台灣人口老化之空間分析

Abstract

隨著社會經濟之變遷，台灣嬰幼兒死亡率降低、出生率下降、高齡人口死亡率減少、經濟富裕、醫藥發達，人民平均壽命逐漸延長，自1993年起65歲以上老人佔總人口比率達7.10%，正式步入高齡化社會。使得研究人口老化之學者如雨後春筍般地湧出，過去研究老化因素之學者曾藉由歷史脈絡、社會經濟因子等試圖找出人口老化之因素，但卻未考量地理空間之分佈對老化之影響。本研究結合地理學之空間概念以及人口學之靜態觀點與其他社經變數予以分析，並關注最小平方法後之殘差，其空間分佈是否為隨機分佈，有無產生空間相依性或空間異質性之問題。 本研究發現各年度年齡結構指數與社經人口變項之關係稍有不同，但與高中職人口比呈負相關、與性別比比呈負相關、與有偶且同居人口比呈負相關、與淨遷徙率呈正相關則同時存在於兩年度之資料中，表示高中職人口比、性別比或有偶且同居人口比越高，年齡結構則越老年化；而淨遷徙率越高，年齡結構則越年輕化。此外，光以傳統的最小平方法來分析空間資料略顯不足，受到空間相依性與空間異質性之影響，將會使得其估計結果有所偏誤、不具有效性。故對模型進行檢測，發現在空間相依性方面，以空間落遲模型優於最小平方法；而在空間異質性方面，則以地理加權平均優於最小平方法。

Along with the development of social economy, mortality rate and birth rate reducing, economy wealthily, medicine is developed and the growth of life expectancy. Above 65 years old person divided by the total population reach 7.10% in 1993, officially marched into aging population society. So there were many scholars researched aging population, in the past studied scholar of aging population focused on population’s demographic history, the social economy factor and so on, attempted to discover the factor of the aging population. But actually they had not considered the distribution of the geography space. Therefore, this article gathers the geography and the static demography concept to analyze, and some factors of economy and society. And test the spatiall pattern of Ordinary Least Square’s residul, if it is random. If the answer is not, two broad classes of spatial effects may be distinguished, referred to as spatial dependence and spatial heterogeneity. This research discovery: the relationship of Age structure index and the variables of population and society have slightly difference in 1990 and 2000. But something are the same. The relationship of high education ratio, sex ratio and copule rate between age structure index are negative. And the realationship of net migration ration between age structure index is positive. Indicate that less high education ratio, sex ratio or copule rate and more net migration ration will increase Age structure. In addition, the residul having spatial pattern is not good to use Ordinary Least Square. Therefore carries on the examination to the model, discovered in the spatial dependence aspect, Spatial Lag model is better than Ordinary Least Square. Otherwise, in spatial heterogeneity aspect, Geographically Weighted Regression is better than Ordinary Least Square.