Estimating an affine term structure model of interest rates with correlated noise

Kalman filtering for the affine term structure model of interest rates is typically applied under the assumption of white noise. However, correlated noise frequently occurs during actual data processing. The accuracy and reliability of the filter are compromised if the correlated noise is assumed to be white noise. This paper develops a measurement expansion scheme for the affine term structure model based on the whitening properties of the Kalman filter, enabling latent factor estimation under the general assumption of correlated noise. The simulation results indicate that the estimation based on the measurement expansion scheme achieves higher accuracy compared to the traditional method.