Title: The Impact of Circular Migration on the Position of Married Women in Rural China
Abstract: Abstract This study examines the impact of migration on women's positions in Chinese rural households. A number of studies have found that rural Chinese migrant women experience more autonomy and freedom in urban areas than they would at home. But do these experiences carry over into marriage when they return to rural areas? Using a survey of more than 3,000 married, rural women in Anhui and Sichuan provinces and controlling for potential endogeneity of migration and return, this paper explores four main categories of women's status: women's views on male/female relationships, women's roles in household decision making, women's relationships with their husbands, and women's views concerning parents and children. It concludes that for women from Anhui and Sichuan, migration has some statistically significant lasting effects on a woman's position in the household, though the effects are not always positive, nor are they universal. Keywords: Women's empowermentinternal migrationChinese women Acknowledgments The authors would like to thank Zhengming Xie and Bin Bin Lou of the China Population Information Research Center for all their work in data collection and also acknowledge funding from the Beijing Office of the Ford Foundation. Thanks as well to Sarah Cooke, Dirk Early, Maggie Maurer-Fazio, Nancy Riley, and three anonymous referees for their useful comments and suggestions. Notes 1 The system of household registration in China is being reevaluated and revised, and there are signs that the situation regarding migrant settlement might change dramatically in the near future (Roberts Citation2007). 2 Margaret Maurer-Fazio, James Hughes, and Dandan Zhang (2005) identify a substantial increase in the labor-force participation of older rural women between 1990 and 2000, which includes on-farm labor. Much of this may be caused by the migration of the younger generation and consequent loss of their farm labor. 3 Women in the China Rural Poverty Survey listed agricultural labor requirements as the most important variable affecting their migration decisions after their children (Yang Du, Albert Park, and Sangui Wang 2005). 4 This data has been used for several other studies published in both Chinese and English. Chinese readers may be interested in the book Renkou liudong yu nongcun funu fazhan [Migration and rural women's development], edited by Zhenzhen Zheng and Zhenming Xie (2004), in which each chapter is based on the RMS as well as focus groups and individual interviews that were conducted at the same time. One of the chapters by Rachel Connelly, Kenneth Roberts, Zhenming Xie, and Zhenzhen Zheng, "Waichu dagong dui nongcun funu diwei de yingxiang" [The effect of migration on rural women's decision making], focuses on similar questions to this paper but uses a simple descriptive approach that does not consider the statistical issues of selection of out-migrants, nor the selection of return. In terms of articles published in English, Lou et al. (Citation2004) describe the impact of migration on the lives of young women using mainly the qualitative information, while Kenneth Roberts, Rachel Connelly, Zhenming Xie, and Zhenzhen Zheng use the same survey data as this paper but consider the diversity of patterns of women's migration uncovered through this data collection. Roberts et al. (2004) was one of the first papers to argue that married women were migrating in large numbers at the turn of the twenty-first century. Before that there was a sense that the vast majority of migrant women were single. In unpublished analysis of the data, we found that if we ignored all migration episodes that happened after 1995, we too would have concluded that the majority of migration was by single women, but there were still a substantial number of married migrants in the pre-1995 sample. 5 According to Nancy Riley, "Joint is such a nebulous concept. If a woman has a tiny little role in a decision, that's joint. But that kind of role (say she says to her husband, showing a tiny doubt, 'really, you think you want to buy that multiplex TV? Oh …') is different from an equally shared decision or one where the woman takes a larger role in the decision. But they all fall under the rubric 'joint'" (2003). 6 Even this statement is subject to some doubt based on the interesting work done by Sharon J. Ghuman, Helen J. Lee, and Herbert L. Smith (2002), who show that questions measuring women's empowerment must be interpreted within the context of their society and are difficult to compare across locations. 7 Marrying a man who is not one's first boyfriend could be a direct result of migration if single migrant women find their spouses while they are away from home. However, most women in the sample did marry a man from their home county, and there is no substantial difference in the proportion marrying outside the county by migration status. It is also possible that not marrying one's first boyfriend was not the migrant woman's choice. There have been cases in which the boyfriend rejects the migrant girlfriend upon her return, questioning her fidelity, or finding her changed by the experience. 