University of Southern California Dissertations and Theses

Does differential sensitivity to aggregate earnings shocks drive post-earnings-announcement drift?

(USC Thesis Other)

Does differential sensitivity to aggregate earnings shocks drive post-earnings-announcement drift?

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

DOES DIFFERENTIAL SENSITIVITY TO AGGREGATE EARNINGS
SHOCKS DRIVE
POST-EARNINGS-ANNOUNCEMENT DRIFT?

By

SURESH NALLAREDDY

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BUSINESS ADMINISTRATION)

December 2012

Copyright 2012 Suresh Nallareddy
ii

Dedication

This dissertation is dedicated to my parents Chandra Sekhar Reddy and Devasena Reddy,
and to my wife Sahitya Nelavai.
iii

Acknowledgements

I express my deepest gratitude to my dissertation chair, K.R. Subramanyam, for his
continuous insight, support and guidance in the development of this paper. I also wish to
thank other members of my dissertation committee Wayne Ferson, Maria Ogneva, and
Robert Dekle for their comments and suggestions. This paper also benefited from helpful
comments and suggestions from Dan Amiram, Karthik Balakrishnan, Patricia Dechow,
Richard Frankel, David Huelsbeck, Zhaoyang Gu, Mingyi Hung, Sudarshan Jayaraman,
Alon Kalay, Mozaffar Khan, Urooj Khan, Yaniv Konchitchki, Yuri Loktionov, Chad
Larson, Alastair Lawrence, Xiumin Martin, Jeff McMullin, Ganapathi Narayanamoorthy,
Doron Nissim, Panos Patatoukas, Mark Peecher, Stephen Penman, Scott Richardson,
Joshua Ronen, Stephen Ryan, Gil Sadka, Tatiana Sandino, Bryce Schonberger,
Lakshman Shivakumar, Pervin Shroff, Richard Sloan, Theodore Sougiannis, James
Stekelberg, Karen Ton, Dushyantkumar Vyas, Kara Wells, Biqin Xie, Alicia Yancy, Paul
Zarowin, and Jieying Zhang. I am also grateful to participants at Columbia University,
Florida International University, London Business School, New York University,
University of California, Berkeley, University of Illinois at Urbana-Champaign,
University of Minnesota, University of Southern California, and University of
Washington in St. Louis. All errors are my own.

iv

Table of Contents

Dedication ........................................................................................................................... ii
Acknowledgements ............................................................................................................ iii
List of Tables ..................................................................................................................... vi
List of Figures ................................................................................................................... vii
Abstract ............................................................................................................................ viii
Chapter 1: Introduction ....................................................................................................... 1
Chapter 2: Return Decomposition and Testable Implications ............................................ 8
Chapter 3: Data and Preliminary Analysis ........................................................................ 11
3.1 Sample and Data................................................................................................. 11
3.2 Replication of Returns to PEAD Strategy .......................................................... 11
3.3 Risk Exposures of PEAD Portfolios .................................................................. 14
Chapter 4: Do Aggregate Earnings Shocks Explain PEAD Returns? .............................. 17
4.1 Estimation of Aggregate Earnings Shocks ......................................................... 17
4.2 Earnings Shock Betas for PEAD Portfolios ....................................................... 20
4.3 PEAD Returns in Periods with Extreme Aggregate Earnings Shocks ............... 21
4.4 PEAD Returns and Systematic Earnings Shocks ............................................... 27
4. 5 What Drives Aggregate Earnings Shocks? The Role of
Macroeconomic Shocks .................................................................................... 32
4.6 Do Aggregate Non-Earnings Shocks Drive PEAD Returns? ............................. 40
Chapter 5: Alternative Explanations and Additional Analysis ......................................... 42
5.1 Do Aggregate Earnings Shocks Explain PEAD Returns Beyond the
Investor Naivety Hypothesis? ............................................................................ 42
5.2 Do Aggregate Earnings Shocks Explain PEAD Returns Beyond
Analyst Underreaction to Prior Earnings Information? ..................................... 45
5.3 Do Aggregate Earnings Shocks Explain PEAD Returns Beyond the Inflation
Illusion Hypothesis? ........................................................................................... 46
5.4 Do Results Hold for Firms Without Analyst Following? .................................. 48
Chapter 6: Conclusions ..................................................................................................... 49
v

References ......................................................................................................................... 50
Appendix ........................................................................................................................... 58

vi

List of Tables

Table 1: Post-Earnings-Announcement Drift Portfolio Characteristics: Realized Returns,
Risk Exposures, Expected Returns, and Earnings Shock Betas ......................... 14
Table 2: PEAD Strategy Returns for Low and High Aggregate Earnings
Shock Quarters .................................................................................................... 24
Table 3: Systematic Earnings Shocks and PEAD Strategy Returns ................................. 29
Table 4: Aggregate Earnings Shocks and Macroeconomic Variables .............................. 33
Table 5: Investor Naivety and Analyst Underreaction to Prior Earnings Information ..... 43
Table 6: The Inflation Illusion Hypothesis ....................................................................... 46

vii

List of Figures

Figure 1: Time Line for Variable Measurement ............................................................... 12
Figure 2: Time Series Graph of PEAD Strategy Returns & Aggregate
Earnings Shocks ................................................................................................ 22

viii

Abstract

This paper finds that returns to the post-earnings-announcement drift (PEAD) strategy
result from differential sensitivity of individual stock returns to aggregate earnings
shocks. Larger negative aggregate earnings shocks are associated with higher PEAD
returns, because stocks in the PEAD’s sell portfolio are more sensitive to aggregate
earnings shocks than those in the buy portfolio. Such differential sensitivity to aggregate
earnings shocks drives a significant portion of PEAD returns. During the 1985 to 2009
sample period, investors were on average negatively surprised by aggregate earnings
shocks, leading to average positive returns to the PEAD strategy. Further analysis
suggests that macroeconomic shocks (that work through aggregate earnings shocks)
explain the variation in PEAD returns.
1

Chapter 1: Introduction

Post-earnings-announcement drift (PEAD)—the phenomenon whereby stocks
with positive quarterly earnings surprises outperform stocks with negative quarterly
earnings surprises over subsequent months—is arguably one of the most puzzling and
enduring patterns in the cross-section of equity returns (Fama, 1998). One explanation for
PEAD is that investors underreact to earnings news because they underestimate the
implications of current earnings for future earnings (Bernard and Thomas, 1989, 1990).
However, if the returns to the PEAD strategy are indeed driven by investor naiveté, then
they should have been arbitraged away over time, especially given the profitability and
pervasiveness of the PEAD strategy.
1
Yet PEAD has persisted for over 40 years, which
raises questions about whether PEAD does in fact represent market mispricing or whether
it is driven by other factors.
In this paper, I show that PEAD arises due to differential sensitivity of individual
stock returns to aggregate earnings shocks. Specifically, I find that the returns of stocks in
the PEAD’s sell portfolio (i.e., stocks in the lowest earnings surprise decile) are
significantly more sensitive to aggregate earnings shocks than those of stocks in the buy
portfolio (i.e., stocks in the highest earnings surprise decile). Because aggregate earnings
shocks have been mostly negative over the past 25 years, the sell portfolio generates
more negative returns on average than the buy portfolio, giving rise to the PEAD return
pattern. Consistent with this conjecture, I find that most of the returns to the PEAD

1
There is mixed evidence on whether transaction costs (more broadly, limits of arbitrage) constrain
arbitrageurs from exploiting PEAD returns. Richardson, Tuna, and Wysocki (2010) review this literature
and conclude that PEAD strategy returns are lower after accounting for transaction costs but are still
significant.
2

strategy arise from the component of returns that is correlated with aggregate earnings
shocks and that there are no significant differences across the PEAD portfolios in the
component of returns uncorrelated with aggregate earnings shocks. Further, the
magnitude of PEAD strategy returns is significantly larger for those quarters with the
most negative aggregate earnings shocks relative to those quarters with the least negative
shocks. In fact, there are no significant PEAD returns in those quarters that fall in the
tercile of the least negative aggregate earnings shocks.
I begin my analysis by measuring aggregate earnings shocks. I define aggregate
earnings shocks as the component of quarterly market returns that represents changes in
aggregate earnings expectations. I measure aggregate earnings shocks as the equally
weighted average of individual stocks’ earnings shocks, which in turn are measured as
returns driven by current-period revisions to expectations of future earnings (Easton and
Monahan, 2005).
2
I focus on the earnings shock component of market returns—rather
than total market returns—because past research shows that PEAD is likely driven by the
components of returns that are related to earnings expectation revisions,
3
rather than by

2
The earnings shock component is conceptually similar to the cash flow shock component in the Campbell
and Shiller (1988) and Campbell (1991) return decomposition framework. Empirically I measure earnings
shocks from analyst forecast revisions. Since analysts do not update their earnings forecasts in a timely
fashion, earnings shock estimates derived from analyst forecasts provide a lower bound estimate of true
earnings shocks (Chan and Zhao, 2010). Consistent with this argument, regressing market returns on
aggregate earnings shocks and different proxies of expected returns yields a coefficient estimate in the
range of 1.67-1.71, which is greater than the theoretical coefficient of 1 (Campbell and Shiller, 1988). As a
robustness check, I re-estimate the results using the fitted value of returns on earnings shocks. Overall, the
results using this proxy are similar to those using the proxy derived from analyst forecast revisions.

3
See, for example, Bernard and Thomas (1989, 1990), Abarbanell and Bernard (1992), Ranjan and Sloan
(1998), Soffer and Lys (1999), Brown and Han (2000), and Narayanamoorthy (2005).

3

expected returns.
4
Using various methods, in untabulated analysis I find that PEAD arises
primarily from earnings shocks and not from other return components (i.e., those related
to expected returns or discount rate shocks). Over the 1985 to 2009 sample period, I find
that aggregate earnings shocks are mostly negative, although there is a fair amount of
variation over time.
I next determine the sensitivity of the PEAD portfolio’s earnings shocks to
aggregate earnings shocks by estimating earnings shock betas. To do so, I regress the
quarterly earnings shocks of each PEAD portfolio on the aggregate earnings shocks over
the 12 quarters prior to the formation of the PEAD strategy. The earnings shock beta
captures the sensitivity of earnings expectation revisions in the respective PEAD
portfolios to aggregate earnings expectation revisions.
5
I find that the earnings shock beta
of the sell portfolio (2.49) is 2.8 times as large as that of the buy portfolio (0.89).
To examine whether the PEAD strategy returns can be attributed to the
differential sensitivity of PEAD portfolio returns to aggregate earnings shocks, I conduct
two separate tests. First, I decompose the PEAD strategy returns into systematic earnings
shocks, which arise from the sensitivity of the PEAD portfolios to aggregate earnings
shocks, and the residual return, which is uncorrelated with aggregate earnings shocks. I
measure the systematic earnings shocks for each PEAD portfolio as the portfolio’s
respective earnings-shock beta multiplied by the aggregate earnings shocks. I find that

4
See, for example, Bernard and Thomas (1989, 1990), Bernard, Thomas, and Wahlen (1997), Chordia and
Shivakumar (2006), and Wu and Zhang (2011).

5
Conceptually, earnings shock betas are similar to bad beta (Campbell and Vuolteenaho, 2004; Khan,
2008; Da and Warachka, 2009). The key difference is that bad beta is the covariance between portfolio
returns and aggregate earnings news.
4

systematic earnings shocks account for a significant portion of PEAD strategy returns
(1.88% out of total PEAD returns of 2.20% per quarter). In contrast, I do not find any
pattern in the residual returns for the PEAD portfolios. These results strongly suggest that
PEAD arises primarily because of the differential sensitivity of the sell versus the buy
PEAD portfolios to aggregate earnings shocks.
Recall that I argue that PEAD returns arise because the aggregate earnings shocks
are negative on average over my sample period: the sell portfolio, which has a higher
covariance with aggregate earnings shocks than the buy portfolio, generates more
negative returns on average, and this differential sensitivity gives rise to the PEAD return
pattern. To confirm this conjecture, however, I need to show that the pattern in PEAD
returns over time is predictably related to time-series variation in the direction/magnitude
of aggregate earnings shocks. Accordingly, in my second test on whether the PEAD
strategy returns can be attributed to the differential sensitivity of PEAD portfolio returns
to aggregate earnings shocks, I examine returns to the PEAD strategy after partitioning
the sample period on the direction/magnitude of aggregate earnings shocks. I find that the
decile (quintile, quartile, tercile) of the sample quarters with the most negative aggregate
earnings shocks has statistically significant PEAD returns of 4.37% (3.92%, 3.61%,
2.85%) per quarter. In contrast, the decile (quintile, quartile, tercile) of quarters with the
least negative aggregate earnings shocks has statistically insignificant returns of -0.33%
(0.15%, 0.04%, 0.31%) per quarter. Differences in PEAD returns between these
respective partitions are statistically significant at the 5% level. These results suggest that
variation in aggregate earnings shocks is associated with variation in PEAD returns;
5

specifically, more negative aggregate earnings shocks are associated with higher PEAD
returns.
As I use revisions in analysts’ earnings forecasts to identify earnings shocks, one
concern that arises is whether my measure of aggregate earnings shocks merely captures
a form of aggregate analyst optimism rather than economic shocks that drive aggregate
earnings expectation revisions. The fact that the aggregate earnings shocks are mostly
negative accentuates this concern. To investigate whether my measure of aggregate
earnings shocks represents systematic shocks to the economy, I examine the association
between aggregate earnings shocks and various macroeconomic variables. I find that
aggregate earnings shocks are significantly associated with current-period revisions to
expectations about future periods’ consumption growth, industrial production, and
inflation. Further, macroeconomic variables explain 73% of the variation in aggregate
earnings shocks.
6
These results suggest that my measure of aggregate earnings shocks is
indeed picking up economic shocks that affect earnings expectations.
I also investigate whether the component of aggregate earnings shocks driven by
macroeconomic shocks can better explain PEAD strategy returns than the component
unrelated to macroeconomic shocks. To this end, I decompose aggregate earnings shocks
into the fitted aggregate earnings shock that is driven by macroeconomic shocks,
estimated as the fitted value from regressing aggregate earnings shocks on
macroeconomic shocks, and the residual aggregate earnings shock that is unrelated to

6
In contrast, these variables explain only 18% of total market returns, which is only significantly correlated
with the risk-free rate.

6

macroeconomic shocks. I find that fitted aggregate earnings shocks explain variation in
PEAD returns better than the residual aggregate earnings shocks. Specifically, I find that
quarters with more negative fitted aggregate earnings shocks have significantly higher
PEAD returns than quarters with less negative fitted values. In contrast, quarters with
more negative residual aggregate earnings shocks do not have statistically higher PEAD
returns than quarters with less negative residual values. This evidence suggests that
macroeconomic shocks (that work through aggregate earnings shocks) explain the
variation in PEAD returns.
Next, I examine whether the relationship that I document between aggregate
earnings shocks and PEAD returns is driven by existing explanations for PEAD returns. I
find that aggregate earnings shocks explain variation in PEAD returns beyond existing
explanations such as the investor naivety hypothesis in Bernard and Thomas (1989,
1990), analyst underreaction to prior earnings information in Abarbanell and Bernard
(1992), or the inflation illusion explanation in Chordia and Shivakumar (2005).
Overall, I find that differential sensitivity of the sell versus the buy portfolios to
aggregate earnings shocks explains a significant portion of the variation in PEAD
strategy returns. It is important to note that the framework that I provide is not a risk-
based explanation. Even if aggregate earnings shocks are a priced risk factor (Campbell
and Vuolteenaho, 2004; Ball, Sadka, and Sadka, 2009; Da and Warachka, 2009), I find
that earnings shock betas for the sell portfolio are larger than those for the buy portfolio,
suggesting that the PEAD strategy has negative risk exposure to aggregate earnings
shocks.
7

This paper contributes to the literature in several ways. First, I present a
framework that sheds light on when PEAD strategy returns occur and document that
higher negative aggregate earnings shocks are associated with higher PEAD returns. I
show that a significant portion of the returns to the PEAD strategy are driven by
differential sensitivity of the sell versus the buy portfolios to aggregate earnings shocks.
Second, I present evidence that suggests economic shocks (which work through
aggregate earnings shocks) better explain variation in PEAD returns than alternative
explanations based on investor or analyst underreaction to prior earnings information.
Third, I contribute to the literature that links macroeconomic variables to earnings by
documenting that 73% of the variation in aggregate earnings revisions is attributable to
changes in macroeconomic conditions. Finally, the framework developed in this paper
can be used to understand other asset pricing anomalies for which sorting on the anomaly
variable is essentially sorting on the aggregate earnings shock sensitivity.

8

) 1 ( ) (
1 1 1 1 + + + +
− + =
t t t t t
DN EN r E r
Chapter 2: Return Decomposition and Testable Implications

In this section, I rely on the return decomposition framework to understand the
drivers of realized returns to the PEAD strategy. Realized returns can be decomposed into
expected returns, earnings shocks,
7
and discount rate shocks (Campbell and Shiller, 1988;
Campbell, 1991; Vuolteenaho, 2002).

where r
t+1
stands for realized returns for quarter t+1, ) (r E for expected returns, EN for
earnings shocks (earnings news) and DN for discount rate shocks (discount rate news).
Earnings shocks are defined as returns attributable to current-period revisions to
expectations about future earnings. Discount rate shocks are defined as returns driven by
current-period revisions to expected returns. Realized returns to the PEAD strategy (buy
portfolio (B) - sell portfolio (S)) can be similarly decomposed as

I examine each of these three components (differences in expected returns, earnings
shocks, and discount rate shocks) to understand the drivers of PEAD realized returns.
Evidence from the existing literature suggests that expected returns (the first term
in equation (2)) do not drive realized returns to the PEAD strategy (Bernard and Thomas,
1989, 1990; Bernard, Thomas, and Wahlen, 1997; Chordia and Shivakumar, 2006; Wu
and Zhang, 2011).

7
Financial economists often refer to earnings shocks as cash flow shocks.
) 2 ( ) ( ) ( )) ( ) ( (
1 1 1 1 1 1 1 1
S
t
B
t
S
t
B
t t
S
t t
B
t
S
t
B
t
DN DN EN EN r E r E r r
+ + + + + + + +
− − − + − = −
9

Turning next to earnings shocks (the second term in equation (2)), these can be
further decomposed into a systematic component that is related to aggregate earnings
shocks and an idiosyncratic component as follows:

S
t
B
t
M
t
S
t EN
B
t EN
S
t
M
t
S
t EN
B
t
M
t
B
t EN
S
t
B
t
EN
EN EN EN EN
1 1 1 , ,
1 1 , 1 1 , 1 1
) (
) ( ) ( ) (
+ + +
+ + + + + +
− + − =
+ − + = −
ε ε β β
ε β ε β
(3)
where
M
t
EN
1 +

denotes aggregate earnings shocks,
B
t EN ,
β captures the sensitivity of the buy
portfolio’s earnings shocks to aggregate earnings shocks (earnings shock beta for the buy
portfolio),
S
t EN ,
β captures the sensitivity of the sell portfolio’s earnings shocks to
aggregate earnings shocks (earnings shock beta for the sell portfolio), and
t EN
M
t
EN
, 1
β
+
captures the portion of earnings shocks for a portfolio that co-moves with
aggregate earnings shocks.
If aggregate earnings shocks drive most of the variation in realized returns to the
PEAD strategy, then mean-zero aggregate earnings shocks in a given time period should
translate into mean-zero returns to the PEAD strategy. Indeed, on average aggregate
earnings shocks should be mean-zero in expectation or over a long sample period.
However, evidence suggests that in finite samples – even 100 years of data is considered
a finite sample – aggregate earnings shocks can be non-mean zero (Elton, 1999; Fama
and French, 2002; Lundblad, 2007).
8
Hence, in a finite sample, we can observe non-zero
returns to the PEAD strategy.

8
Earnings shocks and/or discount rate shocks can be non-zero for long time series of data (see Elton, 1999
for a review). Lundblad (2007) shows that even 100 years of time-series data have nonzero shocks. As a
recent example, over the last 40 years ending March 2009, long-term government bonds outperformed the
S&P 500 index by 0.12% per year. Fama & French argue that this is an outcome of non-mean zero earnings
10

Testable implications from equation (3) are as follows: (1) if the earnings shock
beta of the buy portfolio is smaller than that of the sell portfolio (i.e., 0
, ,
< −
S
t EN
B
t EN
β β ),
then periods with higher negative aggregate earnings shocks translate into higher PEAD
returns; and (2) if the earnings shock beta of the buy portfolio is greater than that of the
sell portfolio (i.e., 0
, ,
> −
S
t EN
B
t EN
β β ), then periods with higher positive aggregate earnings
shocks translate into higher PEAD returns.
Similar arguments can be made for discount rate shocks (the third term in
equation (2)), but the testable implications will have signs that run in the opposite
direction given that discount rate shocks are negatively related to realized returns.

and/or discount rate shocks (http://www.dimensional.com/famafrench/2009/08/qa-bonds-for-the-long-
run.html#more).
11

Chapter 3: Data and Preliminary Analysis

3.1 Sample and Data
The sample comes from the intersection of I/B/E/S, CRSP, and Compustat for the
1985 to 2009 period. All variables except the risk (common) factors are measured at the
quarterly frequency. Portfolios are formed on a calendar date basis, and hence I restrict
the sample to firms with a December fiscal year-end to align the data across firms. I also
restrict the sample to ordinary common shares (share codes 10, 11) that are traded on the
NYSE, AMEX, or NASDAQ exchanges. I drop firms with a stock price less than $1 as
well as firm-quarters with negative book values of equity. The final sample comprises
162,278 firm-quarter observations. Monthly values of the three Fama-French factors–
market (MKT), size (SMB), and book-to-market (HML)–come from Professor Kenneth
French’s website.
9
I obtain macroeconomic data from the Federal Reserve Bank of St.
Louis.

3.2 Replication of Returns to PEAD Strategy
The PEAD strategy is executed by taking a long position in the decile of firms
with the most positive earnings surprise (buy portfolio) and a short position in the decile
of firms with the most negative earnings surprise (sell portfolio). Specifically, at the
beginning of each quarter, 10 portfolios are formed on standardized unexpected earnings

9
http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html

12

(SUE) and are held for three subsequent months. Figure 1 presents the timeline of
variable measurement.
Figure 1: Time Line for Variable Measurement

Standardized unexpected earnings (SUE) are estimated as reported earnings
minus expected earnings, deflated by stock price. Specifically, SUE in quarter t is
measured as:

it
it t it
t i
P
X E X
SUE
)) ( (
1
,
−
−
= (4)
where X
it
is realized earnings for firm i in quarter t, E
t-1
(X
it
) is the consensus analyst
forecast (median analyst forecast 90 days prior to the earnings announcement date) for
firm i in quarter t
10
, and P
it
is price per share for firm i at the end of quarter t. The
advantage of deflating by price rather than the standard deviation of forecast errors is that
deflating by price does not require a long time series of data, mitigating problems

10
Data come from the I/B/E/S unadjusted files rather than the adjusted files to avoid potential rounding
problems identified by Payne and Thomas (2003). Actual and consensus data are from I/B/E/S unadjusted
files and adjustments are made using the CRSP adjustment factor.

t

End of
Quarter
t-1 t+1 t+2
E t-1(X t)

R t+1
E-Shocks t+1
AGG_E_Shock t+1
+1

P t

Portfolio
formation
date

Portfolio
ending
date

13

associated with survivorship bias (Livnat and Mendenhall, 2006). This measure of SUE is
similar to those used in Abarbanell and Bernard (1992), Mendenhall (2004), Francis et al.
(2004), and Livnat and Mendenhall (2006).
11,12

The starting point of my analysis is to compute realized returns to the PEAD
strategy. For each SUE portfolio, I estimate equally weighted average realized returns for
the 3-month period starting from the portfolio formation date. Time-series means of
portfolio returns are reported in Table 1. Consistent with over 40 years of literature on
PEAD, I find statistically significant returns to the PEAD strategy.
13
Specifically, the sell
portfolio observes returns of -1.03%, while the buy portfolio realizes returns of 1.14%.
As a result, the hedge portfolio (buy-sell portfolio) has an average return of 2.16% over
the three-month holding period.

11
My primary analysis is restricted to firms with analyst following. As a robustness check, I also examine
returns to the PEAD strategy for firms without analyst following. The sample for this analysis is drawn
from CRSP and Compustat. For these firms, E
t-1
(X
it
) is replaced with earnings from a seasonal random walk
model.

12
With the exception of Livnat and Mendenhall (2006) and Mendenhall (2004), who also use I/B/E/S data,
the source of analyst earnings estimate data in these studies differ. In particular, Abarbanell and Bernard
(1992) extract data from Value Line, while Francis et al. (2004) obtain data from Zacks Investment
Research.

