ROBUST OPTIMIZATION MODELS FOR MANAGING CALLABLE BOND PORTFOLIOS

A major sector of the bond markets is currently represented by instrum ents with embedded call options. The complexity of bonds with call fea tures, coupled with the recent increase in volatility, has raised the risks as well as the potential rewards for bond holders. These complex ities, however, make it difficult for the portfolio manager to evaluat e individual securities and their associated risks in order to success fully construct bond portfolios. Traditional bond portfolio management methods are inadequate, particularly when interest-rate-dependent cas hflows are involved. In this paper we integrate traditional simulation models for bond pricing with recent developments in robust optimizati on to develop tools for the management of portfolios of callable bonds . Two models are developed: a single-period model that imposes robustn ess by penalizing downside tracking error, and a multi-stage stochasti c program with recourse. Both models are applied to create a portfolio to track a callable bond index. The models are backtested using ex po ste market data over the period from January 1992 to March 1993, and t hey perform constistently well.