健康保險能促進生育率嗎？臺灣全民健保對生育率的影響 - 政大學術集成

全文

(1)International Master’s Program in International Studies National Chengchi University 國立政治大學國際研究英語碩士學位學程. 政治大. ‧. ‧ 國. 學. 立 Can Health Insurance Boost Fertility? The Fertility Effect of National Health Insurance in Taiwan 健康保險能促進生育率嗎？臺灣全民健保對生育率的影響 n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Elyse Mark 麥麗施 Adviser: Dr. Yu-hsuan Su. June 2018. DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(2) Abstract The implementation of Taiwanese National Health Insurance in 1995 affected almost every Taiwanese citizen’s healthcare options and coverage. Inclusion of low-cost prenatal and postnatal care, delivery services, and child healthcare coverage indicates the implicit pronatalism of NHI policy. Given the endogeneity of healthcare availability to childbearing decisions, this study uses the 1995 implementation of NHI to estimate the impact of national healthcare on Taiwanese fertility. We first approach this question using OLS regression to test for correlation. 政治大. between completed childbearing and an NHI dummy, among other demographic factors. In the. 立. second approach, difference-in-differences methodologies estimate the effect of 1995 NHI. ‧ 國. 學. implementation on the treatment group of Taiwanese women and state-insured control group. Our data sources are the 1979-2016 waves of the Women’s Marriage, Fertility, and Employment. ‧. Survey conducted by the Statistical Department of the Taiwanese Ministry of the Interior. We. y. Nat. sit. hypothesize that the 1995 NHI policy change increased the fertility of Taiwanese women around. n. al. er. io. the cutoff date despite sub-replacement national fertility beginning in the 1980s. OLS analysis. i n U. v. fails to show a positive correlation between NHI and women’s fertility in Taiwan, however.. Ch. engchi. Further, DID estimations using a maternal age and a conception time both reject the hypothesis that lowered cost of childbearing through NHI stimulated Taiwanese fertility. Keywords: Taiwan, National Health Insurance, fertility, pronatalism, difference in differences. i DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(3) 摘要自 1995 年臺灣實施全民健康保險起，健保幾乎影響了所有臺灣居民的醫療選擇。臺灣的健保包括產前和產後檢查、分娩服務、兒童保健的費用補助，表明健保政策潛在的鼓勵生育性。本研究利用 1995 年全民健保的實施估計國家醫療保健對臺灣生育率的影響，首先用普通最小平方法估計完成生育子女數與健保以及其他的人口因素之間的相關性，並使用差異中之差異方法來估計健保實施對實驗組和對照組的影響。我們的數據來源是内政部 1979 年至 2016 年的婦女婚育與就業調查。我們假設 1995 年的健保政策減緩了臺灣婦女. 治政大自 1980 年代開始的生育率下降趨勢。然而，普通最小平分析未能顯示健保與臺灣婦女生立通過健保降低生育成本會刺激臺灣生育的假設。. 關鍵字：臺灣，全民健康保險，生育率，差異中之差異. Nat. n. al. er. io. sit. y. ‧. ‧ 國. 學. 育率的正相關。此外，使用母親年齡和受孕時間作為截止點的差異中之差異估計也拒絕了. Ch. engchi. i n U. v. ii DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(4) Acknowledgements To my adviser, Professor Yu-hsuan Su, thank you for your support and guidance throughout this project, during a year filled with exciting new challenges for both of us. I couldn’t have done it without you! Special thanks to the Republic of China Ministry of Science and Technology, which provided funding for me to travel to the Population Association of America’s 2018 Annual Meeting in Denver, CO to present preliminary findings from this study. Feedback from this. 政治大 share my learning about Taiwan with scholars from the United States and beyond. 立. experience was instrumental in revision of the study, and I am thankful for the opportunity to. To my classmates in IMPIS: thank you for four semesters of companionship and. ‧ 國. 學. commiseration. No matter which countries we call home, it’s nice to know that grad students. ‧. worldwide share a love of coffee and conversation You all have made IMPIS a place of mutual. sit. y. Nat. encouragement and friendship that has helped me grow.. io. er. Finally, I am very grateful to the staff at Fulbright Taiwan for their administrative and financial support for my degree work at National Chengchi University. Without the generosity of. n. al. Ch. i n U. v. Fulbright Taiwan, I would not have come to this beautiful island or had the opportunity to. engchi. explore new and exciting research interests in international studies and demography, among peers from all over the world. Over the past two years, conversations with researchers and grantees working in variety of fields have not only sparked my academic curiosity, but also led me to lifelong friendships along the way. Thank you.. iii DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(5) Table of Contents Abstract ............................................................................................................................................ i Acknowledgements ........................................................................................................................ iii List of Figures ................................................................................................................................. v List of Tables ................................................................................................................................. vi 1. Introduction ................................................................................................................................. 1 2. Background ................................................................................................................................. 3 2.1 Taiwanese Fertility Transition .............................................................................................. 3. 政治大 3. Literature Review........................................................................................................................ 7 立 2.2 Pronatal Fertility Policy ........................................................................................................ 6. ‧ 國. 學. 3.1 Low Fertility in the 20th Century........................................................................................... 7 3.1.1 Theoretical Explanations for Low Fertility .................................................................... 8 3.1.2 Pronatal policy. ............................................................................................................. 14. ‧. 3.2 NHI Implementation and Taiwan ........................................................................................ 22 3.2.1 NHI and the Taiwanese Population. ............................................................................. 22. y. Nat. sit. 3.2.2 NHI and Taiwanese Fertility. ....................................................................................... 25. n. al. er. io. 4. Methods..................................................................................................................................... 28. v. 4.1 Theoretical Framework ....................................................................................................... 28. Ch. i n U. 4.2 Data ..................................................................................................................................... 28. engchi. 4.3 Empirical Methods .............................................................................................................. 32 4.3.1 Graphical Analysis. ...................................................................................................... 33 4.3.2 Multivariate Regression. ............................................................................................... 34 4.3.3 DID Approach.. ............................................................................................................ 36 5. Results ....................................................................................................................................... 39 5.1 Multivariate Regression ...................................................................................................... 39 5.2 DID Approach 1: Maternal Age .......................................................................................... 43 5.3 DID Approach 2: Monthly Birthrate ................................................................................... 47 6. Conclusion ................................................................................................................................ 50 Bibliography ................................................................................................................................. 53 iv DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(6) List of Figures Figure 1: Fertility change in Taiwan, 1975-2016 ........................................................................... 5 Figure 2: Number of births by maternal age group for 1959 and 1971 cohorts ........................... 32 Figure 3: Mean number of live births by maternal birth year (WMFES 2016) ............................ 34 Figure 4: Monthly Birthrate DID .................................................................................................. 48. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. v DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(7) List of Tables Table 1: Fertility of Married Women............................................................................................ 29 Table 2: Characteristics of Married Women by Treatment Status ............................................... 31 Table 3: OLS Summary Statistics ................................................................................................. 