Estimating fiscal multipliers by combining statistical identification with potentially endogenous proxies
研究发现，不同代理变量导致财政乘数估计矛盾，可能源于代理变量外生性假设被违反。本文提出贝叶斯非高斯SVAR方法处理潜在内生代理变量，发现增加政府支出比减税更能刺激经济。
Summary Different proxy variables used in fiscal policy structural vector autoregressions (SVARs) lead to contradicting conclusions regarding the size of fiscal multipliers. Our analysis suggests that the conflicting results may stem from violations of the proxy exogeneity assumptions. We propose a novel approach to include proxy variables in a Bayesian non-Gaussian SVAR, tailored to accommodate potentially endogenous proxies. Using our model, we find that increasing government spending is more effective in stimulating the economy than reducing taxes.