臺灣低生育率之空間分析

Abstract

在過去研究中，對於人口轉型導致生育率下降的原因分別有社經結構調適過程與 (創新) 傳播擴散過程的兩種解釋途徑，前者認為生育率的下降是社會邁向現代化過程中對於結構轉變的調適結果，強調社經結構對於行為者的生育態度及決策的影響，而後者則認為社會網絡與人際互動等人際傳播方式對大眾所形塑的減少生育或是延後生育觀念以及對避孕知識與方法的認識並採納，是促使生育率普遍下降的主要原因，重視空間鄰近關係所形成的影響效果。換句話說，傳播擴散觀點認為住得越近的人，其生育行為會越相似，也就是所謂的鄰近效應，因此本文的目的即在檢驗臺灣普遍低生育率現象是否是受到鄰近效應所影響。 本文應用地理資訊系統 (GIS) 分析1980年代後臺灣生育率的空間變異，探索結果顯示，臺灣生育率下降很大的原因來自於「城市/鄉村差距」與「平地/山地差距」的逐漸縮小，生育率有趨於同質化的傾向；此外空間指標Moran’s I 值隨著歷年的時間變化不斷上升，證實臺灣生育率有強烈的空間自相關；而LISA空間群聚圖反應出大都會地區 (臺北、臺南、高雄) 有穩固的低-低生育率聚集現象，而高-高生育率則普遍聚集於原住民比例較高的山地鄉，顯示族群也是影響臺灣生育率重要的變數之一。 從空間探索分析證實臺灣生育率存在空間聚集，本文進一步透過類空間迴歸模型，同時考慮時間及空間變數，分別檢驗社經結構、族群分布及鄰近關係對生育率空間變異的影響。結果顯示，1981年及2001年在加入空間鄰近關係變數後，迴歸模型解釋力大幅提升，證實鄰近效應對於生育率有極顯著的正相關影響，支持傳播擴散理論的觀點；表現在生育率歷年變化上，生育率持續的下降，說明了鄰近鄉鎮的低生育率形成負向回饋影響目標鄉鎮生育率的下降。1991年的迴歸模型顯示，鄰近效應影響效果較低且空間迴歸解釋力並不突出，說明影響生育率變化的因素主要源自於社經結構及族群分布的影響，尤其以族群影響更為顯著，偏向社經結構調適的觀點；表現在生育率歷年變化上，生育率維持低度穩定狀態。因此本文認為，社經結構的調適雖然會造成低生育率，但使生育率持續下降的原因則是受到鄰近效應的影響，進一步觀察2001年後臺灣極低生育率現象，其生育率的變化是持續下降而非平穩狀態，推測鄰近效應的影響仍是非常關鍵的因素。

In the past researches, socioeconomic adjustment and innovation diffusion were two perspectives for explaining why fertility declined caused by the population transition. According to socioeconomic adjustment perspective, fertility transition is related to motivational forces stemming from changes in socioeconomic conditions prior to and during fertility decline. According to innovation diffusion perspective, an idea of less birth has been molded by social networks and interpersonal connection which makes people accept contraception. In other words, people live nearby have similar fertility behaviors, which is called the neighborhood effect. The purpose of this study is to verify whether Taiwan’s low fertility rate is influenced by the neighborhood effect or not. We use the Geographic Information System (GIS) to analyze spatial differences of Taiwan’s fertility rate after 1980s. It shows that the fertility rate divergences between “urban vs. rural areas” and “aborigine vs. non-aborigine areas” are both becoming homogeneous. Moran’s I increases when time passes, which proves that Taiwan’s fertility rate has high spatial autocorrelation. The LISA cluster map also shows that big cities like Taipei, Tainan and Kaohsiung have a stable low-low fertility rate cluster; a high-high fertility rate cluster generally exist in mountain areas where there is a higher aborigine population, which means ethnicity is an important factor that affects Taiwan’s fertility rate. Next, we build a matrix-spatial regression model of fertility to test the socioeconomic structure, ethnic distribution and the neighborhood relation factors, which considers both time and spatial dimensions. The results show that in 1981 and 2001, after adding spatial factors, R2 has a big increase which shows how the neighborhood effect plays an important role on fertility decline, supporting the view of diffusion theory. Near county’s low fertility creates a negative feedback mechanism, influencing the focus county’s fertility and leading the whole fertility rate downward. In 1991, regression model shows that the neighborhood effect does not dominant; fertility rate is effected by socioeconomic structure and especially ethnic distribution factors. Partial to the view of adjustment theory, the fertility rate maintains low but stable. Therefore, we point out that although the socioeconomic adjustment process might cause a low fertility rate, it was the neighborhood effect that bought fertility downward. We observe that Taiwan’s lowest-low fertility from 2001 till now wasn’t stable but continuing downward, so we conjecture that the neighborhood effect is still a very important factor to influence the fertility rate.