Conditionally parametric fits for CAPM betas

The CAPM model assumes stock returns to be a linear function of the market return. However, there is considerable evidence that the beta stability assumption commonly used when estimating the model is invalid. Nonparametric regression methods are used to examine the stability of beta coefficients in German stock returns. Since local polynomial regression is used for estimation, known methods for testing the stability and for bandwidth choice can be used. For some returns the test indicates time-varying betas. For these returns conditionally parametric fits are calculated.