Uncovering the Skewness News Impact Curve

We investigate, within flexible semiparametric and parametric frameworks, the shape of the news impact curve (NIC) for the conditional skewness of stock returns, that is, how past returns affect present skewness. We find that returns may impact skewness in ways that sharply differ from those proposed in earlier literature. The skewness NIC may exhibit sign asymmetry, other types of nonlinearity, and even non-monotonicity. In particular, the newly discovered "rotated S"-shape of the skewness NIC for the S&P500 index is intriguing. We explore, among other things, properties of skewness NIC estimates and conditional density forecasts, the term structure of the skewness NIC, and previously documented approaches to its modeling.