13
Since Ball and Brown (1968), who first document the PEAD phenomenon, many papers document such
drift for different samples and using methods; see, e.g., Jones and Litzenberger (1970), Latane, Joy, and
Jone (1970), Brown and Kennelly (1972,; Joy, Litzenberger, and McEnally (1977), Latane and Jones
(1979), Watts (1978), Rendleman, Jones, and Latane (1982), Foster, Olsen, and Shevlin (1984),
Rendleman, Jones, and Latane (1987), and Freeman and Tse (1989), among others.
14

Table 1: Post-Earnings-Announcement Drift Portfolio Characteristics: Realized
Returns, Risk Exposures, Expected Returns, and Earnings Shock Betas
Portfolio SUE Returns CAPM-Beta SMB-Beta HML-Beta ICC_GLS ICC_PEG E-Beta
Sell -3.84 -1.03 1.37 1.80 0.08 2.69 4.37 2.49
2 -0.40 0.26 1.06 1.22 0.11 2.39 3.43 1.01
3 -0.15 0.82 1.02 1.10 -0.07 2.25 3.05 0.78
4 -0.06 0.24 0.97 0.77 -0.09 2.22 2.91 0.56
5 0.00 0.79 1.10 0.88 -0.41 2.15 2.63 0.49
6 0.04 0.85 1.08 1.02 -0.33 2.07 2.62 0.56
7 0.09 0.80 1.05 1.05 -0.28 2.17 2.77 0.50
8 0.17 1.03 1.12 1.19 -0.17 2.25 2.96 0.63
9 0.34 1.18 1.17 1.24 -0.11 2.33 3.21 0.76
Buy 1.53 1.14 1.27 1.49 0.01 2.50 3.94 0.89
Buy_Sell 5.37 2.16 -0.10 -0.31 -0.08 -0.19 -0.43 -1.60
t-stat (11.47)*** (4.31)*** (-2.74)*** (-3.81)*** (-0.52) -(0.54) (-5.82)*** (8.46)***

Table 1 reports the realized returns, risk exposures, expected returns, and earnings shock betas of
portfolios sorted on standardized unexpected earnings (SUE). Sample data cover the period 1985
to 2009. The sample includes firms with December fiscal year-end and portfolios are formed at
the beginning of each quarter. Every quarter, 10 portfolios are formed on SUE and portfolios are
held for the subsequent three months. All variables are measured at the quarterly frequency.
Returns are calculated as the continuously compounded three-month return estimated after the
portfolio formation date. CAPM-Beta is the market beta. SMB-Beta and HML-Beta are the size
and book-to-market factor betas. ICC_GLS is the implied cost of capital measure derived from
the GLS model (Gebhart et al., 2001). ICC_PEG is the implied cost of capital measure derived
from the PEG model (Easton, 2004). E-Beta is the earnings shock beta, calculated as the
covariance between portfolio earnings shocks and aggregate earnings shocks, divided by the
variance of aggregate earnings shocks. Betas are estimated at the portfolio level using rolling 12-
quarter data and the time-series means of the rolling betas are reported. ***indicates significance
at the 1 percent level, ** at the 5 percent level, and * at the 10 percent level.

3.3 Risk Exposures of PEAD Portfolios
The return decomposition framework suggests that realized returns can be
decomposed into expected returns, earnings shocks, and discount rate shocks. In this
section, I examine whether differences in expected returns for the buy versus sell PEAD
portfolios can explain the realized returns to the PEAD strategy. Specifically, I estimate
15

implied cost of capital and risk exposures using standard asset pricing models to measure
expected returns for the PEAD portfolios.
Implied cost of capital estimates are used extensively in the accounting and
finance literatures as a proxy for expected returns. I employ two different methods to
estimate implied cost of capital. The first method follows Easton (2004) and is derived
from the Ohlson and Juettner-Nauroth (2005) abnormal earnings growth model. The
second method comes from Gebhardt, Lee, and Swaminathan (2001). Appendix describes
the two estimation procedures.
In addition to implied cost of capital, I estimate risk exposures for the PEAD
portfolios using two models: the Capital Asset Pricing Model (CAPM; Sharpe, 1964) and
the Fama and French three-factor model (Fama and French, 1994). The CAPM beta is
estimated using a time-series regression of quarterly portfolio returns on quarterly market
returns:

t mt
CAPM
p pt
r a r ε β + + =
,
(5)
where
CAPM
p,
β is the beta for portfolio p and
mt
r is the market return. Betas are estimated
using a rolling 12-quarter window. Exposures to the Fama and French three factors are
estimated using a time-series regression of quarterly portfolio returns on quarterly market
return, SMB, and HML factors:
R
p,t
= α
p
+ β
p
r
m,t
+ γ
p
SMB
t
+ δ
p
HML
t
+ ε
pt
(6)

where γ
p
captures the common factor exposure to the size factor (SMB) and δ
p
captures
the common factor exposure to the book-to-market factor (HML).
16

Table 1 reports the time-series means of the risk exposure and implied cost of
capital estimates. The results suggest that the sell portfolio is marginally riskier than the
buy portfolio. Specifically, the sell portfolio has a higher CAPM beta, SMB beta, and
implied cost of capital than the buy portfolio. This evidence is consistent with findings in
prior literature (Bernard and Thomas, 1989, 1990; Bernard, Thomas, and Wahlen, 1997;
Chordia and Shivakumar, 2006; Wu and Zhang, 2011).
The results further suggest that the relationship between SUE portfolios and risk
is U-shaped. The sell portfolio is the most risky, the middle portfolio is the least risky,
and the buy portfolio is slightly less risky than the sell portfolio. The middle portfolios
are the least risky as a result of portfolio formation. Firms that are not doing well
operationally and are volatile are in the sell portfolio, while firms that are doing well
operationally and are volatile are in the buy portfolio. This leaves the most stable, low
volatility firms in the middle portfolio. Hence, the middle portfolios are the least risky as
compared to the extreme portfolios.
Overall, the evidence from several risk estimation methods suggests that sell
portfolios are marginally riskier than buy portfolios. Expected returns are therefore
unable to explain realized returns to the PEAD strategy.

17

Chapter 4: Do Aggregate Earnings Shocks Explain PEAD Returns?

4.1 Estimation of Aggregate Earnings Shocks
Aggregate earnings shocks are defined as aggregate returns driven by current-
period revisions to expectations about future aggregate earnings (Campbell and Shiller,
1988; Vuolteenaho, 2002). The aggregate earnings shock for quarter t is estimated as the
equally weighted average of firm-level earnings shocks as follows:
∑
=
=
t
N
i
t i
t
t
Shock E
N
Shock E Agg
1
,
) _ (
1
_ _ (7)
where Agg_E_Shock
t
is the aggregate earnings shock in quarter t, E_Shock
i,t
is the
earnings shock for firm i in quarter t, and N
t
is the number of firms in quarter t.
14

I estimate firm-level earnings shock following the Campbell and Shiller (1988)
and Vuolteenaho (2002) return decomposition. In particular, firm-level earnings shock is
defined as returns driven by current-period revisions to expectations of future earnings:






Δ =
∑
∞
=
+
−
+ +
1
,
1
1 1 ,
_
n
n s i
n
t t i
roe E Shock E ρ (8)

where E_Shock
i,t+1
is the earnings shock for firm i in quarter t+1, [] .
1 +
Δ
t
E is the change in
expectations from quarter t to t+1, s is the current year,
n s i
roe
+ ,
is the natural log of the
gross accounting rate of return on equity for firm i in year s+n, and ρ monotonically
increases in dividend yield and is slightly less than one.

14
As a robustness check, I also estimate the aggregate earnings shocks as the value-weighted average of
firm-level earnings shocks. Value-weighting results in aggregate earnings shocks which are smoother and
on average lower in magnitude. Overall, the results using value-weighted aggregate earnings shocks are
similar to those using the equally-weighted aggregate earnings shocks.
18

I use analyst earnings forecast revisions to proxy for investor expectation
revisions. Relying on analyst earnings forecasts to measure investor expectations
introduces two limitations. First, while a theoretical definition of an earnings shock
requires infinite-horizon earnings expectations, analyst earnings forecasts are available
only for finite horizons. Further, evidence suggests that analyst forecasts for the longer
term are too noisy and have low accuracy (Chan, Karceski, and Lakonishok, 2003). To
overcome these limitations, I restrict attention to revisions to current-year earnings
forecasts and one-year-ahead earnings forecasts, and capitalize revisions to one-year-
ahead earnings forecasts as an approximation for the sum of future-period earnings
expectation revisions as described below (Easton and Monahan, 2005). Second, analysts
do not update their earnings forecasts in a timely fashion (Lys and Sohn, 1990;
Abarbanell, 1991; Ali, Klein, and Rosenfeld, 1992; Hughes, Liu, and Su, 2008;
Konchitchki, Lou, Sadka, and Sadka, 2011). As a result, earnings shock estimates derived
from analyst forecasts provide a lower bound estimate of true earnings shocks (Chan and
Zhao, 2010).
15

Following Easton and Monahan (2005), I operationalize firm-level earnings
shocks as follows:

15
Another approach to estimate an earnings shock is to use the Vector Autoregression (VAR) approach as
described in Campbell and Vuolteenaho (2004). Earnings shock estimates derived using this approach are
subject to many limitations such as selection of state variables, method of estimation, etc. (Chen and Zhao,
2009).
19

2 , 1 , , 1 ,
* 1
_
+ + +








−
+ + =
s i s i s i t i
rev rev rev ews N E
ω ρ
ρ
)
(9)
where








+ =
− +
+
+
1 ,
,
,
1 log
n s i
n s i
n s i
B
REV
rev ,
n s i
REV
+ ,
is the revision to the year s+n median
analyst forecast made during quarter t+1 for firm i, s is the current year, and
n s
B
+
is the
year s+ n book value of equity for firm i. In the above equation, the first term (
s i
rev
,
)
captures the part of earnings shocks driven by realized forecast errors for year s earnings
reported in year s+1.
16
The second term (
1 , + s i
rev ) captures the part of earnings shocks
driven by revisions to current-year earnings expectations during the quarter. The final
term (
2 ,
) * 1 (
+
−
s i
rev ω ρ ρ ) captures the portion of earnings shocks driven by capitalized
revisions to one-year-ahead earnings expectations during the quarter. Here,ω

represents
the persistence in revisions to one-year-ahead earnings expectations and is estimated by
running the following regression for each of the 48 Fama and French industries:

1 , 1 , + +
+ + =
t t i t i
rev rev ξ ω σ (10)
where rev
i,t
is the natural log of the gross accounting rate of return for firm i in quarter t,
ω is estimated for each industry, and all firms within an industry are assigned the same
persistence coefficient. ρ estimates are from Easton and Monahan (2005).
17
To reduce the
influence of outliers, I winsorize earnings shock estimates at the 1% and 99% levels.

16
Year s earnings announcement made in year s+1 occurs only during the first quarter of year s+1. For the
second, third, and fourth quarters this term is equal to zero.

17
As a robustness check, following Ogneva (2010), I also estimate earnings shocks by assuming that ρ is a
cross- sectional constant equal to 0.91. Earnings shocks are lower on average using this method but the
main findings are qualitatively similar.
20

Figure 2 presents a time-series graph of aggregate earnings shock estimates.
Aggregate earnings shocks are on average -1.21% per quarter for the 1985 to 2009
sample period. Consistent with the findings in Chava and Purnanandam (2010) and Chan,
Karceski, and Lakonishok (2003), aggregate earnings shocks are on average negative.

4.2 Earnings Shock Betas for PEAD Portfolios
Given that the difference in expected returns cannot explain realized returns to the
PEAD strategy, I next examine whether the second component of the return
decomposition framework (i.e., earnings shocks) can explain PEAD returns. Specifically,
I examine whether the earnings shocks that co-move with aggregate earnings shocks
explain the PEAD returns.
As described in the return decomposition framework, the relationship between
aggregate earnings shocks and PEAD returns is conditional on the differences in earnings
shock betas for the buy and sell portfolio firms. In this section, I estimate the earnings
shock betas for these portfolios. Earnings shock betas are defined as the covariance
between portfolio earnings shocks and aggregate earnings shocks. Portfolio earnings
shocks are estimated as the equally weighted average of firm-level earnings shocks. I
estimate earnings shock betas using a time-series regression of quarterly portfolio
earnings shocks on quarterly aggregate earnings shocks,

t t
p
t EN t p
Shock E AGG a Shock E ε β + + = _ _ _
, ,
(11)
21

where
p
t EN ,
β is the earnings shock beta for portfolio p and AGG_E_Shock
t
is the aggregate
earnings shock for quarter t. Earnings shock betas are estimated using rolling 12-quarter
data for each portfolio; the means of the rolling 12-quarter betas are reported in Table 1.
Earnings shock betas vary significantly across the PEAD portfolios. In particular, the
earnings shock beta for the sell portfolio is 2.8 times as large as that of the buy portfolio,
which implies that firms in the sell portfolio are 2.8 times as sensitive to aggregate
earnings shocks as firms in the buy portfolio. Overall, the evidence suggests that the sell
portfolio has a much higher earnings shock beta than the buy portfolio.

4.3 PEAD Returns in Periods with Extreme Aggregate Earnings Shocks
Because the sell portfolio is more sensitive to aggregate earnings shocks, higher
negative aggregate earnings shocks translate into higher negative returns for the sell
portfolio relative to the buy portfolio. This result in higher returns for the PEAD (buy-
sell) strategy, that is, higher negative aggregate earnings shocks are associated with
higher PEAD returns. In this section, I explore the extent to which this differential
earnings shock sensitivity drives realized returns to the PEAD strategy.
22

Figure 2: Time Series Graph of PEAD Strategy Returns & Aggregate Earnings
Shocks
-7
-5
-3
-1
1
3
5
7
-20
-15
-10
-5
0
5
10
15
20
%
%
PEAD
AGG_E_Shocks

Figure 2 shows the time series of PEAD strategy returns and aggregate earnings shocks. The
sample data cover the period 1985 to 2009. Every quarter, 10 portfolios are formed on SUE and
are held for the subsequent three months. Returns to the highest decile of SUE minus the lowest
decile of SUE are reported on the y-axis. Aggregate earnings shocks (AGG_E_Shocks) are
aggregate returns driven by earnings expectation revisions and are measured as the equally
weighted average of firm-level returns from earnings expectation revisions.

