Abnormal trading behavior of specific types of shareholders before US firm bankruptcy and its implications for firm bankruptcy prediction

• Author: William J. Moser,Christine Cheng,Stewart Jones
• DOI: http://doi.org/10.1111/jbfa.12338
• Date: 01 October 2018
• Published date: 01 October 2018

DOI: 10.1111/jbfa.12338

Abnormal trading behavior of specific types of

shareholders before US firm bankruptcy and its

implications for firm bankruptcy prediction

Christine Cheng1Stewart Jones2William J. Moser3

1PattersonSchool of Accountancy, The Univer-

sityof Mississippi, MS, USA

2TheUniversity of Sydney Business School, The

Universityof Sydney, NSW, Australia

3FarmerSchool of Business, Miami U niversity

Oxford,OH, USA

Correspondence

WilliamJ. Moser, 2027 Farmer School of Busi-

ness,800 E. High St., Oxford, OH 45056.

Email:moserwj@miamioh.edu

JELClassification: G11, G20, G33

Abstract

This paper examines the trading behavior of US corporate insid-

ers and certain groups of institutional investors (short-term, tran-

sient, top-performing, and those with fiduciary responsibility) in

the eight quarters leading up to a US firm bankruptcy filing. Using

a matched sample based on year, industry, and a probability of

future bankruptcy model, we find that US corporate insiders dis-

play abnormal reduced net trading activity in the quarters before

bankruptcy, with corporate insiders ‘going quiet’ immediately pre-

ceding a US bankruptcy. In contrast, we find that our identified

types of institutional shareholders display abnormal selling activity

several quarters before a US bankruptcy. We then use this infor-

mation to enhance the bankruptcy-predictive capabilities of recent

machine-learning techniques such as gradient boosting, as well as

the probability of future bankruptcy model. We find that the vari-

ables measuring the absolute value of net purchases by US corpo-

rate insiders in the two quarters prior to bankruptcy, along with

the changes in ownership by specific types of institutional share-

holders, improvethe out-of-sample predictive capabilities of our two

different bankruptcy prediction models. Overall, we find that spe-

cific types of shareholder display abnormal trading in the quarters

preceding US firm bankruptcy, and such information improves the

out-of-sample accuracy of firm bankruptcy prediction models.

KEYWORDS

bankruptcy prediction, financial distress, insider ownership,

institutional ownership

1INTRODUCTION

In the 1980s, pre-bankruptcy shareholders of US firms were able to recover at least part of their investmentapproxi-

mately 78% of the time after a firm went through the bankruptcy process (LoPucki & Whitford, 1993). However,a new

survey,using more recent data, indicates that less than 10% of the pre-bankruptcy shareholders of US firms were able

1100 c

2018 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/jbfa JBus Fin Acc. 2018;45:1100–1138.

CHENG ET AL.1101

to recover any portion of their investment(Wood, 2011). Additionally, according to data provided by New Generation

Research, Inc., the combined asset value of the 52 US public companies that filed for bankruptcy in 2014 was US$ 71.8

billion.1The convergence of the decline in recovery rate for equity holders in US bankruptcies and the potential for

substantial losses makes bankruptcy a catastrophic event for shareholders. As a result, shareholders in US firms have

increased and renewed interest in improving methods for predicting impending bankruptcy.

Lennox(1999) finds that bankruptcy prediction models, such as those developed by Altman (1968), Merton (1974),

Neophytou and Molinero (2004), Ohlson (1980), Shumway (2001), Dewaelheyns and ValHulle (2006) and Campbell,

Hilscher, and Szilagyi (2008), are generallyable to outperform bankruptcy predictions based on auditor-issued going

concern opinions. Therefore, Lennox (1999) suggests that stakeholders rely on these bankruptcy prediction models

for decision making. These existing US bankruptcy prediction measures focus primarily on accounting financial state-

ment information and market return data. However,Piotroski and Roulstone (2004) identify two potentially impor-

tant sources of firm-specific information: (1) corporate insider trading activity, and (2) trading behavior of specific

categories of institutional investors. The findings by Piotroski and Roulstone (2004) suggest that adding informa-

tion related to investor trading activity to the current US bankruptcy prediction models mayenhance their predictive

capabilities.

