THE MEASUREMENT OF EARNINGS MOBILITY AMONG OCCUPATIONAL GROUPS*
| Date | 01 February 1982 |
| Published date | 01 February 1982 |
| Author | A. Markandya |
| DOI | http://doi.org/10.1111/j.1467-9485.1982.tb00437.x |
Scottish Journalof Political
Economy,
Val.
29,
No.
I,
February
1982
0
19x2
Scottish Economic Society
003~9292/82/00050000 $02.00
THE MEASUREMENT
OF
EARNINGS
MOBILITY AMONG OCCUPATIONAL GROUPS*
A.
MARKANDYA
University College, London
I
INTR~DUCTION
The stability, or otherwise, over time, of the occupational structure
of
earnings
is a subject of considerable interest. Among the many papers in the literature
that deal with this question, one can identify perhaps three underlying issues.
The first, and the one with the longest history, is the study of changes in
earnings differentials over time.' The measurement of occupational dif-
ferentials in such studies usually consists of tabulating or charting the
movement of earnings of a specific group relative to some central measure of
all earnings. Such information, though illuminating in many ways does not
provide an
overall
measure of changes in differentials. One might think that
any standard measure of dispersion, such as the coefficient of variation' would
do for this purpose, but this is clearly unsatisfactory in that it does not identify
links between groups over time. The same measure would emerge, for example,
if one took any permutation
of
the earnings of the different groups. Hence if
earnings were
xi, x2
and
x3
for three groups in the first period and
xi,
xi
and
xj
in the second period then the change in the measure of dispersion is
independent of how
xi,
xi
and
xi
are permuted between the groups
1,2
and
3.
Clearly, however, such a permutation affects differentials in most interesting
cases.
The second issue that emerges from this literature is the importance of the
length of the time period over which earnings are measured on the observed
inequality of earnings. It is easy to see that the length of this accounting period
can affect the measurement of inequality-random variations will tend to get
averaged out and secular trends more
fully
incorporated as the period is
lengthened. The relationship between short and long period inequality has
*I
would like to thank Holly Sutherland for computational assistance, and David Ulph
for
helpful conversations on this subject.
All
errors are my own.
'
The literature on occupational earnings differentials and their determinants is truly vast.
A
list
of
the postwar papers
in
this field would be incomplete, however, without reference to Lebergott
(1947), Slichter (1950), Bell (1951), Reder (1955), Keat (1960), and Scitovsky (1966).
A
summary of
many of the relevant results is to be found in Phelps-Brown (1977). More recently Dean (1978),
Eliot and Fallick (1979), Metz and Stager (1979), Quinn (1979) and Brown
et
nl.
(1980) have
addressed themselves to the same question.
Keat (1960)
for
example uses the coefficient
of
variation as a measure
of
earnings differentials.
Date
of
receipt
of
final manuscript
:
2
July 1981.
75
To continue reading
Request your trial