36 Table 4: Ordinary Least Squares Results ...................................................................................... 41 Table 5: Difference in Differences Estimation ............................................................................. 43 Table 6: DID Results for 3 Cutoff Ages (35, 40, and 45 years old) ............................................. 44. 政治大 Table 8: Difference in Differences Estimation (Birthrate) ........................................................... 47 立 Table 7: Difference in Differences Results by Income-education Cell ........................................ 46. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. vi DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(8) 1. Introduction In the 20th century, Taiwanese fertility underwent a dramatic transformation, declining from a peak total fertility rate of 7.04 births per woman in 1951 to below-replacement level fertility in 1984. Since then, fertility rates have stagnated below 2.1 births per woman until the present day (G. H. Lee & Lee, 2014). Low fertility is not an isolated Taiwanese problem, but rather a demographic crisis endemic to many economically developed countries, and felt acutely in Western Europe and East Asia. Sub-replacement fertility does not automatically bring about. 政治大. population decline. When the momentum of a large childbearing cohort is lost, however, and in. 立. the absence of immigration to offset low birthrate, low fertility causes to population shrinkage.. ‧ 國. 學. While this might lower the environmental burden of a large population, sustained low fertility threatens state capacity to provide pensions for aging populations, promote stable economic. ‧. growth, and fund vital public services like education or healthcare. From Japan to Norway,. y. Nat. sit. nations around the world have undertaken natalist policies in an attempt to boost fertility,. n. al. er. io. ranging from the provision of public childcare to anti-discrimination employment law and paid. i n U. v. parental leave (Nakajima & Tanaka, 2014; Rindfuss, Guilkey, Morgan, & Kravdal, 2010).. Ch. engchi. Previous studies on the efficacy of pronatal policy aimed at increasing domestic fertility through cash or non-cash incentives in developed countries have yielded mixed results. Most governments work concurrently through implicit and explicit policy channels to promote childbearing, which presents analytical challenges for researchers hoping to identify successful pronatal incentives. Because fertility decisions lie at the intersection of social, economic, and cultural crossroads, intercountry generalization is unreliable, if not impossible. Little consensus exists regarding whether or not policy incentives are able to stimulate fertility at all, much less which types of incentive structures are the most effective. 1 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(9) This paper narrows the broader academic debate over pronatal policy solutions to subreplacement fertility by concentrating on the Taiwanese case during the 1990s. The primary purpose of the study is to determine the effect of national health insurance (NHI) implementation on the childbearing decisions of Taiwanese women, working on the premise that NHI coverage almost universally lowered financial barriers to childrearing. Given its inclusion of various birth and child healthcare benefits, NHI may be considered a form of implicit pronatal policy in Taiwan, and one that circumvented contemporary political opposition to other more explicit. 治政大Statistical Department of the Employment Survey (WMFES), conducted periodically by the 立 forms of pronatal policy. Data for this paper are from the Women’s Marriage, Fertility, and. Taiwanese Ministry of the Interior since 1979. The study uses multivariate OLS regression and. ‧ 國. 學. difference-in-differences (DID) designs to analyze the fertility effects of the universal. ‧. implementation of Taiwanese NHI in 1995, examining conditional fertility and monthly birthrate as outcome variables, respectively.. y. Nat. er. io. sit. The organization of the paper is as follows. The next section provides a brief background on the Taiwanese demographic transition. Section 3 reviews the wider literature on fertility. n. al. Ch. i n U. v. decline in developed countries and Taiwanese national health insurance. In Section 4, I elaborate. engchi. on the empirical methods data sources of this study. Section 5 reports findings from the OLS and DID estimations, and the final chapter concludes with further discussion of results and policy implications in contemporary Taiwan.. 2 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(10) 2. Background 2.1 Taiwanese Fertility Transition Taiwan has undergone a well-documented demographic reversal over the last century, transforming from an island with high fertility to low fertility in the 1950s, and from low fertility to sub-replacement levels beginning in the 1980s. Fertility decline on the island has lowered fertility from its peak TFR of 7.04 children born per woman in 1951 to a stagnant 1.75 between 1986 and 1997, decreasing again to approximately 1.12 by 2007, where it remains stable today.. 政治大 Combined with an increase in birth rate after World War II, Taiwan’s fertility peaked in the 立 Beginning in the 1920s, the decreasing death rate fast-tracked Taiwan’s population growth.. ‧ 國. 學. 1950s with a crude annual birth rate of over 40 children per 1000 population. The combination of a household registration system and periodic national knowledge, attitude, and practice (KAP). ‧. sit. Nat. trends, making Taiwanese fertility an ideal case for exploration.. y. family planning surveys since 1965 reflects long-standing state interest in tracking population. io. er. The 1950s to 1980s was generally a period of fertility reduction for Taiwanese women of. al. childbearing age. Government introduction of the 1964 Family Planning Program, combined. n. iv n C with increased opportunities for private h economic gain, radically e n g c h i U influenced Taiwanese. childbearing preferences. The program employed civil servants as Pre-pregnancy Health Workers (PPHWs) or part-time Village Health Education Nurses (VHENs) in each of Taiwan’s 361 townships, tasking them with traveling door-to-door to private residences to recruit women into the program (Montgomery & Casterline, 1993). Through the family planning program, PPHWs provided recruits with various contraceptives such as intrauterine devices, oral contraceptives, condoms, and distributed propaganda encouraging smaller family sizes.. 3 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(11) Presence of the family planning program during Taiwan’s fertility decline from the 1960s onward and high usage of contraceptives by Taiwanese women indicates the program was successful. In their analysis of diffusion of fertility control in Taiwan, however, Montgomery and Casterline (1993) find that the actual estimate of the program impact varies dramatically depending on statistical design. Weighted least squares regression shows the program reducing marital fertility no more than 5% from 1968-80, with increases in marital fertility in the early years of the program, while their diffusion based curves put that number somewhere between 5. 治政 increased Taiwanese fertility decline 1954-1970, pointing to 大 evidence from Taichung (a target 立 and 22%. Li (1973) supports the claim that the family planning program neither induced nor. area of the program) to show that action programs had little appreciable effect on childbearing.. ‧ 國. 學. Rather, they point to decreased infant mortality as likely explanatory variable throughout the. ‧. country, and rising educational levels as an important urban factor.. Nuptiality and martial fertility have also been cited as causes for Taiwanese fertility. y. Nat. er. io. sit. decline. Sun, Lin, and Freedman (1978) first showed that about 1/3 of the decrease in birth rate 1961-1984 was caused by changing marriage patterns, while in contrast Feeney (1991) claims. n. al. Ch. i n U. v. that nuptiality contributed nothing to fertility change in this period. His study shows a sustained. engchi. high level of marriage and motherhood for Taiwanese women throughout the 1980s, asserting contrast between his and the Chang, Freedman, and Sun (1987) study is due to methodological differences related to his focus on women’s probability of first marriage and Chang et al. examining age specific birth rate. However, his later claim that Taiwan “may not be below replacement level after all” (in the 1990s) belies the strength of that argument (Feeney, 1991, p.. 4 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(12) Fertility change in Taiwan, 1975-2016 35.0. 3.5. 30.0. 2 1.5 25.0. 1. 政治大. 0.5. 立. 2015. 2013. 2011. 2009. 2007. 2005. 2003. 2001. 1999. 1997. 1995. 1993. 1989. 1987. 1985. 1983. ‧ 國. 1981. 1979. 1977. 20.0. 學. 1975. 0. 