Market Anomalies and Abnormal Returns

As mentioned in prior studies, there is not one accepted model that presents a measure for expected return by withstanding a certain level of risk. Now a new approach has been provided by Penman and Zhou (2014) that the accounting variables are linked to the expected return. In this approach if the market assesses the earnings along with the risk, then the variables that predict earnings and earnings growth, will also be able to predict the return. They argue that if it is proven that the aforementioned factors predict future earnings and earnings growth in the same direction that predicted future returns, it will be proven that the return predicted by these variables is not abnormal.
Research Hypotheses: The main question of this research is to investigate whether the variables that predict future earnings or growth in future earnings can predict future returns and future realized returns in the same manner. This article is based on sign of predictive variables in predicting future earnings and comparison with the predicted future returns. Accordingly, the following four hypotheses were explained and tested:Hypothesis 1: The variables that predict future earnings are able to predict future returns in the same direction.
Hypothesis 2: The variables that predict future earnings growth are able to predict future returns in the same direction.
Hypothesis 3: The variables that predict future earnings are able to predict future realized returns in the same direction.
Hypothesis 4: The variables that predict future earnings growth are able to predict future realized returns in the same direction.
The time period used in this study is the 11-year period between 2003 and 2013 and last year for estimating models is 2011. The population of this research consists of the companies listed in the Tehran Stock Exchange. A sample was selected from this population using screening method. Four models were specified for predicting each of the four indexes (future earnings, future earnings growth, future return and future realized return) and were estimated using pool and panel data approach. The variables in these four models that were indexes of anomaly variables include: Working capital accruals, Change in net operating assets, ROA, Investments, Net shares issuance, External Financing and Momentum. Then the probability and the sign of these variables in these four models were compared mutually and each of the hypotheses were investigated using these signs.
Results showed that four variables (Accruals, External Financing, Momentum and ROA) which are able to forecast the future earnings are also able to forecast future return in the same direction. Therefore that return is predictable and referred to as "abnormal" is not suitable for them. But that return would be "required return" according to Penman and Zhou (2014). That means the return of which is expected to occur. For this reason, it is clear that this return is consistent with the hypothesis of rational expectations.
Discussion and According to our results, researchers and investors should note that market inefficiency is detectable not only through the accounting variables but also through investigating whether these returns are consistent with the hypothesis of rational expectations. Consistency of returns with the assumption of rational expectations is an important condition which is usually ignored in this field.