8 None of these questions are unambiguously indicators of increased empowerment. It is certainly possible that other forces are at work, such as an increased reticence to report beatings rather than a decline in actual beatings. However, the substantial number of women reporting incidents of domestic violence is some evidence that women do not feel the need to hide this fact. 9 One might consider including family income level, but income is certainly potentially endogenously related to having migrated. Thus, we must think of our estimation as a reduced form estimation in the sense that some of the effect of migration and return we observe may be the result an indirect process by which migration leads to potentially higher family income of migrants, and higher income leads to changes in women's position. Migration of women would also be expected to affect the proportion of monetary income contributed by the wife, and that too may contribute to changes in women's status. This effect is also subsumed in our reduced form coefficient of migration/return. 10 This model is similar to the one used in William Greene (Citation1998). We cannot separate the determinants of migration from return since we only observe in our data women who both migrated and returned or women who never migrated. Women who migrated and did not return are absent from our sample. One could theoretically estimate the decision to migrate and the decision to return separately using a method for partial observability as proposed by Dale J. Poirier (Citation1980), but the estimation procedure requires a variable that is thought to affect out-migration but is not expected to affect return and vice versa. This seems a difficult task conceptually, much less empirically. Individual characteristics surely affect both, and village characteristics would affect both the push and the pull (or lack of it) back. Later in this paper, we explore two sub-samples of the married women chosen to reduce the selectivity of return. 11 The migration/return equation is potentially identified by five variables included as variables that affect migration probabilities and excluded from the personal and contextual variables affecting women's position. These variables are: the proportion of the village estimated by local authorities to be currently migrating; the distance of the village to the nearest bus stop; the distance to the nearest train station; and two indicators of whether the terrain is flat, hilly, or mountainous. Du, Park, and Wang (2005) use a variable similar to the proportion currently migrating to identify migration propensity in their model. In addition, the estimation strategy we use allows for a robust estimation of the standard errors and corrects the standard errors for the possible clustering caused by multiple women in the sample coming from the same village. There are thirty-eight villages in our data. We used Stata's biprobit to estimate the bivariate probit and the marginal effects. 12 We used Stata's heckprob procedure, which uses a maximum likelihood estimation procedure. Again, we calculate robust standard errors adjusted for the clustering of the sample in villages. 13 The tables with a full set of results are available from the authors upon request. Education, age, and having a husband who migrated often affect a woman's position variables. The coefficients of the migration/return equation are remarkably consistent across the women's position variables. Identifiers, the proportion of the villagers in the sample who have ever migrated, distance to the bus stop, and the terrain of the village are usually key predictors of having migrated and returned. In particular, the proportion of the village women who have ever migrated is always statistically significant and serves as a strong identifier of the ever migrated/returned equation. Having attended primary school, attended middle school, age, and whether one's husband ever migrated are also consistently predictors of migration/return. 14 One might be concerned that a woman's ideal living arrangement was highly correlated with her actual living arrangement, but the correlation is only 0.43. Clearly, many women imagine an ideal arrangement that is different from their actual one. 15 Tables showing the full set of results for Models 1 and 2 are available from the authors upon request. The results shown in Tables 3 and 4 are the marginal effects of only the variables directly related to characteristics of migration. 16 Brown (Citation2009) argues that one of the purposes of dowries in a rural Chinese marriage is that the woman's parents can transfer goods to her in order to increase her power in the marriage. We readily acknowledge that both the propensity to argue with one's husband and the probability of reporting being a victim of domestic violence are imperfect measures of empowerment at best, full of potential effects in both directions. Violence against one's wife could be a backlash against her growing empowerment upon returning from migration or even just frustration that she is somehow different than she was. 17 Women in these villages are allowed to have two children since the villages are quite remote. 18 The correlation is essentially zero in the last third of the equations.
Publication Year: 2010
Publication Date: 2010-01-01
Language: en
Type: article
Indexed In: ['crossref']
Access and Citation
Cited By Count: 53
AI Researcher Chatbot
Get quick answers to your questions about the article from our AI researcher chatbot