Women's Wealth and Intimate Partner Violence: Insights from Ecuador and Ghana

ABSTRACT

Intimate partner violence (IPV) by men against their partners is one of the most glaring indicators of women's lack of empowerment. Drawing upon the 2010 Ecuador Household Asset Survey (EAFF) and the 2010 Ghana Household Asset Survey (GHAS), nationally representative surveys for Ecuador and Ghana, respectively, this study investigates the relationship between women's ownership of assets and physical and emotional abuse by spouses against currently partnered women over the previous twelve months. It uses the value of a woman's total assets compared to those of her partner as the main proxy for a woman's bargaining power. Differentiating between physical and emotional violence in both countries, the study finds that women's share of couple wealth is significantly associated with lower odds of physical violence in Ecuador and emotional violence in Ghana. Moreover, the association between women's share of couple wealth and IPV is contingent on the household's position in the wealth distribution.

KEYWORDS:

• Spousal violence
• intrahousehold bargaining
• Africa
• Latin America

JEL Codes:

• D31
• J12
• O15

NOTES ON CONTRIBUTORS

Zachary B. Catanzarite is an interdisciplinary social scientist with M.A. degrees in both sociology and psychology and a focus on methodology and statistical analysis. He has particular interest in the intersection of social structure, culture, and mental health in low-income countries and has been working as a statistical analyst and research associate with the Gender Asset Gap project.

Carmen Diana Deere is Distinguished Professor of Latin American Studies and Food and Resource Economics at the University of Florida. She is co-author of Empowering Women: Land and Property Rights in Latin America (University of Pittsburgh Press, 2001) among other books, and co-editor of two special issues of Feminist Economics, on “Women and the Distribution of Wealth” (2006) and “Gender and International Migration” (2012). During 2009–10 she was a Visiting Scholar at FLACSO-Ecuador and directed the Ecuador Household Assets Study; she is co-primary investigator of the Gender Assets Gap comparative project.

Abena D. Oduro is Senior Lecturer at the Department of Economics, University of Ghana. She is co-author of Measuring the Gender Asset Gap in Ghana (University of Ghana and Woeli Publishing Services, 2011). Her current research interests are women and entrepreneurship and gender and asset ownership. She is co-primary investigator of the Gender Assets Gap comparative project.

ACKNOWLEDGMENTS

The authors are grateful to the Vanguard Charities Endowment Program for their financial support of the Gender Assets Gap project under which this research was undertaken, as well as to the other co-PIs of this project, Cheryl Doss, Caren Grown, and Hema Swaminatham, for their helpful suggestions. An earlier version of this paper was presented at the World Bank Workshop on Gender and Assets, June 14, 2012, Washington, DC.

Notes

¹ Herein we use “spouse” to refer to both married and unmarried partners, and for the purposes of this study, we only consider partnership between a man and a woman.

² In partial community property regimes, all assets purchased during marriage belong equally to each partner. Assets acquired prior to marriage and inherited assets (whether before or after marriage) are treated as separate property. In separation of property regimes, all assets irrespective of how and when they are acquired are regarded as individual property unless explicit arrangements are made for these to be joint marital property. See Deere, Oduro, Swaminathan, and Doss (2013) for a detailed analysis.

³ Ecuador's 1995 Law Against Violence Toward Women and the Family recognizes physical, psychological, and sexual violence not covered by the country's criminal code (República de Ecuador 1995). The implementation of the law was fairly haphazard until the government of Rafael Correa took office in 2007. Ghana passed its first Domestic Violence Act (Act 732) in 2007, criminalizing physical, sexual, emotional, and economic abuse.

⁴ The different results regarding the relationship between employment and IPV, however, could be due to potential endogeneity, as Andrés Villarreal (2007) and Bhattacharyya, Bedi, and Chhachhi (2011) argue.

⁵ See Macmillan and Gartner (1999), Catherine Kaukinen (2004), Resko (2010), and Lori L. Heise (2011), among others.

⁶ This comparative project, which also includes the state of Karnataka in India, was funded by the MDG3 Fund of the Dutch Foreign Ministry. The Karnataka survey did not include a module on IPV and thus is not included in this analysis.

⁷ Respondents were asked to value their assets in three ways: by their potential market value (what they might be sold for today); their replacement cost (that is, to construct the dwelling today); and, for the main dwelling and land, their potential rental value. Our estimates of gross wealth in this study are based on their potential sales value and include, besides for immovable property, agricultural tools, equipment and installations, livestock, businesses, all consumer durables, and all forms of financial assets, both formal and informal.

⁸ We are cognizant that in the study of IPV it is standard practice to pair respondents and enumerators by gender and not doing so may result in the underreporting of IPV. While the Ecuador survey intended to follow this practice, during the pilot survey it became evident that this would increase the costs of the survey beyond our budget. We tested whether our practice introduced a bias in the Ecuador results and found that the probability of reporting physical and emotional abuse is independent of the gender of the enumerator (Wald Chi Square test, p=0.540 for physical violence and p=0.776 for emotional abuse).

⁹ Household residency was defined as not being away for more than six months during the previous year. We exclude the partnered women in the sample whose spouses have been away for longer than this threshold, since information on their individual characteristics was not gathered in the household inventory.

¹⁰ Our survey results cannot be precisely compared to these other national surveys since they used different sampling procedures and the latter include partnered women irrespective of whether their spouses currently reside in the household. Our IRB protocol did not allow us to interview women under 18 years of age; we include women age 50 and over in our analysis to maximize the number of observations. Since the Ghana 2008 DHS survey was limited to women 18–49 years of age, comparable data for the GHAS are presented in Table 1; the Ecuador Gender Violence Survey 2011 interviewed women age 50 and over, hence that comparison is presented. The estimates presented are unweighted. We acknowledge that underreporting is a serious issue, particularly for Ghana. However, we do obtain interesting and comparable results regarding the factors differentiating physical violence from emotional abuse that could provide the basis for further research in the future.

¹¹ We explored the potential endogeneity between women's share of wealth and IPV using parents’ ownership of immovable property and whether they were literate as instrumental variables. Unfortunately, there were too many missing observations on these variables, which would have greatly reduced the sample size. We then ran a one-way ANOVA test for each country, which revealed no significant difference between physical and no violence and for Ecuador finds women who report emotional violence to have greater shares of couple wealth. This would appear contrary to the notion that couple share of wealth suffers from problems of endogeneity, as we either find no evidence of a direct causal relationship or find evidence opposite to what we would expect.

¹² Ideally, one would be able to control for the quality of employment as well, an exercise not undertaken for this paper.

¹³ Due to empty cells in certain subpopulations, we also fit models with education and education difference as continuous variables and collapsed employment and earnings categories into “equality” and “non-equality.” The results were comparable and thus we present our detailed models so as to gain specificity in our conclusions. There are two categories for employment and earnings instead of three in the model for Ghana because of missing observations in some categories.

¹⁴ These regression models are unweighted and do not take into account survey design effects.