The purpose of this research note is to compare between Rogers model and net migration model on the precision of population projections by prefecture in Japan and to consider potentiality of a combined model. On conditions of the same population data in 1995-2000 and the same assumption, Rogers model applies for projections generally better than net migration model does. Although we could not find any uniform difference between the two models by age-groups, there seem to be the tendency that net migration model is effective in populated regions and Rogers model be in depopulated regions. We suggested a combined model by net migration model and Rogers model and projected the regional populations again. As a result, a combined model applies generally better than the two models and better in all age-groups than Rogers model. Only if the trend of migration flows can be predicted, a combined model may be effective for increasing the precision of the regional population projections better than net migration model commonly used.

View full abstract

To make policies for dealing with a declining population and regional revitalization, local governments are using the population projections of the National Institute of Population and Social Security Research in Japan (IPSS). Some local governments are also developing and using their own population projections. Although IPSS population projections have a process which makes the sum of the population projections by region match the population projection of Japan (controlling), the population projections by the local governments lack this process. How does the process of controlling influence the results of population projection by region? How does this influence force us to be careful when using results? These are the research questions of this study. In this study, the population projections by prefecture are developed using the cohort component method with four cases which differ from the component of the projections; the launch year population and the assumptions of the future net migration rate, and with two cases whether a controlling process is occurring. The influence of controlling is analyzed through comparing these projection results. It was found that the rates of difference by age and sex when controlling is used are not consistent in magnitude and direction of adjustment by components of projections. And the amount of difference by controlling is too large to ignore, particularly in prefectures which have large populations. Conversely, in rural prefectures, comparing the population projections where the assumption of a net migration rate keeps the value observed in the base period (with no controlling influence) and that where the assumption of a net migration rate declines from the value observed in the base period (with a controlling influence) yields the result that there are many prefectures in which the latter is smaller than the former. Without understanding the influence of controlling, misunderstandings concerning the relationship between the assumption of the net migration rate and the result of the population projections can arise. Therefore, the results of this analysis show that it is important to widely distribute information about the influence of controlling on the population projections by region.

View full abstract