Robust Constrained Combinations Of Forecasts

Paper shows that, under assumption that the single forecasts which enter the combination are unbiased, imposing some constraints on coordinates of <em>M </em>-estimator (of corresponding regression coefficients) leads to a gain in the asymptotic variance of one-step forward prediction evaluated by means of combined forecast. Numerical illustration offers a possibility to create an idea about the magnitude of the gain, and it also hints which type of <em>M </em>-estimator is to be used.