Analysis of the Effects of Macroeconomic Conditions on the First Marriage *
April 25, 2014
Keiji Katsura, Graduate School of Economics, Hosei University
Seiritsu Ogura, Hosei Institute on Aging, Hosei University
Abstract
The purpose of this paper is to find out if the macroeconomic indicators such as the economic
growth rate or unemployment rate affect the timing of the first marriages. We have analyzed the
timing of the first marriages by Discrete Time Method using the first marriage data in “The Japanese Longitudinal Survey on Employment and Fertility (LOSEF): the 2011 Internet Version”
(conducted by professors Takayama, Inagaki and Oshio), a retrospective panel data for education, job history, as well as the adolescent family environments for more than 4 thousand individuals whose birth years span three decades, starting from 1950. We have found that, for both
men and women, good economic conditions shortens the time to first marriages, while higher
unemployment in graduating years tend to delay them. We have confirmed that for men, regular
employment is important to marriage, but it is not the case for women, which is what the
standard search theory for marriage suggests. On the other hand, we have found men and women from high income families tend to marry early, contrary to what has been found in previous
works.
*This paper is based on a part of the Master’s Thesis of Keiji Katsura presented to Hosei University in March 2014. The research was supported by the special grant of Ministry of Education and Technology (22000001). We are particularly grateful to professor
Inagaki and professor Takayama for making “The Internet Survey on the Public Pension Records, and Living and Working Conditions” available for this research.
1
Analysis of the Effects of Macroeconomic Conditions on the First Marriage
Keiji Katsura
Seiritsu Ogura
1.
Introduction
There is a consensus in Japan that the low fertility rates of the last two decades are
the results of increase in the number of single women and the age at first marriage
(Iwasawa 2002). Many economic demographers now believe that, ceteris paribus, marriage rate increases with the increase in the male earnings, but decreases with the increase in the female earnings (Blau et al 2000). A number of Japanese empirical studies have shown the increase in the proportion of women getting higher education, getting good jobs and the reduced gap in earnings between women and men can explain a
large part of the decline in marriage and fertility (Ogura and Dekle1992, Takayama et
al 2000, Ogura and Kadoda 2010).
Although a large number of studies have investigated the effects of business cycles
and fertility and marriage for the U.S. and the other developed countries, most of these
studies have been based on observations of aggregated data of these countries (Butz
and Ward 1979, Winegarden 1984, Macunovich 1994, Sobotka et al 2010, Goldstein et
al 2013), including the ones for Japan (Ogawa and Mason 1986, Lee and Gan 1989,
Ogawa and Retherford 1993, Ueno 1998, Retherford, Ogawa and Matsuoka 2001, Ogawa 2003). There have been a limited number of micro-econometric studies, most of
which are based on cohorts born between relatively narrow intervals.
Based on the assumption of differential opportunity costs of unit time between working and non-working wives, a well-known study using the aggregated data of the U.S.
presented an evidence of the counter-cyclical relationship of fertility (Butz and Ward
1979). However, a subsequent analysis cast serious doubts on the robustness of their
findings (Macunovich 1993). Recent experience in Europe and Japan suggest that in a
severe recession, as many young men out of schools cannot find stable employment,
young women are not tempted to marry them, and fertility rate falls (Ogura and Kadoda 2010). Thus the relationship between economic conditions and marriage may not
be a simple one.
The purpose of this paper is to examine the effects of macroeconomic conditions on
marriage, by estimating a discrete survival function on the time to first marriage starting at age 18. Our estimation is based on a panel data called “The Japanese Longitudinal
Survey on Employment and Fertility (LOSEF): the 2011 Internet Version” (conducted by professors Takayama, Inagaki and Oshio), which is a retrospective panel data for education, job
history, as well as the adolescent family environments for more than 4 thousand individuals.
