Out-of-sample performance of discrete-time spot interest rate models.

We provide a comprehensive analysis of the out-of-sample performance of a wide variety of mark rate models in forecasting the probability density of coming time interest rates. Although the greatest in quantity parsimonious models perform best in forecasting the conditional mean of many financial time series, we find that the mark rate models that incorporate conditional heteroscedasticity and exces kurtosis or heavy tails have better density forecasts. Generalized autoregressive conditional heteroscedasticity significantly improves the modeling of the conditional variance and kurtosis, whereas regime switching and pass by a leaps improve the modeling of the marginal density of interest rates. Our analysis exhibits that the sophisticated spot