ for those four years. For the other
four years, it is classified as a non-REB. A similar classification applies to REB1-3. The years in which a bank is
classified as a REB tend to be consecutive.
14 The banks in the REB4+ group have a clear asset-allocation strategy. Event though, the REB1-3 banks do not
have a consistent strategy, for completeness, we include the REB1-3 results in our tables.
15 Since the data files we employ are domestic only, our analysis and reported data exclude foreign activities of
our sample banks.
16 Sheshunoff Information Services is a part of the Thomson financial information services firm. It provides a BankSearch CD, a comprehensive database that includes financial information on every federally
insured bank, S&L, holding company, and credit union. Further information can be obtained from http://www.
sheshunoff.com/bksearch/bsnation.htm website.
63
Table 1 shows the number and size distribution of banks in our three, mutually exclusive
categories by year. Panel A describes the distribution of banks in different asset-size classes
while Panels B and C focus on aggregate and average assets and real-estate loans of our
bank groups. Most commercial banks, including REBs, are community banks with total
assets between $10 million and $500 million. For example, at year-end 1996, 89.6% of all
commercial banks were in this size range.
Although consolidation has led to a steady decline in the total number of commercial
banks, the number of REBs increased by 64% from 1724 (1989) to 2835 (1996). The
number of banks that became REBs increased by 63% from 1685 (1989) to 2740 (1996)
while large REBs (those with total assets greater than $1 billion) more than doubled from
39 (1989) to 95 (1996). Over this same period, the number of thrift institutions declined
from 3087 (1989) to 1924 (1996) and 96 thrifts converted to commercial banks.17 If these
banks remained traditional thrifts after converting (i.e., are classified as REBs), they would
account for only 8.6% of the increase in REBs over our test period.
Panels B and C of Table 1 show that the increases in the number of REBs during our
sample period are highlighted by 235% growth of their aggregate total domestic assets,
which increased from $321.6 billion in 1989 to $754.9 billion in 1996. This growth rate
is substantially higher than the modest 34% increase in the aggregate domestic assets of
the commercial-banking industry. On an aggregate basis, REBs accounted for 19% of the
industrys assets in 1996, almost doubling its 1989 position. A 238-percent increase in the
total amount of real-estate loans held by REBs, totaling $366.9 billion at year-end 1996,
paralleled their asset growth.
Table 2 shows the organizational structure of REBs at year-end 1992 and 1996. Organizational structure does not seem to have a major effect on whether or not a bank specializes
in real-estate lending as we find roughly similar proportions across the three most common
organizational forms of banking (i.e., independent, one-bank and multi-bank BHC).
5. Empirical ndings
Table 3 presents three measures of capital adequacy (Panel A) and our risk index (Panel
B). Panel A shows the ratio of equity capital to total assets (CAP, one of the components
of the risk index) and the total risk-based capital (RBC) ratio and tier-1 RBC ratio. The
data reveal that REBs hold substantially less capital than non-REBs, and the differences
are statistically significant at the one-percent level or better. The important implication for
our purposes is that if RBC requirements accounted for interest-rate risk, then presumably
REBs would have higher RBC ratios.
Panel B presents the risk index (ZRISK) based on the time-series approach described
in section three.18 The average bank safety index, ZRISK, increased over our test period,
which suggests an average decrease in risk. This increase in bank safety is consistent with
the observed decline in bank failures, which dropped from 206 in 1989 to six in 1996.19
17
These data are from the Office of Thrift Supervision.

It is possible to construct the aggregate index in a manner similar to Eisenbeis and Kwast, but their resulting
risk measure is hard to interpret.
19 FDIC, Annual Report (1999), p. 119.
18
64
Table 2
The organizational structure of real-estate-lending banks at year-end 1992 and at year-end 1996
Organizational structure REB type
Independent
One-bank BHC
Multi-bank BHC
Total
Panel A: year-end of 1992

Non-REB
REB1-3
REB4+
2,533
94
628
4,002
100
715
2,644
131
617
9,179
325
1,960
3,255
22.2
4,817
16.9
3,392
22.1
1,556
187
445
3,039
444
728
2,099
422
609
6,694
1,053
1,782
2,188
28.9
4,211
27.8
3,130
32.9
9,529
29.7
Total
Percent REB (%)
Panel B: year-end of 1996
Non-REB
REB1-3
REB4+
Total
Percent REB (%)
11,464
19.9
Notes: an independent bank is one not reporting any Financial High Holder Identifier (FHHI). Financial high
holder refers to the entity that is a bank holding company or a foreign banking organization, as defined in regulation
K: the relationship is direct, the basis for the relationship is equity, and the percent of equity is greater than 50%. If
only one commercial bank reports the same FHHI, this commercial bank is considered to be a part of One-Bank
BHC. In this sense, our term One-Bank BHC is in fact One-Commercial-Bank BHC, as BHCs may own
several different (non-commercial) banking institutions (e.g. savings bank, Edge Act Bank, etc.) If two or more
commercial banks report common FHHI, they are considered to be a part of a Multi-Bank BHC structure.
Source: report of condition and report of income data tapes supplied by NTIS, 19891996.
From 1989 through 1991, the average REB4+ bank was safer than the average non-REB
bank. However, after 1992, the average REB4+ bank is riskier than the average non-REB
bank.20 The values of the ZRISK index in 1996 for REB4+ group imply the upper bound
of probability of book value insolvency of an average bank is about 0.016%.21
Since 1992, we reject the null hypothesis that REBs have the same risk (probability of
insolvency) as non-REBs. The interesting thing about the break point is that risk-based
capital (RBC) requirements were fully implemented in 1992, although the change also is
likely due to improved economic conditions. This finding suggests that banks strongly react
to the regulatory incentives and the lack of RBC requirements for interest-rate risk, ceteris
paribus.
5.1. The effects of real-estate lending on bank risk measures
To complement our univariate comparisons of bank risk, we estimate a multivariate
least-squares regression to evaluate the determinants of bank risk. We use the individual
bank safety index, ZRISK, for our dependent-variable specification. Our focal explanatory
variable is exposure to real-estate lending, which we measure in two ways: the ratio of
20 Keep in mind that since ZRISK is a safety index, higher values of ZRISK mean lower risk. In this case, the
average REB has a lower index, 56 than the average non-REB, 63.
21 It is important to keep in mind that the 0.016% is a probability of an average bank, and thus it is not possible to
make a prediction about the absolute number of bankruptcies just from this probability and the number of banks
in the population.
65
Table 3
Bank capital, regulatory capital adequacy, and bank risk measures
Year
Equity to total assets

(mean, %)
Non-REB
REB1-3
Total risk-based capital

ratio (mean, %)
REB4+
Non-REB
REB1-3
Tier 1 risk-based capital

ratio (mean, %)
REB4+
Panel A: bank capital and regulatory capital adequacy

1989
9.47
7.73***
8.59*** n/a
n/a
n/a
8.42*** 16.52
11.09*** 12.18***
1990
9.46
7.83***
1991
9.51
8.13***
8.52*** 18.60
13.91*** 14.49***
1992
9.75
9.17*
8.65*** 19.21
14.51*** 14.46***
1993 10.07
9.29***
8.97*** 20.15
15.82*** 15.55***
1994 10.14
9.10***
9.06*** 20.25
15.54*** 15.73***
1995 10.93
9.52***
9.40*** 20.19
15.78*** 15.70***
1996 11.23
9.69***
9.36*** 19.95
15.53*** 15.17***
Year
Non-REB
REB1-3
REB4+
n/a
15.39
16.93
17.99
18.44
18.62
18.62
18.80
n/a
n/a
9.96*** 11.07***
12.21*** 12.91***
13.32*** 13.40***
14.20*** 13.85***
13.97*** 14.07***
14.45*** 14.12***
14.39*** 13.97***
ZRISK (safety) index mean

