Modeling the Prepayment Risk of a Fixed Rate MBS Portfolio

This paper, explains why prepayment risk makes an MBS portfolio less attractive and discusses three different models for pricing this risk. The applicability of these three models are tested in Indian conditions by using them to price the prepayment risk for a synthetically created portfolio comprised of HDFC Home Loans with varying coupons and maturity periods. The methodology used in this paper is general and can be applied in real life situations for a loan portfolio with unequal loans and varying maturities. It is found that two of the models (constant prepayment rate and variable prepayment rate), which do not capture the impact of interest movements on the prepayment rates compared to the third model (elastic prepayment rate), which does so, underestimate the prepayment risk and hence overestimate the value of the portfolio. However, the use of elastic prepayment involves more complicated computation which increases the time and cost of implementation. And the model using a constant prepayment rate is a better approximation of prepayment risk than the model using a variable prepayment rate. Hence, if computational complexities restrain the use of the elastic prepayment model, the constant prepayment model could be preferred over the variable prepayment model. Finally, a sensitivity analysis of the portfolio returns shows that none of the three models are significantly impacted by changes in non-interest rate factors like age of the loan and average age of the population. This proves that such factors do not significantly affect the prepayment risk