Consistent estimation in regression models for the drift function in some continuous time models

A regression-based consistent estimation method for the drift function in some continuous time models is suggested, and its limiting distribution is derived. The accuracy of the new estimation method is examined via some finite sample Monte Carlo simulation studies. The proposed approach is employed to estimate the drift function for the U.S. Treasury Bill yields data, assuming the appropriateness of the model under consideration for it. Our approach offers an explanation for versatile conclusions on the shape of drift function in the existing literature.