intra-industry effects of negative stock price surprises

19
ORIGINAL RESEARCH Intra-industry effects of negative stock price surprises Aigbe Akhigbe Jeff Madura Anna D. Martin Ó Springer Science+Business Media New York 2014 Abstract We find that a pronounced stock price decline of one firm yields negative valuation effects for industry rivals, on average. We test whether the impact is conditioned on a measure of default likelihood of rivals derived from the option pricing framework. The stock price contagion effects are more pronounced for rivals with the greatest default likelihood. The contagion effects are also conditioned on the degree of the surprise, characteristics of the firm experiencing the negative surprise (such as its relative size), characteristics of the rival firms (such as their similarity to the firm experiencing the negative surprise), and characteristics of the corresponding industry (such as degree of concentration). The sensitivity of industry rivals and portfolios to negative stock price surprises changes during the 2007–2008 financial crisis, which may be because stocks had already been priced to reflect pessimistic outlooks, or because the market anticipated restructuring or government intervention that could prevent the collapse of firms with the greatest default likelihood. Keywords Negative surprise Intra-industry Default likelihood Distress JEL Classification G30 G33 A. Akhigbe College of Business Administration, University of Akron, Akron, OH 44325, USA e-mail: [email protected] J. Madura College of Business, Florida Atlantic University, Boca Raton, FL 33431, USA e-mail: [email protected] A. D. Martin (&) Tobin College of Business, St. John’s University, Jamaica, NY 11439, USA e-mail: [email protected] 123 Rev Quant Finan Acc DOI 10.1007/s11156-014-0446-4

Upload: anna-d

Post on 23-Dec-2016

212 views

Category:

Documents


0 download

TRANSCRIPT

ORI GINAL RESEARCH

Intra-industry effects of negative stock price surprises

Aigbe Akhigbe • Jeff Madura • Anna D. Martin

� Springer Science+Business Media New York 2014

Abstract We find that a pronounced stock price decline of one firm yields negative

valuation effects for industry rivals, on average. We test whether the impact is conditioned

on a measure of default likelihood of rivals derived from the option pricing framework.

The stock price contagion effects are more pronounced for rivals with the greatest default

likelihood. The contagion effects are also conditioned on the degree of the surprise,

characteristics of the firm experiencing the negative surprise (such as its relative size),

characteristics of the rival firms (such as their similarity to the firm experiencing the

negative surprise), and characteristics of the corresponding industry (such as degree of

concentration). The sensitivity of industry rivals and portfolios to negative stock price

surprises changes during the 2007–2008 financial crisis, which may be because stocks had

already been priced to reflect pessimistic outlooks, or because the market anticipated

restructuring or government intervention that could prevent the collapse of firms with the

greatest default likelihood.

Keywords Negative surprise � Intra-industry � Default likelihood � Distress

JEL Classification G30 � G33

A. AkhigbeCollege of Business Administration, University of Akron, Akron, OH 44325, USAe-mail: [email protected]

J. MaduraCollege of Business, Florida Atlantic University, Boca Raton, FL 33431, USAe-mail: [email protected]

A. D. Martin (&)Tobin College of Business, St. John’s University, Jamaica, NY 11439, USAe-mail: [email protected]

123

Rev Quant Finan AccDOI 10.1007/s11156-014-0446-4

1 Introduction

Research has investigated whether a firm’s financial distress can emit a negative signal

throughout the industry. Earlier studies provide evidence that corporate defaults are cor-

related (e.g., Lucas 1995; Pedrosa and Roll 1998). More recently, research shows that

default contagion may occur through firm-specific and industry variables (e.g., Chava and

Jarrow 2004; Lando and Nielsen 2010; Huang and Lee 2013). Das et al. (2007) offer three

explanations for default clustering that are also applicable to the proposition of intra-

industry stock price effects of distress, which is the focus of our study. First, firms in the

same industry are likely exposed to common risk factors. Second, firms within the industry

may have direct cash flow linkages. Third, when investors learn about the distress of one

firm, they may more closely scrutinize corresponding industry competitors that are exposed

to potential distress.

Lang and Stulz (1992), Ferris et al. (1997), Jorion and Zhang (2007) and Hertzel et al.

(2008) examine whether the negative signal associated with a bankruptcy can trigger stock

price effects for rival firms in the corresponding industry. These studies predominantly

show that the average stock price response of industry rivals to bankruptcies is negative

and significant, supporting an intra-industry contagion effect.

Jorion and Zhang (2007) mainly focus on the impact of bankruptcy filings and large

jumps in CDS spreads on the CDS spreads of rival firms, but they also evaluate the impact

on rival stock prices. They find that Chapter 11 (Chapter 7) bankruptcy events elicit

significant intra-industry contagion (competitive) CDS spread effects, but do not find

significant stock price effects from these bankruptcy events. They also find that large jumps

in CDS spreads result in strong contagion effects, in terms of both CDS spread and stock

price effects on rivals, and assert that they are due to the unanticipated nature of these

events.

Our study uniquely examines whether firms that experience a substantial decline in their

stock price generate intra-industry stock price effects. Certainly bankruptcies are more

severe, but a pronounced stock price decline may be an early signal of financial distress

and/or a signal that the firm is experiencing adversity, thus may also transmit industry

information. Based on the strong contagion effects in response to unanticipated jump

events documented in Jorion and Zhang (2007), we expect that negative stock price sur-

prises likely trigger significant contagion effects since these shocks are also unanticipated.

While negative surprise events are not as severe as bankruptcies, they do occur more

frequently. We identify 2,368 firms with a minimum of $1 billion market capitalization

between 1998 and 2011 that have experienced a one-day stock price drop of 15 % or more,

which is nearly one major stock price drop across relatively large firms every day. In

comparison, Hertzel et al. (2008) draw their sample of 250 firms from a population of

1,695 bankruptcy filings between 1978 and 2004, which equates to about five filings per

month on average. The events ultimately included in their sample occur less than once per

month. Furthermore, their sample has an average distress-day abnormal return of -26 %,

whereas we focus on a less stringent requirement of a -15 % nominal change in stock

price and do not require the firm to eventually file for bankruptcy. The typical information

content of a bankruptcy filing reveals a firm that has already been subject to very weak

performance and a declining stock price, and seeks a formal restructuring of its debt as a

desperate strategy to rebound. Conversely, the typical firms in our sample are not neces-

sarily even close to bankruptcy, and many of them were performing well prior to the

negative surprise. Yet, even though the negative surprise is not as severe as a bankruptcy

signal, we believe that it can transmit contagion effects throughout the industry. Our

A. Akhigbe et al.

123

assessment of more common events offers insight on the transmission of stock price effects

that occur even without extreme events such as bankruptcies.

One explanation for the previously documented negative intra-industry stock price

effects due to bankruptcy filings is that the bankruptcy draws attention to the default

likelihood of all firms in the industry, causing a contagion effect. Furthermore, market

participants may reduce their projections of expected future cash flows for all firms in the

industry, as the bankruptcy announcement could signal a shrinking industry. Thus, we

hypothesize that rival firms of negative surprise firms experience significantly negative

valuation effects. Additionally, as the large stock price drop may prompt the market to

focus attention on the default likelihood of rivals and/or on adverse industry conditions, we

hypothesize that greater contagion stock price effects occur for those rivals with greater

default likelihood.

Of the previous studies that examine intra-industry stock price effects of distress, only

the study by Lang and Stulz (1992) provides empirical evidence that financial leverage (as

a proxy for default likelihood) has some influence on the contagion stock price effects.1

They report that rival portfolios with above median degrees of leverage experience sig-

nificantly negative abnormal returns, but do not detect leverage as a significant factor in

cross-sectional analyses on rival responses. Jorion and Zhang (2007) and Hertzel et al.

