We argue that transitional dynamics play a critical role in the evaluation of punitive incarceration reform on crime, inequality and the macroeconomy. Individuals' past choices related to crime and employment under old policies have persistent consequences that limit their future responses to policy changes. Novel cohort evidence is provided in support of this mechanism. A quantitative model of this theory calibrated using restricted administrative data predicts nuanced, non-monotone dynamics of crime and incarceration similar to the U.S. experience following a single permanent increase in punitive incarceration in the 1980s. Increased inequality and declining employment accompany these changes and are borne unequally across generations.