Non-REB
Panel B: risk measures: bank safety index, ZRISK

1989
40.76
1990
42.02
1991
45.29
1992
48.86
1993
53.23
1994
54.67
1995
59.02
1996
63.04
REB1-3
REB4+
37.14**
31.86***
33.61***
38.47***
42.06***
51.09**
54.19***
58.83***
51.03***
47.68***
47.65**
50.08
50.78**
52.19**
53.45***
56.05***
Source: report of condition and report of income data tapes supplied by NTIS, 19891996, and WRDS. Description:
Panel A: total risk-based capital ratio: total risk-based capital (RBC) by risk weighted assets (adjusted), based
on RBC definitions for risk-based premiums (RBP), variable name in RIS/WRDS database: RBCRWAJP. Tier 1
risk-based capital ratio: Tier 1 risk-based capital by risk weighted assets (adjusted), based on RBP, variable name
RBC1RWAP. Panel B: ZRISK is calculated for each individual bank from a time series of banks ROA in the
preceding 6 years using: ZRISK = [ROA + (equity-to-assets)]/[S.D. of ROA].
* Significantly different from non-REB at the 10% level.
** Significantly different from non-REB at the 5% level.
*** Significantly different from non-REB at the 1% level.
real-estate-loans to total assets (RETA) and a binary variable (REDUMMY = 1 for a REB,
and 0 otherwise). The regressions control for fixed effects, such as year, and other bank
characteristics, such as profitability (average ROA over the past 6 years), size, and geographical region.22 This analysis provides some insight into the risk-return tradeoff banks
make when deciding to specialize or not specialize in real-estate lending.
Table 4 presents our regression findings. The results for the ZRISK specifications suggest
that REBs are more risky than non-REBs for a given size and profitability level. The param22
For our geographical tests, we employed the following nine regions (states): New England (ME, NH, VT, MA,
CT, RI), Mid-Atlantic (NY, NJ, DE, PA, MD, VA, WV), Southeast (NC, SC, TN, GA, FL, AL, MS), Ohio Valley
(OH, KY, IN, IL), North Central (MI, MN, WI), Plains (IO, KN, NB, MO, ND, SD), Southwest (LA, AK, OK,
TX, NM, AZ), Rocky Mountain (NV, CO, UT, MT, WY, ID) and West Coast (CA, OR, WA).
66
Explanatory variables
Dependent variable
Intercept
ROAMEAN
RETA
ZRISK
ZRISK
62.6***
12.8***
0.132***
(1.78)
59.3*** (1.69)
(1.35)
12.9*** (1.35)
REDUMMY
(0.012)
3.93*** (0.402)
TASS
Adjusted R-square
0.440***
0.145
0.145
(0.055)
0.427*** (0.053)
Comment: The negative sign of the RETA and REDUMMY parameter estimates suggest higher risk of real-estate lending banks. Recall that; ZRISK is a safety index and
lower value means lower safety or higher probability of insolvency. Notes: REG and YEAR dummy-variables estimates not reported. Bank Risk is measured either by
a standard deviation of return-on-assets ROASTD, or a safety index ZRISK[= (CAP + ROA)/ROASTD], where CAP denotes a ratio of bank equity capital to total assets,
and ROA denotes return-on-assets. BankRisk variable = intercept + 1 ROAMEAN + 2 REvar + 3 TASS + I = 1,8 i REGi + I = 1,7 i YEARi + ; ROAMEAN denotes
banks mean return-on-assets over estimation period for any given year the estimation period is 6 previous years. REvar is a real-estate exposure measure either
RETA, a ratio of banks total real-estate loans to total assets, or REDUMMY, a dummy variable for real-estate banks as defined earlier. REG is a region dummy variable
that excludes the Mid-Atlantic region. Footnote 21 defines the regions and the states in each region. YEAR is a year dummy for each year from 1989 to 1995 (excluding
year 1996). The model is estimated with rates and ratios expressed in %, and TASS in billions. Parameter estimates of selected variables (Whites heteroskedasticity
consistent standard errors in parentheses). * Significantly different from non-REB at the 10% level. ** Significantly different from non-REB at the 5% level.
*** Significantly different from non-REB at the 1% level.
Table 4
Ordinary least squares regression estimates for models of bank risk
67
Table 5
Bank profitability
Year
ROA (mean, %)
Non-REB
ROE (mean, %)
REB1-3
REB4+
Panel A: return on assets (ROA) and return on equity (ROE)

1989
0.67
0.02**
0.89**
1990
0.68
0.16**
0.70
1991
0.79
0.28**
0.65**
1992
1.02
0.52**
0.82**
1993
1.14
0.87**
0.98**
1994
1.10
0.99**
1.04**
1995
1.10
1.06
1.13
1996
1.09
1.05
1.17**
Year
NIM (mean, %)
Non-REB
Non-REB
REB1-3
REB4+
7.55
7.43
8.56
11.24
11.97
11.64
11.11
11.03
3.72**
1.93**
1.32**
5.36**
10.86**
11.32
11.79**
11.59**
11.17**
8.32**
7.40**
9.10**
10.93**
11.34**
12.28**
12.96**
Net noninterest income (mean, %)

