Partners or Rivals? Exploring the Relationship between Men’s and Women’s Empowerment in Bangladesh

Abstract

The aim of this study is to investigate the statistical relationship between husbands’ empowerment and wives’ empowerment. Using panel data from Bangladesh that include over 4,000 husband-wife pairs, we regress husband’s empowerment on wife’s empowerment and a set of covariates. We use a domain-based framework, similar to the Women’s Empowerment in Agriculture Index, to capture the multidimensional nature of empowerment. The findings reveal that the interplay between spouses’ empowerment varies across domains. Husbands and wives act as partners with regard to community influence and daily workload adequacy. Yet, they act as rivals with regard to input in productive decisions and ownership of assets. Additionally, we identify variables that are consistently associated with husbands’ and wives’ empowerment. For wives, their income and employment are positive correlates of empowerment. For husbands, their wives’ employment is a negative correlate, whereas household wealth is a positive correlate. Notably, the results also suggest that the gender asset gap rises with household wealth.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. Please refer to Appendix 1 for the English translation of all survey items used to construct the empowerment variables in this paper.

2. A related variable, control over use of income, which is based on the WEAI variable in the domain of income, yields very similar results. The correlation between input in productive decisions and control over use of income is greater than 0.90.

3. To complement the regression results, we examine in detail the pattern of zeros and missing values across rounds (see Table 4).

4. In lower income countries, it is reasonable to assume that joint ownership is associated with less control than sole ownership (Ambler et al., 2017; Fafchamps & Quisumbing, 2002; Jacobs & Kes, 2015; Roy et al., 2015). Nevertheless, we investigate the implications of assigning a value of 1 to joint ownership (see Table 3).

5. To complement the regression results, we examine in detail the pattern of zeros and missing values across rounds (see Table 4).

6. About 39 per cent of husbands have missing time use information in round 1. The issue appears to be an anomaly in the BIHS data, since all values are missing for husbands above a certain household identification number.

7. For both group membership and speaking in public, the coefficient on wife’s empowerment is positive and significant in multivariate regressions of husband’s empowerment on wife’s empowerment. We prefer community influence to the other variables, because it exhibits greater variation. For comparison, 83 per cent of husbands and 72 per cent of wives have a zero value for group membership.

8. In models without household fixed effects (see Table 3), we include controls for husband and wife’s age at round 1, husband and wife’s education, wife’s pre-marital assets, and religion (Muslim). These factors are not included in panel regression models because they are largely time-invariant.

9. The preferred coding is to assign a value of 0.5 to joint ownership, while the alternative coding is to assign a value of 1. Please refer to the discussion in subsection 3.2.