اثرات طولانی مدت شرایط بازار کار بر تشکیل خانواده برای جوانان ژاپنی

This study aims to examine how each cohort’s family formation is affected by labor market conditions experienced in youth in Japan. Although the deterioration in youth employment opportunities has often been blamed for Japan’s declining fertility rate, the effect of slack labor market conditions on fertility is theoretically unclear. We estimate the effects of regional labor market conditions at entry to the labor market and contemporaneous conditions on fertility, controlling for nation-wide year effects and prefecture fixed effects, and find the following. First, high school-educated women who experienced a recession while entering the labor market are less likely to have children. In contrast, a recession rather increases fertility among college-educated women. When summed up, the aggregate impact of labor market conditions experienced in youth on fertility is weak. Second, the unemployment rate at entry to the labor market is positively correlated with the probability of having two or more children conditionally on having at least one child. Third, the contemporaneous unemployment rate is negatively correlated with marriage of women in the local labor market, although the correlation is weak and concentrated on the less educated group.

In this paper we examine how Japanese women’s fertility is affected by labor market conditions experienced in youth using the dataset of individual-level information on fertility history. As is the case in many developed countries, in Japan, the fertility rate has been falling for decades. At the same time, unemployment and non-regular employment among youth have been rising (Fig. 1). Because of the concurrence of these events, the deterioration of youth employment has often been blamed for the declining fertility rate (e.g. Yamada, 2007 and Matsuda, 2009).Economic theories, however, do not predict that deterioration in youth employment prospects necessarily leads to lower fertility rates. A recession can have a positive substitution effect because worse employment opportunities for women lower the opportunity costs of childbearing. Therefore, the observed negative correlation between the unemployment rate and the fertility rate in Japan might be a mere coincidence. Indeed, empirical studies using data from the United States have established that worse opportunities for women in the local labor market can actually increase marriage and fertility rates, implying that the positive substitution effect tends to dominate the negative income effects that derive from a deterioration of male employment opportunities ( Butz and Ward, 1979, Schultz, 1994, Blau et al., 2000, Dehejia and Lleras-Muney, 2004 and Kondo, forthcoming). 1 In Japan, however, the fertility rate remains low in a recession, suggesting that the negative income effect dominates the positive substitution effect on fertility. We suppose that two important factors peculiar to Japanese labor market are in play for suppressing the demand for children by cohorts that faced a recession in their youth: less flexibility for inter-temporal substitution of labor supply and the persistent effect of a recession experienced at entry to the labor market. The first factor is based on the thin mid-career job market. In Japan, most firms prefer to hire new high school or college graduates over mid-career workers, and they favor employees who have stayed at the firm since their graduation in internal promotions. Therefore, workers have incentives to remain as long as possible at the same firm, and quitting a job is considered as a serious disadvantage. If a woman quits a job due to marriage or childbearing, it is hard for her to find a fulltime regular job after a few years’ break. As a result, a considerable number of women choose to keep their jobs over marriage and childbearing.2 Furthermore, the number of non-regular workers on temporary contracts has increased rapidly among unmarried young women since the late 1990s. Since few of them can take even an unpaid maternity leave, it is even more difficult for them to keep their jobs while raising children. In fact, women on such non-regular employment tend to delay marriage and have fewer children (Nagase, 2002, Sakai and Higuchi, 2005 and Kitamura and Sakamoto, 2007). Hence it is harder in Japan to adjust labor supply inter-temporarily, and the positive substitution effect may be weaker. Another important factor peculiar to Japan is that labor market conditions in the year of graduation have very persistent effects on employment and earnings for the affected cohort. As in many other countries, new school graduates are hit more severely by a recession than prim-aged workers because it is easier for firms to reduce hiring than firing incumbent employees. In addition, as described above, Japanese firms prefer to hire new school graduates and favor employees who stay at the same firm for long time, thus there are few opportunities for cohorts who enter labor market during a recession to catch up by changing their jobs. Consequently, the labor market conditions at graduation have very persistent effects on employment and earnings in subsequent years (Genda et al., 2010). Such a decline in permanent income may affect the total number of desired children and the number of people who will never have a child. Furthermore, earning capacity is an important factor that Japanese women care about when choosing a spouse,3 thus, a persistent decline in income of young men (i.e. their potential spouses) may discourage marriage. Therefore, in