The more we know, the less we agree: A test of the trading horizon heterogeneity theory

• Published date: 01 February 2022
• Author: Lili Dai,Jerry T. Parwada,Donald W. Winchester,Bohui Zhang
• Date: 01 February 2022
• DOI: http://doi.org/10.1111/fire.12282

DOI: 10.1111/fire.12282

ORIGINAL ARTICLE

The more we know, the less we agree: A test of the

trading horizon heterogeneity theory

Lili Dai1Jerry T.Parwada2Donald W. Winchester3

Bohui Zhang4

1School of Accounting, UNSW Business

School, UNSW, Sydney,New South Wales,

Australia

2School of Banking and Finance, UNSW

Business School, UNSW, Sydney,New South

Wales,Australia

3AustralianInstitute of Business, Adelaide,

South Australia,Australia

4Shenzhen Finance Institute, Chinese

University of Hong Kong,Shenzhen, China

Correspondence

DonaldW. Winchester,Australian Institute of

Business,Adelaide 5000, SA, Australia.

Email:donald.winchester@aib.edu.au

Abstract

We examinethe Kondor theoretical explanation of an endur-

ing puzzle: trading volumes and stock return volatility peak

after the release of public information. Using a compre-

hensive data set of institutional holdings and earnings

announcements, we find supporting evidence that the pro-

portion of short-term investors is positively associated with

post-announcement spikes in trading volume and return

volatility. This finding survives in the identification test

based on the annual reconstitutions of the Russell 1000 and

2000 indices. We show our results largely withstand sev-

eral alternative explanations related to the constitution of

institutional investors, informed trading,and heterogeneous

beliefs.

KEYWORDS

higher order expectations, noisy rationalexpectations model, public

information, trading volume

JEL CLASSIFICATION

D82, D84, G12, G14

1INTRODUCTION

Over the past five decades and across various markets, the literature documents a substantial increase in trading

volume after a public information release.1This stylized fact defies the common intuition that public information

1The literatureshows that trading volume increases substantially after earnings announcements in the stock market (Chae, 2005; Kandel & Pearson, 1995;

Krinsky & Lee,1996). For example, earnings announcements trigger significant abnormal trading volume, even when the announcement returns are close to

zero(Kandel & Pearson, 1995). A similar trading pattern around public news releases is documented in the bond market (Fleming & Remolona, 1999; Green,

2004)and in the currency market (Evans & Lyons, 2008; Love & Payne,2008).

Financial Review. 2022;57:45–67. wileyonlinelibrary.com/journal/fire ©2021 The Eastern Finance Association 45

46 DAI ET AL.

generates consensus among investorsand thereby should lower their trading incentives. A question pertaining to this

puzzle is often raised by financial economists: How can investorsreceive the same public information and draw oppo-

site conclusions? Toanswer this question, Kondor (2012) considers the impact of heterogeneous trading horizons and

develops a new theory which shows that public announcements can polarize investor valuation of a stock without

polarizing their fundamental expectations. We test the Kondor theory on how the presence of short-horizon traders

in a stock affects abnormal trading volume around earningsannouncements.

In the Kondor (2012) model, short-horizon traders buy stocks now for resale to other investors later.In pursuit of

trading profits, short-horizon traders care about future sale prices more than fundamentals. A public announcement

reduces uncertainty about fundamentals but increases disagreement among short-horizon traders who try to guess

otherinvestors’ fundamental expectations. This disagreement among short-horizon traders translates into speculative

trading after the announcement. As such, public announcements are accompanied by tradingvolume spikes as long as

a certain fraction of investors havea short trading horizon. Kondor’s novel theory goes a long way toward solving the

enduring trading volume puzzle, which standard models do not address.

Kondor(2012, p. 1177) states, in particular, that the trading horizon heterogeneity theory is testable: “. .. myresults

imply that the trading volume of assets with a larger share of short-horizon investors in their investor base should

respond more strongly to public announcements. This is testable, as recent work has constructed empirical proxies

for the investment horizon of the investorbase of financial assets.” Following this theoretical motivation, we examine

whether tradingvolume and return volatility around earnings announcements are indeed more pronounced for stocks

with a larger proportion of short-horizon traders.

We choose quarterly (annual) earnings announcements as public information releases, for two reasons. First, earn-

ings announcements contain the most value-relevant information about a firm’s past profitability,which investors use

to project future operationalperformance (Beyer et al., 2010). Second, the Securities and Exchange Commission (SEC)

has mandated quarterly reporting for all public firms in the United States since 1970, rendering the timing and content

of public announcements comparable across firms pursuant to regulation.

Our empirical analyses support the Kondor’s (2012) theory through four pieces of evidence. First, when there is

a larger share of short-horizon traders in a stock’s investor base, we observe a more pronounced abnormal trading

volume and stock return volatility in post-earnings announcement windows. This finding is statistically robust and

economically significant in both univariate and multivariate analyses. For example,the multivariate test shows that a

1 standard deviation increase in the proportion of short-horizon traders is associated with about a 23% increase in

abnormal trading volume and an 11% increase in abnormal return volatility,both relative to their sample means.

Second, endogeneity appears as an important considerationfor our empirical tests because (1) both trading activity

andinvestor horizon can be endogenously determined by unobservable factors, and (2) short-term investors may favor

stocks with large transaction volume and return volatility around earnings announcements. Tomitigate these endo-

geneity concerns, we stress test our findings using the annual reconstitutions of the Russell 1000 and 2000 indices as

an exogenous shock to the proportion of short-term investors. The Russell 1000 and 2000 indices areformed annu-

ally based on stock marketcapitalization. The largest 1000 stocks are included in the Russell 1000 Index, and the next

2000 stocks are in the Russell 2000 Index. Therefore, close to the Russell 1000/2000 threshold, the inclusion in one

of the two Russell indices is quasi-random with respect to firm fundamentals. Because the Russell indices are value-

weighted,stocks just included and just excluded in the Russell 1000 Index have a large difference in their weights in the

indices. This large discontinuity in Russell index weights produces a substantial difference in interest from short-term

investors, confirmed across different studies. For example, Crane et al. (2016) find that a causal and positive effect

of index inclusion on institutional ownership is evident for transient investors, but not for dedicated owners. Simi-

lar evidence excluding long-term or dedicated investorsfrom those that react to Russell Index reconstitutions at the

threshold is provided byBoone and White (2015) and Cremers et al. (2020), among others. Accordingly, exploiting this

threshold as a source of exogenous variation in transientownership, we implement a regression discontinuity design

(RDD) and obtain evidence consistent with our baseline analyses.