Return predictability: The dual signaling hypothesis of stock splits

• DOI: http://doi.org/10.1111/fire.12192
• Published date: 01 November 2019
• Date: 01 November 2019
• Author: Ghada Ismail,Lei Gao,Ahmed Elnahas

DOI: 10.1111/fire.12192

ORIGINAL ARTICLE

Return predictability: The dual signaling

hypothesis of stock splits

Ahmed Elnahas1Lei Gao2Ghada Ismail3

1Robert C. VackarCollege of Business &

Entrepreneurship, University of TexasRio

GrandeValley, Edinburg, Texas

2Ivy College of Business, Iowa State University,

Ames, Iowa

3FogelmanCollege of Business and Economics,

The University of Memphis, Memphis, Tennessee

Correspondence

AhmedElnahas, Robert C. Vackar College of

Business& Entrepreneurship, University of Texas

RioGrande Valley,Edinburg, TX 78539.

Email:ahmed.elnahas@UTRGV.edu

LeiGao, Ivy College of Business, Iowa State Uni-

versity,3342 Gerdin Business Building, Ames, IA

50011.

Email:lgao@iastate.edu

GhadaIsmail, College of Business and Economics,

Universityof Memphis, Fogelman, 3675 Central

Avenue#437, Memphis, TN 38152.

Email:gismail@memphis.edu

Abstract

This paper aims to differentiate between optimistic splits and

overoptimistic/opportunistic splits. Although markets do not dis-

tinguish between these two groups at the split announcement

time, optimistic (overoptimistic/opportunistic) splits precede pos-

itive (negative) long-term buy-and-hold abnormal returns. Using

the calendar month portfolio approach, we show that the zero-

investment, ex ante identifiable, and fully implementable trading

strategy proposed in this paper can generate economically and

statistically significant positive abnormal returns. Our findings

indicate that pre-split earnings management and how it relates

to managers’ incentives, is an omitted variable in the studies of

post-split long-term abnormal returns.

KEYWORDS

dual-signaling hypothesis, earnings management, long-term stock

returns, stock splits

JEL CLASSIFICATIONS

G11, G12, G14, G35, M41

1INTRODUCTION

Since the seminal work of Fama, Fisher,Jensen, and Roll (1969; henceforth, FFJR), the finance literature provides sev-

eral alternative explanations for conducting a stock split. A manager can conduct a stock split to signal positive pri-

vate information (Desai & Jain, 1997; Lakonishok & Lev,1987; McNichols & Dravid, 1990), to lower a stock price to a

preferable range (Dyl & Elliott, 2006; Lamoureux & Poon, 1987), and to boost stock liquidity (Lin, Singh, & Yu,2009;

Muscarella & Vetsuypens, 1996).1However, Weld, Michaely, Thaler, and Benartzi (2009) argue that none of these

explanationsactually justify conducting a stock split. In order to further explore the signaling hypothesis, recent papers

have investigated the dual use of a stock split and earnings management. For example, Louis and Robinson (2005)

argue that combining a split and discretionary accruals may bean effective means of communicating managers’ private

1Another explanationfor the use of stock splits is the desire of companies to supply shares at lower prices when investors are more willing to pay premiums

forcheaper stocks. For a comprehensive review of different motives for stock splits, please see Minnick and Raman (2014).

Financial Review.2019;54:801–831. wileyonlinelibrary.com/journal/fire c

2019 The Eastern Finance Association 801

802 ELNAHAS ET AL.

information, rather than a means of deceiving shareholders. They state that “the stock split signal lends credibility to

the accrual signal whereas the accrual signal reinforces the split signal.”2By contrast, Guo, Liu, and Song (2008) argue

that managers support their already inflated stock prices (due to aggressive earnings management) by announcing

stock splits. They argue that managers use that approach to delay stock price corrections before stock-financed acqui-

sitions. The extant literatureon stock split return predictability does not differentiate between announcement and/or

long-term returns of optimistic splits (Louis & Robinson, 2005) and opportunistic splits (Guo et al., 2008). This paper

aims to contribute to this gap in the stock split return predictability literature.

Investigating the dual use of stock splits and earnings management, and its implications on the announcement as

well as long-term split returns would enable us to better understand the market reaction to different types of splits

and to better understand results in the prior literature.

In this paper,we try to differentiate between two groups of stock splitters. The first group (optimistic splitters) con-

sists of firms that conduct stock splits without a high degree of earnings management to conveypositive private infor-

mation. The second group (overoptimistic/ opportunistic splitters) consists of firms that combine stock splits with a

high degree of earnings management as a reflection of overestimationof, or to send a false signal about, future earnings

streams. Specifically,we use the degree of pre-split earnings management (through both discretionary accruals [Dacc]

and real activities management [RAM] measured by abnormal cash flows [Acfo]) as a differentiating factor between

these two groups of stock splits. Our results show that investorsdo not differentiate between these two types of stock

splits at the time of the announcement. However, these two types havestrikingly different long-term returns. Splits

conducted by optimistic firms are followed by significantly positive long-term abnormal returns. By contrast, splits

conducted by overoptimistic/opportunistic firms are followed by significantly negative long-term abnormal returns.

Tothe best of our knowledge, this paper provides the first empirical evidence that long-term returns are significantly

negative for an exante identifiable group of stock splits.

Using tercile double ranking, we construct nine portfolios of stock splits ranked based on the degree of earn-

ings management (henceforth, portfolios M1:M9). Portfolio M1 includes firms in the bottom discretionary accruals

and RAM terciles prior to a stock split. At the other extreme, portfolio M9 includes firms in the top discretionary

accruals and RAM terciles prior to a stock split. We argue that portfolios at the top of this ranking are more asso-

ciated with managerial overoptimism/opportunism while portfolios at the bottom of this ranking are more associ-

ated with managerial optimism. We first investigate whether investors differentiate between optimistic splits and

overoptimistic/opportunistic splits at the time of the split announcement. Our results indicate that investors do not

differentiate between these two groups of splits at the time of the announcement. Inversely, there is weakevidence

that investors have a better reactionto overoptimistic/opportunistic splits. This evidence is consistent with Louis and

Robinson (2005).

Next, we investigate the long-term returns of optimistic splits and overoptimistic/opportunistic splits. The main

premises of this paper are that optimistic splits are expectedto be followed by positive abnormal returns while overop-

timistic/opportunisticsplits are expected to observe return reversals in the post-split period. Our buy-and-hold abnor-

mal returns (BHAR) results show that the 1-year BHAR for portfolio M1 is 5.5%, a figure in stark contrast to that of

portfolio M9, which has a 1-year BHAR of −16% during a comparable period. In addition to using the BHAR approach,

we test our conjectures using the calendar month portfolio analysis. Results of the calendar month analysis show that

the equally (value) weighted zero-investment portfolio that buys portfolio M1 and sells short portfolio M9 would gain

100 (90) basis points per month for a 12-month holding period.

Since earnings management estimates are mandatory for our portfolio formation, the use of split events conducted

beforethe release of annual reports would create a look-ahead bias in our results. So, to ensure that our proposed strat-

egy is fully implementable and to avoidthe look-ahead bias issue, we exclude all splits announced within several annual

report release periods (45, 60, and 75 days after the fiscal year end). This filtrationdoes not hinder the profitability of

the calendar month trading strategy.

2SeeLouis and Robinson (2005, p. 361).