We estimate a model that allows for dynamic and interdependent responses of morbidity in different local areas to economic conditions at the local and national level, with statistical selection of optimal local area. We apply this approach to quarterly British data on chronic health conditions for those of working age over the period 2002-2016. We find strong and robust counter-cyclical relationships for overall chronic health, and for five broad types of health conditions. Chronic health conditions therefore increase in poor economic times. There is considerable spatial heterogeneity across local areas, with the counter-cyclical relationship being strongest in poorer local areas with more traditional industrial structures. We find that feedback effects are quantitatively important across local areas, and dynamic effects that differ by health condition. Consequently, the standard panel data model commonly used in the literature considerably under-estimates the extent of the countercyclical relationship in our context.