#14 – Momentum Effect in Stocks

Authors: Goetzmann, Huang

Title: Momentum in Imperial Russia

Link: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2663482

Abstract:

Some of the leading theories of momentum have different empirical predictions about its profitability conditional on market composition and structure. The overconfidence explanation provided by Daniel, Hirshleifer, and Subrahmanyam (1998), for example, predicts lower momentum profits in markets with more sophisticated investors. The information-based theory of Hong and Stein (1999) predicts lower momentum profits in markets with lower informational frictions. The institutional theory of Vayanos and Woolley (2013) predicts lower momentum profits in markets with less agency. In this paper we use a dataset from a major 19th century equity market to test these predictions. Over this period there was no evidence of delegated management in Imperial Russia. A regulatory change in 1893 made speculating on the St. Petersburg stock market more accessible to small investors. We find a momentum effect that is similar in magnitude to those in modern markets, and stronger during the post-1893 period than during the pre-1893 period, consistent with the overconfidence theory of momentum.

Notable quotations from the academic research paper:

"Momentum strategies, also known as relative strength strategies, are a class of long-short trading strategies that buy past winners and sell past losers. Jegadeesh and Titman (1993) first document the profitability of momentum strategies in a sample of U.S. stocks during the period from 1965 to 1989. Many other studies extend this initial finding by documenting momentum profitability in the post-1989 period, in equity markets outside of the U.S., and in other asset classes While there is an abundance of empirical evidence for momentum, the debate on its underlying mechanism remains unsettled. For instance, Daniel, Hirshleifer, and Subrahmanyam (1998) propose a model in which investor overcondence about the precision of private information generates the momentum effect. On the other hand, in Hong and Stein's (1999) model, the interaction of boundedly rational agents and the slow diffusion of information generate initial underreaction and subsequent overreaction. More recently, Lou (2012) and Vayanos and Woolley (2013) propose explanations of momentum driven by flows into and between institutional money managers.

While these theories all generate the momentum effect, they have different predictions about momentum profitability conditional of market composition and structure. Tests of these predictions require significant variation for identification. In this paper we use a comprehensive dataset of monthly stock returns from the St. Petersburg Stock Exchange in the 19th and early 20th centuries to test these predictions. The St. Petersburg Stock Exchange provides an ideal setting for investigating momentum because: 1) it is, as yet, an untouched sample for finance researchers; 2) there was no evidence of a delegated management structure in the Russian Empire over this period; and 3) a regulatory change in 1893 substantially reduced the costs of speculative trading for less sophisticated investors. The institutional theory predicts a muted momentum effect in such a market as the institutions that the theory relies on to generate momentum were absent. The overcondence theory predicts lower momentum profits during the pre-1893 period than during the post-1893 period, because in the later period there was more market participation by those who are more susceptible to being overconfident. In contrast, the information diffusion theory predicts higher momentum profits during the pre-1893 period, because market participation was narrower and information flow was slower.

Our results are consistent with the investor overcondence theory. Despite the absence of a delegated management structure in our setting, we find that past medium-term winners outperform past medium-term losers by as much as 74 basis points per month, which is similar in magnitude to momentum prots in modern markets. Exposure to the market factor cannot explain this outperformance. In addition, we find that the momentum effect is small and insignificant during the pre-1893 period, but large and highly significant during the post-1893 period. A placebo test using momentum returns from the London Stock Exchange shows that the same empirical regularity is not observed in a market that did not undergo a similar regulatory change.

Daniel and Moskowitz (2013) document that the momentum trade occasionally experiences large crashes. This suggests that, whatever the mechanism, momentum profits could compensate for an infrequently occurring risk factor. Based on this idea, they propose a method to manage this extreme left-tail risk.Interestingly, in our Russian sample extending more than 40 years, we find that while momentum returns are somewhat negatively skewed, extraordinary crashes like those that occurred in the U.S. are absent."

Are you looking for more strategies to read about? Check http://quantpedia.com/Screener

Do you want to see performance of trading systems we described? Check http://quantpedia.com/Chart/Performance

Do you want to know more about us? Check http://quantpedia.com/Home/About