We use two steps to evaluate whether information related to investor trading activity in US firms may enhance

existing bankruptcy prediction models. First, we investigatewhether US corporate insiders and specific types of insti-

tutional investorsengage in abnormal trading within several quarters of firm bankruptcy. Second, we evaluate whether

the observed trading activity of US corporate insiders and that of specific types of institutional shareholders, in con-

junction with other bankruptcy prediction variables, enhances the out-of-sample predictive performance of existing

bankruptcy prediction models. To do so, we rely on the recent studies by Jones (2017), and Jones, Johnstone, and

Wilson (2015, 2017) that introduce a series of machine-learning techniques to the corporate bankruptcy and credit

ratings literature. As an additional check, we consider whether the inclusion of the trading behavior of the specific

groups of investorsthat do demonstrate abnormal trading behavior improves the out-of-sample predictive capabilities

of the machine-learning bankruptcy prediction model after controlling for the variables from the Campbell et al. (2008)

bankruptcy prediction model and the Jones et al. (2017) model.

Corporate insiders haveaccess to information not necessarily disclosed to the market. Since the trend in the United

States is for equity holders in bankrupt firms to recover none of their investment (Wood, 2011), corporate insiders

stand to lose a significant amount oftheir non-diversified(Fama, 1980) personal wealth in the event of firmbankruptcy.

Despite access to private information, the ability of US corporate insiders to trade on non-public information is con-

strainedby the Securities Exchange Act of 1934, promulgation of SEC Regulation FD in 2002, enhanced enforcement of

insider trading regulations by the SEC, and increased risk of litigation byexisting shareholders (Gao, Ma, & Ng, 2015).2

Several empirical studies provide evidence that these trading disincentives maysuppress insider trades prior to bad

news announcements (Gao et al., 2015; Korczak, Korczak, & Lasfer, 2010; Marin & Oliver, 2008). However, despite

these safeguards, other academic studies provide evidence that insiders trade prior to bad news events (Agrawal &

Cooper, 2015; Griffin, Lont, & McClune, 2014; Ke,Huddart, & Petroni, 2003). Korczak et al. (2010, p. 369) note that

insiders’ trading decisions ‘result from a trade-off between their incentives to capitalize on foreknowledge of the dis-

closure and the risk of regulatory scrutinyand reputational loss’. While the existenceof other academic studies examin-

ing corporate insider trading before the release of negative firm news maybe insightful for our issue, firm bankruptcy

represents an extreme environment of permanent loss in which we can examine US corporate insider trading deci-

sions. Thus, the conflicting results in these studies may reflect differences in settings, in which US corporate insiders

must weigh the gains from using their insider information for personal profit against the risk of regulatory scrutiny and

reputational loss that insider trading lawsuits maybring.

1NewGeneration Research, Inc. www.bankruptcydata.com website valid as of 6/30/2017.

2TheCornerstone Research Database maintained by Stanford Law School reports a dramatic increase in the number of Security Class Action Lawsuits against

CorporateInsiders in 2001 and 2002, corresponding with the collapse of Enron Corporation, WorldCom, Inc. and Tyco International Ltd.

1102 CHENG ET AL.

Unlike managers who have private information, institutional investorsmay expend considerable time and effort to

acquire and process all available information. Initial research in the area of institutional investors or institutional own-

ership in the periods preceding bankruptcy treats these investors as a homogeneous group. Forinstance, Frino, Jones,

Lepone, and Wong (2014) examine aggregate institutional ownership in the periods leading up to firm bankruptcy.

