This paper provides new empirical evidence on tax buoyancy (tax revenues responsiveness to changes in economic activity) over the period 1990-2020 using a large panel of 185 countries. This study compares short-term and long-term buoyancy coefficients for total tax revenues and different individual taxes by reviewing and contrasting a range of estimators. Our results broadly confirm the main body of the literature on long-term buoyancy hovering around one. We find evidence of lower estimates for short-term buoyancy relative to previous literature, suggesting a limited automatic stabilization power of taxes. As a robustness exercise, in addition to changes in tax rates, we introduce novel control variables for tax exemptions and bases to disentangle discretionary from automatic tax revenue changes. The marginal changes in the results when controlling for policy actions suggest that, on average, the economic cycle does not necessarily influence tax reforms.