Empirical analysis of politically-themed stocks using text mining techniques and entropy-based network dynamics – focus on the Republic of Korea's case

Abstract

Politically-themed stocks refer mainly to stocks that benefit from pledges of politicians and policies of political parties. This study gave the empirical analysis of the politically-themed stocks in the Republic of Korea and constructed politically-themed stock networks based on the Republic of Korea’s politically-themed stocks derived mainly from politicians. To select politically-themed stocks, we calculated the political sentiment index for 12 politicians using text mining techniques and search analytics and selected stocks belonging to the politically-themed stock candidates. We then determined politically-themed stocks based on the statistically significant causal relationship from the rate of change of politician sentiment index to the excess returns of the politically-themed stock candidates derived from asset pricing models. To examine correlations and causal relationships between politically-themed stocks, we constructed politically-themed stock networks using entropy-based approaches, normalized mutual information, and effective transfer entropy. After that, we analyzed the correlations and causal relationships between selected politically-themed stocks using the normalized mutual information volume and efficient transfer entropy, which are entropy-based measures, and constructed politically-themed stock networks. We determined whether and how large the correlations and causal relationships occurred between the politically-themed stocks in the stock market during the study period by analyzing the established networks. We empirically analyzed politically-themed stocks before-and-after real-world situations from the schematized network, focusing on dynamic changes of politically-themed stock networks. We verified that politically-themed stocks are significantly affected by external information flows about politicians based on the results. Moreover, there exist nonlinear correlations and causal relationships between politically-themed stocks. Besides, we confirmed that when political events are closer, the correlations between excess returns of politically-themed stocks were strengthened, and the causal relationships were weakened. These trends are eased after political events. Furthermore, we developed the investment strategy using the politician sentiment index, politically-themed stock networks, and portfolio optimization methods based on the empirical analysis results. As a result, we confirmed that this strategy could benchmark the Republic of Korea's major market indices.