Time-Varying Modeling of Systematic Risk: using High-Frequency Characterization of Tehran Stock Exchange

We decompose time-varying beta for stock into beta for continuous systematic risk and beta for discontinuous systematic risk. Brownian motion is assumed as nature of price movements in our modeling. Our empirical research is based on high-frequency data for stocks from Tehran Stock Exchange. Our market portfolio experiences 136 days out of 243 trading days with jumps which is a considerable ratio. Using 1200 monthly (5200 weekly) estimations, 100 stocks for 12 months (52 weeks), 2400 (10400) betas are calculated. No general trend or constancy has been seen in continuous or discrete betas, and no general correlation between them. Existence and importance of both continuous and discrete betas are demonstrated by related tests. Comparing continuous and discrete beta, show that, in addition to greater significance of discrete beta, the estimated jump beta is higher than the continuous beta almost 87% of the time; and on average jump betas are 180% higher than continuous betas. Both greater significance and greater values are resulted for discrete risk premium.