Research

I study how the allocation of national stimulus across US counties affects aggregate job creation. Using county-level variation in American Recovery and Reinvestment Act (2009–2012) outlays, I find that local employment responses to government spending are largest in mid-sized counties (about 100,000 residents) and substantially smaller in both small and large counties (about 10,000 and 1 million residents).

Evidence on local production structure, purchasing patterns, and hiring points to two opposing forces. While small counties leak most of the demand they receive to other places, large counties turn the demand they retain into fewer jobs, consistent with greater worker bargaining power and higher recruiting costs. Mid-sized counties balance the two. I interpret and rationalize these patterns using a multi-region New Keynesian model in which locations are linked through trade, and labor markets feature search and matching frictions.

I reallocate funds from large to mid-sized counties until large counties receive the same funding per adult as small counties. This reallocation would have created about 91,400 additional job-years without increasing total spending. At the baseline cost per job-year, this gain is equivalent to about $4.4 billion in additional stimulus spending. The spatial allocation of stimulus is thus an important dimension of stabilization policy distinct from the size of the program.