(2008) acknowledge that a small sample size can make it difficult to detect significant

cross-sectional factors. In our cross-sectional analyses, we address two key limitations of

these previous studies. First, by examining negative surprises that are not as severe as

bankruptcies, we assess a very large sample, which enables us to uncover significant cross-

sectional factors. Second, we estimate the default likelihood from an option pricing

framework and use it as a cross-sectional factor. While this volatility-adjusted leverage

measure is frequently used in default contagion literature (e.g., Das et al. 2007; Duffie et al.

2007; Lando and Nielsen 2010; Huang and Lee 2013), it has not yet been used to explain

intra-industry stock price effects of distress; we believe that this measure is advantageous

over a balance sheet measure such as financial leverage because of its reliance on infor-

mation embedded in equity prices and volatility.

Our results demonstrate that a pronounced stock price decline of one firm yields neg-

ative valuation effects for industry rivals. This finding extends the results of previous

studies on intra-industry stock price effects of (1) bankruptcy announcements (e.g., Lang

and Stulz 1992; Ferris et al. 1997; Hertzel et al. 2008) and (2) unanticipated jumps in CDS

spreads in Jorion and Zhang (2007). Unlike previous studies, we are able to show that

default likelihood, derived from an option pricing framework, is an underlying reason for

the intra-industry contagion stock price effects. Leverage, Tobin’s Q, similarity between

rivals and negative surprise firms, degree of the surprise, relative size of the negative

surprise firm, industry concentration and industry performance are also found to explain

the contagion effects that accrue to industry rivals in response to negative surprises.

The remainder of the paper is organized as follows. In Sect. 2, we provide an overview

of previous literature on intra-industry stock price effects and discuss in more detail those

studies that specifically examine intra-industry effects from bankruptcies. In Sect. 3, we

describe the sample, data, and our methods for estimating valuation and default likelihood.

In Sect. 4, we describe: (1) the results for valuation effects that occur in response to

1 Jorion and Zhang (2007) find leverage to significantly explain the cross-sectional variation in the CDSspread effects that result from CDS jump events. They do not evaluate the cross-sectional variation in thestock price effects that result from CDS jump events or Chapter 11 and Chapter 7 bankruptcies.

Negative stock price surprises

123

negative stock price surprises, and (2) the analysis of factors that may influence the cross-

sectional variation in these valuation effects. We provide a summary in Sect. 5.

2 Literature on intra-industry stock price effects

Numerous studies have tested how negative information about a single firm could have an

impact on the stock prices of other firms in the industry. The general impetus for these

studies is that because of asymmetric information, the release of new public information

about one firm is used by market participants to make valuation inferences about corre-

sponding rival firms in the same industry. Foster (1981) and Clinch and Sinclair (1987) find

that negative earnings information about a firm can cause contagion effects within the

corresponding industry. Fenn and Cole (1994) find that negative news about First Exec-

utive and Travelers (leading to asset writedowns) has contagion effects that are most

pronounced for rivals with riskier assets. Aharony and Swary (1983; 1996) determine that

negative news about one bank causes contagion effects within the banking industry. Ak-

higbe et al. (1997) find that bond rating downgrades causes contagion effects within the

corresponding industry. Laux et al. (1998) and Kohers (1999) show that dividend reduc-

tions cause contagion effects throughout the corresponding industry.

More recently, Govindaraj et al. (2004) document favorable competitive effects in

response to a product recall of Firestone tires by the Bridgestone Corporation. Akhigbe

et al. (2006) find that analyst downgrades lead to negative industry effects. Xu et al. (2006)

show that an accounting irregularity at one firm can cause contagion effects within the

corresponding industry. Chen et al. (2007) find that delayed new product announcements

harm industry rivals. Hertzel et al. (2008) find that negative news about a firm can be

transmitted to other firms that are part of the supply chain.

In addition to the aforementioned studies, some studies have focused specifically on the

impact of a firm’s bankruptcy on the stock prices of rival firms. Lang and Stulz (1992)

explain that while the bankruptcy of a firm could signal problems for other rival firms, it

could allow rival firms an opportunity to increase market share that may be lost by the

bankrupt firm, and therefore cause competitive effects. They examine the intra-industry

effects of 59 bankruptcy filings between 1970 and 1989 by firms with more than $120

million in liabilities, and find industry rival portfolios react negatively to the bankruptcy

announcements on average. In their cross-sectional analyses, they find that the interaction

of leverage and industry concentration significantly influence the wealth effects of the

rivals. That is, firms that jointly have greater leverage and operate in more competitive

industries suffer to a greater extent when one of their competitors files for bankruptcy.

Ferris et al. (1997) study a sample of 279 bankruptcies between 1979 and 1989 without

restrictions on firm size, and also find the average intra-industry effects to be negative and

significant. They categorize rival firms as likely to default or not, based on whether the

rival firm actually files for bankruptcy within the next 3 years, and find greater contagion

effects for the rivals that end up in bankruptcy.

Using a more recent time period of 2001 through 2004, Jorion and Zhang (2007) find

their sample of 272 Chapter 11 bankruptcy events elicit intra-industry contagion effects,

and their sample of 22 Chapter 7 bankruptcy events elicit competitive effects. They also

assess 170 unanticipated distress events, as captured by jumps in credit default swap rates,

and find that these jumps generate contagion CDS spread and stock price effects. They

analyze the cross-sectional variation in the CDS spread effects and find larger contagion

effects for rival portfolios that are more strongly correlated with the event firm and have

A. Akhigbe et al.

123

greater leverage, and when the distressed firm is larger. Their cross-sectional results also

show greater competitive effects for distress events in more concentrated industries. Using

a sample of 250 bankruptcy filings, Hertzel et al. (2008) evaluate valuation consequences

for rivals, suppliers, and customers. Because financial distress typically begins prior to the

bankruptcy filing, they also examine the largest abnormal drop in the market value of the

filing firm in the year prior to filing and document intra-industry stock price effects. They

find that larger bankruptcies and bankruptcies in more concentrated industries are asso-

ciated with greater contagion effects.

3 Sample and methodology

The sample of firms with negative stock price surprises is formed over the 1998–2011 time

period. We identify all firms traded on the NYSE, AMEX or NASDAQ stock exchanges

with a minimum market capitalization of $1 billion and that experience a stock price

decline of at least 15 % in a single day. Table 1 shows the sample distribution over time

and across industries for the 2,368 firms that meet these criteria. The yearly distribution in

Panel A shows that 17 % of the stock price drops occurred in 2000, which coincides with

the collapse of the technology sector. In the subsequent years, 2001 and 2002, 16 % of the

negative surprises occurred. It is in these years that the financial markets uncovered a

variety of frauds that led to the passage of the Sarbanes–Oxley legislation, and the 9/11

terror attacks occurred. Reflecting the turmoil of the 2007–2008 financial crisis period, we

have approximately 25 % of our negative surprises over 2007 and 2008. The industry

distribution presented in Panel B tabulates the number of negative surprise events for each

two-digit SIC industry that represents 5 % or more of the 2,368 observations. The business

services industry, with 19 % of the observations, has the largest proportion of negative

surprise events. Also, technology-related industries that suffered from the eruption of the

technology bubble are heavily represented; 13 % of the observations are in electronics and

other electric equipment and 8 % are in industrial and commercial machinery and com-

puter equipment.

3.1 Measuring abnormal returns

We use event study methodology to estimate the daily abnormal stock returns (ARs) for the

2,368 negative surprise firms, 21,936 individual event-rival firms, and 2,368 equally-

weighted rival portfolios.2 Daily abnormal returns are calculated for the period surrounding

the date of the negative surprise, t0:

ARkt ¼ Rkt � ak þ bkRmtð Þ ð1Þ

where ARkt is the daily abnormal return for firm, rival, or rival portfolio k, Rkt is the daily

return for firm, rival, or rival portfolio k, Rmt is the daily return on the CRSP equally-

weighted index, and the parameters ak and bk are obtained from the market model that is

estimated with daily returns over the period t-120 to t-21 relative to the date of the negative

surprise.

