Does Geopolitical Risk Matter? Evidence from South Korea

ABSTRACT

This study explores the impact of geopolitical events on the stock return behavior of inter-Korean economic cooperation-related firms depending on the North Korean regime. We document empirical evidence showing that cross-sectional stock return tends to react positively to positive geopolitical events under the current regime in North Korea (Kim Jong-un), whereas negative geopolitical events have limited impact. Conversely, we find that negative geopolitical events yielded more pronounced effects on the stock returns of related firms under the former regime (Kim Jong-il). In addition, this study investigates the role of geopolitical shock in the evolution of aggregate economic variables of South Korea using Caldara and Iacoviello’s (2018) geopolitical risk index. Geopolitical shock is found to yield no statistically meaningful impact on stock price index, industrial output, employment, or gross trade volume. Furthermore, aggregate stock market variables are found to be immune to geopolitical shock in South Korea. These results indicate that market participants estimate the escalation of geopolitical risk into full-scale war as unlikely.

KEYWORDS:

• Geopolitical risk
• North Korea
• stock return
• inter-Korean economic cooperation

Disclosure Statement

No potential conflict of interest was reported by the author.

Notes

1. ‘Trinity of uncertainty’ refers to economic uncertainty, policy uncertainty, and geopolitical uncertainty.

2. The study by Kim, Kang, and Lee (2018) is a closely related study that discusses the tone of the media in English-based newspapers – especially U.K. news media, which can predict North Korea’s future military aggressions.

3. A similar approach can be found in Pyo and Kim (2019); they demonstrate the existence of links between a sentiment index derived from newspaper articles and asset prices in South Korea.

4. The GPR has also been utilized to assess the impact of geopolitical risk in other contexts. For instance, Gupta et al. (2019) have shown that geopolitical risk measured by the GPR negatively affects trade flow between countries using a gravity model. On the other hand, Bilgin, Gozgor, and Karabulut (2018) have shown that government spending is positively affected by geopolitical risk, which indicates that a government has incentive to mitigate the increased risk by providing compensation in the form of government spending.

5. Barro (2006) argued that the equity risk premium puzzle can be explained to some extent by macroeconomic disasters. Specifically, he considered all historical disaster-equivalent events and used them for structural modeling, from which the magnitude and probability of loss in the event of a macroeconomic disaster were inferred.

6. In the literature, ‘disaster’ usually refers to macroeconomic disasters owing to which GDP per capita falls by more than 15%.

7. Variables such as the GPR are by nature less likely to be affected by the current economic situation compared to other uncertainty indices (e.g., Baker 2016). Thus, in estimating the effects of geopolitical risk, reverse-causality or endogeneity problems can be avoided; as major economic indicators deteriorate (improve), the uncertainty index tends to increase (decrease). However, there may be cases where geopolitical risk changes as economic conditions change, but the likelihood is low compared to other uncertainty indices. This feature of the GPR makes it a good candidate for estimating the direct effects of geopolitical risk even with reduced-form empirical models.

8. See Blomberg, Hess, and Orphanides (2004), Tavares (1988), Glick and Taylor (2010), Moretti, Steinwender, and Reenen (2014), and Aghion et al. (2018) for the economic effects of war and terrorism.

9. The Korean Peninsula has gone through repeated conflicts and reconciliations throughout the post-Korean War, but a full-scale war never occurred thereafter. Of course, it would be reasonable to expect that the level of geopolitical risk perceived by market participants can be affected even by a minor local military conflict.

10. In Appendix A, we provide results-as a robustness check-based on the alternative sample specification in terms of the composition of firms and geopolitical events.

11. Calculation of abnormal return (i.e. actual return minus normal return) must be preceded by defining the normal return of a stock. In asset pricing literature, the normal return (or equilibrium return, in terms of the risk-reward relationship, used interchangeably) of a given stock is usually regarded as the return implied by various asset pricing models, such as capital asset pricing and various multi-factor models (e.g. Fama and French (2015) and their various extensions). Since the behavior of abnormal returns is largely influenced by the choice of asset pricing model used, we choose to use the excess return or raw return of a given stock to prevent model selection errors, by which we can investigate the overall picture of post-event stock return movements.

12. Asymmetry between the number of positive events and the number of negative events is due to the nature of the inter-Korean relationship in recent years.

13. In Appendix A, we expand the event sample to include all minor events and conduct the same exercise as a robustness test.

14. Though North Korea’s former leader Kim Jong-il died on 17 December 2011, Kim Jong-un did not assume full power immediately following his father’s death. Therefore, in this paper, we assume Kim Jong-un’s regime started at the beginning of 2012.

15. In particular, they count articles that contain geopolitical risk-related terms such as ‘nuclear war,’ ‘war risk,’ and ‘terrorist attack,’ and normalize by dividing the total with the total number of articles published during the prespecified period.

16. Caldara and Iacoviello (2018) provide an individual country’s GPR. See https://www.matteoiacoviello.com/gpr.htm.

17. The KOSPI 200 index consists of 200 large-cap stocks that capture overall stock market performance well; it accounts for about 80% of total market capitalization of the KOSPI stock market.

18. However, it should be noted that when considering longer horizons after Day 1, the difference between the two groups quickly disappears.

19. It should be noted that all responses to the geopolitical shock are not statistically significant as a 95% confidence interval of functions contains a zero.

20. We estimate the structural VAR in a similar context with the change in the composition of endogenous variables. The endogenous variables are specified as follows:

$y=\left[GPR,r,vol\right]$

where GPR is the GPR, r denotes a stock return, and vol denotes the trading volume. We also included a dummy variable for the month of January to capture the January effect.

21. In terms of a theoretical game-like setting, full-scale war against the U.S. and South Korea would be the worst scenario for the North Korean regime, having observed the devastating consequences demonstrated through the elimination of previous leaders of Iraq and Al-Qaeda by U.S. forces. However, we cannot completely rule out the possibility of irrational thinking by the North Korean regime, which can be seen as a black swan event in the probability of war.

22. Stock return volatility tends to increase more with bad news than with good news. However, the negativity effect in stock returns is not universal; for example, Chen, Noronha., and Singal (2004) documented empirical evidence of the ‘positivity effect’ in stock returns.

23. For details, refer to Table B1 and Table B2.

Additional information

Funding

This research was supported by Changwon National University in 2019 ~ 2020.