Revisiting the effect of statutory pension ages on participation and the average age of retirement in OECD countries
https://doi.org/10.3326/pse.45.2.4  Published online: June 6, 2021 Figure 1
Comparing policy effects of a stylised shift in participation and econometric predictions Effect of a oneyear increase in statutory retirement ages on participation rates, German males, 2015 Note: Both panels illustrate the estimated effect of a oneyear increase in both the statutory minimum and normal retirement ages. Panel A uses a stylised shift in the actual ageparticipation profile, whereas panel B uses the baseline pooledcountry estimated equation reported in Geppert et al. (2019). The size of the effect on labour force participation and the average retirement ages is proportional to the shaded area. Source: Authors’ calculations. Figure 2
Age fixed effects in the baseline model. Effect on labour force participation at different ages (percentage points) Note: Age fixed effects from the baseline model, taken as equation [1] in table 1 of Geppert et al. (2019). Source: Authors’ calculations. Figure 3
Sensitivity of the age of retirement to a change in statutory retirement ages. Estimated effect on the average effective retirement age of an increase in the statutory retirement age of one year Note: The number in brackets at the start of each label on the xaxis refers to the equation number in table 1. Successive bars in each panel show the effect of changing one characteristic relative to previous bars. Figure 4
Modelling of statutory retirement ages Note: The figures illustrate alternative ways of modelling statutory retirement ages when the minimum retirement age is 60 and the normal retirement age is 65. Source: Authors’ calculation. Table 1
Variant pooled estimations explaining labour force participation
Notes to Table 1: All equations are estimated with OLS. All equations include country fixed effects, gender fixed effects, an education fixed effect, a gendereducation interaction effect, genderage interaction effect and an educationage interaction. No equations include time fixed effects. The unemployment gap is a fiveyear moving average. The pension wealth variable is described in more detail in Geppert et al (2019), as are other variables in the baseline model (1). Model (1) is the baseline model in Geppert et al. (2019) and is the only equation to include age fixed effects. Model (2) = Model (1), but replaces age fixed effects with a linear age variable. Model (3) = Model (1), but replaces age fixed effects with a quadratic age variable. Model (4) = Model (2), but replaces dummy variables for ages above the minimum and normal retirement ages with a ‘transition to retirement’ variable, which is a function of age and takes a value equal to: zero for ages below the minimum retirement age; one for ages above or equal to the normal retirement age; and for ages between the two statutory ages it assumes a value between zero and one proportional to the distance between them (see the main text for further explanation). Model (5) = Model (4), but differentiates the coefficient on the ‘transition to retirement’ variable for three groups of countries through the addition of a variable which is the product of the transition variable and two distinct dummy variables. A first dummy variable is created for countries (Canada, Ireland, United Kingdom and United States) where voluntary private pension systems are important. A second dummy variable is created for countries where there is some evidence of the prevalence of early retirement outside the oldage pension system (France, Greece, Italy, Poland, Netherlands and Spain). The third group of countries, referred to in the lower part of the table as the ‘Majority of countries’ have no dummy variable. Model (6) = Model (5), but the early retirement coefficient on the ‘transition to retirement’ variable is differentiated only for individuals with low or medium education. Two additional variables are added for individuals with low or medium education in early retirement countries only, namely: 1year and 2year pipeline to retirement variables defined using dummy variables at the age two years and one year preceding the minimum age of retirement, respectively. Model (7) = Model (6) but with quadratic, rather than linear, age effects. Model (8) = Model (6) but estimated on the sample 200017. Model (9) = Model (7) but is estimated on the sample 200017. Model (10) = Model (6) but estimated on the sample 201017. Model (11) = Model (6) but is estimated only on the year 2015. The basis for the calculations in the lower part of the table summarising the implied effect on the participation rate of those aged 5574 and on the average age of retirement from increasing the statutory age of retirement by one year is given in the Annex. Figure 5
The simulated effect of eliminating early retirement pathways. Stylised ageparticipation profile of low and mediumeducated workers Note: The chart compares a stylised ageparticipation profile of low and mediumeducated workers in countries classified as ‟early retirement” countries with those in the majority of countries. The profiles are generated using equation (6) in table 1, using: differential responses to the minimum and normal statutory ages of retirement, which for the purposes of this example are assumed to be 60 and 65: the pipeline effects (for early retirement countries only); common linear age effects; a constant difference between the two groups of countries equal to the difference in the average country fixed effects for each group. The black dashed line (and black shaded area) simulates the effect of eliminating early retirement pathways on the assumption that the response of early retirement countries to statutory ages becomes the same as the majority of countries and the pipeline effects are eliminated. A shift to the gray line (and gray shaded area) further assumes that eliminating early retirement pathways would also imply the average country fixed effect for early retirement countries becomes the same as for the majority of countries. These two effects combined would imply an increase in the average participation rate for the age group 5574 of 6 percentage points and an increase in the average age of retirement (ΔAAR) by 20 months. Figure 6
The evolution of labour force participation and retirement ages in Germany Source: OECD (2019), Eurostat and authors’ calculations. Figure A1
Older age population distribution of the average OECD country. Size of age group as a percentage of age group aged 5574 Note: Shares are calculated as unweighted averages of OECD countries. Source: United Nations population estimates and projections. This is an Open Access article distributed under a Creative Commons AttributionNonCommercial 4.0 International License (CCBYNC) which permits non commercial use and redistribution, as long as you give appropriate credit, provide a link to the license, and indicate if changes were made.

June, 2021 II/2021 