Since rival firm returns are estimated within the same industry and over the same period

of time, the rival firm returns may not be independent. Therefore, we also measure the

2 There are 2,438 unique rival firms included in the sample of individual event-rival firms. For the sample ofrival portfolios, the mean (median) number of rivals per portfolio is 9.3 (4.0).

Negative stock price surprises

123

abnormal return using portfolios of rivals in Eq. (1), where the portfolio consists of all

rivals with the same four-digit SIC code as the negative surprise firm and are assigned an

equal weight within the portfolio. This method is advantageous because it controls for

contemporaneous correlation, but it (1) implicitly allows greater weight to rival firms in

more concentrated industries, because each rival is assigned a relatively high weight within

an industry that contains a small number of firms and (2) can reduce the variation across

the sample of rival portfolios due to the averaging process. Given the advantages and

disadvantages of both approaches, we analyze both individual rivals and rival portfolios,

following Song and Walking (2000). Z-statistics from Mikkelson and Partch (1988) are

Table 1 Distribution of negative stock price surprises

Year Frequency Percent Cumulativefrequency

Cumulativepercent

Panel A: sample distribution by year

1998 146 6.17 146 6.17

1999 167 7.05 313 13.22

2000 403 17.02 716 30.24

2001 194 8.19 910 38.43

2002 188 7.94 1,098 46.37

2003 86 3.63 1,184 50.00

2004 108 4.56 1,292 54.56

2005 98 4.14 1,390 58.70

2006 98 4.14 1,488 62.84

2007 138 5.83 1,626 68.67

2008 447 18.88 2,073 87.54

2009 89 3.76 2,162 91.30

2010 68 2.87 2,230 94.17

2011 138 5.83 2,368 100.00

Total 2,368 100.00 2,368 100.00

Industry Frequency Percent Cumulativefrequency

Cumulativepercent

Panel B: sample distribution by industry

Business services 456 19.26 456 19.26

Electronics and other electrical equipment(excl. computer equipment)

312 13.18 768 32.43

Chemicals and allied products 265 11.19 1,033 43.62

Industrial and commercial machineryand computer equipment

193 8.15 1,226 51.77

Communications 127 5.36 1,353 57.14

Measuring, analyzing and controllinginstruments; photographic, medicaland optical goods; watches and clocks

123 5.19 1,476 62.33

All others (54 other two-digit SICs) 892 37.67 2,368 100.00

Total 2,368 100.00 2,368 100.00

This table reports the sample distribution of 2,368 firms with a minimum of 15 % daily decline in stockprice over 1998 to 2011. Panel A (Panel B) provides the number of observations by year (by industry)

A. Akhigbe et al.

123

computed to test for statistical significance of the cumulative standardized average ARs for

the firms and rival portfolios.

3.2 Measuring default likelihood

The default likelihood measure used in this study is based on option pricing methodology

following previous studies (e.g., Vassalou and Xing 2004; Hillegeist et al. 2004; Duffie

et al. 2007; Huang and Lee 2013).3 This default likelihood measure (DL) uses information

embedded in equity data as a proxy for likelihood of default per firm per day. More

specifically, for each rival firm, we average the daily DL as described in Eq. (2) below over

the 60 days prior to the negative surprise event window:

DLt ¼ N �lnðVA;t=XtÞ þ ðlt � 1

2r2

A;tÞTrA;tT

!ð2Þ

where DLt = default likelihood measure on day t, VA,t = market value of assets on day t,

Xt = book value of total liabilities on day t, gathered for each year from Compustat,

lt = drift on day t (the daily rolling mean of the change in lnVA over the prior year),

rA,t = annualized standard deviation of asset returns on day t, T = time to maturity and is

set equal to 1, N = cumulative density function of the standard normal distribution. VA,t

and rA,t = unobservable variables estimated simultaneously using an iterative process.

Based on the notion that equity is like holding a call option on the value of the firm’s

assets, the Black–Scholes–Merton option pricing model can be used as the basis on which

to simultaneously estimate the two unobservable variables, VA,t and rA,t :

VE;t ¼ VA;tNðd1;tÞ � Xte�rtT Nðd2;tÞ ð3Þ

where, VE,t = market value of equity from CRSP at the end of day t, rt = risk-free rate on

day t (the daily nominal one-year U.S. T-bill rate), d1,t and d2,t = defined below in Eqs. (4)

and (5), respectively, and the remaining variables are previously defined.

d1;t ¼lnðVA;t=XtÞ þ rt þ 1

2r2

A;t

� �T

rA;tTð4Þ

d2;t ¼ d1;t � rA;t

ffiffiffiffiTp

ð5Þ

The iterative process we use to simultaneously estimate VA,t and rA,t applies a Newton

search algorithm to Eq. (3) and the optimal hedge equation, rE = (VA e-T N(d1) rA)/VE.

This process conforms to that in Hillegeist et al. (2004). For a starting value of rA,t we use

the volatility of equity on day t, rE,t. We calculate the annualized, rolling rE,t as the

standard deviation of daily equity returns over the previous year multiplied by the square

root of the number of trading days in the year. We then use the daily values of rE,t and VE,t,

to compute the initial daily value of rA,t, where rA,t = rE,t VE,t/(VE,t ? Xt). VA for each day

t is computed using Eq. (3) and the optimal hedge equation, where the Newton search

algorithm identifies values for VA and rA.

3 We acknowledge that the study by Bharath and Shumway (2008) shows the Merton distance to defaultmodel may not be a sufficient statistic for predicting default likelihood and argues that most of its marginalbenefit comes from its functional form. Nevertheless, the default likelihood measure has been shown tooutperform the traditional Altman-Z types of corporate default predictors that are based on accounting data(see Hillegeist et al. 2004).

Negative stock price surprises

123

Using the estimated daily values of VA and rA from this iterative process, l, T, and X,

we ultimately calculate the default likelihood measure, DL, per day for each rival firm in

our sample with Eq. (2). The default likelihood measure by construction is a function of

leverage, returns, and volatility. Regression analyses show the estimated default measures

are statistically significantly related to each of these three factors. Standardized regression

coefficients show that the most influential factor is leverage, followed by volatility, and

lastly returns.4 For brevity, these results are not tabulated.

4 Analyses and results

Results from estimating abnormal returns are reported in Table 2 for the 2,368 firms

experiencing the negative surprises in Panel A, along with the 21,936 individual event-

rival firms in Panel B and the 2,368 rival portfolios in Panel C. In Panel A, the 2,368 firms

experiencing the negative surprises have an average -19.00 % CAR for the [-1, 0]

window. The [? 1, ?2] window is slightly positive, 0.42 %, resulting in an average

-18.58 % CAR for the [-1, ?2] event window. The [? 3, ?20] post-event window does

not detect significant longer-term or persistent effects from the negative stock price surprises.

Panels B and C show that the individual rivals and rival portfolios experience negative

and significant abnormal returns of -1.15 and -3.63 %, respectively, over the two-day

window. Even though the rivals have significant and positive wealth effects over the [? 1,

?2] time period, the rival effects over the [-1, ?2] event window in both Panels B and C

are negative and significant, -0.47 and -3.39 %, respectively. These findings extend the

results of previous studies on intra-industry effects of bankruptcy announcements (e.g.,

Lang and Stulz 1992; Ferris et al. 1997; Jorion and Zhang 2007; Hertzel et al. 2008) by

providing evidence that large stock price declines transmit net contagion stock price effects

to the industry, even though negative surprises are not as drastic as bankruptcy.

