Risk Overhang and Loan Portfolio Decisions: Small Business Loan Supply before and during the Financial Crisis

• Author: GӦKHAN TORNA,ANDREW WINTON,ANNE GRON,ROBERT DEYOUNG
• Published date: 01 December 2015
• Date: 01 December 2015
• DOI: http://doi.org/10.1111/jofi.12356

THE JOURNAL OF FINANCE •VOL. LXX, NO. 6 •DECEMBER 2015

Risk Overhang and Loan Portfolio Decisions:

Small Business Loan Supply before and during

the Financial Crisis

ROBERT DEYOUNG, ANNE GRON, GӦKHAN TORNA, and ANDREW WINTON∗

ABSTRACT

We estimate a structural model of bank portfolio lending and find that the typical U.S.

community bank reduced its business lending during the global financial crisis. The

decline in business credit was driven by increased risk overhang effects (consistent

with a reduction in the liquidity of assets held on bank balance sheets) and by reduced

loan supply elasticities suggestive of credit rationing (consistent with an increase

in lender risk aversion). Nevertheless, we identify a group of strategically focused

relationship banks that made and maintained higher levels of business loans during

the crisis.

SMALL BUSINESSES,DEFINED AS HAVING fewer than 500 employees, employ about

one-half of the U.S. labor force and create nearly two-thirds of net new private

sector jobs in the United States annually (U.S. Small Business Administration

(2014)). Virtually all of these small firms are privately held and lack access to

public capital markets. Toensure access to credit, these informationally opaque

businesses establish close borrower–lender relationships with small “commu-

nity banks” (e.g., Petersen and Rajan (1994), Berger et al. (1997), Berger et al.

(2005)). This confluence of small firms and small banks is important for macroe-

conomic growth both in the United States and elsewhere: Berger, Hasan, and

Klapper (2004) find a strong positive link between a large, healthy small bank-

ing sector and macroeconomic growth across 49 developed and developing

nations.

The financial crisis took a toll on the U.S. small banking sector. About 6%

of all commercial banks and thrift institutions failed between 2007 and 2012,

∗DeYoung is at Kansas University. Gron is at NERA Economic Consulting. Torna is at State

University of New York at Stony Brook. Winton is at University of Minnesota. The opinions

expressed in this paper do not necessarily reflect the views of NERA Economic Consulting. We

thank three anonymous referees, Allen Berger, Lamont Black, Paolo Fulghieri, Ted Juhl, Evren

Ӧrs, Michael Roberts (the Editor), Elu von Thadden, Greg Udell, and seminar participants at

Bangor University, the Bank of Canada, the Federal Deposit Insurance Corporation, the Federal

Reserve Bank of Chicago, the Federal Reserve Bank of New York, the Financial Intermediation

Society,the University of Groningen, the University of Kansas, the University of Limoges, and the

University of Mannheim for their insightful comments and suggestions. The opinions expressed in

this paper do not necessarily reflect the views of NERA Economic Consulting. The authors have no

material financial or non-financial interests related to this research, as identified in the Journal

of Finance’s disclosure policy.

DOI: 10.1111/jofi.12356

2451

2452 The Journal of Finance R

and 411 of the 478 insolvencies were small institutions with assets less than

$1 billion.1As the FDIC fashioned resolutions for these insolvent banks, it is

understandable that some of their small business clients would suffer interrup-

tions to, reductions in, or even outright loss of their credit lines.2But whether

the stress of the financial crisis caused healthy banks in the United States

to reduce credit to small and medium enterprises (SMEs) remains an open

question. A reduction in SME credit supply by healthy banks—that is, a credit

crunch or credit rationing—would have had procyclical effects, exacerbating

the economic downturn by denying firms the short-term credit necessary to

finance increased inventories and retain workers. This would be antithetical to

the idea of small business relationship banking, which carries with it the pre-

sumption that additional credit will be available when needed. In this paper,

we investigate whether, how, and why small U.S. banks reduced small business

credit supply during the financial crisis.

Some evidence has emerged from European economies, where credit reg-

istries provide researchers highly detailed data on loans and loan applications,

that the financial crisis was accompanied by reduced credit supply to SMEs

(e.g., Popov and Udell (2012), Cotugno, Monferra, and Sampagnaro (2012),

Jimen´

ez et al. (2012)). These studies document that credit supply declined

more during the crisis as banks experienced financial stress (low levels of eq-

uity capital, poorly performing loan portfolios) but to a lesser extent for SMEs

with strong bank–borrower relationships. While this evidence is informative,

it remains incomplete. First, none of the extant studies examine SME credit in

the United States, where research has focused on the syndicated loan supply

to large firms during the crisis (e.g., Ivashina and Scharfstein (2010a,2010b),

Chodorow-Reich (2014)) due to a lack of systematic loan-level data for SMEs.

