Idiosyncratic Volatility, Institutional Ownership, and Investment Horizon

• Published date: 01 September 2015
• Author: Doina C. Chichernea,Blerina Bela Zykaj,Alex Petkevich
• Date: 01 September 2015
• DOI: http://doi.org/10.1111/j.1468-036X.2013.12033.x

Idiosyncratic Volatility, Institutional

Ownership, and Investment Horizon

Doina C. Chichernea, Alex Petkevich and

Blerina Bela Zykaj

John and Lillian Neff Department of Finance, College of Business and Innovation, University of Toledo,

Mail Stop 103, 2801 W. Bancroft Street, Toledo, Ohio 43606‐3390, USA

E-mails: Doina.Chichernea@utoledo.edu; Alexey.Petkevich@utoledo.edu; Blerina.Reca@utoledo.edu

Abstract

This paper reevaluates the cross‐sectional effect of institutional ownership on

idiosyncratic volatility by conditioning on institutions’investment horizon. Prior

literature establishes a positive link between growing institutional ownership and

idiosyncratic volatility. However, this effect may vary depending on the type of

institutional ownership. We document that short‐term (long‐term) institutional

ownership is positively (negatively) linked to idiosyncratic volatility in the cross

section. These opposite effects persist after controlling for institutional preferences

and information‐based trading and remain qualitatively unchanged after

controlling for endogeneity. This suggests that short‐term (long‐term) institutions

exhibit higher (lower) trading activity, which increases (decreases) idiosyncratic

volatility.

Keywords: institutional investors, idiosyncratic volatility, investment horizon,

trading preference

JEL classification: G12, G23

1. Introduction

This article investigates the cross‐sectional relation between institutional ownership and

idiosyncratic volatility (IV), accounting for differences in the trading horizon of

institutional investors. The connection between institutions and IV has been extensively

studied in the literature, mainly in the context of identifying the determinants of aggregate

IV. For example, Campbell et al. (2001), Dennis and Strickland (2004), Sias (1996), and

The authors wish to thank an anonymous referee and John Doukas (the editor) for insightful

comments and useful suggestions. We are also thankful to Sachin Modi, Lawrence

Kryzanowski, Andrey Ukhov, and participants at the Financial Management Association,

European Financial Management Association and Midwest Finance Association Meetings

for helpful comments. Any remaining errors or omissions are the authors’alone.

European Financial Management, Vol. 21, No. 4, 2015, 613–645

doi: 10.1111/j.1468-036X.2013.12033.x

© 2013 John Wiley & Sons Ltd

Xu and Malkiel (2003) find a positive relation between total institutional ownership and

IV and argue that the upward trend in institutional ownership is primarily responsible for

the increase in aggregate idiosyncratic risk over time. Conversely, Zhang (2010) shows

that since 2001 aggregate IV consistently declined, while institutional ownership has

maintained an upward trend. Brandt et al. (2010) argues that, given the trend reversal of

aggregate IV, its episodic increase is more likely to be due to retail investors than to

institutional investors.

1

Rather than focus on the aggregate time‐series trend, our study investigates whether

different types of institutional ownership have different effects on the cross‐sectional

variation in IV. We argue that ignoring the heterogeneity of institutions and studying them

as a group can produce confounding results. Today’s trading environment is characterised

by not only a significant increase in the presence of institutional investors,

2

but also large

variation in their trading behavior, due to their different objectives, characteristics, limits,

and so forth. The fact that we often see institutions on both sides of the same trade is the

best supporting evidence that not all institutional trades are similarly motivated.

We verify this claim by examining the cross‐sectional relation between IV and

institutional ownership conditional on the investment horizon of institutional investors.

There are at least two reasons to believe that the relation between institutional ownership

and idiosyncratic risk is conditional on the investment horizon of institutional investors.

First, the average investment horizon of shareholders is indicative of the amount of

trading a particular stock will face. Ownership by institutions with different investment

horizons indicates different trading frequencies. In particular, stocks with primarily short‐

term institutional ownership are exposed to higher trading activity, while the opposite is

true for stocks with primarily long‐term institutional ownership. In this case, short and

long‐term institutional ownership represents trading (or lack thereof) of a stock and hence

should have opposite effects on the level of IV in the cross‐section. The relation between

trading volume and volatility has long been documented in the literature.

3

Therefore, if

institutional ownership is indicative of stock trading volume, we expect short‐term

institutional ownership to increase IV and long‐term institutional ownership to decrease it.

The remainder of this study refers to this as the trading argument.

