We propose and implement a coherent statistical framework
for combining theoretical and empirical models of macroeconomic activity. The
framework is Bayesian, and enables the formal yet probabilistic incorporation
of uncertainty regarding the parameterization of theoretical models. The approach
is illustrated using a neoclassical business-cycle model which builds on the Greenwood,
Hercowitz, and Huffman (1988) variable-utilization framework to study out-of-sample
forecasting of output and investment. The forecasts so produced are comparable
with those from a Bayesian vector autoregression.