CMStatistics 2018: Start Registration
View Submission - CFE
A0592
Title: Likelihood-based dynamic asset pricing Authors:  Stefan Reitz - (Germany)
Dennis Umlandt - Kiel University (Germany) [presenting]
Abstract: A new parametric approach is proposed to estimate linear factor models with time-varying risk premia. In contrast to recent contributions in the literature, our framework abstains from introducing instrument variables to describe the time variation of risk prices. This is particularly useful in situations where instrument variables are unavailable or of poor quality and misspecification should be circumvented. Risk prices are derived from a generalized autoregressive score (GAS) model where parameters dynamics are driven by the scaled score of the observation density. Estimation and inference is conducted by likelihood maximization. We assess the potential improvement for predicting risk prices in a simulation study. Moreover, several applications to classical factor pricing models like the CAPM and the Fama-French 3-factor model are presented.