The fiscal reaction function (FRF) remains a valuable tool for gaining insights into a country’s fiscal sustainability and output stability. However, there is currently no consensus on how to model it. Previous research has primarily focused on questioning the dominant cubic reaction form and enlarging the sample size. This document investigates the best functional form for the FRF by adopting a nonlinear autoregressive distributed lag approach that accounts for a potential structural break in the data across periods such as the COVID-19 crisis and the Global Financial Crisis. Contrary to what is often assumed empirically, the sample aims for generalization rather than heterogeneity reduction. Using data for G20 members’ economies over the period 1998-2021, this thesis finds evidence of a fiscal reaction function with a fourth-degree polynomial (quartic) function form that shows a limited or insufficient fiscal space to overcome future crises.