Abstract:
Governments perform national, labor-intensive censuses on a regular schedule. Censuses represent many of the largest peacetime expansions and contractions in federal hiring. The predetermined occurrence and scale of the census offers an economic experiment in the effects of temporary government hiring. This paper describes the construction of a data series on census hiring in the United States since 1950 and also collects available data on census employment in England and Wales, Canada, Korea, and Japan. Regressing total employment changes on census hiring yields coefficients extremely close to 1, indicating that there is no spillover from census hiring to the rest of the economy. Using census hiring and occurrence as instruments for government hiring in the US, Canada, and Korea, I estimate the effect of federal hiring on overall employment. Different samples yield varying jobs multipliers, with point estimates varying from -0.01 to 1.48. Including Korean and Canadian data yields lower multipliers, while including pre-1990 US data yields higher multipliers. In no specification can I reject the hypothesis that the job multiplier equals 1. In all specifications, standard errors are large enough that I can reject neither Keynesian nor crowd-out effects.
