Is a fiscal stimulus effective? This classical question has received significant research attention since the collapse of the global financial services firm Lehman Brothers. Although most studies agree on the existence of Keynesian multiplier effects, several studies also demonstrate the existence of non-Keynesian effects. What explains this lack of consensus in the literature? In this paper, we aim to bridge the two views by estimating a near-vector autoregressive system that includes interaction terms of fiscal instruments, and the debt-to-gross domestic product (GDP) or the primary-deficit-to-GDP ratios. Moreover, to embed the dynamics of the debt-to-GDP ratio in the analysis, we explicitly incorporate the government budget constraint. By computing and comparing the impulse response functions, we find Keynesian effects when fiscal conditions are sound, and non-Keynesian effects when the primary deficit is large.