This paper empirically studies the U.S. multiplier effects of government investment, government consumption and total government purchases on output. We explore dependencies of the multipliers on states of the economy, measured in different ways. Using local projections with instrumental variables, we find that a model without state-dependencies and using total government spending (instead of its components) provides the best fit to post-WWII data. These results are robust to various alternative specifications. We account for the COVID-19 period with a pandemic stringency index and for monetary policy shocks with a shadow interest rate. The government spending multiplier is approximately 0.5.

本文实证研究了美国政府投资、政府消费和政府总购买对产出的乘数效应。我们探讨乘数对经济状况的依赖性，以不同的方式衡量。使用带有工具变量的本地预测，我们发现不依赖于国家并使用政府总支出（而不是其组成部分）的模型最适合二战后的数据。这些结果对于各种替代规范都是稳健的。我们用大流行严格指数来解释 COVID-19 时期，并用影子利率来解释货币政策冲击。政府支出乘数约为0.5。