Abstract: Recent advancements in econometrics have predominantly centered on the refinement of identification strategies to estimate causal effects. In this paper, we conduct a replication exercise to examine to which extent the presence of outliers in the data can affect the outcomes of interest. Our analysis is based on 99 articles that were published in 2017 in nine top economic journals, with available data for replications. After replicating the results using robust estimators (MDPD and MM), we observe that 50% of the causal parameters change the magnitude in a range from 23% to 60% in absolute value when considering robust estimates. Similarly, we observe changes in parameter sign for a share of parameters ranging from 7,8% to 20,5%. None of these changes can be attributed to a large set of observables for which can control. This finding prompts further discussion and encourages a more precise analysis of outliers when addressing causal estimates. (joint work with Federico Belotti and Alessandro Mistretta)