Investor sentiment and management earnings forecast bias

• Author: Helen Hurwitz
• Published date: 01 January 2018
• DOI: http://doi.org/10.1111/jbfa.12282
• Date: 01 January 2018

DOI: 10.1111/jbfa.12282

Investor sentiment and management earnings

forecast bias

Helen Hurwitz

JohnCook School of Business, Saint Louis Univer-

sity,3674 Lindell Blvd., Saint Louis, MO63108,

USA

Correspondence

HelenHurwitz, John Cook School of Business,

SaintLouis University, 3674 Lindell Blvd., Saint

Louis,MO 63108, USA.

Email:helen.hurwitz@slu.edu

Abstract

This study investigates whether investor sentiment is associated

with behavioral bias in managers’ annual earnings forecasts that

are generally issued early in the year when uncertainty is relatively

high. I provide evidence that management earnings forecast opti-

mism increases with investor sentiment. Furthermore, I find that

managers’ annual earnings forecasts are more pessimistic during

low-sentiment periods than during normal-sentiment periods. Since

managers lack incentives to further deflate stock prices during a

low-sentimentperiod, this evidence indicates that sentiment-related

management earnings forecast bias is likely to be unintentional. In

addition, I find that the relationship between management earnings

forecast bias and investorsentiment is stronger for firms with higher

uncertainty, consistent with investor sentiment having a greater

influence on management earnings forecasts when uncertainty is

higher.

KEYWORDS

investor sentiment, management earnings forecast bias, managerial

behavioral bias

1INTRODUCTION

This study examines whether managers are susceptible to prevailingmarket sentiment when forecasting annual earn-

ings well before the yearend. Sentiment can be broadly defined as beliefs about future cash flows or discount rates that

are not consistent with prevailing fundamentals (e.g., Baker & Wurgler,2006). Prior evidence suggests that investors

and analysts are susceptible to sentiment and can form overly optimistic or pessimistic earnings expectations (e.g.,

Bergman & Roychowdhury,2008; Delong, Shleifer, Summers, & Waldman, 1990).Although managers have an informa-

tion advantage over outsiders about their own firms’ performance, they are not immune to behavioral bias and prior

research finds that they exhibit attribution bias and overconfidence (e.g., Libby & Rennekamp, 2012; Malmendier &

Tate,2005, 2008). Brown, Christensen, Elliott, & Mergenthaler (2012) suggest that sentiment affects managers’ non-

GAAP earnings disclosures, probably through its influence on managers’ own earnings expectations. I thus expectthat

managers, like investors and analysts, are susceptible to sentiment and that forecast bias in managers’ annual earn-

ings forecasts is associated with sentiment. This research has important implications for market participants relying

166 c

2017 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/jbfa JBus Fin Acc. 2018;45:166–183.

HURWITZ 167

on management earnings forecasts to form earnings expectations and for managers who are concerned about their

forecast reputations.

Thebehavioral finance literature defines investor sentiment as optimism or pessimism about stocks in general or the

deviation of investor beliefs about future firm performance from fundamental values (Baker & Wurgler,2006, 2007;

Delong et al., 1990; Morck, Shleifer,& Vishny, 1990). Prior research suggests that analysts’ earnings forecasts are more

optimistic during high-sentiment periods and that this bias in analysts’ earnings forecasts is likely to be unintentional

(Bergman & Roychowdhury,2008; Clement, Hales, & Xue, 2011; Hribar & McInnis, 2012). Compared to analysts, man-

agers haveprivate information about their own firms’ performance and thus may be affected less by investor sentiment

when forecasting earnings. However, investmentbanks devote a significant amount of resources to macroeconomic

research. With access to macroeconomic expertise, analysts are likelyto have an information advantage in identifying

investor sentiment and be more objective in forecasting earnings. Evidence of unintentional bias in analysts’ earnings

forecasts related to sentiment thus indicates that managers can be susceptible to investorsentiment when forecasting

annual earnings relatively early in the year.

Prior research provides evidence that managers exhibit behavioral bias, such as overconfidence and attribution

bias (Baker & Wurgler,2012; Ben-David, Graham, & Harvey, 2013; Bergman & Roychowdhury, 2008; Hribar & Yang,

2016; Libby & Rennekamp, 2012; Malmendier & Tate,2005, 2008). For example, Bergman & Roychowdhury (2008)

and Libby & Rennekamp (2012) suggest that managers place greater weight on internal, rather than external, factors

for good firm performance and that this attribution bias increases managers’ confidence in improved future firm per-

formance and thus affects their forecasting behavior. As stock prices increase with investor sentiment, attribution

bias can lead managers to overestimate the extent to which they contribute to good stock price performance and

thus become more confident in future firm performance. As a result, managers are likely to provide more optimistic

annual earnings forecasts as investorsentiment increases. This expectation is consistent with Brown et al.’s (2012) evi-

dence that managers tend to disclose non-GAAP earnings that exceedGAAP earnings when investor sentiment is high

and that managers’ sentiment-related earnings expectations play a role in their decisions about non-GAAP earnings

disclosures.

I further examine whether the expected sentiment-related management earnings forecast bias is attributable to

unintentional bias or managerial opportunism to perpetuate investor sentiment to influence stock prices. Since man-

agers have incentives to maintain or increase stock prices due to concerns about job security and compensation,

greater forecast optimism during high-sentiment periods can be attributed to either unintentional bias or manage-

rial opportunism. However, during low-sentiment periods, the managerial opportunism view predicts that managers

will provide more optimistic earnings forecasts to correct pessimistic investor sentiment and boost stock prices while

the unintentional bias view predicts that managers can be susceptible to pessimistic sentiment and provide more pes-

simistic earnings forecasts. An investigation of forecast bias in managers’ annual earnings forecasts during high- and

low-sentiment periods separately will shed light on this research question.

Furthermore, prior research finds that the influence of investor sentiment on stock valuation and analyst earn-

ings forecast bias is larger for firms with higher uncertainty (Baker & Wurgler, 2006, 2007; Hribar & McInnis, 2012;

Lemmon & Portniaguina, 2006; Mian & Sankaraguruswamy,2012; Qiu & Welch, 2006; Seybert & Yang, 2012). Since it

is generally more difficult to forecast the earnings of firms with higher uncertainty, sentiment-related management

earnings forecast bias is likely to be more pronounced for firms with higher uncertainty. Following prior research,

I use firm size, age, stock return volatility, market-to-book ratio and dividend payment to proxy for uncertainty

and investigate the role of uncertainty in the relationship between investor sentiment and management earnings

forecast bias.

Using the survey-based sentiment index constructed by the Michigan Consumer Research Center and a sam-

ple of 6,555 management forecasts of annual earnings per share (EPS) issued during 1998–2010, I find a positive

relationship between management earnings forecast bias and investor sentiment, indicating that management earn-

ings forecast optimism increases with sentiment. When replacing the continuous sentiment measure with indicator

variables for high- and low-sentiment periods, I find that managers’ annual earnings forecasts are more pessimistic

during low-sentiment periods than during normal-sentiment periods. This result suggests that sentiment-related