Trading networks in Korean financial markets

Abstract

In spite of tremendous research on the financial traders and markets, the connectedness of traders in financial markets have not been investigated by previous literature. However, the information of traders’ inter-connectedness across the financial markets can be the key to understand how the financial market function and how the market volatility changes. In this thesis, the traders’ connectedness across five different Korean financial markets is investigated with network theory.

In first paper, I investigates financial traders’ network structures across different capital markets. The influential traders and markets are found within the network based on the network structures estimated with Granger causality and generalized variance decomposition. I also find strong connections between traders and particular conditions. Then, the contributions of traders’ connectedness measures to the financial market volatility are examined. The result shows that traders’ influence is shown to be not necessarily related to the financial market volatility.

In second paper, financial traders’ network structures across different capital markets are studied with nonlinear Granger causality and nonlinear generalized variance decomposition methods. Traders with stronger influence to other traders and the conditions under which the changes of their influence occur, are found. I also investigate contributions of traders’ connectedness measures to the financial market volatility. Several influential traders’ connectedness measures contribute to the market volatility, whereas other traders’ connectedness measures do not contribute. In addition, I also find sensitive traders to the shocks of influential traders’ daily net trading volumes.

In third paper I study financial traders’ network structure in particular with the expectation on other traders’ trading on next day across different financial markets. As a proxy of the expectation, the forecasted values of traders’ trading volumes on next day with machine learning technic are applied. I estimate financial traders’ network structure utilising the expectation of a trader’s trading volume on other traders. I find influential traders such as foreign investors within the network. Then, I implement 3-phased impulse response analysis in order to capture the connections between financial market volatility, traders’ connectedness measures and traders’ actual trading volumes. The result shows the evidences that traders’ connectedness measures and their trading volume can be functioned as financial market volatility spill over channel.

Date of Award	19 Jun 2019
Original language	English
Awarding Institution	University of Bath
Supervisor	Christopher Martin (Supervisor) & Andreas Krause (Supervisor)