REB1-3
REB4+
Non-REB
REB1-3
Panel B: net interest margin (NIM) and net noninterest income (burden)
4.25**
2.52
3.00**
1989
3.96
4.14**
**
**
1990
3.88
4.15
4.20
2.47
3.05**
1991
3.93
4.21**
4.15**
2.49
3.22**
**
**
1992
4.16
4.49
4.37
2.46
3.12**
1993
4.12
4.33**
4.38**
2.42
2.75**
**
**
1994
4.17
4.35
4.49
2.42
2.67**
1995
4.08
4.20**
4.37**
2.30
2.46**
**
**
1996
4.04
4.15
4.31
2.20
2.40**
REB4+
2.64**
2.68**
2.71**
2.73**
2.70**
2.72**
2.50**
2.40**
Description: NIM (net interest margin) is defined as: (interest income interest expense)/total assets although
the burden is usually defined as (noninterest income noninterest expense)/total assets, we measure it as:
(noninterest expense noninterest income)/total assets. Source: report of condition and report of income data
tapes supplied by NTIS, 19891996.
** Statistically different from non-REB at the 5% level or better.
eter estimates for both RETA and REDUMMY are significantly negative, which means
lower safety, or equivalently, greater riskiness of real-estate banks in terms of higher probability of bankruptcy. Given the regulatory focus on capital adequacy as a measure of bank
safety and soundness, our results should prove useful for safety-net managers, especially
since most REBs as community banks that are less influenced by outside or market discipline. Hovakimian & Kane (2000) argue that regulatory discipline focuses too narrowly on
accounting measures of bank risk and equity capital. Their analysis, however, focused on
123 large BHCs for which market data were available. In contrast, we investigate the population of insured commercial banks, which overall only can be monitored with accounting
data.
5.2. Protability
Table 5 presents two accounting measures of profitability (return on assets, ROA and
return on equity, ROE, in Panel A) and the two key components of bank profitability: net
interest income (NIM) and net noninterest income in Panel B. For 1989 and 1996, REB4+
68
banks are significantly more profitable in terms of ROA, but from 1991 through 1994 they
are significantly less profitable. Since REBs use more leverage (CAP) than non-REBs, they
have a higher ROE in 4 years. Together ROA and CAP (Table 3) comprise the numerator
of the risk index.23
5.3. Net interest margin and net noninterest income
A banks net interest margin (NIM) and its net noninterest income drive its profitability,
where NIM is the difference between interest income and interest expense, scaled by assets,
while net noninterest income is the difference between noninterest expense and noninterest
income, scaled by assets.24 Panel B of Table 5 presents these data for our sample. The
results show a trade-off, that is, REBs have consistently and significantly higher net interest
margins but they have consistently and significantly lower (more negative) net noninterest
incomes (burden). The latter suggests that REBs are not more operationally efficient in
terms of controlling noninterest expenses and generating noninterest income.
Altogether, the results on profitability and net interest margins suggest that real-estate
lending is at least comparably profitable as other asset-allocation strategies. If REBs are less
risky (as presumed by regulators and lower capital requirements), absent market inefficiencies in the lending market, one would expect real-estate lending strategies to be consistently
less profitable. We leave it to future research to establish whether the REB profitability level
is due to some market inefficiency or whether it is consistent with our other results suggesting that REBs are indeed at least as risky as non-REBs.
Our results indicate that the original Basel committee regulations underrated the risk
of real-estate lending by assigning it a 50% risk weight, as opposed to the 100% weight
for the most of the other loan categories. While it is true that real-estate loans have lower
credit risk, what matters for deposit insurance pricing is the total portfolio risk that includes
interest rate risk. Of course, ignoring interest-rate risk is the culprit as the Basel Accord
focused solely on credit risk.
5.4. Geographic differences in performance
Eisenbeis et al. (1996) document that REBs in Texas performed worse than banks in
the rest of the U.S. We also found geographic differences (see footnote 21) in the performance of REBs. For example, the problems with real-estate lending in New England could
be associated with the regions high proportion of REBs, where the ratio of REBs to all
commercial banks averaged 65.3% during our sample period and remained high at 68.9%
in 1996. New Englands crisis in real-estate lending during 1989 and the early 1990s manifested itself in extremely low ROAs, which over our test period averaged 0.32% for banks
23 The third component of the risk index, ROA, showed a pattern similar to the one for ZRISK, that is, from
1989 through 1992, ROA for REB4+ was lower than the one for non-REB but after that it was higher. For 1996,
for example, the standard deviations were 0.34% (Non-REB) and 0.38% (REB4+), significantly different at the
five-percent level. On balance, the risk implications are the same as those for the ZRISK measure.
24 Since almost all banks have noninterest expenses greater than noninterest income, the net figure can be described
as a banks burden.
69
Table 6
Loan quality and measures of credit-risk exposure
Year
Loan-loss provision
(mean, % of total loans)
Non-REB
REB1-3
Net loan losses (mean, % of

total loans)
REB4+
Panel A: loan-loss provisions and net loan losses

1989
0.95
1.31**
0.51**
1990
0.83
1.51**
0.70**
**
1991
0.75
1.66
0.77
1992
0.63
0.93**
0.74**
0.49**
1993
0.37
0.57**
1994
0.27
0.29
0.34**
1995
0.33
0.26**
0.29
1996
0.44
0.31**
0.26**
Year
Past-due loans >90 days,

accruing (mean, % of total loans)
Non-REB
REB1-3
Panel B: past-due and nonaccrual loans

1992
0.44
0.56
1993
0.38
0.38
1994
0.35
0.26**
1995
0.39
0.25**
1996
0.44
0.29**
Non-REB
REB1-3
REB4+
0.83
0.70
0.68
0.55
0.32
0.24
0.27
0.35
0.88
1.09**
1.36**
0.70**
0.46
0.18**
0.17**
0.17**
0.34**
0.45**
0.56**
0.54
0.40**
0.27
0.23**
0.19**
Total nonaccrual loans

REB4+
Non-REB
REB1-3
REB4+
0.43
0.35
0.29**
0.31**
0.34**
1.15
0.94
0.74
0.71
0.70
1.61**
1.02
0.66
0.54**
0.49**
1.41**
1.25**
1.00**
0.89**
0.80**
Notes: net loan losses are defined as loan charge-offs minus loan recoveries. In Panel B, data for 19891991 are
not available. Source: report of condition and report of income data tapes supplied by NTIS, 19891996.
in the REB1-3 group and 0.38% for banks in the REB4+ group. The number of real-estate
banks increased in every region, except for New England. Since the only decline was in
New England, it might be associated with the competition REBs in that region faced from
mutual savings banks (MSBs). From year-end 1991 to year-end 1996, the number of MSBs
in New England declined from 396 to 331 (Statistics on Banking, FDIC). A similar pattern
is observed for REB1-3. This increase in the number of REBs is in sharp contrast to the
number of non-REBs, which declined by 39% from 10,978 (1989) to 6694 (1996). Most
REBs (1520, 54%) are located in three regions: Mid-Atlantic, Southeast, and Ohio Valley.
Since the Ohio-Valley, North-Central, and Plains regions had more stable economic environments over this period, ROAs for banks in these regions remained high and relatively
stable during our test period.
5.5. Credit risk and measures of loan quality
The provision for loan loss (PLL) is a non-cash outlay. It is an expense item that runs
through a banks income-expense statement to its loan-loss reserve (LLR) or allowance for
loan and lease losses (ALLL), a contra-asset account on the balance sheet. The Appendix A
describes in more detail the relationship between the PLL and the LLR. Panel A of Table 6
70
shows the PLL as a percent of total loans for our sample banks. Loan-loss provisions (as
percent of total assets, not shown in table) for REB4+ were significantly smaller than the
PLL for non-REBs only in 1989 and 1996. This result may be explained in part by the higher
average loan-to-asset ratios for real-estate banks as compared to non-REBs. Adjustment for
different loan-to-asset ratios supports this notion as REB4+ provisioned more only from
1992 to 1994 (Table 6, Panel A), a pattern that might be related to the real-estate crisis of
the late 1980s and early 1990s.25
Panels A and B of Table 6 present alternative measures of credit-risk exposure including
net loan losses, past due loans, and nonaccrual loans. In general, the REB4+ group has lower
net loan losses relative to loans than non-REBs and they exhibit less variation in this loanloss ratio. Past due loans for REBs were lower or not different from those at non-REBs. On
the other hand, nonaccrual loans were consistently higher for long-term, real-estate lenders
(REB4+) relative to the loans of non-specialized banks.
On balance, these results for bank credit quality provide mixed evidence. Judging by
some measures (such as net loan losses/total loans, Panel A, Table 6), the real-estate loans
are of higher credit quality, which supports the assumptions behind the original Basel riskbased-capital regulations. However, other measures (such as PLL/total loans, nonaccrual
loans/total loans) do not indicate superior credit quality of real-estate loans. Consistent with
our conclusions in previous sections, the reasons for a 50%, risk-based weight for real-estate
loans seem at best questionable.
5.6. Balance-sheet traits: loan, and real-estate-loan categories

