Strategic Gender Egalitarianism in Rural China: The Impacts of Husbands’ Migration on Gender Relations

Abstract

When men migrate for work and couples live separately, structures of gender relations may be altered in response. Our research in rural China about husbands’ and wives’ perceptions of household decision-making and gender-related attitudes examines hypotheses derived from Connell’s (2002) theory of gender relations using the Gansu Survey of Children and Families. Consistent with previous research, we find that both migrant husbands and their wives perceive that women have more decision-making power over household production and consumption. In contrast, the effect of husbands’ migration on their own and their wives’ gender attitudes are less consistent. Both migrant husbands and their wives are more likely than non-migrant spouses to agree that women and men can achieve the same given equal opportunities. Moreover, migrant husbands hold more gender egalitarian views towards their children’s education than non-migrant husbands. However, migrant husband and their wives are no more likely to question sons’ obligations or men’s and women’s roles in the family than their non-migrant counterparts. We interpret our findings as strategic gender egalitarianism, that is, gender egalitarianism born of necessity, particularly economic necessity. Strategic gender egalitarianism does not, however, challenge dominant gender structures that define family relationships and allocate power unequally.

Introduction

While gender is an important determinant of who migrates (Donato & Gabbacia 2015), the meaning of gender is also transformed by migration (Grasmuck & Pessar 1991; Herrera, 2013; Hondagneu-Sotelo, 1992, 1994; Mahler & Pessar, 2006). Generally, men are more likely to migrate for work than women, especially when regional labor markets and gender ideology support this pattern (Donato & Gabbacia 2015). But the conditions that create this gendered pattern may also alter how gender is performed, especially for male migrants who are not accompanied by their wives and families. Many researchers find that married couples who live separately because of husbands’ migrations hold more gender egalitarian beliefs and exhibit gender neutral behaviors (Hondagneu-Sotelo, 1992, 1994; Kyle, 2000; Peleikis, 2000; Schafer, 2000). If these findings are generalizable, then migration may be a powerful force transforming structures of gender relations.

Migration may alter gender relations by forcing migrants to adopt new behaviors and exposing migrants to new ideas. Most scholars argue that migration-related separation forces men and women to take on roles typically associated with the “opposite sex”, thus demonstrating that it is both possible and even desirable to have more flexible gender roles. Adding nuance to this argument, Parrado and Flippen (2005) argue that the contrast between the structural contexts of migrants’ origin and destination communities influence the ease with which migrants take on these new roles. To better understand how migration alters gender ideology and gendered practices, researchers must examine gender relations within migrant households across a range of contexts using a range of methods (Pessar & Mahler, 2003; Curran, Shafer, Donato, & Garip, 2006).

Rural China is an ideal location to study migration and married couples’ gendered practices and attitudes. During the past four decades of economic reforms, gender relations within families have changed dramatically as a result of massive rural-to-urban migration, the growth of “left-behind” wives and children in rural areas (Lu, 2012), and women’s increased labor force participation in both agricultural and non-agricultural economic sectors (Chang, MacPhail, & Dong, 2011; De Brauw, Li, Liu, Rozelle, & Zhang, 2008). In this paper, by examining migrant husbands’ and wives’ household decision-making power and gender ideology relative to non-migrant couples, we focus on how husbands’ migration influences attitudes toward two structures of gender relations. The first is labor, production, and consumption, and the second is cathexis, or the organization of people’s emotional and symbolic relations within the family (Connell 2002). By focusing on these two structures of gender relations, we argue that migration alters the structure of labor, production, and consumption through the adoption of strategic gender egalitarianism while leaving the crucial components of the structure of cathexis unchallenged.

Literature Review

Gender and Migration

Instead of viewing migration as a household decision, as in the New Economics of Labor Migration (Stark & Bloom, 1985), gender and migration scholars propose that migration decisions are shaped by gender ideology and inequality within the family (Donato & Gabbacia, 2015; Grasmuck & Pessar, 1991; Mahler & Pessar, 2006; Hondagneu-Sotelo, 1994; Paul, 2015). Consequently, gender is strongly associated with who migrates and who is left behind. A common pattern of migration that draws on gender ideology and practice is the split household strategy, in which married couples live separately with husbands migrating for employment and wives remaining in the origin community to perform domestic and farm labor and care for children and elderly parents. The split household strategy maintains the structure of gender relations in some domains, while altering that structure in others since gendered practices are adapted when a spouse is absent (Hondagneu-Sotelo, 1992; Kanaiaupuni, 2000; Kyle, 2000; Peleikis, 2000; Schafer, 2000).

