Exogenous retirement has no impact on the qualitative predictions of the model. Quantitatively, the volatility of hours worked is higher compared to Modell for two reasons. By extending the working period, agents benefit from education for a longer period of time, which increases the incentive to substitute work with schooling. Further, the volatility of hours worked of older agents is higher. Thus, aggregate volatility increases as we include older agents. However, with endogenous retirement, the fraction of retired people to active population generated by the model is too high compared to the data. For this reason, in our preferred specification, retirement is mandatory. This allows US to correctly match the number of retired people relative to workers. It is also worth mentioning that the average labor force participation rate for individuals aged 65 and over is 14% for the period 1962-2012.17 Therefore, by setting exogenous mandatory retirement at age 47 in the model (and 64 in reality), only a small fraction of the population is excluded from our analysis.