Managing Transaction Costs in a Dynamic Trading Strategy

We derive an explicit solution to a continuous time dynamic portfolio problem assuming investors maximize their welfare from a consumption stream in an incomplete market where returns to the securities are predictable but costly to trade. The solution is phrased in terms of a risk-sensitive Riccati equation. We show that the optimal trading strategy is to target a portfolio that is the optimal solution to a frictionless (or 'no-cost') dynamic portfolio problem but where the returns to the assets have been adjusted for costs; that is they have been expressed on a net rather than gross basis. The legacy portfolio (the inherited undesirable positions) are then traded away in line with a backward-looking optimal execution problem. We show that the utility gradient is a stochastic discount factor that prices the assets net returns. Thus we are able to generalise some of the results of the martingale approach to dynamic portfolio theory to market with frictions