Who Bequeaths, Who Rules: How the Right to Bequeath Affects Intrahousehold Bargaining Li Han and Xinzheng Shi * January 2018 Abstract We examine the influence of changes in status-bequest rules on the bargaining outcomes within local-local marriages in urban China, where a reform in 1998 ended women’s monopoly on the transmission of residency permits (hukou ) to children. This reform arguably weakened the intrahousehold bargaining position of women with priv- ileged local urban hukou by improving local urban men’s divorce options in the remar- riage market. We find that the allocation of resources within pre-existing local-local urban marriages responded to this reform in favor of men at the cost of female-favored consumption and investment in children. This response was stronger in settings in which local men have better prospects of marrying migrants after divorce. Keywords: hukou ; bequest; intrahousehold bargaining; urban China. * Li Han ([email protected]) is affiliated with the Hong Kong University of Science and Technology. Xinzheng Shi (corresponding author; [email protected]) is affiliated with Tsinghua University. We gratefully acknowledge the China National Bureau of Statistics for providing the Urban Household Survey data. We thank Andrew Foster and two anonymous referees for their insightful comments. We also thank seminar participants at Lingnan University, the Hong Kong University of Science and Technology, Peking University, Renmin University, and Shanghai University of Finance and Economics for helpful comments. Xinzheng Shi acknowledges financial support from the National Natural Science Foundation of China (Project ID 71673155). All remaining errors are our own. 1
40
Embed
Who Bequeaths, Who Rules: How the Right to Bequeath A ects ...crm.sem.tsinghua.edu.cn/UploadFiles/File/201801/20180126002203… · Bequeath A ects Intrahousehold Bargaining Li Han
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Who Bequeaths, Who Rules: How the Right to
Bequeath Affects Intrahousehold Bargaining
Li Han and Xinzheng Shi ∗
January 2018
Abstract
We examine the influence of changes in status-bequest rules on the bargaining
outcomes within local-local marriages in urban China, where a reform in 1998 ended
women’s monopoly on the transmission of residency permits (hukou) to children. This
reform arguably weakened the intrahousehold bargaining position of women with priv-
ileged local urban hukou by improving local urban men’s divorce options in the remar-
riage market. We find that the allocation of resources within pre-existing local-local
urban marriages responded to this reform in favor of men at the cost of female-favored
consumption and investment in children. This response was stronger in settings in
which local men have better prospects of marrying migrants after divorce.
where Yict denotes shares of expenditure on different types of good for household i in city c
at year t; Mig densityc,2000 is the density of female migrants aged between 20 and 45 in city
c, which is computed using the 2000 population census data; Post is an indicator for the
post-reform period, which takes the value 1 for years after 1998; Xict is a vector of covariates
including couples characteristics (husbands’ and wives’ age and years of schooling), family
characteristics (the natural logarithm of total expenditure per capita, family size, and the
age structure of family members).12
Cities with different female migrant densities could have different macroeconomic cy-
cles, which might also be correlated with household expenditure patterns. To control for
these macroeconomic cycles, we include certain city-level macroeconomic variables (GDP
per capita, GDP growth rate, shares of GDP in different industries, and average wage) in
the regressions. In addition, an SOE reform starting in 1998 laid off many SOE employees.13
If more SOE employees were laid off in cities with more female migrants, then the change
in household expenditure patterns could be due to the change in the bargaining position
12The age structure of the family includes the proportion of male family members aged 0–6, 7–18, 19–60,
and above 60, and the proportion of female family members aged 0–6, 718, and 19–60. The proportion of
female family members older than 60 is omitted to avoid collinearity.13See Wu (2003) for a detailed description of the SOE reform.
16
between husband and wife because of the change in their relative economic status induced
by the SOE reform, leading to bias in our estimates. Therefore, we control for the city-level
share of SOE employees in the regression. All these city-level variables are included in Mct.
We also include city fixed effects δc and year fixed effects λt. The coefficient α1 thus
captures how household bargaining outcomes changed after the policy change in cities with
different densities of female migrants. We do not control for the post-reform dummy because
the effect has been included in the year fixed effects. We also do not control for female migrant
density because it is a city level time invariant variable and thus absorbed by the city fixed
effects. Standard errors are calculated by clustering over the city level to deal with any
heteroskadasticity problem.
