In contrast to representative agent models with flat-rate taxes, this framework allows the authors to distinguish between marginal tax rates and the empirical proxies that are typically used for these rates such as the share of tax revenue, or government expenditures, in GDP. The analysis then illustrates how the endogenous nature of these proxy variables causes them to be weakly correlated, or even increase, with economic growth. This study, therefore, helps explain why cross-country regressions have mostly failed to uncover the distortional growth effects of taxes. In fact, while past U.S. tax reforms appear to have contributed only small increases in per capita GDP growth, the authors' analysis nevertheless suggests that differences in tax codes across countries explain a 2.5 percent variation in cross-sectional growth rates. Finally, the authors show that progressivity also introduces significant lags in the effects of tax changes on output growth.