Individuals whose higher mortality is associated with
a lower socioeconomic status will, in fact, lose more in relative terms
if the age of pension eligibility is raised. The loss may not be as much in
absolute terms, since the pension levels of the higher
socioeconomic group are likely to be greater. Nonetheless, it can create
what could be considered an unfair situation.
in New Zealand can choose whether to identify as Māori for the census and other records. (Māori currently make up approxi- mately 15 percent of the total population.) In the 1990s, the
accuracy of the life tables came into question since ethnicity on
a death record was supplied by family or friends and wouldn’t
necessarily reflect an individual’s own census declaration. This
led to establishment of the New Zealand Census-Mortality
Study (NZCMS), in which death records have been matched
to census responses.
The data obtained allow one to study mortality in the context of socioeconomic characteristics reported in the census.
It always was apparent that Māori mortality was substantially
higher than that of non-Māori. A key finding from the NZCMS
over the years is that much of that difference in longevity—but
by no means all—can be explained by socioeconomic factors.
Regression analysis conducted by the NZCMS under the
direction of Tony Blakely, director of the Health Inequalities
Research Programme at the University of Otago, Wellington,
suggests that the unequal socioeconomic position between the
two ethnic groups was responsible for approximately one-third
(older adults) to one-half (working-age adults) of the present
disparity in mortality between Māori and non-Māori New
Zealanders. And this likely is an underestimate because not all
dimensions of socioeconomic position have been captured in
the regression (although other factors, such as institutional racism, disparities in access to and quality of health care, and other
culturally influenced lifestyle choices, play some part). But it’s
still possible that socioeconomic factors could explain up to 75
percent of the excess mortality.
For the purposes of a paper I gave at the 2011 Society of Actuaries Living to 100 and Beyond International Symposium, I
used the most recent non-Māori mortality table as a proxy for
higher socioeconomic status mortality. I constructed a second
comparator table as a proxy for lower socioeconomic status
mortality by adding a fraction of the difference between Māori
and non-Māori mortality. The fraction I chose was 75 percent
up to age 55, decreasing linearly thereafter to 50 percent of the
difference from age 80 onward. Calculations derived from these
data, therefore, capture real population socioeconomic mortality differences in a coherent and credible fashion.
Comparative Scenarios
It should be noted that the life tables are period tables—they
provide mortality rates as of the date of census. There’s no lack
of evidence that mortality rates are changing. But what are the
implications of different rates of improvement for different socioeconomic groups?
I addressed this question by considering four mortality scenarios. The first assumes no improvement. This is obviously
limited in its application. Expectation of life based on period
rates is a snapshot measure, not a projection. The second,
third, and fourth scenarios assume the base mortality rates—
the proxy for higher socioeconomic status mortality—decrease
1 percent per annum over all ages. This is a broad-brush approach (improvement, for example, is likely to vary by age and
cohort) but offers a reasonable first cut.
The comparator table rates—the proxy for lower socioeconomic status mortality—are then adjusted in three different
ways. For the second scenario, the comparator table rates decrease in such a way as to keep the difference in life expectancy
constant between the tables. The recalculated life expectancy at
age 20 and age 65, based on the improved rates of mortality, is
higher in each case, but the absolute difference stays the same.
For the third scenario, the rates in the comparator table are
assumed to reduce 1 percent per annum over all ages, the same
relative improvement as for the base table. Although the absolute reduction in mortality rates is greater in the comparator
table—because the rates of mortality are higher—the difference in life expectancy, under this scenario, increases. In other
words, an equal percent rate of reduction applied to the rates in
each table doesn’t reduce the absolute difference in life expectancy between the tables. Rather, the converse happens.
The fourth scenario assumes the rate of decrease in the comparator table to be 2 percent per annum over all ages—twice the
improvement in the base rates—until rates converge. In this
case, the absolute difference in life expectancy does decrease.
Just how likely are these scenarios? As I noted earlier, some
improvement in mortality generally is anticipated, regardless
of socioeconomic status. In New Zealand, projections by the
government statistician assume something like an overall
decrease in mortality rates for the population as a whole of
