Government programs can improve local labor markets, but do they? : a re-analysis of Ham, Swenson, Imrohoroğlu, and Song (2011)

Research on the effects of enterprise zones - especially state programs - has generally failed to find evidence of beneficial effects such job growth or poverty reduction. In contrast, Ham, Swenson, Imrohoroğlu, and Song (2011, hereafter HSIS) present evidence that state and federal enterprise zones (EZs) established in the 1990s substantially reduced poverty. However, their estimates of the effects of EZs in reducing poverty are badly overstated for two reasons. First, HSIS have a substantial error in their data on poverty rates by Census tract, which accounts for most of the estimated impact of state EZs that they find. Second, their estimates of the effects of federal Empowerment Zones (EMPZs) and Enterprise Communities (ENTCs) appear to be strongly influenced by selection of areas that experienced negative shocks. An estimator based on comparing federally designated zones to more-comparable areas that applied for and were rejected as zones, or became zones in the future, yields much smaller estimates than those in HSIS. And the large poverty-reduction effects of ENTCs that HSIS found are largely spurious - not surprisingly, given that ENTCs received meager benefits and had no hiring credits.