Abstract
The Brenner hypothesis is essentially that economic cycles, characterized by unemployment and fluctuations in per capita income can have profound negative implications for population health. A number of subsequent studies have identified shortcomings in Brenner's model and have reported results that for the most part contradict his results. This paper argues that the failure to account for the time‐series properties (i.e. the potential for unit root behaviour) of macro level data is a key omission in Brenner's and other subsequent studies. To address this omission an error correction model specification was applied to American data for the period 1948–1996. The findings suggest that economic cycles do have a permanent effect on population health. Paradoxically, they also suggest that economic growth and increases in unemployment reduce aggregate mortality risk. A need for measures of economic change that are perhaps more sensitive to the effects of economic cycles on groups that may be at greater risk of unemployment was identified. Copyright © 2004 John Wiley & Sons, Ltd.