Using proprietary ASX data, Frino et al. (2014) find that most institutional investors retain their equity interests

through the bankruptcy process, with only active institutional investors gradually selling their equity interests in the

115 days leading up to the Australianfirm's bankruptcy announcement.

Frinoet al.’s (2014) finding regarding active institutional investors is consistent with the largely US-based research,

which examines institutional investor trading behavior prior to various non-bankruptcy news events.This US-based

research generally recognizes the heterogeneity of institutional investors and attempts to partition them into cat-

egories based on different investment horizons, different investment objectives and styles, and different fiduciary

or legal restrictions (Yan & Zhang, 2009). Previous research shows that specific types of institutional shareholders

are able to consistently display superior information-processing capabilities by outperforming other types of insti-

tutional shareholders in particular information environments (Yan & Zhang, 2009). We focus on four institutional

investor groups empirically identified by previous research as outperforming other groups of institutional investors

ahead of certain information events, either due to superior information processing3or an aversion to risk: (1) insti-

tutional investors with a short-term investment horizon (Yan& Zhang, 2009), (2) transient institutional shareholders

(Bushee, 1998), (3) top-performing institutional investors (Adebambo, Brockman, & Yan,2015), and (4) those institu-

tional investors with fiduciary responsibilities (DelGuercio, 1996).

Despite the evidencethat these four institutional investorgroups tend to outperform other groups ahead of certain

information events,Ramalingegowda (2014)is the only US-based study that uses the Bushee (1998) institutional own-

ership classification system and examines dedicated, transient and quasi-indexerinstitutional investor trading behav-

ior ahead of bankruptcy announcements. He finds that as opposed to transient institutional shareholders, dedicated

institutionalinvestors are generally more likely to display abnormal equity sales before US firm bankruptcy.Ramalinge-

gowda's (2014) findings are consistent with the notion that due to their long-term relationship with the firm, dedicated

institutional investors possess superior information-processing capabilities and use them to sell their equity positions

before firm bankruptcy.However, Ramalingegowda's (2014) results conflict with the theory in Yan and Zhang (2009),

who suggest that transient or short-term institutional investors are more likely to sell their equity interests before

US bankruptcy,while dedicated or long-term institutional shareholders tend not to engage in activetrading behavior.

Thus, the two studies that offer insight specifically into institutional investor tradingbehavior offer mixed evidence as

to which groups of institutional investors will haveabnormal trading behavior in advance of US firm bankruptcy.

Using a matched sample design, we investigateabnormal trading behavior for each of our investor types. We match

firms that filed for bankruptcy between 1992 and 2012 with similarly distressed firms that survived financial distress.

Our evidence suggests that SEC enforcement, along with shareholder litigation risk, may influence corporate insid-

ers to forgo trading several quarters ahead of a bankruptcy filing. Furthermore, we find that short-term institutional

shareholders, transient institutional shareholders, top-performing institutional shareholders, and institutional share-

holders with fiduciary responsibilities demonstrateabnormal trading behavior by selling equity in firms that ultimately

declare bankruptcy while retaining the equity in firms that ultimately survive financial distress. In our supplemental

section, contrary to Ramalingegowda (2014), we do not find any evidence that dedicated institutional shareholders

display abnormal trading in the quarters before US firm bankruptcy.

After examining the abnormal trading of specific types of shareholders before US firm bankruptcy,we next inves-

tigate whether the equity trading by specific types of institutional shareholders and the absence of net trading

by corporate insiders provides incremental information to US bankruptcy prediction models. First, we employ the

advanced machine-learning technique gradient boosting to compare the relativevariable importance (RVIs) of different

bankruptcy predictors. In recent studies, Jones (2017) and Jones et al. (2015, 2017) introduced a series of advanced

3Consistent with Ramalingegowda (2014), Yanand Zhang (2009), and Adebambo, Brockman, and Yan (2015), we define the term ‘superior information pro-

cessing’as the institutional shareholders’ superior processing of both non-public and public information.