Do management earnings forecasts fully reflect information in past earnings changes?

• DOI: https://doi.org/10.1108/IJAIM-11-2017-0144
• Published date: 05 August 2019
• Pages: 373-406
• Date: 05 August 2019
• Author: Guojin Gong,Yue Li,Ling Zhou
• Subject Matter: Accounting & Finance,Accounting/accountancy,Accounting methods/systems

Do management earnings

forecasts fully reflect information

in past earnings changes?

Guojin Gong

Pennsylvania State University, University Park, Pennsylvania, USA

Yue Li

University of Illinois, Urbana-Champaign, Illinois, USA, and

Ling Zhou

DepartmentofAccounting,UniversityofNewMexico,Albuquerque,NewMexico,USA

Abstract

Purpose –It has been widely documented that investors and analysts underreact to information in

past earnings changes, a fundamental performance indicator. The purpose of this paper is to

examine whether managers’voluntary disclosure efficiently incorporates information in past

earnings changes, whether analysts recognize and fully anticipate the potential inefficiency in

management forecasts and whether managers’potential forecasting inefficiency entirely results

from intentional disclosure strategies or at least partly reflects managers’unintentional information

processing biases.

Design/methodology/approach –Archival data were used to empirically test the relation between

managementearnings forecast errors and past earnings changes.

Findings –Results show that managers underreact to past earnings changes when projecting future

earnings and analystsrecognize, but fail to fully anticipate, the predictablebias associated with past earnings

changes in management forecasts. Moreover, analysts appear to underreact more to past earnings changes

when management forecasts exhibit greater underestimation of earnings change persistence. Further

analyses suggest that the underestimation of earnings change persistence is at least partly attributable to

managers’unintentionalinformation processing bias.

Originality/value –This study contribute s to the voluntary disclosure litera ture by demonstrating the

limitation in the informational value of management forecasts. The findingsindicate that the effectiveness

of voluntary disclosure in mitigating market mispricing is inherently limited by the inefficiency in

management forecasts. This study can help market participants to better use management forecasts to

form more accurate earnings expectations. Moreover, our evidence suggests a managerial information

processing bias with respect to past earnings changes, which may affect managers’operational,

investment or financing decisions.

Keywords Earnings persistence, Underreaction, Management earnings forecasts,

Past earnings changes

Paper type Research paper

JEL classification –G14, M41

The authors thank Michael Jung and workshop participants at the 2010 American Accounting

Association annual meeting, the 2011 Financial Accounting and Reporting Section meeting, the 2013

Desert Finance Festival and University of Illinois at Chicago for helpful comments and suggestions.

Data Availability: Data are available from public sources indicated in the text.

Information in

past earnings

changes

373

Received29 November 2017

Revised27 February 2018

Accepted30 March 2018

InternationalJournal of

Accounting& Information

Management

Vol.27 No. 3, 2019

pp. 373-406

© Emerald Publishing Limited

1834-7649

DOI 10.1108/IJAIM-11-2017-0144

The current issue and full text archive of this journal is available on Emerald Insight at:

www.emeraldinsight.com/1834-7649.htm

1. Introduction

It has been widely documented that investors and analysts underreact to information in past

earnings changes, a fundamental performance indicator (Bernard and Thomas, 1989;

Abarbanell and Bernard, 1992). Kimbrough (2005) shows that the initiation of conference calls

reduces investors’underreaction to past earnings changes, indicating that voluntary disclosure

from managers could help investors better understand information in past earnings changes.

However, it does not answer the question whether voluntary disclosure itself fully incorporates

information in past earnings changes. In this paper, this question is investigated by testing the

relation between management earnings forecast errors and past earnings changes.

Understanding the efficiency of voluntary disclosure in incorporating public information can

shed light on the role of voluntary disclosure in influencing market inefficiency. Given

voluntary disclosure significantly influences market expectations (Patell, 1976;Penman, 1980;

Baginski and Hassell, 1990), examining how efficiently voluntary disclosure incorporates

information can assist market participants to better use voluntary disclosure in forming

expectations and making investment decisions. Specifically, the following research questions

are examined. Do management earnings forecasts fully reflect the implications of past earnings

changes for future earnings? If not, do analysts recognize and fully anticipate this inefficiency

in management forecasts? This study also examines whether managers’forecasting

inefficiency entirely results from intentional disclosure strategies or at least partly reflects

managers’unintentional information processing biases.

