A model for portfolio management with mortgage-backed securities

This paper presents a stochastic programming model for the mamagement of large portfolios of mortgage-backed securities (MBS). It is a two-stage, multiperiod model, whereby portfolio decisions made here-and-now are influenced by uncertain information about the future. In particular, uncertainty in both the prepayment activity of the MBSs in the portfolio, as well as uncertainty about the future reinvestment rates is considered. A simulation procedure is used to generate interest rate paths and prepayment behavior, and the stochastic program can be extremely large. Solution of the resulting large-scale programs is particularly challenging. The paper shows that with massively parallel computing technology, the proposed models are indeed solvable. Empirical results on a Connection Machine CM-2 are reported.