What Drives Emerging Stock Market Returns? A Factor-Augmented VAR Approach

ABSTRACT

This paper explores the dynamic relationship between global economic factors and emerging stock returns within a factor-augmented VAR model. I find that favorable global growth and stock market shocks have significant positive effects on emerging equity returns, whereas global uncertainty and US dollar exchange rate shocks cause a substantial fall in the returns. Global oil shocks lead to a transient increase in emerging stock returns, followed by a gradual decline. Variance decomposition analysis implies that the global uncertainty shock is the most important in the short run, explaining more than 30% of the fluctuation in emerging stock returns, while the US dollar exchange rate shock becomes the most critical in the long run, explaining more than 40%. These findings have crucial implications for international investors, as well as for policymakers in emerging market economies.

KEYWORDS:

• Emerging stock returns
• factor-augmented VAR
• global economic factors

JEL:

• F36
• F65
• G15

Acknowledgments

I would like to thank the Editor (Paresh Kumar Narayan) and anonymous referees for their helpful comments and suggestions.

Notes

1. Oil demand in emerging market economies such as China and India has grown rapidly and overtaken industrialized countries for the first time since 2013, according to the International Energy Agency (IEA).

2. Bekaert and Harvey (2017) show that the world GDP share of emerging market economies has dramatically increased, accounting for only 16% in 1987 and reaching to more than 40% in 2016. Moreover, for equity market capitalization, the emerging market represented less than 1% of the world market in 1987, but it has increased by more than a 10-fold in recent years.

3. Regional economic factors and contagion effects between emerging stock markets may also influence emerging equity returns. This paper, however, focuses on global economic factors not only because they are common factors, but also because the main goal of this study is to examine which global factors are important in accounting for variations in emerging stock returns.

4. There are opposing arguments by Sardorsky (2000), Zhang et al. (2008), and Akram (2009) showing that US dollars affect oil prices because a weaker dollar induces international buyers of oil to pay more dollars for oil. In the Robustness Section, I consider alternative ordering restrictions in the FAVAR to check the sensitivity of the main empirical results and find that they remain intact.

5. Bernanke, Boivin, and Eliasz (2005)show that both two-step and one-step procedures generate very similar results, indicating no great benefit from the more computationally intensive Bayesian methods.

6. It would be possible to employ a dynamic factor model suggested Kose, Otrok, and Whileman (2003), but their method is especially useful to estimate various common factors simultaneously. The global factors extracted with the dynamic factor model produce quite similar results obtained by the first principal component.

7. In the Robustness Section, I apply an alternative estimation method of local linear projections developed by Jordà (2005) to check the sensitivity of the model misspecification.

8. Two different criteria of Akaike information and Schwarz-Bayesian Information indicate much shorter lag length, choosing p = 4 and p = 3, respectively. In the Robustness Section, I will consider different values of lag length in the VAR model and find that the key empirical results remain unchanged.

9. The EPU index reflects media coverage containing the term of uncertainty and items related to economy and economic policies, capturing the near and longer term economic concerns. The index can be easily downloaded from http://www.policyuncertainty.com.

10. Another interpretation of the short-run positive reaction of emerging stock returns can be attributed to a low short-run price elasticity of oil demand that raises revenues for oil-exporting countries with the increased oil price as described in Hamilton (2009)

11. The capital flow data captures the net portfolio purchases of nonresidents from residents, which is provided by Mandalinci and Mumtaz (2019) in a quarterly frequency and covers the period 2001:Q1–2014:Q3. Nominal exchange rates for each emerging country can be obtained from the Bloomberg.

12. The short-and long-term interest rates are available at OECD database and the 10-year Treasury yield of the US comes from the FRED. The sample periods for the short-term rates and long-term rate spreads cover 2000:1–2018:12 and 2003:1–2018:12, respectively, which are determined by data availability.

13. Jordà (2005) proves that impulse response functions obtained by local projection methods are consistent and asymptotically normal, as well as more accurate than standard methods.

14. To extract the unobserved common component for global short-term interest rates, I use the short-term interest rates available at OECD database and policy rates for developed countries in Table 1. Following Basher, Haug, and Sadorsky (2012), I order the global short-term rate after global growth, global oil price, global uncertainty variables, before the US dollar exchange rate and global stock return variables.