4. Data for the thirteen countries

In this chapter, we present data for each of the thirteen countries which are included in the database. For all except two of these countries, there are data for the period 1960-70, the period 1970-80, the period 1980-90 and for the cohorts born in 1871-80. In each case there are separate data for males and females, so there are eight datasets for each country. (The two exceptions are Iceland and the Netherlands, for which the cohort data are incomplete, so that there are only six datasets for these two countries).

        This is a immense amount of data, and it is easiest to show the main features graphically. We shall give two figures for each country, one for males and one for females, each showing the data for each of  the datasets concerned. We shall also give two more figures showing the results when the data for the thirteen countries are pooled. For many individual countries the data become somewhat erratic above age 100 or even 95, so the pooled data give a clearer picture at these extreme ages.

        Figure 4 plots the force of mortality at each age, on a logarithmic scale. On this scale, straight lines correspond to the Gompertz law.

        Readers will obviously wish to study the figures for the countries in which they are particularly interested, but there are a few more general observations which can be made. These concern the changes over time, the differences between male and female mortality, the differences between the countries, the relationship between the period and cohort data, and the evidence for departures from the law of Gompertz:

(a) Many of the countries show very marked falls in the force of mortality between the successive periods, and particularly between 1970-80 and 1980-90. These falls are particularly marked in Finland, France and Japan in the case of males, and additionally in Denmark, Iceland, Switzerland and Sweden in the case of females. These falls extend generally right up to age 100 and sometimes beyond. The long-held view that mortality rates above age 80 are almost immune to change is clearly untrue. More detailed studies of these falls in the force of mortality have already been given by Kannisto (1994) and by Kannisto et al (1994).

(b) In most countries, but not all, the fall has been larger at age 80 than at age 100. The main exceptions to this rule are Norway and the Netherlands.

(c ) In most countries, the difference between male and female mortality is larger at age 80 than at age 100. This is the difference between the logarithms of the force of mortality. The implication is that the sex ratio of mortality, i.e. the ratio of the force of mortality for males to the force of mortality for females, becomes closer to 1 as age increases.

(d) The figures in this chapter are not really designed for the purpose of comparing one country with another. Such comparisons have already been given by Kannisto (1994, 1996), in a study which also includes some countries in Eastern Europe. In general terms, the mortality rates in the countries which are included in the present study are comparatively close together except for Iceland, which is well below the rest.

(e) In all the countries, the cohort data are consistent with the period data. A simple way to describe the relationship is as follows. The cohorts born in 1871-80 reached age 90 in 1961-70. Thus the cohort data for age 90 should be close to the period data for age 90 in 1960-70. Similarly, at age 100 the cohort data should be close to the period data in 1970-80.

(f) It can be seen by eye that the observed values of the logarithm of mortality do not lie on exact straight lines, but fall below them as age increases. This is direct evidence that the force of mortality gradually falls below the values predicted by the Gompertz law at very high ages.

Updated by V.Castanova, 1 March 1999