Nonparametric estimation of conditional beta pricing models

Abstract

We propose a new procedure to estimate and test conditional beta pricing models which allows for flexibility in the dynamics of assets' covariances with risk factors and market prices of risk (MPR). The method can be seen as a nonparametric version of the two-pass approach commonly employed in the context of unconditional models. In the first stage, conditional covariances are estimated nonparametrically for each asset and period using the time-series of previous data. In the second stage, time-varying MPR are estimated from the cross-section of returns and covariances, using the entire sample and allowing for heteroscedastic and cross-sectionally correlated errors. We prove the desirable properties of consistency and asymptotic normality of the estimators. Finally, an empirical application to the term structure of interest rates illustrates the method and highlights several drawbacks of existing parametric models.