The main purpose of this paper is to use the Gibbs-sampling model method as an alternative technique to estimate mortgage payment rates. We use this method due to the face that available data on mortgage loans in many emerging markets, like Taiwan, are somewhat limited. The empirical results confirm that one advantage of this approach is that the coefficient estimate of each parameter exhibits its own exact posterior distribution rather than a mere point estimate. More than that, this approach allows for finite-sample inferences within the model estimates. We conclude, therefore, that the realized distributional form can be inferred from the statistical results. This significantly enhances the measurement of mortgage prepayment probabilities, particularly when the data are scant.