Second, recent literature documents that institutions’investment horizon is a good

indicator of informational trading. For example, Yan and Zhang (2009) document that

short‐term institutions can forecast likely future stock returns due to their informational

advantage. On the other hand, Cremers and Pareek (2011) study the effect of institutional

1

Brandt et al. (2010) show that by 2003, aggregate IV had decreased to pre‐1990s levels. They

find no evidence that institutional investors caused the increase in IV in early 2000. As such,

the authors conclude that the increase could have been caused by retail investors.

2

See Jones and Lipson (2003) and Kaniel et al. (2008) for evidence of institutional investor

dominance in the financial market.

3

See Karpoff (1987) for a review on the early literature in this area. His synthesis of previous

research concludes that while there is a positive correlation between the absolute value of price

changes and volume, the relation between price changes per se and volume is non‐monotonic

and asymmetric (i.e., the correlation between volume and positive price changes is positive,

while that between volume and negative price changes is negative). Schwert (1989) and

Gallant et al. (1992) also provide good reviews of the empirical and theoretical research in this

area.

© 2013 John Wiley & Sons Ltd

614 Doina C. Chichernea, Alex Petkevich, Blerina Bela Zykaj

investment horizon on the efficiency of stock prices and conclude that short‐term

institutions are behaviorally biased and mostly overconfident and their presence helps

explain many stock returns anomalies. Regardless of whether this behavior is generated

by overconfidence or informational advantage, this line of thinking provides convincing

reasons to believe that institutions’investment horizon affects the relation between

institutional ownership and IV.

This argument also raises an important point, namely, that institutions with different

horizons can have different attitudes (preferences) toward idiosyncratic risk exposure.

Whether they are better informed (Yan and Zhang, 2009) or overconfident (Cremers and

Pareek, 2011), short‐term institutions are more likely to prefer stocks with higher IV

because they can take advantage of the (perceived) mispricing. The opposite is true for

institutions with longer investment horizons. Therefore, the investment horizon of

institutional investors could capture their preferences and hence influence the relation

between institutional ownership and IV. We henceforth refer to this as the preference

argument.

Following Yan and Zhang (2009), we use quarterly institutional holdings to construct

an investment horizon measure based on institutions’portfolio turnover. Based on this

measure, we classify institutions as short‐term (high‐turnover) and long‐term (low‐

turnover) investors and calculate the percentage ownership of short‐and long‐term

institutions at the stock level. Using daily data, we follow Ang et al. (2006) to estimate

quarterly IV relative to the Fama–French three‐factor model. Although our focus is not on

aggregate trends, we show that our measures are, in the aggregate, consistent with

previous literature. Specifically, we confirm that aggregate IV increases until 2001

(Campbell et al., 2001; Xu and Malkiel, 2003) and reverses after 2001, with the exception

of the 2008 financial crisis (Brandt et al., 2010; Zhang, 2010). Conversely, total

institutional ownership consistently increases during our entire sample period (from 20%

in 1980 to almost 60% in 2010). After decomposing total institutional ownership into its

short‐and long‐term components, we find that that aggregate long‐term institutional

ownership consistently dominates its short‐term counterpart after 2001.

After documenting the general aggregate dynamics of our variables of interest, we

focus on the cross‐sectional relation between IV and components of institutional

ownership. First, we examine the average IV of portfolios sorted independently by their

previous quarter institutional ownership and size. Univariate analysis shows that for the

entire sample period stocks with higher short‐term (long‐term) institutional ownership

have higher (lower) subsequent‐quarter IV. This relation is less pronounced among small

stocks, which can be explained by the low level of institutional presence in these stocks

overall.

Motivated by the above results, we use a standard Fama and MacBeth (1973) approach

to test whether our conclusions persist after controlling for other determinants of IV, such

as size, illiquidity, past IV, earnings volatility, and performance. Short‐term (long‐term)

institutional ownership is persistently positively (negatively) linked to future IV in the

cross section, providing additional support for the trading hypothesis. Subsample analysis

shows that although the relation between short‐term institutional ownership and IV is

relatively stable in the pre and post‐2001 period, the relation between IV and long‐term

ownership seems to weaken after 2001. Further investigation shows that changes in the

relative proportions of small capitalisation stocks in the market after 2001 might be

responsible for this result. Given the high level of IV and low institutional presence within

the group of small stocks, we expect that the relationship between ownership measures

© 2013 John Wiley & Sons Ltd

Idiosyncratic Volatility, Institutional Ownership, and Investment Horizon 615