臺灣生育行為之社會決定因素

背景：臺灣於1984年完成生育轉型後，近年總生育率又快速地下降，成為世界上生育率最低的地區。可惜的是，過去研究大多為單一層次之橫斷性研究設計，且僅聚焦於解釋總生育率或生育數量之變異，較少利用近年之長期資料及多層次分析方法，探討區位特質之脈絡效應，並將生育年齡納入考量，以釐清影響臺灣各鄉鎮市區總生育率與年齡別生育率以及個人生育行為之社會決定因素。目的：具體而言，本研究利用臺灣近年生育率及生育行為調查資料，並結合多層次分析方法，旨在：（一）描述生育率之時空變遷趨勢；（二）探討影響生育率變遷之社會決定因素；（三）探討影響個人生育行為之社會決定因素。方法：本研究資料來自歷年內政部戶政司及統計處人口統計、主計總處之縣市統計及「婦女婚育與就業調查」資料。依變項包括1998至2015年之總生育率及年齡別生育率，以及個人生育事件與生育年齡，解釋變項包括人口密度、高等教育比例、失業率、服務業比例、女性勞動參與率、扶老比、可支配所得中位數、醫療可近性和嬰兒死亡率等社會決定因素。在分析方法部分，首先利用地理資訊系統地圖描述臺灣生育率之時空分布，並採用多年期縱橫資料分析方法及多層次存活分析模型，探討影響地區生育率及個人生育行為之社會決定因素。結果：（一）在生育率變遷趨勢方面：臺灣總生育率在1998年後的快速跌落，主要是受到30歲以下生育率快速下降之影響，且生育率下降之過程是由主要都市擴散到周圍地區。（二）在生育率變遷之社會決定因素方面：2000年時，地區人口密度和總生育率為負向關係，但在生育率下降到極低水準的過程中，不同人口密度地區之生育率差距縮小，負向關係逐漸消失。縱橫資料分析模型顯示，高等教育擴張是解釋生育率變遷最重要的因素，高等教育比例增加1%，總生育率下降千分之13.6。另外，失業率與服務業比例對於生育率有負向影響；而女性勞動參與率和扶老比與30歲以下生育率為負向關係，與30歲以上生育率轉變為正向關係；地區收入水準及醫療可近性與生育率呈現正向關係；嬰兒死亡率和生育率之關係則相對較小。（三）在個人生育行為之社會決定因素方面：多階層存活分析顯示，個人教育程度越高，其越晚開始生育，且終生不生育的機率也越高。若個人處在高失業率或高等教育擴張程度較高的地區，其生育機率較低，且高等教育擴張抑制生育行為之脈絡效應，對於教育程度為高中職以下者影響較大，特別是對生育第二胎之影響。結論：高等教育的快速擴張，是造成臺灣1998年後生育率急遽下降的重要因素，另外失業率所帶來之經濟不確定性，以及服務業比例之提升，亦不利於生育行為。而高等教育擴張與低生育率之關係，除了受到個人教育程度的組成效應影響之外，地區高等教育擴張也會產生脈絡效應，延後並減少低教育程度者之生育行為。

關鍵字

生育行為；人口密度；高等教育擴張；經濟不確定性；縱橫資料分析；多層次存活分析；社會決定因素

並列摘要

Background: After the completion of fertility transition in 1984, the total fertility rate in Taiwan has rapidly declined to the world's lowest fertility level in recent years. However, most of past studies conducted a single-level analysis with a cross-sectional design, while focused on explaining the variation of total fertility rate or the number of children. Research is needed to use longitudinal data and multilevel analysis methods to investigate the contextual effect of social determinants on fertility and individual fertility behaviors. Objectives: In specific, the aims of this study were to: (1) describe the temporal and spatial trends of fertility rates in Taiwan; (2) examine the social determinants of fertility rates, and (3) examine the impact of social determinants on individual fertility behaviors. Methods: The data were obtained from the Ministry of the Interior and the Directorate General of Budget, Accounting and Statistics of Executive Yuan. The dependent variables were the total fertility rates and age-specific fertility rates from 1998 to 2015, as well as personal birth events and childbearing age. The explanatory variables include population density, tertiary education proportion, unemployment rate, proportion of service industry, female labor participation rates, old-age dependency ratio, median disposable income, health care accessibility, and infant mortality rates. In the analysis, I used the geographic information system map to describe the spatial and temporal distribution of fertility in Taiwan. Furthermore, panel data and multilevel survival analysis were conducted to examine the effect of social determinants on fertility and individual fertility behaviors. Results: Firstly, the rapid decline in total fertility rates in Taiwan after 1998 was mainly influenced by the rapid fall of the fertility rate of women under 30 years old. The decline in fertility was spread from major cities to the surrounding areas. Second, the association between population density and the total fertility rates were negative in 2000. However, the negative relationship gradually disappeared because of the fertility gap between areas with different population density level were narrowed during the period of fertility declined to lowest-low level. The panel data model analysis showed that the tertiary education expansion is the most important factor in explaining the fertility change. In addition, the unemployment rate and the proportion of services were suppressed fertility. While the female labor participation rate and the old-age dependency ratio were negatively related to the fertility rate of women under 30 years old, and the positive relationship was shown with the fertility of women over 30 years. The regional income level and health care accessibility were positively related to the fertility rate. The association between infant mortality rate and fertility is relatively small. Finally, multilevel survival analysis showed that the higher the individual educational attainment, the later they begin to have a child. If an individual lives in an area with higher tertiary education expansion, their fertility would be lower. This contextual effect is stronger for the low educational population, especially for the second childbearing. Moreover, the unemployment rate also reduced the chances for a person to have a child. Conclusion: The rapid expansion of tertiary education is an important social determinant of fertility decline after 1998 in Taiwan. In addition, the economic uncertainty which caused by the unemployment rate, as well as the proportion of service industry, also discourages fertility behaviors. In addition to the compositional effect of personal educational attainment, the degree of higher education expansion in an area will also induce a contextual effect, to reduce and delay the childbearing for the lower educational population.

並列關鍵字

fertility behaviors ； population density ； higher education expansion ； economic uncertainty ； panel analysis ； multilevel survival analysis ； social determinants