1991. Fertility Rate. 2.5. Maternal Age. 3. replacement rate fertility. average age of mothers at birth. average age of mothers at first birth. ‧. total fertility rate. Figure 1: Fertility change in Taiwan, 1975-2016. y. Nat. io. sit. 476). Chang (2006) found that 2/3 of the decrease in birth rate 1965-1980 was attributable to. n. al. er. declining marital fertility, and the remaining 1/3 due to nuptiality patterns.. Ch. i n U. v. In 1983, the combination of an extremely low death rate and large cohort of women of. engchi. childbearing age resulted in a high population growth rate, which prompted the Taiwanese government to push a second wave of the family planning program. During the 1980s, statistics show fertility reduction for women under 30 years of age and a trend of delaying childbearing until the late twenties; by the 2000s the average age of mothers at first birth was into the early thirties (see Figure 1; plotted with data from R.O.C. Ministry of the Interior). Despite overall fertility decline, martial fertility actually increased between 1985 and 2007 (Luoh, 2007 as cited in Lee & Lin, 2016). Taiwan has extremely low extramarital fertility, and childbearing has been. 5 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(13) named as the de facto cause of marriage for many Taiwanese women (K. Chen & Yang, 2005 as cited in Lee & Lin 2016). 2.2 Pronatal Fertility Policy Taiwanese fertility policy reflects its demographic reversal, lagged by two decades. Clear antinatalism underlies the 1964 national family planning program, which boosted contraceptive use and promoted the social and economic advantages of small families; by 1992, however, the revised Guideline for Population Policy acknowledged the need for more stable population. 政治大 progress” obstructed the state’s ability to increase Taiwanese family size, as did pressure to 立. growth in Taiwan. The “link between Taiwan’s family planning program and rapid economic. adopt a universal pension plan in the late 1990s (M. Lee & Lin, 2016, p. 270).. ‧ 國. 學. Explicit pronatal policy did not pass until the 2008 publication of a Pronatal Population. ‧. Policy White Paper, due in part to backlash from feminist and environmentalist groups. sit. y. Nat. throughout the 1990s and 2000s, who questioned the gendered and environmental consequences. io. er. of increasing the national population. Actual quantitative fertility targets were not published until the 2013 revision of that paper, along with strategies to address low fertility. In the midst of this. n. al. Ch. i n U. v. controversy, 1995 passage of National Health Insurance (NHI) in Taiwan included several. engchi. incentives for higher fertility, such as free prenatal and neonatal care, child delivery, child health insurance, one-month cash benefit for childbirth, and complete premium coverage for lowincome families. By the end of 1995, 97% of the population had registered for NHI coverage (L. Chen, Chen, & Yang, 2008). Prior to NHI, only women covered by Government Employee Insurance had access to free prenatal care through public health insurance. Given the pronatal incentives in NHI, the present study aims to determine the effect of nationwide implementation of National Health Insurance on Taiwanese fertility.. 6 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(14) 3. Literature Review 3.1 Low Fertility in the 20th Century Low fertility is a relatively recent social phenomenon. Populations today have greater access than ever before to family planning tools such as contraceptives and voluntary sterilization procedures, especially in economically developed areas where a variety of cultural, economic, and political factors convene to decrease the demand for children. The majority of historical fertility studies have analyzed the influence of variables such as cultural norms, social. 政治大 that lower fertility is a demographic goal, particularly for developing countries. If so many 立. or health policy, female labor force participation, or education on fertility, under the assumption. ‧ 國. policy, then why do other states consider low fertility a threat?. 學. developing countries strive for low fertility through family planning programs or other state. ‧. From a governmental perspective, low fertility is essentially an issue of potential “market. sit. y. Nat. failure,” wherein population reductions threaten support for governmental programs and the. io. er. private market (Demeny, 1986, p. 339). More specifically, below-replacement fertility decreases. al. the tax base needed in many richer countries to finance retirement benefits, state-sponsored. n. iv n C medical care for the elderly, among other towards vulnerable segments of the hprograms e n g ctargeted hi U population. A growing body of literature not only acknowledges the increasing importance of low fertility in other areas of the world, but also addresses the effectiveness of pronatal policies adopted by those countries vulnerable to this demographic trend. In a broad sense, the abstract justification for pronatal government interventions derives from the preservation of the state and protection of its citizens’ economic and social well-being. Nearly half of the world population lives in countries with below replacement-level fertility. During the twentieth century, the entire European continent has come to experience. 7 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(15) serious fertility decline and population growth rates nearing zero, particularly in Western and Northern European nations (Chesnais, 1996; Legge Jr & Alford, 1986). Even in southern Europe, where Mediterranean countries are commonly understood to have “traditional, Catholic, and family oriented” societies valuing larger family sizes, fertility has been at sub-replacement levels since the 1990s (Chesnais, 1996, p. 729). Today, low fertility remains pervasive throughout Europe and poses a serious demographic problem in North America and East Asia. Despite the widespread nature of this trend, the literature explaining low fertility trends remains divided,. 政治大. ranging from sociological or cultural arguments based on industrial modernization to microeconomic household theories.. 立. 3.1.1 Theoretical Explanations for Low Fertility. Theoretical explanations for low. ‧ 國. 學. fertility in developed countries vary widely, blaming industrial modernization, microeconomic. ‧. household theories, and even values-based trends that defy direct state intervention. Below, I have organized current and historical fertility theories into three categories: modern transition. y. Nat. er. io. sit. theory, the economic model of fertility, and other sundry approaches. Although I only list these three headings, throughout history, many scholars have attempted to summarize their. n. al. Ch. i n U. v. understandings of population into fertility theories; these three segments simply represent the. engchi. most politically and academically significant theories of fertility decline over the past two hundred years. Demographic Transition Theory. Demographic transition (DT) theory, also known as modern transition theory, has its roots in a nineteenth century Malthusian approach to fertility decline popularized in Europe and the United States. This theory links voluntary lower fertility with expanded economic opportunity for social mobility. DT theory flourished in the beginning of the twentieth century, as the European demographic transition from a high mortality-high. 8 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(16) fertility region to low mortality-low fertility one slowed. Key proponents include Davis (1945), Notestein (1945) and Landry (1987). DT theory claims that fertility decline arises from the “structural transformation of the economic and social systems generated by the industrial revolution,” given inevitable shifts in cultural norms, values, and socioeconomic classes that arise due to industrialization (Demeny, 1986, p. 343). In early industrialization stages, fertility is high and uncontrolled and mortality is declining because of increasing quality of life. For example, the industrialization improves food availability, public health, urbanization, agriculture,. 治政大 theory used to explain fertility Demographic transition theory has been the predominant 立. and contraceptive technologies, among other factors (Szreter, 1993).. trends in policymaking (Demeny, 1986), and the correlation between fertility decline and. ‧ 國. 學. socioeconomic development at the national level is well accepted. This theory, however,. ‧. assumes that industrialization’s population effects are mainly economic, and does not recognize the role of social changes like gender equity or labor force participation (Demeny, 1986; Zaidi &. y. Nat. er. io. sit. Morgan, 2017). In addition, it fails to account for the role of migration in population size. The second demographic transition framework is an extension of demographic transition theory for. n. al. Ch. i n U. v. already low-fertility countries, put forth in the 1980s by Lesthaeghe and van de Kaa (Van de. engchi. Kaa, 1987; Zaidi & Morgan, 2017). It attributes the sub-replacement fertility of industrialized countries in North America and Western Europe to complete control of fertility, and childbearing is influenced by factors like the popularity of contraceptives and increased female labor force participation (Van de Kaa, 1987, 2001). Further, its proponents argue that childbearing decisions increasingly depend on value-based, rather than purely economic, factors (Zaidi & Morgan, 2017).. 