UCLA

This thesis uses high-frequency data to estimate the stochastic discount factor. The high-frequency data used is sampled at one-second frequency. The fundamental equation of asset pricing is based on the continuous-time no-arbitrage theory. For empirical estimation, I apply the general method of moments to estimate the market price of risk for the risk factors, which consist of exchange-traded funds (ETFs). In Chapter 1, I estimate a one-factor model using the ETF SPY (an SPDR ETF that tracks S&P 500 index) as the risk factor. The estimated risk prices are significant over 2/3 of the sample, and the time series shows plausible patterns of the overall riskiness of the market. An additional factor using IWM (the Russell 2000 ETF that tracks the performance of the small-cap equity market) as the second factor is incorporated into the model in Chapter 2 to arrive at a two-factor model. Adding IWM improves the performance of the model and the estimation precision substantially: the risk price of SPY is almost always significant and the risk price of IWM is significant for about 2/3 of the sample. In Chapter 3 I extend the two-factor model by adding a third factor. Adding a third factor improves the performance of the model to a modest extent, but the large-cap factor SPY followed by the small-cap factor IWM are predominant.