Do Government Spending Multipliers Depend on the Sign of the Shock?

We analyze whether government spending multipliers differ by the sign of the shock, i.e., do shocks that raise government spending lead to different multipliers than shocks that lower government spending. Using aggregate historical U.S. data, we first apply Ben Zeev’s (2020) nonlinear diagnostic tests and find evidence of nonlinearities in the impulse response functions of both government spending and GDP. We then extend Ramey and Zubairy’s (2018) framework to allow for asymmetric effects as a type of state dependence to estimate multipliers. While the estimates imply differences in the impulse response functions by sign of the shock, the resulting multipliers estimates do not differ by sign of the shock, either quantitatively or statistically. Thus, we find no evidence of asymmetry of government spending multipliers. We compare our results and methods to some recent work that found asymmetries.