9 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(17) The Economic Model. During the 1960s, the school of new household economics 1 emerged to rival the explanatory power of demographic transition theory regarding fertility decisions. On a foundation of microeconomic analysis and statistical evidence, rather than normative assumptions, this model understands fertility by placing children in a household economic framework. In his earliest analysis of fertility, Becker (1960) explains that children may be consumed as a source of “psychic income or utility,” and that past a certain age they may also contribute pecuniary income to the family, thus acting as a productive good (Becker, 1960,. 治政 level, speaking of opportunity cost of childbearing. That cost大 is the foregone earnings that 立. p. 213). Becker’s later works further develop theories on the allocation of time at the household. parents sacrifice, as measured by “the amount of time used per dollar of goods and the cost of. ‧ 國. 學. unit per time” (Becker, 1965, p. 503). Becker theorizes that households in richer countries forfeit. ‧. earnings in exchange for psychic income gained by having and raising children. Since childcare is not monetarily productive, its relative opportunity cost is higher for families enjoying higher. er. io. sit. y. Nat. incomes.. Critics of this microeconomic model argue that it emphasizes empirical data to the. n. al. Ch. i n U. v. exclusion of significant cultural, social, and historical influences on fertility (Demeny, 1986).. engchi. While Demeny (1986) also claims that “policy applications of the findings from the new types of [microeconomic] analyses were scarce,” today we see that many pronatal policies actually do rely on the concept of the lowering the financial cost of children to incentivize fertility (p. 344). Rational choice theory is a variant of the economic model, related to, and often studied in conjunction with new household economics vis-à-vis fertility. Childbearing is defined as a purely rational economic decision, determined by strict cost-benefit analysis, and influenced by parental. 1. This school of fertility theory is also known as new home economics theory (Gauthier, 2007).. 10 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(18) income constraints and consumer preferences. McDonald (2002) notes that the benefits of children are psychological rather than economic, especially in societies where elder care does not remain within the family. He thus categorizes children as consumption instead of investment goods. Gauthier (2007) separates the cost of children into direct and indirect categories, with indirect costs as the foregone (monetary) earnings given up during the childbearing/rearing process; these are equivalent to Becker’s opportunity cost of childbearing. Direct costs, on the other hand, are “the actual dollar expenditure on the child less any financial benefits” received by. 治政大 may affect later order births child services (McDonald, 2002, p. 424). The direct cost of children 立. parents because of the child or children, such as financial transfers, tax exemptions, or subsidized. more significantly than indirect costs, which level as the number of children increases, and may. ‧ 國. 學. primarily affect first-order births.. ‧. One caveat to the concept of the psychological value of children is the effect of birth order on perceived benefits of the child. McDonald (2002) writes:. y. Nat. er. io. sit. Having the first child provides benefits including the status of being a parent, “being a family”, having offspring who will carry on the family, meeting the expectations of. n. al. Ch. i n U. v. others, having a baby who will be fun and will grow up and love you, fulfilling childhood. engchi. dreams, or providing vicarious pleasure from the child's success. The decision to have a second child may be more related to the strength of the notion that each child should have at least one sibling or the desire for a child of the other sex. Those who have a third child may value at least three children as a “real” family, or they may be still trying for a child of the sex that they don’t have … It is likely that the level of the net psychological benefits threshold falls as birth order rise… [and] that the level of the threshold falls as people get older (p. 422-23).. 11 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(19) According to rational choice fertility theory, decreasing the cost of children or increasing parental income through pronatal policy should increase demand for children. Because raising the psychological benefits of children is not within the scope of policy, the welfare state approach lowers the costs of children through financial transfers, the tax system, or subsidized services (McDonald, 2002, p. 424). As seen above, however, theoretically policy may affect birth spacing more than overall fertility, by incentivizing earlier childbearing rather than having more children (Gauthier, 2007).. 治政大 (2) Parents and individuals An increase in income is expected to increase demand for children; 立 Gauthier (2007) observes five main assumptions of the economic model of fertility. (1). have complete information about the cost of childbearing and births and perfectly planned; (3). ‧ 國. 學. Marriage and childbearing are economically rational choices; (4) Policies may impact fertility by. ‧. decreasing the cost of children or increasing parental income; (5) Childbearing preferences are homogenous among the family or household. Two key criticisms of the economic model involve. y. Nat. er. io. sit. the first and fifth points. Regarding the first assumption, an increase in income could increase the quality of children rather than the quantity. This means expenditure on children could rise (e.g.. n. al. Ch. i n U. v. more expensive education, extracurricular development, health care) and increase the overall. engchi. cost of children, keeping the demand for children low. The fifth assumption reveals that the economic model ignores the role of gender dynamics in fertility. If the demand for children is heterogeneous in the household, then local gender dynamics may affect how pronatal policy influences childbearing. Risk aversion theory describes another facet of rational choice theory, minus the assumption that individuals and households have perfect information on the costs and benefits of childbearing. Without that assumption, couples consider children a source of potential economic,. 12 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(20) interpersonal, or romantic risk. For example, childbearing almost guarantees a period of lower income, delayed return to the workforce by one or both parents, and higher consumption costs (McDonald, 2002). The best, and possibly the only, policy solution according to this theory is the welfare state approach, which lowers risk of childbearing throughout economic and social spheres for citizens in general. Additional Approaches. In the literature, there are several other notable theories of fertility change. Easterlin (1975) proposes a cyclical pattern of fertility that fluctuates between. 治政 that couples will try to achieve a minimum standard of living大 that is equal to or higher than that 立 baby booms and low fertility periods. Fertility reflects relative income in this theory, meaning. of their childhood. Childbearing trends are predicated upon expected/realized economic wealth. ‧ 國. 學. and subject to spontaneous recovery of higher fertility, decreasing the need for pronatal state. ‧. interventions (Demeny, 1986).. Post-materialist fertility theory suggests that a sustained level of development in wealthy. y. Nat. er. io. sit. countries has caused an ideational shift from materialism to self-realization. When values such as personal growth, liberalism, and individual freedom are celebrated over traditional family values,. n. al. Ch. i n U. v. fertility decreases. Post-materialist values theory dovetails with second demographic transition. engchi. theory. As previously mentioned, second demographic transition theory proposes a post-WWII decoupling of marriage and procreation in advanced countries, which, along with increasingly available contraceptive technologies, leads to below-replacement fertility. It is assumed that shifting ideational systems prompt these changing demographic patterns (Van de Kaa, 2001; Zaidi & Morgan, 2017). McDonald (2002) suggests that reason for this discrepancy is the unaddressed issue of gender equity in childbearing decisions. His gender equity theory of fertility divides society into. 13 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(21) individual-oriented institutions (i.e. education and market employment) and family-oriented institutions (e.g. social security, the industrial complex). In advanced modern societies, he claims, gender equity is high and rising within education systems and the much of the labor force. That said, the traditional male breadwinner model still prevails as the foundation of many family-oriented institutions, and if children are present, obliges women to be caretakers and men economic providers within traditional heterosexual relationships (McDonald, 2002). Childbearing limits the increased economic and career opportunities afforded to women by rising. 治政大 and Japan well, where there is fertility of traditional patriarchal societies such as southern Europe 立 gender equity, and women likewise limit their fertility. This theory of fertility explains the low. a large discrepancy of gender equity between individual and family systems.. ‧ 國. 