To distinguish domains that may be affected by migration, we draw on Connell’s (1987; 2002) conceptualization of structures of gender relations. Connell (2002) describes three dimensions of gender relations, including structures of labor, production, and consumption (or labor for short); institutional power; and cathexis. Typically, these structures maintain men’s advantage over women. However, migration may alter these structures to produce a more egalitarian distribution of resources and power within couples, and perhaps, the larger family. For example, Flippen and Parrado (2005) find that Mexican women in the U.S. are more likely to participate in the labor force but, on average, report less power in their marital relationships and hold more conservative representations of gender than their non-migrant counterparts who reside in the Mexico. Thus, migration resulted in gender role transformation in some domains and gender role continuity in others (Flippen, 2014; Flippen & Parrado, 2015; Parrado & Flippen, 2005; Parrado, Flippen, & McQuiston, 2005).

In this paper, we focus on two of these gender structures – labor and cathexis – to examine whether migration alters the gender attitudes of men and women in rural China. We ask whether husbands and wives in households split by migration show evidence of women’s greater involvement in decision-making about household production and consumption and more egalitarian attitudes toward children, filial obligations and gender roles compared to husbands and wives in non-migrating households. In the following sections, we provide background for our hypotheses about split household strategies and the structures of labor and cathexis.

Split household strategies and the gender structure of labor

Split-household strategies are prevalent in China for married rural-to-urban migrants because the household registration (hukou) system discourages male rural migrants from migrating with their families. In 1958, the Communist Party of China established the hukou system, a household registration system that serves to deter and control migration from rural to urban areas. The system was relaxed in 1978 after China launched its economic reform and the central government allowed more rural people to migrate in response to growing urban labor demand (Chan, 2010). Despite this demand, it is difficult for most rural-to-urban migrants to meet the central and local governments’ requirements for transferring to urban hukou status (Chan & Zhang, 1999; Chan, 2010).

Without urban household registration, rural-to-urban migrants are denied the social welfare benefits received by urban residents. Until the mid-1990s, this meant that rural migrant children could not attend urban schools (Human Rights in China, 2002). Reforms to this policy still present considerable obstacles for migrant children in school enrollment (Sa, 2004; Dong, 2010). As a result, a large number of rural households adopt split-household strategies with wives and children remaining behind while the husbands migrate to urban areas for employment.

Most previous research on couples engaging split household strategies focuses on the structure of labor, production, and consumption. Generally, women with migrant husbands take on more family responsibilities and often generate income during spousal separation, a pattern found in Mexico (Hondagneu-Sotelo, 1992), Ecuador (Kyle, 2000), West Africa (Peleikis, 2000), and Zimbabwe (Schafer, 2000). This pattern is also evident among non-migrant women with migrant husbands in rural China. Rural women have increased their participation in both unpaid farming and paid off-farm work despite slackening demand for agricultural labor due to technology-induced increases in productivity (Chang et al., 2011; De Brauw et al., 2008; Mu & Van De Walle, 2011; Wang, 1999). Most non-migrant wives farm to fulfill the subsistence needs of their family (Chang et al., 2011). Building on this research, we hypothesize that women in migrant households have greater decision-making power over production and consumption than women in non-migrant households (Hypothesis 1).

Split household strategies and the gender structure of cathexis

Gendered attitudes toward children

The social reforms producing rural-to-urban migration that began in the 1970s coincided with the Chinese government’s family planning program designed to limit population growth¹. However, while urban family planning boards restricted married couples to having one child, rural couples were allowed to have a second child if the firstborn child was a daughter. Consequently, families in rural China are gendered by the political and cultural contexts and gender inequalities in sibling composition and investments in children persist (Greenhalgh & Winckler, 2005).

This urban-rural difference is evident in parents’ attitudes toward boys and girls. In urban families, girls in single child families receive as much parental time, energy and financial investments as boys in single child families (Liu, 2017). In contrast, sons in rural areas might be treated more favorably than their sisters, who must participate in domestic activities or paid labor instead of or in addition to education (Basu & Van, 1998; Fong, 2004; Liu, 2017; Tsui & Rich, 2002). When fathers migrate to urban areas they may observe the more egalitarian treatment of urban children and adopt a more egalitarian attitude toward their own children. Hence, we hypothesize that migrant husbands are more likely to hold gender egalitarian attitudes towards children than nonmigrant husbands (Hypothesis 2a). Since nonmigrant women are not exposed to urban gender ideology, we hypothesize that there is no significant difference in gender egalitarian attitudes towards children between women with and without migrant husbands (Hypothesis 2b).

Attitudes toward filial obligations

The traditional patrilineal family system supports son preference (Cohen, 2005; Das Gupta et al. 2003). In this system, married couples are expected to live with or close to the husband’s parents and work together with an understanding that sons will support their parents in old age and, in return, inherit the family assets. The absence of social pensions in rural China fortifies this multi-generational solidarity between parents and their married sons and daughters-in-law (Guilmoto, 2009; Ebenstein & Leung, 2010). Sons also confer symbolic status to their parents since they represent the household in family and community rituals and maintain the lineage (Cohen, 2005; Guilmoto, 2009; Das Gupta et al. 2003). Migration in no way disrupts this practice, therefore we hypothesize that husband’s migration will have no significant impact on the belief that sons should provide old-age support to their parents (Hypothesis 3).

