Early retirement schemes (2)

1
Early Retirement Behaviour in the
Netherlands Evidence from a Policy Reform
Data Model In the empirical analysis we use the
Income Panel data set (IPO) of Statistics
Netherlands. This data set is based on
administrative data from the Dutch National Tax
Office and concerns the period 1989 2000. The
data set has a panel structure and contains
information about 75 thousand individuals. We
only consider the individuals aged between 55 and
65, which are participant in one of the selected
pension funds and which are not left-censored.
This results in a sample of 2973 individuals. To
estimate the effect of the transition to a
prepension scheme, we use a duration model. The
duration of an individual is defined as the time
that elapses between his 55th birthday and his
moment of retirement. We assume a mixed
proportional hazard form and we estimate both the
baseline hazard and the unobserved heterogeneity
nonparametrically. A actuarially fair prepension
scheme does not provide incentives to retire at
specific ages. A VUT scheme however provided two
different incentives an incentive to postpone
retirement to individuals younger than the
standard retirement age and an incentive to
retire to individuals older than the standard
retirement age. Therefore, we use two dummy
variables to model the effect of the transition
from VUT to prepension incentive to wait and
incentive to retire.
Abstract The Dutch labor force participation rate
of elderly is among the lowest of Europe and
early retirement schemes play an important role.
Already in the early 1990s, unions and employer
organizations recognized the adverse incentive
effects of the generous and actuarially unfair
PAYG schemes and decided to transform these to
less generous and actuarially fair capital funded
schemes. The starting dates of the transitional
arrangements varied by sector. In this study, we
exploit the variation in starting dates to
estimate the impact of the policy reform on early
retirement behavior. We use a large
administrative dataset, the Dutch Income Panel of
the National Tax Office, to estimate hazard rate
models for early retirement. We conclude that the
policy reform induces workers to postpone early
retirement. Model simulations show that the
transitional scheme has already led to average
retirement postponement by 8 months, which will
become almost a year once the transition is
completed.
Early retirement schemes (1)
The first early retirement scheme in the
Netherlands was introduced in the late seventies
and was called the VUT scheme. The exact
conditions of this scheme varied by sector, but a
typical scheme provided a benefit from age 60
until 65 that was equal to 80 of the last earned
wage. This replacement rate did not increase if
retirement was postponed, so the scheme was not
actuarially fair.
Results
The table on the right shows the estimation
results for the baseline hazard and the two dummy
variables. Incentive to retire is significantly
positive, which means that individuals aged 60
and over had a higher hazard rate in the VUT
scheme than in the prepension scheme. On the
other hand, incentive to wait is not significant,
implying that the transition had no effect on
individuals aged between 55 and
60. Specifications with a double baseline (one
for the VUT scheme and one for the transitional
scheme) or with financial variables like the
pension wealth, the peak value and the option
value were also estimated, but the performance
of these models was less.
Coefficient Coefficient Std error
Age 55 -5.70 (0.60)
Age 56 -5.67 (0.63)
Age 57 -5.09 (0.62)
Age 58 -5.41 (0.63)
Age 59 -4.47 (0.63)
Age 60 -3.04 (0.61)
Age 61 -0.50 (1.14)
Age 62 -1.41 (1.18)
Age 6364 -1.72 (1.25)

Incentive to retire 2.56 (0.32)
Incentive to wait 0.32 (0.21)
The VUT scheme was financed by a pay-as-you-go
(PAYG) system, which implied that its
sustainability was threatened by the ageing of
the Dutch population. Because of this, unions and
employer organizations decided to transform the
schemes to capital funded schemes, the so-called
prepension schemes. These schemes are less
generous the standard retirement age is higher
and the replacement rate is lower (see figures).
Early retirement schemes (2) To spare older
workers, the pension funds introduced
transitional schemes in which the replacement
rate depended on the date of birth. The moment of
introduction and the exact conditions varied by
sector. The table below shows the replacement
rates in the VUT scheme, the transitional scheme
and the prepension scheme for ABP, the pension
fund of the government. Note that the VUT scheme
provided a clear incentive to retire at age 60,
whereas the prepension scheme is actuarially
fair. A similar table is available for six other
pension funds, which did not introduce a
prepension scheme during the period of
investigation. We use the participants of these
pension funds as a control group.
VUT Transitional Transitional Transitional Transitional Prepension
Date of retirem. lt Apr.97 97-03  97-03  gt Apr.03 gt Apr.03 gt Apr.97
Age/Date of birth   lt Apr.42 42-47 lt Apr.42 42-47 gt Apr.47
55 0 27 25 27 25 18
56 0 30 28 30 28 21
57 0 35 32 35 32 24
58 0 40 38 40 38 28
59 0 48 45 48 45 33
60 80 59 55 59 55 40
61 80 75 70 75 70 51
62 80 75 70 100 93 70
63 80 75 70 100 100 100
64 80 75 70 100 100 100
Rob Euwals (CPB, IZA, CEPR), Daniel van Vuuren
(CPB, VU University Amsterdam), Ronald Wolthoff
(Tinbergen Institute, VU University Amsterdam),
wolthoff_at_tinbergen.nl