order to predict Japan’s future demographic trends, it is very important to explore the extent to which the persistent effect of recessions in the year of graduation affects women’s family formation behaviors, in addition to the effects of contemporaneous labor market conditions. Existing studies on the relationship between labor fertility and labor market conditions in Japan vary significantly in their conclusions, depending on the data and methodologies used. Among the studies using prefecture-level aggregate data, on the one hand, cross-sectional analyses using Census 2000 data report a positive correlation between employment opportunities for young men and contemporaneous marriage and fertility rates (Ogawa, 2003, Ohta, 2007 and Kitamura and Miyazaki, 2009).4 On the other hand, Ogura and Dekle (1992) apply the first-difference method to prefecture-level panel data from Census 1970–1985 data and find a negligible effect of men’s earnings on marriage and fertility rates. All of these studies report a modest negative effect of better labor market opportunities for women on marriage and fertility; therefore, the overall effect of a recession on marriage and fertility rates remains ambiguous. Furthermore, studies using micro-level data also vary in their conclusion: Shimizu (2002) finds that a decrease in women’s earnings raises fertility using Japanese General Social Surveys, whereas Higuchi et al. (2007) find that the prefecture-level active job opening to vacancy ratio (yuko kyujin bairitsu) does not affect the fertility of married women using data from the Japan Panel Study of Consumers. Lastly, Abe (1999) and Higuchi (2001) find that women who graduate from school in a year when labor demand is slack tend to marry and bear children earlier, while the contemporaneous unemployment rate delays marriage and childbearing. Such a lack of consensus among existing studies can be attributed to the lack of an appropriate dataset that provides individual-level information on fertility and marriage history and has a sufficiently large sample size. We aim to overcome this limitation by constructing individual-level fertility history data from the Employment Status Survey (ESS), supplemented by prefecture-level panel data for marriage histories from the Census and Vital Statistics. These datasets are augmented with the regional unemployment rates taken from the Labour Force Survey. By exploiting geographical variation in unemployment rates, we control for nation-wide time trends by including year dummies. Further, the ESS’s large sample size allows us to investigate the effect for subsamples divided by education, which sheds light on differential effects across skill or income levels. Another contribution of this study is that we control for nation-wide year effects and region fixed effects to obtain estimates that are more robust to trend changes.5 We limit our sample to women because the effects of the unemployment rate on mobility and educational attainment, which are discussed in the Appendix, are not negligible for males in the ESS.6 Our estimates suggest that high-school educated women who experienced a recession at entry to the labor market are less likely to have a child. In contrast, a recession increases fertility among college-educated women. This weak positive income effect among more educated women is consistent with existing evidence that income loss due to a recession at entry is smaller for college-educated men, who are more likely to marry with college educated women, than for high-school educated men ( Genda et al., 2010). That is, college educated women are less likely to be credit constrained during a recession and thus have more room to adjust the timing of childbearing and labor supply according to the changes in their opportunity costs. However, this positive effect for more educated women is canceled out by the negative effect for less educated women. Consequently, the overall impact of labor market conditions experienced in youth on family formation is weak. The rest of this paper is organized as follows. The next section reviews the theoretical background for the effect of labor market conditions on fertility. The third section describes our data, while the fourth reports our empirical findings. The final section provides concluding remarks.

We have shown that the aggregate effect of local labor market conditions on fertility is subtle. However, a breakdown by educational background reveals that the effect of unemployment rate at entry to the labor market on fertility is negative for less educated women and positive for more educated women. The positive effect for more educated women is consistent with the substitution effect typically observed by existing studies in the United States. The negative effect for less educated women may be attributable to a stronger income effect, the lack of job security during the maternity leave, or the lack of matching opportunities for non-regular female employees. In any case, the overall effect of labor market conditions experienced in youth on family formation is weak, despite the well-documented persistent effect of labor market conditions at entry on subsequent employment and earnings. Although it might have delayed marriage timing slightly, the deteriorating employment prospects for Japanese youth is not likely to be the primary reason of the declining fertility. This result implies that, even though restoring the employment prospects of young Japanese is an important policy challenge on its own, it is unlikely to be an effective policy for increasing the fertility rate.