In the main analysis, we focus on female-favored and male-favored consumption expendi-
tures. The UHS data contain information on two categories of women-specific consumption–
women’s clothes and cosmetics. Expenditures on women’s clothes have been used in studies
such as Lundberg, Pollak, and Wales (1997) and Browning et al.(1994). Although there is no
direct empirical evidence, women are the main consumers of cosmetics products in China.14
Because previous studies show that women tend to invest more in children than men (e.g.,
Duflo, 2003), we also examine child-related expenditures including expenditures on chil-
dren’s clothes and children’s education. Male-favored consumption includes men’s clothes,
cigarettes, and alcohol. As expenditures on women’s clothes, expenditures on men’s clothes
have also been used by Lundberg, Pollak, and Wales (1997) and Browning et al.(1994).
Wang (2014) shows men consume much more cigarettes and alcohol than women and uses
these two variables as men-specific consumption.15 Data for 1997 (pre-reform) and 1999
(post-reform) are used in estimating Equation (1).
The assumption underlying Equation (1) is that female migrant density in the year 2000,
Mig densityc,2000, should not be correlated with the error term. This assumption would be
14A China Daily article provides some description. See http :
//www.chinadaily.com.cn/english/doc/2005− 06/07/content449333.htm15Headgear and footwear are potential gender-specific expenditures, but the UHS do not collect the infor-
mation by gender.
17
violated if the policy change induced a change in migration patterns in different cities. To
address this concern, we use the city-level density of female migrants aged 20-45 years in
1990 as an IV for the density in 2000.
Concerns remain as to whether the IV is correlated with the error term. One possible
channel through which the IV might correlate with the error term is that expenditure in
cities with a high migrant density in 1990 could have followed a different time trend from
cities with low migrant densities had there been no policy change. Our estimates would
capture the difference in the two trends if this were the case. To check whether the parallel
trends assumption is valid, we conduct a placebo test using data from the three years before
the policy change, i.e., 1995, 1996, and 1997.
Another possible channel through which the IV might correlate with the error term is
if migrant density in 1990 was correlated with the macroeconomic status of different cities,
which could affect the patterns of household expenditure. Including particular macroeco-
nomic variables in the regression as per our approach can mitigate this concern. In addition,
we conduct another placebo test by investigating the effect of the policy change on other
household expenditure items, i.e., gender-neutral goods. If the IV estimates of Equation (1)
are driven by unobserved macroeconomic shocks, we should observe a similar consumption
pattern for gender-neutral goods. Any difference in patterns would suggest that the changes
are not likely to be driven by macroeconomic shocks.
One could still be concerned that macroeconomic shocks in different cities might be
gender specific. For example, if local women in cities with higher proportions of female
migrants in 1990 experienced slower growth in work opportunities between 1997 and 1999,
then our estimates could be contaminated by the worsened bargaining position of women due
to their relatively inferior economic status within households. We test whether this concern
materializes by investigating whether the policy affects the probability of being employed
and the probability of participating in the labor force for women and men, respectively.
Absence of a significant policy change effect would suggest that this concern may not be a
problem.
A further concern arises with the use of multiple cross-sectional data. Because our sample
18
does not constitute a panel, theoretically the composition of households in the sample may
change from year to year. In particular, if more local–local households in cities with more
female migrants dissolved after the policy change, our estimated effects would suffer from
selection bias. Moreover, as pointed out in Lafortune et al. (2017), newly formed households
after the policy change could respond differently to the policy, also leading to bias in our
estimates. To address this concern, we conduct another robustness check by restricting
our sample to those couples who married before 1998. As the UHS data do not contain
information on the year of marriage, we restrict the 1999 sample to households with children
older than 2 years, as the couples in these households are most likely to have married before
1998.
To explore and ensure the validity of our results, we also conduct other robustness checks
such as reduced form regressions and permutation tests.
7 Empirical Results
7.1 Main Results
Table 2 presents the OLS results for Equation (1). Columns (1)–(4) show estimation results
for women-favored expenditures, including women’s clothes, cosmetics, children’s clothes,
and children’s education, respectively; while columns (5)–(7) show estimation results for
men-favored expenditures, including men’s clothes, cigarettes, and alcohol, respectively. The
coefficients on the interaction of the post-reform dummy and female migrant density are
negative and statistically significant at the 5% level for cosmetics, 10% level for children’s
clothes, and 1% level for children’s education (-0.005, -0.005 and -0.053, respectively). Al-
though the coefficients of the interaction Post ×Mig density are statistically insignificant
for expenditures on men-favored expenditures, they are all positive.