This data offers the marriage experience of individuals whose birth years span more than three
decades, starting from 1950 to 1981. The macroeconomic conditions when they were most marriageable ranged from the last phase of high growth era to the glacier period of the last two
decades, with bubble economy in the middle. Such a wide swing in economic environments
happening in relatively short period of time provides a very good opportunity to examine their
effects on marriage behavior. Our unit of time for observation is one year and our analysis
covers more than four decades, starting in 1968 and ending in 2011.
Following the previous studies, we control for such individual characteristics as the
types of current employment, number of years of education, and types of first employment after graduation. Various socioeconomic characteristics of the household in which
the person grew up are controlled too. Macroeconomic conditions are represented by
the current rates of economic growth and the rate of unemployment in the graduating
2
year. The latter variable captures the long-lasting effects of the macroeconomic conditions at the time of graduation on subsequent personal income levels and employment.
To preview our results, pooling all samples, we find that good macroeconomic conditions promote marriage for both men and women, but their effects are not large. A high
unemployment rate at the time of graduation works as a persistent inhibitor of marriage for both men and women. The strongest promoter of marriage is regular employment for men, but it is irregular employment for women. The strongest inhibitor of
marriage is being a student for both men and women. Being unemployed immediately
after graduation seems to have a persistent negative effect on marriage, for both men
and women.
Estimating by birth cohorts revealed somewhat different stories; for cohorts who
reached age 20 prior to 1990 in the bubble economy, good economy still promotes marriage, and bad economic condition at the time of graduation still delays marriage. On
the other hand, for women who reached age 20 after the burst of the bubble economy, a
higher rate of unemployment at the time of graduation promoted marriage. The last
point may indicate that marriage may be now a new risk-sharing arrangement on the
part of women in today’s unprecedented economic hardships, although more research
on this point is needed.
2.
2.1
Background for our research
Changes in marriage market in the last fifty years
Increases in the age at first marriage and in the number of unmarried women have
contributed to the rapid decline in fertility in Japan since 1970. As the number of babies born outside of marriage accounts for only 2 % of all babies born in Japan, the decline in marriage rates has been directly translated to the declines in fertility rates
(Ogura 1994, Iwasawa 2002).
In terms of marriage rate, we can divide the post WWII history of Japan into two distinct periods; namely, the period of universal marriage that lasted until the mid-1980’s,
and the second period of individual search.
During the period of universal marriage, almost everyone got married, many with the
help of “arranged marriage” 1 . The marriageable age changed a few years over the
years, having been pushed by college education; most women got married before they
reached 25 years old, and most men got married before the age 30. The universal marriage during this period was supported by what economists termed as “Japanese system of corporate management”, or lifetime employment with seniority based wages. In
the formal labor market, regular jobs were reserved for men, and only a small number
of jobs like teachers or nurses were open to women. Every year, large firms hired exclusively graduating men and women, with an understanding that men were expected to
work for life, but women, given mostly manual or subsidiary jobs, were expected to
leave when they find someone to marry. If women had stayed on the job, without regular promotions given to men, their earnings would have been one-third or one-quarter
of male counterparts. Most women did not have an alternative life-plan to marriage.
The period of universal marriage had began with the era of high economic growth
In spite of the word “arranged”, only the initial introduction in the search process was
arranged, and the marriage itself had not been arranged. Both families, choosing the
acceptable ones from the lists of candidates who had circulated their bios and photographs for formal introduction, meet a few times before they decide to get married or
not.
1
3
around 1950, and it managed to last almost three decades until the middle of 1980’s.
Why did this period come to an end? Simply put, it ended because Japanese firms, no
longer could sustain the sexual discrimination in the labor market, began hiring women for jobs once reserved for men, and gave women an alternative to marriage (Ogura
2001). Several factors including the currency shock of 1971, and oil-shock of 1973, contributed to this change. On the supply side, by 1970’s, middle-income families, which
used to send only their sons to college, could now afford to send their daughters too,
and hence the supply of women with higher education was abundant.
