우리나라 소지역 인구추계를 위한 방법론 비교 분석

Abstract

Despite the recent increase in interest in population projections for small-areas such are si-gun-gu level units in South Korea, methodological studies on related matters are being conducted very limited. Therefore, the main objective of this study is to compare and analyze the characteristics of various methodological alternatives for small-area population projections using observation data. To this end, based on a review of previous studies on population projections for small areas, major methodologies were selected. In performing population projections for the selected methodologies, 20 execution-level methods are derived according to the applicable detailed plans or settings. For each method, population projections were performed targeting for year 2020 based on data from 2001-2010 at the si-gun-gu level in South Korea, and the results were analyzed. First, the compromise-share method and the biregional cohort-component method turned out to be almost equally superior in terms of the overall performance for the total population projection. Second, other characteristics were revealed when the estimation results were further examined in terms of population structure and spatio-temporal heterogeneity. The results of the biregional cohort-component method were excellent except for the population aged 0-4 and 25-34. In the short-term projections (5 years), the compromise-share method was the best,but the increase in error due to the increase in the estimation period was the lowest in the biregional cohort-component method. Areas with an error of more than 10% accounted for about 25%. They often experienced a great migration pattern change or a serious fluctuation in population growth rate during the estimation period. This study is expected to be a major reference in conducting population projections for various small-areas by presenting an empirical basis for the small-area population projections.