Measuring Risk Transmission through Connectedness: Evidence from Stock and Foreign Exchange Markets

Lu Yu, Fordham University


This dissertation studies connectedness both on the Chinese stock market and the foreign exchange market. The networks of connectedness are constructed through generalized variance decomposition, based on forecast error variance. For the Chinese stock market, we study the connectedness of risk spillovers among publicly traded financial institutions, with 5-min high-frequency data, covering the 2010-2019 period. We analyze from two aspects: the risk spillover with realized variances, and the asymmetric risk spillover with realized semivariances. We find financial institutions cluster strongly by sectors. The systemic total connectedness rises when the market is more volatile, which surges to the peak as crisis erupted. While the financial market prevails in optimistic mood, it slumps to be dominated by the bad volatility during crisis. Both the total systemic connectedness of financial institutions and the risk spillover asymmetry shed lights on the future stock market performance. That is, a faster increase in the total systemic connectedness and higher weekly pessimistic sentiment predict a higher volatility of future stock market. Moreover, we find that actively traded smaller banks and undervalued non-banks are more important in propagating risks. In addition, larger and undervalued non-bank and non-insurance firms are more influential in propagating bad volatility and lead to more downside risks, which become even stronger during crisis. For the foreign exchange market, we study with 5 liquid pairs traded in 6 geographic trading segments, and various sampling frequencies. We find the New York and Japan segments are the main risk/information providers. The foreign exchange pairs of the currencies which mainly circulate during the previous trading segment tend to contribute as the largest risk givers in the current segment, which points towards heat wave type of information flow. We also find that pairs at higher sampling frequency spread more risks than those at lower frequencies, though we do not find a clear information flow between them.

Subject Area

Economics|Early childhood education

Recommended Citation

Yu, Lu, "Measuring Risk Transmission through Connectedness: Evidence from Stock and Foreign Exchange Markets" (2020). ETD Collection for Fordham University. AAI27961295.