ASYMPTOTIC INFERENCE FOR THE PARAMETERS OF A DISCRETE-TIME SQUARE-ROOT PROCESS

This paper is concerned with asymptotic properties of the maximum likelihood estimators for the discrete-time square-root process. This process and its generalizations are employed in financial literature as models for movements of asset prices. the considered process is nonergodic and therefore standard maximum likelihood theory does not apply. the nonstandard asymptotic theory is developed. Strong consistency of the estimators is established, joint asymptotic distribution of the properly normalized estimators is obtained and confidence intervals for the parameters are constructed. the results of the small simulation study are reported. Copyright 1994 Blackwell Publishers.