Parameter estimation in CKLS model by continuous observations (Q2667620)

The CIR process is an important SDE model of the form \[ dX_{t} =(a-b X_t )dt + \sigma X_t^\beta d W_{t}, \] where \(a, b, \sigma >0\) and \(\beta \in (\frac{1}{2}, 1)\) are parameters and \(\{ W_t, \,\,\, : \,\,\, t >0 \}\) is the Wiener process, finding various applications, among others in financial mathematics (e.g., in short rate modelling). This paper studies the problem of parameter estimation in this model in terms of continuous observations of a sample path \(\{X_{t}, \,\,\, : \,\,\, t \in [0,T]\}\). The properties of the maximum likelihood estimator (MLE) as well as an alternative estimator for the drift are studied, and in particular strong consistency properties for both estimators and asymptotic normality for the MLE, whereas the problem of estimating \(\sigma\) and \(\beta\) is treated by considering the realized quadratic variations. The performance of the estimators is further studied using simulation results.