Abstract

In studying the economic cycle dependency of fiscal multipliers in Chile, we implement an independent component analysis for structural shock identification within a non-linear vector autoregressive setting with generalized impulse response functions. Thereby we relax more restrictive assumptions adopted in previous studies, namely the a-priori assumption of a recursive model structure and the use of linear impulse response functions. As a result, we cannot fully confirm core insights from more restrictive structural models: we find no significant differences in neither government spending nor government revenues multipliers when comparing different states of the economy. Moreover, our estimates imply that fiscal multipliers in Chile do not differ significantly from zero.