In this article we characterize and estimate the process for short-ter
m interest rates using federal funds interest rate data. We presume th
at we are observing a discrete-time sample of a stationary scalar diff
usion. We concentrate on a class of models in which the local volatili
ty elasticity is constant and the drift has a flexible specification.
To accommodate missing observations and to break the link, between ''e
conomic time'' and calendar time, we model the sampling scheme as an i
ncreasing process that is not directly observed We propose and impleme
nt two new methods for estimation. We find evidence for a volatility e
lasticity between one and one-half and two. When interest rates are hi
gh, local mean reversion is small and the mechanism for inducing stati
onarity is the increased volatility of the diffusion process.