The Discretization Bias for Processes of the Short-Term Interest Rate : An Empirical Analysis

This paper compares difference continuous-time specifications for the short-term interest rate dynamics on five European markets. We propose a general specification which encompasses nine well-known processes of the financial literature. A classical estimation of the parameters leads us to the choice of simple models like the Ornstein-Uhlenbeck process of Vasicek (1977) or the “Square Root” process of Cox, Ingersoll and Ross (1985). Then we focus on the discretization bias and a methodology to correct it is proposed. The results confirm the existence of a discretization bias. Finally, we show that this simulation-based methodology can be applied to the estimation of the parameters of correlated diffusion processes.