Bank Sectoral Concentration and Risk: Evidence from a Worldwide Sample of Banks

• Published date: 01 September 2022
• Author: Thorsten Beck,Olivier De Jonghe,Klaas Mulier
• Date: 01 September 2022
• DOI: http://doi.org/10.1111/jmcb.12920

DOI: 10.1111/jmcb.12920

THORSTEN BECK

OLIVIER DE JONGHE

KLAAS MULIER

Bank Sectoral Concentration and Risk: Evidence

from a Worldwide Sample of Banks

We propose a novel,stock-return based, technique to measure three aspects

of banks’ sectoral concentration that feature prominently in episodes of bank

risk: specialization (capturing high exposures),differentiation (capturing de-

viation from peer banks), and nancial sector exposure (capturing direct

connectedness) and show external validity for these measures. Wend that

both individual and systemic bank risk decrease with specialization. Dif-

ferentiation is particularly and positively related to individual bank risk,

whereas direct connectedness of banks is particularly and positively related

to systemic bank risk. These ndings inform the theoretical and policy de-

bate on the relationship between sectoral concentration and banks’ stability.

JEL codes: G01, G21, G28, L5

Keywords: bank concentration, bank risk, differentiation, factor model,

sectoral specialization, systemic stability

Declaration of interest: none. The authors would like to thank the editor Bob DeYoung, two anony-

mous referees, Lamont Black, Charles Calomiris, Fabio Castiglionesi, John Duca, José Liberti, Martin

Melecky, Geoffrey Miller, Phong Ngo, Alex Popov, Joao Santos, Glenn Schepens, Jason Sturgess, Wolf

Wagner,and seminar participants at the European Central Bank, FEBS conference in Surrey,the NYU-Law

“2014 Law & Banking/Finance conference,” the Sydney Banking and Financial Stability Conference, the

2018 IBEFA-ASSAconference, the Norges Bank, the NY Fed, dePaul University, University of Glasgow,

National University of Singapore, Nanyang Business School, Singapore Management University,CESifo-

Munich, Australian National University, University of Melbourne, Massey University, Cardiff Business

School, University of Edinburgh Business School, Universita Cattolica de Milano, Frankfurt School of

Finance and Management, and Essex Business School for interesting discussions and helpful comments.

This paper started as a background paper for the World DevelopmentReport 2014 entitled Risk and Op-

portunity: Managing Risk for Development. The viewsexpressed are solely those of the authors and do not

necessarily represent the views of the World Bank or the National Bank of Belgium. This paper won the

CoPFiR Awardfor the Best Policy-relevant Research Paper at the rst JRC-EC-Community of Practice in

Financial Research workshop.

T B is a Professor at the Florence School of Banking and Finance,EUI, and CEPR Re-

search Fellow(E-mail: Thorsten.Beck@eui.eu). O D J is at the National Bank of Belgium,

and an Associate Professor at Tilburg University(E-mail: olivier.dejonghe@nbb.be). K M is

an Associate Professor at Ghent University (E-mail: klaas.mulier@ugent.be).

Received June 5, 2020; and accepted in revised form May 12, 2021.

Journal of Money, Credit and Banking, Vol. 54, No. 6 (September 2022)

© 2022 The Authors. Journal of Money, Credit and Banking published by Wiley Periodicals

LLC on behalf of Ohio State University.

This is an open access article under the terms of the Creative Commons Attribution License,

which permits use, distribution and reproduction in any medium, provided the original work

is properly cited.

1706 :MONEY,CREDIT AND BANKING

A   W  . (2004), nine of the 13 major individ-

ual and systemic banking crises of the twentieth century were caused by credit con-

centration of banks. Theoretical models trying to rationalize this are typically built

around one or more of the following aspects: (i) specialized versus diversied asset

portfolios, (ii) differentiated versus nondifferentiated asset portfolios (vis-à-vis peer

banks), and (iii) counterparty risk within the nancial system.1However, theory has

not given a clear answer to howthese different aspects relate to bank stability. Further-

more, trying to underpin this anecdotal evidence with solid empirical relationships,

let alone testing the theoretical models, has proven difcult because we lack appro-

priate data on banks’ asset concentration—especially in a uniform way across banks

and countries. Such exercise is further complicated by the fact that concentration can

arise on both the asset and liability side, or even off-balance sheet.

This paper lls this gap by constructing comprehensive, stock-return based mea-

sures of concentration that overcome several shortcomings of existing measures. We

develop measures of specialization (are banks’ exposures diversied or not?), differ-

entiation (are banks’ exposures different from peer banks?), and nancial sector ex-

posure (how directly connected is a bank with the nancial sector?), using a uniform

setup for banks from different countries and provide external validity for these new

measures. We then include these three dimensions of bank concentration in models

explaining individual bank risk and systemic bank risk and test competing theories

on the relationship between concentration and stability.

