Unconventional Confidence Bands in the Literature on the Government Spending Multiplier

Ryan Murphy, Southern Methodist University

Abstract

Macroeconomists have applied many versions of vector autoregression to measuring the size of the government spending multiplier. Very frequently in the literature on that multiplier, the statistical significance of results is held to an unconventionally low standard of one standard error. This paper will document the pervasiveness of the issue among thirty-one papers. Five of the six papers with at least 500 Google citations use that unconventionally narrow band.