Table 7 reveals that the ratio of total loans to assets was at least 13%age points larger
for REBs relative to non-REBs.26 By definition, real-estate banks had substantially higher
proportions of real-estate loans in their asset portfolios. Also, total real-estate loans for
non-REBs increased from 21.3% of assets in 1989 to 24.3% of assets in 1996, indicating a
general trend towards real-estate lending. Real-estate loans held by REB4+ are, on average,
close to 50% of their total assets. Per force, REBs are less involved in other forms of lending.
Panel B of Table 7 presents the composition of real-estate loans as a percentage of total
assets. The data reveal that more than one-half of real-estate loans, relative to total assets, are
secured by 14 family residential properties, assets relatively more exposed to interest-rate
risk.
25 An income-smoothing hypothesis suggests that banks have higher PLLs when their earnings are up (saving for
a rainy day) and vice versa. Sinkey (1998) describes the SECs challenge to a banks alleged income smoothing.
Moser (1998) analyzes the similarities and differences between PLLs and credit derivatives.
26 Results on asset composition (not shown in tables) reveal that REBs hold substantially more loans than liquid
assets. This contrasts sharply with the average portfolio of non-REBs that is split equally between loans and liquid
assets. Liquid assets are defined to include cash and interest and noninterest bearing balances, securities, federal
funds sold, and trading assets. This result cannot be attributed to the sample selection, as Non-REBs can hold
assets other than real-estate loans.
Year
Total real estate loans

(mean, % of total assets)
Non-REB
Consumer loans (mean, %

of total assets)
Commercial (loans mean,

% of total assets)
Other Loans (mean, % of

total assets)
REB1-3
REB4+
Non-REB
REB1-3
REB4+
Non-REB
REB1-3
REB4+
Non-REB
REB1-3
REB4+
Panel A: loan composition

1989
21.27
48.55**
1990
21.52
47.96**
1991
21.97
46.67**
1992
22.49
45.40**
1993
23.02
44.43**
1994
23.77
44.33**
1995
24.02
44.81**
1996
24.31
45.99**
49.13**
49.48**
49.64**
49.69**
49.98**
51.34**
50.78**
51.65**
11.22
10.90
10.24
9.63
9.72
10.27
10.28
10.38
9.44**
8.48**
7.80**
8.30**
8.18**
9.08**
8.77**
8.67**
9.14**
8.61**
7.85**
7.21**
6.94**
7.19**
7.14**
6.92**
11.10
10.58
9.63
9.04
8.93
9.30
9.69
10.02
10.88
11.37**
10.79**
10.61**
10.21**
9.62
9.13**
9.81
8.97**
9.07**
8.41**
7.97**
7.83**
7.94**
7.96**
8.30**
6.11
6.32
6.36
6.29
6.82
7.45
7.34
7.53
0.60**
0.79**
1.02**
0.95**
1.60**
2.29**
2.22**
2.18**
0.80**
0.85**
0.60**
0.57**
0.66**
0.66**
0.75**
0.76**
Year
1-4 family residential property

loans (mean, % of total assets)
Non-REB
REB1-3
Panel B: selected real-estate loan categories

1989
11.42
26.46**
1990
11.64
26.06**
1991
11.89
25.33**
1992
12.06
24.79**
1993
12.15
24.03**
1994
12.32
23.90**
1995
12.26
24.77**
1996
12.36
24.78**
Construction and land develop.

Non-farm non-residential property

REB4+
Non-REB
REB1-3
REB4+
Non-REB
REB1-3
REB4+
28.60**
28.69**
28.28**
27.98**
27.73**
28.14**
27.93**
27.91**
1.52
1.42
1.25
1.19
1.23
1.33
1.47
1.53
5.33**
5.28**
4.54**
3.77**
3.54**
3.53**
3.31**
3.70**
4.80**
4.64**
3.97**
3.58**
3.61**
3.87**
3.90**
4.24**
5.45
5.50
5.69
5.87
6.03
6.21
6.34
6.38
14.14**
14.02**
13.62**
13.48**
13.04**
12.56**
12.58**
13.36**
12.95**
13.31**
14.27**
14.79**
15.06**
15.58**
15.20**
15.74**
Note: the other two categories of real-estate loans, multifamily residential property loans, and loans secured by farm land are not reported. Source: report of condition
and report of income data tapes supplied by NTIS, 19891996.
Table 7
Loans and real-estate loans composition
71
72
Year
Distribution of sample banks all

banks
Non-REB
REB1-3
Number of banks with any holdings of

interest-rate derivatives (user banks)
REB4+
All
Non-REB
REB1-3
REB4+
Panel A: distribution of all commercial banks vs. commercial banks with IRD holdings (user banks)
1993
8,532
343
2,096
10,971
466
32
108
1994
7,745
642
2,065
10,452
434
46
103
1995
7,314
709
1,921
9,944
378
43
99
1996
6,694
1,053
1,782
9,529
330
34
91
Year
Notional amounts of interest-rate

derivatives to total assets (mean, %)
User banks to all banks (%)

All
Non-REB
REB1-3
REB4+
All
606
583
520
455
5.46
5.60
5.17
4.93
9.33
7.17
6.06
3.23
5.15
4.99
5.15
5.11
5.52
5.58
5.23
4.77
Aggregate notional amounts of Interest-rate

derivatives (millions of dollars)
Total assets (mean millions of dollars)
Non-REB
REB1-3
REB4+
All
Non-REB
REB1-3
REB4+
All
Non-REB
REB1-3
REB4+
All
Panel B: user banks

1993
63.85
1994
80.05
1995
83.01
1996
84.95
11.75**
15.76**
30.08*
9.28**
12.67**
16.54**
18.30**
13.66**
51.98
63.75
66.31
65.04
7,159,406
9,856,448
10,165,605
13,387,130
21,199
10,541
892,967
13,844
31,764
57,841
36,889
25,753
7,212,369
9,924,830
11,095,462
13,426,727
4,265
4,941
5,648
7,377
2,018
1,584
5,842
3,701
1,295
1,955
1,900
2,003
3,618
4,149
4,951
6,027
Year 1996
Aggregate notional amounts of IR

derivatives (millions of dollars)
Panel C: interest-rate derivatives held by the top bank holding companies## (BHCs) in 1996
Top BHCs (all)
12,985,323
Top BHCs/excl. largest 10
454,457
130,811
82,985
Aggregate total assets

(millions of dollars)
Number of affiliated banks
4,057,994
2,206,401
1,585,877
1,253,835
3,140
3,033
2,896
2,781
Sources: Panels A and B: report of condition and report of income data tapes supplied by NTIS, 19891996. Panel C: BankSearch, Sheshunoff Information Services
Inc., Holding Companies, version 2.8, Data: December 1996.
# IRDs include: futures and forward IR-contracts, exchange traded and OTC IR-options (purchased and written), and IR-swaps.
## Top bank holding company (BHC) refers to the highest financial holder of the bank group, and data are from consolidated statements. There are 1314 highest financial
holder BHCs on BankSearch, Sheshunoff BHC database that reported total assets and IR-derivatives in 1996.
* Statistically different from non-REB at the 10% level or better.
Table 8
The use of interest-rate derivatives (IRDs)#
73
5.7. Managing interest-rate risk