New concentration measures. In the rst part of the paper, we develop a new

methodology to identify banks’ strategic choices with respect to concentration in sec-

toral exposures. Weapply this methodology to estimate the sectoral concentration of

1,716 banks across 34 countries over the period 2002–12. The underlying assumption

of this methodology is that one can identify a bank’s sectoral concentration choices

from the covariation between its stock returns and the returns on sectoral indices, and

thus relies on market participants’ information on bank choices. Earnings calls tran-

scripts, for instance, indicate that analysts do have information about sectoral expo-

sures. During earnings calls, analysts ask questions in terms of actual exposure to and

performance across economic sectors but also in terms of hedging instruments used

to hedge against sectoral concentration. We provide several examples from earnings

calls transcripts in Section 1.1. We use an extended factor model and relate banks’

stock returns to the returns on nine global sectoral indices, a nancial sector index,

and a set of common factors.2This allows us to see whether a bank’s exposures are

1. See Diamond (1984) and Boyd and Prescott (1986) for seminal work on the benets of diversi-

cation or Winton (1999) for theoretical and Acharya, Hasan, and Saunders (2006) for empirical work on

the benets of specialization. See, for example, Acharya and Yorulmazer (2007, 2008), Wagner (2010),

Allen, Babus, and Carletti (2012), Greenwood, Landier,and Thesmar (2015), or Cai et al. (2018) for work

on differentiation. See, for example, Allen and Gale (2000), Freixas, Parigi, and Rochet (2000), Gorton

and Metrick (2012), or Allen and Carletti (2013) for work on counterparty risk.

2. These are the returns on the global market index, the local market index, a real estate investment

trust, and the global small-minus-big, high-minus-low, and momentum factor.

THORSTEN BECK, OLIVIER DE JONGHE, AND KLAAS MULIER :1707

well diversied (meaning that its returns are only exposed to the set of common fac-

tors) or whether a bank is also signicantly exposed to certain sectors (meaning that

its returns exhibit signicant exposures to sectoral indices over and above the other

factors in the model).3

We then dene bank sectoral specialization as the percentage variation of the

bank’s stock returns that is incrementally explainedby the sector-specic indices over

and above the variation explained by the other factors. Next, we dene bank sectoral

differentiation as the Euclidean distance between a bank’s estimated sectoral expo-

sures and the average sectoral exposures of all other banks in the same country and

year. Finally,we dene a bank’s nancial sector exposure as the estimated sensitivity

of its stock returns to the returns on the nancial sector index.

To shed more light on the informational content of our measures, we hand-collect

actual sectoral lending exposures from the notes to the annual statements for a small

subsample covering the largest banks in our sample. We then show external validity

for our newly created measures of sectoral specialization, differentiation, and nan-

cial sector exposure by documenting statistically and economically signicant corre-

lations with their accounting-based counterparts.

Strengths and limitations. Our approach has a number of advantages compared

to other data or methods used in the literature. First, our identied exposures relate

to the lending exposures of banks, but capture more. They also take into account

banks’ securities holdings and derivative positions through which banks might hedge

excessive sectoral lending exposures (or, alternatively, create such positions when

there is no sectoral lending exposure). Furthermore, they also account for sectoral

exposures at the liability side of banks’ balance sheets (e.g., sectoral concentration

in corporate deposits). This is important given the increasing complexity of banks

and their increasing focus on nonlending activities (Demirguc-Kunt and Huizinga

2010). A second advantage is that our approach allows to cover a signicantly wider

range of banks and countries than in previous studies, which either relied on credit

register data for single country studies (e.g., De Jonghe et al. 2020) or relied on syn-

dicated loan exposures of large, international (mainly U.S.-based) banks (e.g., Cai

et al. 2018). On average, the banks in our sample cover nearly 70% of all banking as-

sets in their respective countries, which increases the external validity of the results

and relevance for policymakers. A third advantage is that this methodology can be

applied to identify a wide range of strategic bank choices that are usually either only

available to insiders or not available in a homogeneous way (through, for instance,

earnings calls). Our methodology can be used to construct time-varying indicators of

a bank’s exposure to certain sectors (as in this paper), but it is quite general and could

also be implemented to determine a bank’s exposure to certain geographical areas,

3. This approach is similar to returns-based analysis used to deconstruct mutual fund returns in ex-

posures to investment strategies or asset classes. This offers an indirect identication of exposures that

are otherwise difcult to observe, like sectoral exposures, which cannot be easily deducted from nancial

statements or annual reports. Researchers have used this to identify mutual fund exposures to large vs.

small stocks or value vs. growth stocks (see, e.g., Sharpe (1992) and Brown and Goetzmann (1997)).