Option pricing with dynamic conditional skewness

In this paper, we develop a discrete-time affine option-pricing model that explicitly incorporates the dynamics of conditional skewness. The new proposed model features different dynamics for conditional skewness and variance. To stress the difference in information, we use alternative realized measures constructed from high-frequency historical returns to update skewness and variance dynamics. By Fourier inversion, we derive closed-form option valuation formulas. Empirically, the flexibility that the model offers for conditional skewness as well as high-frequency information from the underlying asset contribute to superior performance upon benchmark models using S&P 500 index options. Overall, the joint modeling of dynamic conditional skewness and realized measures leads to an out-of-sample gain of 12.25% in pricing accuracy. The improvements are more pronounced for deep in-the-money calls, options with shorter maturities, and during highly volatile periods.