Research on the U.S. equity market indicates that “old” or intermediate momentum (12 months ago to 7 months ago) is much more important than “new” or recent momentum (6 months ago to two months ago, incorporating a skip-month to avoid short-term reversal) in predicting future stock returns. Do other equity markets confirm this finding? In their September 2011 preliminary paper entitled “Is Momentum an Echo?”, Amit Goyal and Sunil Wahal investigate whether other country equity markets behave similarly. Using regressions, single-sorts on past stock returns and double-sorts on intermediate and recent past stock returns, along with country-specific risk factors (market, size, book-to-market), for 36 non-U.S. country equity markets during 1991 through 2009, they find that:
- Regression tests across country markets over the sample period indicate that:
- The value premium (based on book-to-market ratio) is consistently strong.
- The size effect is relatively weak.
- Prior-month return relates negatively to next-month return (short-term reversal).
- In markets with strong momentum effects, both intermediate and recent returns relate positively to next-month return, but the strength of the two relationships is not statistically different in 35 of 36 markets. In the one market for which the difference is significant (UK), recent past returns are more predictive than intermediate past returns.
- For value-weighted winner-minus-loser single-sort momentum portfolios:
- None of 16 emerging markets exhibit a reliable difference in future returns between portfolios formed on “old” or “new” momentum.
- Two of 20 developed markets (UK and Sweden) exhibit differences, but in both cases recent past returns are more effective than intermediate past returns.
- For value-weighted intermediate winner-minus-loser and recent winner-minus-loser double-sort momentum portfolios:
- In most countries, spreads between extreme double-sorted portfolios based on intermediate versus recent momentum are small and highly volatile.
- In places where the spreads are large, they generally indicate that “new” momentum is more important than “old” momentum. For example, the average monthly gross return of the UK winner-minus-loser portfolio based on intermediate (recent) momentum among stocks in the recent (intermediate) loser portfolio is 0.76% (1.38%).
- Randomness is a possible explanation for the unusual U.S. results in the original research. If differences between intermediate and recent momentum portfolio returns are normally distributed across country markets with zero mean and some variance, one or two false positives are very likely.
In summary, evidence from multiple tests applied to monthly returns for 36 non-U.S. equity markets does not support a general belief that “old” momentum is more important than “new” momentum in predicting future stock returns.
Cautions regarding findings include:
- Returns presented are gross, not net, and monthly momentum strategies tend to have high turnover. Including reasonable estimated trading frictions would materially reduce these returns.
- Moreover, trading frictions may vary considerably across markets and types of stocks, such that findings based on estimated net returns may differ from those based on gross returns.
- Statistical significance tests assume well-behaved distributions. Wildness in distributions would undermine validity of the tests.