As robustness checks, we evaluate whether the valuation effects are affected by using

alternative cutoffs instead of a 15 % negative stock return. Panels D and E report the

valuation effects when we instead use 10, 20 and 25 % as our cutoff. The individual rivals

and rival portfolios experience negative and significant abnormal returns over the [-1, 0]

and [-1, ?2] windows, consistent with the results when we use a 15 % cutoff. From this

point forward, we use the 15 % cutoff to define our sample, since a 15 % one-day stock

price drop is a substantial decline that clearly generates negative rival effects and it allows

us to examine a sufficiently large sample for our cross-sectional analyses.

We evaluate the cross-sectional variation in the rival valuation effects using ordinary

least squares regressions. The results of the simulations in Karafiath (2009) show that no

other estimators have clear advantages over ordinary least squares in the presence of

‘‘event clustering.’’ We estimate the following model using both the 21,936 individual

event-rivals and the 2,368 rival portfolios:

CARk ¼ h0 þ h1Defaultk þ h2Leveragek þ h3TobinQk þ h4Similarityk

þ h5PostPerformancek þ h6DegreeOfSurprisek þ h7RelativeSizek

þ h8Herfindahlk þ h9IndustryReturnsk þ x

ð6Þ

where the dependent variable is the rival CAR [-1, ?2], and each of the explanatory

variables is discussed and defined below. Table 3 reports summary statistics for the

4 Das et al. (2006) also report that leverage and volatility are the two largest factors explaining covariationin conditional default probabilities.

A. Akhigbe et al.

123

Table 2 Valuation effects for negative stock price surprises

Day/Window N AR/CAR % z value % positive

Panel A: ARs and CARs for firms that experience negative surprises

-5 2,368 0.02 -1.34 45

-4 2,368 0.07 0.19 48

-3 2,368 -0.02 -1.59 45

-2 2,368 0.31 1.37 48

-1 2,368 0.06 -2.10** 45

0 2,368 -19.06 -349.68*** 0

?1 2,368 0.41 5.51*** 51

?2 2,368 0.00 1.16 48

?3 2,368 -0.13 -0.50 48

?4 2,368 -0.10 0.71 49

?5 2,368 -0.10 -0.80 49

[- 1, 0] 2,368 -19.00 -247.23*** 3

[? 1, ?2] 2,368 0.42 4.59*** 51

[- 1, ?2] 2,368 -18.58 -169.70*** 7

[? 3, ?20] 2,368 -1.57 0.30 47

Panel B: ARs and CARs for individual event-rivals

-5 21,936 -0.00 -1.32 48

-4 21,936 0.10 5.90*** 49

-3 21,936 0.05 2.94*** 48

-2 21,936 -0.08 -1.02 48

-1 21,936 -0.19 -7.58*** 46

0 21,936 -0.96 -53.66*** 38

?1 21,936 0.46 26.27*** 53

?2 21,936 0.22 10.02*** 50

?3 21,936 0.09 7.60*** 49

?4 21,936 0.04 4.00*** 49

?5 21,936 -0.09 -1.12 47

[- 1, 0] 21,936 -1.15 -43.68*** 39

[? 1, ?2] 21,936 0.68 24.64*** 53

[- 1, ?2] 21,936 -0.47 -13.46*** 46

[? 3, ?20] 21,936 0.59 9.45*** 52

Panel C: ARs and CARs for rival portfolios

-5 2,368 -0.25 -6.49*** 45

-4 2,368 -0.36 -9.69*** 43

-3 2,368 -0.44 -10.14*** 43

-2 2,368 -0.14 -4.13*** 46

-1 2,368 -0.06 -11.80*** 43

0 2,368 -3.03 -62.22*** 22

?1 2,368 0.21 4.28*** 49

?2 2,368 0.03 -1.96** 47

?3 2,368 -0.19 -4.30*** 47

?4 2,368 -0.07 -2.00** 47

Negative stock price surprises

123

Table 2 continued

Day/Window N AR/CAR % z value % positive

?5 2,368 -0.12 -2.57** 47

[- 1, 0] 2,368 -3.63 -54.01*** 24

[? 1, ?2] 2,368 0.24 1.63 51

[- 1, ?2] 2,368 -3.39 -37.22*** 32

[? 3, ?20] 2,368 -1.60 -7.95*** 47

Panel D: robustness tests on price drop cutoff for individual event-rivals

10 %

[- 1, 0] 54,985 -1.20 -70.49*** 40

[? 1, ?2] 54,985 0.27 20.31*** 51

[- 1, ?2] 54,985 -0.94 -35.18*** 44

[? 3, ?20] 54,985 -0.96 -3.75*** 49

20 %

[- 1, 0] 8,347 -1.19 -29.92*** 38

[? 1, ?2] 8,347 0.25 4.30*** 50

[- 1, ?2] 8,347 -0.95 -18.07*** 43

[? 3, ?20] 8,347 0.73 6.67*** 52

25 %

[- 1, 0] 4,061 -1.23 -19.85*** 39

[? 1, ?2] 4,061 0.12 1.45 49

[- 1, ?2] 4,061 -1.11 -13.07*** 42

[? 3, ?20] 4,061 -0.12 -0.90 51

Panel E: robustness tests on price drop cutoff for rival portfolios

10 %

[- 1, 0] 6,366 -4.38 -123.60*** 21

[? 1, ?2] 6,366 0.34 10.31*** 52

[- 1, ?2] 6,366 -4.03 -79.94*** 27

[? 3, ?20] 6,366 -1.92 -17.69*** 47

20 %

[- 1, 0] 897 -2.71 -26.56*** 29

[? 1, ?2] 897 -0.04 -1.65* 48

[- 1, ?2] 897 -2.75 -20.05*** 33

[? 3, ?20] 897 -0.61 -1.82* 49

25 %

[- 1, 0] 416 -2.50 -15.84*** 31

[? 1, ?2] 416 -0.41 -3.45*** 47

[- 1, ?2] 416 -2.90 -13.51*** 34

[? 3, ?20] 416 -0.98 -2.28** 48

This table shows daily abnormal returns (ARs) for 10 days around the negative stock price surprise as wellas mean cumulative abnormal returns (CARs) over various windows for the 2,368 negative surprise firmsalong with 21,936 individual event-rivals and 2,368 rival portfolios. AR is the difference between actual andexpected returns, where expected returns are based on the market model that is estimated over the periodt-120 to t-21 relative to the date of the negative stock price surprise

*, **, and *** denote significance at the 10, 5 and 1 % levels, respectively

A. Akhigbe et al.

123

variables used in Eq. (6). The variable definitions below are described for the individual

rivals; the variable definitions for the rival portfolios are the equally-weighted averages of

the individual rivals that comprise the portfolio.

4.1 Explanatory variables used in cross-sectional model

We hypothesize that the degree of the revision in the rivals’ stock prices is greater for those

with higher default likelihood. We expect that the negative surprise events would result in

additional scrutiny to the default likelihood of all members of the industry. With this added

focus on the likelihood of default, rival firms with greater default likelihood may be

penalized to a greater extent. The variable Default is calculated as the daily average of the

DL from Eq. (2) over the 60 days prior to the event [-61, -1], so it is not influenced by the

negative surprise event.

It can be seen in Table 3 that the default likelihood measure is zero for more than half of

the sample of 21,936 event-rival firms. As a result, we also create and utilize a high default

indicator variable to capture the rival firms with distance to default in the top quartile of the

sample to handle the nonlinearity in the data. High Default is set equal to one for rivals

with Default in the top quartile of the sample, and zero otherwise. For the rival portfolios,

High Default captures the proportion of the rivals in the industry portfolio with default

likelihood in the top quartile of the sample. As reported in Table 3, the mean (median)

value for High Default for the rival portfolio is 23 % (7 %).

Industry effects are expected to be worse for rival firms that have higher financial

leverage. Rival firms with lower leverage have more financial flexibility to deal with cash

flow pressure, and may even be able to capitalize during weak industry conditions by

seizing market share (e.g., Opler and Titman 1994). Leverage is measured as the ratio of

long-term debt to total assets measured at the end of the calendar year prior to the cor-

responding negative surprise, using Compustat data. The summary statistics in Table 3

report the average rival portfolio leverage is 17 % leverage and the median rival portfolio

leverage is 14 %.