Figure 2 plots the time series of realized returns to the PEAD strategy and
aggregate earnings shocks. The correlation between aggregate earnings shocks and
returns to the PEAD strategy is -0.20. This finding is consistent with the conjecture from
the return decomposition analysis that if the buy portfolio has a smaller earnings shock
beta than the sell portfolio, then higher negative aggregate earnings shocks lead to higher
PEAD strategy returns.
To more formally test the conjecture that quarters with higher negative aggregate
earnings shocks have higher PEAD returns, I partition the sample quarters into low and
high aggregate earnings shock quarters. Quarters with aggregate earnings shocks in the
23

bottom (top) decile, quintile, quartile, tercile, and half are defined as low (high) aggregate
earnings shock quarters, i.e., Low AE (High AE), for the decile, quintile, quartile, tercile,
and median partitions, respectively. Quarters in the low decile (quintile, quartile, tercile,
median) partition have mean aggregate earnings shocks of -3.23% (-2.49%,-2.29%, -
2.05%, -1.75%), respectively. Table 2 reports realized returns for the PEAD strategy in
these partitions.
Realized returns for the partitions suggest that quarters with low aggregate
earnings shocks (i.e., quarters with more negative aggregate earnings shocks) have
positive and significant returns to the PEAD strategy. In contrast, quarters with high
aggregate earnings shocks (i.e., quarters with less negative aggregate earnings shocks)
have returns to the PEAD strategy that are not economically or statistically different from
zero, with the exception of the median partition. Differences in PEAD returns between
the high and low partitions are statistically different from each other. This evidence is
consistent with the conjecture that higher negative aggregate earnings shocks are
associated with higher PEAD strategy returns.

24
Table 2: PEAD Strategy Returns for Low and High Aggregate Earnings Shock Quarters

Full Low AE High AE Low AE High AE Low AE High AE Low AE High AE Low AE High AE
AGG_ES -1.21 -3.23 -0.16 -2.49 -0.37 -2.29 -0.44 -2.05 -0.53 -1.75 -0.67
Sell -1.03 -11.21 6.24 -11.40 4.54 -10.09 6.29 -6.73 5.40 -5.82 3.87
2 0.26 -5.45 4.10 -6.49 3.97 -5.87 4.91 -3.72 4.36 -2.76 3.35
3 0.82 -6.32 2.91 -4.38 2.99 -3.62 4.27 -2.02 3.78 -1.43 3.11
4 0.24 -7.13 2.96 -6.02 2.89 -4.55 3.49 -2.63 3.55 -2.26 2.73
5 0.79 -6.19 4.04 -4.83 3.86 -3.70 4.63 -1.69 3.76 -1.07 2.65
6 0.85 -7.02 3.04 -5.83 3.16 -4.53 4.33 -2.58 3.71 -1.41 3.15
7 0.80 -5.91 3.15 -4.81 3.26 -3.91 4.42 -2.34 3.86 -1.48 3.13
8 1.03 -6.00 4.12 -5.32 3.77 -4.35 4.65 -2.37 4.24 -1.32 3.82
9 1.18 -6.09 4.26 -5.60 3.67 -4.44 5.12 -2.81 4.42 -1.68 4.09
Buy 1.14 -6.84 5.91 -7.48 4.69 -6.48 6.33 -3.88 5.71 -2.73 5.10
Buy_Sell 2.16 4.37 -0.33 3.92 0.15 3.61 0.04 2.85 0.31 3.09 1.22
t-stat (4.31)*** (2.57)** (-0.41) (2.73)*** (0.23) (2.99)*** (0.05) (2.59)** (0.55) (3.69)*** (2.35)**
Diff
t-stat
Tercile Quintile Median
Partition Partition Partition
Decile
4.71
(2.50)**
3.77
(2.57)**
Partition
Quartile
Partition
(2.39)**
1.87
(1.89)*
PEAD Portfolio Returns
3.57 2.54
(2.06)**

Table 2 reports the time-series means of returns to SUE-sorted portfolios for the full sample as well as for the sample partitioned into low and high
aggregate earnings shock quarters. The sample includes firms with December fiscal year-end and portfolios are formed at the beginning of each
quarter. Every quarter, 10 portfolios are formed on SUE and portfolios are held for the subsequent three months. Aggregate earnings shocks are
aggregate returns driven by earnings expectation revisions and are measured as the equally weighted average of firm-level returns from earnings

25

Table 2 (Continued)

expectation revisions. Quarters with aggregate earnings shocks in the bottom (top) decile, quintile, quartile, tercile, and half are defined as low
(high) aggregate earnings shocks, i.e., Low AE (High AE), for the decile, quintile, quartile, tercile, and median partitions, respectively. AGG_ES is
the average aggregate earnings shock for the partition. Realized returns are measured at the quarterly frequency and the time-series means are
reported. Returns are calculated as the continuously compounded three-month return estimated after the portfolio formation date. Sample data for
this analysis cover the period 1985 to 2009. *** indicates significance at the 1 percent level, ** at the 5 percent level, and * at the 10 percent level.

26

More specifically, quarters in the low aggregate earnings shock decile (quintile,
quartile, tercile, median) partition have statistically significant PEAD strategy returns of
4.37% (3.92%, 3.61%, 2.85%, 3.09%), whereas quarters in the high aggregate earnings
shock decile (quintile, quartile, tercile) partition have statistically insignificant PEAD
strategy returns of -0.33% (0.15%, 0.04%, 0.31%). Return patterns for the buy and sell
portfolios in these quarters are striking. In the low aggregate earnings shock partitions,
both the buy and the sell portfolios have negative returns, but the sell portfolio has larger
negative returns than the buy portfolio. In these quarters with significantly high positive
PEAD returns, there is no evidence of return drift in the direction of earnings surprises
for the buy portfolios. This evidence is contrary to the view that returns to the PEAD
strategy arise because the sell portfolio continues to lose and the buy portfolio continues
to gain. Further, in the high aggregate earnings shock partitions, returns to the buy and
sell portfolios are both positive. This evidence is again contrary to the view that returns
move in the direction of earnings surprises.
Even more striking, when quarters with insignificant returns to the PEAD strategy
are excluded, returns to the PEAD strategy are positive as both the buy and sell portfolios
have negative returns but the sell portfolio has returns that are more negative than the buy
portfolio. In particular, after excluding the high aggregate earnings shocks quartile
(tercile), returns to the sell portfolio are -3.36% (-4.23%), returns to the buy portfolio are
-0.52% (-1.14%), and returns to the PEAD strategy are 2.85% (3.09%). In comparison, if
I include the high aggregate earnings shocks quartile (tercile), returns to the sell portfolio
are -1.15% and returns to the buy portfolio are 1.05%.
27

The evidence therefore suggests that the conventional view that sell portfolios
continue to lose and buy portfolios continue to gain is supported only when the quarters
with insignificant returns to the PEAD strategy are included. That is, the conventional
view is an on-average effect. Periods that have significant returns to the PEAD strategy
are those in which both the sell and the buy portfolios have negative returns. Periods that
have insignificant returns to the PEAD strategy are those in which both the sell and the
buy portfolios have positive returns. Combining these two sets of periods yields an
average effect in which the sell portfolio experiences negative returns and the buy
portfolio experiences positive returns. Hence, the conventional view that the sell portfolio
continues to lose and the buy portfolio continues to gain does not hold as an explanation
for the PEAD phenomenon.
In summary, returns to the PEAD strategy are driven by aggregate earnings
shocks, and higher positive returns occur in periods with higher negative aggregate
earnings shocks. Additionally, the conventional view that returns to the PEAD strategy
occur because sell portfolio stocks continue to experience negative returns and buy
portfolio stocks continue to experience positive returns is an average effect and cannot
explain PEAD returns.

4.4 PEAD Returns and Systematic Earnings Shocks
Having established that variation in PEAD returns is associated with variation in
aggregate earnings shocks, I next examine what fraction of PEAD strategy returns are
driven by aggregate earnings shocks. To do so, I decompose PEAD strategy returns into
systematic earnings shocks (defined as portfolio returns attributable to covariation with
28

aggregate earnings shocks) and residual returns (defined as total returns minus systematic
earnings shocks). Systematic earnings shocks are measured as follows:
) _ _ (
1 , 1 , + +
=
t
p
t EN t p
Shocks E AGG shocks earnings Systematic β (12)
where
p
t EN ,
β is the earnings shock beta for portfolio p in quarter t. The earnings shock beta
measures the covariance between portfolio earnings shocks and aggregate earnings
shocks, divided by the variance of aggregate earnings shocks. It is estimated using rolling
12-quarter data (t-11 to t).
Table 3, Panel A reports systematic earnings shocks and residual returns for the
PEAD portfolios. Of 2.2% total realized PEAD returns, 1.88% is driven by systematic
earnings shocks, on average. That is, a significant portion of total returns to the PEAD
strategy is attributable to aggregate earnings shocks. Further, differences in residual
returns for the PEAD strategy are not statistically different from zero. This evidence
suggests that a significant fraction of PEAD returns are driven by aggregate earnings
shocks.
Two additional pieces of evidence further suggest that systematic earnings shocks
are more stable over time and are able to capture the core returns to the PEAD strategy.
First, the standard error of the systematic earnings shock estimate is one-fourth that of
realized returns. Second, the regression of PEAD realized returns on systematic earnings
shocks for the PEAD strategy (i.e., systematic earnings shock of the buy portfolio minus
that of the sell portfolio) presented in Panel B of Table 3 shows that a 1% change in
systematic earnings shocks is associated with a 0.97% change in PEAD returns.
Moreover, even though the R
2
of the model is only 6.59%, the intercept is not statistically
29

different from zero and the coefficient on systematic earnings shock is not statistically
different from one. This evidence suggests that systematic earnings shocks capture the
stable component of PEAD returns and the component of PEAD returns unrelated to
systematic earnings shocks is not statistically different from zero.
Table 3: Systematic Earnings Shocks and PEAD Strategy Returns
Panel A: Decomposition of PEAD Realized Returns into Systematic Earnings Shocks and
Residual Returns
Sell -1.15 2.49 -3.01 1.86 2.31 -2.85 1.70
2 0.22 1.01 -1.29 1.51 0.93 -1.14 1.36
3 0.73 0.78 -0.94 1.67 0.70 -0.87 1.59
4 -0.22 0.56 -0.70 0.48 0.48 -0.59 0.37
5 0.66 0.49 -0.64 1.30 0.36 -0.45 1.11
6 0.59 0.56 -0.70 1.28 0.51 -0.63 1.22
7 0.59 0.50 -0.63 1.22 0.46 -0.57 1.16
8 0.94 0.63 -0.79 1.74 0.57 -0.70 1.64
9 0.86 0.76 -1.00 1.86 0.69 -0.85 1.71
Buy 1.05 0.89 -1.13 2.18 0.77 -0.95 2.00
Buy_Sell 2.20 -1.60 1.88 0.32 -1.54 1.90 0.30
t-stat (4.02)*** (8.46)*** (12.67)*** (0.61) (8.21)*** (12.71)*** (0.57)
Standard Err 0.55 0.13 0.15 0.52 0.12 0.15 0.53
Rolling Beta Cross Validation and Rolling Beta
Cross
Validation
E-Beta
Rolling
E-Beta Portfolio
Realized
Returns
Systematic
Earnings
Shocks
Residual
Returns
Systematic
Earnings
Shocks
Residual
Returns

30

Table 3 (Continued)

Panel B: Regression of PEAD Realized Returns on Systematic Earnings Shocks for the
PEAD Strategy

t t t
Strategy PEAD for Shocks Earnings Systematic PEAD ε λ λ + + = _ _ _ _ _
1 0

const Adj R
2
PEAD Strategy Returns 0.38 0.97
t-stat (0.53) (2.44)** 6.59%
p value = 0.95
H
0
:

Coefficient on Systematic_Earnings
_Shocks_for_PEAD_Strategy=1
Systematic_
Earnings _Shocks_
for_PEAD_Strategy

Table 3 reports the time-series means of realized returns and systematic earnings shocks to SUE-
sorted portfolios. The sample includes firms with December fiscal year-end and portfolios are
formed at the beginning of each quarter. Every quarter, 10 portfolios are formed on SUE and
portfolios are held for the subsequent three months. Realized returns and systematic earnings
shocks are measured at the quarterly frequency and the time-series means are reported. Returns
are calculated as the continuously compounded three-month return estimated after the portfolio
formation date. Aggregate Earnings Shock is the aggregate return driven by earnings expectation
revisions and is measured as the equally weighted average of firm-level returns from earnings
expectation revisions. Systematic earnings shocks are estimated as follows:

) _ _ (
1 , 1 , + +
=
t
p
t EN t p
Shock E AGG shocks earnings Systematic β
,

where
p
t EN ,
β
is the earnings shock beta for portfolio p in quarter t. Earnings shock beta is estimated
using rolling 12-quarter data (t-11 to t) and measured as the covariance between portfolio
earnings shocks and aggregate earnings shocks, divided by the variance of aggregate earnings
shocks. Time-series means of rolling earnings shock betas are reported under the column Rolling
E-Beta. Because of the rolling 12-quarter data restriction to estimate the earnings shock betas,
the sample data for this analysis cover the period 1988 to 2009. Residual returns are defined as
realized returns minus systematic earnings shocks. Systematic earnings shocks using the cross
validation method are estimated using the earnings shock betas derived as follows: (1) aggregate
earnings shocks are re-estimated after omitting the firms in portfolio p, and (2) the earnings shock
beta is estimated using the aggregate earnings shocks from (1). Time-series means of earnings
shock betas estimated using this approach are reported under the column Cross Validation E-
Beta.

Panel B reports the regression of PEAD strategy realized returns on systematic earnings shocks
for the PEAD strategy. Systematic_earnings_shocks_for_PEAD_strategy are defined as the buy
portfolio’s systematic earnings shocks minus the sell portfolio’s earnings shocks. *** indicates
significance at the 1 percent level, ** at the 5 percent level, and * at the 10 percent level.

31

One of the limitations of estimating the earnings shock beta for portfolio p (p is
any of the 10 SUE portfolios) is that aggregate earnings shocks also contain portfolio p’s
earnings shocks as aggregate earnings shocks are estimated using the equally weighted
average of firm-level earnings shocks. Thus alternatively, I estimate the earnings shock
beta for portfolio p after removing the firms in portfolio p when estimating aggregate
earnings shock. Earnings shock betas estimated using this approach are lower across all
portfolios compared to the estimates derived from specification (11). However, because
the earnings shock betas derived using this approach are lower for both the buy and the
sell portfolios, the effect on the buy-sell strategy portfolio earnings shock beta is only
marginal (it is reduced by 0.06). Systematic earnings shocks estimated using this
approach are reported under the column Cross Validation and Rolling Beta in Table 3.
Overall, the results using this alternative approach to estimating earnings shock betas are
similar to those using earnings shock betas derived from specification (11).
Table 3, Panel A also shows that the distribution of systematic earnings shocks
has an inverted U-shape across the decile portfolios. Hence, the difference in systematic
earnings shocks is higher for the middle minus sell (portfolio5 minus sell) portfolio than
the hedge (buy minus sell) portfolio. However, the expected returns for the middle
portfolio are significantly lower than those for the sell portfolio (as discussed in Section
3.3 and presented in Table 1). The difference in expected returns for the middle minus
sell portfolio is thus significantly negative and offsets the effects of systematic earnings
shocks to drive the total returns lower. In contrast, the difference in expected returns
between the buy and sell portfolios is marginal. Therefore, overall realized returns for the
middle minus sell portfolio are lower than those for the buy minus sell portfolio.
32

4. 5 What Drives Aggregate Earnings Shocks? The Role of Macroeconomic Shocks
Given that I use analyst forecast revisions to measure revisions to investor
expectations, one potential concern is that my measure of aggregate earnings shocks
merely reflects behavioral biases in analyst forecast revisions such as optimism (Dechow
and Sloan, 1997) or walking down to beatable forecasts (Richardson, Teoh, and Wysocki,
2004) rather than the economic shocks that drive the revisions in earnings expectations.
In this section, I investigate the drivers of aggregate earnings shocks by examining the
relationship between aggregate earnings shocks and macroeconomic shocks. Further, I
investigate the extent to which the portion of aggregate earnings shocks that is
attributable to macroeconomic shocks explains PEAD strategy returns.
To capture the portion of aggregate earnings shocks that is associated with
economic shocks during the quarter, I regress aggregate earnings shocks on lagged,
current, and future macroeconomic variables. The purpose of having future economic
variables in the specification is to capture the part of aggregate earnings shocks that is
driven by current-quarter revisions to expectations about future macroeconomic
activities. Specifically, the relationship between aggregate earnings shocks and
macroeconomic shocks is estimated as follows:
t t t t t t t
t t t t t t t t t
Inf Rf IndP
Def Term GDP Cons Shock E AGG
ε α α α
α α α α α
τ
τ
τ τ
τ
τ τ
τ
τ
τ
τ
τ τ
τ
τ τ
τ
τ τ
τ
τ
+ + + +
+ + + + =
+
− =
+ +
− =
+ +
− =
+
+
− =
+ +
− =
+ +
− =
+ +
− =
+
∑ ∑ ∑
∑ ∑ ∑ ∑
1
1
, 7
1
1
, 6
1
1
, 5
1
1
, 4
1
1
, 3
1
1
, 2
1
1
, 1 0
_ _
(13)

33

where t is the current quarter, AGG_E_Shock is the aggregate earnings shock for the
current quarter, Cons is the per capita growth rate of personal consumption, GDP is the
per capita growth rate of gross domestic product, Term is the yield spread between the
10-year zero-coupon T-bond and three-month T-bills, Def is the yield spread between
Baa- and Aaa-rated corporate bonds, IndP is industrial production growth, Rf is the three-
month T-bill rate, and Inf is the change in the consumer price index. Economic data come
from the Federal Reserve Bank of St. Louis website.
Table 4: Aggregate Earnings Shocks and Macroeconomic Variables
Panel A: Descriptive Statistics
Variable(*100) Mean 10% 25% Median 75% 90% STDDEV
AGG_E_Shock -1.21 -2.23 -1.40 -1.05 -0.74 -0.23 0.90
MKT_Ret 2.78 -9.87 -1.06 3.68 7.88 13.57 8.89
Cons 0.63 0.26 0.47 0.63 0.82 1.09 0.39
GDP 1.26 0.60 0.99 1.33 1.64 2.05 0.66
IndP 0.48 -1.24 0.25 0.72 1.26 1.77 1.28
Term 1.80 0.33 0.87 1.71 2.80 3.32 1.16
Def 0.99 0.64 0.71 0.91 1.16 1.36 0.41
RF 1.07 0.29 0.70 1.19 1.37 1.78 0.55
INF 0.72 0.33 0.54 0.75 0.95 1.18 0.52

34

Table 4 (Continued)

Panel B: Multiple Regression: Aggregate Earnings Shocks and Macroeconomic Variables

t t t t t t t
t t t t t t t t t
Inf Rf IndP
Def Term GDP Cons Shock E AGG
ε α α α
α α α α α
τ
τ
τ τ
τ
τ τ
τ
τ
τ
τ
τ τ
τ
τ τ
τ
τ τ
τ
τ
+ + + +
+ + + + =
+
− =
+ +
− =
+ +
− =
+
+
− =
+ +
− =
+ +
− =
+ +
− =
+
∑ ∑ ∑
∑ ∑ ∑ ∑
1
1
, 7
1
1
, 6
1
1
, 5
1
1
, 4
1
1
, 3
1
1
, 2
1
1
, 1 0
_ _

t-1 t t+1
Cons 1.02 1.48 0.81
(1.86)* (2.66)**** (1.51)
GDP 0.15 0.15 -0.17
(1.15) (1.21) (-1.29)
Term -0.01 0.06 -0.18
(-0.06) (0.25) (-1.08)
Def 0.04 -0.5 -0.44
(0.15) (-1.07) (-1.24)
IndP 0.06 -0.05 0.21
(0.70) (-0.57) (2.47)**
Rf -0.72 -0.61 0.66
(-1.48) (-0.92) (-1.37)
Inf -0.25 -0.77 -0.79
(-0.64) (-1.97)* (-2.05)**
Intercept -0.39 Adj R
2
72.66%
(-0.93)
Quarter (Lag/Lead)

35
Table 4 (Continued)

Panel C: PEAD Strategy Returns for Low and High Fitted Aggregate Earnings Shock Quarters
Full Low AE High AE Low AE High AE Low AE High AE Low AE High AE Low AE High AE
Fitted_AGG_ES -1.21 -3.04 -0.13 -2.38 -0.34 -2.24 -0.42 -2.01 -0.54 -1.74 -0.68
Sell -1.02 -11.21 2.31 -10.82 3.82 -11.18 3.48 -8.07 2.87 -5.11 3.08
Buy 1.14 -6.83 2.27 -7.05 4.55 -7.18 4.63 -4.11 4.21 -2.12 4.39
Buy_Sell 2.16 4.38 -0.04 3.77 0.73 4.00 1.15 3.96 1.34 2.99 1.31
t-stat (4.31)*** (2.57)** (-0.04) (2.53)** 0.68 (3.03)*** 0.97 (3.70)*** (1.87)* (3.86)*** (2.95)***
Diff
t-stat (2.26)** (1.89)* (1.97)* (2.22)** (1.78)*
PEAD Portfolio Returns
4.42 3.04 2.85 2.62 1.68
Decile Quintile Quartile Tercile Median
Partition Partition Partition Partition Partition

Panel D: PEAD Strategy Returns for Low and High Residual Aggregate Earnings Shock Quarters

Full Low AE High AE Low AE High AE Low AE High AE Low AE High AE Low AE High AE
Residual_AGG_ES 0.00 -0.58 0.57 -0.45 0.47 -0.41 0.42 -0.35 0.36 -0.25 0.25
Sell -1.02 -5.09 1.12 -2.91 2.29 -4.22 2.44 -4.17 3.41 -3.34 1.31
Buy 1.14 -2.43 2.11 -0.77 3.50 -1.89 3.72 -1.23 5.43 -0.54 2.82
Buy_Sell 2.16 2.66 0.99 2.14 1.21 2.33 1.28 2.94 2.02 2.80 1.51
t-stat (4.31)*** (2.15)** (0.25) (2.87)*** (1.12) (3.29)*** (1.37) (4.08)*** (1.97)* (4.31)*** (2.01)**
Diff
t-stat
1.67 0.93 1.05 0.92 1.29
(0.39) (0.71) (0.89) (1.24) (0.74)
Partition Partition Partition Partition Partition
PEAD Portfolio Returns
Decile Quintile Quartile Tercile Median
36

Table 4 (Continued)

Panel E: Systematic Macro Earnings Shocks, Residual Macro Earnings Shocks and PEAD
Strategy Returns

Sell -1.15 -2.65 -0.03
2 0.22 -1.26 -0.01
3 0.73 -0.86 -0.05
4 -0.22 -0.53 -0.08
5 0.66 -0.50 -0.02
6 0.59 -0.80 0.01
7 0.59 -0.65 0.01
8 0.94 -0.77 0.01
9 0.86 -0.95 0.01
Buy 1.05 -1.08 -0.01
Buy_Sell 2.20 1.57 0.01
t-stat (4.02)*** (12.28)*** (0.16)
Standard Err 0.55 0.13 0.06
Rolling Beta
Portfolio
Realized
Returns
Systematic
Macro
Earnings
Residual
Macro
Earnings

Panel A of Table 4 reports descriptive statistics for aggregate earnings shocks and
macroeconomic variables. Aggregate earnings shocks (AGG_E_Shock) are aggregate returns
driven by earnings expectation revisions and are measured as the equally weighted average of
firm-level returns from earnings expectation revisions. MKT_Ret is the equally weighted sample
firm return. Macroeconomic data are obtained from the Federal Reserve Bank of St. Louis. Cons
is the per capita growth rate of personal consumption. GDP is the per capita growth rate of gross
domestic product. Term is the yield spread between the 10-year zero-coupon T-bond and the
three-month T-bill. Def is the yield spread between Baa- and Aaa-rated corporate bonds. IndP is
industrial production growth. Rf is the three-month T-bill rate. Inf is the change in the consumer
price index.

Panel B of Table 4 reports the relationship between aggregate earnings shocks and
macroeconomic variables. ***indicates significance at the 1 percent level, ** at the 5 percent
level, and * at the 10 percent level.

Panel C of Table 4 reports the time-series means of returns to the SUE-sorted portfolios for the
full sample as well as for the sample partitioned into low and high fitted aggregate earnings shock
quarters. Fitted aggregate earnings shocks are the fitted value from regressions of aggregate
37

Table 4 (Continued)

earnings shocks on macroeconomic variables. Quarters with fitted aggregate earnings shocks in
the bottom (top) decile, quintile, quartile, tercile, and half are defined as low (high) fitted
aggregate earnings shocks, i.e., Low AE (High AE), for the decile, quintile, quartile, tercile, and
median partitions, respectively. Fitted_AGG_ES is the average fitted aggregate earnings shock for
the partition.

Panel D of Table 4 reports the PEAD returns for the residual aggregate earnings shock partitions.
Residual aggregate earnings shocks are the residual value from regressions of aggregate earnings
shocks on macroeconomic variables. Residual_AGG_ES is the average residual aggregate
earnings shock for the partition.

Panel E of Table 4 reports the time-series means of systematic macro earnings shocks (defined as
portfolio returns attributable to covariance with aggregate earnings shocks that are driven by
macroeconomic shocks). Systematic macro earnings shocks are estimated by (1) estimating fitted
values from the regression of aggregate earnings shocks on macroeconomic shocks (fitted
aggregate earnings shocks); (2) estimating fitted values from the regression of portfolio earnings
shocks on macroeconomic shocks (fitted portfolio earnings shocks); (3) estimating earnings
shock betas by regressing the past 12-quarter fitted portfolio earnings shocks on fitted aggregate
earnings shocks; and (4) each quarter, estimating systematic macro earnings shocks by
multiplying earnings shock betas by fitted aggregate earnings shocks. Residual macro earnings
shocks are estimated using the same procedure as for the systematic macro earnings shocks
except that fitted values are replaced with residual values.

Panel B of Table 4 reports results from a regression of aggregate earnings shocks
on the macroeconomic variables. I find that macroeconomic shocks explain 73% of the
variation in aggregate earnings shocks. Aggregate earnings shocks are significantly
associated with consumption growth, industrial production, and inflation. In contrast,
these variables explain only 18% of the total market return, which is significantly
correlated only with the risk-free rate (untabulated). Overall, the evidence suggests that
most of the variation in aggregate earnings shocks is explained by macroeconomic
shocks.
Next, I examine whether the variation in PEAD returns is attributable to fitted
aggregate earnings shocks (i.e., the fitted values from the regression of aggregate
38

earnings shocks on macroeconomic shocks) or the residual aggregate earnings shocks
(i.e., the residuals from the regression of aggregate earnings shocks on macroeconomic
shocks). Quarters with fitted aggregate earnings shocks in the bottom (top) decile,
quintile, quartile, tercile, and half are defined as low (high) aggregate earnings shock
quarters, i.e., Low AE (High AE), for the decile, quintile, quartile, tercile, and median
partitions, respectively. The same approach is used to partition residual aggregate
earnings shocks. Partitions of fitted aggregate earnings shocks, partitions of residual
aggregate earnings shocks, and realized returns to the PEAD strategy in these partitions
are reported in Panels C and D of Table 4.
Partitioning on fitted aggregate earnings shocks suggests that the PEAD strategy
yields higher positive returns in quarters with low fitted values of aggregate earnings
shocks. That is, in periods when macroeconomic shocks drive investors to revise their
aggregate earnings expectations downward, returns to the PEAD strategy are higher. In
addition, the PEAD strategy yields lower and statistically insignificant returns (with the
exception of the tercile and median partitions) in quarters with high fitted values of
aggregate earnings shocks. That is, returns to the PEAD strategy are lower and even
statistically insignificant when macroeconomic shocks drive aggregate earnings shocks
close to zero or positive. These differences in PEAD returns between fitted low and high
partitions are statistically different from each other with a difference in returns of 4.42%
(3.04%, 2.85%, 2.62%, 1.68%) for the decile (quintile, quartile, tercile, median) splits,
respectively. In contrast, even though the quarters with low residual values of aggregate
earnings shocks have higher PEAD returns than those in the high residual aggregate
earnings shock partition, differences in PEAD returns are not statistically different
39

between these partitions. Overall, the evidence suggests that returns to the PEAD strategy
are related to macroeconomic shocks.
18

Finally, I investigate what fraction of PEAD returns are attributable to aggregate
earnings shocks that are related to macroeconomic shocks. Specifically, I estimate
systematic macro earnings shocks (defined as portfolio returns attributable to covariation
with the aggregate earnings shocks that are driven by macroeconomic shocks). I estimate
systematic macro earnings shocks as follows: (1) I estimate fitted values from the
regression of aggregate earnings shocks on macroeconomic shocks (fitted aggregate
earnings shocks), (2) I estimate fitted values from the regression of portfolio earnings
shocks on macroeconomic shocks (fitted portfolio earnings shocks). (3) I estimate
earnings shock betas by regressing past 12-quarter fitted portfolio earnings shocks on
fitted aggregate earnings shocks, and (4) each quarter, I estimate systematic macro
earnings shocks by multiplying earnings shock beta with fitted aggregate earnings
shocks. Residual macro earnings shocks are estimated using the same procedure as
systematic macro earnings shocks except that fitted values are replaced with residual
values. The time-series means of systematic macro earnings shocks are reported in Panel
E of Table 4. The evidence suggests that, of 2.20% total PEAD returns, systematic
earnings shocks account for 1.57%, on average. That is, a significant portion of the
PEAD returns can be attributable to earnings shocks driven by macroeconomic shocks.
Overall, the evidence suggests that most of the variation in aggregate earnings
shocks is attributable to macroeconomic shocks, and that the aggregate earnings shocks

18
Chordia and Shivakumar (2006) examine the relation between PEAD returns and macroeconomic
activity. They find that current-period PEAD returns are counter-cyclically related to future-period
economic activity. Evidence presented in this paper suggests that the counter-cyclical result may be an
outcome of a positive relation between aggregate earnings shocks and business cycle variables.
40

attributable to macroeconomic shocks drive a significant portion of PEAD returns.
However, it is important to note that the evidence presented here cannot completely rule
out the possibility that some fraction of aggregate earnings shocks reflects some form of
aggregate analyst optimism.

4.6 Do Aggregate Non-Earnings Shocks Drive PEAD Returns?
Thus far, I have shown that expected returns cannot explain PEAD returns while
aggregate earnings shocks explain a significant part of realized returns to the PEAD
strategy. In this section, I examine whether aggregate non-earnings shocks can explain
the variation in PEAD returns. Aggregate non-earnings shocks are measured as aggregate
realized returns minus aggregate earnings shocks. Aggregate non-earnings shocks capture
the sum of aggregate discount rate shocks and aggregate expected returns.
19

To examine the relationship between aggregate non-earnings shocks and PEAD
returns, I make various partitions on aggregate non-earnings shocks. Quarters with
aggregate non-earnings shocks in the bottom (top) decile, quintile, quartile, tercile, and
half are defined as low (high) aggregate non-earnings shocks for the decile, quintile,
quartile, tercile, and median partitions, respectively. In untabulated results, I do not find
any relationship between aggregate non-earnings shock partitions and PEAD returns,
despite the fact that aggregate non-earnings shocks vary considerably in these partitions.
For example, aggregate non-earnings shocks average -10.98% for quarters in the lowest

19
As a robustness check, I also estimate aggregate non-earnings shocks as residuals from a time series
regression of quarterly returns on aggregate earnings shocks. Overall the conclusions using this measure are
similar to those using the measure derived using aggregate realized returns minus aggregate earnings
shocks.

41

quartile and 10.48% for quarters in the highest quartile. However, the difference in the
PEAD returns between these two partitions is only 0.89%, which is not statistically
significant. The observation that significant variation in aggregate non-earnings shocks
exists between the quartiles but very little cross-sectional variation (i.e., for the PEAD
portfolios) exists within a quartile is driven by the fact that risk exposure differences
between the extreme PEAD portfolios is marginal (as shown in Table 1 and discussed in
Section 3.3). If risk exposure differences are marginal then the significant driver of
aggregate non-earnings shocks is changes in risk premiums. However, changes in risk
premiums are constant in the cross-section and therefore cannot explain returns to a
cross-sectional strategy such as PEAD. In summary, I do not find any relationship
between aggregate non-earnings shocks and PEAD returns.

42

Chapter 5: Alternative Explanations and Additional Analysis

5.1 Do Aggregate Earnings Shocks Explain PEAD Returns Beyond the Investor Naivety
Hypothesis?
In this section, I examine whether the relationship that I document between
aggregate earnings shocks and PEAD returns is driven by the investor naivety hypothesis.
The investor naivety hypothesis posits that investors fail to understand the time-series
properties of earnings (Bernard and Thomas, 1989, 1990).
20
According to this hypothesis,
firms that have positive (negative) earnings surprises in the current quarter will also have
positive (negative) earnings surprises in the next quarter; however, because investors are
naïve and do not understand this phenomenon, firms with positive (negative) earnings
surprises in the current quarter will surprise investors with positive (negative) earnings in
the next quarter and hence will generate positive (negative) returns.
To examine whether the relationship between aggregate earnings shocks and
PEAD returns is driven by investor naivety, I partition the quarters into low and high
aggregate earnings shock quarters. Quarters with aggregate earnings shocks in the bottom
(top) quartile are defined as low (high) aggregate earnings shock quarters, i.e., Low AE
(High AE). If investor naivety explains the variation in PEAD returns, then the buy
portfolio (sell portfolio) should have positive (negative) one-quarter-ahead earnings
surprises and the magnitude of the one-quarter-ahead earnings surprises should vary with
the magnitude of PEAD returns.

20
Ball and Bartov (1996) and Markov and Tamayo (2006) argue that predictability does not imply
irrationality; predictability could also be an outcome of parameter uncertainty, with investors/analysts
learning about parameters over time.
43

Table 5: Investor Naivety and Analyst Underreaction to Prior Earnings Information

Panel A: Investor Naivety: Seasonal Random Walk Earnings for Low and High Aggregate
Earnings Shock Quarters

Portolio
RW (*100) t t+1 t t+1
SELL -5.13 -5.77 -1.92 -0.58
2 -0.58 -1.39 -0.26 -0.04
3 -0.33 -0.86 0.11 0.07
4 -0.28 -0.58 0.20 0.18
5 -0.08 -0.15 0.24 0.18
6 0.01 -0.30 0.26 0.26
7 0.03 -0.21 0.35 0.37
8 0.05 -0.22 0.47 0.45
9 0.10 -0.64 0.76 0.46
BUY 0.89 -0.48 1.72 1.77
PEAD returns 3.61% 0.04%
Random Walk Earnings Changes: Portfolios Formed at Quarter t
Low AE High AE

Panel B: Analyst Underreaction: Analyst Forecast Errors for Low and High Aggregate
Earnings Shock Quarters

Portolio
AF (*100) t t+1 t t+1
SELL -5.03 -3.78 -2.99 -1.18
2 -0.45 -1.16 -0.36 -0.23
3 -0.18 -0.27 -0.12 -0.07
4 -0.09 -0.21 -0.04 -0.02
5 -0.01 -0.09 0.01 0.00
6 0.03 -0.35 0.06 0.03
7 0.08 0.01 0.11 0.05
8 0.16 -0.07 0.20 0.06
9 0.33 -0.09 0.37 0.13
BUY 1.55 -0.43 1.64 0.27
PEAD returns 3.61% 0.04%
Analyst Forecast Errors: Portfolios Formed at Quarter t
Low AE High AE

44

Table 5 (Continued)

Panel A of Table 5 reports the seasonally differenced quarterly earnings for the Low AE and High
AE subsamples. Quarters with aggregate earnings shocks in the bottom (top) quartile are defined
as quarters with low (high) aggregate earnings shocks, i.e., Low AE (High AE). Every quarter, 10
portfolios are formed on SUE. One-quarter-ahead seasonally differenced quarterly earnings are
tracked for each portfolio and time-series means are reported under the column (t+1). SUE
t
is
measured as current earnings minus four-quarter-ago earnings, scaled by price per share at the
end of quarter t. PEAD returns are the returns to the PEAD strategy in these partitions.

Panel B reports the quarterly analyst forecast errors for the Low AE and High AE subsamples.
The estimation methodology is the same as in Panel A, except that seasonally differenced
quarterly earnings are replaced with analyst forecast errors. Analyst forecast errors are measured
as earnings minus median analyst forecasts 90 days prior to the earnings announcement, scaled by
price per share at the end of quarter t.

Panel A of Table 5 reports the earnings surprises (seasonally differenced quarterly
earnings) for the portfolios formed at quarter t as well as the one-quarter-ahead earnings
surprises for these portfolios. The low aggregate earnings shock partition (Low AE) has
PEAD returns of 3.61% on average, which is 1.7 times the PEAD return for the full
sample. If investor naivety explains PEAD returns, then one should observe negative
surprises for the sell portfolio and positive surprises for the buy portfolio. However, in
this partition all the portfolios have negative one-quarter-ahead earnings surprises. This
evidence suggests that investor naivety does not drive the relationship between aggregate
earnings shocks and PEAD returns. Further, the high aggregate earnings shock partition
does provide supportive evidence for the investor naivety hypothesis: the buy (sell)
portfolio has positive (negative) earnings surprises. However, the returns to the PEAD
strategy in this partition are both economically and statistically insignificant (0.04%).
Overall, the evidence suggests that the relationship between aggregate earnings shocks
and PEAD returns is not driven by investor naivety.
45

5.2 Do Aggregate Earnings Shocks Explain PEAD Returns Beyond Analyst
Underreaction to Prior Earnings Information?
In this section, I examine whether the relationship between aggregate earnings
shocks and PEAD returns is driven by analyst underreaction to prior earnings
information. Abarbanell and Bernard (1992) argue that analysts underreact to prior
earnings information and hence analyst forecast errors are predictable. If analyst
underreaction drives PEAD returns, then one should observe positive (negative) surprises
to the buy (sell) portfolio. Panel B of Table 5 reports the analyst forecast errors for the
portfolios formed at quarter t and also the one-quarter-ahead forecast errors for these
portfolios.
Examining the one-quarter-ahead analyst forecast errors for the alternative
aggregate earnings shock partitions leads to several observations. First, in the low
aggregate earnings shock partition, in which PEAD returns are 1.7 times the PEAD
returns of the full sample, all of the PEAD portfolios have negative one-quarter-ahead
analyst forecast errors. Second, despite the fact that the high aggregate earnings shock
partition has positive (negative) analyst forecast errors for the buy (sell) portfolio, PEAD
returns in this partition are both economically and statistically insignificantly different
from zero (0.04%). The evidence therefore suggests that the relationship between
aggregate earnings shocks and PEAD returns is not driven by analyst underreaction to
prior earnings information.

46

5.3 Do Aggregate Earnings Shocks Explain PEAD Returns Beyond the Inflation Illusion
Hypothesis?
In this section, I examine whether aggregate earnings shocks can explain the
variation in PEAD returns after controlling for the effect of inflation illusion on PEAD
returns as documented by Chordia and Shivakumar (2005). The inflation illusion
hypothesis posits that investors do not understand the implications of inflation in
forecasting earnings. As a result, investors forecast the same nominal earnings rather than
the same real earnings in high and low inflation periods, in which case stocks prices are
overvalued in low inflation periods and undervalued in high inflation periods (Modigliani
and Cohn, 1979; Campbell and Vuolteenaho, 2004).
21

Table 6: The Inflation Illusion Hypothesis
Panel A: Inflation Illusion Hypothesis Test

Portfolio Low INF High INF Low AE High AE
Sell -0.29 -5.57 -7.88 5.79
2 0.65 -2.94 -4.28 4.68
3 0.61 -1.49 -2.17 3.87
4 -1.16 -2.96 -2.88 2.44
5 1.47 -0.75 -2.12 3.87
6 0.09 -0.17 -2.71 3.74
7 0.60 -0.68 -2.61 4.26
8 0.94 -0.68 -2.59 4.59
9 0.50 -1.13 -3.16 4.52
Buy 1.48 -2.11 -4.92 5.91
Buy_Sell 1.77 3.47 2.96 0.13
t-stat (2.04)** (2.51)** (2.14)** (0.16)
Lagged Inflation
Inflation Adjusted
AGG_E_Shocks

21
Konchitchki (2011) also documents that investors do not fully reflect the implications of inflation effects
for future cash flows.
47

Table 6 (Continued)

Panel B: Double Sorts on Lagged Inflation and Aggregate Earnings Shocks

Low High Diff t-stat
Low 3.01 (3.74)*** 0.39 (0.47) 2.62 (2.26)**
High 3.19 (1.96)* 1.91 (2.84)*** 1.28 (0.78)
Diff 0.18 1.52
t-stat (0.10) (1.45)
AGG_E_Shocks
L_Inflation

Panel A of Table 6 reports the time-series means of returns to the SUE-sorted portfolios in low
and high lagged inflation quarters. INF is the quarter-over-quarter change in the consumer price
index. Quarters with lagged inflation less than the 25
th
percentile are defined as low inflation
periods (Low INF) and those above the 75
th
percentile are defined as high inflation periods (High
INF). Inflation-adjusted aggregate earnings shocks are the residuals from the regression of
aggregate earnings shocks on lagged inflation. *** indicates significance at the 1 percent level, **
at the 5 percent level, and * at the 10 percent level.

Panel B reports the PEAD returns to the 2x2 double sorts between lagged inflation and aggregate
earnings shocks. All quarters are divided into four groups based on independent sorts on the
lagged inflation and aggregate earnings shocks. For each group, time-series averages of the
returns to the PEAD strategy are estimated. Diff is the time-series average of the difference in
returns between extreme groups. All numbers are in percentages. t-statistics are reported next to
the coefficient estimates in brackets. *** indicates significance at the 1 percent level, ** at the 5
percent level, and * at the 10 percent level.

Chordia and Shivakumar (2005) find that inflation predicts returns to the PEAD
strategy. Panel A of Table 6 presents results on the inflation illusion hypothesis. Quarters
in which lagged inflation (lagged by one quarter) is less than the 25
th
percentile are
defined as low inflation quarters (Low INF) and quarters in which lagged inflation is
above the 75
th
percentile are defined as high inflation quarters (High INF). Consistent
with the findings of Chordia and Shivakumar (2005), I find that lagged inflation and
returns to the PEAD strategy are related. Specifically, quarters with low inflation have
lower one-quarter-ahead PEAD strategy returns of 1.77% while the returns to the PEAD
strategy more than double to 3.47% in high inflation quarters.
48

Aggregate earnings shocks could be related to PEAD returns because aggregate
earnings shocks and the inflation illusion are related. I investigate this possibility using
two methods. First, I estimate aggregate earnings shock residuals by regressing aggregate
earnings shocks on lagged inflation. The intuition is that aggregate earnings shock
residuals control for the predictable part of earnings revisions that are driven by the
inflation illusion effect. Second, I double sort (i.e., independently sort) on aggregate
earnings shocks and lagged inflation and examine the returns to the PEAD strategy.
Results are presented in Panel B of Table 6.
The evidence from both methods suggests that even after controlling for inflation
illusion effects, returns to the PEAD strategy are inversely related to aggregate earnings
shocks. As such, aggregate earnings shocks explain PEAD returns even after controlling
for the inflation illusion effect.

5.4 Do Results Hold for Firms Without Analyst Following?
Finally, I examine the robustness of the main findings to firms with no analyst
following. For these firms, SUE is estimated from the seasonal random walk model, and
the classification of quarters into low and high aggregate earnings shocks is the same as
described in Section 4.3 (i.e., aggregate earnings shocks are derived from the firms with
analyst following). The findings from the no-analyst-following firms are consistent with
those for the analyst-following firms. Specifically, aggregate earnings shocks and PEAD
returns are negatively correlated: low aggregate earnings shock quarters have significant
PEAD returns, while high aggregate earnings shock quarters have statistically
insignificant PEAD strategy returns with the exception of the median partition.
49

Chapter 6: Conclusions

This paper shows that returns to the post-earnings-announcement drift (PEAD)
strategy are driven by the differential sensitivity of the sell versus the buy PEAD
portfolio returns to aggregate earnings shocks. The higher sensitivity of the sell portfolio
(relative to the buy portfolio) to aggregate earnings shocks implies that larger negative
aggregate earnings shocks translate into higher negative returns for the sell portfolio than
for the buy portfolio. Therefore, larger negative aggregate earnings shocks are associated
with higher PEAD strategy (buy minus sell portfolio) returns. Aggregate earnings shocks
on average should be mean-zero in expectation or over a long sample period. However,
the 1985 to 2009 sample period is dominated by large negative aggregate earnings
shocks, which explains why the PEAD strategy generates positive returns on average
over this period.
The evidence presented in this paper also suggests that a significant portion of the
variation in aggregate earnings shocks is explained by macroeconomic shocks. Further
analysis suggests that macroeconomic shocks working through aggregate earnings shocks
drive the variation in PEAD strategy returns. Specifically, when macroeconomic shocks
contribute to downward revisions to aggregate expectations, PEAD strategy returns are
positive.

50

References
Abarbanell, J., 1991. Do analysts’ earnings forecasts incorporate information in prior stock price
changes? Journal of Accounting and Economics 14, 147–165.
Abarbanell, J., Bernard, V., 1992. Tests of analysts’ overreaction/underreaction to earnings
information as an explanation for anomalous stock price behavior. Journal of Finance 47,
1181–1207.
Abarbanell, J., Lehavy, R., 2003. Biased forecasts or biased earnings? The role of reported
earnings in explaining apparent bias and over/underreaction in analysts’ earnings forecasts.
Journal of Accounting and Economics 36, 105–146.
Ali, A., Klein, A., Rosenfeld, J., 1992. Analysts use of information about permanent and
transitory earnings components in forecasting annual EPS. The Accounting Review 67, 183–
198.
Balakrishnan, K., Bartov, E., Faurel, L., 2010. Post loss/profit announcement drift. Journal of
Accounting and Economics 50, 20–41.
Ball, R., 1978. Anomalies in relationships between securities, yields and yield surrogates.
Journal of Financial Economics 6, 103–126.
Ball, R. and Bartov, E., 1996. How naïve is the stock market’s use of earnings information?
Journal of Accounting and Economics 21, 319–337.
Ball, R., and Brown, P., 1968. An empirical evaluation of accounting numbers. Journal of
Accounting Research 6, 159–178.
Ball, R., Kothari, SP., Watts, R., 1993. Economic determinants of the relation between earnings
changes and stock returns. The Accounting Review 68, 622–638.
Ball, R., Sadka, G., Sadka, R., 2009. Aggregate earnings and asset prices. Journal of Accounting
Research 47, 1097–1133.
Bartov, E., Krinsky, I., Radhakrishnan, S., 2000. Investor sophistication and patterns in stock
returns after earnings announcement. The Accounting Review 75, 43–63.
51

Basu, S., Markov, S., Shivakumar, L., 2010. Inflation, earnings forecasts, and post earnings
announcement drift. Review of Accounting Studies 15, 403–440.
Bernard, V., Thomas, J., 1989. Post-earnings announcement drift: delayed price response or risk
premium. Journal of Accounting Research 27, 1–36.
Bernard, V., Thomas, J., 1990. Evidence that stock prices do not fully reflect the implications of
current earnings for future earnings. Journal of Accounting and Economics 13, 305–340.
Bhushan, R., 1994. An informational efficiency perspective on the post-earnings announcement
drift. Journal of Accounting and Economics 18, 45–65.
Brown, P., Ball, R., 1967. Some preliminary findings on the association between the earnings of
a firm, its industry, and the economy. Journal of Accounting Research 5, 55-77.
Brown, L., Han, J., 2000. Do stock prices fully reflect the implications of current earnings for
future earnings for ar1 firms? Journal of Accounting Research 38, 149–164.
Brown, P., Kennelly, J., 1972. The information content of quarterly earnings: an extension and
some further evidence. Journal of Business 45, 403–415.
Campbell, J., Shiller, R., 1988. The dividend-price ratio and expectations of future dividends and
discount factors. Review of Financial Studies 1, 195–228.
Campbell, J., Vuolteenaho, T., 2004. Bad beta, good beta. American Economic Review 94,
1249–1275.
Campbell, J., Vuolteenaho, T., 2004. Inflation illusion and stock prices. American Economic
Review Papers and Proceedings 94, 19–23.
Cao, S., Narayanamoorthy, G., 2012. Earnings volatility, post-earnings announcement drift and
trading frictions. Journal of Accounting Research 50, 41–74.
Chan, L., Narasimhan, J., Lakonishok, J., 1996. Momentum strategies. Journal of Finance 51,
1681–1713.
Chan, L., Karceski, J., Lakonishok, J., 2003. The level and persistence of growth rates. Journal of
Finance 58, 643–684.
52

Chava, S., Purnanandam, A., 2010. Is default risk negatively related to stock returns? Review of
Financial Studies 23, 2523–2559.
Chen, L., Zhao, X., 2010. What drives stock price movement?. Available at SSRN:
http://ssrn.com/abstract=1121893
Chen, L., Zhao, X., 2009. Return decomposition. Review of Financial Studies 22, 5213–5249.
Choi, J., Kalay, A., Sadka, G., 2011. Earnings news and aggregate stock returns. Working paper,
Columbia University.
Chordia, R., Goyal, A., Sadka, G., Sadka, R., Shivakumar, L., 2009. Liquidity and the post
earnings-announcement drift. Financial Analyst Journal 65, 18–32.
Chordia, T., Shivakumar, L., 2005. Inflation illusion and post-earnings-announcement drift.
Journal of Accounting Research 43, 521–556.
Chordia, T., Shivakumar, L., 2006. Earnings and price momentum. Journal of Financial
Economics 80, 627–656.
Collins, D., Hribar, P., 2000. Earnings-based and accrual-based market anomalies: one effect or
two? Journal of Accounting and Economics 29, 101–123.
Cready, W., Gurun, U., 2010. Aggregate market reaction to earnings announcements. Journal of
Accounting Research 48, 289–334.
Da, Z., Warachka, M., 2009. Cashflow risk, systematic earnings revisions, and the cross-section
of stock returns. Journal of Financial Economics 94, 448–468.
Dechow, P., Sloan, R., 1997. Returns to contrarian investment strategies: tests of naïve
expectations hypotheses. Journal of Financial Economics 41, 3–27.
Dontoh, A., Ronen, J., Sarath, B., 2003. On the rationality of the post-announcement drift,
Review of Accounting Studies 8, 69-104.
Doyle, J., Lundholm, R., Soliman, M., 2006. The extreme future stock returns following I/B/E/S
earnings surprises. Journal of Accounting Research 44, 849–88.
53

Easton, P., 2004. PE ratios, PEG ratios, and estimating the implied expected rate of return on
equity capital. Accounting Review 79, 73–95
Easton, P., Monahan, S., 2005. An evaluation of accounting-based measures of expected returns.
The Accounting Review 80, 501–538.
Fama, E., 1998. Market efficiency, long-term returns and behavioral finance. Journal of
Financial Economics 49, 283–306.
Fama, E., French, K., 1993. Common risk factors in the returns on stocks and bonds. Journal of
Financial Economics 33, 3–56.
Ferson, W., Harvey, C., 1991. The variation of economic risk premiums. Journal of Political
Economy 99, 285–315.
Foster, G., Olsen, C., Shevlin, T., 1984. Earnings releases, anomalies and the behavior of
security returns. The Accounting Review 59, 574–603.
Francis, J., Lafond, R., Olsson, P., Schipper, K., 2007. Information uncertainty and post
earnings-announcement-drift. Journal of Business Finance and Accounting 34, 403–443.
Freeman, R., Tse, S., 1989. The multiperiod information content of accounting earnings:
confirmations and contradictions of previous earnings reports. Journal of Accounting Research
27, 49–84.
Gebhardt, W., Lee, C., Swaminathan, B., 2001. Toward an implied cost of capital. Journal of
Accounting Research 39, 135–176
Gode, D., Mohanram, P., 2003. Inferring the cost of capital estimates using the Ohlson-Juettner
model. Review of Accounting Studies 8, 399–431.
Hann, R., Ogneva, M., Sapriza, H., 2012. Forecasting the macroeconomy: analysts versus
economists. Working paper, University of Maryland.
Hirshleifer, D., Lim, S., Teoh, S., 2009. Driven to distraction: extraneous events and
underreaction to earnings news. Journal of Finance 64, 2289–2325
54

Hughes, J., Liu, J., Su., W., 2008. On the relation between predictable market returns and
predictable analyst forecast errors. Review of Accounting Studies 13, 266–291.
Jacob, J., Lys, T., Sabino, J., 1999. Autocorrelation structure of forecast errors from time-series
models: alternative assessments of the causes of post-earnings announcement drift. Journal of
Accounting and Economics 28, 329–358.
Jegadeesh, N., Titman, T., 1993. Returns to buying winners and selling losers: implications for
stock market efficiency. Journal of Finance 48, 65–91.
Jones, C., Litzenberger, R., 1970. Quarterly earnings reports and intermediate stock price trends.
Journal of Finance 25, 143–148.
Joy, M., Litzenberger, R., Mcenally, R., 1977. The adjustment of stock prices to announcements
of unanticipated changes in quarterly earnings. Journal of Accounting Research 15, 207–25.
Khan, M., 2008. Are accruals mispriced? evidence from tests of an intertemporal capital asset
pricing model. Journal of Accounting and Economics 45, 55–77.
Konchitchki, Y., 2011. Inflation and nominal financial reporting: implications for performance
and stock prices. The Accounting Review 86, 1045–1085.
Konchitchki, Y., Lou, X., Sadka, G., Sadka, R., 2011. Underreaction or risk? expected earnings
and the post-earnings-announcement drift. Working paper, University of California, Berkeley.
Kothari, S., 2001. Capital markets research in accounting. Journal of Accounting and Economics
31, 105–231.
Kothari, S., Lewellen, J., Warner, J., 2006. Stock returns, aggregate earnings surprises, and
behavioral finance. Journal of Financial Economics 79, 537–568.
Latane, H., Jones, C., 1979. Standardized unexpected earnings — 1971–1977. Journal of Finance
34, 717–724.
Latane, H., Joy, M., Jones, C., 1970. Quarterly data, sort-rank routines, and security evaluation.
The Journal of Business 43, 427–38.
55

Liang, L., 2003. Post-earnings announcement drift and market participants’ information
processing biases. Review of Accounting Studies 8, 321–45.
Livnat, J., Mendenhall, R., 2006. Comparing the post-earnings announcement drift for surprises
calculated from analyst and time series forecasts. Journal of Accounting Research 44,177–205.
Livnat, J., Petrovits, C., 2009. Investor sentiment, post-earnings announcement drift and
accruals. Working paper, New York University.
Lys, T., Sohn, S., 1990. The association between revisions of financial analysts earnings
forecasts and security-price changes. Journal of Accounting and Economics 13, 341–363.
Magee, R., 1974. Industry-wide commonalities in earnings. Journal of Accounting Research 12,
270–287.
Mendenhall, R., 2004. Arbitrage risk and post-earnings-announcement drift. Journal of Business
77: 875–94.
Modigliani, F., Cohn, R., 1979. Inflation, rational valuation and the market. Financial Analyst
Journal 35, 24–44.
Narayanamoorthy, G., 2005. Conservatism and cross-sectional variation in the post-earnings
announcement drift. Journal of Accounting Research 44, 763–89.
Nekrasov, A., Ogneva, A., 2011. Using earnings forecasts to simultaneously estimate firm
specific cost of equity and long-term growth. Review of Accounting Studies 16, 414–457.
Ng, J., Rusticus, T., Verdi, R., 2008. Implications of transaction costs for the post earnings
announcement drift. Journal of Accounting Research 46, 661–696.
Ogneva, M., 2012. Accrual quality, realized returns, and expected returns: the importance of
controlling for cash flow shocks. The Accounting Review 87, 1414-1444.
Ohlson, J., Juettner-Nauroth, B., 2005. Expected EPS and EPS growth as determinants of value.
Review of Accounting Studies 10, 349–365
Patatoukas, P., 2011. Detecting news in aggregate accounting earnings: implications for stock
market valuation. Working paper, University of California, Berkeley.
56

Payne, J., Thomas B., 2003. The implications of using stock-split adjusted I/B/E/S data in
empirical research. The Accounting Review 78, 1049–1067
Penman, S., 1984. Abnormal returns to strategies based on the timing of earnings reports. Journal
of Accounting and Economics 6, 165–183.
Rangan, S., Sloan, R., 1998. Implications of the integral approach to quarterly reporting for the
post-earnings announcement drift. The Accounting Review 73, 353–371.
Rendleman, R., Jones, C., Latane, H., 1982. Empirical anomalies based on unexpected earnings
and the importance of risk adjustments. Journal of Financial Economics 10, 269–87.
Rendleman, R., Jones, C., Latane, H., 1987. Further insight into the standardized unexpected
earnings anomaly: size and serial correlation effects. Financial Review 22, 131–144.
Richardson, S., Tuna, I., Wysocki, P., 2010. Accounting anomalies and fundamental analysis: a
review of recent research advances. Journal of Accounting and Economics 50, 410–454.
Sadka, G., 2007. Understanding stock price volatility: the role of earnings. Journal of Accounting
research 48, 199–228.
Sadka, G., Sadka, R., 2009. Predictability and the earnings-returns relation. Journal of Financial
Economics 94, 87–106.
Shivakumar, L., 2006. Accruals, cash flows, and the post-earnings-announcement drift. Journal
of Business Finance and Accounting 33, 1–25.
Shivakumar, L., 2007. Aggregate earnings, stock market returns and macroeconomic activity: a
discussion of ’does earnings guidance affect market returns? The nature and information
content of aggregate earnings guidance’. Journal of Accounting and Economics 44, 64–73.
Soffer, L., Lys, T., 1999. Post-earnings-announcement-drift and the dissemination of predictable
information. Contemporary Accounting Research 16, 305–331.
Vuolteenaho, T., 2002. What drives firm-level stock returns? Journal of Finance 57, 233–264.
Watts, R., 1978. Systematic abnormal returns after quarterly earnings announcements. Journal of
Financial Economics 6, 127–150.
57

Wu, J.,Zhang, L., 2011. Do anomalies exist ex ante?. Working paper. Ohio State University.
Zhang, Y., 2008. Analyst responsiveness and the post-earnings-announcement drift. Journal of
Accounting and Economics 46, 201–215.

58

Appendix: Estimation of Implied Cost of Capital

Implied cost of capital estimates are used extensively in the accounting and
finance literature as proxies for expected returns. I use two different methods to estimate
implied cost of capital. The first method follows Easton (2004) and is derived from the
Ohlson and Juettner-Nauroth (2005) abnormal earnings growth model. Implied cost of
capital estimated using this approach is denoted by r
PEG
.
t i
s i s i PEG
t i
P
eps eps
r
,
1 , 2 ,
,
) (
+ +
−
=
(A1)

where eps
i,s+n
is the n-years-ahead consensus forecast at the end of quarter t of year s,
with s the current year, and P
it
is price per share for firm i at the end of quarter t.

The second method that I use to estimate implied cost of capital follows Gebhardt,
Lee, and Swaminathan (2001) and is denoted by r
GLS
. From the following valuation
equation, r
GLS
is derived through an iterative procedure:

∑
=
+ − + +
+
−
+
+
−
+ =
11
1
11
, ,
11 , ,
,
1 , , ,
) 1 (
) (
) 1 (
) (
n
GLS
t i
GLS
t i
s i
GLS
t i
n GLS
t i
n s i
GLS
t i n s i
it it
r r
b r indROE
r
b r roe
bps P
(A2)

where
1 , , , − + + +
=
n s i n s i n s i
bps eps roe for n=1, 2. For n>2, roe mean-reverts to the industry
median roe (IndRoe), IndRoe is the median roe for all firms in the same industry
spanning year s-4 through year s with positive earnings and positive book value of equity.
Industry classification follows the Fama and French (1997) industry definitions.
n s i
eps
+ ,
is
the n-years-ahead consensus forecast at the end of quarter t. bps is book value per share
and is estimated using the clean surplus relation for n>1.
22

22
To estimate the book value of equity using the clean surplus relation, the dividend payout ratio is
required. I estimate the dividend payout ratio as dividends over earnings for profitable firms and dividends
over 6% of total assets for unprofitable firms (Gebhardt, Lee, and Swaminathan, 2001).

Abstract (if available)

Abstract This paper finds that returns to the post-earnings-announcement drift (PEAD) strategy result from differential sensitivity of individual stock returns to aggregate earnings shocks. Larger negative aggregate earnings shocks are associated with higher PEAD returns, because stocks in the PEAD’s sell portfolio are more sensitive to aggregate earnings shocks than those in the buy portfolio. Such differential sensitivity to aggregate earnings shocks drives a significant portion of PEAD returns. During the 1985 to 2009 sample period, investors were on average negatively surprised by aggregate earnings shocks, leading to average positive returns to the PEAD strategy. Further analysis suggests that macroeconomic shocks (that work through aggregate earnings shocks) explain the variation in PEAD returns.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Accrual quality and expected returns: the importance of controlling for cash flow shocks

PDF

Disclosure distance and earnings announcement returns

PDF

The term structure of CAPM alphas and betas

PDF

Essays in asset pricing

PDF

“What you see is all there is”: The effects of media co‐coverage on investors’ peer selection

PDF

Three essays in derivatives, trading and liquidity

PDF

Essays in empirical asset pricing

PDF

Real asset liquidity and asset impairments

PDF

Private disclosure contracts within supply chains and managerial learning from stock prices

PDF

Does fair value accounting exacerbate the pro-cyclicality of bank lending?

PDF

Who manages the firm matters: the incremental effect of individual managers on accounting quality

PDF

Two essays on major macroeconomic shocks in the Japanese economy: demographic shocks and financial shocks

PDF

Can I borrow your footnotes? Learning and network benefits of footnote similarity

PDF

Two essays on financial econometrics

PDF

Mutual fund screening versus weighting

PDF

Evolution of returns to scale and investor flows during the life cycle of active asset management

PDF

Share repurchases: how important is market timing?

PDF

Essays on the effect of cognitive constraints on financial decision-making

PDF

Essays on real options

PDF

Essays on delegated portfolio management under market imperfections

Asset Metadata

Creator Nallareddy, Suresh (author)

Core Title Does differential sensitivity to aggregate earnings shocks drive post-earnings-announcement drift?

School Marshall School of Business

Degree Doctor of Philosophy

Degree Program Business Administration

Publication Date 09/28/2012

Defense Date 06/01/2012

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag earnings beta,OAI-PMH Harvest,post earnings announcement drift

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Subramanyam, K.R. (committee chair), Dekle, Robert (committee member), Ferson, Wayne (committee member), Ogneva, Maria (committee member)

Creator Email sn2520@columbia.edu,suresh.nallareddy@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-101134

Unique identifier UC11289190

Identifier usctheses-c3-101134 (legacy record id)

Legacy Identifier etd-Nallareddy-1217.pdf

Dmrecord 101134

Document Type Dissertation

Rights Nallareddy, Suresh

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA