Investment Decisions Under Model Uncertainty: An Application Using Exchanger Rate and Interest Rate Forecasts

In this paper we analyse the effect of model uncertainty on the wealth and utility outcomes of an investment decision. We compute optimal portfolio weights for domestic and foreign assets and using these weights we construct end investment horizon wealth and utility ratios. Model uncertainty is accounted for using a Bayesian type Model Averaging (BMA), where the Schwartz Bayeisan and Akaike information criterion (SBC and AIC) are used to form model weights. SBC, AIC and an arithmetic averages, as well as their twelve component models are then used to generate forecasts of the nominal exchange rate, the nominal domestic interest rate and the nominal foreign interest rate as inputs to the investment decision. We find that for our US-UK, 1981m1-2003m6 monthly application model uncertainty suggests a wide range of optimal portfolio weights for the allocation between domestic and foreign assets. Models which are not selected as the SBC or AIC best models nonetheless perform well relative to the no prediction benchmark case when evaluated using wealth and utility criteria