This suggests that there are some common features of these industries such as the scale
of production or the type of skills employed which make them more amenable to
unionisation.
There were, however, some important differences between the countries. The US
samples of both men and women were more highly educated than the samples of the
other countries. The aggregate result for men of a steeper age earnings profile in the US
than in the other countries cannot be explained solely by the mix of educational
endowments. There was greater variation in earnings with age for American men for
most education groups taken separately. The earnings for each education group in the
US continued to grow with age for longer than in the other countries.
The evidence we have presented here also shows greater variation in average
industry earnings for men in the US than in the other countries. It appears that industries
which are high paying on average are also relatively high paying for young men. There
was greater variation in average industry earnings for these young men in Great Britain
and the US than in Australia.
The relationship between age and earnings was different for men than for women
in each country. Womens' earnings varied less with age than did mens' particularly for
women over thirty. In general, for each education category there was more variation in
earnings with age for women in the US than in Australia or Great Britain. There was
also more variation in earnings by industry of employment in the US than in the other
countries.
The preceding chapter set out various theories which aim to explain why earnings
should differ with age. Among the important potential explanators were educational
attainment, industry of employment and the role of unionisation. This chapter has set out
the facts on the relationship between these variables and earnings for our samples of
men and women working full-time in the three countries. While there were many
similarilties between the countries in terms of the effect of education and industry of
employment on earnings, a general result was that earnings showed greater variation in
the US than in the other countries. The remaining chapters look firstly at the
8 0
results of a comparison of the relationship between age and earnings for men and
women in each country.
Footnotes
1. We have outlined in the preceding chapter the major explanations offered for the
positive relationship between age and earnings. While these in general do not consider
age of itself to be important in explaining earnings, working experience or time spent
with a particular employer are central to these theories. These factors are closely related
to age.
2. See Bowman (1987) for a discussion of cohort, period and age effects on individuals.
3. The full definitions of the variables are presented in Appendix A.
4. The income data in the Australian Census is presented in categories while the data for
the other countries is for actual earnings. The mid point of each category was chosen
and an estimate made for the open-ended category. The 1981 Income and Housing
Survey for Australia presents the actual data on income but categorises age. The age
earnings profiles for full-time men derived from this sample showed a similar profile for
men to the Census although the level of the profile was somewhat higher. As the general
story from the data was similar for the two Australian data sets, it was decided to opt for
the larger Census data set using actual age rather than age by category.
5. In all countries, there appeared to be a decline in average weekly earnings for men at
the end o f their working lives but in the US, the fall was smaller than in Australia and
Great Britain. At the age of sixty four, the American average weekly earnings were
twelve per cent below their peak but in Australia and Great Britain, they were sixteen
and eighteen per cent respectively below the peak. It is important to remember that
although cross section age earnings profiles typically show a decline in earnings at the
end of working life, this need not imply that earnings are reduced with age for each
(college graduates, high school graduates and elementary school graduates), that there
has been "no systematic tendency for time series profiles [of the earnings of a cohort] to
decline even though cross section ones do" (p219). Hanoch and Honig (1986) in
contrast, concluded that there was a decline in earnings among older men in their US
sample which reflected a reduction in hours worked. There are several reasons why the
aggregate cross section age earnings profile may show lower average earnings for older
workers. As older workers are less well educated than younger ones, this can create the
appearance of a declining age earnings profile in the cross section. The average earnings
of full-time workers over sixty may also be effected by selective retirement if those with
a greater earnings potential tend to retire earlier than those without. However, looking at
an individual’s earnings over time, economic growth may prevent a decline in their real
value.
6. In another comparison of age earnings profiles between countries, Hashimoto and
Raisain (1985) found that Japanese male workers had steeper earnings profiles than
those of American male workers. They argued that the growth in earnings attributable to
the period spent with one employer was greater in Japan than in the US.
7. A close correlation between earnings by industry in Australia and the United States
was also found when a more disaggregated industry breakdown was used by Gregory
and Daly (1990). This result held for both men and women.
8. See Metcalf (1990) for a discussion of some of the changes in British industrial
relations in the 1980’s.
9. The issue of selectivity bias in estimated earnings regressions for women is one
which has received considerable attention in the literature. We discuss this question
further in chapter 7 footnote 5.
10. The results for women aged 60 to 64 should be treated with particular caution. The
number o f older women in the samples was rather small, especially in Great Britain
where the state pension became available to women at age 60 encouraging earlier
follows, there were also some small numbers of observations in some of the cells,