Gender ideology

Previous research finding that married couples who live separately because of husbands’ migrations exhibit more gender egalitarian behaviors suggests that this may also be found in rural China (Hondagneu-Sotelo, 1992, 1994; Kyle, 2000; Peleikis, 2000; Schafer, 2000). Husbands’ experiences in providing for their own domestic needs while in urban areas and their exposure to gender egalitarian urban areas may challenge their previously held views. Similarly, non-migrant women with migrant husbands may hold more gender egalitarian views because of their greater autonomy and decision-making around household production and consumption. Therefore, we hypothesize that husbands and wives in migrant households will have more gender egalitarian attitudes than those in non-migrant households (Hypothesis 4).

Data and Methods

The Gansu Survey of Children and Families (GSCF) is a longitudinal, multi-level study of children’s wellbeing in rural area of the Gansu province, one of the poorest areas in China. It provides measures of family and household characteristics, children’s education enrollment and performance, school environment, and community-level characteristics for a representative sample of 2,000 children from 100 villages in rural Gansu. Using village household registration lists, the GSCF selected children aged 9 to 12 years old in 2000. We used the second wave of GSCF data, which consisted of a household survey and separate surveys for mothers and fathers of the sample children in 2004.

Dependent variables

We have two groups of dependent variables. The structure of labor map onto survey items in the GSCF data about decision-making power over children’s schooling, purchase of new durable goods, farming, husbandry, family finance, parenting, and household management. Respondents are allowed to answer (1) you decide; (2) decide together; or (3) your spouse decides. To create a dichotomous variable measuring wives’ decision-making, we code husbands’ responses as (1) your spouse decides, and (0) decide together or you decide, and wives’ responses as (1) you decide, and (0) decide together or your spouse decides

The structure of cathexis maps onto survey questions about gender attitudes. A series of questions in GSCF measure husbands’ and wives’ gender attitudes towards children, gender roles, and adult sons’ filial obligations. The questions about attitudes towards children focus on whether they agree that boys and girls have equivalent academic capability as well as equal educational opportunities, whether parents should encourage girls to have their own opinions, and whether it is useless to send daughters to school. The questions about gender egalitarianism ask whether men and women can achieve the same given equal opportunities, whether important family decisions should be made by men, and whether a couple should share housework if both of them work full time. One final question measures filial obligations which asks whether parents should rely on sons for old-age support. Husbands and wives were asked separately to what extent they agreed with the statements using the following response options: 1=disagree; 2=no opinion; 3=agree. Our dichotomous variables denote whether the respondent agree with the statements (1), or disagrees or has no opinion (0).

Key independent variable: migrant husband

A migrant husband is defined as a male household head and wage earner who was away from home for at least three months and whose workplace was not in the same place that his spouse and children lived in the past year. We created one dummy variable for husband’s migration status in 2003 (the survey was implemented in 2004). If the husband migrated, the value will be “1”, otherwise the value will be “0”. We also deleted households with migrant wives (about 2 percent of the sample) and deceased or divorced parents (about 3 percent of the sample) to avoid any confounding effects of parental separation on gender dynamics within the households.

Our model includes a set of controls that may be associated with patterns of decision-making and attitudes. We include the age and years of education of both husbands and wives and number of children into our model to control for family characteristics. Moreover, since the gender dynamics in households with only boys or only girls are quite different from those households with both boys and girls, we include dummy variables to control for the effects of sibling sex composition (only girl(s), only boy(s), and both girl(s) and boy(s)) with households that have only boys as the reference group. We also include household wealth as measured by family fixed assets and durable goods. Fixed assets are transportation equipment, such as trucks, cars and farm tools. Durable goods are household items of value such as televisions, radios, furniture, and jewelry. The measure of household wealth is highly skewed, so we take the natural logarithm of household wealth to minimize the effects of extreme cases.

Since all of our dependent variables are dichotomous, logistic regression is used to estimate the impacts of husband’s migration on separate items of family decision-making power and gender ideology.² Furthermore, both sets of logistic regression models predicting family decision-making power are used to predict whether the wives in the households make family decisions or not.

Results

Table 1 shows the descriptive results for all independent variables based on our full sample. Sample sizes vary between outcomes because of differences in non-response to the dependent variables. The overall sample size is 1,763. Nineteen percent of the households have a migrant husband. The average age of husbands and wives are 41.5 and 39.2. Husbands have more education than wives in our sample (mean of 7.1 years of education for husbands compared to mean of 4.4 for wives). On average each household has 2.3 children, ranging from only one child up to 6 children. While 9.7 percent of the households have daughters only, 23.3 percent of the households have sons only and the majority of the households have both sons and daughters (67 percent).

