How Economic Empowerment Reduces Women’s Reproductive Health Vulnerability in Tanzania

Abstract

This article uses data from Northern Tanzania to analyse how economic empowerment helps women reduce their reproductive health (RH) vulnerability. It analyses the effect of women’s employment and economic contribution to their household on health care use at three phases in the reproductive cycle: before pregnancy, during pregnancy and at child birth. Economic empowerment shows a positive effect on health seeking behaviour during pregnancy and at child birth, which remains robust after controlling for bargaining power and selection bias. This indicates that any policy that increases women’s economic empowerment can have a direct positive impact on women’s RH.

Acknowledgements

We gratefully acknowledge financial support of the PopDev program of NWO-WOTRO and the William and Flora Hewlett Foundation, the collaboration with Dr. Switbert Kamazima from Muhimbili University of Health and Allied Sciences (MUHAS) from Tanzania and the support from the Tanzanian National Bureau of Statistics (NBS). We also thank Bart van Rijsbergen, Janine Huisman and 14 Tanzanian research assistants for their support during the fieldwork. This article benefitted from useful comments on earlier versions by Ruerd Ruben, Anouka van Eerdewijk, Dereck Chitama and two anonymous referees. The dataset and code used for this paper is available on request from the corresponding author.

Notes

1. A research permit from the Tanzanian Commission for Sciences and Technology (COSTECH) and an ethical clearance from the Muhimbili University for Health and Allied Sciences (MUHAS), as well as consent from regional and district officers were obtained. The Tanzanian NBS provided the information required to revisit the women interviewed in 2004. The data collected in 2010 is authorised by MUHAS. Prior to the interview, regional officers of the NBS visited the enumeration areas to ask for permission and to track the women to be interviewed. The respondents were first informed by their local leaders and asked for consent. At the start of the interview, consent was asked again by the researchers.

2. It is realistic to assume that RH may also have an impact on women’s contribution to the household income. Storeng et al. (2008), found in Burkina Faso that near-miss pregnancy complications put strain on intra-household relations, as women often felt responsible for the depletion of income (due to high health costs) and were blamed for it. In response, many women tried to minimise spending on their own needs and tried to become self-sufficient by resuming domestic and agricultural work often before they felt physically ready for it.

3. It should be noted that feminist scholars would probably come to a similar hypothesis, but they would not attribute it to a common set of preferences among household members, but rather point to a lack of control over the earnings and the resilience of gender ideologies (for example Kabeer, 1997).

4. This assumption is also recognised by Lundberg and Pollak (1996, p. 146–147) who state: ‘If divorcing partners maintain ownership of income received separately within marriage, the demands emerging from marital bargaining will depend not on total family income but on the income received by the husband and income received by the wife.’

5. Although we analyse various types of household models, and power dynamics are likely to change if women increase their income share, we do not suggest policies to intervene with these dynamics as such.

6. We acknowledge that working and earning an income are only two of the possible channels to establish empowerment.

7. This decision leads to a loss in precision. However, the number of observations in the middle categories are too low (5%). We have run analyses with different categories, and concluded that the current distinction is the best reflection.

8. An alternative would have been to focus on sole decision-making by the respondent. This correlates strongly with the indicator used in the analyses (.6671). Running the analyses with the alternative indicator does lead to minor differences in the results, but the main effect of economic empowerment remains intact. Results available upon request.

9. Cronbach’s alpha on these four items is .8593. The loading on the factor is between .6379 and .8518.

10. Cronbach’s alpha is .9200, with factor loadings between .7571 and .9302.

11. Although we have tried to limit omitted variable bias as much as possible, it remains a potential concern as long as earning an income is not randomised. One possible omitted attribute might be related to personality characteristics of the women.

12. Working away from home does not necessarily mean being formally employed. It also includes women who work on land located further away from the home. Among the 15 per cent who contributes to the household financially, the largest share (70%) works in agriculture.

13. We expected that women who earn an income are also more likely to control household expenses as it is (partly) their money to be spent, hence this result is rather unexpected. It might be explained by the indicators of employment: women who are not working or working at home have the highest control over money. Most likely, they are in charge of daily businesses at home.

14. It is rather unexpected that women’s economic contribution is significantly correlated with decision-making but not with control over household expenses, despite that some household decisions have financial implications (small and large household purchases). However, the decision making index is broader than pure financial decisions (including the decision what to buy).

15. Ordinary Least Square regression results are preferred over Poisson regression for interpretation purposes. Poisson regression leads to similar results.

16. Based on a White test we conclude that there is no indication of heteroskedasticity in the analyses based on the year of pregnancy. See the Online Appendix for more details.

17. Likelihood-ratio test suggests that there is no heteroskedasticity based on the year of pregnancy (details can be found in the Online Appendix).