In literature, many researchers focus on information contained in stochastic volatility dynamics, such as CBOE VIX index and its risk premium. However, there are relatively fewer studies on stochastic skewness dynamics. Simple linear regression indicates that stochastic volatility and stochastic skewness have different information about economic fundamentals: regressing CBOE VIX index on CBOE Skewness index, ranging from 1990 to 2013, yields R2 only 2%. Motivated by such striking yet simple empirical result and advances in reduced-form option pricing literature, we build one discrete-time equilibrium option pricing model with non-normal innovation. In our framework, market return skewness is driven by a stochastic process independent of volatility. Thus volatility and skewness dynamics possess different information about stochastic investment opportunity. Specifically, we use Variance Gamma distribution to generate economic growth shocks. Variance gamma innovation is able to be decomposed into two shocks, one of them is strictly positive, and the other is strictly negative. Our skewness risk factor controls the frequency and magnitude of these two shocks, therefore, controls the switching behavior of economic condition, e.g. from good to bad state. We show in our model that stochastic skewness risk should be priced, provided that investors have recursive utility function. Empirically, using Fama-Macbeth two pass regression and a panel of S&P 500 index option returns, we find that stochastic skewness has positive, statistically significant risk premium, and its sign does not change across our sample period. Moreover, our results show that skewness risk has superior explanation power for index option returns, compared with performance of volatility risk and Merton-type of jump risk. The shock of skewness risk premium is able to predict short-term market excess returns, the R2 is as high as 7%.