Table 1.

Descriptive Statistics for Independent Variables

Key independent variable
Father’s migration in 2003 (%)	18.60%
	Mean	Sth. Dev.	Min	Max
Controls
Father’s age	41.52	5.04	30	66
Mother’s age	39.16	4.41	29	68
Father’s education	7.13	3.47	0	15
Mother’s education	4.36	3.46	0	13
Logged family wealth	9.83	0.93	5.75	15.56
Number of children	2.34	0.73	1	6
Sibling sex composition (%)
Only boys household	23.3%
Only girls household	9.7%
Households with both boys and girls	67.0%
Observations	N=17633

³The descriptive statistics are calculated based on our sample for mother’s survey (N=1763). Overall, the descriptive statistics do not vary too much if we use different samples based on different models.

Split household strategies and the gender structure of labor: decision-making power

The majority of the husbands and wives perceived that they shared decision-making power over most labor, production, and consumption measures. The only exception to this is management of family finances, for which both husbands and wives most commonly perceived that husbands have sole decision-making power, followed by wives, and it is least common for spouses to share decision-making power (Figures 1 and 2). Among those husbands and wives who did not perceive shared decision-making power in their marriages, husbands were more likely to perceive that they made these decisions alone and wives were more likely to perceive that their husbands made these decisions alone. The share of husbands who thought their wives had sole family decision-making power is generally small, with the highest proportion of husbands (29%) saying that their wives had sole power over family finances and the lowest percentage (5%) saying their wives had sole power over children’s schooling. The share of wives who thought they had sole family decision-making power is also small, but slightly higher than husbands who thought their wives had sole decision-making power, and has a similar ordering of the items as the husbands. Although the overwhelming pattern of response is shared decision-making, husbands are more likely to be perceived by themselves and their wives to have more power.

Figure 1.

Husbands’ Perception of Family Decision-Making Power

Figure 2.

Wives’ Perceptions of Family Decision-Making Power

We hypothesize that husbands’ migration is associated with wives having greater decision-making power over household production and consumption compared to wives in nonmigrant households (Hypothesis 1). Our logistic regression results show that, in general, migrant husbands and women with migrant husbands are more likely to report that women have sole decision-making power in four out of seven domains, although the domains differ between husbands and wives (Table 2). To facilitate interpretation, we report the predicted probabilities for each of the four equation domains in which we observe a statistically significant relationship, holding all other variables at their mean values (Figures 3 and 4). Both migrant husbands and wives of migrant husbands have higher probabilities of agreeing that women had sole decision-making power in farming, buying or selling livestock, and managing family finances compared to non-migrant husbands and wives of migrant husbands. Migrant husbands are more likely to agree that their wives have sole decision-making power about parenting, while wives of migrant husbands are more likely to agree that they have sole decision-making power in the purchase of new durable goods. Neither children’s schooling nor household management are significantly associated with husband’s migration.

Table 2.

Logistic Regression Models Predicting Husbands’ and Wives’ Perceptions of Family Decision-Making Power (Family Decisions Made by Wives) (Odds ratios)

