On the usefulness of macroeconomic forecasts as inputs to forecasting models

A forecasting model for y t based on its relationship to exogenous variables (e.g. x̌ t ) must use x̌ t , the forecast of x̌ t . An example is given where commercially available x̌ t 's are sufficiently inaccurate that a univariate model for y t appears preferable. For a variety of types of models inclusion of an exogenous variable x̌ t is shown to worsen the y t forecasts whenever x̌ t must itself be forecast by x̌ t and MSE (x̌ t ) > Var (x̌ t ). Tests with forecasts from a variety of sources indicate that, with a few notable exceptions, MSE (x̌ t ) > Var (x̌ t ) is common for macroeconomic forecasts more than a quarter or two ahead. Thus, either: available medium range forecasts for many macroeconomic variables (e.g. the GNP growth rate) are not an improvement over the sample mean (so that such variables are not useful explanatory variables in forecasting models), and/or the suboptimization involved in directly replacing x̌ t by x̌ t is a luxury that we cannot afford.