Estimation and Analysis of Multivariate Jump Diffusion Models with Jump Clustering and Contagion Effects

Abstract

Extreme negative equity return across markets during financial crisis is one of the main concerned topics in the financial world. When such unexpected events occur, the adverse effect tends to trigger further losses and spreads over to other markets. This research develops a novel financial model that tries to capture these jump clustering and contagion effects. The model is developed to handle a large number of assets. Based on empirical evidence on developed and emerging markets from weekly data since 1999 to 2014, we find that jumps in developed markets have negative means whereas the means of jumps in emerging markets are mixed. Arrivals of jumps in both markets slightly increase their jump intensities. The jumps from the developed markets cause a higher contagion effect to the emerging markets than jumps from the emerging markets do to the developed markets although stochastic jump intensities are not statistically significant. The developed markets are more integrated among themselves whereas correlation among the emerging markets are low. Based on daily data of the United States and Latin America markets from 1999 to 2015, we find the evidences on jump clustering and jump contagion effects along with their jump asymmetries. The United States market reacts more to unexpected movements in the Latin America market than the Latin America does to the United States market. The model provides the distribution of the jump intensities given the past information on returns, which can be useful for applications in risk management and asset allocation.