Raphael Corbi

We exploit a series of discontinuities, at several population thresholds, in the allocation mechanism of federal transfers to municipal governments in Brazil to identify the causal effect of municipal spending on local labor markets, using a ‘fuzzy’ regression discontinuity design. Our estimates imply a cost per job of about 8; 000 US dollars per year, mostly driven by employment in services, and a local income multiplier of around two. A currency union model with nominal rigidities and liquidity constraints implies that the stimulative effects would have been substantially smaller if local government spending was financed by local tax revenues rather than regional transfers.