A Study on the Firm-specific Component of Stock Returns

Abstract

This thesis investigates the firm-specific component on stock returns in Korean stock market. In the first chapter, I analyze the relation between absolute firm-specific component on stock returns and future stock returns, using all non-financial firms in Korean stock market for the period from 1987 to 2011. I find that portfolios with low absolute firm-specific component outperform those with high absolute firm-specific return over the next period. This result is always significant regardless of formation period and holding period of portfolio. Furthermore, the negative relationship between absolute firm-specific component and future stock returns is robust to controlling for firm size, book-to-market ratio, systematic risk, idiosyncratic volatility, and trading volume. Finally, to examine the possibility that this result is induced from specific short period, I divide the sample into 5-year subperiod and analyze the performance. Except the first 5-year subperiod, the stocks in lower absolute-firm-specific-component-based portfolio outperform in every subperiod, and this relation is increasingly stronger in recent period than in the past. Considering the first chapter that shows the negative relationship between firm-specific component on stock returns and future stock returns, this result that absolute firm-specific component has the predictability of future stock price can be caused from the asymmetric effect of positive and negative firm-specific component on future return. To reveal the cause of the negative relationship between absolute firm-specific component on stock returns and future stock returns, I investigate another test using not absolute firm-specific component but firm-specific component. In the second chapter, I examine the relation between firm-specific component on stock returns and future stock returns, using all non-financial firms for the period from January 1987 to July 2014. I find that contrarian strategy using weekly firm-specific component is more powerful than general contrarian strategy using weekly total return. Specifically, contrarian profit of firm-specific component is higher, more significant and steadier than that of total return. This result is neither due to size or idiosyncratic volatility effect nor due to very high profit in special periods. Furthermore, all contrarian portfolios based on past firm-specific component generate significantly and continuously positive profits regardless of whether they belong to total-return-based contrarian portfolios, while stocks both in total-return-based and not in firm-specific-component-based contrarian portfolios cannot earn significant positive profits. In long-term period, stocks only in firm-specific-component-based contrarian portfolios outperform other stocks during 3 years, however stocks only in extreme deciles of past total-return earn negative and low profits during same years. I decompose the contrarian profit based on firm-specific component applying Lo and MacKinlay (1990). As a result, the profit is attributed to negative autocovariances in individual firm-specific component rather than positive cross-serial covariances across securities firm-specific component. Further decomposing Lo and MacKinlays decomposition into each winners and losers and into between them to calculate their own and mutual relation in auto- and cross-serial covariances, I reveal that winners are strongly negatively autocorrelated although losers are positively autocorrelated and cross-serial correlated. Therefore, when I consider that size of winners is bigger than that of other portfolios, it is the main source of contrarian profit based on firm-specific component that investors overreact to big stocks good firm-specific news. And cross-serial covariances within losers are another source.