Within the current debate about the magnitude of fiscal multipliers and the persistent effect of a fiscal policy shock, we propose an empirical analysis that aims at two main objectives: exploring the possibly different effects of specific classes of government spending and assessing the degree of persistence in the dynamic impact on GDP. In particular, we identify and characterize the historical role played by different components of public expenditure in the US by breaking-down the total government expenditure into investment and consumption. Using quarterly data for the period 1947-2017, in the spirit of Blanchard and Perotti (2002) and Auerbach and Gorodnichenko (2012) models, we implement a time series analysis based on a Structural Vector Autoregressive (SVAR) methodology that considers a set of alternative specifications in order to achieve robust conclusions.

We choose SVAR model for a number of reasons: (i) SVAR model allows us to isolate the exogenous component of the fiscal variables via a chosen identification strategy; (ii) it gives us the chance to consider the effects of a broader set of fiscal interventions; (iii) it allows us to provide an objective quantitative estimate of the effects of an average increase in different categories of government spending; (iv) Finally, it is important not to disregard the possibly rich dynamic interactions between the relevant variables. A SVAR framework is in this respect the most general approach, giving us the possibility not only to assess the effects of a fiscal stimulus over different time horizons, but also to characterize the time profile of a typical government spending episode, while accounting for all possible dynamic feedback within the economy.

Our results indicate that total public expenditure has a significant impact multiplier equal to 0.68, a significant peak multiplier of 1.72 and a significant long-run multiplier of 0.79. On the contrary, taxes show a negative but not significant effect on GDP. When considering the two components of public spending, we obtain a significant impact multiplier of 1.86 and 0.36, a significant peak multiplier of 3.15 and 1.95, and a long-run multiplier of 1.65 and 1.36, for investment and consumption respectively. The main results of analysis can be summarised as follows: (i) spending multipliers are generally larger than 1 and bigger than tax multiplier (in absolute value); (ii) public investment generates a larger effect on GDP than public consumption; (iii) fiscal policies shocks generate persistent and positive effect on the output level.