One reason that business cycle turning points are hard to understand and predict may be that economists typically employ linear models, while a growing body of research suggests that many economic variables interact in a nonlinear fashion. This study measures the contribution of changes in military spending to business cycles. It uses data at the state level, which offer the advantages of a great diversity and volume of data relative to the aggregate, and a framework capable of capturing nonlinear and asymmetric relationships.

The results suggest that military spending is a significant determinant of economic activity at the state level, with a modest impact on most states and a sizable impact on those with a large exposure to the military sector. The transmission of military spending changes to personal income (and to employment) appears to be nonlinear and asymmetric, with large cutbacks having proportionally larger responses than either large contract awards or small changes.