An Application of Interest Rate Models to RMBS and SME Conditional Prepayment Rates and CPR Time Series Analysis

The ABS Research team at Investment Bank X has asked the new summer intern to come up with an innovative way to forecast Conditional Prepayment Rates (CPR). The intern has some knowledge about interest rates models and decides to apply the same approach to model CPR with the goal of testing the following aspects: a) CPR and Interest Rates follow similar dynamics. b) The stochastic component of CPR and Interest Rates are negatively correlated. c) CPR is a mean reverting process, emphazised in each regime of the economy. In the first part of the project, we analyse the time series of Spanish annual and monthly CPR from a portfolio of RMBS and SME Collateral Loan Obligations (CLOs). In the second part of the project we estimate CIR and Vasicek models applied to CPR and test their dynamics, fit and prediction performance with historical data. We then run a future simulation forecast of the model with Monte-Carlo Method. We simulate a forecast taking into account regime switches in the economy by performing a Monte-Carlo simulation with a dynamic calibration of the model. We modify the stochastic differential equation by including other variables that are correlated with CPR. In order to measure explanatory power in correlation, we run a Principal Component Analysis. We dynamically simulate this modified version of the model. We check results consistency with real CPR dynamics and provide a conclusion