LEARNING BANKS' EXPOSURE TO SYSTEMATIC RISK: EVIDENCE FROM THE FINANCIAL CRISIS OF 2008
Author | Ariel M. Viale,Jeff Madura |
Published date | 01 February 2014 |
Date | 01 February 2014 |
DOI | http://doi.org/10.1111/jfir.12029 |
LEARNING BANKS’EXPOSURE TO SYSTEMATIC RISK: EVIDENCE FROM
THE FINANCIAL CRISIS OF 2008
Ariel M. Viale and Jeff Madura
Florida Atlantic University
Abstract
Using a two‐state Markov regime‐switching intertemporal capital asset pricing model,
we find that the exposure to systematic risk of bank stocks varies with size and the state of
the economy. Across large banks, those with higher net interest margins and lower capital
ratios are the most exposed to unexpected shifts in the term structure of interest rates.
Small banks with higher net interest rate margins and that rely less on noninterest income
are the most exposed to unexpected shifts in the stance of monetary policy. We apply the
asset pricing model out of sample to assess its ability to detect troubled banks during a
financial crisis.
JEL Classification: G120, G210
I. Introduction
There were various indications that the market pricing of bank risk was overly optimistic
before the financial crisis of 2008 occurred. Commercial banks were aggressively
pursuing expansion into mortgage securitization without recognition that a housing
market bubble had developed. Their diversification of business operations may have
reduced their transparency and created a false perception that their exposure to a weak
economy was limited. The securitization of mortgages and other debt instruments
allowed them to reduce their capital and may have created a false perception that
banks were insulated from mortgage defaults. Banks increasingly used short‐term funding
(with asset‐backed commercial paper) to finance long‐term investments in mortgages.
Credit rating agencies consistently assigned high ratings to pools of mortgages of
questionable quality that were securitized. According to Goodhart (2008), central banks,
the Bank for International Settlements, and the International Monetary Fund expressed
their concerns about how the market was underestimating bank risk before the financial
crisis.
The market may have underestimated the risk exposure of banks because of the
presumption that these banks can employ risk management strategies based on derivatives
(see Instefjord 2005; Lepetit et al., 2008). In this regard, Gersbach and Wenzelburger
(2008) develop a model where banks attempt to fully incorporate systematic risk into the
We are especially grateful to the associate editor Ken Cyree and an anonymous referee whose comments and
suggestions helped us improve the article. The views expressed are those of the authors. All remaining errors are
their own responsibility.
The Journal of Financial Research Vol. XXXVII, No. 1 Pages 75–97 Spring 2014
75
© 2014 The Southern Finance Association and the Southwestern Finance Association
RAWLS COLLEGE OF BUSINESS, TEXAS TECH UNIVERSITY
PUBLISHED FOR THE SOUTHERN AND SOUTHWESTERN
FINANCE ASSOCIATIONS BY WILEY-BLACKWELL PUBLISHING
pricing of their loans. Their research shows that the premiums incorporated by banks to
account for systematic risk in the pricing of their loans were too small and that banks may
have severe exposure if their capital levels are insufficient. Moreover, the risk perception
of some banks before the financial crisis of 2008 may have been clouded by the view that a
two‐tier (too big to fail) regulatory policy exists.
If banks are relatively less transparent than other firms so that investors are unable
to correctly price their exposure to systemic risk, how effective is market discipline? We
attempt to determine whether rational investors equipped with a modified (i.e., regime
switching) version of Merton’s (1973) intertemporal capital asset pricing model (ICAPM)
(as implemented by Petkova 2006) would have detected the financial distress of banks that
failed during the financial crisis. We apply a four‐factor model that accounts for market
risk, shocks to the term structure of interest rates, shocks to the default spread, and shocks
to the stance of monetary policy with a state‐dependent asset pricing equation conditional
on good and bad economic times.
The first step in our analysis is to estimate an empirical asset pricing model for
bank stock returns. Next, we assess whether the two‐state Markov regime‐switching
(MRS)–ICAPM is able to correctly detect the ex ante risk exposure of large and small
public banks that failed or were rescued during the financial crisis of 2008. We calculate
the cost of equity out of sample for each of these banks and compare it with the cost of
equity of its corresponding benchmark portfolio sorted by size. We also allow for a
structural break in the data‐generating process (DGP) of stock returns during bad
economic times and repeat the comparative analysis by calculating the implied cost of
equity of each bank in response to the financial crisis.
Our study is related to that of Fahlenbrach, Prilmeier, and Stulz (2012), who
evaluate the relative performance of banks during the 2008 financial crisis versus their
relative performance during 1998 when the Long Term Capital Management (LTCM)
fund was rescued after Russia defaulted on its debt. They also control for bank
characteristics and find the performance is worse during the 2008 financial crisis for banks
that were smaller, had high book‐to‐market values, and had low betas. Their focus is on
the ability to identify ex ante risky banks from their past performance based on a previous
crisis. According to their hypothesis, bank managers adapt slowly to the changing
economic environment. Fahlenbrach, Prilmeier, and Stulz focus on a bank managers’
learning model, which may identify the bank characteristics that can cause some banks to
be persistently exposed to risk. Their assessment of bank systematic risk is focused on the
tail of the distribution of stock returns.
Conversely, our focus is on whether bank failures could have been anticipated by
investors during the financial crisis of 2008 using a standard dynamic asset pricing model
in finance. We focus on the investors’learning model to identify the risks inherent in the
banking function. Our analysis uses the information contained in the whole (bimodal)
distribution of stock returns. We develop an intertemporal asset pricing model over a
long‐term period, which explicitly accounts for low‐frequency structural breaks in stock
returns.
It can be argued that because bank operations constantly change, asset pricing
models will not serve as effective ex ante predictors of a bank’sfinancial distress.
However, our regime‐switching model includes a learning mechanism that accounts for
76 The Journal of Financial Research
To continue reading
Request your trial