Sinkey & Carter (1997) and others have shown that most commercial banks do not use
derivatives. Table 8, which presents the use of interest-rate derivatives, reveals that this is
also true for REBs. Only about 5% of commercial banks use interest-rate derivatives, which
is also true for real-estate lenders. We identify about 100 banks in the REB4+ category
out of some 2000 banks that report holding any interest-rate derivatives. Despite higher
interest-rate exposure, REBs are no more likely to use interest-rate derivatives (IRDs) than
non-REBs. While REBs do not appear to increase their derivatives use in response to their
asset strategy, they do use floating-rate loans to manage a portion of their interest-rate risk.
Thus, the derivatives they do use might be a crucial portion of their overall risk-management
strategy.27
Banks that use derivatives rank among the largest in the industry. For example, in 1996,
their average total assets were $2 billion for the REB4+ group, $3.7 billion for REB1-3, and
$7.4 billion for non-REB. The aggregate notional amounts of IRDs held by commercial
banks totaled $13.4 trillion in 1996, while REBs held only $39.6 billion of this amount
(0.3%). This disparity can be traced to the large amounts of IRDs held in trading accounts
by money-center banks, which do not tend to specialize in real-estate lending.
For banks affiliated with holding companies, potential hedging activities may originate in
other bank or non-bank subsidiaries so that the consolidated BHCs risk exposure is reduced
or offset. To investigate this possibility, we use the 1996 Sheshunoff database to examine
the notional amounts of IRDs of consolidated bank holding companies. Panel C of Table 8
presents the findings. At year-end 1996, the total notional value of IRDs held by BHCs was
almost $13 trillion.28 Further investigation reveals that a major part of these activities is
attributable to the 10 largest BHCs (with 107 affiliated banks) as they hold approximately
$12.5 trillion in IRDs. Excluding the 30 largest BHCs, which have 359 affiliated banks, the
remaining BHCs with 2781 affiliated banks report only $83 billion in notional amounts of
IRDs. Since the potential benefits and efficiencies resulting from hedging by a consolidated
BHC may accrue only to about one-third of all banks and since the volume of IRDs rapidly
decreases as BHC size decreases, we conclude that the majority of banks do not enjoy
hedging protection of a bank-holding company.
One of our most interesting findings is a potential peso problem facing real-estate
banks stemming from the large proportion of fixed-rate loans they hold. Table 9 shows
that REBs make greater use of floating-rate loans to manage interest-rate risk,29 including
greater use of adjustable-rate mortgages. Nevertheless, the ratio of average total fixed-rate
loans to total assets for real-estate lending banks is greater than the ratio for banks with
smaller exposure to usually long-term, real-estate loans. The asset composition of real-estate
banks makes them vulnerable to unexpected increases in interest rates. For example, the
average real-estate bank held almost 52% of its total assets in real-estate loans with more
27 A Journal referee has suggested that to the extent that changes in the level of derivatives might provide
information about REB decision-making at the margin, our results could be driven by what the average REB does.
28 These data are for the highest financial holder of the banking group. Since it is not uncommon for one BHC
to own another BHC, using the highest holding-company code avoids double counting.
29 Floating-rate loans make up a greater proportion of total loans for REBs than for less-specialized banks.
74
Table 9
The use of floating-rate loans and adjustable rate real-estate loans
Year
Total fixed-rate loans

Non-REB
REB1-3
Total floating-rate loans

REB4+
Non-REB
Panel A: fixed- and floating-rate loans as a percent of total assets

37.80**
17.19
1989
33.18
39.02**
1990
33.30
37.41**
37.86**
16.67
1991
32.68
36.02**
36.81**
16.17
1992
31.66
34.69**
35.49**
16.48
1993
32.17
35.30**
35.61**
17.08
1994
33.41
38.18**
35.88**
18.19
1995
33.05
38.32**
35.70**
19.07
1996
33.95
40.16**
37.08**
19.02
Year
REB1-3
REB4+
30.59**
31.29**
30.15**
31.16**
29.89**
27.99**
27.37**
27.24**
31.03**
30.77**
30.17**
30.52**
30.38**
31.98**
31.59**
31.20**
Adjustable-rate family mortgages#

(mean, % of total mortgages)##
Total floating-rate loans

Non-REB
REB1-3
REB4+
Panel B: relative use of floating-rate and ARMs

45.02**
1989
33.15
43.38**
44.66**
1990
32.38
45.17**
**
1991
32.44
44.91
44.86**
1992
33.51
47.08**
46.18**
**
1993
34.01
46.12
46.14**
1994
34.72
42.64**
47.57**
**
1995
36.12
41.99
47.29**
1996
35.40
40.76**
46.07**
Non-REB
REB1-3
REB4+
23.02
23.02
23.25
23.47
23.36
23.30
24.63
24.30
30.96**
31.18**
33.16**
30.59**
32.89**
34.20**
34.69**
35.97**
36.12**
35.10**
# Adjustable rate loans secured by 14 family residential properties.
## Real-estate loans secured by 14 family residential properties.
than 50% of them in residential property loans, only one-third of which had adjustable rates.
At year-end 1996, the average REB had 18% of its assets in fixed-rate mortgages.30
Although we do not have data for the duration of bank loans, a reasonable estimate of
the duration of a banks real-estate loan portfolio is 10 years. Assuming a 10-year duration
for the average mortgage, a 1% increase in interest-rates would lead to approximately a
10% decrease in the market value of a REBs mortgages and a 1.8% decrease in the market
value of its assets, with a non-negligible effect on its net worth. Given that the S&L crisis
of the 1980s was attributed mainly to the excessive interest-rate risk of thrifts, our results
suggest the potential for a similar crisis among REBs.
Assessment of interest-rate risk and its potential effect on a banks earnings and net
worth requires data on the maturity or repricing of its assets and liabilities. We use maturity
This figure is derived as follows: fixed-rate mortgages = 18% = (27.91%) (135.10%), where from Panel B of
Table 7 the ratio of residential loans/total assets (year 1996, REB4+) is 27.91% and from Panel B of Table 9 the ratio
for adjustable-rate family mortgages (year 1996, REB4+) is 35.10%. The comparable figure for non-real-estate
banks in 1996 is: proportion of fixed-rate mortgages = 9.5% = (12.36%) (123.3%).
30
75
Table 10
GAP analysis: rate-sensitive assets and rate-sensitive liabilities#
Year
Rate-sensitive
assets (mean, % of
total assets)
Non-REB
Panel A
1989 49.77
1990 48.12
1991 46.09
1992 45.29
1993 45.19
1994 44.20
1995 47.20
1996 44.70
Rate-sensitive
liabilities (mean,
% of total assets)
GAP12 interest-rate
measure (mean, % of
total assets)
REB1-3
REB4+
Non-REB
REB1-3
REB4+
Non-REB
REB1-3
REB4+
53.38**
52.16**
50.51**
50.56**
50.51**
46.28**
47.84
45.49
54.76**
53.29**
51.52**
50.80**
50.29**
49.30**
50.39**
47.31**
75.31
76.01
76.45
75.23
74.35
73.92
72.78
72.51
78.24**
78.20**
78.16**
75.69
74.48
74.73**
74.85**
74.77**
76.74**
78.13**
78.27**
77.09**
75.78**
75.23**
74.97**
75.18**
25.54
27.89
30.37
29.95
29.16
29.72
25.58
27.81
24.86
26.06**
27.79**
25.13**
23.97**
28.46
27.01**
29.28**
21.98**
24.84**
26.75**
26.29**
25.49**
25.93**
24.57**
27.87
** Statistically different from non-REB at the 5% level; GAP analysis performed for commercial banks with
domestic offices only.
# Denitions: GAP12 interest-rate measure is defined as the difference between rate-sensitive assets and ratesensitive liabilities; rate sensitive assets assets repricing within 12 months = (federal funds sold) + (securities
purchased under agreements to resell) + (customers liability) + (trading assets) + (fixed and floating debt securities maturing or repricing within 12 months) + (fixed and floating loans maturing or repricing within 12 months);
rate sensitive liabilities liabilities repricing within 12 months = (federal funds purchased) + (securities sold
under agreements to repurchase) + (banks liability on acceptances executed and outstanding) + (trading liabilities) + (other borrowed money) + (demand notes issued to the U.S. Treasury) + (time and savings deposits) (large
long-term time deposits); time and savings deposits = all non-transaction and transaction deposit accounts with
the exception of demand deposits (e.g. with the exception of deposits with zero interest); large long-term
deposits = time deposits of $100,000 or more with remaining maturity or repricing frequency of more than 1
year; small longer-term deposits = time deposits of less than $100,000 and remaining maturity or repricing frequency of more than 1 year; GAP12 best estimate = (GAP12 proxy) plus (small longer-term deposits): e.g.,
GAP12 adjusted for the small, longer-term deposits originally included in the rate sensitive liabilities. This is our
best estimate of the GAP12. Call reports contain data on small, longer-term deposits for years after 1996 only.
GAP analysis, a common tool of interest-rate-risk management, to compare the interest-rate