As a result, in the economic “bubble” of the second half of 1980’s, the second babyboomers reached their marriageable age, and many of them discovered that the economic gains from marriage were no longer overwhelming. Instead of going through the
motion of “arranged marriage” as their preceding generations had done, they would
rather keep on working until she would meet a marriageable man. For the first time in
modern history of Japan, women were given a right to remain single for life, without
condemning herself to a lifetime of poverty.
There are two points we would like to make on the relationship between women and
the labor market at this moment. First, after some initial confusion, marriage still remains a lifetime goal of most women. The proportions of women who give higher priority to work over marriage are still small. According to the National Survey on Marriage
and Childbirth, in 2010, 89.4% of surveyed single women replied that they are going to
get married eventually, and only 6.8% of them replied they intend to remain single for
life. These proportions have remained stable in the last two decades2.
Secondly, women’s pursuit of work opportunities and higher income still seems to be
conditioned by their marital status. In particular, in the last three decades, we have
not observed an increased tendency among “married” women to return to work. If we
look at the changes in the labor force participation rates during this period, among single women, it has increased by more than 10% points3. Among married women, however, it has changed little4. Comparing the labor force participation rates between 2001
and 2011, the rates have increased in 25-29 year old class and in 30-34 year old class,
but changed little in the other age classes (Fig 1-3-1). These changes seem to indicate
that many women keep on working a few more years after marriage, but once they
leave their jobs to have children, they do not return to labor market.
2-2.Previous studies on the determinants of marriages
In explaining why people get married, Becker (1973) likened marriage to trades between men and women. The economic gains from marriage, according to Becker (1973),
are based on the different comparative advantages of men and women. Men have comparative advantage in market labor and women have comparative advantage in household labor.
A sufficiently large difference in market wage rates between men and women enables
almost all women to get better off by getting married, and trading her market labor for
household labor. As the difference narrows, however, a woman can benefit from marriage only if her partner’s earnings exceed a certain minimum level, and hence she
starts searching for a marriage partner. Many previous studies presume that such a
2 The only exceptions are the second half of 1980’s and the first half of 1990’s. Among
women between ages 30-34, the proportion of “single for life” jumped to 16.9% in
1987,but dropped to 12.6% in1992. In subsequent surveys, the proportion fell back to
below 10%.
3 From 52.6% in 1980 to 63.2% in 2011.
4 From 52.6% in 1980 to 49.0% in 2011.
4
search is similar to a job search in labor economics. Just as a job searcher sets the minimum acceptable wage offer and accepts any offer above it (MacCall 1970), a woman in
search of a marriageable partner sets a minimum acceptable earning level, which we
will call her reservation wage, and accepts the first offer above it.
Most of the empirical works on the timing of marriage can be regarded as applications of this search model with reservation wages. For many women, the reservation
wage is an increasing function of their own earning; for example, it can be simply their
current earnings multiplied by some factor similar to an equivalent scale. As such, an
individual attribute that enhances her earning capacity (e.g. higher education, regular
employment etc.) increases her reservation wage, reduces the number of qualified men,
and delays the timing of marriage. In contrast, an attribute associated with a lower
earning capacity (e.g. compulsory education only, irregular employment or unemployment) decreases the reservation wage, and shortens the time to marriage.
There are other factors that need to be considered. In Japan, the role of education
may be one of them. In the seniority-based system of Japanese firms, there is very
small difference in wages among the workers of the same age in their twenties, but the
difference widens as time goes on, with education serving as the most important signal
for the growth in earnings in the middle age. The increase in the number of women
with college education, however, tends to decrease the number of qualified men.
For most Japanese women, moreover, the reservation wage is not just a function of
their own earnings, but also a function of the standard of living they would like to
have. Women from higher income households may ask for a higher reservation wage
than their own earnings can justify. For example, Higuchi and Abe (1999) and Ogawa
(2002) argued that single women tend to set the reservation wage on the basis of their
own father’s income. The observed delay in marriage, according to Higuchi and Abe
(1999), reflects the smaller number of young men who can meet reach the same income
as achieved by their father’s generation, in the stalled macro economy.
3.
