Using State Pension Shocks to Estimate Fiscal Multipliers since the Great Recession

One painful lesson from the Great Recession is that there remain limits to what is achievable with conventional monetary policy. The severity of the downturn, despite near zero interest rates, has renewed interest in fiscal stimulus among academics and policy makers alike. Holding the general question of fiscal multipliers aside, the ongoing political debate regarding past and future stimulus raises the pressing question of the current size of the multiplier in the United States. This paper aims to provide evidence on that question. Has government spending raised income and employment since 2008? The canonical literature estimating fiscal policy multipliers uses either time-series models (Blanchard and Perotti 2002) or variation in national defense spending due to exogenous military events (Barro and Redlick 2011, Ramey 2011). In recent years there have been a number of studies estimating fiscal policy multipliers at the regional level using an instrumental variables approach. For example, Nakamura and Steinsson (2011) isolate variation in local military spending due to changes in national defense outlays, and Serratto and Wingender (2010) estimate the effect of spending due to Census overand undercounting. These studies, as well as several others 1 , generally find that exogenous spending shocks have significant and sizeable income and employment effects. The identification strategies used in most of these papers are not easily adapted to study the specific effects of spending in the US since 2008. A number of papers, such as Sacerdote (2011), Chodow-Reich et al. (2012), and Wilson (2012), have estimated fiscal multipliers in this recession using local variation in ARRA stimulus outlays. Like the regional-level literature discussed above, these papers generally find that spending increased both local income and employment. The importance of the question, though, warrants investigation using an additional, independent source of variation. In this paper, I exploit the identification strategy used in Shoag (2010) to recover exogenous spending changes associated with