Given that the business, banking, and financial environments in the United

States and Europe are substantially different, one cannot assume that small

business credit supply behaved similarly on both sides of the Atlantic. Sec-

ond, while the studies cited above find well-identified statistical associations

between macroeconomic conditions and SME credit supply, they stop short of

modeling the underlying lender behaviors that drive these empirical associa-

tions and the channels through which these associations might occur. Third,

these studies typically do not differentiate banks by their business strategies,

an important distinction that could drive banks’ lending behaviors during eco-

nomic downturns.

We derive a theoretical loan supply function from a model of loan portfolio

optimization with market imperfections, following Froot and Stein (1998). In

our model, a capital-constrained bank (i.e., one facing imperfect capital mar-

kets) might reject an otherwise profitable lending opportunity if it cannot make

1Data from the Federal Deposit Insurance Corporation website.

2The FDIC arranged “purchase and assumption” resolutions for 427 of these failed banks. In

these transactions, the FDIC arranges for a healthy bank to acquire all of the assets of the failed

bank, so clients of these failed banks were unlikely to fully lose access to credit. In the other 51

bank insolvencies, the FDIC seized the failed bank’s assets and disposed of them piecemeal over

time; clients of these banks were more likely to fully lose access to new credit.

Risk Overhang and Loan Portfolio Decisions 2453

room on its balance sheet by cheaply selling off any of its existing loans (i.e.,

imperfect loan markets) and the returns on its existing loans covary positively

with the returns on the new lending opportunity. The illiquidity of its existing

loans, which we characterize as “loan overhang,” combined with costly external

capital effectively makes the bank risk averse, that is, less likely to make risky

but positive NPV loans. Because capital and credit markets tend to become

more imperfect during recessions, our theory predicts increased lender risk

aversion and reduced SME loan supply during the financial crisis. Thus, in the

course of testing empirically whether U.S. banks reduced and/or rationed credit

to SMEs during the financial crisis, we also perform an important empirical

test of financial intermediation theory.

We estimate the model using quarterly observations of community banks

with assets less than $2 billion (2010 dollars) operating in U.S. metropoli-

tan markets between 1991 and 2010.3These banks are too small to make or

participate in loans to large firms, and they tend to hold mixed portfolios of

consumer loans, real estate loans, and SME loans. Importantly, these banks

are typically consistent with the maintained assumptions of our theory model:

SME loans are illiquid assets and must be held in portfolio where they lock

up scarce equity capital; small banks are seldom publicly traded, rarely have

public credit ratings, and face relatively inelastic deposit markets, all of which

make external capital expensive; and the owners of these banks tend to invest

a disproportionate share of their (or their family’s) wealth in the bank, and this

lack of diversification further encourages risk-averse business practices (Spong

and Sullivan (2007)).

Our empirical results largely confirm the predictions of our theory. We find

strong evidence of loan overhang effects. All else equal, banks make fewer

new business loans when their portfolios contain large amounts of preexisting

business loans, and make more new business loans when their portfolios con-

tain large amounts of loans to other sectors (e.g., consumer loans) that covary

negatively with business loans. Consistent with increases in loan illiquidity

and lender risk aversion during times of economic uncertainty, the negative

impact of overhanging business loans on new business lending is substantially

stronger during the financial crisis. Regarding the main question of our inves-

tigation, we find that new SME lending declined during the financial crisis for

the average bank in our data, and this finding looks like credit rationing: prior

to the crisis there is a strong positive relationship between business loan sup-

ply and the expected returns on business loans, but during the crisis business

loan supply became insensitive (perfectly inelastic) to expected loan returns.

This new evidence complements the finding of Montorial-Garriga and Wang

(2012) that small business borrowers were more likely to be rationed out of the

bank loan market during the financial crisis than large firms. Importantly, we

identify a small set of banks that strategically focused on making illiquid loans

3For decades, both bank regulators and bank researchers used $1 billion as a convenient upper

size threshold to define the U.S. community bank sector (DeYoung,Hunter, and Udell (2004)). Our

$2 billion threshold is similarly convenient, but recognizes several decades of inflation.