Regional differentials of death rates and their effects on an accuracy of a regional population projection in Japan
Keita Suga, National Institute of Population and Social Security Research, Japan
We assess the role of mortality assumption under the population aging on the accuracy of population projections for the case of Japanese municipalities. Japanese society is the most advanced in terms of population aging in the world in 2010 and further population aging and decline is anticipated. Japan is not the only country predicted for a future population decline, but more than one half of European countries is projected to lose their population for 2035-40. The fundamentals underlying the future population decline focus on a relative increase in the gross death rate due to population aging, given a level of fertility and a scarce international migration. The extent that mortality assumptions influence the projection results will be expanded when population ages and under its consequence of a population decline. There are substantial variations in population aging among Japanese municipalities. This variation provide a source for the assessment of mortality effects on an accuracy of population projections to derive implications to future population projection for not only Japanese regions but also countries for which population aging and declines are anticipated. Related with the mortality assumption, we identify four potential sources that cause discrepancy in population projections. We conduct a simulation of population projections by distorting one of the four discrepancies each time separately with keeping other components at the realized values, then we compare the projected population with the census population realized ex-post. Results show that the forecast errors for the survival rates in 5 years beyond are negligible for most of regions and at most 5% even for municipalities with less than 1,000 inhabitants.
Presented in Poster Session 1