Testing for Asset Pricing Errors
Prof. Merrick Li
Assistant Professor in the Department of Economics
The Chinese University of Hong Kong
Prof. Xiye Yang
We introduce a testing procedure to discriminate between two broad classes of models for asset prices. Under the null hypothesis of no-arbitrage, the asset prices evolve as a semimartingale with time-varying risk premium, stochastic volatility and jumps; under the alternative, a pricing error that reflects various market frictions or biases is added. Our approach analyzes the autocovariances of observed returns over different time horizons. We propose a method that combines information from returns across multiple horizons to detect pricing errors. Our theory reveals that analysing a small set of horizons suffices to identify a wide range of pricing errors that can be endogenous, nonstationary, serially correlated, and weakly identified.