學. 3.1.2 Pronatal policy. The prevalence of low fertility over the past half century has led to. ‧. a variety of pronatal policies intended to stabilize population growth. Pronatal policies may be divided into two kinds, nonmaterial and material incentives for childbearing.. y. Nat. er. io. sit. Nonmaterial pronatal policy. Nonmaterial pronatal policy promotes fertility through nonfinancial channels, and falls into distinct categories: anti-discrimination legislature,. n. al. Ch. i n U. v. restrictive fertility legislature, or values reeducation programs. In general, these policies provide. engchi. broad legal support for combining employment and family, or for social changes conducive to increased childbearing. McDonald (2002) describes several options in the policy toolbox to address low fertility. Antidiscrimination legislation preventing unfair hiring practices based on gender, relationship, or family status supports the growth of economically sustainable families, as does guaranteed maternity and paternity leave with full or partial benefits. Policy ensuring a parent’s return to work after childbirth guarantees some economic stability for new families; some policies may even allow a part-time return to work with continued partial benefits.. 14 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(22) Parental leave policy immediately raises questions about who pays leave benefits, what the criteria are for receiving those benefits, and how long of a leave period is guaranteed. That said, parental leave policies may help people reconcile work with family, and gender-neutral policies may even encourage greater gender equity in childrearing responsibilities (McDonald, 2002). European evidence in France and former East Germany shows that role incompatibility for working mothers has had a significant impact on fertility (Legge Jr & Alford, 1986), and that parental leave policies were correlated with higher fertility in Canada, Finland, Norway, and. 治政 research from the latter half of the twentieth century suggests大 that maternal leave policies and 立. Sweden (Gauthier, 2007). Rindfuss and Brewster (1996) find, however, that in the United States,. cash benefits did not promote childbearing in the absence of other family policy. Thus while. ‧ 國. 學. parental leave policy is associated with more stable fertility among low fertility countries,. ‧. evidence remains mixed as to its effect on fertility in isolation.. Values reeducation ties back to the second demographic transition / post-materialist. y. Nat. er. io. sit. approach to low fertility and would (hypothetically) involve governmental support for programs to reshape citizen’s attitudes towards family size. Demeny (1986) argues that pronatal values. n. al. Ch. i n U. v. reeducation is not an appropriate route for politically developed countries, given that:. engchi. Democratic states are ill-equipped to engage in specialized value education of their citizens: values are embedded in and conveyed by the deeper institutional structures…Ministerial exhortations, posters of happy three-child families, and medals to heroine mothers are neither well-received nor effective in influencing fertility (p. 347). On a similar note, restricting access to contraceptives, abortion, and other family planning methods presents an illiberal approach to pronatal policy unsuitable for modern democracies. In countries where contraceptive knowledge and use has already become widespread, and where the. 15 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(23) internet enables rapid communication, such retroactive measures may not be possible. Furthermore, evidence from Legge Jr and Alford (1986) in Romania indicates that abortion restriction is both unlikely to raise fertility and very likely to harm the health of women and children. After World War II, many Eastern European countries promoted population policy to combat fears about a diminishing labor force and regional population distribution; Romania accomplished this through extreme abortion restriction and childlessness taxes. In general, for Eastern Europe in the 1950s-60s, birth control access was rarely available, and abortion and the. 治政大to illegal abortion to prevent in the absence of legal family planning methods, women turned 立. pullout method were the primary means of birth control. The Romanian case demonstrated that. births, with severe health consequences. Moreover, results show that the economic incentive. ‧ 國. 學. structures of contemporary Hungary and former East Germany had a more lasting positive effect. ‧. on birthrate in those regions.. Material pronatal policy. Financial incentives for childbearing span a wide range of. y. Nat. er. io. sit. options, from direct cash infusions for new parents to subsidized goods and services for family support. Cash-based pronatal policy uses direct monetary incentives to motivate limited. n. al. Ch. i n U. v. childbearing among women and couples. In a broad review of thirteen macro-level studies on. engchi. family cash benefits and fertility, Gauthier (2007) concludes that in general, higher child benefits (of any of these material policy types) are associated with higher fertility among a diverse range of methodologies. However, she asserts that these aggregate studies show that cash benefits are more likely to affect birth timing than completed fertility, and proposes they may only speed up the timeline of childbearing decisions. Micro-level data reviewed also supports the correlation between child or family benefits and fertility, but with inconclusive results regarding magnitude of effect on birth order.. 16 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(24) Specifically, material incentives may include financial assistance measures like periodic or one-time birth benefits, and tax exemptions for parents. Periodic payment schemes provide regular cash transfers for parents by the state, depending on the age of the child or its birth order. A key flaw of such policies is reliance on the assumption that the household equally shares income and childrearing responsibility, and failure to address the role of gender equity in the family. The state similarly might dispense one-time child benefits to parents, usually paid upon the birth of a child or at another, significant age (for example, when the child begins school).. 治政大 and raising a child (Demeny, supplementing parental income in recognition of the cost of having 立 Both of these types of redistributive policy are intended to increase horizontal equity by. 1986; McDonald, 2002).. ‧ 國. 學. Tax deductions, exemptions, or credits given to parents are another form of redistributive. ‧. fertility policy. Whittington, Alm, and Peters (1990) examine implicit pronatalism in U.S. income tax personal exemptions for dependents, showing that personal exemptions led to a. y. Nat. er. io. sit. significant increase in birthrate for middle-income families. The obvious reason for this type of tax-based pronatal policy is to provide a child subsidy, which offsets the costs of raising a child. n. al. Ch. i n U. v. to adulthood. In the Whittington et al. (1990) study, all specifications demonstrated that the. engchi. personal exemption policy had a positive and significant impact on birthrates in the U.S. Even stronger evidence of a pronatal effect from tax incentives comes from Canada, in analysis of the Quebec Allowance for Newborn Children policy. Milligan (2005) shows that women eligible for the maximum child benefit (C$8000) increased their fertility by 16.9%, and women on average increased childbearing by 12%. Milligan reconciles these findings with previously inconclusive or weak findings from U.S. welfare literature by interacting income with the policy dummy, finding a strong income effect that indicates unexplained heterogeneity in the sample. He. 17 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(25) hypothesizes that the stronger relationship between policy and childbearing for higher income women may be tied to the higher incidence of planned births in that income bracket, for which previous studies failed to acknowledge. Besides direct monetary transfers or tax savings, another form of material incentive for childbearing is subsidized goods or services for children and the family. Housing subsidies (via periodic, one-time, or tax-based payments) may contribute to a sense of economic security that encourages childbearing. In most economically developed countries, public education systems. 治政大and material pronatal policy focus child welfare services. Many contemporary studies on fertility 立 offer free schooling for children until adulthood, and these may also be considered subsidized. on childcare availability, with mixed results (Gauthier, 2007). The French, Norwegian, and. ‧ 國. 