How sensitive are retirement decisions to financial incentives: A stated preference analysis

We analyze stated preference data on retirement. Survey respondents ofages 25 and older in the Netherlands were given hypothetical retirement scenarios describing the age(s) of (partial and full) retirement and corresponding replacement rates. Several types of retirement trajectories were considered {with retirement before, at, or after the standard retirement age (65 years), with and without gradual retirement, and with various replacement rates during partial and full retirement. The data were collected in 2006, 2007 and 2008,partly for the same respondents.The SP data are used to estimate an intertemporal utility model in which the individual’s utility is the discounted sum of within period utilities that depend on employment status (working, partially retired, or (fully) retired) and income in that period. Parameters of the utility function vary with observed and unobserved respondent characteristics and the year of data collection. The estimated model is used to analyze how retirement preferences differ by back-ground characteristics and how they evolve over the survey years. Simulating the choice of the retirement age under actuarially fair and unfair trade-offs, we then analyze how the preferred retirement age changes if pension income levels change irrespective of the retirement age (the “(pension) income effect”), or if the pension benefit accrual induced by delaying retirement changes (the”price” or “substitution” effect).Confirming most findings in the international literature, we find large effects of financial incentives on the preferred retirement age, often even larger than the effects found with revealed preferences, in line with the fact that we allow for exible choices without imposing restrictions like mandatory retire-ment at age 65. Introducing gradual retirement opportunities after the normal retirement age would stimulate participation after age 65. We find that for trade-offs involving gradual retirement, the replacement rate after full retirement is given much more weight than the replacement rate during gradual retirement. Our simulations with choices between actuarially fair retirement scenarios at ages between 60 and 70 show that an increase in life-time pension incomes by 10% would lower the average retirement age by 3 months (the “income effect”). Changing the compensation for delaying retirement fromactuarially fair to 50% of what would be actuarially fair would reduce the average retirement age by 9.7 months.