Abstract

The role of the jumps has been focused in the time series analysis of asset prices. Eraker et al (2003) first incorporate the jump process in the stochastic volatility model. However Nakajima (2012) concluded the effect of inclusion of the jump process has only limited effect in the stochastic volatility models for the stock prices or the exchange rate data. This study is to develop jump process stochastic volatility model on the interest rate of the sovereign bonds. The performance among the different specification of the model is compared by calculating the marginal likelihood by the method presented in Watanabe (2009) which improved the modified harmonic mean method of Geweke (1999). Model performance is significantly worsened if the data includes the period of the economic crisis. The inclusion of jump process is expected to help overcome this problem in applying the stochastic volatility model to the data observed during the financial crisis period.