Rival firms may be more susceptible to adverse stock price reactions in response to

negative signals when they have been performing strongly because the signal may reflect a

more pronounced shock. We use Tobin’s Q as a proxy for firm performance. Tobin Q is a

dummy variable that equals one when the market value of equity is greater than the book

value of equity, and zero otherwise.5 The market value of equity is measured at the time of

the negative surprise using CRSP data and book value of equity is the year end value

gathered from Compustat for the calendar year-end prior to the negative surprise. From the

summary statistics in Table 3, we see that the majority of the rivals and rival portfolios

have market value of equity greater than book value of equity.

Rival firms should experience greater contagion effects if they are more similar to the

negative surprise firm. Similarity is measured as the correlation in daily stock returns

between the negative surprise firm and the rival firm over the period t-120 to t-21 relative to

the date of the negative surprise. The summary statistics in Table 3 report the correlation in

stock returns for the mean and median rival portfolio is 50 %.

5 We also use the ratio of market value of equity to book value of equity in the regressions. Since thiscontinuous version of Tobin’s Q is not significant, we report the results when we include the dummyvariable. Lang et al. (1989) also convert Tobin’s Q into a categorical variable for part of their analysis. Forour analysis, the continuous variable may not be significant because the relationship between CARs and Qmay be nonlinear.

Negative stock price surprises

123

We expect that a larger negative surprise to the firm would elicit a more pronounced

negative market response for the firm’s rivals. Degree of Surprise is the percent drop in

the share price of the negative surprise firm. The average share price drop is approximately

21 %, and the median drop is approximately 18 %.

It can be argued that the contagion effects are more profound when the negative surprise

firm is relatively large. To the extent that the larger firms are market share leaders, rivals

are more likely to also suffer when these industry leaders experience a negative shock. The

Relative Size variable is measured using CRSP data as the ratio of the equity market value

of the negative surprise firm to the median industry equity market value one month prior to

the negative surprise (i.e., t-21 relative to the date of the negative surprise). As reported in

Table 3, the mean (median) value for Relative Size for the rival portfolio is 3.3749

(2.2201 %).

When a firm experiences a pronounced negative surprise, the negative effects on the

industry may depend on industry conditions. Rivals in less competitive industries may not

be adequately prepared to handle the added challenge and suffer to a greater extent. Thus,

it can be argued that greater contagion effects would result in less competitive industries.

An alternative argument is presented by Lang and Stulz (1992). Due to customer shifts

in demand as a result of bankruptcies, they argue that the strength of intra-industry

competitive effects would be greater in less competitive industries. However, we believe

that this competitive repositioning benefit is less likely to occur with our less severe

distress events than with bankruptcy events.

Table 3 Summary statistics for rival firm characteristics

Variables Individual rivals Rival portfolios

Mean Q1 Q2 Q3 Mean Q1 Q2 Q3

Rival CAR -0.0047 -0.0448 -0.0043 0.0029 -0.0338 -0.0658 -0.0215 0.0086

Default 0.0092 0.0000 0.0000 0.0006 0.0094 0.0000 0.0002 0.0050

High default 0.2500 0 0 0 0.2295 0.0000 0.0690 0.4000

Leverage 0.1550 0.0099 0.1042 0.2376 0.1674 0.0850 0.1366 0.2217

Tobin Q 0.6052 0 1 1 0.9134 1 1 1

Similarity 0.5419 0.3900 0.5526 0.6900 0.4971 0.3441 0.5083 0.6593

Degree ofsurprise

-0.2090 -0.1721 -0.1862 -0.2223 -0.2084 -0.2254 -0.1823 -0.1628

Relative size 1.9505 0.3299 0.5547 1.1956 3.3749 1.1069 2.2201 4.7090

Herfindahl 0.2147 0.1044 0.1713 0.2915 0.3862 0.1870 0.3045 0.5044

Industry returns 0.1349 -0.0400 0.0795 0.2309 0.1096 -0.0659 0.0664 0.2212

This table presents summary statistics for 21,936 individual event-rivals and 2,368 rival portfolios. RivalCAR is the event CAR [- 1, ?2]; Default is the daily average of the DL from equation (2) over the 60 daysprior to the event [- 61, -1]; High Default equals one for rivals with Default in the top quartile of thesample, and zero otherwise; Leverage is the long-term debt/total assets for the rival; Tobin Q is market valueof equity [ book value of equity; Similarity is the correlation in stock returns between the negative surprisefirm and the rival firm; Degree of Surprise is the percent share price drop for the negative surprise firm;Relative Size is the negative surprise firm market value/market value of median firm in the industry;Herfindahl is the sum of the squared market shares of all firms in the industry; Industry Returns is the %change in the value of all firms in the industry. For the rival portfolios, the variables are calculated asequally-weighted averages of the values for the individual rivals that comprise the portfolio, except HighDefault is the proportion

A. Akhigbe et al.

123

We use the Herfindahl index to reflect industry concentration, where an industry is

considered to be less competitive, or more concentrated, with a greater Herfindahl index.

Specifically, Herfindahl is defined as the sum of the squared market shares of all firms in

the industry measured at the calendar year-end prior to the negative surprise event, using

Compustat data. The summary statistics in Table 3 show the mean (median) Herfindahl for

the rival portfolios is 0.3862 (0.3045).

Rival firms may be more susceptible to adverse stock price reactions in response to a

negative surprise when the industry has been performing strongly. If the industry has been

performing strongly, the market may be less likely to expect and more shocked to see a

negative surprise within the industry. We use Industry Returns as a proxy for industry

performance. It is calculated using CRSP data as the percent change in the value of all the

firms in the industry over the period t-120 to t-21 relative to the date of the negative

surprise. Table 3 shows the mean (median) industry performance for the sample of rival

portfolios to be positive 10.96 % (6.64 %).

4.2 Results of cross-sectional analyses

In Tables 4 and 5, we present the regression results from estimating equation (6). We

organize the explanatory variables into three groups: (1) related to rival firms, (2) related to

negative surprise firms, and (3) related to industry. Table 4 presents the results using the

21,936 event-rivals, and Table 5 presents the results using the 2,368 rival portfolios. We

evaluate the regression model separately for negative surprise events that occur during the

2007–2008 financial crisis period; Models 1 and 2 analyze negative shocks that occur in

the non-crisis period and only differ by the proxy that is used to measure the default

likelihood whereas Model 3 analyzes the shocks that occur in the crisis period. Following

previous studies (e.g., Brunnermeier 2009; Fahlenbrach et al. 2012; Gorton and Metrick

2012), we define the 2007–2008 financial crisis period as July 2007 through December

2008.6

Also, these same three models are estimated using only the first observation of a

negative surprise to occur in an industry within 20 trading days. This subset of first

observations provides a robustness check on the possibility that event clustering is driving

our results. Indeed, the results for this subset are consistent, with few exceptions. Thus, the

following discussion related to Tables 4 and 5 primarily focuses on the full sample of

negative surprises. The variance inflation factors across the models range from 1.0 to 1.3,

which suggests that multicollinearity is not unduly influencing the statistical testing of the

coefficients.