However, as discussed in Section 6, OLS estimates might be inconsistent if migration
had already responded to the policy change before 2000. To address this issue, we use the
density of female migrants aged 20–45 years in 1990 as an IV for this density in 2000.
19
The density of female migrants in 1990 is a strong predictor of the density in 2000. The
first stage result is presented in column (1) of Table 3. The coefficient of the interaction
of the post-reform dummy and female migrant density in 1990 is 7.47 and it is statistically
significant at the 1% level. This result means that the density of female migrants in 2000
is 7.47 percentage points higher in cities where this density was 1 percentage point higher
in 1990. The F-value shown in the last row is 11.65, which is higher than the conventional
lower bound for a valid IV.
The second-stage results of the IV estimation are shown in columns (2)–(8) of Table 3.
These IV results are even stronger than the OLS results. The coefficients of the interac-
tion of the post-reform dummy and female migrant density are statistically significant in all
columns. The estimated coefficients for women-favored expenditures are all negative, equal
to -0.039 (women’s clothes in column (2)), -0.008 (cosmetics in column (3)), -0.013 (children’s
clothes in column (4)), and -0.062 (children’s education in column (5)). The coefficients for
men-favored expenditures are positive, equal to 0.043 (men’s clothes in column (6)), 0.065
(cigarettes in column (7)), and 0.016 (alcohol in column (8)). Compared with the aver-
age share of expenditure on women’s clothes (0.037, see Table 1), a one-standard-deviation
increase (approximately 0.08) in female migrant density leads to a 8.4% (the product of
0.08 and 0.039 divided by 0.037) decrease in expenditure on women’s clothes. A similar
calculation shows that a one-standard-deviation increase in female migrant density leads to
10.7%, 17.3%, and 9.9% decreases in expenditure on cosmetics, children’s clothes, and chil-
dren’s education, respectively; however, the same change in female migrant density leads to
11.9%, 10.8%, and 9.1% increases in expenditures on men’s clothes, cigarettes, and alcohol,
respectively.
Taken together, these results indicate that the 1998 change in the hukou bequest rule
resulted in a decrease in women-favored expenditures but an increase in men-favored ex-
penditures of local–local couples. Indeed, beyond women-specific consumption, investment
in children’s education is also negatively affected, thus the adverse effect may also carry
through to the next generation.16
16One caveat we need to acknowledge is that for local-migrant households which are not the focus of our
20
7.2 Heterogeneous Effects
If the estimated policy effect arises through the remarriage market channel, we would expect
the bargaining power of husbands with better education to increase more because they have
greater advantages in remarriage markets. In this section, we investigate the heterogeneous
effects of the policy change in terms of husbands’ schooling years.
In Equation (1) we include interactions between Post×Mig densityc,2000 and husbands
schooling years (together with all double interaction terms). We use interactions of female
migrant density in 1990 with relevant variables as IVs for the interactions of female migrant
density in 2000 with the same variables. Table 4 reports the results. Therein, the coefficient
of the triple interaction is negative and statistically significant for expenditures on women’s
clothes. Although coefficients for other outcome variables are not significant, they are nega-
tive for expenditures on cosmetics and children’s education, and they are positive for men’s
clothes and cigarettes. These results suggest that the effects of the 1998 change in the hukou
bequest rule are stronger for local men who are better educated and therefore have more
outside options.
7.3 Robustness Checks
We conduct several robustness checks to explore the validity of our main results.
Testing for Pre-existing Time Trends. One threat to the validity of the IV estimation
is that the IV could be correlated with pre-existing time trends that drive the estimation
results. To address this concern, we conduct a placebo test by estimating Equation (1) us-
ing data from 1995-1997. In estimating this equation, we replace Post ×Mig densityc,2000
with interactions of years 1996 and 1997 dummies with female migrant density in 2000,
Dummy1996×Mig densityc,2000 and Dummy1997×Mig densityc,2000. We use Dummy1996×
Mig densityc,1990 and Dummy1997×Mig densityc,1990 as IVs. If pre-existing trends are cor-
related with the IV, we would expect the interaction terms to be statistically significant. The
paper, the status of children would be better because of the policy change.
21
estimation results are presented in Appendix Table 2. All the coefficients of the interaction
terms are statistically insignificant. These results thus give us more confidence that the IV
estimates are not driven by pre-existing time trends.
Effects on Gender-neutral Expenditures. Although we have included macroeconomic vari-
ables in the regressions, one may still query whether unobserved macroeconomic variables
are correlated with the IV, leading to biases in our estimates. To address this, we estimate
effects of the policy change on food items, a gender-neutral expenditure. If the main results
shown in Table 3 are driven by unobserved macroeconomic variables, we should observe
the same pattern for gender-neutral goods. Appendix Table 3 presents effects of the policy
change on gender-neutral expenditure. Column (1) shows the share of expenditure on food.
Columns (2)–(4) report expenditures on the three most commonly consumed food items,
namely rice, pork, and vegetables. In all four columns, the coefficients on the interaction
term of the post-reform dummy and female migrant density in the year 2000 are not statis-
tically significant, suggesting that unobservable macroeconomic shocks did not lead to the
results shown in Table 3.
Effects on Labor Market Status. One could still be concerned that macroeconomic shocks
in different cities might be gender specific. For example, local women in cities with a higher
ratio of female migrants could experience slower growth in work opportunities from 1997
to 1999 due to the more severe competition from female migrants. Therefore, the negative
effects of the policy change could be due to women’s worsened bargaining position within
households because of their inferior economic status. To address this concern, we investi-
gate effects of the policy change on the probability of being employed and the probability
of participating in the labor force for women and men, respectively. We define a person as
participating in the labor force if he or she is employed or without a job but searching for
employment. Appendix Table 4 presents the results. Therein, none of the coefficients of the
interaction term Post×Mig densityc,2000 using Post×Mig densityc,1990 as an IV are signif-
icant for either outcome variable, whether the female or male sample is used. These findings
suggest that our results in the main analysis are not driven by the worsening economic status
of women in cities with higher proportions of female migrants.
22
Effects of Possible Dissolution or the Formation of Families. As the data used in our
empirical analysis do not constitute a panel, another concern is that changes in the com-
position of households may confound our results, especially when the policy change affects
the marriage matching pattern.17 To rule out this concern, we conduct an analysis using
households formed prior to the policy change. Because the UHS data do not include infor-
mation on years of marriage, we restrict households in the 1999 sample to those with children
older than 2 years. This group of households is most likely to have been formed before the
policy change. The results estimated using the 1997 sample and the sub-sample in 1999 are
reported in Appendix Table 5. The coefficients of the interaction of the post-reform dummy
and female migrant density are qualitatively similar to the main results.18
Reduced Form Regressions. To further confirm our results, we follow suggestions pro-
posed by Angrist and Pischke (2009) and conduct reduced form regressions. Replacing
Mig densityc,2000 with Mig densityc,1990, we estimate Equation (1) using OLS. Results are
shown in Appendix Table 8. All coefficients of Post × Mig densityc,1990 are statistically
significant. They are negative for women-favored expenditures but positive for men-favored
expenditures. The reduced form regression results thus re-confirm our main findings.
Permutation Tests. To address the concern that our main results could be driven by ran-
dom factors, we conduct permutation tests.19 We randomly assign female migrant densities
17A related issue is that it could be easier for migrant women who married local men to acquire local hukou
after the policy change, and then these couples may be included in our post-reform sample. These migrant
women could have weaker bargaining power within households because they were not born locally even if
they have local hukou. If there are more such cases in cities with higher proportions of female migrants, then
our estimates are biased.18We also estimate Equation (1) using the 1997 sample and newly formed households in the 1999 sample
(i.e., households without any children older than 2 years). The results, shown in Appendix Table 6, are much
weaker. We then estimate long-run effects by including two more years of data (2000 and 2001); the results,
shown in Appendix Table 7, suggest that although policy change effects still exist three years after the policy
was introduced (particularly for children’s clothes), they become weaker. These findings are consistent with
Lafortune et al. (2017) who suggest that newly formed households could respond to the policy before union
which offsets the effects of the policy change.19For similar exercises, see, e.g., Chetty, Looney, and Kroft, 2009; La Ferrara, Chong, and Duryea, 2012;
23
to cities and estimate impacts of the policy change on all outcome variables. We repeat this
exercise 1000 times. Appendix Figure 2 presents histograms of all 1000 p-values for each
outcome variable (panel A for women-favored expenditures and panel B for men-favored
expenditures). The majority of the p-values are larger than 10% and thus these permutation
tests provide additional evidence supporting the validity of our main results.
7.4 Extension
We have shown in Section 7.1 that, on average, the policy change reduces household expen-
ditures on children. Such an effect could be different for boys and girls. Anticipating that
girls’ status will deteriorate in the marriage market in the future, parents might compensate
girls by spending more on them now, leading to weaker effects of the policy change on girls,
compared to boys. Furthermore, substitution effects existing among different categories of
consumption will lead to changes in other types of expenditures. To test this hypothesis,
we investigate the heterogeneous effects of the policy change in terms of children’s gender.
Because expenditures are measured at the household level, we restrict the sample to house-
holds having only one child. Because parents are more likely to make decisions for non-adult
children, we further restrict the sample to households with one child aged 0-18 years old.
The estimation results are shown in Table 5. The coefficient of the triple interaction
for expenditures on children’s education is positive and statistically significant at the 5%
level. It shows that the policy change has weaker effects on households’ expenditures on
girls’ education compared to boys. This is consistent with our hypothesis that parents
compensate girls by spending more on them in anticipation of their worsening status in
future marriages. We can also see from Table 5 that the coefficient of the triple interaction
for women’s clothes is negative and statistically significant, meaning that expenditures on
women’s clothes decrease more for households having girls. This suggests that expenditures
on women’s clothes might be switched to expenditures on girls’ education.20
and Cai, et al., 2016.20We conduct the same exercise using the sample of households having one only child older than 18 years.
Results in Appendix Table 9 show that no coefficients of the triple interaction are significant, which could
24
8 Mechanisms
If, as our theoretical framework suggests, the policy effects on intrahousehold bargaining
mainly arise through improving local urban men’s divorce options, two key elements must
have played a role. Firstly, divorce is a credible threat, especially in cities with more female
migrants. Secondly, people care about the hukou status of their prospective children. In
addition, our empirical strategy is based on the assumption that households can perceive
the effect of female migrant density on local urban men’s remarrying prospects. We construct
the following tests to provide evidence for these conditions.
Firstly, Table 6 shows the correlation between female migrant densities and the increase
in divorce rates in cities.21 We observe that the coefficients on female migrant density in
1990 (column 1), female migrant density in 2000 (column 2), and female migrant density in
2000 (using female migrant density in 1990 as an IV, column 3) are all positive. Thus, cities
with higher female migrant densities tended to exhibit faster growth in divorce rates from
1990 to 2000, providing suggestive evidence that husbands’ threats to divorce their wives are
more credible in cities with higher densities of female migrants.
Secondly, we investigate whether the policy change has stronger effects on younger couples
because they are more likely to have children after they divorce and thus they may care
more about prospective children. For this purpose, we use the sample including households
where both couples are older than or equal to 40 and households where both couples are
younger than 40. The results are shown in Table 7. The coefficients of the triple interaction
of the post-reform dummy, female migrant density in 2000, and the dummy for younger
households are significant for expenditures on cosmetics (negative, column 2), children’s
education (negative, column 4), and cigarettes (positive, column 6). This provides suggestive
evidence for the hypothesis that the policy change has stronger effects on couples who are
more likely to have children after divorce.
be because parents are less likely to make decisions for their adult children.21Divorce rate is defined as the ratio of divorced individuals to all individuals who ever married (including
currently married, widowed and divorced).
25
Thirdly, we investigate whether effects of the policy change are stronger for households
where husbands work in industries with relatively high female migrant densities. These
husbands have more chances to meet migrant women and therefore have better outside
options. We use the 1990 population census to calculate female migrant density for each
2-digit industry,22 and then we define industries with female migrant densities above the
mean as industries with a high migrant density (i.e., High Mig Density industry).23 Table
8 presents the regression results. The coefficients of the triple interaction are significant for
children’s clothes (negative, column 3) and cigarettes (positive, column 6). This shows that
the effects are indeed stronger for households where husbands work in industries with a high
density of female migrants, suggesting that meeting migrant women via work is an important
channel for husbands to get a sense of the density of female migrants in their locality.
9 Conclusion
In the context of urban China, we examine how the bargaining outcomes of couples with the
same valuable local hukou responded to a policy change that granted men the same rights
as women to pass on their hukou status to their children. This policy change results in a
decrease in female-favored consumption and an increase in male-favored consumption. This
finding suggests a worsening relative position of local urban females in local-local marriages.
Our finding illustrates that the regulation on who has the right to bequeath status to the next
generation matters for intrahousehold bargaining of couples with the same status through
changing their outside options in the remarriage market. This conclusion can also transfer
to the more general case of bequeathing property to the next generation. In such a case, one
possible channel through which the more resourceful partner in a marriage enjoys greater
bargaining power is that his/her bequeath potential makes him/her more attractive in the
remarriage market.
22UHS data only provides industry information at the 2-digit level.23Households where husbands are not working are assigned zero because their chances of meeting migrant
women are lower.
26
Although the hukou system is unique to China, our results may be useful for under-
standing similar changes in other contexts. Many societies have changed from adhering to a
patrilineal status bequeathing rule to an ambilineal bequeathing rule, which grants females
equal rights to pass on their status. This change is often considered as empowering females.
Although the alleged goal of this type of female empowerment is usually to help increase
the position of females in inter-status marriages, our results suggest that changes targeting
the relative minority in the marriage market can have broad impacts on the majority of
marriages in which both husband and wife have the same status. In particular, our findings
suggest that this type of female empowerment increases the bargaining power of females
who have high status regardless of marriage type. However, the mechanism that we have
examined implies that this type of empowerment may well result in a “within-gender war”
instead of a “gender war” in the sense that not all women can gain. Low-status females may
be unable to benefit from this type of empowerment because the relative demand for them
may be lower as high-status women become more sought after.
There are several limitations to our study. Firstly, we are unable to track divorced couples
because of data limitations. Family dissolution itself is an outcome of within-household
bargaining which is worth exploring. Nevertheless, given the fairly low divorce rates in the
time period we cover, our study captures the changes in the majority of marriages that
sustain through time. Secondly, we are unable to explore the effect of changes in status
bequest rights on marriages involving at least one low-status migrant because the UHS data
used herein do not include information on those with non-local urban hukou. Analysis of
these marriages would help us understand the general equilibrium in the marriage market
and thus future research in this direction is merited.
27
References
[1] Anderson, Siwan. 2003. “Why dowry payments declined with modernization in Europe
but are rising in India.” Journal of Political Economy, vol. 111, 269-310.
Robust standard errors in parentheses are calculated by clustering over city level. * significant at 10%; ** significant at 5%; *** significant at 1%.
(1) Year dummies and city dummies are controlled in all columns.
(2) Household demographic structure includes family size, ratio of male family members aged 0-6, 7-18, 19-60, and above 60, ratio of female family members
aged 0-6, 7-18, and 19-60. Ratio of female family members aged above 60 is omitted to avoid multi-collinearity. Macroeconomic variables include log GDP
per capita, GDP growth rate, ratio of GDP in primary industry, ratio of GDP in secondary industry, log average wage in city level, and ratio of SOE employees.
36
Table 4 Heterogeneous Effects in Terms of Husband's Schooling Years
Robust standard errors in parentheses are calculated by clustering over city level. * significant at 10%; ** significant at 5%; *** significant at 1%.
(1) Year dummies and city dummies are controlled in all columns.
(2) Household demographic structure includes family size, ratio of male family members aged 0-6, 7-18, 19-60, and above 60, ratio of female family
members aged 0-6, 7-18, and 19-60. Ratio of female family members aged above 60 is omitted to avoid multi-collinearity. Macroeconomic variables include
log GDP per capita, GDP growth rate, ratio of GDP in primary industry, ratio of GDP in secondary industry, log average wage in city level, and ratio of
SOE employees.
(3) The sample used in this table includes households with only one 0-18 years old child.
38
Table 6 Impact of the Policy on City Divorce Rates
(1) (2) (3)
Δ Divorce Rate between 1990 and
2000
Mig_density1990 0.514*
(0.294)
Mig_density2000 0.008
(0.009)
Mig_density2000 (Mig_density1990 as an IV) 0.122
(0.073)
Constant 0.004*** 0.005*** 0.008*
(0.001) (0.001) (0.004)
Observations 47 47 47
R-squared 0.056 0.011
Robust standard errors are in parentheses. * significant at 10%; ** significant at 5%; *** significant at 1%.
(1) Divorce rate is calculated as the ratio of divorced individuals over those who ever married, using data
from 1990 and 2000 population census.
39
Table 7 Heterogeneous Effects in Terms of the Ages of Spouses