We consider the general estimation problem of the drift parameter matrix in
a multi-factors Vasicek model. We also develop estimation theory for the drift
parameters under natural restrictions. In particular, we propose shrinkage and
pretest estimators when the natural restrictions may or may not hold. Based on
the asymptotic properties of both unrestricted and restricted maximum likelihood
estimators (MLE), we examine the relative performance of the shrinkage and
pretest estimators. Finally, we appraise the properties of the listed estimators
based on a simulation study, insofar as applied implementation of the procedure
is concerned.