The effect of political volatility on capital markets in EU accession and neighborhood countries

Abstract

Political instability has the potential to disrupt financial markets. But how do political institutions affect financial movements in an environment where all institutions are in flux? This paper looks at the effects of formal and informal political volatility in the new EU countries of central and eastern Europe, in the Eastern Neighborhood, and farther afield in Central Asia to answer this question. Using asymmetric GARCH modeling on monthly data, I find that informal political volatility has a significant negative effect on stock returns, while formal political institutions generate much higher financial volatility than changes in monetary policy.

Keywords:

• volatility
• political institutions
• elections
• terrorism
• EU
• stock markets

JEL Classifications:

• G20
• O43
• P30

Acknowledgements

The author wishes to thank Sander M. Ijmker for his excellent research assistance, Grzegorz Poniatowski for his suggestions, Ali Kutan and participants at the 7th annual Conference on “Economic Challenges in Enlarged Europe” in Tallinn in June 2015 for their comments, and the organizers and attendees of the Portsmouth-Fordham Conference on Banking and Finance in September 2016 (especially Fotios Pasiouras, Mohammad Hasan, and Renatas Kizys) for their insights. I also wish to thank two anonymous referees and the Editor in Chief for their excellent comments.

Notes

1. A further benefit to the GJR-GARCH model is how EGARCH models the conditional variance. EGARCH places extra weight on more recent observations, making each shock that enters the system additive. However, in the case of institutional changes, which may amplify over time, the volatility shock is more likely to be multiplicative, a feature that the GJR-GARCH model captures (Duan 1997).

2. GARCH was chosen over system-GMM as the high-frequency nature of the data, serial correlation, and its sheer size made GMM inappropriate.

3. The countries included in the data-set are Belarus, Bosnia, Bulgaria, Croatia, the Czech Republic, Estonia, Hungary, Kazakhstan, the Kyrgyz Republic, Latvia, Lithuania, Macedonia, Mongolia, Poland, Romania, Russia, Serbia, Slovakia, Slovenia, and Ukraine.

4. ARCH, GARCH, and AR terms are all significant at the 1% level, but are not shown in the interest of space. The leverage effect of the GJR-GARCH model also shows that it is a more “correct” model of political volatility.

5. A helpful reviewer noted that there might be multicollinearity from GDP growth, as M2 is already included in the base model. However, given that we are using growth metrics (growth of GDP vs. growth of M2) or volatility metrics (growth of GDP vs. volatility of M2 growth), this is less of a theoretical problem. Econometrically, there is only a slight positive correlation between the two growth variables (0.22) over the full data-set, and even less correlation between GDP growth and the volatility of M2 growth (−0.16).

6. In most cases, these models also have the best log likelihood results as well.

7. Additional robustness tests on the type of party in power are shown in the online supplementary material.

8. Unfortunately, it is not possible to disentangle how much of Poland’s policy uncertainty is captured by the Europe variable and what remains at a home-grown level.