Macroeconomic conditions and marriage
3-1.Previous studies and the macroeconomic conditions
Previous studies on marriage focused on the relationship between the delay in
marriage and the increase in reservation wage as well as in the opportunity costs of
childbirth, and analyzed the effects of the individual properties on their marital behaviors. Few included macroeconomic indicators in their analyses5. In this paper, we are
going to analyze the effects of macroeconomic variables on marriage behaviors.
3-2.Macroeconomic conditions at the time of graduation
When are the marriage decisions of young men and women most susceptive to macroeconomic conditions? The answer should be at the beginning of their work life, with
no savings to buffer the variation in income. There are a number of interesting studies
on the impacts of macroeconomic conditions at graduation on subsequent employments
and incomes in Japan. For example, Sato (2008) has shown that the labor market conditions at graduation influence not only the qualities of the first jobs but also the subsequent jobs, and, in particular, those who have been employed as irregular workers at
We have found one study (XXXX) that has included the unemployment rate as an explanatory variable of the timing of first marriage, and found it does not have a statistical significance.
5
5
graduation tend to be employed as irregular workers in subsequent jobs as well, facing
a long-term insecurity. Also Sakai and Higuchi (2005) have shown that those who had
become part-time workers following graduation tend to marry late and have children
late, due to their low income. Similarly, Ohta, Genda and Kondo (2005) have shown
that individuals in cohorts who had their first jobs in difficult economic times continue
to face low real wage rates and unstable employment, compared with those in cohorts
who had gotten their first jobs in good economic times. While in most developed countries, the effects of the initial conditions are shown to last for at most 10 years before
they disappear, in Japan, the end of these effects are not yet established (文献).
3-3.Effects of individual characteristics
Among the individual characteristics that influence the probability of marriage, the
most important property is the type of employment. According to Nagase (2002), irregular employees tend to marry later than regular employees. Subsequently, Mizuochi
(2006) has shown that male regular employees marry earlier than male irregularly
employees, but found no such difference among women with different employment
types.
Another important aspect of individual characteristics is the education level. Comparing the individuals with different education levels, Ohtani (1989) has found, while
among women less than 22 years old, women with more education had lower rates of
first marriage than women with 9 years of education, that among men under the age
25, those with less education had lower rates of first marriage. Subsequently, Mizuochi
(2006) has found that, for both men and women, from the time they leave their last
schools, more educated individuals take less time to get married than do less educated.
Nozaki (2007) also confirmed the finding of Mizuochi (2006), by observing that, while
individuals with less education enter the labor market at younger ages and get married
younger, it actually takes them more time, from the time they get the first job, for them
to get married, and showed that those with college education or more take the least
amount of time to get married from the time of graduation.
4.Methodology
4-1.Data
Our analysis is based on the first marriage data in “The Japanese Longitudinal Survey on
Employment and Fertility (LOSEF): the 2011 Internet Version” (conducted by professors Takayama, Inagaki and Oshio), which is a retrospective panel data for education, job history, as well
as the adolescent family environments for more than 5 thousand individuals whose birth years
span three decades, starting from mid-1950’s. The survey is designed to take advantage of
the periodical provision of individual pension and employment records from the public
pension authority, and ask the respondents to add retrospectively the information on
education, the changes in jobs, types of employment, marriage, births, living arrangements with parents, location, types of jobs as well as the information on current living
conditions, family and lifestyles.
The dataset contained information of 5,953 individuals under the age 61. We have
excluded some samples following our exclusion criteria. The first are the individuals
who had married under the age 18. Legally speaking, to get married under the age 20,
one needs a parent’s consent. In addition, men have to be over the age 18, and women
have to be over the age 16. Only a small fraction of population has married under the
age 18, so this exclusion should not matter too much. In addition, as the individuals
thus excluded are distributed uniformly across generations, this exclusion should not
influence our estimation results too much.
6
The second exclusion concerns the observations above some age. Since our interest is
on the first marriage than can produce children, we wanted to put some reasonable upper limit on the age. We have selected age 41 as such a limit, and individuals who married after the age had been treated as right censored data.
