Is the supermultiplier currently nil? - A replication study of Deleidi and Mazzucato (2021)
复制Deleidi和Mazzucato（2021）的研究，发现其关于美国政府任务导向支出对私人研发和经济活动产生正向溢出效应的结论不稳健，取决于数据变换和样本期选择，1985年后数据甚至显示临时挤出效应。
Analyzing US macro data via a structural vector-autoregressive model (SVAR), Deleidi and Mazzucato (2021) find strong positive spillovers from mission-oriented government spending on private research and development, as well as on overall economic activity (“crowding in”). Deleidi and Mazzucato apply the SVAR to first-differenced data despite the possibility of cointegration. The replication shows that the result hinges on the transformation of the data and the choice of the sample period. The time variation of the estimation results is substantial. When estimating the model with data starting after 1985, the results point to a temporary “crowding out” of private research and development spending. • Findings in Deleidi and Mazzucato (2021) about the “supermultiplier” are not robust. • Transformation of the data matters (first differences vs. level). • Results exhibit substantial time variation. • Results for the most recent part of the sample indicate temporary crowding out of private R&D.