The economics of beauty is now a burgeoning field of research. Not only the magnitude but also the direction of the beauty effect on labor outcomes is a matter of discussion. In this work, I conduct a quantitative synthesis of 418 estimates of the effect of beauty on worker's productivity, as reported in 37 studies. The estimates are tested for publication selection using informal testing of the funnel plot as well as formal testing methods. The results provide substantial evidence of selective reporting: positive estimates of the beauty effect are preferred in literature. The set of 21 explanatory variables was collected to determine the sources of heterogeneity in the reported estimates. To account for the model uncertainty, I employ the Bayesian and Frequentist model averaging. The results indicate that differences in the reported estimates appear to be driven by choice of study design and sources of real heterogeneity, such as geographical regions and individual characteristics of respondents. The type of occupation and gender of respondents have no impact on the estimates of beauty effect concerning productivity. The average beauty effect is probably much lower than commonly believed based on the available empirical literature.