SKEWNESS AND THE ASYMMETRY IN EARNINGS ANNOUNCEMENT RETURNS

• Date: 01 June 2015
• Author: J. Michael Pinegar,Ryan J. Whitby,Benjamin M. Blau
• DOI: http://doi.org/10.1111/jfir.12056
• Published date: 01 June 2015

SKEWNESS AND THE ASYMMETRY IN

EARNINGS ANNOUNCEMENT RETURNS

Benjamin M. Blau

Utah State Unviersity

J. Michael Pinegar

Brigham Young University

Ryan J. Whitby

Utah State University

Abstract

Much of traditional asset pricing theory rests on the assumption of normality in the

distribution of stock returns. A growing body of research suggests that skewness in the

return distributions can affect asset prices. In this article we attempt to empirically

identify factors that influence return skewness. Consistent with the theoretical literature,

we find that prices during the postearnings announcement period are more convex for

firms that have tighter short-sale constraints and for firms that experience greater

disagreement among investors. Perhaps more important, we also find that price

convexity is a key determinant in the skewness of stocks.

JEL Classification: G10, G12, G14

I. Introduction

Much of traditional asset pricing theory rests on the assumption of normality in the

distribution of stock returns. However, prior research documents negative skewness in

market returns (e.g., Black 1976; Christie 1982; Schwert 1989; Bekaert and Wu 2000)

and positive skewness in individual stock returns (Harvey and Siddique 1999, 2000;

Chen, Hong, and Stein 2001). The presence of skewness might introduce bias into some

of the asset pricing models that are used by both academics and practitioners.

Furthermore, more recent research suggests that investors might have preferences for

skewness, that will affect asset prices more generally. For instance, Barberis and Huang

(2008) use prospect theory to show that some investors will overweight the tails in return

distributions, thus leading to skewness preferences. Barberis and Huang show that these

preferences, if strong enough, can lead to stock price premiums and subsequent

underperformance. Empirical evidence seems to confirm these theoretical predictions as

The authors gratefully acknowledge helpful comments from Tyler Brough, Jared DeLisle, Anders Ekholm,

Paul Koch, Jason Smith, Zhe Zhang, the editor (Scott Hein), and seminar participants at Utah State University and

the 2012 Financial Management Association annual meetings.

The Journal of Financial Research Vol. XXXVIII, No. 2 Pages 145–168 Summer 2015

145

© 2015 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

stocks with the most positive skewness significantly underperform stocks with the most

negative skewness (Mitton and Vorkink 2007; Kumar 2009; Boyer, Mitton, and Vorkink

2010; Kumar, Page, and Spalt 2011). Given the evidence that indicates that skewness

affects asset prices, an important extension to this growing body of research might be to

identify factors that lead to skewness in the distribution of stock returns. The main

objective of this study is to provide these tests.

Underlying most theories of skewnes s are the assumptions of short-sale

constraints and heterogene ous expectations (Hong and Stei n 2003; Diamond and

Verrecchia 1987). Xu (2007) nicely develops the theoretical r elation between both

short-sale constraints an d heterogeneous beliefs and skewness. He argues that in t he

presence of constraints and he terogeneity, stock prices react more to good news than to

bad news, which will result in convex prices across the news signal and lead to more

positive return skewness. I n some initial tests, Xu shows that, in general, postearnings

announcement returns are indeed convex and that both short-sale co nstraints and

heterogeneous beliefs dire ctly affect skewness. In this article, we extend these

empirical results by examining several other predictions in Xu that have not been

tested. In particular, we test whether the price convexity in postannouncement returns

is directly related to both shor t-sale constraints and hetero geneous beliefs.

Furthermore, we test whether pr ice convexity affects the skewness in stock return

distributions.

Consistent with the predict ions in Xu (2007), we first find that price responses

following earnings announc ements are more convex for firms that ha ve tighter short-

sale constraints. These results hold in both univariate sorts and multivariate tests, and

suggest that stocks prices reac t more to good news than to bad news. Second, we find

weak evidence that postanno uncement price convexity is al so driven by stocks with

greater heterogeneity when approximated with return volatility. We do not find a

significant link between price convexity and share turnov er, which has also been

shown to proxy for heterogene ous beliefs (Xu 2007; Berkman et al. 2009). Therefore,

the evidence between price convex ity and heterogeneity is weak at best . Third, we

examine whether price convexity is positively related to skewness. Results show that

the convexity in postannounc ement returns is a key determinant in contemporaneous

skewness for our sample of NYSE stocks. We are not, however, able to draw similar

conclusions when focusing on our sam ple of NASDAQ-listed stocks. To the ext ent

that both short-sale constraints and heterogeneous belief s can affect the skewness of

returns (as in Xu 2007), our empirical tes ts show that the relation between const raints

and skewness can be explained by constraint-induced price convexity. After

estimating this convexity, we show that NYSE-listed stocks with the gr eatest

convexity generally have the hi ghest return skewness. Combine d with our earlier

findings, these results seem to indicate that although the lack of short selling and

greater disagreement among i nvestors can lead to higher skewne ss, the results from

our tests suggest that the mechanism through which this occurs is throu gh the

convexity associated with postannouncement returns. Give n the growing body of

research that discusses potential biases in asset prices cau sed by skewness, our

findings contribute to this literature by identifying how stock returndistributions might

become more skewed.

146 The Journal of Financial Research