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7
	Children’s schooling	Purchase of New Durable Goods	Farming	Buying or Selling Livestock	Managing Family Finance	Parenting	Household Management
PANEL A: Husbands’ Perceptions of Wives’ Family Decision-Making Power
Husband’s migration	1.04 (0.35)	1.08 (0.20)	2.36*** (0.48)	2.11*** (0.47)	1.29* (0.20)	1.76** (0.40)	1.33 (0.28)
Husband’s age	1.01 (0.04)	1.00 (0.02)	0.98 (0.03)	1.05* (0.03)	1.03* (0.02)	0.99 (0.03)	1.01 (0.03)
Wife’s age	0.98 (0.04)	1.02 (0.03)	1.00 (0.04)	0.93* (0.04)	0.97 (0.02)	1.01 (0.04)	1.00 (0.03)
Husband’s education	0.90*** (0.03)	1.04* (0.02)	1.00 (0.03)	1.01 (0.03)	0.99 (0.02)	0.95* (0.03)	0.96 (0.02)
Wife’s education	1.04 (0.04)	1.09*** (0.02)	1.06** (0.03)	1.08*** (0.03)	1.07*** (0.02)	1.08*** (0.03)	1.10*** (0.03)
Logged family wealth	1.07 (0.14)	0.95 (0.07)	1.01 (0.10)	0.90 (0.09)	0.91 (0.06)	0.92 (0.09)	0.80*** (0.07)
Number of children	1.08 (0.18)	1.03 (0.10)	1.05 (0.14)	1.12 (0.16)	0.98 (0.08)	1.13 (0.15)	0.95 (0.12)
Sibling sex composition (reference category: only boys)
Only girls household	1.54 (0.65)	1.18 (0.30)	0.91 (0.30)	0.84 (0.30)	0.89 (0.20)	1.24 (0.42)	0.93 (0.28)
Both boys and girls household	1.14 (0.36)	1.08 (0.19)	0.82 (0.18)	0.75 (0.17)	1.07 (0.15)	0.95 (0.22)	0.93 (0.19)
Constant	0.06* (0.10)	0.07*** (0.07)	0.17 (0.23)	0.21 (0.30)	0.77 (0.62)	0.15 (0.20)	0.86 (0.98)
Observations	1,664	1,664	1,663	1,661	1,664	1,663	1,664
Pseudo R-squared	0.02	0.02	0.02	0.03	0.01	0.02	0.02
PANEL B: Wives’ Perceptions of Their Own Family Decision-Making Power
Husband’s migration	0.90 (0.22)	1.70*** (0.26)	2.13*** (0.32)	1.74*** (0.28)	1.37** (0.18)	1.11 (0.21)	1.24 (0.20)
Husband’s age	0.94* (0.03)	0.97 (0.02)	1.00 (0.02)	0.96* (0.03)	1.02 (0.02)	1.00 (0.03)	0.99 (0.02)
Wife’s age	1.04 (0.04)	1.06** (0.03)	0.98 (0.03)	1.03 (0.03)	1.00 (0.02)	1.02 (0.03)	1.02 (0.03)
Husband’s education	0.94** (0.03)	0.98 (0.02)	0.98 (0.02)	0.98 (0.02)	0.99 (0.02)	0.99 (0.02)	0.95** (0.02)
Wife’s education	1.03 (0.03)	1.08*** (0.02)	1.04* (0.02)	0.99 (0.02)	1.02 (0.02)	0.99 (0.02)	1.04** (0.02)
Logged family wealth	0.89 (0.09)	0.95 (0.07)	1.02 (0.08)	1.11 (0.09)	1.04 (0.06)	0.90 (0.07)	0.92 (0.07)
Number of children	1.03 (0.15)	1.02 (0.10)	1.10 (0.11)	1.14 (0.12)	0.98 (0.08)	0.93 (0.10)	0.86 (0.09)
Sibling sex composition (reference category: only boys)
Only girls household	0.62 (0.23)	0.94 (0.23)	0.97 (0.26)	0.96 (0.27)	0.88 (0.18)	0.86 (0.24)	0.96 (0.23)
Both boys and girls household	0.75 (0.17)	0.86 (0.14)	1.13 (0.20)	1.15 (0.21)	1.05 (0.14)	0.91 (0.17)	1.00 (0.16)
Constant	1.07 (1.44)	0.10** (0.10)	0.21 (0.21)	0.09** (0.09)	0.14*** (0.10)	0.31 (0.33)	0.73 (0.67)
Observations	1,754	1,754	1,754	1,753	1,754	1,754	1,754
Pseudo R-squared	0.01	0.02	0.02	0.01	0.005	0.004	0.01

^***p<0.01

^**p<0.05

^*p<0.1

Figure 3.

Predicated Probabilities of Family Decisions Made by Wives (Husbands’ Perceptions)

Figure 4.

Predicated Probabilities of Family Decisions Made by Themselves (Wives’ Perceptions)

Although we find statistically significant relationships, we are cautious in our interpretations since the pseudo R-squared values in these models are quite small (0.01 to 0.02). This reflects the high level of agreement among respondents in their perceptions of family decision-making power. These models also show that wives’ education is a consistent and significant predictor of husbands’ perception of their family decision-making power across nearly all domains (Table 2). Wives’ education is less consistently associated with wives’ perception of their own family decision-making power. Overall, the results confirm our first hypothesis, that husbands’ migration is associated with husbands’ and wives’ perceptions of her sole decision-making over household production and consumption.

Split household strategies and the gender structure of cathexis

Following the presentation format in the previous section, Figures 5 and 6 show the distribution of responses to each of the eight measures of gender attitudes. Tables 3 and 4 present the results of the logistic regression estimating the model for eight outcomes associated with the structure of cathexis. To facilitate interpretation, the predicted probabilities of the models with significant effects for husbands’ migration are shown in Figures 7 and 8. These models hold all other variables at their mean to isolate the effect of husbands’ migration on the outcome variable.

Figure 5.

Husbands’ Gender Values

Figure 6.

Wives’ Gender Values

Table 3.

