|Job Market Paper||589 KB|
I show that the effectiveness of fiscal stimulus as a tool for counter-cyclical stabilization depends not only on how much of changes in income households spend, but also on the composition of that expenditure. By combining expenditure and production data I measure the extent to which households cut back expenditure on labor-intensive goods upon unemployment. The result is quantitatively relevant because the labor share with which different final consumption goods are produced varies widely across the economy. I find that upon unemployment, a household reduces demand for other workers' labor by 6.5%, 15% more than what is implied when heterogeneity in production and expenditure is ignored. In the context of the Great Recession, the expenditure response to unemployment accounts for a fifth of the drop in labor compensation. Using a multi-good, multi-sector New-Keynesian model with heterogeneous agents, I show that my findings have significant implications for the targeting and evaluation of fiscal stimulus. First, the fiscal multiplier of government purchases of labor-intensive goods is almost five times larger than for purchases of capital-intensive goods. Second, the heterogeneity I document explains the lower effectiveness to stimulate the economy of capital-intensive military spending, as compared to highly labor-intensive general government spending. Third, I show that the decline of the labor share in the last decades has reduced the effectiveness of fiscal policy.