In Table 4, when we analyze Models 1 and 2, the Default variable is not significant, but

High Default is negative and significant. This finding suggests that a negative shock has a

more pronounced and adverse valuation effect on industry rivals with the greatest default

likelihood. The significance and direction of influence of the remaining variables are

consistent across Models 1 and 2. The coefficient of Leverage is negative and significant. It

appears that rival firms with higher leverage are penalized to a greater extent, perhaps

because they do not have the financial flexibility to deal with cash flow pressure (Opler and

Titman 1994). The coefficient of Tobin Q is negative, indicating that rivals with strong

performance are more adversely affected by the negative surprise. The coefficient on

Similarity is found to be negative and significant, consistent with our argument that rivals

that are more similar to the negative surprise firm suffer greater contagion in response to

6 The results are essentially the same when January 2007 is used as the beginning of the crisis period.

Negative stock price surprises

123

Ta

ble

4C

ross

-sec

tional

resu

lts

for

rival

effe

cts

from

neg

ativ

esu

rpri

ses:

indiv

idual

rival

s

Fir

stO

bse

rvat

ion

On

ly

Mo

del

1M

od

el2

Mo

del

3M

odel

1M

odel

2M

odel

3V

aria

ble

sN

on

-Cri

sis

No

n-C

risi

sC

risi

sN

on

-Cri

sis

No

n-C

risi

sC

risi

s

Inte

rcep

t0

.02

34

(8.0

0**

*)

0.0

24

6(8

.41

**

*)

0.0

00

9(0

.09

)0

.018

7(5

.65

**

*)

0.0

19

7(5

.93

**

*)

-0

.024

6(-

2.4

2*

*)

Rel

ate

dto

riva

lfi

rms

Def

ault

-0

.005

2(-

0.4

5)

--

0.0

01

4(0

.12

)–

Hig

hd

efau

lt-

-0

.00

65

(-4

.77*

**

)0

.04

78

(9.7

7*

**

)–

-0

.004

6(-

2.9

2*

**

)0

.010

2(1

.73

*)

Lev

erag

e-

0.0

14

5(-

3.9

4*

**

)-

0.0

12

6(-

3.4

2*

**

)-

0.0

06

1(-

0.8

1)

-0

.00

90

(-2

.16*

*)

-0

.008

0(-

1.9

2*

)0

.018

6(2

.00

**

)

To

bin

Q-

0.0

06

0(-

4.7

5*

**

)-

0.0

06

4(-

5.0

4*

**

)-

0.0

09

0(-

2.7

7*

**

)-

0.0

03

8(-

2.5

5*

**

)-

0.0

04

3(-

2.9

0*

**

)-

0.0

07

6(-

2.0

2*

*)

Sim

ilar

ity

-0

.018

4(-

5.6

6*

**

)-

0.0

16

3(-

4.9

7*

**

)-

0.0

08

9(-

0.9

5)

-0

.02

59

(-7

.36*

**

)-

0.0

24

2(-

6.8

2*

**

)-

0.0

00

2(-

0.0

2)

Rel

ate

dto

neg

ati

vesu

rpri

sefi

rms

Deg

ree

of

surp

rise

0.0

54

2(6

.72

**

*)

0.0

56

4(7

.06

**

*)

-0

.069

2(-

2.6

5*

**

)0

.027

9(2

.97

**

*)

0.0

29

6(3

.16

**

*)

-0

.090

3(-

3.6

6*

**

)

Rel

ativ

esi

ze-

0.0

00

3(-

4.6

1*

**

)-

0.0

00

3(-

4.5

8*

**

)0

.00

16

(4.4

8*

**

)-

0.0

00

7(-

3.0

9*

**

)-

0.0

00

7(-

3.0

3*

**

)0

.003

1(1

0.2

8*

**

)

Rel

ate

dto

ind

ust

ry

Her

fin

dah

l-

0.0

15

7(-

4.3

1*

**

)-

0.0

16

0(-

4.3

9*

**

)-

0.0

56

7(-

5.6

7*

**

)-

0.0

11

7(-

3.2

2*

**

)-

0.0

11

8(-

3.2

5*

**

)-

0.0

25

0(-

2.7

4*

**

)

Ind

ust

ryre

turn

s0

.00

79

(4.1

3**

*)

0.0

07

4(3

.85

**

*)

0.0

07

6(0

.75

)-

0.0

13

7(-

3.8

4*

**

)-

0.0

13

7(-

3.8

3*

**

)-

0.0

32

1(-

3.0

6*

**

)

N1

7,5

76

17

,57

64

,36

08

,553

8,5

53

1,8

28

Ad

j.R

20

.01

09

0.0

11

70

.03

86

0.0

10

70

.011

70

.091

4

F-v

alu

e2

4.1

8*

**

27

.03

**

*2

2.8

7*

**

12

.56

**

*1

3.4

6*

**

23

.97

**

*

Th

ista

ble

repo

rts

reg

ress

ion

resu

lts

toan

alyze

the

cro

ss-s

ecti

on

alv

aria

tio

nin

the

CA

Rs

[-1

,?

2]

of

indiv

idual

rival

sof

firm

sth

atex

per

ience

neg

ativ

esu

rpri

ses.

The

subse

t,F

irst

Ob

serv

ati

on

On

ly,in

clu

des

just

the

firs

tn

egat

ive

surp

rise

too

ccu

rin

anin

du

stry

wit

hin

20

trad

ing

day

s.D

efa

ult

isth

ed

aily

aver

age

of

the

DL

from

equ

atio

n(2

)o

ver

the

60

day

sp

rio

rto

the

even

t[-

61

,-

1];

Hig

hD

efa

ult

equ

als

on

efo

rri

val

sw

ith

Def

ault

inth

eto

pq

uar

tile

of

the

sam

ple

,an

dze

roo

ther

wis

e;L

ever

age

isth

elo

ng

-ter

md

ebt/

tota

las

sets

for

the

riv

al;

To

bin

Qis

mar

ket

val

ue

of

equ

ity

/boo

kv

alue

of

equ

ity

;S

imil

ari

tyis

corr

elat

ion

inst

ock

retu

rns

bet

wee

nth

en

egat

ive

surp

rise

firm

and

the

riv

alfi

rmp

rio

rto

the

neg

ativ

esu

rpri

se;

Deg

ree

of

Su

rpri

seis

the

per

cen

tsh

are

pri

ced

rop

for

the

neg

ativ

esu

rpri

sefi

rm;

Rel

ati

veS

ize

isth

en

egat

ive

surp

rise

firm

mar

ket

val

ue/

mar

ket

val

ue

of

med

ian

firm

inth

ein

du

stry

;H

erfi

nda

hl

isth

esu

mo

fth

esq

uar

edm

ark

etsh

ares

of

all

firm

sin

the

indu

stry

;In

du

stry

Ret

urn

sis

the

%ch

ang

ein

the

val

ue

of

all

firm

sin

the

ind

ust

ry

*,

**,

and

***

den

ote

signifi

cance

atth

e10,

5an

d1

%le

vel

s,re

spec

tivel

y

A. Akhigbe et al.

123

Ta

ble

5C

ross

-sec

tional

resu

lts

for

rival

effe

cts

from

neg

ativ

esu

rpri

ses:

rival

port

foli

os

Fir

sto

bse

rvat

ion

on

ly

Mo

del

1M

odel

2M

odel

3M

odel

1M

odel

2M

odel

3V

aria

ble

sN

on-c

risi

sN

on-c

risi

sC

risi

sN

on-c

risi

sN

on-c

risi

sC

risi

s

Inte

rcep

t0

.019

5(1

.87

*)

0.0

29

1(2

.74

**

*)

-0

.022

8(-

1.0

8)

0.0

27

6(2

.50

**

)0

.03

39

(3.0

1**

)-

0.0

23

9(-

1.0

2)

Rel

ate

dto

riva

lfi

rms

Def

ault

0.0

17

8(0

.38

)–

–0

.00

47

(0.1

0)

––

Hig

hd

efau

lt–

-0

.000

2(-

3.0

2*

**

)0

.00

01

(0.8

4)

–-

0.0

00

1(-

2.2

1*

*)

-0

.000

0(-

0.0

7)

Lev

erag

e-

0.0

22

3(-

1.6

2)

-0

.019

6(-

1.4

3)

-0

.117

3(-

4.3

5*

**

)-

0.0

31

5(-

2.2

4*

*)

-0

.029

2(-

2.0

8*

*)

