Volatility measures as predictors of extreme returns

• Author: Lorne N. Switzer,Yun Zhao,Cagdas Tahaoglu
• DOI: http://doi.org/10.1016/j.rfe.2017.04.001
• Date: 01 November 2017
• Published date: 01 November 2017

Volatility measures as predictors of extreme returns☆

Lorne N. Switzer ⁎, Cagdas Tahaoglu, Yun Zhao

John MolsonSchool of Business, ConcordiaUniversity, 1455 De MaisonneuveBlvd. W., Montréal,Québec H3G 1M8, Canada

abstractarticle info

Articlehistory:

Received1 December 2016

Accepted5 April 2017

Availableonline 7 April 2017

JEL classification:

G10

G11

G14

G17

This paper examinesthe relationship betweenvolatility and the probabilityof occurrence of expected extreme

returns in the Canadian market. Fourmeasures of volatility are examined: implied volatility from firm option

prices, conditional volatility calculated using an EGARCH model, idiosyncratic volatility, and expected shortfall. A

significantly positive relationship is observed between a firm's idiosyncratic volatility and the probability of occur-

rence of an extreme return in the subsequent month for firms. A 10% increase in idiosyncratic volatility in a given

month is associated with the probability of an extreme shock in the subsequent month (top or bottom 1.5% of the

returns distribution) of 26.4%.Other firm characteristics, includingfirm age, price, volume and book-to-market

ratio, are also shown to be significantly relatedto subsequent firm extreme returns. The effects of conditional

and implied volatility are mixed. The E-GARCH and expected shortfallmeasures of conditional volatility are con-

sistentwith mean reversion:high shortterm realizationsof conditional vol atility foreshadow a lower probability

of extreme returns.

© 2017 Elsevier Inc. All rights reserved.

Keywords:

Extremereturns

Impliedvolatility

Conditionalvolatility

Idiosyncraticvolatility

Expectedshortfall

1. Introduction

The study of extreme returns has been of increased interest to re-

searchers and practitioners in recent years.Analyzing extreme returns

is important because of their potential to explain investor behavior. For

example, An (2016) finds that investors sell more when they have stocks

with both larger unrealized gains and larger unrealized losses. Since the

higher selling pressure of stocks with extreme returns leads to lower cur-

rent prices and higherfuture returns, such investor's trading behavior

can affect equilibrium price dynamics. Bali, Cakici, and Whitelaw

(2011) provide evidence indicatinga negative and significant relation

between the maximum daily return over the past one month (MAX)

and expected stock returns. This result is consistent with under-diversi-

fication of portfolios and investor preferences for stocks with lottery like

payoffs. Longin (2000,2016) shows the importance of focusing on ex-

treme values for portfolio stress testing. Fodor, Krieger, Mauck, and

Peterson (2013) demonstrate that, in the U.S. market, when the ex

ante (predicted) extreme return stocks are removed from a portfolio,

the betas decrease and the overall portfolio performance improves.

Identifying and excluding predicting extremereturn firms is therefore

a challenge,but may be of great importance for investors. The purpose

of this study is to try to identify factors that may be helpful in predicting

extreme returns in the Canadian stock market.

Most extant work in this area looks at US markets. Fodor et al. (2013)

show a positive relationship between implied volatility and extreme

returns from option prices in the U.S. market. This study extends Fodor

et al. (2013) by looking at the extremereturn predictive ability of im-

plied volatility from option prices, as well as from conditional volatility

calculated from an EGARCH model, idiosyncratic volatility calculated by

using the Fama and French (2015) five-factor model, and expected

shortfall.

Other explanatory factors,which are shown to have predictive abil-

ity for extremereturns in past research, are also included in this study.

These factorsinclude: firm size, book-to-marketratios, trading volume

and equity price.

This study focuses on firm data in the Canadian stockmarket from

2001 to 2014.Probit regressionmodels are estimatedto identify the re-

lationship between the probability of occurrence of expected extreme

returns and firm characteristics . The analyses are conducted for the

market as a whole, as well as forfirms in the natural resource industry,

an importantsector for the Canadian economy,which has experienced

considerablevolatility in recent years.

Four volatility models are tested as predictorsof the probability of

extreme return behavior. The first model uses implied volatility from

option prices. The second model uses conditional volatility from an

Reviewof Financial Economics35 (2017) 1–10

☆Finance Department, ConcordiaUniversity. Financial support from the Autorité des

MarchésFinanciers and the SSHRC to Switzeris gratefullyacknowledged. We would like

to thank the Editor, Gera ld Whitney, an anonymous re feree, Menachem Brenner ,

François Longin,and seminar participants at the 2016 IRMC conference in Jerusalem for

their valuablecomments and suggestions.

⁎Correspondingauthor.

E-mailaddress: lorne.switzer@concordia.ca(L.N. Switzer).

http://dx.doi.org/10.1016/j.rfe.2017.04.001

1058-3300/©2017 Elsevier Inc. All rightsreserved.

Contents listsavailable at ScienceDirect

Review of Financial Economics

journal homepage: www.elsevier.com/locate/rfe