Although past earnings changes is a fundamental performance indicator that have

important implications for investors and analysts(Brown and Rozeff, 1979), to the authors’

knowledge, there is no prior study that examines how efficiently management forecasts

incorporate past earningschanges. It is ex ante unclear whether management forecasts fully

reflect information in past earnings changes. On the one hand, managers, as corporate

decision makers and financial information suppliers, presumably possess superior private

information about their firms’future performance and the sustainability of past earnings

trend. Management earnings forecasts thus could be free from predicable errors associated

with past earnings changes. Onthe other hand, managers may suffer from cognitive biases

in processing publicly available information similar to investors and analysts (Baker et al.,

2007). Moreover, managersmay strategically bias their forecasts in relation to past earnings

changes for the purpose of reaping private benefits. Cognitive biases and/or disclosure

strategies could induce predicable errors associated with past earnings changes in

management earningsforecasts.

Using the First Call’s Company Issued Guidelines (“CIG”) database, a significantly

negative association was determined betweenmanagement forecast errors for quarter tþ1

earnings (defined as forecasted earnings minus actual earnings scaled by price) and

seasonal earnings changes for quarter t. This finding suggests that management forecasts

are more optimistic (pessimistic)when their firms experience a larger decline (improvement)

in quarterly earnings,which is consistent with managers underestimating the persistenceof

past earning changes when projectingfuture earnings[1]. This negative association remains

significant after controlling for the self-selection nature of management forecasts as well as

a comprehensive set of managerial incentivesand firm characteristics. The underestimation

of the persistence of past earnings changes is the single most important factor driving

management forecasterrors in this study’s model.

Next, the analysts’understanding of the predictablebias (associated with past earnings

changes) in management forecasts is examined. Analysts recognized this predictable bias

when reacting to management forecasts; however, analysts fail to fully incorporate

this predictable biasin their earnings forecasts. Specifically, analysts seem to underestimate

IJAIM

27,3

374

this predictable bias, as evidenced by a significantly positive association between the

predictable bias in management forecasts associated with past earnings changes and

analysts’forecast errors (defined as analysts’earnings forecasts minus actual earnings

scaled by price) issued immediately afterwards. Moreover, analysts’underreaction to past

earnings changes is more pronounced when management forecasts exhibit greater

underreaction to past earnings changes. Taken together, the evidence supports the notion

that analysts’forecasting inefficiency is affected by managers’underestimation of the

persistence of pastearning changes[2].

Finally, the potential explanations for the negative association between management

forecast errors and past earnings changes are investigated. Two previously documented

disclosure strategies –bundlingand conservatism –are examined. Managers may “bundle”

optimistic forecasts with negativeearnings news to counterbalance bad news, resulting in a

negative relation between management forecast errors and negative earnings changes.

Managers may also “conservatively”estimate (i.e. underestimate) the persistence of positive

earnings changes to guide down market expectations to avoid negative earnings surprises.

Note that neither strategy plays a major role in the negative association between

management forecasterrors and earnings changes.

Second, the authors examined whether unintentional information processing bias

contributes to our finding of a negative relation between management forecast errors and

past earnings changes. As explained further in Section V, it is inferred that managers’true

beliefs about future firm performanceis based on their trading activities. Furthermore,there

is consistent evidence that managers’underestimation of the persistence of past earnings

changes is attributable to managers’information processing bias. Moreover, the negative

association between management forecast errors and past earnings changes is more

pronounced for range forecasts than for point forecasts, consistent with the behavioral

prediction that cognitive biasesare more likely to arise with greater forecasting uncertainty

(Hirshleifer,2001).

Note that the results obtained cannot completely rule out the possibility that certain

disclosure strategies (other than bundling with negative earnings news and conservative

towards positive earnings news)contribute to the underreaction of management forecasts to

past earnings changes. Nevertheless, the evidence suggests that unintentional information

processing bias, at least partly, gives rise to the negative relation between management

forecast errors and past earnings changes. Note that the study does not intend to test

specific cognitive biases that underlie managers’inefficient forecasting behavior, which is

beyond the scope of our study.

This study contributes to the voluntary disclosure literature by demonstrating the

limitation in the informational value of management forecasts. Prior research has shown

that management’s voluntary disclosure helps reduce market mispricing of earnings

information such as post-earnings-announcement drift (Kimbrough, 2005). Note that

management forecasts containa predicable bias in relation to past earningschanges. Given

that past earnings changes isan important performance indicator and has been extensively

studied, it is surprising that it remains a major driver of management forecast errors.

Moreover, analysts fail to fully anticipate this inefficiency imbedded in management

forecasts and analysts’underreaction to past earnings changes appears positively related

with management’s underreaction. The results indicate that the effectiveness of voluntary

disclosure (in particular, management forecasts) in mitigating market mispricing is

inherently limitedby the inefficiency in management forecasts.

Furthermore, studies have shown that management earnings forecasts significantly influence

market earnings expectations (Patell, 1976;Penman, 1980;Baginski and Hassell, 1990). The

Information in

past earnings

changes

375