學. Swedish governments have famously taken this approach via subsidized nursery schools. ‧. (McDonald, 2002). In Norway, Rindfuss et al. (2010) find that increasing publicly subsidized childcare not only speeds up birth timing, but also increases overall levels of childbearing.. y. Nat. er. io. sit. Further, they find that childcare availability could increase fertility by up to 0.5-0.7 children per woman; or, put another way, if the results of this study are generalizable to other sub-. n. al. Ch. i n U. v. replacement fertility countries, increasing public childcare from 0% to 60% availability could. engchi. bring total fertility back up to replacement level. In Japan, G. H. Lee and Lee (2014) find that increased childcare availability through the Angel Plan and New Angel Plan (largescale national pronatal policies) between 1971 and 2009 had little effect on fertility. Additionally, they did not find evidence that role incompatibility of working mothers limits fertility, as proposed by Rindfuss and Brewster (1996). Pronatal healthcare policy. Healthcare and medical insurance policies fall into the category of material incentives for childbearing, since they function by lowering the monetary. 18 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(26) cost of children in a similar fashion as tax credits or cash payment systems do. Analysis of the RAND Health Insurance Experiment in the U.S. by Leibowitz (1990) shows that women randomly assigned to free medical care for 3-5 years evidenced a 29% higher birthrate than those assigned to cost-sharing insurance. This study indicates a pronatal effect of medical care programs, but notes that this effect may speed birth timing in the short term rather than increase fertility in the long run. In the United States, there is a robust literature on the state-sponsored low-income health insurance program known as Medicaid. Between the 1980s and early 1990s,. 治政大healthcare regardless of family the United States, extending coverage of prenatal care and child 立 significant reform of Medicaid enrollment criteria doubled eligibility rates for women 15-44 in. structure and other limiting factors (Zavodny & Bitler, 2010). According to standard fertility. ‧ 國. 學. theory, decreasing the cost of childbearing and childcare should increase fertility; analysis of. reform, however.. ‧. panel birth certificate and U.S. census data reveals only weak fertility effects of Medicaid. y. Nat. er. io. sit. Zavodny and Bitler (2010) explore the effect of Medicaid eligibility expansion on birthrate through OLS regression, disaggregating results by maternal race (white and black only),. n. al. Ch. i n U. v. educational attainment, and marital status. They find that the expansion had little substantial. engchi. effect on overall fertility during the period 1982-1996, and a 1% increase in eligibility was associated with a statistically significant but small 0.9% increase in birthrate for white women (a 0.3% birthrate increase in the black sample was statistically insignificant). This small boost in fertility was significant and larger among less-educated white women (those without a high school diploma). The pronatal effects of increased eligibility were generally larger for unmarried or less-educated women as compared to married, women with beyond a high school education in either racial group—but those results were found to be largely statistically insignificant at the. 19 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(27) 10% level. Looking specifically at enrollment expansions between 1987 and 1991, Joyce, Kaestner, and Kwan (1998) find that raising the income ceiling for Medicaid enrollment increased birthrates among white American women by 5%, with no significant effect among black women. Though pronatal effects of Medicaid were found after eligibility expansion, increases in birthrate were only statistically significant for certain demographic groups. Further Medicaid analysis by DeLeire, Lopoo, and Simon (2011) using state fixedeffects modeling find a small but positive relationship between Medicaid eligibility expansions. 治政 eligibility correlated with a 1.2% increase in births for white 大 women, and 2.4% increase for black 立 in the 1980s-90s and births, using state birth count as the dependent variable. A 1% increase in. women. These results were not robust to state and year fixed effects across cells disaggregated. ‧ 國. 學. by race, maternal age, education, and marital status, though, and once fixed effects were. ‧. introduced the magnitude of any fertility effect often decreased to a trivial extent. Both of the above studies conclude that there is no robust relationship between recent Medicaid eligibility. y. Nat. er. io. sit. expansion and childbearing patterns in the United States.. Similar research on a 2005 U.S. insurance mandate reform by Schmidt (2007) analyzes. n. al. Ch. i n U. v. the effect of increased coverage of infertility services on fertility rates, under the same premise. engchi. that decreasing the cost of health services could stimulate usage of those services, and thus, increase childbearing. Difference-in-differences modeling on birth certificate and U.S. census data finds that mandate states were associated with a small negative effect on birthrate overall (with a coefficient of -0.05). The presence of an infertility mandate increased first birth rates for white women 35 and older by 8%, significant to various robustness tests. But like DeLeire et al. (2011), once control variables (including demographic and geographic information) were introduced into regression, no significant effects of the mandates were found for any of the. 20 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(28) women in different age or racial groups. These studies are in line with the larger literature on the fertility effects of Medicaid and other insurance reform in the U.S.; while they seem to imply some positive correlation between health insurance and fertility, they also indicate that this relationship is either weak or not well understood. Relatively fewer studies have approached the potential pronatal effects of healthcare policy outside of the United States. Nakajima and Tanaka (2014) determined that subsidized health checkup services increased the probability of couples having at least one child in Japan.. 治政大 S.-S. Lee (2009) similarly finds the probability of a couple having a child by 0.5-0.9%. In Korea, 立 Their results show that a 10% increase in state expenditure for health checkup services increases. that introduction of the state-sponsored “health and nutrition system” supporting maternal and. ‧ 國. 學. child health as part of the First Basic Planning for Low Fertility and Aged Society policy, had a. ‧. positive effect on first and second-order birthrate in logistic regression in 2007. Other sections of this major population policy, such as maternal labor law reform and childhood educational. y. Nat. er. io. sit. reform, also had positive fertility effects during this period.. Debate continues over the usefulness of fertility policy, nonmaterial and material alike.. n. al. Ch. i n U. v. Population scholars, including Paul Demeny (1986, 2005) and Anne Gauthier (2007), argue that. engchi. policy is more likely to affect fertility timing than completed fertility; they further claim that the types of pronatalist policy acceptable to liberal developed cultures is often expensive and ineffective. The literature on pronatal policy and fertility remains inconclusive, in part because such policies often combine financial incentives for childbirth with non-cash incentives, making it difficult to pinpoint which program, if any, stimulated fertility. For example, in Frejka and Zakharov (2013), results show that two waves of comprehensive pronatal policy in Russia (during 1981 and 2007) that included material and nonmaterial birth incentives both failed to. 21 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(29) promote domestic fertility. Did none of the policies, from parental leave, to maternal capital, to child cash benefits, have any effect on domestic childbearing decision-making? Because of the bundling of cash and noncash benefits, it is difficult to compare the effectiveness of the individual policies from historical analysis alone. Analysis of the fertility effects of healthcare policy implementation or reform, on the other hand, often benefits from the availability of information about timing and benefits of individual policy changes. In addition, policy initiatives like Medicaid eligibility reform or. 治政大 period – this lends itself more readily to empirical assessment. 立. Taiwanese NHI implementation affect a large portion of the population in a relatively short. 3.2 NHI Implementation and Taiwan. ‧ 國. 學. 3.2.1 NHI and the Taiwanese Population. Of studies addressing the effect of NHI. ‧. implementation on population in Taiwan, only two have focused on increased prenatal. sit. y. Nat. healthcare, one of the program’s key offerings in 1995. An initial study by L. Chen et al. (2008). io. er. focused on the different ways women perceive prenatal care access pre- and post-NHI along the urban/rural divide, given that NHI implementation dramatically increased the number of. n. al. Ch. i n U. v. facilities where women could seek prenatal care nationwide. Using national survey data from. engchi. two cohorts before (1990-1992) and after (1998-1999) NHI implementation, L. Chen et al. (2008) examined where Taiwanese women sought prenatal care and the perceived convenience of that care for urban and rural residents. The study found that post-NHI provision rural women increased their demand for prenatal care in big hospitals, while urban women had no change in usage; there was no perceived change in convenience of transportation to care facilities by rural women, however, urban women did report increased convenience. Differences in transportation. 