A Pitfall with DSGE–Based, Estimated, Government Spending Multipliers

In this paper, we study issues related to the estimation of long–run government spending multiplier (GSM) in a Dynamic Stochastic General Equilibrium (DSGE) context. We stress a potential source of bias in the GSM arising from the combination of (i) Edgeworth complementarity between private consumption and government expenditures and (ii) countercyclical government expenditures. We find that the degree of Edgeworth complementarity and the cyclicality of policy interact through cross–equation restrictions, paving the way for potential biases. It turns out that the GSM increases with the degree of Edgeworth complementarity between private consumption and government expenditures. Thus, any bias in the degree of Edgeworth complementarity translates into a biased GSM.