II.Summary of the Methodology and Assumptions for the Population Projections


1. Period of Projection
The period of projection is 50 years, from 2001 to 2050.

2. Method of Projection
The cohort component method is used for this projection, as with the previous report. This method takes into consideration international migration while calculating the ages of the existing population using the future life table. It also uses the future fertility rate to calculate future births and obtain the number of survivors for the population that is expected to accrue. Five items, (1) base population, (2) future survival rate, (3) future fertility rate, (4) future sex ratio at birth, and (5) future international migration numbers (rates), are required to project the population using the cohort component method.

3. Base Population
As for the starting point of the projection, called the base population, the male and female population figures, classified by age group (including non-Japanese residents) as of October 1, 2000, excerpted from the Population Census of Japan compiled by the Statistics Bureau, Management and Coordination Agency, were used. The "age unknown" figure was distributed over all age groups.


4. Assumption of the Survival Rate (Future Life Table)
In order to project the population from one year to the next, survival rates by age and sex are needed, and, to obtain future survival rates, future life tables may be constructed.

There are three main ways to construct a future life table: the empirical method, the mathematical method, and the relational model method. This projection has adopted the Lee and Carter model, which is based on the relational model, and modified it to suit the purpose of this projection. The Lee and Carter model describes age-specific changes of mortality with a single index of mortality by decomposing a matrix of age-specific death rates into the 'average' mortality age schedule, the general level of mortality (mortality index), the age-specific changes in the mortality schedule when the general level of mortality changes, and an error term. Non-linear curves were fitted to the data after 1970 in order to reflect the changes in the level of mortality that had gradually been easing off during the past 30 years. The data for 1995 were excluded due to the influence of the Great Hanshin Earthquake, the future life tables were constructed separately for 2001 because of the very low reported death counts in February of that year, and the final fittings were done. The future life tables were constructed from the assumed age- and sex-specific death rates until 2050, based upon the parameters obtained through the above procedures.

According to the future life tables, the life expectancy, 77.64 years for males and 84.62 years for females in the year 2000, is expected to extend to 78.11 years for males and 85.20 years for females in 2005, 79.76 years for males and 87.52 years for females in 2025, and, in 2050, 80.95 years for males and 89.22 years for females (see Table 6, Figure 6).

5. Assumptions of Fertility Rates
The age-specific fertility rates are required in order to project the number of births in future. There are two methods used to estimate future fertility rates: the period-fertility method and the cohort-fertility method. The latter was adopted for this projection. The cohort-fertility method observes the birth process per female birth cohort on an annual basis, and forecasts the level of completed fertility1 and the birth timing for cohorts in which the birth process is incomplete. The age-specific fertility rate on an annual basis and total fertility rates can be obtained by converting the estimated cohort fertility data into annual data. Due to an extensive uncertainty in future fertility, three assumptions (medium, high, and low variant projections) are compiled and fertility rates are projected for each of them.

(1) Assumption for the Medium Variant

  1. The mean age of marriage for cohort has advanced from 24.4 years for the cohort born in 1950 to 27.8 years for the cohort born in 1985; this tendency will not change for cohorts born in 2000 and after.
  2. The proportion never married has advanced from 4.9% for the cohort born in 1950 to 16.8% for the cohort born in 1985; it will not change for cohorts born in 2000 and after.
  3. The completed number of births from married persons is affected by later marriage, later childbearing, and changes in reproductive behavior of couples; it has advanced from 2.14 for the cohort born between 1948 and 1952 to 1.72 for the cohort born in 1985; it will not change for cohorts born in 2000 and after.
  4. The distribution of completed fertility among females appears as follows, and remains consistent for cohorts born in 2000 and after.


In this case, the total fertility rate will decline from 1.36 in 2000 to 1.31 in 2007. Thereafter, a gradual upward change is predicted, and in 2049 the rate will be 1.39(see Table 7, Figure 7).

(2) Assumption for the High Variant

  1. The mean age of marriage for cohort has advanced from 24.4 years for the cohort born in 1950 to 27.3 years for the cohort born in 1985; this tendency will not change for cohorts born 2000 and after.
  2. The proportion never married has advanced from 4.9% for the cohort born in 1950 to 13.3% for the cohort born in 1985; it will not change for cohorts born in 2000 and after.
  3. The completed number of births from married persons is affected by later marriage and later childbearing; it has advanced from 2.14 for the cohort born between 1948 and 1952 to 1.93 for the cohort born in 1985; it will not change for cohorts born in 2000 and after.
  4. The distribution of completed fertility among females appears as follows, and remains consistent for cohorts born in 2000 and after.

In this case, the total fertility rate will turn upward immediately from 1.36 in 2000, reaching 1.63 in 2049(see Table 7, Figure 7).

(3) Assumption for the Low Variant

  1. The mean age of marriage for cohort has advanced from 24.4 years for the cohort born in 1950 to 28.7 years for the cohort born in 1985; this tendency will not change for cohorts born in 2000 and after.
  2. The proportion never married has advanced from 4.9% for the cohort born in 1950 to 22.6% for the cohort born in 1985; it will not change for cohorts born in 2000 and after.
  3. The completed number of births from married persons is affected by later marriage, later childbearing, and changes in reproductive behaviors of couples; it has advanced from 2.14 for the cohort born between 1948 and 1952 to 1.49 for the cohort born in 1985; it will not change for cohorts born in 2000 and after.
  4. The distribution of completed fertility among female appears as follows, and remains consistent for cohorts born in 2000 and after.

In this case, the total fertility rate continues to decline from 1.36 in 2000, reaching 1.10 in 2049 (see Table 7, Figure 7).

6. Assumption of Sex Ratio at Births
As for the sex ratio at births which divide the future number of newborns into male and female, the female to male ratio is set to 100:105.5 based on the results of the last five years, and remains consistent from 2001 onward (see Table 8, Figure 8).

7. Assumption of International Migration
International migration varies according to advances in internationalization and economic activities of Japan. Additionally, it is affected by the policy concerning international migration and the economic and social conditions of other countries.

For the past projections of international migration, it was assumed that the age-specific net (entries minus exits) international migration rate by sex was constant. However, the international migration trend differs for Japanese and non-Japanese population. Additionally, migration, especially non-Japanese migration, does not depend on the population size and the structure of Japanese population. This projection has different international migration figures for Japanese and non-Japanese population. That is, two assumptions are made: the age-specific net international migration rate for Japanese population and the amount of net (entries minus exits) international migrants for non-Japanese population.

Because international migration for Japanese population is relatively stable, the assumptions are made as follows: first, obtain the average value of the annual net international migration rate between 1995 and 2000, and adjust the rate to remove the blurring which occurs due to random fluctuation as constant for 2001 onward. Because the parent population of migration is Japanese, projection of the Japanese-only population is required. This population is calculated by multiplying the projected age- and sex-specific population and the proportion of Japanese population (obtained from the Population Census of Japan in 2000 and the number of births in the Vital Statistics).
As for international migration of non-Japanese population, net-migration is more or less in excess and tends to be increasing, so a regression line is applied per sex for results from 1970 and after. However, for the years around 1990 when drastic fluctuations occurred, the years except 1988 to 1995 when the divergence from the overall tendency is apparent are used and extrapolated using logistic curves, thereby obtaining the excess net-migrations per sex of non-Japanese population in the future. The age distribution of net-migrants is fixed as the average value between 1995 and 2000 (see Figure 9 to Figure 11).


Return to the Contents

Go back to the IPSS top page