Predicting Equity Returns in Emerging Markets

ABSTRACT

This study investigates the relation between firm-specific attributes and future equity returns in 23 emerging markets. Equal-weighted portfolio returns reveal strong evidence of short-term momentum (rather than reversal) and medium-term return momentum. We also find evidence that market beta, book-to-market ratio and downside risk metrics predict equity returns, however, these relations get weaker once value-weighting is used. In univariate regressions, smaller firms with higher idiosyncratic volatility, lottery-like characteristics and stock-specific downside risk are associated with higher future returns, however, these relations disappear in a multivariate setting. We conclude that the most robust cross-sectional effects are short- and medium-term return momentum.

KEYWORDS:

• Tail risk
• momentum
• anomalies
• cross-section of equity returns
• emerging markets

JEL:

• G10
• G11
• G12

Notes

1. To make sure that daily non-zero equity returns are not driven by changes in exchange rates, we implement this last screen using total return indices denominated in local currencies.

2. The sample beginning year for each country is presented in Panel C of Table I of the online appendix.

3. These firm-specific attributes are collectively defined in detail in the appendix.

4. Our results are qualitatively the same if we measure lottery demand by the maximum daily return of a stock during a given month.

5. We discuss and present country-specific means and correlation matrices for these firm-specific attributes and country equity indices in Section A1 and Table I of the online appendix.

6. We also check whether any significant return differences between the extreme deciles can be explained by the international asset pricing factors of Fama and French (2017) which constructs global market, size, value, profitability and investment factors. Empirically, we regress the monthly return differences between extreme firm-specific attribute deciles on the global asset pricing factors and observe whether the intercept terms (or alphas) obtained from these regressions are statistically significant.

7. Asparouhova, Bessembinder, and Kalcheva (2010) argue that microstructure noise in security prices biases the results of empirical asset pricing specifications and suggest a methodological correction to eliminate such biases. We follow their suggestion and perform each monthly cross-sectional regression using a weighted-least squares (WLS) specification where each return is weighted by the observed gross return on the same stock in the prior period. Our results are qualitatively the same using this methodology.

8. To make these results more digestible for the readers, we provide a summary of significant results in Table II of the online appendix. This table presents the incidence of significantly positive, significantly negative and insignificant results for the subsequent portfolio and regression analyses across markets. Throughout our discussion, statistical significance is meant to indicate statistical significance at the 5% level (two-tailed).

9. Recent literature presents evidence for short-term momentum patterns. For example, Zaremba, Long, and Karathanasopoulos (2019) find evidence for short-term return momentum in equity indices, government bonds, T-bills, commodities and currencies in international markets. The takeaway from this study is that “short-term reversal may be a far less pervasive anomaly than is commonly believed, and the dominant pattern may be – surprisingly – short-term one-month momentum. For the universe of individual stocks, studies such as Avramov et al. (2017), Arnott et al. (2019) and Gupta and Kelly (2019) construct various factor-based strategies and document significant persistence at the monthly frequency.

10. The abnormal returns or alphas associated with the equal-weighted returns to the zero-cost strategies are discussed and presented in Section A2 and Table III of the online appendix.

11. The abnormal returns or alphas associated with the value-weighted returns to the zero-cost strategies are discussed and presented in Section A2 and Table IV of the online appendix.

12. We discuss and present Sharpe ratios associated with these value-weighted zero-cost strategies in Section A3 and Table V of the online appendix. We also discuss the results from some robustness tests in Section A4 of the online appendix. Specifically, we divide stocks in each country into quintiles rather than deciles and we omit the stocks whose size is smaller than the median firm size in each country-month from the sample. The results are presented in Tables VI and VII of the online appendix, respectively.

13. Although unreported, a similar reversal in signs is also observed for the lottery demand measure if MAX is included in the specifications rather than IVOL.

14. Our main analyses are carried out for each country seperately. In the online appendix, we also conduct univariate, bivariate and multivariate analyses for four distinct regions (Europe, Asia, Latin America and Africa, respectively). The results are qualitatively robust and they are discussed and presented in Section A5 and Tables IX to XI of the online appendix.