Did Investors Herd during the Financial Crisis? Evidence from the US Financial Industry

Published date01 March 2018
Date01 March 2018
AuthorM. Humayun Kabir
School of Economics and Finance, Massey University, Palmerston North, New Zealand
We examine the herding behavior of investors in the US nancial industry,
especially commercial banks, S&Ls, investment and insurance rms during global
nancial crisis of 2008 towards own sub-sector and market consensus using
augmented cross sectional absolute deviation of returns (CSAD) model. After
distinguishing between fundamental and non-fundamental information, we nd
agreaterinuence of global nancial crisis on spurious herding for commercial
and investment banks, and such herding increases in the down market and with
conditional volatility of returns, but adverse herding is prevalent among investors
during normal period in response to fundamental information. We also nd that
herding intensity on fundamental information is relatively high with market
consensus for all nancial institutions except insurance rms in high volatility
regime, and intentional herding is only signicant and limited to S&Ls and
investment banks in high volatility regime. Our ndings suggest limited
spillover effects of herding when investors face non-fundamental information.
JEL Codes: G11; G12; G21; G29
Accepted: 30 May 2017
Herding behavior emerges when individuals ignore or suppress their own belief
or private signal and follow the actions of other individuals or investors. Early
studies focus on rational herding showing that such behavior may be optimal
as individuals, who are followed by others may possess important information
(Bikhchandani et al. 1992), and managers or analysts want to protect reputation
(Scharfstein and Stein 1990; Trueman 1994; Graham 1999) or insure against
underperformance with peers (Rajan 2006). However, Froot et al. (1992) show
that naïve investors may herd in short investment horizon to exploit the
informed investors. Other studies depart from investorsrationality and put more
emphasis on the presence of irrational and psychological or sentiment that affect
trading behavior leading to herd phenomena (Shleifer and Summers 1990;
DeLong et al. 1991; Cipriani and Guarino 2005). If the correlated behavior of
investors due to suppression of private information leads to the situation when
© 2017 International Review of Finance Ltd. 2017
International Review of Finance, 2017
DOI: 10.1111/ir.12140
Did Investors Herd during the
Financial Crisis? Evidence from
the US Financial Industry
International Review of Finance, 18:1, 2018: pp. 59–90
DOI:10.1111/irfi .12140
© 2017 International Review of Finance Ltd. 2017
the market price fails to reect fundamental information, herding can result in
extreme effects in mispricing and inefciency in the asset market. Studies based
on trading data by institutional investors nd mixed evidence of herding
(Lakonishok et al. 1992; Grinblatt et al. 1995; Nofsinger and Sias 1999; Sias
2004; Choi and Sias 2009; Brown et al. 2013; Holmes et al. 2013).
Methodologically different from such studies, Gleason et al. (2004), Tan et al.
(2008), Zhou and Lai (2009), Chiang and Zheng (2010), Gębka and Wohar
(2013), and Chiang et al. (2013) use aggregate market data to detect herd
behavior by investors across markets or sectors following Chang et al. (2000).
The purpose of the present study is to examine the herding behavior of
investors in the US nancial industry, especially commercial banks, savings and
loan institutions (S&L), investment and insurance rms during global nancial
crisis of 2008 following the methodology of Chang et al. (2000). The epicenter
of the 20072008 nancial crisis was the banks and other nancial institutions.
The crisis was triggered by the revelation of losses faced by major nancial
institutions from subprime mortgages. Later, subsequent losses due to
counterparty exposures resulted in systemic risk with collapse and near-failure
of major banks. Beltrattia and Stulz (2012) nd that the buy-and-hold dollar
return of larger banks worldwide was 51.84% with standard dev iation of
27.74% during the period of 20072008 crisis, which is extremely high. The
interconnectedness and the complexity of nancial structure and institutions
have contributed to such systemic failure. For example, although insurance
industry did not have direct exposure to the nancial crisis, American
International Group (AIG), the worlds largest insurance company, was on the
verge of collapse. The fact is that AIG was a complex nancial group, consisting
of 70 US-based insurance companies and another more than 170 other nancial
service companies. The complexity of such interconnections created through
credit default swaps and securities lending resulted in AIG to be an important
counterparty to other systemically important banks. The losses and systemic risk
during 20072008 reached to such an extent that Fed and Treasury had to come
up with the plan to rescue banks, and prevent disorderly winding of exposed
nancial institutions.
