Which size and evolution of the government expenditure multiplier in France (1980-2010)?

The importance of the stimulus packages that were injected in most advanced economies from the start of the financial crisis and the speed at which budgets are now being consolidated in Europe has revived the long-lasting debate on the size of fiscal multipliers. In this study, we focus on government expenditures on goods and services. Our conclusion following Blanchard and Perotti (2002) for the identification of government spending shocks is that the multiplier is significant and not far from 1 on impact and becomes statistically insignificant after about 3 years in France. We provide numerous robustness checks concerning the definition of expenditures, assumptions about data stationarity, the role of expectations and the choice of the sample. Moreover, using a time-varying SVAR model, our main findings are (1) that the multiplier did not evolve significantly at any horizon since the beginning of the 1980s and (2) that the variance of shocks hitting the economy evolves a lot more than the model autoregressive parameters. Even in alternative specifications where the Bayesian priors are pushed towards time-variation, the main evolution that we uncover is a (non-significant) decrease of the medium term expenditure multiplier, partly linked to a more aggressive monetary policy since the 1990s. We do not find evidence of an increase of the multiplier during every recession in France, contrary to the finding of Auerbach and Gorodnichenko (2012) for the United States. At least, business cycle conditions do not seem to be the main driver of the evolution of the expenditure multiplier in the last 30 years in France.