Journal of Risk

Bias-corrected estimators for the Vasicek model: an application in risk measure estimation

Zi-Yi Guo

  • Our Monte Carlo simulation results indicate that the widely-used least squares estimations of the Vasicek model suffer significant small-sample biases even if the sample length reaches as long as 30 years.
  • Bias-corrected estimators could substantially reduce the small sample biases of the least squares estimations, and further project much more accurate value-at-risk and potential future exposure estimates.
  • Empirical applications to a variety of time series are in general in line with the Monte Carlo simulation results.

We evaluate the usefulness of bias-correction methods in enhancing the Vasicek model for market risk and counterparty risk management practices. The naive bias-corrected estimator, the Tang and Chen bias-corrected estimator and the Bao et al bias-corrected estimator are selected to be compared against the benchmark least squares (LS) estimator. Our Monte Carlo experiment shows that the bias-corrected estimators substantially reduce the small sample bias of the LS estimator for the Vasicek model and project much more accurate value-at-risk and potential future exposure estimations. Even if the sample length is as long as 30 years, the improvements are still significant, especially for the cases where the mean-reversion parameter is close to zero. The applications to real data further demonstrate that the small sample bias of the LS estimator cannot be ignored and one should consider bias-corrected estimators for the Vasicek model.

Sorry, our subscription options are not loading right now

Please try again later. Get in touch with our customer services team if this issue persists.

New to View our subscription options

If you already have an account, please sign in here.

You need to sign in to use this feature. If you don’t have a account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: