Heterogeneity in the Effect of Economic Downturns on Health Care Labor Supply

The downstream effects of economic downturns have been of longstanding research and policy interest. Within health economics, specifically, both demand- and supply-side literatures have emerged. The former has shown fluctuations in health and insurance status as well as care consumption in tandem with the business cycle, while the latter has investigated how health care workers respond to macroeconomic changes. The set of supply-side studies is smaller, however, and narrower in focus. Most of the related work concerns low- to mid-skill level nurses employed within the hospital or nursing home industries. Although these contributions are important, they cannot offer a complete picture of how recessions influence the health care workforce across the gamut of occupations—which vary in training requirements as well as scarcity.

To partly fill these gaps, we exploit a complete census of actively licensed health care workers in North Carolina from 2000-2015, which contain information on the full range of worker skill types. We first estimate standard two-way fixed effects models to capture how the supply of a given occupation type changes with an increase in the local unemployment rate. We then exploit heterogeneity in the incidence of the Great Recession to examine differential changes in health care labor supply between the most versus least affected areas. Doing so allows us to directly estimate the impact of the recent recession using a dose-response empirical approach. Across both identification strategies, we examine fluctuations in aggregate labor supply as well as changes relative to the size of the local population (i.e., provider-to-patients ratio).

Our findings reveal substantive heterogeneity across workers and geographic areas. Nearly all occupations exhibit declines in total workers during a worsening local economy—though for many categories the drop in available supply is in proportion to population losses. However, some notable exceptions are found within rural areas. A 1-percentage point increase in the unemployment rate leads to a 6% decline in the supply of registered nurses and a 2% decline in the registered nurse-to-population ratio. Similarly, rural areas lose 6% of their physicians and 12% of their physician assistants with each 1-percentage point increase in the unemployment rate, which then translates to a 2% and 4% reduction in the prevailing physician-to-population ratio and PA-to-population ratio, respectively. The strongest effects of a worsening business climate on non-rural areas are found among occupational therapists and psychologists. The number of occupational therapists declines by 4% and psychologists by 6% with a rising unemployment rate. The corresponding provider-to-population ratio changes are a negative 3% and 4%, respectively. We also find that areas differentially (i.e., most) impacted by the Great Recession of 2008-2009 have fairly persistent and negative effects on their provider-to-population ratios for registered nurses, psychologists, optometrists, and physical therapists relative to areas experiencing a lighter downturn.

Overall, we see consistent evidence of procyclical changes in health care labor supply leading to worsening provider-to-population ratios over a range of occupations. The stronger effects in rural areas are also likely to exacerbate pre-existing health care delivery challenges.