-0

.108

2(-

3.7

2*

**

)

To

bin

Q0

.002

5(0

.35

)-

0.0

04

9(-

0.6

6)

0.0

22

0(2

.23

**

)0

.00

08

(0.1

1)

-0

.004

4(-

0.5

9)

0.0

15

4(1

.34

)

Sim

ilar

ity

-0

.060

3(-

7.3

8*

**

)-

0.0

55

6(-

6.7

5*

**

)-

0.0

56

0(-

2.9

3*

**

)-

0.0

63

5(-

7.1

7*

**

)-

0.0

59

8(-

6.7

0*

**

)-

0.0

33

7(-

1.6

2)

Rel

ate

dto

neg

ati

vesu

rpri

sefi

rms

Deg

ree

of

surp

rise

0.0

05

8(0

.28

)0

.00

84

(0.4

1)

-0

.137

7(-

2.4

8*

**

)0

.02

94

(1.2

8)

0.0

30

0(1

.31

)-

0.1

61

8(-

2.8

3*

**

)

Rel

ativ

esi

ze-

0.0

01

2(-

2.8

1*

**

)-

0.0

01

2(-

2.7

9*

**

)0

.00

04

(0.2

8)

-0

.000

3(-

0.5

9)

-0

.000

3(-

0.5

6)

0.0

01

2(0

.83

)

Rel

ate

dto

ind

ust

ry

Her

fin

dah

l-

0.0

18

8(-

3.1

1*

**

)-

0.0

19

2(-

3.2

0*

**

)-

0.0

31

3(-

2.3

3*

*)

-0

.021

8(-

3.4

4*

**

)-

0.0

21

9(-

3.4

7*

**

)-

0.0

39

3(-

2.8

0*

**

)

Ind

ust

ryre

turn

s-

0.0

58

8(-

11

.63

**

*)

-0

.059

4(-

11

.77

**

*)

-0

.025

0(-

1.3

7)

-0

.038

5(-

5.3

6*

**

)-

0.0

38

6(-

5.3

8*

**

)-

0.0

31

0(-

1.6

2)

N1

,806

1,8

06

56

21

,22

51

,22

53

45

Ad

j.R

20

.088

70

.09

32

0.0

71

80

.05

16

0.0

55

40

.099

0

F-v

alu

e2

2.9

5*

**

24

.19

**

*6

.42

**

*9

.32

**

*9

.97

**

*5

.72*

**

This

table

report

sre

gre

ssio

nre

sult

sto

anal

yze

the

cross

-sec

tional

var

iati

on

inth

eC

AR

s[-

1,?

2]

of

riv

alp

ort

foli

os

of

firm

sth

atex

per

ience

neg

ativ

esu

rpri

ses.

Th

esu

bse

to

fF

irst

Ob

serv

ati

on

sin

clu

des

just

the

firs

tn

egat

ive

surp

rise

too

ccu

rin

anin

du

stry

wit

hin

20

trad

ing

day

s.D

efa

ult

isth

ed

aily

aver

age

of

the

DL

fro

meq

uat

ion

(2)

ov

erth

e6

0d

ays

pri

or

toth

eev

ent

[-6

1,

-1

];H

igh

Def

ault

equ

als

on

efo

rri

val

sw

ith

Def

ault

inth

eto

pq

uar

tile

of

the

sam

ple

,an

dze

roo

ther

wis

e;L

ever

age

isth

elo

ng

-ter

md

ebt/

tota

las

sets

for

the

riv

al;

To

bin

Qis

riv

alm

ark

etv

alu

eo

feq

uit

y/b

oo

kv

alu

eo

feq

uit

y;

Sim

ila

rity

isth

eco

rrel

atio

nin

stock

retu

rns

bet

wee

nth

en

egat

ive

surp

rise

firm

and

the

riv

alfi

rmp

rio

rto

the

neg

ativ

esu

rpri

se;

Deg

ree

of

Su

rpri

seis

the

per

cen

tsh

are

pri

ced

rop

for

the

neg

ativ

esu

rpri

sefi

rm;

Rel

ati

veS

ize

isth

en

egat

ive

surp

rise

firm

mar

ket

val

ue/

mar

ket

val

ue

of

med

ian

firm

inth

ein

du

stry

;H

erfi

nda

hl

isth

esu

mo

fth

esq

uar

edm

ark

etsh

ares

of

all

firm

sin

the

ind

ust

ry;

Ind

ust

ryR

etu

rns

isth

e%

chan

ge

inth

ev

alue

of

all

firm

sin

the

indu

stry

.T

he

var

iab

les

are

calc

ula

ted

for

each

riv

alp

ort

foli

oto

be

equ

ally

-wei

gh

ted

aver

ages

of

the

ind

ivid

ual

riv

als

that

com

pri

seth

ep

ort

foli

o

*,

**

and

***

den

ote

signifi

cance

atth

e10,

5an

d1

%le

vel

s,re

spec

tivel

y

Negative stock price surprises

123

the negative surprise. The coefficient of Degree of Surprise is positive, implying that rivals

experience more pronounced adverse valuation effects in response to larger negative

surprises.

The coefficient of Relative Size is negative and significant. This result indicates that

rivals are more adversely affected when the firm experiencing the negative surprise is

relatively large. Industry concentration is also found to be an important factor. The

coefficient of Herfindahl is negative and significant, consistent with the argument that

rivals in less competitive industries suffer to a greater extent.

Lastly, the coefficient of Industry Returns is positive and significant. However, when we

examine Models 1 and 2 for the subset of first observations, we find the coefficient of

Industry Returns is negative and significant. These results imply that the rivals are more

adversely affected in response to the first negative surprise within the industry when the

industry return was more favorable. These results imply that when the industry is per-

forming well, the rivals experience more pronounced adverse effects in response to the first

negative surprise observation, but less pronounced adverse effects in response to sub-

sequent negative surprise observations in the same industry.

In Table 4, when we analyze Model 3 that is focused on negative surprises that occur

during the 2007–2008 financial crisis period, the results are quite similar to Model 2. One

notable exception is that the coefficients on High Default and Leverage are no longer

negative and significant, but are now positive and significant.7 This suggests that a negative

shock has a less pronounced adverse effect on industry rivals with the greatest default

likelihood. During the financial crisis, stock prices of most firms were systematically

depressed, even without the force of negative industry-specific signals. Industry rivals with

the greatest default likelihood might have already been priced during the crisis to reflect a

pessimistic outlook, so that the effects of an additional negative industry-specific shock

were diluted. Conversely, other industry rivals with lower default likelihood might not

have been priced as low in response to the general outlook of the crisis, thus these rivals

may be more exposed to negative industry-specific shocks.8 An alternative argument is that

the market may have expected beneficial restructuring or government intervention aimed at

preventing the collapse of firms with the greatest default likelihood, which could reduce

the adverse effects on these firms during the crisis. These arguments offer an explanation

for a positive relationship between the default likelihood and valuation effect among rival

firms.

In addition, the coefficients on Degree of Surprise and Relative Size also change signs

when we examine the sample of negative surprises during the crisis period. These coef-

ficients show that industry rivals benefit to a greater extent with larger negative surprises

and when the firms experiencing the negative shocks are relatively large. During the crisis,

valuations were already depressed. Therefore, the negative news may have triggered a

flight to quality within the industry, rather than a complete sell-off of all rival stocks in

response to a negative surprise.

In Table 5, when Models 1 and 2 are estimated for the sample of rival portfolios, the

results for the default variables are consistent with those shown in Table 4 for the indi-

vidual rivals. As with Table 4, the Default variable is not significant, while the coefficient

7 The coefficient on Leverage for Model 3 is positive and significant when examining the subset of firstobservations of negative surprises within an industry within 20 trading days.8 A comparison of CARs between rivals with high default and those without high default during the crisisperiod shows that the CARs are significantly lower for rivals without high default likelihood. This com-parison lends support to this interpretation.

