The dynamic effects of government spending: a FAVAR approach

The aim of the following work is to assess the dynamic effects of government spending on an extensive set of variables via macroeconometric modeling. As argued in Fatas and Mihov (2001), the following analysis of government spending is not a restrictive focus, but it explicitly aims at resolving a conflict among competing theories. Considering that the intrinsic nature of fiscal policy is to be predicted by economic agents, namely, there exists a problem of fiscal foresight, the exogenous shock in government expenditure cannot be regarded as the structural ones. Hence, it is crucial to build a time series that unmistakably conveys fiscal news to both econometrician and economic agents. In Auerbach and Gorodnichenko (2012) and in Fragetta and Gasteiger (2014), it is proposed to consider the effects of the shock in the forecasting of the growth rate of government spending, which is defined as purified spending shock; the latter fiscal variable should contain the relevant information regarding changes in public expenditure and aim at resolving the fiscal foresight issue. Furthermore, the econometrician can include in the VAR only a limited number of variables. The narrowness of the included variables could lead to non-fundamental shocks and, thus, to biased estimates. In fact, small scale VARs might suffer of deficiency of information given that the information set spanned by the endogenous variables in the VAR might be smaller than the one detained by economic agents. A way to uncover the comovements in the economy is to extract factors from a large informational dataset via Principal Component Analysis. Thus, including principal components that are consistent estimates of the factors in the model in order to build a factor-augmented VAR (FAVAR) as in Bernanke et al. (2005) should amend the non-fundamentalness problem. Therefore, applying the above-mentioned methodology, the impact on key macroeconomic variables is similar across VAR and FAVAR specifications, besides the effect on inflation that is negative and significant in the FAVAR estimation; the latter result is quite puzzling and it relates to the interaction between the forward-looking nature of inflation and the information-augmenting factors.