Logistic Regression Models Predicting Husbands’ Gender Values (Odds Ratios)

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
	Boys and girls have equivalent academic capability	Boys and girls should have equal education opportunities	Parents should encourage girls to have their own opinions just like boys	It is useless to send daughters to school	Parents should rely on sons for old-age support	Men and women can achieve the same given equal opportunities	If both wife and husband work full-time, they should share the housework	Important family decisions should be made by men
Husband’s migration	2.69** (1.16)	2.08** (0.68)	1.27 (0.26)	1.24 (0.24)	1.14 (0.17)	1.75* (0.51)	0.85 (0.15)	0.91 (0.14)
Husband’s age	0.98 (0.03)	1.04 (0.03)	1.00 (0.02)	1.01 (0.02)	1.02 (0.02)	0.97 (0.03)	0.95** (0.02)	1.01 (0.02)
Wife’s age	1.03 (0.04)	0.99 (0.04)	1.03 (0.03)	0.95* (0.03)	0.98 (0.02)	1.05 (0.03)	1.05** (0.03)	1.02 (0.02)
Husband’s education	1.03 (0.03)	1.01 (0.03)	1.01 (0.02)	0.96* (0.02)	0.95*** (0.02)	1.03 (0.03)	0.98 (0.02)	0.96** (0.02)
Wife’s education	1.05 (0.04)	1.10*** (0.03)	1.06** (0.02)	0.93*** (0.02)	0.95*** (0.02)	1.04 (0.03)	1.02 (0.02)	0.98 (0.02)
Logged wealth in 2004	1.34** (0.15)	0.96 (0.10)	0.89 (0.07)	0.92 (0.07)	0.95 (0.05)	1.09 (0.10)	0.98 (0.07)	0.91* (0.05)
Number of children	0.95 (0.15)	0.80* (0.11)	1.00 (0.10)	0.94 (0.10)	0.97 (0.07)	0.80* (0.10)	0.85* (0.08)	1.01 (0.08)
Sibling sex composition (reference category: only boys)
Only girls household	0.99 (0.36)	4.34*** (2.13)	1.79** (0.51)	1.10 (0.29)	0.38*** (0.08)	1.72 (0.60)	1.57* (0.41)	1.14 (0.23)
Both boys and girls household	1.59* (0.42)	1.53* (0.34)	1.17 (0.20)	1.03 (0.18)	0.79* (0.10)	1.39 (0.30)	1.11 (0.18)	1.05 (0.14)
Constant	0.49 (0.71)	2.89 (3.74)	3.30 (3.24)	3.64 (3.63)	4.39** (3.23)	2.82 (3.38)	7.88** (7.03)	1.75 (1.33)
Observations	1,664	1,664	1,664	1,663	1,664	1,663	1,664	1,664
Pseudo R-squared	0.04	0.04	0.01	0.03	0.03	0.02	0.01	0.01

^***p<0.01

^**p<0.05

^*p<0.1

Table 4.

Logistic Regression Models Predicting Wives’ Gender Values (Odds Ratios)

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
	Boys and girls have equivalent academic capability	Boys and girls should have equal education opportunities	Parents should encourage girls to have their own opinions just like boys.	It is useless to send daughters to school	Parents should rely on sons for old-age support	Men and women can achieve the same given equal opportunities	If both wife and husband work full-time, they should share the housework.	Important family decisions should be made by men
Husband’s migration	1.32 (0.34)	1.30 (0.31)	1.24 (0.21)	0.89 (0.16)	1.28* (0.16)	1.55** (0.33)	1.04 (0.16)	1.35** (0.18)
Husband’s age	0.10 (0.03)	0.97 (0.03)	0.99 (0.02)	0.97 (0.02)	1.03* (0.02)	0.98 (0.02)	1.00 (0.02)	1.01 (0.02)
Wife’s age	1.00 (0.04)	1.03 (0.03)	1.02 (0.03)	0.99 (0.03)	0.98 (0.02)	1.03 (0.03)	1.01 (0.02)	1.00 (0.02)
Husband’s education	1.08*** (0.03)	1.04 (0.03)	1.03 (0.02)	0.94*** (0.02)	0.98 (0.02)	1.04* (0.02)	0.98 (0.02)	0.99 (0.02)
Wife’s education	1.01 (0.03)	1.02 (0.03)	1.05** (0.02)	0.93*** (0.02)	0.94*** (0.02)	1.04 (0.03)	1.05** (0.02)	0.96*** (0.02)
Logged wealth in 2004	1.11 (0.12)	0.96 (0.09)	1.12 (0.08)	0.92 (0.07)	0.89** (0.05)	1.26*** (0.11)	1.01 (0.07)	1.00 (0.06)
Number of children	1.05 (0.15)	1.07 (0.14)	1.18* (0.12)	0.98 (0.10)	1.13* (0.08)	1.00 (0.11)	1.03 (0.09)	1.11 (0.09)
Sibling sex composition (reference category: only boys)
Only girls household	0.91 (0.30)	1.51 (0.52)	1.50 (0.38)	0.59* (0.17)	0.57*** (0.11)	1.48 (0.50)	0.63** (0.14)	1.06 (0.20)
Both boys and girls household	1.19 (0.28)	1.19 (0.25)	1.04 (0.16)	0.80 (0.14)	0.90 (0.11)	0.80 (0.16)	0.72** (0.11)	1.24* (0.16)
Constant	2.20 (2.90)	6.74 (8.24)	0.51 (0.46)	4.97 (4.94)	2.02 (1.44)	0.45 (0.49)	3.10 (2.61)	0.98 (0.72)
Observations	1,763	1,763	1,763	1,762	1,763	1,763	1,763	1,763
Pseudo R-squared	0.02	0.01	0.01	0.03	0.03	0.03	0.01	0.01

^***p<0.01

^**p<0.05

^*p<0.1

Figure 7.