22 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(30) convenience are ascribed to the tendency of rural women to seek prenatal care in large hospitals rather than local clinics, due to their higher perceived value. C.-S. Chen, Liu, and Chen (2003) further analyzed changes to Taiwanese prenatal care usage before and after the 1995 NHI implementation, finding increased usage of healthcare services primarily among unemployed women and various aboriginal communities of central Taiwan. Data from the Taiwan Maternal and Infant Health Survey included two cohorts of Taiwanese women from all 23 administrative districts, the first who gave birth during May 1989. 治政大 C.-S. Chen et al. (2003) found to estimate the effect of NHI on prenatal care patterns in Taiwan, 立. and the second who gave birth during February 1996. Using a two-part negative binomial model. that maternity clinics experienced the highest volume of visits post-NHI, previously uninsured. ‧ 國. 學. working women and government employees increased their care use compared to female labor. ‧. workers and farmers, and care use increased most dramatically in central Taiwan. The authors imply that some of the consumption patterns seen in the study might have been due to societal. y. Nat. er. io. sit. trends, such as women delaying marriage, and as a result, more frequently turning to high technology solutions to help them conceive for multiple births at higher ages. This supports the. n. al. Ch. i n U. v. hypothesis that increased access to prenatal care from NHI implementation would influence. engchi. women’s fertility choices, as argued in the present study.. L. Chen et al. (2008) and C.-S. Chen et al. (2003) both indicated a significant uptake in care usage by rural women after NHI implementation, especially of in central Taiwan (a more rural area), despite different data sources and methods. At the same time, L. Chen et al. (2008) notes that the probability of rural women seeking prenatal care in large hospital increased 6.54 times between 1990 and 1999, which seems to contradict C.-S. Chen et al. (2003)’s finding that medical clinics saw a greater increase in demand for prenatal services in the post-NHI period.. 23 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(31) Both of these studies analyze usage of prenatal care services and various demographic characteristics – location of residence, insurance, industry of employment. These discrepancies possibly stem from unobserved differences in the birth cohorts selected by the two studies, or from recall bias issues. While NHI introduction overall increased demand for Taiwanese prenatal care overall and especially among rural women, it is unclear where exactly that demand was concentrated (in clinics or hospitals) and what reasons underlie women’s treatment preferences. Chou, Grossman, and Liu (2014) went beyond examining the effect of NHI on prenatal. 治政大 insurance data from the 1990 census data instead of randomized surveys, the study incorporates 立 care usage by analyzing the impact of NHI on Taiwanese post-neonatal infant mortality. Using. public and private sectors from before and after NHI implementation to determine whether NHI. ‧ 國. 學. provision lowered infant mortality. Before NHI implementation, only government employees. ‧. had full insurance coverage for childbirth and infant healthcare, but after 1995, coverage was extended to private sector and agricultural workers via NHI. Results show that NHI introduction. y. Nat. er. io. sit. decreased neonatal mortality by 8-16% for those born in farm households, but not in other sectors. Farm families in general have fewer financial, educational, and health advantages than. n. al. Ch. i n U. v. those in other sectors; the findings of this study thus “suggest that health insurance improves. engchi. infant health outcomes of population subgroups characterized by lower levels of education, income, and health” (p. 90). A final finding notes the marked reduction in the mortality rate of preterm infants in the post-NHI data, a difference of between 20-41% (p. 89). Overall, Chou et al. (2014) conclude that Taiwanese national health insurance improved neo-natal infant mortality rates in Taiwan, but not equally among different industries and population subgroups. Besides its direct effects on infant mortality and healthcare usage, Chou, Liu, and Hammitt (2003) also studied the indirect effect of NHI implementation on precautionary savings. 24 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(32) in Taiwan, using difference in differences methodology. They posited NHI as an ideal natural experiment to test the effect of insurance on savings, given the universality of its implementation and its lack of means-tested eligibility. Rather than redirecting healthcare savings into household savings, national health insurance decreased savings by 8.6-13.7%, with this reductive effect particularly strong for households with the highest financial risk—those in the lowest savings bracket. This study also uses a control group based on joint employment of husband and wife, where households with at least one government employee constituted the control group, like the. 治政大back to fertility through the The effects of NHI on precautionary saving may be tied 立. present study does.. concept of “yang er fang lao” (養兒防老), or raising children to provide for family members in. ‧ 國. 學. old age, as discussed by Lai and Tung (2015). In Taiwan, familial transfers have long been a. ‧. primary source of income security for the elderly, along with personal savings and public (state-. sit. y. Nat. sponsored) programs. Lai and Tung (2015) found that even as consumption level of the elderly. io. er. increased between 1985 and 2005, and health consumption costs increased from 9% to 17% during this period, the role of family transfers decreased significantly. If NHI constitutes a public. al. n. iv n C program that reduced income uncertainty enough to affect precautionary savings, it may also hengchi U have affected childbearing decisions that formerly played an important part in long-term financial planning. 3.2.2 NHI and Taiwanese Fertility. Despite a rich history of literature analyzing. population change, there remains a lack of consensus on the main drivers of low fertility today. Certainly, economic and social factors must jointly influence childbearing decisions in any culture, and the fertility decline is widely considered a social effect of industrialization and modernization in developed countries. Evidence from pronatal policies of the last half century is. 25 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(33) equally mixed; because of the mixed nature of reform, it is sometimes impossible to separate the effects of financial and nonfinancial fertility incentives and determine which is most effective. A variety of methodological challenges afflicts contemporary fertility studies. For country-level studies, period total fertility rate (TFR) is often the statistic of choice for the outcome variable. TFR, however, is a hypothetical statistic that generalizes average number of births per 1,000 women of average childbearing age, regardless of marital status and actual age distribution of mothers. The preferred statistic (when available) appears to be cohort fertility rate. 治政大 type of fertility data, period or other factors. Hoem (2008) reports an ongoing debate over which 立 (CFR), which by nature offers a higher level of detail on age-specific fertility, birth order, and. cohort, offers more insight. Period data may reveal short-term effects of policy, he claims, while. ‧ 國. 學. cohort data is more suited to examination of long-term social change.. ‧. In addition, endogeneity is an unavoidable issue in demography or fertility studies. Unobserved variables affecting both fertility and its potential causal factors, whether political,. y. Nat. er. io. sit. social or economic, undermine the validity of isolated statistical findings. Fertility change may drive policy; policy change may encourage or discourage fertility. Ashraf, Weil, and Wilde. n. al. Ch. i n U. v. (2013) portray academic frustration with these obstacles: “We detect a general cynicism. engchi. regarding the ability of social scientists to say anything useful about the economic effects of fertility – the issue is viewed as political rather than scientific, and conclusions from empirical analyses are assumed to reflect the preexisting views of authors” (p. 33). Despite these obstacles, the seeming inevitability of population aging and shrinkage spurs continued research into the underlying mechanisms of fertility change. In Taiwan, fertility has been a relevant policy topic for decades, first in the 1960s when the state promoted smaller family sizes through the 1964 family planning program, and again. 