risk of REBs and our group of less-specialized banks.31 Table 10 presents data on banks
12-month GAP positions measured as rate-sensitive assets minus rate-sensitive liabilities.32
The average REB4+ bank decreased the amount of its short-term assets from 54.8% in 1989
to 47.3% in 1996, while rate-sensitive liabilities decreased only slightly from 76.7% in 1989
to 75.2% in 1996. This trend results in a deteriorating negative GAP position, which declined
from 22% of total assets in 1989 to 28% in 1996. These data also reveal that REBs do,
to some extent, manage their interest-rate risk from an on-balance sheet perspective, as their
GAPs are better than GAPs of less specialized banks.
31 Duration is a more precise measure of interest-rate risk but it is not readily available. In any case, GAP measures
provide useful (at least) qualitative assessment of banks interest-rate risk.
32 Table 10 provides our working definition of rate-sensitive assets and rate-sensitive liabilities. Following
Flannery & James (1984), GAP12 is defined as the difference between rate-sensitive assets and rate-sensitive
liabilities with remaining maturity or repricing frequency of less than 12 months.
76
A word of caution, however, is in order. Since banks are not required to report their GAP
positions, our GAP proxy is only an estimate. However, beginning in 1996, banks were
required to report an additional detail in the form of small, longer-term deposits, which
permits construction of a more precise estimate of a banks GAP. These data, not shown
in tabular form, reveal that the addition of small, longer-term deposits (e.g., 9.0% of total
assets for REB4+) to our GAP measures only marginally reduces the observed negative gap
positions to a range of a negative 1921%. This value corresponds to the already reported
proportion of real-estate bank assets held in fixed-rate residential mortgages.
As a tool of risk management, securitization provides banks with a way to remove risk
from their balance sheets. To investigate banks securitization activities, we analyze data
on loans transferred with recourse.33 Since data on loans sold without recourse are not
available, the data we report are only an approximation of banks securitization activities.
We find little securitization activity by our sample banks and, not unexpectedly, domination
by large banks. For 1996, we identify only 175 banks (1.8% of the population), with average
assets of $8.3 billion that report any loan-trading activity. Over the period 19891996, we
have 88,027 bank-years for which mortgages sold (with recourse) are reported as zero
compared to 1275 bank-years with mortgages sold reported. These numbers pale in relation
to the $1.1 trillion in total real-estate loans held by all commercial banks at year-end 1996,
of which $367 billion were held by REBs (Panel C, Table 1).
On balance, we conclude that, over our test period, REBs did not take full advantage of
securitization as a risk-management tool and, similar to derivatives activities, securitization
was dominated by the largest banking organizations. While the failure of REBs to take
full advantage of these risk-management techniques is puzzling, it is not inconsistent with
the dichotomy that exists with respect to risk management between large complex banking
organizations (LCBOs) and community banks.
6. Summary, conclusions, and policy implications

