A number of researches have been done aiming at clarifying the demographic factors causing the decline in the total fertility rate since 1970's in Japan. Among these, decompositions of the decline using age-specific married fertility have been making people convinced that married fertility have not affected the decline and that decreasing proportion never-married solely has brought about it. We have been criticizing those decompositions as inappropriate and lately showed an alternative decomposition basing upon a cohort fertility model. On this line, we have built a more sophisticated model that reproduces the annual fertility rates by the input of nuptiality and fertility rates on cohort base. We conducted 16 simulations utilizing this new model with four cohort variables, i.e. lifetime ever-married rate, average age at first marriage, lifetime ever-married fertility rate, and average length between first marriage and births, setting these four variables either change according to the real change (including projected one) or remain stable at the value for the cohort born in 1933-34. Using the results of these 16 simulations, we decomposed the decline in TFR by the contributions of the changes in the four variables. For the input to the simulations, we prepared cohort first marriage rates and cohort fertility rates which were composed by the annual female age-specific first marriage rates and female age-specific fertility rates calculated anew not using the vital rates published by the government institutions because of the inconsistency between their numerators and denominators resulting in the distorted cohort cumulated rates. We found that the lifetime fertility rates for cohorts born after 1944-45 mostly dropped under 2 and projected 1.62 for the cohort born in 1964-65 and 1.390 in the end. Also, we projected the lifetime ever-married rate (at age 50.0) for cohorts born after 1951-52 as under 94 percent, 79.2 percent for the cohort born in 1967-68 and 71.7 percent in the end. These ultimate rates are more or less susceptible to the recent trends in rates and need to be reprojected according to the change in the trends in the future, if any. The results of the simulations revealed that more than half (56.8%) of the decline in the total fertility rate, 0.748 from 2.138 in 1970 to 1.386 in 2000 (projected) was accounted for by the decline in the level of cohort first marriage rate (increase of never-marriedness), 13.5 percent by the delay of first marriage (increase in average age at first marriage), 24.5 percent by the decline in the level of ever-married fertility (cohort lifetime ever-married fertility rate) and 5.3 percent by the delay in ever-married fertility. Therefore the contributions to the decline by the changes in marriage and by married fertility were concluded to be 7 and 3 respectively. This corroborates the importance of the policy directly influencing the couple fertility, not only that of the policy targeting the late marriage or marriage avoidance. Using the cohort fertility rates, we calculated Ryder's quantum index and tempo index for the period total fertility rates, which nicely agreed with the amount of the tempo effects of cohort fertility calculated by our simulations. Also, using the cohort fertility rates, we estimated the cumulated decline in total fertility rates applying the formula proposed by us in a recent paper and found that it matched fairly well the amount of the decline assigned to the tempo effect of cohort fertility via the simulations.

View full abstract

[in Japanese]

View full abstract