After these exclusions, for our analysis of survival time, we had data for 4,044 individuals, born between 1956 and 1981, during the period of 1968 to 2011. Although
there were slightly more males (2,082 individuals, or 51.5%) than females (1,962 individuals, or 48.5%) in the sample, females experienced more “failures”, with males accounting for 1,478 “failures”, and females accounting for 1,644 failures.
4-2.Methodology
For our analysis, we use the analysis of survival time in this paper. This method enables us to analyze the causal effects of the factors that may contribute to the onset or
the end of a phenomenon under observation. The advantage of this analysis is that it
takes account of censoring or censored data, which is important in realistic survival
time data. We define the failure of our analysis as the first marriage of an individual.
For each individual, we start counting the number of years from the moment an individual reaches age 18, end our counting when the individual gets married, but we end
the counting when the individual reaches age 41, or end by year 2011, treating all the
samples as right-censored for which no marriages have been reported.
Among the available models in the survival analysis, we have chosen a discrete time
method. We have started our estimation using a standard, semi-parametric, Cox proportional hazard model that had been used in the previous works on marriage. Unlike
non-parametric models, a proportional hazard model allows us to introduce covariates
and to examine their effects on survival time, without worrying about the risks of misspecification, by selecting a wrong distribution function. For a proportional hazard
model to be valid, however, each covariate in the data has to satisfy the proportionality
assumption, which, in practice, is very difficult. As a global statistical test for such proportionality, after estimating our proportional hazard models, we have computed
Schoenfeld residuals, and carried out chi-square tests. Unfortunately, we have found
the result significantly different from zero, indicating that our data did not support the
proportionality assumption (Table 4.2.1).
As our next step, we have tried to choose the best model among the popular parametric survival models (Table 3-4-2). We have compared the goodness of fit measures in the
estimation results of exponential distribution, Weibull distribution, Gompertz distribution, log normal distribution, log-logistic distribution, and gamma distribution, but
found relatively small differences. The AIC did not give us a clear-cut choice either, and
we could not resolve our concern about the misspecification (Table 4.2.2).
As a result, we have chosen a discrete time method that can handle both proportional
hazard and other types of hazards (Allison 1982). Moreover, since our data is a retrospective data in which events are reconstructed by the year, this analytical model is a
particularly suitable one.
4-3.Selection of variables
Following the specifications of previous studies, as our main explanatory variable, we
have selected the type of current employment. The type of our current employment
consists of four categories (regular employment, self-employment, non-regular employment, and student), out of which we have chosen regular employment as the base
category. In addition to the type of current employment, we have added an indicator
variable for the non-regular employment of the initial job. This will make it possible to
analyze the lasting effects of the initial condition at subsequent ages.
7
As to the education level variable, we have used the total number of years at school.
High school graduates are given 18 years, graduates of junior colleges or special vocational schools 20 years, college graduates 22 years, and those with post graduate degrees 24 years. The variable is treated as a continuous variable.
As to the macroeconomic indicators, we have selected the rate of economic growth of
the previous year at each age, as well as the rate of unemployment of the last year in
school. The former is a time-varying variable, while the latter is a time-invariant variable.
We also used a number of indicator variables to capture the effects of pre-adolescent
family backgrounds, including “My family subscribed to a newspaper”, “My family had
own bathroom”, and “My parents constantly quarreled”, respectively, at the time when
the respondent was 15 years old. Also the Rikert-type scale rating (1: significantly lower,..., 5: significantly higher) of the household income at age 15 relative to the neighbors’ was used in its original form.
5.Results and discussions
5-1.Results of Pooled Cohort Data (both sexes, separate male and female)
We present our estimation results using all cohorts in our sample, first using pooled
male and female data, then separate results using male data and female data.(Table
5.1.1)(Table 5.1.2)(Table 5.1.3)
First of all, the estimation results of pooled cohort data indicated that women tend to
marry earlier than men, and hence we decided to estimate our survival functions separately for men and for women.
The non-regular employment dummy of the first job shows a significantly delaying
effect on marriage, for both men and women. The dummy is an indicator variable for
the employment categories of the self-employed, part-time workers, and temporary
workers.
Regarding the effects of current employment, however, there is a subtle difference
between men and women. For men, all the types of employment other than regular
employment (i.e., self-employment, irregular employment, and student) significantly
delay marriage. Being a student naturally has the largest negative effect, followed by
irregular employed. Even self-employed has a negative effect. On the other hand, for
women, being a student has the largest negative effect, but, contrary to men, irregularly employment now has a significant positive effect. In other words, women’s own irregular employment promotes marriage, rather than delaying it. It is consistent with
the reservation wage hypothesis6; irregular employment reduces it and increases the
number of qualified single men.
The number of years of education does not have a significant effect for men, one way
or the other, but it delays marriage significantly for women.
Macroeconomic factors have significant effects on marriage. A higher rate of economic growth shortens time to marriage for both men and women, but the effect is larger
for women. On the other hand, a higher rate of unemployment in the graduating year
Another possibility is the self-selection; namely those who want to marry earlier opt
for irregular employment, rather than for regular employment. In our data, however,
we did not worry about this possibility because the null hypothesis of the exogeneity
assumption of all employment types could not be rejected (p>0.53).
6
8
delays marriage. In terms of the absolute value, the coefficient of annual growth rate is
less than one-half of the macro unemployment rate. The effect of the macro unemployment, in turn, is less than one-half of the non-regular employment.
Family background variables at age 15 do not seem to have significant effects on the
time to marriage, except the household income. Moreover, the income variable has positive effects for both men and women. The result is natural for men. It probably indicates that those men raised in higher income households enjoy unobserved advantages
in the search such as graduating more prestigious schools, better education, or more
wealth.
Not only the coefficient of income for women is positive, but also it is almost twice
that of men. This result contradicts a naïve prediction that, ceteris paribus, daughters
from rich families, having set reservation wages higher by observing their fathers’
earnings, will spend more time looking for marriage partners. There are several ways
to explain this result. Firstly, daughters may not simply set reservation wages by observing their fathers’ earnings, but by observing wage distribution of men in their own
cohort. Alternatively, our result may indicate that more affluent families have several
ways to improve the odds of their daughters search. Secondly, richer families can afford
more intensive search for qualified partners for their daughters in each time period.
Thirdly, more affluent parents can give various financial incentives for their daughters
to marry desirable candidates by side-payments. Fourthly, even without direct payments, marrying daughters from affluent families can have long-term benefits for men.
5-2.Estimation results by cohorts: results and observations
We have shown three estimation results for each birth cohorts (1950’s, 1960’s and
1970’s) pooling data of both sexes, male data and female data, separately. As we have
discussed earlier, those born in 1950’s spent their marriageable age in the period of
universal marriage, those born in 1970’s or later came after its collapse, and those born
in 1960’s are somewhere in between. In terms of macroeconomic conditions, those born
in 1960’s had been exposed to the extremely good macroeconomic conditions in their
marriageable age, those born in 1970’s have been affected by the deep recession during
the lost decades.
For women, employment categories do not seem to have significant effects, except for
student. For men, however, interesting results have been obtained. Among men who
were born in the 1950’s, and reached adulthood in the 1970’s, no significant effects of
employment categories were observed either, except for students. This is consistent
with the practice of the universal marriage. But among men who were born in the
1960’s, and reached adulthood in the 1980’s during the bubble economy, in addition to
students, irregular employment showed a clear negative effect. Among men who were
born in the 1970’s, and reached adulthood in the 1990’s after the burst of bubble economy, self-employed were added to the significant negative list. One possible interpretation of these results is that, with the increase in economic uncertainty since 1980, the
regularly employed men have become more attractive as partners of marriage, due to
their economic stability.
Number of years of education showed modest, but consistent negative effects for
women in each of these three cohorts. For men, it did not have a consistent sign, except
for the small but significant negative sign for those born in the 1970’s. The future stability in employment and earnings of male workers may be completely captured by employment status.
As to family background, we should point out that a higher household income increas9
es the hazard of marriage both for women who reached adulthood in the 1980’s, or in
the bubble economy, and for women who reached adulthood in the 1990’s, or the first of
the lost decades. In other words, it is a universal phenomenon that surfaced in the
1980’s, as universal marriage collapsed, but it is independent of macroeconomic conditions.
As to macroeconomic variables, significant effects, obtained in cohort-pooled samples,
now have been lost. This is not very surprising. Due to data splitting by cohort, their
variations within each cohort were significantly reduced, resulting in significant increase in standard errors of the estimated coefficients.
One interesting result is the positive sign of the macro unemployment rate in the
graduating year obtained from women in 1970’s cohort. Given the negative sign obtained from the cohort-pooled sample, this positive effect stands out as one of the characteristics of this 1970’s cohort. In addition, the women in this cohort are characterized
by a large, statistically significant, negative effect of the initial non-regular employment effect. Unlike previous cohorts, even the women belonging to the least-demanding
class in terms of reservation wage have experienced delays in marriage. On the other
hand, those succeeded in getting regular employment have married earlier. Marriage
in this cohort seems to be a form of insurance for the increased income instability coming from the deep economic recessions of the 1990’s and early 2000’s.
Concluding Remarks
In this paper, by estimating a discrete survival model to the time to marriage, we
have shown that time to marriage is affected not only by individual properties, but also
by the macroeconomic conditions of their graduating years. We have shown that job
characteristics associated with higher or stable income for male workers shortens the
time to marriage, but it’s the other way around for female workers. We have also found
that throughout the normal business cycles, a low economic growth delays marriage,
while a higher growth rate shortens the time to marriage. The unemployment rate in
graduating year works in similar ways; a high unemployment rate tends to delay marriage.
Most of these findings are consistent with the predictions of a search model for a marriage partner, except the effect of high family income on women. Unlike the previous
presumption, we have found high family income shortens the time to marriage, rather
than delay it. A high family income of women may result in higher reservation wage,
but it may also enable her to engage in more intensive search or receive side payments
from family.
Another interesting finding is on the effect of unemployment rate during the long, deep
recession of the 1990’s. In this period, contrary to the normal business cycle, we have
found that a higher unemployment rate shortens the time to marriage among the regularly employed men and women. While further research is needed to confirm this result, we can interpret it as an insurance policy against the extreme employment uncertainty during this period.
【注:関連データ】
*1.国立社会保障・人口問題研究所「第 14 回出生動向基本調査」「結婚と出産に関する
全国調査 独身者調査の結果概要」
*2.厚生労働省 「平成 23 年版働く女性の事実」「女性の配偶関係別労働力率」
【出典】
10
・岩澤美帆(2002)「近年の期間TFR変動における結婚行動および夫婦の出生行動の変
化の寄与について」『人口問題研究』58(3):15-44.
・大谷憲司(1989)「初婚確率と第 1 子出生確率の Proportional Hazards Model 分析」
『人口問題研究』第 45 巻第 2 号、pp.46-50
・太田聰一・玄田有史・近藤絢子(2007)「溶けない氷河―世代効果の展望」、『日本労働
研究雑誌』、No.569,pp.4-16.
・ 小椋正立「2020 年までの日本人人口予測」『日本経済研究』No.27, pp.77-101,
1994 年 3 月
・ 小椋正立・Decle, Robert(1992)「1970 年以降の出生率低下とその原因-県別・年齢
別データからのアプローチ」『日本経済研究』22 巻 pp.16-76.
・玄田有史(2004)『ジョブ・クリエーション』日本経済新聞社
・厚生労働省「平成 22 年度「出生に関する統計」の概況」
(H22 年 12 月 9 日発表)
・ 厚生労働省「平成 23 年版働く女性の事実」「女性の配偶関係別労働力率」
・国立社会保障・人口問題研究所「第 14 回出生動向基本調査」「結婚と出産に関する
全国調査 独身者調査の結果概要」
・国立社会保障・人口問題研究所「第 14 回出生動向基本調査」「結婚と出産に関する
全国調査 夫婦調査の結果概要」
・ 近藤絢子(2014)「不況と少子化—失業率と出生率・既婚率の都道府県パネル分析」
『経済志林』第 81 巻第 2・3・4合併号 pp.109-126
・酒井正・樋口美雄(2005) 「フリーターのその後」『日本労働研究雑誌』535 号, 29-41.
・佐藤一磨 (2008) 「学卒時の雇用情勢は初職離職後の就業行動にも影響しているのか」
KEIO/KYOTO GLOBAL COE DISCUSSION PAPER SERIES DP2008-003
・佐藤一磨(2010)「景気後退期の就業行動の変化」『日本の家計行動のダイナミズム
[ ]
経済危機下の家計行動の変容』慶應義塾大学出版
・永瀬伸子(2002)「若年層の雇用の非正規化と結婚行動」
『人口問題研究』第 58 巻
第 2 号,pp22-35
・野崎裕子(2007)「雇用不安時代における女性の高学歴化と結婚タイミング-JGSS
データによる検証-」日本版 General Social Surveys 研究論文集[6] JGSS で見
た日本人の意識と行動 JGSS Research Series No.3
・樋口美雄・阿部正浩(1999)「経済変動と女性の結婚・出産・就業のタイミング-固定
効果と変動要因の分析-」『パネルデータからみた現代女性 結婚・出産・就
業・
消費・貯蓄』東洋経済新報社、pp25-65
・両角良子(2002)「女性の出産・就業,高齢者就業:ミクロデータによる分析のサーベ
イ」
『家計のミクロ統計分析』 第 7 章 pp.163-190 統計情報研究開発センター
(Sinfonica)研究叢書
・水落正明(2006)「学卒直後の雇用状態が結婚タイミングに与える影響」『生活経済学
研究』22・23,pp167-176
Becker, G.S.(1960), “An Economic Analysis of Fertility,” in Demographic and
Economic Change in Developed Countries,Princeton University
Press, Princeton.
Becker, G.S.(1973), “A Theory of Marriage: Part I”, Journal of Political Economy, Vol 81,
No.4 (July/August 1973), pp.813-846
Blau, F., Kahn, L.M. & J. Waldfogel, “Understanding young women’s Marriage Decisions: The Role Of Labor And Marriage Market Conditions,“, Industrial and
Labor Relations Review, Vol.53, No.4 (July 2000), pp.623-647.
Lichter, D.T., McLaughlin, D.K., Kephart, G., and D.J.Landry, 1992, “Race and the Re11
treat from Marriage: A Shortage of Marriageable Men?” American Sociological
Review, Vol.57, No.6 (December 1992), pp.781-99.
Mare, R.D., and C.Winship, “Socioeconomic Change and the Decline of Marriage for
Blacks and Whites.” The Urban Underclass, Brookings Institution, pp.175-202
Ogura,S. & T.Kadoda, “ Effects of Public Policies and the Labor Market on the Fertility
of Japanese Women; Analyses of Municipal Data”, in Takayama,N. and
Werding, M. eds., Fertility and Public Policy, The MIT Press, pp.81-110, Dec
2010
Ogura,S. “What Went Wrong with the 1994 Official Population Projection of Japan?”
in Ogura.S., Tachibanaki, T. & D.A.Wise, eds., Aging Issues in the U.S. and
Japan, pp.361-402, The University of Chicago Press, Jan 2001
Schltz, T.P. , “Marital Status and Fertility in the United States: Welfare and Labor in
the Market Effects.” Journal of Human Resources, Vol.29, No.2 (Spring),
pp.639-69
White, L., “A Note on Racial Differences in the Effect of Female Economic Opportunity
on Marriage Rates.” Demography, Vol.18, No3. (August), pp. 349-54
12
ダウンロード

Paper