Uncovering Publication Bias in Fiscal Multiplier Estimates

Our study provides an integrated overview of fiscal multiplier estimates, a key parameter assessing the economy’s response to government interventions. Using a comprehensive database comprising 131 studies and over 3200 observations, we employ both linear and non- linear meta-analysis methods. Notably, our study marks the first application of Bayesian Model Averaging (BMA) in the context of meta-analysis for fiscal multipliers. Our results reveal a positive but moderate fiscal multiplier effect, ranging from 0.75 to 0.83 across different models. Significantly, our findings diverge from prior research by identifying a publication selection bias, largely attributed to our innovative use of BMA for heterogeneity investigation.