Relabelling, retirement, and regret

Jonathan Gruber, Ohto Kanninen, Terhi Ravaska 22 November 2020



Developed countries around the world face enormous long-run deficits in their public pension systems. As a result, pension reform is a constant source of public-policy debate. A wide variety of reform strategies have been contemplated and/or implemented around the world. 

A common approach to addressing such fiscal deficits is to reform the underlying structure of pension plans – most notably by changing either the ‘early retirement age’ at which individuals can first qualify for benefits or the ‘normal retirement age’ around which benefit determination is centred.

A large literature shows a causal relationship between these ages and ‘excess retirement’ (Gruber and Wise 1999, Börsch-Supan and Coile 2019, Manoli and Weber 2016, Seibold 2019). What is less well understood, however, is the reason for such spikes in retirement hazards. 

In a standard model of optimising retirement behaviour, what should matter is the financial incentives to retire at a particular age. However, existing evidence suggests that financial incentives alone cannot explain these spikes; the ‘excess’ retirement at legislated ages suggests that there are other behavioural mechanisms at play in driving retirement. 

Contemporaneous changes in retirement ages and changes in financial incentives make it difficult to measure convincingly the impact of the retirement-age norm itself. Discerning the effect of the age norm requires labelling of retirement ages to change without an associated change in financial incentives. If changing retirement behaviour is possible through extreme ‘nudges’ like retirement age relabelling, working lives could be extended without hurting those who need to retire earlier.

We study a reform in Finland that allows us to separate financial incentives and norms associated with retirement age (Gruber et al. 2020). Before 2005, retirees in Finland faced an early-retirement regime which ran from age 60 to age 65, with normal retirement at age 65. A reform in 2005 introduced a new ‘flexible’ retirement age at 63 – opening up a new and unanticipated retirement route at ages 63 and 64. 

While the reform also included changes in financial incentives, these changes were both modest and more continuous across cohorts than was this relabelling, allowing us to separate the financial incentives from the norms. We analyse the impact of this reform using data on 100% of the workers in Finland, which allows us to include large samples of workers at and around the key retirement ages at the time of reform.

The key findings are illustrated in Figure 1. The figure shows age at the time of the reform on the horizontal axis and percentage changes on the vertical axis. We demarcate three areas through vertical dashed lines: those who were too young to be relabelled, those relabelled at some point over the next 12 months (to allow a focus on annual retirement), and those who were already too old to be relabelled.

Figure 1 Pension incentives, labelling, and retirement rates in 2005 vs 2004

Notes: Pension wealth increased on 1 January 2005 due to the reform as a function of age. The reform changed the marginal accrual rate as a proportion of accrued pension calculated for a 12-month period as a function of age, earnings, and accrued pension. Relabelling is defined as reaching full retirement age due to the 2005 reform in 12 months. The means are estimated for bimonthly age bins. Retirement is measured as claiming old-age pension and estimated as a t-test of the difference in 2005 and 2004 for monthly birth bins. Error bars show 95% confidence intervals.

The figure shows the key financial changes that were associated with reform. The solid line shows the reform-induced change in pension wealth, which jumps up at age 62 due to a much larger pension bonus for those working past age 62, rises slowly to age 63, then declines steadily. The dashed line (and associated confidence interval) shows the reform-induced pension accrual rate (the percent rise in pension from working an additional year), which moves only slightly over time because the large financial bonus at age 62 was offset by removing a previous bonus for retiring before 65.

The points, and associated confidence intervals, show the difference between the 2005 cohort (which was eligible for reform) and the 2004 cohort (which was ineligible) in their retirement rates at each age. 

The impact of relabelling is clear. There is little pre-existing difference in retirement rates before age 62; there is then an enormous jump in the retirement-rate difference, from near zero to 40%. The differential retirement rate remains escalated in the entire relabelling range, with no clear pattern. It then immediately jumps down once the relabelling regime is over (once the retirement age labels are the same in both years). After age 64, the retirement rate remains somewhat lower, as those who remain in the labour force in the relabelled cohort are less likely to retire at each age.

The implied effect of relabelling can be computed most cleanly by comparing the red and green dots right before and after the discontinuity at age 64. We estimate that relabelling led to a 40pp change in the retirement hazard. We confirm this magnitude using hazard models that exploit cross-individual variation in financial incentives, relabelling dates, and the exact month of retirement.

Individuals who face self-control problems may retire excessively relative to their long-run preferences, leading to ‘regret’ among early retirees. Quantitatively measuring such regret is difficult in most settings, but in ours, we have an excellent revealed preference measure of regret: marginal increases in return to work among those induced to retire early by the reform. If we see an increase in return to work due to relabelling, it is more consistent with regret over retirement for the incremental workers who are induced to retire by relabelling. 

And in fact, we find exactly that: workers who are incrementally retiring due to the reform are 50% likelier to return to the labour market relative to the baseline. Such a pattern is consistent with individuals regretting their decision to respond to the rebelling of retirement dates.

Focal retirement ages are a central feature of social security programmes around the world. These focal ages can play an important role in setting retirement expectations and norms. We find that retirement-age relabelling is a particularly powerful determinant of date of retirement. Both graphical evidence and estimated hazard models reveal an enormous change in retirement when individuals face a newly defined ‘normal retirement’ age. 

Our findings suggest that such relabelling is as powerful as vast changes in pension wealth or dynamic pension retirement incentives. But we also find that there is a marginal increase in regret among those who respond to this change in retirement ages – suggesting a potential source of welfare loss from inducing excess retirement.


Börsch-Supan, A, and C Coile (eds) (2018), Social security programs and retirement around the world: Reforms and retirement incentives, Chicago: University of Chicago Press.

Gruber, J, and D A Wise (eds) (1999), Social security and retirement around the world, Chicago: University of Chicago Press.

Gruber, J, O Kanninen and T Ravaska (2020), “Relabeling, retirement and regret”, NBER Working Paper 27534.

Manoli, D, and A Weber (2016), “Nonparametric evidence on the effects of financial incentives on retirement decisions”, American Economic Journal: Economic Policy 8(4): 160–82.

Seibold, A (2019), “Reference dependence in retirement behavior: Evidence from German pension discontinuities”, CESifo Working Papers 7799.



Topics:  Welfare state and social Europe

Tags:  Finland, pensions, Retirement, retirement age, Social security

Ford Professor of Economics, MIT

Research Coordinator, Labour Institute for Economic Research

Senior Researcher, VATT


CEPR Policy Research