A. Akhigbe et al.

123

of High Default is negative and significant. However, there are some differences in results

when applying Models 1 and 2 to the rival portfolios (Table 5) as compared to the indi-

vidual rivals (Table 4). The main differences are that the coefficient for Tobin Q is no

longer significant, while the Industry Return coefficient is negative and significant. These

results imply that the recent industry stock price performance may substitute for the firm-

specific past performance (as measured by Tobin Q) when models are applied to the rival

portfolios in the corresponding industry. In addition, the coefficients on Leverage and

Degree of Surprise are no longer significant during the non-crisis period.

When we examine the negative surprises that occur within the 2007–2008 financial

crisis period (Model 3), High Default is not significant. As previously discussed, it is

plausible that during these extreme economic conditions the resulting restructurings and

bailouts interferes with market sensitivity to default conditions. However, the coefficient of

Leverage is negative and significant when applying Model 3, which suggests weaker

effects for more highly levered rival portfolios in response to negative surprises during the

2007–2008 financial crisis period. In addition, the rival portfolios with strong performance

experience more favorable effects and those that are more similar to the firm experiencing

the negative surprise experience more unfavorable effect.

These same three models are estimated again using only the first negative surprise for a

firm within 20 trading days, as a robustness check on the possibility that event clustering is

driving our results. The results are consistent, with one exception that the coefficient of

Relative Size becomes insignificant.

5 Summary

Studies have determined that specific news about financial distress such as bankruptcy

announcements of one firm can emit a negative signal about its corresponding industry.

We extend the literature by examining a large sample of negative surprises to determine

whether a large decline in a stock’s price transmits net contagion effects to the industry.

While large stock price declines are not as severe as bankruptcy announcements, they are

unanticipated negative surprises that may transmit industry information. Our analyses

show that a pronounced stock price decline of one firm yields significant negative valuation

effects for industry rivals, on average.

We evaluate various factors to explain the cross-sectional variation in the rival valuation

effects. A negative surprise about one firm likely compels market participants to scrutinize

the default likelihood of rivals and/or industry conditions and penalize them accordingly.

We focus on a measure of default likelihood of rivals that has not yet been used to explain

the intra-industry stock price effects of distress. A key contribution of our cross-sectional

analyses is that we find negative shocks have more pronounced adverse effects on rivals

with the greatest default likelihood. The impact of the negative surprise on rivals is also

conditioned on the degree of the surprise, characteristics of the firm experiencing the

negative surprise (such as its relative size), characteristics of the rival firms (such as their

similarity to the firm experiencing the negative surprise), and characteristics of the cor-

responding industry (such as degree of concentration).

We also find that the sensitivity of industry rivals and portfolios to negative surprises

changes during the 2007–2008 financial crisis. The rivals with the greatest default likeli-

hood do not suffer more pronounced negative effects during this period, which may be

because their stocks had already been priced to reflect pessimistic outlooks, so that the

effects of an additional shock were diluted. Alternatively, the market may have expected

Negative stock price surprises

123

restructuring or government intervention that could prevent the collapse of firms with the

greatest default likelihood.

References

Aharony J, Swary I (1983) Contagion effects of bank failures: evidence from capital markets. J Bus56:305–322

Aharony J, Swary I (1996) Additional evidence on the information-based contagion effects of bank failures.J Bank Financ 20:57–69

Akhigbe A, Madura J, Whyte AM (1997) Intra-industry effects of bond rating adjustments. J Financ Res20:545–561

Akhigbe A, Madura J, Newman M (2006) Industry effect of analyst stock revision. J Financ Res 29:181–198Bharath ST, Shumway T (2008) Forecasting default with the Merton distance to default model. Rev Financ

Stud 21:1339–1368Brunnermeier MK (2009) Deciphering the liquidity and credit crunch. J Econ Perspect 23:77–100Chava S, Jarrow RA (2004) Bankruptcy prediction with industry effects. Rev Financ 8:537–569Chen SS, Chung TY, Ho KW, Lee CF (2007) Intra-industry effects of delayed new product introductions.

Rev Pac Basin Financ Mark Policies 10:415–443Clinch GJ, Sinclair NA (1987) Intra-industry information releases: a recursive systems approach. J Account

Econ 9:89–106Das S, Freed L, Geng G, Kapadia N (2006) Correlated default risk. J Fixed Inc 16:7–32Das SR, Duffie D, Kapadia N, Saita L (2007) Common failings: how corporate defaults are correlated.

J Financ 62:93–117Duffie D, Saita L, Wang K (2007) Multi-period corporate default prediction with stochastic covariates.

J Financ Econ 83:635–665Fahlenbrach R, Prilmeier R, Stulz RM (2012) This time is the same: using bank performance in 1998 to

explain bank performance during the recent financial crisis. J Financ 67:2139–2185Fenn GW, Cole RA (1994) Announcements of asset-quality problems and contagion effects in the life

insurance industry. J Financ Econ 35:181–198Ferris SP, Jayaraman N, Makhija AK (1997) The response of competitors to announcements of bankruptcy:

an empirical examination of contagion and competitive effects. J Corp Financ 3:367–395Foster G (1981) Intra-industry information transfer associated with earnings releases. J Account Econ

3:201–232Gorton G, Metrick A (2012) Securitized banking and the run on repo. J Financ Econ 104:425–451Govindaraj S, Jaggi B, Lin B (2004) Market overreaction to product recall revisited-The case of Firestone

tires and the Ford Explorer. Rev Quant Financ Account 23:31–54Hertzel MG, Li Z, Officer MS, Rodgers KF (2008) Inter-firm linkages and the wealth effects of financial

distress along the supply chain. J Financ Econ 87:374–387Hillegeist SA, Keating EK, Cram DP, Lundstedt KG (2004) Assessing the probability of bankruptcy. Rev

Account Stud 9:5–34Huang HH, Lee HH (2013) Product market competition and credit risk. J Bank Financ 37:324–340Jorion P, Zhang G (2007) Good and bad credit contagion: evidence from credit default swaps. J Financ Econ

84:860–883Karafiath I (2009) Is there a viable alternative to ordinary least squares regression when security abnormal

returns are the dependent variable? Rev Quant Financ Accout 32:17–31Kohers N (1999) The industry-wide implications of dividend omission and initiation announcements and the

determinants of information transfer. Financ Rev 34:137–158Lando D, Nielsen MS (2010) Correlation in corporate defaults: contagion or conditional independence?

J Financ Intermediation 19:355–372Lang LHP, Stulz RM (1992) Contagion and competitive intra-industry effects of bankruptcy announce-

ments: an empirical analysis. J Financ Econ 32:45–60Lang LHP, Stulz RM, Walkling RA (1989) Managerial performance, Tobin’s Q, and the gains from

successful tender offers. J Financ Econ 24:37–54Laux PA, Starks LT, Yoon PS (1998) The relative importance of competition and contagion in intra-industry

information transfers: an investigation of dividend announcements. Financ Manag 27:5–16Lucas D (1995) Default correlation and credit analysis. J Fixed Inc 9:76–87Mikkelson WH, Partch MM (1988) Withdrawn security offerings. J Financ Quant Anal 23:119–133

A. Akhigbe et al.

123

Opler TC, Titman S (1994) Financial distress and corporate performance. J Financ 49:1015–1040Pedrosa M, Roll R (1998) Systematic risk in corporate bond spreads. J Fixed Inc 8:7–26Song MH, Walking RA (2000) Abnormal returns to rivals of acquisition targets: a test of the ‘acquisition

probability hypothesis’. J Financ Econ 55:143–171Vassalou M, Xing Y (2004) Default risk in equity returns. J Financ 59:831–868Xu T, Najand M, Ziegenfuss DE (2006) Intra-industry effects of earnings restatements due to accounting

irregularities. J Bus Financ Account 33:696–714

Negative stock price surprises

123