The Productive Government Spending Multiplier, In and Out of The Zero Lower Bound

Recently, a series of papers have argued that output multipliers of government spending can be potentially large during times when the Zero Lower Bound on nominal interest rates is binding (Christiano et al. (2011)). This literature generally considers "excess-savings" liquidity traps and identifies the reaction of real interest rates —that follows the effect of government purchases on marginal cost and, hence, inflation —as the main channel of propagation. Here, I show that taking explicitly into account the fact that government spending is productive can mitigate this result. The higher the share of productive government spending in total stimulus spending, the lower the gap between the government spending multipliers in and out of the Zero Lower Bound. Furthermore, a sufficient share of productive government spending in total stimulus spending will imply a higher multiplier when the Zero Lower Bound is not binding. It follows that the government spending multiplier need not be unusually large when the economy is in an "excess-savings" liquidity trap. In a "expectationsdriven" liquidity trap (Mertens & Ravn (2010)) however, the government spending multiplier will be larger than in normal times for a sufficient share of productive government spending. But for this to happen, a rise in inflation is still needed. While the predictions of the model with an "expectations-driven" liquidity trap are difficult to compare with the data, I show that the model with an "excess-savings" liquidity trap is at odds with recent empirical evidence on the behavior of key macroeconomic variables in a recession. In contrast, the simple New-keynesian model augmented with a sufficient share of productive government spending is qualitatively consistent with aforementioned evidence.