A Dissertation
by
ERIC KYLE KELLEY
Submitted to the Office of Graduate Studies of
Texas A&M University
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
August 2004
Major Subject: Finance
ABSTRACT
The finding of reversals in weekly returns has been attributed to a combination of microstructure issues and overreaction to information. I provide new evidence eliminating overreaction as a source of reversal. I show that well-known weekly contrarian profits are followed by a long run of momentum profits. In fact, these profits are strong enough to produce a significant momentum effect over the full year following portfolio formation. Thus, the market does not appear to view extreme weekly returns as excessive, as implied by an overreaction story. To the contrary, this return continuation is consistent with underreaction to the news driving extreme weekly returns. This is supported by cross-sectional tests in which I find this week’s news is positively related to next week’s returns. The evidence presented here is consistent with growing evidence that underreaction to firm-specific information is a pervasive feature of price formation. Therefore, if any short-run contrarian profits can be realized, they are better viewed as compensation for providing liquidity than as a reward for arbitrage.
INTRODUCTION
Short-run individual stock returns reverse immediately. Lehmann (1990) finds that contrarian strategies which buy stocks with low weekly returns and sell stocks with high weekly returns generate positive profits over the following week. Jegadeesh (1990) shows that a one-month contrarian strategy is also profitable. This reversal pattern has spawned much discussion over the past decade. Is it due to time variation in risk premia? Is it due to market inefficiencies? Is it spurious? Lehmann highlights that return predictability over such a short horizon cannot plausibly be attributed to time variation in risk premia.1 That leaves the debate to focus on spuriousness or market inefficiencies as the sources of short-horizon predictability.
Bid-ask bounce and nonsynchronous trading are sources of spurious reversals in returns. Kaul and Nimalendran (1990) and Conrad, Kaul, and Nimalendran (1991) show that part of return reversal is due to bid-ask bounce. Lo and MacKinlay (1990) and Boudoukh, Richardson, and Whitelaw (1994) note that nonsynchronous trading contributes to contrarian profits.2 Lead-lag effects and market-maker inventory control are forms of market inefficiencies that can lead to return reversal, but these are not inefficiencies due to cognitive biases. Lo and MacKinlay (1990) provide evidence that that lead-lag effects explain more than half of the contrarian profits of their strategy.
This could reflect delayed reactions of some stocks to a common factor. Jegadeesh and Titman (1995b) observe that market-makers set prices in part to control their inventory, which also induces a return reversal.3 Empirical researchers, however, have had great difficulty in establishing the above sources as the sole drivers of weekly reversals. Controlling for these sources of reversals in various ways, Jegadeesh and Titman (1995a), Cooper (1999), and Subrahmanyam (2003) conclude that contrarian profits are largely due to overreactions to firm-specific news.
With this short-run debate ensuing, other researchers have independently documented persistence in returns after many corporate events, such as unexpected earnings, dividend changes, stock repurchases, stock splits, and seasoned equity offerings, as well as after headline news and cash-flow news.4 Short-run reversals and the conclusion that the market is overreacting are curious in light of the evidence of such perceived underreaction to firm-specific news. In this dissertation, I attempt to reconcile this conflict.
The duration of the return reversal found by Lehmann (1990) and Jegadeesh (1990) is less than four weeks, while the continuations following firm-specific news last up to a year. Therefore, I begin my analysis by extending the holding period of portfolios similar to those used in the short-run literature. Suppose the weekly return reversal is due solely to microstructure issues and not in any way to an intrinsic overreaction to firm-specific news. Suppose also that, as the news literature suggests, there generally is an underreaction to firm-specific news. Since the microstructure effects should dissipate in a few weeks, we should expect to see a continuation in returns once the microstructure issues fade if the market is actually underreacting to news in the formation week.
Consistent with prior studies, I find that a strategy that buys weekly extreme winners and sells weekly extreme losers generates negative profits in the first four weeks of the holding period (one can simply reverse the sign to get contrarian profits). The profits to these weekly portfolios behave quite differently, however, after four weeks.
My key finding is that extreme weekly returns in fact persist for roughly a year once the brief return reversal dissipates. This continuation in returns is robust and steady across the subsequent weeks. Moreover, the continuation easily offsets the brief reversal that follows portfolio formation. In other words, the fifty-two-week post-formation period displays no evidence of a correction. In sum, my findings are inconsistent with an overreaction to news occurring in the formation period.
Additionally, I not only find that the subsequent continuation in returns offsets the initial reversal; I find that the continuation dominates. Across the fifty-two weeks following portfolio formation, the winners continue to outperform the losers. So I find evidence of underreaction. This finding is consistent with the evidence of continuation in returns following firm-specific news, mentioned above. This consistency is remarkable given that the extant literature paints a complexity of reversal and momentum patterns, with reversal in the short-run, momentum in the intermediate-run, and reversal again in the long-run.
This new evidence provides some consistency across the short-run predictability and post-event drift literatures. However, it is still interesting that the event literature rarely documents any immediate reversal. In fact, there is direct evidence of no significant reversal immediately following six of the eight events Daniel, Hirshleifer, and Subrahmanyam (1998) associate with post-event drift. Appendix I provides a list of studies documenting such evidence. In light of this, I consider a subset of the main strategy with earnings announcements during the formation week. These stocks experience both an extreme return and a specific news announcement. After controlling for bid-ask bounce, I find (i) no immediate reversal and (ii) strong return continuation for a full year. Thus, it appears that in this subset strategy the strength of the news offsets any pressure to reverse even in week one.
Pursuing this result further and in a more general context, I find that the initial reversal in returns is not attributable to abnormal firm-specific residual returns (standardized residuals), which I interpret as firm-specific news. Instead, the reversals are strongly related to total returns, consistent with the reversals being due only to microstructure issues. Holding last week’s returns constant, I find a positive relation between last week’s news (standardized residual) and this week’s return. Hence, realizable profits to weekly contrarian strategies, if any, are better viewed as compensation for providing liquidity to the market, rather than as a reward for arbitrage. While some researchers have challenged the overreaction interpretation of weekly reversals, they have been unable to show that microstructure effects can fully explain contrarian profits. Estimating the microstructure effects though is very difficult to do. I contribute to the literature by, in a sense, stepping back and making an observation using a longer window. This approach addresses the source of reversals without explicitly measuring various microstructure effects. The performance of the extreme-weekly return stocks over a long window refutes the overreaction hypothesis.
This dissertation also contributes to a growing body of research suggesting return persistence is a pervasive feature of the price formation process. In concert with the momentum and post event drift literatures, Vuolteenaho (2002) and Chan (2003) provide general evidence of a delayed response to news identified over a monthly horizon. Gutierrez and Pirinsky (2004) show large news shocks measured over periods as short as a month and as long as three years is associated with drift that never reverses. My findings round out this picture of return persistence. Patterns of return predictability across time horizons and events are consistent with each other and perhaps much simpler than originally perceived. The only effect of firm-specific news on future returns is that of continuation.
The remainder of this dissertation is organized as follows. Section II contains a detailed literature review, primarily focusing on return-based predictability. Section III describes the data and portfolio construction. Section IV presents empirical results and robustness tests. Section V concludes.
No comments:
Post a Comment