Excess Capacity and Heterogeneity in the Fiscal Multiplier : Evidence from the Recovery Act

We estimate local multipliers using cross-county variation in expenditure in the ARRA. We use withinstate variation, and include other demographic controls as well as a predicted employment control using an industry shift-share measure. We nd that counties receiving more stimulus expenditures had followed parallel employment trends prior to the ARRA as compared to other counties. We estimate an average annualized employment multiplier of 1.211 job-years per $100K spent per county resident. We nd strong evidence of heterogeneous treatment e ects: the employment response is much greater in counties hit harder by the Great Recession, and hence with likely greater excess capacity. In below median excess capacity counties, the employment multiplier is 0.39. In above median excess capacity counites, the multiplier rises to 2.83. These ndings imply that an employment-maximizing stimulus package targeted to high excess capacity counties would have created 83% more (3.60 million) jobs. While our ndings are consistent with state-dependent scal multipliers, the heterogeneity is not due to the zero-lower bound since our cross-sectional variation in excess capacity holds the interest rate constant. Instead, our ndings suggest that the spatial variation in multipliers re ects variation in the depth of the recession across di erent labor markets. Consistent with the evidence on hysterisis, we nd that the employment impact of the stimulus was long lasting and have likely persisted through the current exapansion. ∗University of Massachusetts Amherst, NBER and IZA †University of Maryland at College Park ‡University of Maryland at College Park ŸEconomic Policy Institute