Predicated Probabilities of Agreeing with Gender Egalitarian Statements Among Husbands

Figure 8.

Predicated Probabilities of Agreeing with Gender Egalitarian/ Conservative Statements Among Wives

As with the logistic regressions for family decision-making power, the fit of these models is only fair, with small pseudo R-square values. This reflects the high levels of agreement among husbands and wives with respect to gender values as seen in Figures 5 and 6. Some of the control variables provide additional explanatory power, particularly husbands’ and wives’ years of education, and in a few cases the gender composition of the sibling set, with more education and households with only daughters associated with more gender egalitarian values.

Gendered attitudes toward children

Overall, both husbands and wives in the sample hold gender egalitarian views towards their children’s education and academic performance (Figures 5 & 6). Almost all husbands and wives agreed that boys and girls had equivalent academic capability if they work hard and that boys and girls should enjoy equal educational opportunities. A smaller majority of husbands and wives agreed that parents should encourage girls to have their own opinions just like boys and disagree that it is useless to send daughters to school.

The regression models show that migrant husbands have more gender egalitarian beliefs about their children’s academic capabilities and education opportunities than nonmigrant husbands (Table 3, Models 1 and 2), but no significant difference is evident between migrant and nonmigrant husbands in encouraging girls to have their own opinions or the value of sending their daughters to school (Table 3, Models 3 and 4). The difference in predicted probabilities are small with 97 percent of migrant husbands agreeing with both statements and 93 percent of nonmigrant husbands agreeing (Figure 7). These results are consistent with hypothesis 2a, that husband’s migration positively influences his gendered attitudes toward children. In contrast, husband’s migration has no significant effect on wives’ gender attitudes toward children. This finding is consistent with hypothesis 2b, that husband’s migration would not affect wives’ gendered attitudes toward children.

Filial obligation

The practice of sons’ and daughter-in-laws’ provision of support to elderly parents persists in China and we do not expect that rural-to-urban migration will affect it. Figures 5 and 6 show that 51% of husbands and 48% of wives agree that they should rely on their sons for old-age support. Husbands’ migration is not associated with agreement with this statement among husbands, however, surprisingly, wives with migrant husbands are more likely to agree that parents should rely on sons for old-age support (Table 3 & 4). In fact, 50% of wives with migrant husbands agree with this statement compared to only 44% of wives with nonmigrant husbands (Figure 8). Thus, there is qualified support for hypothesis 3, that husbands’ migration is not associated with agreement that sons should support their elderly parents.

Gender ideology

We consider three measures of gender ideology (Tables 3 and 4, Models 6, 7, and 8). Among husbands and wives, migration predicts greater agreement with the statement that men and women can achieve equally given equal opportunities (Model 6). Figure 7 shows that 94 percent of migrant husbands and 93 percent of wives of migrant husbands agree with this statement compared to only 91 percent of nonmigrant husbands and 90 percent of wives of nonmigrant husbands. Husbands’ migration is not associated with agreement among husbands or wives that if both husband and wife work full-time they should share the housework. Notably, among wives, husbands’ migration is associated with a less egalitarian attitude; wives of migrant husbands are more likely to agree that important family decisions should be made by men. Migrant husbands are no different from nonmigrant husbands in agreement with this statement. Altogether, these results provide mixed support for hypothesis 4, that migrant husbands and wives are more likely to express gender egalitarian views than non-migrant husbands and wives of husbands who do not migrate. However, the unexpected finding that wives of migrant husbands endorse the statement that men should make important family decisions suggests that these women may resist the decision-making role that they acquire when their husbands migrate.

Discussion

Rural China is a critical site for studying gender and migration due to the massive rural-to-urban migration during the past four decades and the long-standing traditional patrilineal system. Our study shows that the associations between husband’s migration and structures of gender relations are not uniformly positive. A split household strategy caused by rural-to-urban migration challenges the structure of labor but simultaneously reinforces the symbolic status of men as the head of household who should make important decisions as well as reliance on sons for old-age support. We describe this as strategic gender egalitarianism.

