Three Essays In Empirical Studies On Derivatives PDF Download

This thesis is a collection of three essays in empirical studies on derivatives. In the first chapter, I investigate whether credit default swap spreads are affected by how the total risk is decomposed into the systematic risk and the idiosyncratic risk for a given level of the total risk. The risk composition is measured by the systematic risk proportion, defined as the proportion of the systematic variance in the total variance. I find that a firm's systematic risk proportion has a negative and significant effect on its CDS spreads. Moreover, this empirical finding is robust to various alternative specifications and estimations. Therefore, the composition of the total risk is an important determinant of CDS spreads.In the second chapter, I estimate the illiquidity premium in the CDS spreads based on Jarrow's illiquidity-modified Merton model using the transformed-data maximum likelihood estimation method. I find that the average model implied CDS illiquidity premium is about 15 basis points, accounting for 12% of the average level of the CDS spread. I further investigate how this parameter is affected by CDS liquidity measures such as the percentage bid-ask spread and the number of daily CDS spreads available in one month. I find that both liquidity measures are significant determinants of the model implied CDS illiquidity premium. In terms of relative importance, the bid-ask spread is more important than the number of daily CDS spreads statistically and economically.In the third chapter, I investigate the impact of the systematic risk on the volatility spread, i.e, the difference between the risk-neutral volatility and the physical volatility. I find that the systematic risk proportion of an underlying asset has a positive and significant impact on its volatility spread. The risk-neutral volatility in this study is measured with the increasingly popular approach known as the model-free risk-neutral volatility. The surprising positive systematic risk effect was first documented in Duan and Wei (2009) using the Black-Scholes implied volatility. I show that this effect is actually more prominent using the clearly better model-free risk-neutral volatility measure.