26 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(34) when sub-replacement fertility became an issue in the 1980s. A wealth of research has examined Taiwan’s initial transition from high to low fertility (D. S. Freedman, 1975; R. Freedman, Hermalin, & Chang, 1975; R. Freedman & Sun, 1969; Hermalin, 1976; Li, 1973; Rutstein, 1974), as well as its more recent decline from low to lower fertility (R. Freedman, Chang, & Sun, 1994; Keng & Sheu, 2011; M. Lee & Lin, 2016; Montgomery & Casterline, 1993; Parish & Willis, 1993) with regard to the social and economic factors driving population change. The question of how healthcare availability has affected Taiwanese fertility, however, has gone. 治政大implicit form of pronatal policy in There is a lack of studies on universal healthcare as an 立. unanswered.. the broader literature, as well. Given the existing lack of scholarly consensus on which pronatal. ‧ 國. 學. policies are most effective, it is important to continue questioning different avenues through. ‧. which the state might influence childbearing behavior. The present study approaches fertility from an economic perspective, on the foundational assumption that lowering the cost of children. y. Nat. er. io. sit. via state subsidized healthcare will lower the barrier to childbearing. Among Taiwanese fertility studies related to NHI implementation, most focus on child quality outcomes such as infant. n. al. Ch. i n U. v. mortality or usage of health services and not on how increased access to healthcare might affect reproductive decision-making.. engchi. 27 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(35) 4. Methods 4.1 Theoretical Framework The decision to have a child rests on a variety of different medical and non-medical factors including “income, child costs, knowledge, uncertainty, and tastes” (Becker, 1960, p. 231). Women’s fertility and healthcare decisions are jointly determined given that having a child requires women incur the financial cost of pre- and postnatal care, delivery, infant health services, as well as foregone salary from having a child. These factors affect women’s. 政治大. opportunity cost of childbearing and their lifetime allocation of time, influencing the utility. 立. model of the household.. ‧ 國. 學. A theoretical framework for understanding how health insurance might affect women’s childbearing decisions is presented in the Becker and Lewis (1973) child quantity-quality model,. ‧. which posits child quality and quantity are related under household budget constraint. When. sit. y. Nat. child quality (cost required for the rearing of each child) increases, the shadow cost of having. io. er. children (child quantity) also increases. At the same time, the higher the quantity of children, the costlier it is to increase the quality of each child because the cost of goods accumulates (Becker. al. n. iv n C & Lewis, 1973). With the 1995 introduction of NHI in Taiwan, h e n g c h i U low-cost or free access to a. variety of fertility and childcare services was extended to all Taiwanese citizens, permanently lowering the cost of having children. In essence, the shadow cost of child quality fell, allowing parents to recapture some opportunity cost of childbearing and diminishing the effect of the quantity-quality tradeoff. 4.2 Data The data source is the Women’s Marriage, Fertility, and Employment Survey (WMFES) conducted by the Statistical Department of the Taiwanese Ministry of the Interior. This survey,. 28 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(36) conducted periodically since 1979, interviews Taiwanese women aged 15 or older about marriage, childbearing, family composition, employment status, and other factors. Eligible respondents live throughout Taiwan, with women in the military and prisoners excluded, covering approximately 20,000 households in total. The questionnaire covers not only demographic characteristics, but also topics such as marital status and history, employment, childbearing, and child sex preference. Our analysis uses all available WMFES waves 1979-2016 (total of 18 years of surveys).. 治政大variable considered is number of related factors, making it ideal for analysis. The key dependent 立 The WMFES project provides recent detailed information on demographic and family-. 學. ‧ 國. children born per woman at time of survey, which is assumed to equal completed fertility.. Descriptive statistics on average fertility and sex composition of children (number of sons and. 1.98 1.28 1.38. Pre-1960 birth cohort (including 1960) Fertility (live births) 279,690 Son fertility 279,690 Daughter fertility 279,690. 3.61 1.89 1.73. Post-1960 birth cohort Fertility (live births) Son fertility Daughter fertility. 1.78 0.93 0.85. sit. 3.25 1.70 1.55. n. al. Minimum. Maximum. er. Std. Deviation. io. Mean. Whole period Fertility (live births) Son fertility Daughter fertility. Observations. y. Table 1: Fertility of Married Women. Nat. Variable. ‧. daughters) from the WMFES survey are provided in Table 1 for married respondents.. 347,777 347,777 347,777. Ch. 68,087 68,087 68,087. engchi. i n U. v. 0 0 0. 20 20 14. 2.00 1.30 1.43. 0 0 0. 20 20 14. 1.01 0.81 0.85. 0 0 0. 10 10 6. 29 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(37) The average number of births per woman for the whole sample is 3.25 with a mean of 1.70 boys and 1.55 girls. The sample is further divided into two groups: women born before 1960 and after 1960. Those in the pre-1960 birth cohort would be 35 years old in 1995, at the time of NHI implementation. Fertility is significantly higher in the older group, at an average of 3.61 children born per woman; in the younger group, only a mean of 1.78 children born. More sons than daughters are born in both groups, but the effect is more significant in the older cohort, pointing to a possibility of underlying sex preference.. 治政大 These questions relate to employment, educational attainment, and child sex (son) preference. 立 Other questions of interest in the WMFES survey include marital status, spousal. the insurance coverage of the family (families containing a state employee already had state. ‧ 國. 學. health insurance) and power dynamics of the family, which may influence the economic cost of. ‧. childbearing for mothers and influence overall fertility. Additional descriptive statistics on these variables for the whole sample, control group, treatment group, and pre- and post-1960 birth. y. Nat. er. io. sit. cohorts are shown in Table 2. The control group consistently had a higher percentage of spouses who were also government employees, for both the pre- and post-treatment groups. Educational. n. al. Ch. i n U. v. attainment increased over time, and in the highest category (college and above) was higher for. engchi. the control group, with a difference of 47.39% in the whole sample, 43.45% in the pre-treatment group, and 51.75% in the post-treatment group. Child sex preference declined over time and was lower for the control group of government employees than the treatment group in all periods.. 30 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(38) Table 2: Characteristics of Married Women by Treatment Status Whole sample. Whole period Husband is a government employee Educational attainment None-elementary Middle-vocational school College and above Son preference N. 39.24. 7.14. 61.08 30.26 8.66 33.82 347,777. 11.90 34.00 54.10 26.19 14,374. 63.20 30.10 6.71 65.85 333,403. 8.55. 7.33. 16.31 36.65 47.04 33.46 10,237. 75.34 21.07 3.59 38.78 269,453. n. y. sit. er. io. al. 40.60. ‧. 73.18 21.64 5.18 38.58 279,690. Nat. Post-1960 birth cohort Husband is a government employee Educational attainment None-elementary Middle-vocational school College and above Son preference N. 8.47. 政治大. ‧ 國. 立. Treatment group: Private employees %. 學. Pre-1960 birth cohort (inclusive) Husband is a government employee Educational attainment None-elementary Middle-vocational school College and above Son preference N. %. Control group: Government employees %. Ch. 8.17. e11.36 ngchi 65.67 22.97 14.28 68,087. i n U. v 35.90. 6.37. 0.99 27.44 71.57 8.22 4,137. 12.03 68.14 19.82 14.67 63,950. 31 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.

(39) 4.3 Empirical Methods According to statistics from the R.O.C. Ministry of the Interior 2, in 1995 the average age of Taiwanese mothers at birth was 27.7 years. NHI regulations list 34 years old as the minimum age for receiving advanced maternal age NHI insurance benefits; this study uses 35 years as the first cutoff point for DID empirical analysis. Women younger than 35 years old in 1995 were likely to have future fertility decisions affected by NHI childbirth and childcare benefits (treatment group), and women older than 35 were less likely to have their fertility decisions. 治政大 during NHI implementation in Taiwanese women aged 15 years and older (born in or after 1980 立 influenced by NHI (pre-treatment group). For the current study, respondents of interest are. y. al. n. 100,000. sit. io. 120,000. er. 140,000. 80,000. ‧. Nat. 160,000. Number of Births by Age Group for 1959 & 1971 cohorts Source: Ministry of the Interior. 學. 180,000. ‧ 國. 1995. Figure 2 compares data from the Ministry of the Interior on the number of births for the. 60,000. Ch. engchi. i n U. v. 40,000 20,000 0 -20. 20~24. 25~29 born in 1959. 30~34. 35~39. 40~44. Age Group born in 1971. Figure 2: Number of births by maternal age group for 1959 and 1971 cohorts. 2. Data available at: https://www.gender.ey.gov.tw/gecdb/Stat_Statistics_DetailData.aspx?sn=lT4902z3YmLGBZadLKLSzQ%3D%3D. 32 DOI:10.6814/THE.NCCU.IMPIS.011.2018.A06.