Analyzing the possibility of generating excess returns using event based trades

Abstract

Post earnings announcement drift (PEAD) refers to the phenomenon of a drift in stock prices post an announcement provided there is a surprise in the earnings, for up to a period of over 2 months from the announcement date. The said phenomenon is conjectured to generate returns well over market and is said to provide an opportunity to generate profitable trades. Alternative explanations of the phenomenon rely on deviations from the Efficient Markets Hypothesis (EMH), presence of transaction costs or behavioural factors such as disposition effect. The area has attracted substantial research, with Ball and Brown’s 1968 paper paving the way for a body of literature on the existence as well as the causes of PEAD. The scope of this study is restricted to ascertaining if the phenomenon of PEAD is seen in the Indian equity markets and to identifying strategies to make excess profits provided it were to exist. The study has not attempted to identify possible causes of PEAD. It merely attempts to study relationships and generate rules for exploiting the phenomenon, if any. The study was broadly divided into three aspects: a) literature review and design of research methodology, b) data collection, and c) analysis and compilation of results. The first phase resulted in the identification of measures such as % surprise and SUE multiples that capture surprise. Further, alternative strategies to quantify surprises were arrived at – surprises could either be categorized into a) buckets based on percentiles – such as decile or quintile buckets; or buckets based on absolute magnitude of surprise (for ex: surprise between 60 and 80% could fall in one bucket). Each of these alternate quantification strategies is examined for explanatory power of PEAD. In the second phase, a set of companies and time horizon of earnings announcements over which the analysis was carried out was identified. The study was restricted to about 270 companies that had exchange traded futures – this is primarily to exploit any short trading strategies formulated in lieu of the findings of this research. Further, earnings data was restricted to the last 5 years alone – first to adjust for a macroeconomic or business cycle effects, second to adjust for recency of data (and its consequent effect on earnings patterns – data in the recent past will have higher autocorrelation with earnings in the near future) and thirdly, due to a lack of analyst estimates of earnings for periods before that. We primarily looked at Net Income as a proxy for earnings, simply because that is what accrues to shareholders in terms of returns. Further, research by Ball and Brown(1968) has validated Net Income over other measures of earnings. A maximum time window of 2.5 months from the date of the announcement was the chosen as the window of examination for this study. Further, the returns of the stocks in this study are not adjusted for market premium (beta) – as accurate measures of these are not available for the time period under consideration. Comparison with beta adjusted portfolios is beyond the scope of this study. In general, the results of this study point to the presence of phenomenon in the Indian markets. Significant price drifts were observed for periods well into the third month from the announcement, provide there was an earnings surprise to start with. Among the measures employed, excess returns over market appeared to be most appropriate as it strips the impact of market movements and helps examine PEAD as a standalone phenomenon. SUE as a metric of earnings surprise, with bucketing based on absolute values of SUE, emerged as the winner with an R2 of 0.86 and a sensitivity of 0.63 to percentage price drift over a 2.5 month period. The reasons for SUE as the measure of choice are examined both in literature review and in the section capturing the results of our analysis. A portfolio weighted by the magnitude of earnings surprise has been constructed to compare trading strategies and study average weekly returns during the 2.5 month period. Once a case was established for the presence of PEAD for the universe of stocks considered in this study, we further evaluated alternative trading strategies with upside potential significantly higher than a naïve strategy – where a naïve strategy is driven merely by quantum of surprise, whereas advanced strategies take into account other relevant information in addition to surprise like price performance in the week preceding to the event, a complete reversal between market expectations and actual performance etc. The results of these strategies do not appear to be significantly better in comparison the naïve strategy. In future studies, follow up studies could be conducted with the universe of stocks widened and the size effects can be used to see how companies with varying market capitalisation exhibit PEAD. Also, the study could be extended to other emerging markets such as Indonesia, Brazil etc. Of particular relevance could be the construction of alternative portfolios – either an equally weighted portfolio of stocks, market weighted portfolio or a surprise weighted portfolio and the examination of their performance. In summary, PEAD appears to be present in Indian markets. Further study is required to validate these findings and enlarge on the body of work generated herein.