Structural changes in financial markets and in various segments of the financial-services
industry, especially the savings-and-loan component, have presented opportunities and
threats to commercial banks. Many bankers, especially those in local communities, have
reacted by shifting their portfolios toward real-estate lending or housing finance. For example, the number of real-estate-lending banks has increased from 1724 at year-end 1989 to
2835 at year-end 1996. This finding is especially noteworthy given the substantial consolidation in the banking industry, which paralleled the contraction of the thrift industry.
Opportunities for regulatory capital arbitrage, reform of federal deposit insurance, and
recruitment by the FHLB have provided additional incentives for banks to become more
heavily involved in real-estate lending.
A governmental safety net, unless priced and administered properly, enhances riskshifting opportunities for insured depositories. This paper analyzes bank asset structure,
33 Two variables we analyze from the call reports are: (1) loans (mortgages) transferred (i.e., sold or swapped)
with recourse that have been treated as sold for call-report purposes and (2) amount of recourse exposure on these
mortgages as of the report date.
77
risk-taking behavior, and financial risk management of U.S. commercial banks from 1989
through 1996. Focusing on real-estate lending banks (REBs), we find that they have higher
probabilities of insolvency relative to less-specialized banks, a finding that links to the
extent of their real-estate lending. Given the construction of our bank risk index, the higher
riskiness of REBs may be due either to their natural underlying risk exposures or to their
higher risk-level (risk-preference) targets. We also show that REBs are better than the other
banks in managing their interest-rate risk using on-balance sheet techniques but there are
no differences in the use of the off-balance sheet tools.
Collectively, our findings suggest that banks react strongly to regulatory incentives,
which may have unintended evils by raising rather than lowering the economic riskiness
of individual banks. Our analysis is consistent with notions that weaknesses in capital
regulation and deposit-insurance pricing encourage banks to engage in regulatory capital
arbitrage and potentially exploit deposit-insurance subsidies. The increasing number of
banks specializing in real-estate lending, the distorted incentives of regulatory discipline,
and weaknesses in risk-reduction strategies could, in a sharp interest-rate upswing, stress
the condition of the banking system and the FDIC.
We close with some suggestions for future research and some policy concerns. First,
issues dealing with historical, long-term bank insolvencies and how they relate to the bank
level of interest-rate risk-taking present opportunities for future research.34 The S&L crisis
of the 1980s provides an important lesson that neglected interest-rate risk-taking may lead
to economy-wide crises. Work by Kane (1989) and others document that the demise of the
thrift industry was attributed to excessive interest-rate risk and the mispricing of deposit
insurance. The number of commercial bank failures was relatively low and decreasing
during our sample period. For example, 127 banks failed in 1991, 41 in 1993, six in 1995
and the failures stayed in single digits for the rest of the decade.
Second, as REBs comprise only a relatively small part of banking sector assets, we do
not claim that we present evidence of a serious danger of rising systemic risk, defined as
the risk that an event will trigger a loss of economic value or confidence in a substantial
portion of the financial system that is serious enough to have significant adverse effects on
the real economy.35 However, these tendencies deserve attention as smaller banks account
for a disproportionate share of bank loans to small businesses and their failures account for
a disproportionate share of losses to the deposit insurance fund (Gilbert & Sierra, 2003).
Third, risk-based capital requirements encourage banks to engage in regulatory capital
arbitrage, which Jones (2000, p. 36), among others, describes as opportunities for banks
to reduce substantially their regulatory measures of risk with little or no corresponding
reduction in their overall economic risk. By substituting residential mortgages for assets
with higher regulatory-capital requirements, REBs reduce their risk-weighted assets and
lower their capital requirements. In doing so, however, their interest-rate exposure increases.
Based on our findings, this substitution of reduced credit risk for increased interest-rate risk
has resulted in an increase in the overall riskiness of REBs compared to non-REBs. Our
34
We thank a Journal referee for pointing out this topic for further research. A careful examination of bank failures
deserves more space than we can provide in the present paper. Further research should investigate insolvencies
over long-term interest-rate cycles.
35 For a definition, see G10, report on consolidation in the financial sector, January (2001) chapter 3.
78
results confirm Giddys (1985) view that the imposition of regulatory capital requirements
for credit risk would drive banks to substitute noncredit risks such as interest-rate and
operating risks for credit risk. If regulatory discipline (e.g., in the form of risk-based capital
requirements) was more effective, then banks that substitute interest-rate risk for credit risk
should not be able to engage in regulatory capital arbitrage. Consistent with Hovakimian &
Kanes (2000) findings for large BHCs, our results suggest risk-shifting behavior at the bank
level as REBs hold substantially less capital (based on various measures) than non-REBs.
The policy implication seems clear: banks and regulators neglect the interest-rate risk at
their own peril.
On balance, regulatory capital arbitrage and the distorted incentives of regulatory discipline offer plausible explanations for the increase in the number of banks pursuing real-estate
lending strategies and their risk-taking. However, other explanations also are plausible and
provide avenues for future research. Tufano (1996) finds that risk-management practices
are negatively associated with the tenure of firms CFOs. Also, corporate-control issues,
limited access to hedging instruments, and an entrenchment hypothesis grounded in lack
of managerial interests, skills, or preferences might explain why REBs have not adopted
financial-engineering techniques extensively.
Acknowledgements
We thank Scott Barnhart, Ekkehart Boehmer, David Carter, Zack Fulmer, Stephen Jamison, Jim Linck, Marc Lipson, Richard Stehle, Shawn Thomas, and Larry Wall for helpful
comments on earlier drafts of the paper. We also thank Marsha J. Courchane and the participants of the Whither the Community Bank? conference, Federal Reserve Bank Chicago,
March 2003. Part of the work for this paper was completed while Sinkey was the Edward W.
Hiles Professor of Financial Institutions at the University of Georgia and a Visiting Scholar
at the Federal Reserve Bank of Atlanta.
Appendix A. Understanding bank loan-loss reserves: the bathtub analogy and the
role of the securities and exchange commission (SEC) (adapted from Sinkey, 1998)
Think of a banks loan-loss reserve (LLR) as the water level in a bathtub, where the
level reflects the reserves needed to cover future loan losses. The water level, which is a
key variable as it serves as a buffer for absorbing loan losses, can be raised by turning on
a spigot, labeled the provision for loan loss or PLL. The PLL is a discretionary, noncash
outlay that flows into the LLR account on a banks balance sheet. The LLR, which is also
called the allowance for loan and lease losses (ALLL), is a contra-asset account that is
deducted from gross loans to get net loans. When a bank charges off bad loans because
borrowers cannot or will not repay, reserves go down the drain and the water level in the
tub falls. Through collections, however, recoveries are made on some of these bad loans.
Thus, only net loan losses go down the drain and deplete the level of the reserve.
To recap, when loans are charged off as losses, they deplete the loan-loss reserve;
the provision for loan loss is an inflow that augments the stock or level of the reserve.
79
The periodic provision (usually quarterly) is an important variable because it is subject to managerial discretion. Since setting the loan-loss reserve is part art and part
science, stock analysts, investors, auditors, bank regulators, and the SEC watch it
closely.
Because the PLL is subject to managerial discretion, it can be used to manipulate reported
earnings to make them look smoother than they actually are. The SEC is concerned about
this because variability of earnings is a key measure of risk for any business, and risk, of
course, affects value. A bank can smooth its reported earnings by using its PLL judiciously.
When times are good and earnings are up, the bank over-reserves and reports lower earnings;
when times are bad and earnings are down, the bank under-reserves and reports earnings
higher than it could have without the prior over-reserving. The overall result is a smoother
earnings pattern.
The sinister view of this is that the bank is manipulating its earnings to report smoother
earnings to enhance shareholder value. An alternative view is that the bank is behaving like
the hard-working squirrel that stores nuts for the winter. It certainly makes sense to save for
a rainy when the weather is good.
The real problem here is anyone who focuses on accounting earnings. Richard Breeden,
a former chairman of the SEC, said that bank financial statements should have Once
Upon A Time stamped on them. To the extent that the SEC is trying to make bank
accounting earnings less of a fairy tale and more reflective of real earnings, it is to be
applauded.
Usually, however, it is chronic under-reserving that concerns bank regulators and the
SEC the most because this practice overstates a banks earnings and capital. If a bank overreserves, it understates its earnings and capital. Since stock markets value high earnings and
strong capital positions, it is difficult to make a compelling case for over-reserving, smoothing earnings notwithstanding. If bankers do a better job of matching reported earnings with
economic or real income, the divergence between book and market values of loans will be
reduced. Since a banks earning assets are dominated by loans, loan quality is a critical
determinant of a banks earnings and net worth. Under- and over-reserving must be judged
relative to the overall quality of a banks loan portfolio. The most serious case is a bank
with lots of bad loans that is under-reserving.
References
DeMarzo, P., & Duffie, D. (1995). Corporate incentives for hedging and hedge accounting. Review of Financial
Studies, 8, 743772.
Diamond, D. W., & Ragan, R. G. (2000). A theory of bank capital. Journal of Finance, 55, 24312465.
Eisenbeis, R. A., & Kwast, M. L. (1991). Are real estate depositories viable? Evidence from commercial banks.
Journal of Financial Services Research, 524.
Eisenbeis, R. A., Horvitz, P. M., & Cole, R. A. (1996). Commercial banks and real estate lending: The Texas
experience. Journal of Regulatory Economics, 10, 275290.
Esty, B. C. (1997). Organizational form and risk taking in the savings and loan industry. Journal of Financial
Economics, 44, 2555.
Esty, B. C. (1998). The impact of contingent liability on commercial bank risk taking. Journal of Financial
Economics, 47, 189218.
80
Esty, B. C., Tufano, P., & Headley, J. (1994). Banc one corporation: Asset and liability management. Journal of
Applied Corporate Finance (Fall), 7(3), 3365.
Federal Deposit Insurance Corporation. (1999). Annual report (fdic.gov) 1998. Washington, DC.
Flannery, M. J., & James, C. M. (1984). The effect of interest-rate changes on the common stock returns of financial
institutions. Journal of Finance, 39, 11411153.
Froot, K. A., & Stein, J. C. (1998). Risk management, capital budgeting, and capital structure policy for financial
institutions: An integrated approach. Journal of Financial Economics, 47, 5582.
Froot, K. A., Scharfstein, D. S., & Stein, J. C. (1993). Risk management: Coordinating corporate investment and
financing policies. Journal of Finance, 48, 16291658.
Geczy, C., Minton, B. A., & Schrand, C. (1997). Why firms use currency derivatives. Journal of Finance, 52,
13231354.
Giddy, I. (1985). Regulation and off-balance sheet banking. The search for nancial stability: The past fty years
(pp. 165177). Working paper, Federal Reserve Bank of San Francisco.
Gilbert, R. A., Sierra G.E. (2003). The nancial condition of U.S. banks: How different are community banks?
Working paper, Federal Reserve Bank of St. Louis.
Grinblatt, M., & Titman, S. (1998). Financial markets and corporate strategy. Irwin/McGrawHill.
Group of Thirty. (1993). Derivatives: Practices and principles, New York group of thirty.
Hannan, T. H., & Hanweck, G. A. (1988). Bank insolvency risk and the market for large certificates of deposit.
Journal of Money, Credit and Banking, 203211.
Hogan, A. M. B., & Malmquist, D. H. (1999). Barriers to depository uses of derivatives: An empirical analysis.
Journal of Multinational Financial Management, 9, 419440.
Hovakimian, A., & Kane, E. J. (2000). Effectiveness of capital regulation at U.S. commercial banks, 1985 to 1994.
Journal of Finance, 55, 451468.
John, K., John, T. A., & Senbet, L. W. (1991). Risk-shifting incentives of depository institutions: A
new perspective on federal deposit insurance reform. Journal of Banking and Finance, 15, 895
915.
Jones, D. (2000). Emerging problems with the basel capital accord: Regulatory capital arbitrage and related issues.
Journal of Banking and Finance, 24, 3558.
Kane, E. J. (1989). The S&L insurance mess: How did it happen? Working paper. Washington, DC: The Urban
Institute Press.
Kane, E. J. (1995). Three paradigms for the role of capitalization requirements in insured financial institutions.
Journal of Banking and Finance, 19, 431459.
Liang, J. N., & Savage, D. T. (1990). The nonbank activities of bank holding companies. Federal Reserve Bulletin,
280292.
Merton, R. C. (1977). An analytic derivation of the cost of deposit insurance and loan guarantees. Journal of
Banking and Finance, 1, 311.
Merton, R. C. (1995). Financial innovation and the management and regulation of financial institutions. Journal
of Banking and Finance, 19, 461481.
Modigliani, F., & Miller, M. H. (1958). The cost of capital, corporation finance, and the theory of investment.
American Economic Review, 48, 261297.
Moser, J. T. (1998). Credit derivatives and loan-loss provisions: How they differ and how they are the same.
Working paper, Federal Reserve Bank of Chicago.
Nance, D. R., Smith, C. W., Jr., & Smithson, C. W. (1993). On the determinants of corporate hedging. Journal of
Finance, 48, 267284.
Nash, R. C., & Sinkey, J. F., Jr. (1997). On competition, risk, and the hidden assets in the market for bank credit
cards. Journal of Banking and Finance, 21, 89112.
Schrand, C., & Unal, H. (1998). Hedging and coordinated risk management: Evidence from thrift conversions.
Journal of Finance, 53, 9791013.
Sinkey, J. F., Jr. (1998). A bathtub view of loan-loss reserves. American Banker, (3), 34.
Sinkey, J. F. Jr., & Carter, D. A. (1997). Derivatives in U.S. banking: Theory, practice, and empirical evidence
(pp. 4178). Derivatives, Regulation and Banking, North-Holland Publishing Co.
Sinkey, J. F., Jr., & Nash, R. C. (1993). Assessing the riskiness and profitability of credit-card banks. Journal of
Financial Services Research, 7(2), 127150.
81
Smith, C. W., & Stulz, R. (1985). The determinants of firms hedging policies. Journal of Financial and Quantitative
Analysis, 20, 391405.
Smithson, C. (1996). Surveys surveyed. Risk 9 (No. 3, March 1996, pp. 5657).
Smithson, C. (1996b). Who uses what. Risk 9 (No. 5, May 1996, pp. 5051).
Stulz, R. (1990). Managerial discretion and optimal financing policies. Journal of Financial Economics, 26, 328.
Tufano, P. (1996). Who manages risk? An empirical examination of risk-management practices in the gold-mining
industry. Journal of Finance, 4, 10971137.
Whidbee, D. A., & Wohar, M. (1999). Derivative activities and managerial incentives in the banking industry.
Journal of Corporate Finance, 5, 251276.

