PENSIONS and COMPENSATING WAGE VARIATIONS

Publication Date01 Nov 1994
AuthorJohn Creedy
DOIhttp://doi.org/10.1111/j.1467-9485.1994.tb01139.x
Scorrish
Journal
of
Polirical
Economy,
Vol.
41.
No.
4.
November
1994
,.?
Scotlish Economic Society
1994.
Published by Blackwell Publishers.
108
Cowley Road, Oxford
OX4
IJF.
UK
and
238 Main Street. Cambridge.
MA
02142.
USA
PENSIONS AND COMPENSATING WAGE
VARIATIONS
John
Greedy*
I
INTRODUCTION
The theory
of
compensating wage variations in the labour market has been
familiar from the time of Adam Smith’s famous statements; for modern dis-
cussions
of
the theory see Smith (1979), Brown (1980) and Rosen (1986). In the
context
of
pensions it has long been recognized that comparisons
of
earnings
in different firms or occupations should recognize the existence
of
pension
schemes because they provide a form
of
deferred pay. It is also important to
distinguish between compensating variations for individuals and equalizing
differences in the market. The latter only reflect compensating variations at the
margin, where
a
decision regarding labour mobility is relevant. Empirical
studies therefore typically concentrate on the measurement
of
such equalizing
differences.
Attempts to measure wage-pension trade-offs have, however, achieved only
limited success. Studies from the early 1980s include Ehrenberg (1980), Schiller
and Weiss (1980) and Smith (1981) and used a variety of data sets and
approaches. The common feature
of
these studies was that in regressions
of
the
logarithm
of
wages on a set
of
variables, including the logarithm
of
pension
costs, the hypothesis
to
be tested from the basic theory was that the constant
elasticity
of
wages with respect to the pension should be minus unity. There was
little support for this hypothesis and in some cases the coefficient was found
to have the ‘wrong’ sign. This failure to obtain empirical support for the theory
of
compensating variations was initially attributed to the use
of
inappropriate
data and the wrong estimation method (namely ordinary least squares).
In a detailed discussion of the problem
of
estimating such trade-offs, Smith
and Ehrenberg (1983) stressed the need for employer-based data on pension
costs in order to estimate the annual increment in the present value
of
pension
benefits, for comparison with the annual wage. They also suggested that in
many schemes the pension is itself
a
function
of
wages and other pension
characteristics,
so
that when this ‘technical’ relationship is modelled
a
two-
stage-least-squares estimation procedure should be used. The third requirement
stressed by Smith and Ehrenberg was that the variables reflecting the other
factors which influence wages (such as education and age) are not usually found
in employer-based data sets,
so
that employer and employee (household) data
sets need to be combined. Using
a
special set
of
data which satisfied these
requirements, they nevertheless found no evidence
of
a significant trade-off. In
*The University
of
Melbourne, Victoria, Australia
454

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT