Status Inheritance Rules and Intrahousehold Bargaining

Transcription

1 Status Inheritance Rules and Intrahousehold Bargaining Li Han and Xinzheng Shi May, 2015 Abstract This paper studies how changes in the status inheritance rules a ect intrahousehold bargaining outcomes. This question is analyzed in the context of a reform occurring in China that granted men the same rights as women in passing on residency permits (hukou) to their children regardless of their spouse hukou. This reform decreased the relative position of women with privileged urban hukou status in the marriage market. We find that this change decreased the female-favored consumption in the urban-urban marriages that were formed before the reform. This finding shows that the allocation of household resources responds to changes in the relative position in the marriage market. Li Han is an assistant professor in the Division of Social Science, Hong Kong University of Science and Technology. Xinzheng Shi is an associate professor in the School of Economics and Management, Tsinghua University. We thank seminar participants in Lingnan University and Hong Kong University of Science and Technology for helpful comments. All remaining errors are ours. Correspondence author: Xinzheng Shi, shixzh@sem.tsinghua.edu.cn. 1

2 1 Introduction Historically many societies do not grant women the same right as men to confer social or legal status to their children. For example, the inheritance of caste in India once followed the patrilineal inheritance rule; the nationality laws of the majority of the states, including many European countries, did not allow mothers to pass the nationality to children until the late 1970s. Our time has seen a global trend of liberalizing the status inheritance rules as part of gender equality movement. A case in point is the launch of a global campaign to revoke discriminatory laws that prohibit women from changing their nationalities, or conferring their nationalities on their children in It has been generally believed that such changes help improve gender equality in the inter-status marriages. However, few notice that changes in the status inheritance rules could have far-reaching impacts on bargaining power within the couples of the same status by a ecting their relative position in the marriage market. As children are among the most important marital outputs, mate choices should be a ected by the status inheritance rules of forward-looking people who are concerned about the status of their prospective children. Granting equality to the two genders with regard to the status of children likely leads to a rise in the relative demand in the marriage market for the group who have high status but previously were not allowed to pass their status to children, which improves their relative position in the marriage market. Provided that divorce is a credible threat, we would expect that the bargaining position becomes more favorable to this group even if their spouse are of the same status. In contrast, the group who have low status cannot benefit from the acquisition of the right to pass the status, as the relative demand in the marriage market for them decreases. To examine how the changes in the status inheritance rules a ect intrahousehold bargaining, we set our empirical study in the context of China, where a clearly defined inheritance rule exists for a very important legal status hukou. We study how a change in the inheritance rule for hukou a ects the intrahousehold resource allocation in China. For ordinary Chinese people, hereditary hukou (household registration or the class system of residency 1 The Guardian, June 20, Women s nationality is focus of new campaign for gender equality. 2

3 permits) is the most important determinant of socioeconomic status. The e ect is similar to the e ect of caste for Indians. Access to public services such as education and health care and employment opportunities in a locality are linked to the local hukou that a person inherits from his or her parents. Changing hukou through marriage or relocation is tightly controlled (Wu and Treiman, 2004). Moreover, both the rural-urban divide and regional inequality are enormous, which makes urban hukou holders in economically advanced cities e ectively high-social-status individuals. Without local urban hukou, rural migrants who work in cities are considered inferior to local urban people in the urban marriage market. Di erent from the caste system which is largely enforced by social norms, the hukou system is enforced by the government and is subject to exogenous or discrete policy changes. In particular, a policy change in 1998 a ects the status inheritance rule in a significant way and enables us to study the causal link between status inheritance rules and intrahousehold bargaining outcomes. The hukou inheritance had followed the matrilineal rule, i.e., children were granted their mother s hukou, until the law was amended in August 1998 to grant men the same rights as women to transmit hukou to their children. E ectively hukou inheritance has followed the maximum rule ever since, i.e., the child can inherit the higher status of two parents. Han, Li and Zhao (2014) find that this gender-asymmetric policy change caused a striking increase in the likelihood of local urban men marrying migrant women, whereas the likelihood of local urban women marrying migrant men was little a ected in a short run. The changes in the matching pattern is associated with the changes in the relative demand for the two genders with di erent hukou. As the relative demand for local urban men increased while the relative demand for local urban women decreases, the relative position of urban women in the marriage market worsened. The deteriorating position of urban women is likely to lead to a decrease in the bargaining power of women in the urban-urban marriages. We test this hypothesis empirically. Our benchmark identification is a di erence-in-di erence (DID) strategy where the first di erence exploits the 1998 policy change that dramatically increased the substitutability between rural women and urban women in the marriage market. For the second di erence, we exploit the fact that the density of female migrants who are aged years di er by 3

4 cities. The policy change would have an immediate and strong e ect on regions where more potential migrant brides are available. To address the potential endogenous migration driven by the policy change, we use the migrant-to-local ratio in 1990 as an instrument variable (IV) for the post-reform migrant-to-local ratio in Our data mainly draw upon the Urban Household Survey (UHS). The survey collects not only individual level information, but also detailed information on household expenditures. We focus on expenditures on the female-favored goods such as women s, cosmetics, children s, washing machine, and jewelry. Male-favored goods, such as men s, cigarette, alcohol, TV, and refrigerator, are investigated as robustness checks. We use data from the 1990 and 2000 population census to construct ratio of female migrants aged 20 to 45 years for each city. Indeed, the policy change reduced women s bargaining power within households. We find that for cities having one standard deviation higher female migrant ratio, the policy change significantly reduced the expenditures on women by 2.44%, cosmetics by 7.3%, children s by 7.3%, and washing machine by 11%. Several robustness checks show that the estimated results are not driven by other confounding factors such as di erent time trends, the dissolution or formation of families, or macroeconomic cycles. A necessary assumption for the status inheritance rule to matter for the intrahouhold bargaining outcomes is that people care about the status of their prospective children. We thus test this channel from three perspectives. Firstly, we investigate whether the e ects are weaker for households having more children. If the couples have more children, the men s marginal utility of having another child the men will be lower. Therefore, the benefit of husbands divorcing their current wives and having a child with migrant women would be lower. Secondly, we investigate whether the e ects are weaker for older couples. The older the couples are, the less possible is it for them to have another child even if they get divorced. Thirdly, we test whether the policy impact is weaker for couples having a son. Son preferences in China remain strong. Previous studies show that the bargaining power of women who have a son is stronger in the household compared to those who do not (e.g., Li and Wu, 2011). Moreover, given that it is costly to raise a child, one s incentive of having 4

5 another child would be lower if s/he already has a son. Thus the policy impact would be weaker for the couples who have a son. Indeed, we find weaker e ects of the policy change on female-favored expenditures for households having more children and those having a son. We also find that the e ects are stronger for young couples. All the results suggest that the policy impact arises from people s concern about prospective children s status. Our findings illustrate that granting men the equal right as women to pass their status to children increases high-status men bargaining power in the household even if the spouses have high status. Since the global trend of liberalizing status inheritance rules is to grant women a similar right to confer their status to next generation, our finding suggests that it would improve women s bargaining position. Our work makes two contributions. First, we show that granting equal rights to men and women with regard to children s status have a broad impact even on couples who are not directly a ected by this change. Second, our results provide more support for the existing literature on transferable utility models of marriage markets that intrahousehold resource allocation responds to the changes in the conditions of marriage markets (e.g., Becker, 1973, 1974; Browning et al., 1994). The remainder of this paper is organized as follows. Section 2 review the related literatures. Section 3 describes the data and Section 4 introduces the empirical strategy. The main results and robustness checks are reported in Section 5. Section 6 tests for the channels through which the impacts arise. Section 7 concludes the paper. 2 Related Literature This paper contributes to the literature on the distribution of bargaining power within household. The literature that considers the determinants of bargaining weights within couples largely falls into two broad categories. The first category focuses on the gender di erences in economic values such as income or asset ownership. For instance, using a structural model approach, Browning et al. (1994) find that the allocations of expenditures within couples are significantly a ected by their relative incomes, ages and the level of lifetime wealth. Duflo (2003) finds that pensions received by women had a significantly positive impact on the 5

6 anthropometric status (weight for height and height for age) of children especially girls in South Africa whereas no similar e ects were found for pensions received by men. Anderson and Eswaran (2009) find that the relative contribution in earned income is more important in empowering women in Bangladesh; Luke and Munshi (2011) find similar results in India. Wang (2013) finds that transferring the property rights of the house to men increased household expenditures on some goods favored by men and women s time spent on chores, whereas transferring property rights to women decreased household expenditures of some male-favored goods. The second category of this literature focuses on the marriage market conditions. This intuition can be traced back to Becker (1991), who emphasized that the relative supply of males and females in the marriage market is an important determinant of intrahousehold utility allocation. Browning and Chiappori (1998) define distribution factor as a variable that can a ect the intrahousehold decision process without influencing individual preferences or the joint consumption set, for example, the sex ratio and divorce legislation. Chiappori, Fortin and Lacroix (2002) theoretically investigate the e ect of the distribution factors on the within-marriage transfers and estimate the e ect using a structural model. Following this direction, Rangel (2006) finds that the extension of the alimony rights in Brazil reduced the time spent by the women of the cohabiting couples on housework and reduced their labor supply as well. However, the probability of their daughters to be enrolled increased. Edlund, Liu and Liu (2013) find that the importing foreign brides to Taiwan increased domestic brides fertility. Sun and Zhao (2013) find that the lower divorce costs due to China s new divorce law in 2001 led to higher divorce rate and fewer sex-selective abortions. Our study falls into the second category. Although explicit or implicit status inheritance rules regulate the marriage market in most societies, the impact of these rules on the marriage market and intrahousehold distribution of power is little studied. The only exception is the paper by Han, Li and Zhao (2014) which demonstrates that the shift from matrilineal status inheritance rule to the maximum inheritance rule has increased the number of interstatus marriages and worsened the relative position of high-status women in the marriage market. It remains unknown whether and how this change in the status inheritance rule 6

7 has a ected the intrahousehold bargaining outcomes of the couples. To the best of our knowledge, our research is the first study which emphasizes the role of the status inheritance rule in intrahousehold bargaining. 3 Data Our main data draw on the UHS. The UHS was conducted by the National Bureau of Statistics (NBS) in China. The UHS covers all provinces in China and uses a probabilistic sampling and stratified multistage method to select households. It is a rotating panel in which one-third of the sample is replaced each year, and the entire sample is changed every three years. Therefore, the data are essentially repeated cross sections. We have access to data collected in forty-nine cities from nine Chinese provinces with a variant of geographical locations and economic conditions, including Beijing, Liaoning, Zhejiang, Anhui, Hubei, Guangdong, Sichuan, Shaanxi, and Gansu. The mean values and the trends of the most important variables are comparable between our sample and the national sample. The UHS not only contains demographic and income information for every member of the family, but also collects detailed information of household expenditures. Unfortunately, it has no information on assets. Our main analysis utilizes data of years 1997 and Data from years 1996, 2000, and 2001 are used in various robustness checks. We did not use the 1998 data are dropped because the hukou reform took place in August We could not tell whether the information in the 1998 data was collected before or after the reform. For our research purpose, we restrict our sample to households having information on both husband and wife. In total, 25,828 households are kept in the sample. We also compile a city-level data set containing the information on the density of migrants and economic conditions. Our main measure of migrant density is the share of female migrants aged between 20 and 45 years old in the females of the same age cohorts in each city. This measure is constructed for years 1990 and 2000 using a 1% sample of the 1990 population census and a 0.095% sample of the 2000 population census respectively. Both 7

8 waves of population census ask about the location of hukou in the census year and the province of residence 5 years before. A migrant is defined as a person whose hukou is not in the local residence in the census year and was not living in the local province five years before. As no other reliable data are available on the number of migrants in each city in each year, the population census is the most reliable source to compute the density of migrants. However, one disadvantage of using the census data is that census is conducted every 10 years. So we only have the information on the migrant density for these two census years. We also obtain various city level macroeconomic variables including GDP per capita, GDP growth rate, share of GDP in primary, secondary, and tertiary industries respectively, and average wages, from the Chinese City Statistical Yearbook ( excluding 1998). Table 1 presents the summary statistics for the variables in our analysis. All the monetary values are adjusted using provincial consumer price indices. Panel A reports the means and standard deviations for household level variables. The average family size is 3.13 and the average number of children is 0.98, which suggests that most households in our sample are core families. Approximately 48% of families have sons. The average expenditure per capita is 6,070 yuan. Because for most goods it is di cult to tell which gender is the main consumer, we focus on goods that are most likely to be gender specific. Five out of all the categories of expenditures turn out to be female-favored goods, including women, cosmetics, children, washing machine and jewelry. The share of expenditures on women is 0.036, and the shares of expenditures on cosmetics, children, and washing machine are all On average, households spend 0.2% of their total expenditures on jewelry. Male-favored goods include men, cigarettes, and alcohol. The share of expenditures on men is The shares of expenditures on cigarettes and alcohol are respectively and We also examine other electronics such as TV set and refrigerator. The shares of expenditures on these two electronic devices are and respectively. It is worth noting that, given the data limitations, we only examine a small proportion of the total expenditures. The total share of expenditures on the five female-favored goods in our analysis is only while the share of expenditures on male-favored goods is Panel B of Table 1 presents the summary statistics of city level variables. The number 8

9 of migrants has followed an increasing trend since the early 1990s. The average share of migrant women in females aged years is in 2000 while the share is only in Although the density of migrants in 1990 is low, there is substantial variation. The standard deviation of this density is 0.012, which is four times of the mean value. The average GDP per capita is 11,867 yuan, and the average annual growth rate of GDP is 11%. We can also see that on average the primary industry contributes to 18% of GDP, in the secondary industry 45%, and the tertiary industry is 37%. The city level average wage is 8,625 yuan. 4 Empirical Strategy Our benchmark empirical model is the di erence-in-di erences type model. Our DID model employs both the 1998 policy change and the di erences in the density of female migrants aged across cities. The density of female migrants aged is used as a proxy for the intensity of the policy shock to local marriage markets. The rationale behind this proxy is that the 1998 policy change is expected to have bigger e ects on the local marriage market in cities where more migrant women are available. Thus we examine whether the change in intrahousehold bargaining outcomes after the policy change di er across cities with di erent levels of migrant density. In the benchmark specification, we use the density of female migrant aged to proxy the density as of the policy change. That is, we separately estimate the following equation for di erent types of household expenditures. Y ict = Mig density c,2000 Post+ X ict 1 + M ct 2 + t + c + ict (1) where Y ict stands for the expenditure on di erent types of goods for household i in city c at year t; Mig density c,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 of 1 for years after 1998; X ict is a vector of covariates including the couple s characteristics (such as husbands and wives age and years of schooling), family characteristics (such as logarithm form of total expenditures 9

10 per capita, family size and the age structure of family members). M ct is a vector of city level macroeconomic variables (such as GDP per capita, GDP growth rate, shares of GDP in di erent industries, and average wage). They are used to control for di erent macroeconomic shocks in di erent cities which might be correlated with household expenditures. We also include city fixed e ects c and year fixed e ects t. Therefore, the coe cient 1 captures how the household bargaining outcomes changed after the policy change in cities with di erent densities of female migrants. Standard errors are calculated by clustering over city-year level. In the main analysis, we focus on expenditures on female-favored goods, i.e. the shares of expenditures on women s, cosmetics, children s, washing machine, and jewelry. The data for years 1997 (pre-reform) and 1999 (post-reform) are used in estimating Equation (1). The assumption underlying Equation (1) is that the density of female migrants in 2000, Mig density c,2000 should not be correlated with the error term. This assumption would be violated if the policy change induced the change in migration pattern in di erent cities from 1999 to To address this concern, we use the city level density of female migrants aged years in 1990 to instrument for the density in The concern still remains if the instrument variable (IV) is correlated with the error term. One possible channel for the IV to correlate with the error term is that cities with high levels of migrant density in 1990 could have followed di erent time trends from cities with low levels of migrant density had there been no policy change. Our estimates would capture the di erence in the two trends if this were the case. To check whether the parallel trends assumption is valid, we conduct a placebo test by using two years of data before the policy change, i.e and 1997, to estimate Equation (1). If 1 is statistically insignificant in the placebo test, we should have more confidence in the parallel trend assumption. Another possible channel for the IV to correlate with the error term is that the migrant density in 1990 is correlated with macroeconomic status in di erent cities which could a ect the patterns of household expenditure. Including some macroeconomic level variables in the regression can alleviate this problem. In addition, we conduct another placebo test by investigating the e ect of the policy change on other household expenditure items, such as 10

11 male-favored goods. If the IV estimates of Equation (1) are driven by the unobserved macroeconomic shocks, we should see the consumption pattern of female-favored goods is similar to that of male-favored goods or gender-neutral goods. The di erence in those patterns would suggest that the changes are not likely to be driven by macroeconomic shocks. One more concern arises with the use of the multiple cross-sectional data. Because our samples are not a panel data set, theoretically the composition of households in the sample may change from year to year. In particular, if the patterns marriage matching responds to the policy change, the possible change in the composition of the household would confound our estimated impact on the intrahousehold bargaining outcomes. To address this concern, we conduct another robustness check by restricting our sample to those couples who formed marriages before Since the UHS data do not have information on the year of marriage, we restrict the sample to the households having children older than 2 years so that the couples who were wedded after 1998 are excluded. 5 Empirical Results 5.1 Main results Table 2 presents the OLS results for Equation (1). Columns (1)-(5) show the estimation results for the shares of expenditures on women, cosmetics, children, washing machines and jewelry respectively. The coe cients on the interaction of the post dummy and the female migrant density are negative and statistically significant at the 1% level for both women and cosmetics, and it is negative and significant at the 10% level for washing machines. Although the coe cients of the interaction Post M ig density is statistically insignificant for children s jewelry, they are also negative. The estimated coe cient on the interaction between the post dummy and the density of female migrants is for the share of expenditures on women s (column (1)). Given that the standard deviation of the density of female migrants is (see Panel A of Table 1), this result shows that the decrease in the share of expenditures on women 11

12 is 0.11 percentage points higher if the density of female migrants increases by one standard deviation. Since the mean of the share of expenditures on women is 0.036, one standard deviation increase in the density of female migrants leads to a 2.8% decrease in this share (the product of 0.01 and 0.11 divided by 0.036). Similar calculation shows that the share of expenditures on cosmetics and washing machines would decrease by 5.5% and 11.7% respectively as the density of female migrants increases by one standard deviation. This result consistently illustrate that the expenditures on the female-favored goods decreased more in cities where the change in hukou status inheritance rule were expected to have larger e ects on the local marriage market. However, as discussed in Section 4, the OLS estimates might be inconsistent if the migration has already responded to the policy change before To address this issue, we use the density of female migrants aged years in 1990 as an IV for the density in The density of female migrants in 1990 is a strong predictor for that in The first stage result is shown in column 1 of Table 3. The coe cient of the post dummy and the density of female migrants in 1990 is and statistically significant at the 1% level. It means that having a 1 percentage point higher female migrant density in 1990 leads to percentage points higher female migrant density in The F-value shown in the last row is , much higher than the conventional criterion for a valid IV. The second stage results of the IV estimation are shown in columns 2-6 of Table 3. Those IV results are even stronger than the OLS results. Except for the expenditures on jewelry (column 6), the coe cients of the interaction of the post dummy and the female migrant density are statistically significant at least at the 10 percent level in all columns. The estimated coe cients are respectively (women s in column 2), (cosmetics in column 3), (children s in column 4), and (washing machine in column 5). Since the share of expenditures on women s is (see Table 1), the one standard deviation increase (roughly 0.11) of the female migrant density can therefore leads to 2.44% (the product of 0.11 and divided by 0.036) decrease of expenditures on women s. The similar calculation shows that the one standard deviation increase of the female migrant density leads to 7.3%, 7.3%, and 11% decrease in the expenditures on 12

13 cosmetics, children s, and washing machines, respectively. The decrease in the share of expenditure on women is proportionally smaller than the decreases in expenditures on other female-favored goods. The possible reason is that cosmetics and jewelry are more like luxurious goods than women. Since the data used in our empirical analysis do not have a panel data structure, one may concern that the potential changes in the composition of households may confound our results, especially when the policy change a ects the marriage matching pattern. To rule out this concern, we conduct the analysis using households formed prior to the policy change. Because the UHS data do not have the information of marriage year, we restrict our sample to the households having children older than 2 years. This group of households is most likely to be formed before the policy change. The results estimated using this subsample is reported in Table 7. The coe cients of the interaction of the post dummy and the female migrant density are all statistically significant except for that in column (5). They are (column 1), (column 2), (column 3), and (column 4), respectively. These estimates are remarkably similar to the estimated coe cients using the full sample (columns 2-5 of Table 3), suggesting that the possible change in the sample composition would not a ect our estimates. One threat to the validity of the IV estimation is that the IV could be correlated with the pre-existing time trends which drive the estimation results. To address this concern, we conduct a placebo test. That is, we replace the city level migrant density in 2000 in Equation (1) with the migrant density in 1990, and then we estimate Equation (1) using the sample of years 1996 and In estimating this equation, we label year 1997 as the post year. If the pre-existing trends are correlated with the IV, we would expect that the interaction term Post Mig density c,1990 to be statistically significant. The estimation results are presented in Table 5. The coe cients of the interaction term Post Mig density c,1990 are statistically insignificant for all five outcome variables. These results provide us more confidence that the IV is not correlated with the pre-existing time trends. Another threat to the validity of IV is that the IV could be correlated with macroeconomic conditions in di erent cities, which may induce di erent household expenditure patterns. 13

14 Including macroeconomic variables in the regression, as we do in this paper, can alleviate this problem. However, one may still concern that some unobserved macroeconomic variables might be correlated with the IV. To address this concern, we estimate the e ects of the policy change on male-favored expenditures and gender-neutral consumption on electronics such as refrigerators and TV sets. If the estimated e ects of the policy change on femalefavored expenditures are driven by the unobserved macroeconomic variables, we should see the same pattern for male-favored or gender neutral goods. Our estimation results are presented in Table 6. We can see that, contrary to the result for female-favored goods, all the coe cients on the interaction term of the post dummy and the female migrant density in 2000 are positive and statistically significant at least at the 10% level. That is, the share of expenditures on male-favored goods such as men s, cigarette, and alcohol and gender-neutral electronics such as refrigerators and TV sets increased more after the policy change in cities with higher levels of migrant density. This evidence rules out the concern that our main result is driven by unobserved macroeconomic variables. The results above combined illustrate that granting men the same right as women to pass hukou status to children lowers the relative bargaining power of females in the urban-urban marriages. 5.2 The Persistence of the E ects One natural question to ask is whether the adverse e ect of the change in the hukou inheritance rule on urban women persists over time. If so, it implies that the enlarging inequality between men and women due to the policy might be more severe than expected and it should be taken into account in the future policy making process. To investigate this issue, we use two more years (2000 and 2001) of data. We replace the interaction of the post dummy and female migrant density in 2000 in Equation (1) with three interactions, i.e. the interactions of the year dummies for 1999, 2000 and 2001 with the female migrant density in 2000, respectively. As in the above, we use the interactions of these three year dummies and the female migrant density in 1990 as an IV. The results are shown in Table 10. The adverse 14

15 e ect on the consumption of women s (column 1) and washing machine (column 5) persists through the period , while the negative e ect on cosmetics and jewelry is statistically insignificant in years 2000 and For children s, the coe cients of the interactions of year dummies for 1999 and 2001 with the female migrant density are significant while that of the interaction of year dummy for 2000 with the female migrant density is not significant. Overall, the adverse e ects of the policy change on some female-favored consumption are not all transitory. At least some e ects remain even three years after the reform. 6 Testing the Channel Theoretically an assumption necessary for the change in the hukou status inheritance rule to matter for the within-household bargaining is that people care about the status of their prospective child. That is, if the policy e ect arises from the change in the right to pass hukou to their children, we would expect that the policy impact is stronger for those who are more likely to have children in future. Therefore, we test whether this is the case. Firstly, we investigate whether the e ects of the policy change are smaller for households having more children. Having more children reduces husbands marginal utility from children and therefore their incentive to have another one. To conduct this analysis, we add the triple interaction of the number of children, the post dummy and the female migrant density (together with all relevant double interactions) to Equation (1), and then estimate the equation using the same IV (and its interaction with relevant variables). The results are presented in Table 8. The coe cient of interest is that of the triple interaction, which captures to what extent having one more child can o set the adverse policy e ect on femalefavored consumption. We can see that the coe cients on the triple interaction term are positive and statistically significant for washing machine (column 4) and jewelry (column 5). The magnitude of the coe cients is and 0.003, respectively. It shows that the more children do the couples have, the weaker is the negative e ect of the policy change. Secondly, we investigate whether the e ects of the policy change are weaker for older 15

16 couples. The probability for older couples to have another child is lower and therefore the policy impact is expected to be lower. We divide the whole sample into two subsamples, one including couples older than 45 years old and the other one including couples younger than 35 years old. We estimate Equation (1) using these two samples separately. The results are shown in Table 9. Due to the space limit, we only show the coe cient of the interaction of the post dummy and the female migrant density. The first column shows the results for the older sample while the second column shows the results for the younger sample. Except for the coe cient for washing machine, all other coe cients shown in the first column are not significant. However, all the coe cients shown in column 2 are statistically significant and the magnitudes of the coe cients are also larger. These results support that the e ects of the policy change are weaker for couples who are less likely to have more children. Thirdly, we examine whether the policy impacts are weaker for couples who have a son. In China, son preference remains strong. Given that raising a child is costly, one s incentive to have more children would be lower if he/she already has a son. We thus include the triple interaction of the post dummy, the female migrant density and the indicator for having a son (together with all relevant double interactions) in Equation (1), and Table 4 shows the estimation results. As shown in Table 4, the coe cient of the triple interaction is significant at the 1% level for expenditures on washing machine (column 4), equal to For other outcome variables such as women s and children s, the coe cient of the triple interaction is not significant. This result is consistent with the finding in previous literature that women having a son enjoy a higher status within household (Li and Wu, 2011). The above three tests consistently and strongly suggest that the estimated policy impact on intrahousehold bargaining indeed arises from people s concern regarding the status of prospective children. 7 Conclusion This paper explores how changing from the matrilineal hukou status inheritance rule to the maximum inheritance rule a ects the intrahousehold bargaining power of couples who have 16

17 the valuable urban hukou. We find that, granting men the same right as women to pass their hukou status to the next generation results in a decrease in female-favored consumption and an increase in male-favored consumption. This finding suggests a worsening position of urban females in the urban-urban marriages. This e ect is particularly strong for women in relatively weak position, e.g., those who do not have a son. The possible reason for which the e ect arises is that the change in the hukou inheritance rule e ectively increases the substitutability between urban females and rural migrant females in the urban marriage market. Although the hukou system is unique in China, our finding is useful for understanding similar changes in other contexts. Many societies have changed from the patrilineal status inheritance rule to the maximum inheritance rule, which grants the females equal rights to pass on their status. This change is often considered as part of female empowerment. Our finding suggests that this type of female empowerment would increase the bargaining power of females who have high status. However, females with low status may not be able to benefit from this type of female empowerment as their relative position in the marriage market may even deteriorate. This is a direction that merits future research. 17