Treasury established several programs under TARP to
stabilize banking institutions, restart credit markets, and avoid foreclosures.
1 More discussion is presented in literature review in section II.
2 A comprehensive nancial crisis timeline and policy actions have been documented by
Federal Reserve Bank of St Louis, and available online: https://www.stlouisfed.org/nancial-
3 Congress initially authorized $700 billion for TARP in October 2008, which was reduced to
$475 billion by the Dodd-Frank Wall Street Reform and Consumer ProtectionAct (Dodd-Frank
Act). Approximately $250 billion of that amount was committed in programs to stabilize
banking institutions, $27 billion was committed through programs to restart credit markets,
$70 billion was committed to stabilize American International Group (AIG), and Treasury
purchased $20 billion in preferred stock from two institutions, Citigroup Inc. and Bank of
America. Treasury also provided capital to 707 nancial institutions in 48 states, including
more than 450 small and community banks and 22 certied community developmentnancial
institutions (CDFIs). The largest investment was $25 billion and the smallestwas $301,000.
International Review of Finance
© 2017 International Review of Finance Ltd. 20172
International Review of Finance
60 © 2017 International Review of Finance Ltd. 2017
Given the extent and depth of the crisis, it is worthwhile to examine how
investors buying and selling nancial stocks behave in the market when nancial
institutions were facing hurdles as their interdependence through different
channels exerting systemic risks. Did the investors participate based on their
own belief or swayed away by other investors when defaults on mortgages spread
to investment banks and commercial banks via an elaborate network of
derivatives? Did the ripple effects of the demise of some nancial institutions
as well as the evaporation of liquidity through the interaction of market liquidity
and funding liquidity result in nervousness in investorspsyche generating
market transactions following each other? Previous studies concentrate
examining mostly herding in national markets leaving sector-level herding,
especially in the nancial sector which was vital during the nancial crisis of
Diamond and Dybvig (1983), Swary (1986), and Lang and Stulz (1992) have
shown how a bank-specic event or bank-specic trouble can create shock to
other banks and affects the whole nancial sector. Similarly, investors
expectations, beliefs, and memories of past crises (Masson 1998 and
Mullainathan 1998), and information asymmetry between the informed and
uninformed investors (Clavo 1999) can generate contagious effects in response
to a crisis. James (1991) suggests that the direct costs of bank failures are larger
than that of bankruptcy of non-nancial rms. Acharya and Yorulmazer (2008)
develop a model to explore various aspects of systemic risk analyzing the ex-ante
effects of bank failures and losses, and likelihood of information contagion that
induce prot maximizing bank owners to herd with other banks.
Herding behavior is manifested in an increased similarity of returns across
stocks resulting in a lower cross-sectional variability of returns. Chang et al.
(2000) propose a measure of cross-sectional variability of returns, CSAD, cross-
sectional absolute standard deviation with respect to the average return of all
stock returns. Chiang et al. (2013) nd that herding is time-varying and depends
on prevailing market returns and conditional volatility. Motivated by Chiang
et al. (2013), we augment the base CSAD model to explore the possibility of
herding towards own sub-sector as well as the market (for example, commercial
banks herding towards the consensus or equally weighted returns of all
commercial banks, and equally weighted returns of CRSP rms). However, Gębka
and Wohar (2013) argue that there might be an increased dispersion in returns
across assets leading to adverse herding when investors overemphasize their
own view or focus on views dominant among subset of actors excessively
ignoring market information. We also explore such possibility, which might be
due to localized herding (Gębka and Wohar 2013), excessive ight to quality
during market stress (Favero and Giavazzi 2002; Kaminsky et al. 2004; Baur and
4Gębka and Wohar (2013) nd no evidence of herding using DataStream sectoral indices in a
global scale. They also nd that some sector-specic indices like basic materials, consumer
services, and oil and gas reveal tradersirrationality, which could be due to overcondence,
or excessive ight to quality.
Do Investors Herd during the Financial Crisis?
© 2017 International Review of Finance Ltd. 2017 3
Did Investors Herd during the Financial Crisis?
© 2017 International Review of Finance Ltd. 2017 61

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT