Abnormal Returns from Predicting Earnings
LYNN REES firstname.lastname@example.org
Mays Business School, Texas A&M University, College Station, TX 77843-4353
Abstract. This study examines the performance of a trading strategy based on the prediction of ﬁrms
concurrently reporting a positive earnings change and meeting analysts’ earnings forecasts. The evidence
indicates that a model predicting both earnings thresholds concurrently can yield excess returns that are
incremental to predicting only one earnings threshold. Further, I ﬁnd that the prediction of forecast errors
is relatively more important than predicting earnings changes as the incremental beneﬁt from predicting
earnings changes concurrently with forecast errors is small relative to a model that predicts only forecast
errors. The results hold while controlling for various risk factors and known anomalies.
Keywords: earnings thresholds, earnings forecasts, trading strategy
JEL Classiﬁcation: G11, M41
Academic research has shown that ﬁnancial statement analysis can be useful in
predicting future earnings changes, which in turn, can yield abnormal proﬁts.
Burgstahler and Dichev (1997) and Degeorge et al. (1999) provide evidence sug-
gesting that a positive earnings change is an important earnings threshold that
managers strive to obtain. Degeorge et al. (1999) also conclude that meeting analysts’
forecasts is another earnings threshold important to managers.
In this study, I
extend previous research that examines trading strategies based on earnings changes
by introducing an additional earnings threshold of meeting/beating analysts’ earn-
ings forecasts. Speciﬁcally, I examine whether a trading strategy based on the pre-
diction of meeting or beating more than one earnings threshold can be incrementally
more proﬁtable than a trading strategy based on only earnings changes.
Research evidence suggests that successfully predicting whether ﬁrms will meet or
beat forecasted earnings (MBFE) can be especially proﬁtable. Bartov et al. (2002),
Kasznik and McNichols (2002), and Lopez and Rees (2002) have all documented a
signiﬁcant stock price premium (penalty) to meeting/beating (missing) analysts’
earnings forecasts after controlling for the magnitude of the forecast error. These
studies ﬁnd that within a small window surrounding a ﬁrm’s earnings announce-
ment, the stock price reaction can be substantial to an earnings surprise of only a
penny or less. As the return window expands to allow for earnings preannounce-
ments and other information entering the market, the absolute value of the stock
price premium/penalty is increased. Brown and Caylor (2005) ﬁnd that the market
reward to MBFE is greater than meeting other earnings thresholds examined by
Degeorge et al. (1999) and Burgstahler and Dichev (1997).
Besides predicting the outcome of meeting more than one earnings threshold, an
important factor that distinguishes this research from some previous studies is that
Review of Accounting Studies, 10, 465–496, 2005
Ó 2005 Springer Science+Business Media, Inc. Manufactured in The Netherlands.