18 References [1] Anderson, Siwan, and Mukesh Eswaran What determines female autonomy? Evidence from Bangladesh. Journal of Development Economics, 90 (2009), pp [2] Becker, Gary S A Theory of Marriage: Part I. Journal of Political Economy, LXXXI, p [3] Becker, Gary S A Theory of Marriage: Part II. Journal of Political Economy, 82(2), Part 2:, pp [4] Becker, Gary S. A Treatise on the Family. Cambridge, Mass.: Harvard Univ. Press, [5] Bertrand, M., Emir Kamenica and Jessica Pan Gender Identity and Relative Income within Households. revise and resubmit at Quarterly Journal of Economics. [6] Browning, M., F. Bourguignon, P. Chiappori, V. Lechene Income and outcomes: a structural model of intrahousehold allocation. Journal of Political Economics, 102(1994), pp [7] Browning, M, and PA Chiappori E cient intra-household allocations: A general characterization and empirical tests. Econometrica, [8] Chiappori, PA, B Fortin, G Lacroix Marriage market, divorce legislation, and household labor supply. Journal of political Economy, 110 (1), [9] Duflo, Esther Grandmothers and granddaughters: old-age pensions and intrahousehold allocation in South Africa. World Bank Economic Review, 1 (2003), pp [10] Duflo, Esther and Christopher Udry Intrahousehold resource allocation in Cote d Ivoire: social norms, separate accounts and consumption choices. NBER Working Paper, (2004) [11] Edlund, lena, Elaine Liu and Jin-Tan Liu Beggar-Thy-Women: Domestic Responses to Foreign Bride Competition, the Case of Taiwan. Working papers. 18

19 [12] Han, Li, Tao Li and Yaohui Zhao How status inheritance a ects marital sorting: evidence from China. Working paper. [13] Luke, Nancy and Kaivan Munshi Women as agents of change: female income and mobility in India. Journal of Development Economics, vol. 94 (1) (2011), pp [14] Sun, Ang and Yaohui Zhao Divorce, abortion and sex selection at birth: the e ect of the amended divorce law in China. Working paper. [15] Wang, Shing-yi Property rights and intra-household bargaining. Journal of Development Economics, 107, March 2014, Pages

20 Table 1 Summary Statistics Mean S.D. Panel A. Household level variables Total expenditures per capita (yuan) Share of expenditures on women Share of expenditures on cosmetics Share of expenditures on children Share of expenditures on washing machine Share of expenditures on jewelry Share of expenditures on men Share of expenditures on cigaret Share of expenditures on alcohol Share of expenditures on TV Share of expenditures on refrigerator Husbands' schooling years Wives' schooling years Husbands' age Wives' age Have sons Number of children Family size Share of 0-6 years old HH member (male) Share of 7-18 years old HH member (male) Share of years old HH member (male) Share of 60+ years old HH member (male) Share of 0-6 years old HH member (female) Share of 7-18 years old HH member (female) Share of years old HH member (female) Share of 60+ years old HH member (female) OBS Panel B City level variables Ratio of migrant women in Ratio of migrant women in GDP per capita (yuan) GDP growth rate Share of GDP in pimary industry Share of GDP in secondary industry Share of GDP in tertiary industry Average wage (yuan) OBS 49!!!! 20!

21 Table 2 Impact of Local Marriage Market on Intra-household Resource Allocation, OLS (1) (2) (3) (4) (5) Share of Expenditures on: Women Cosmetics Children Washing machine Jewelry Post*Mig_density (0.004)*** (0.001)*** (0.002) (0.004)* (0.001) Husband's age (0.001) (0.000)*** (0.000)** (0.002) (0.001) Wife's age (0.001)*** (0.000)* (0.000)*** (0.002)* (0.001) Husband's schooling years (0.001)*** (0.000) (0.000)*** (0.001) (0.000) Wife's schooling years (0.001)*** (0.000)*** (0.000)** (0.001) (0.000) Ln(total expenditures per capita) (0.001)*** (0.000)*** (0.000)*** (0.001)*** (0.000)*** Household demographic structure YES YES YES YES YES Macroeconomic variables YES YES YES YES YES Observations R-squared Robust standard errors in parentheses are calculated by clustering over city level. * significant at 10%; ** significant at 5%; *** significant at 1%. Year dummies and city dummies are controlled in all columns. 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 Ratio of female family members aged above 60 is omitted to avoid multi-colinearity. Macroeconomic variables include log GDP per capita, GDP growth rate, ratio of GDP in primary industry, ratio of GDP in secondary industry, and log average wage in city level.!!! 21!