Blasko 2006

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Blasko 2006

Diunggah oleh

Hak Cipta:

Format Tersedia

The Quarterly Review of Economics and Finance

Bank asset structure, real-estate lending, and

Department of Accounting and Finance, School of Business Administration,

2. Theoretical, empirical, and banking-industry background

ing decrease in economic risk, which presumably is the intention of capital-adequacy

3. Sample selection and methodological issues

Asset category ($ millions)

Panel A: size distribution of REBs and non-REBs

Panel B: number of REBs and non-REBs and their average assets

Aggregate total domestic assets (millions of dollars)

Aggregate total real estate loans (millions of dollars)

Statistically different from non-REB at the 5% level or better.

Mean total domestic assets (millions of dollars)

3.1. A banks risk index

Although a further assumption about banks return-on-assets distribution would be needed

These data are from the Office of Thrift Supervision.

Panel A: year-end of 1992

Equity to total assets

Total risk-based capital

Tier 1 risk-based capital

Panel A: bank capital and regulatory capital adequacy

ZRISK (safety) index mean

Panel B: risk measures: bank safety index, ZRISK

Panel A: return on assets (ROA) and return on equity (ROE)

Net noninterest income (mean, %)

Net loan losses (mean, % of

Panel A: loan-loss provisions and net loan losses

Past-due loans >90 days,

Panel B: past-due and nonaccrual loans

Total nonaccrual loans

5.6. Balance-sheet traits: loan, and real-estate-loan categories

Total real estate loans

Consumer loans (mean, %

Commercial (loans mean,

Other Loans (mean, % of

Panel A: loan composition

1-4 family residential property

Panel B: selected real-estate loan categories

Construction and land develop.

Non-farm non-residential property

Distribution of sample banks all

Number of banks with any holdings of

Notional amounts of interest-rate

User banks to all banks (%)

Aggregate notional amounts of Interest-rate

Total assets (mean millions of dollars)

Panel B: user banks

Aggregate notional amounts of IR

Aggregate total assets

Number of affiliated banks

5.7. Managing interest-rate risk

Total fixed-rate loans

Total floating-rate loans

Panel A: fixed- and floating-rate loans as a percent of total assets

Adjustable-rate family mortgages#

Total floating-rate loans

Panel B: relative use of floating-rate and ARMs

GAP analysis, a common tool of interest-rate-risk management, to compare the interest-rate

6. Summary, conclusions, and policy implications

Anda mungkin juga menyukai