(a) The age data used are by single years; hence, efforts are made to take into account changes in the age pyramid caused by the past annual fluctuations of births, deaths and overseas migration. The analysis by periods of five years may blur ups and downs of fertility and mortality for a country like Japan, thus rendering the conclusions crude and approximate;
(b) Various time spans are examined in order to analyze every possible combination of time periods;
(c) The rates obtained from forward and backward standardization are averaged so that the residual interaction term can be eliminated.
Table 5 shows the results of decomposition for various time periods 5-year, 10-year, 20-year, 30-year and 35-year.
In the analysis of five-year periods, apart from the effect of the initial age distribution, the effect of fertility was larger than the effect of mortality in the periods 1950-1955 and 1965-1970. The effect of the initial age distribution (Col. 7) is the effect of the previous age distribution, or the cohort effect. It is obvious that, where the baseline age distribution does not have a smooth profile and is characterized by bulges and troughs, such irregularities often determine a good portion of the proportion of the elderly. To be sure, however, that effect of population distribution may itself be attributable to previous histories of fertility and mortality.
However, in the later years under analysis, nearing the year 1985, the effect of mortality becomes larger than that of fertility, again apart from the effect of the age distribution of the initial population. It is argued that, in the earlier half of the 35-year period between 1950 and 1985, the effect of fertility was definitely larger. In contrast to the common implications of the theory of stable population, however, population ageing (in terms of an increase in the proportion of the elderly 65 and over) has been promoted by a decline in overall mortality and consequently by the prolongation of life expectancy, even in the middle of demographic revolution. According to the conventional interpretation of the theory of stable population, fertility is the predominant force causing population ageing, while mortality decline generally has little effect or, at best, promotes population rejuvenation.
If a longer period is used to decompose the change in the percentage of the population aged 65 and over, the effect of fertility is always larger than that of mortality： see, for example, the 35-year time period between 1950 and 1985 in Table 5, where the effect of fertility was even larger than the effect of the initial age distribution.