22 Table 3 Impact of Local Marriage Market on Intra-household Resource Allocation, IV (1) (2) (3) (4) (5) (6) Share of expenditures on: Post* Mig_density2000 Women Cosmetics Children Washing machine Jewelry Post*Mig_density Post*Mig_density2000 (Post*Mig_density1990 as IV) (0.462)*** (0.003)** (0.001)*** (0.001)*** (0.004)* (0.001) Husband's age (0.001) (0.001) (0.000)*** (0.000)** (0.002) (0.001) Wife's age (0.002) (0.001)*** (0.000)* (0.000)*** (0.002)* (0.001) Husband's schooling years (0.001) (0.001)*** (0.000) (0.000)*** (0.001) (0.000) Wife's schooling years (0.001) (0.001)*** (0.000)*** (0.000)** (0.001) (0.000) Ln(total expenditures per capita) (0.001) (0.001)*** (0.000)*** (0.000)*** (0.001)*** (0.000)*** Household demographic structure YES YES YES YES YES YES Macroeconomic variables YES YES YES YES YES YES Observations R-squared F-value Robust standard errors in parentheses are calculated by clustering over city level. * significant at 10%; ** significant at 5%; *** significant at 1%. Year dummies and city dummies are controlled in all columns.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 Ratio of female family members aged above 60 is omitted to avoid multi-colinearity. Macroeconomic variables include log GDP per capita, GDP growth rate, ratio of GDP in primary industry, ratio of GDP in secondary industry, and log average wage in city level.!!!!! 22!

23 Table 4 Using households having children aged at least two years (1) (2) (3) (4) (5) Share of expenditures on: Women Cosmetics Children Washing machine Jewelry Post*Mig_density2000 (Post*Mig_density1990 as IV) (0.003)*** (0.001)*** (0.001)*** (0.004)* (0.001) Husband's age (0.002) (0.000)*** (0.000)** (0.002) (0.001) Wife's age (0.001)*** (0.000) (0.001)*** (0.003) (0.001)* Husband's schooling years (0.001)*** (0.000) (0.001)*** (0.002) (0.000) Wife's schooling years (0.001)*** (0.000)* (0.000)*** (0.002) (0.001) Ln(total expenditures per capita) (0.001)*** (0.000)*** (0.000)*** (0.001)*** (0.000)*** Household demographic structure YES YES YES YES YES Macroeconomic variables YES YES YES YES YES Observations R-squared Robust standard errors in parentheses are calculated by clustering over city level. * significant at 10%; ** significant at 5%; *** significant at 1%. Year dummies and city dummies are controlled in all columns.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 Ratio of female family members aged above 60 is omitted to avoid multicolinearity. Macroeconomic variables include log GDP per capita, GDP growth rate, ratio of GDP in primary industry, ratio of GDP in secondary industry, and log average wage in city level.!!!! 23!

24 Table 5 Pre-trend Testing (1) (2) (3) (4) (5) Share of expenditures on: Women Cosmetics Children Washing machine Jewelry Dummy for 1997*Mig_density (0.017) (0.005) (0.007) (0.041) (0.017) Husband's age (0.001)* (0.000)*** (0.000)*** (0.001) (0.001) Wife's age (0.001)** (0.000)*** (0.000)*** (0.001) (0.001) Husband's schooling years (0.002)*** (0.000) (0.000)*** (0.003) (0.000) Wife's schooling years (0.001)*** (0.000)*** (0.000)** (0.002) (0.001) Ln(total expenditures per capita) (0.001)*** (0.000)*** (0.000)*** (0.001)*** (0.000)*** Household demographic structure YES YES YES YES YES Macroeconomic variables YES YES YES YES YES Observations R-squared Robust standard errors in parentheses are calculated by clustering over city level. * significant at 10%; ** significant at 5%; *** significant at 1%. Year dummies and city dummies are controlled in all columns.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 Ratio of female family members aged above 60 is omitted to avoid multi-colinearity. Macroeconomic variables include log GDP per capita, GDP growth rate, ratio of GDP in primary industry, ratio of GDP in secondary industry, and log average wage in city level.!!!!! 24!

25 Table 6 Impact of Local Marriage Market on Intra-household Resource Allocation (1) (2) (3) (4) (5) Share of expenditures on: Post*Mig_density2000 (Post*Mig_density1990 as IV) Men Cigaret Alcohol TV Refrigerator (0.004)*** (0.006)*** (0.001)* (0.019)*** (0.011)*** Husband's age (0.001)** (0.003) (0.001)* (0.005) (0.003) Wife's age (0.001) (0.003) (0.001) (0.005) (0.003) Husband's schooling years (0.001)*** (0.003)*** (0.001)*** (0.006) (0.003) Wife's schooling years (0.002)*** (0.004)*** (0.001)*** (0.005)*** (0.003) Ln(total expenditures per capita) (0.001)*** (0.002)** (0.001)** (0.005)*** (0.002)*** Household demographic structure YES YES YES YES YES Macroeconomic variables YES YES YES YES YES Observations R-squared Robust standard errors in parentheses are calculated by clustering over city level. * significant at 10%; ** significant at 5%; *** significant at 1%. Year dummies and city dummies are controlled in all columns.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 Ratio of female family members aged above 60 is omitted to avoid multicolinearity. Macroeconomic variables include log GDP per capita, GDP growth rate, ratio of GDP in primary industry, ratio of GDP in secondary industry, and log average wage in city level.!!!! 25!