This paper studies the effects of a series of reforms of the public pension system in Austria in 2000 and 2004. An important element of the reforms was the increase in the early retirement age (ERA), which was phased in linearly over several cohorts. The empirical analysis, based on detailed administrative data, distinguishes between pension entries, which are mechanically affected by the ERA, and job exits, which reflect individual labor supply decisions. The paper presents five main findings. (1) The cohort-wise increase in the early retirement age led to pronounced shifts in the spike of pension entries at the cohort specific early retirement ages. (2) Job exits shifted in an almost parallel fashion, which leaves little room for additional substitution with other social insurance programs. (3) There is evidence of spillover effects from the reform on retirement decisions of exempt groups and on individuals who are not retiring at the early retirement ages. (4) An important mechanism leading to increased employment is that individuals keep their pre-retirement jobs longer. Finally, to quantify the effects of the reform on average retirement ages, we use a regression kink design that exploits the increasing slope in the ERA by birth cohorts and relates it to a corresponding linear increase in the labor force exit and pension claiming ages. We estimate that a one year increase in the ERA leads to a 0.6 to 1.0 year increase in the exit age for men and a 0.5 to 0.7 year increase for women. Joint with Day Manoli.