Strategic gender egalitarianism associated with husbands’ migration results from necessity and affects mostly the structure of labor, production and consumption. It does not challenge the structure of cathexis, or family relations. Women’s gains in decision-making power are largely based on pragmatic considerations, such as managing the home and farm. Such gender egalitarianism coexists with traditional gender relations within the family and accommodates the economic benefits of husbands’ migration. In fact, under the split household strategy women remain economically dependent on migrant husbands since they lack viable local labor market opportunities and engage mostly in subsistence activities (Lu and Tao, 2015). Another crucial component that remains unchallenged is the provision of old-age support, since the absence of social pensions in rural China reinforces this practice (Guilmoto, 2009; Ebenstein & Leung, 2010).

However, challenges to the gender structures of labor and cathexis are evident in migrant husbands’ more gender egalitarian attitudes towards children’s schooling relative to non-migrant husbands. Exposure to gender norms in urban areas may account for this broadening of migrant husbands’ gender attitudes (Hondagneu-Sotelo, 1994; Lu & Tao, 2015), but our findings indicate that this effect does not extend to their wives and their attitudes toward their sons’ filial obligations. These distinct views are not necessarily contradictory. Parents may invest in their daughter’s education for different reasons than they invest in their sons. Better educated daughters are more likely to find marriage partners and highly educated sons are better able to obtain higher paid non-farm employment and support their elderly parents (Cohen, 2005; Ebenstein & Leung, 2010; Das Gupta et al. 2003; Mu & Van De Walle, 2011). Consequently, migrant husbands’ exposure to modern gender norms may enhance education equality but still not shift parental reliance on their sons for old-age support.

Our research has some limitations. First, it is based on cross-sectional data, so we can only observe associations between husbands’ migration and perceived decision-making and gender attitudes of couples. Second, we only observe men’s contemporaneous migration, not their entire migration history. Consequently, we may be underestimating the effect of migration if some husbands who appear as non-migrants have migrated in previous years. Nevertheless, our results suggest new directions for this area of research, specifically, further investigation of the effect of split household strategies on distinct structures of gender relations. Third, although the Gansu Survey of Children and Families instructed the interviewers to interview spouses separately, sometimes in rural households, privacy is not possible. Therefore, some of our respondents might have answered questions in the proximity of their spouse or other family members.

Our research complements the literature on male migration and family gender dynamics by providing a quantitative analysis of men’s migration and gender relations (Curran et al., 2006; Donato, Gabaccia, Holdaway, Manalansan, & Passar, 2006). In our study gender is not merely a dichotomous variable; it is measured by men’s and women’s perceptions of their own decision-making power and family relationships. Therefore, our paper contributes to a growing number of studies in gender and migration that seek to understand gender as a relational process (Donato et al., 2006). Like Flippen and Parrado (2015), we find that in rural China migration transforms some structures of gender relations while leaving others intact. We call this strategic gender egalitarianism because it allows the structure of labor to adapt to migration while maintaining the structure of family relations. Future studies in gender and migration can further study how changes in structural and cultural contexts induced by migration could potentially reconstruct or reinforce gender relations.

Appendix

Table A1:

Descriptive Statistics for Husbands and Wives’ Gender Values

	Husbands’ Gender Values			Wives’ Gender Values
	Disagree	No opinion	Agree	Disagree	No opinion	Agree
	%	%	%	%	%	%
Working hard, both boys and girls can do well at school.	0.9	5.0	94.1	1.70	5.30	93.0
Girls should enjoy same education opportunities as boys.	1.8	6.5	91.7	2.0	6.7	91.3
Parents should encourage girls to have their own opinions just like boys.	2.0	13.9	84.1	3.5	14.9	81.6
It is useless to send daughters to school, since they will eventually marry out.	73.4	11.1	15.5	74.8	10.7	14.5
Parents should rely on their sons for old-age support.	27.0	22.5	50.5	27.9	24.1	48.0
Given equal opportunities, women and men can achieve the same.	2.4	7.0	90.6	3.0	8.5	88.5
If both wife and husband work full-time, they should share the housework.	3.4	17.0	79.6	4.3	17.6	78.1
Important family decisions should be made by men.	19.5	18.1	62.4	21.0	15.5	63.5

Table A2:

Descriptive Statistics for Husbands and Wives’ Perceptions of Family Decision-Making Power

	Husband			Wife
	Your spouse decide	Decide together	You decide	Your spouse decide	Decide together	You decide
	%	%	%	%	%	%
Children’s schooling	4.7	74.1	21.2	18.9	73.7	7.4
Purchase of new durable goods	17.3	57.1	25.6	25.0	56.8	18.2
Farming decisions	9.2	49.5	41.3	31.0	52.9	16.1
Buying or selling livestock	8.0	48.8	43.2	35.7	50.1	14.2
Managing family finance	28.9	27.2	43.9	41.9	27.7	30.4
Having disagreement on parenting	8.2	64.8	27.0	21.7	65.9	12.5
Having disagreement on household management	11.2	62.1	26.7	23.2	59.3	17.5