Over decades much attention has been devoted to the relationship between foreign aid and economic growth, while few studies have focused on the effects of foreign aid on female empowerment. This despite the fact that empowerment of girls and women is a key driver of development, and often an explicit objective of foreign aid. Using geo-coded data on aid project placement and household-level survey responses, Perrotta Berlin, Bonnier and Olofsgård (2023), show that foreign aid has a modest but robust effect on several dimensions of female empowerment. This is the case for both aid in general and gender-targeted aid, highlighting the potential of foreign aid to reduce gender inequalities. It is also found, though, that the impact is contingent on the context, and that there can even be a backlash in male attitudes towards female empowerment in more traditional communities.
The donor community has long been invested in the empowerment of women and girls, and the 2030 Agenda for Sustainable Development also includes gender equality as an explicit goal. Yet surprisingly little quantitative research has tried to make a broader assessment of the effect of foreign aid on gender equality measures.
This policy brief summarises a study by Perrotta Berlin, Bonnier and Olofsgård (2023) which addresses this question by matching the location of aid projects with geo-coded household surveys in Malawi between 2004 and 2010. Analysing the community-level impact on five different female empowerment indices, the study finds foreign aid to affect positively women’s empowerment across several dimensions. Furthermore, the authors find that gender-targeted aid has an additional impact on an index measuring women’s control over sexuality and fertility-related decisions and an index focusing on violence against women.
When considering areas with patrilineal land inheritance traditions, the results however partly shift, especially in relation to men’s attitudes. This implies that the success of foreign aid and gender-targeted aid in reducing gender inequalities may be conditional on the community context.
Gender Equality and Foreign Aid in Malawi
Malawi is highly dependent on foreign aid. Net official development assistance (ODA) has exceeded 10 percent of gross national income yearly since 1975, reaching as high as 23.5 percent in 2016 (World Bank, WDI database).
In recent years, reforms have been undertaken by the Malawian government to improve gender equality. The minimum legal age of marriage was raised from 15 to 18 through the 2015 Marriage, Divorce and Family Relations Bill, and the 2013 Gender Equality Act strengthened the legislation concerning gender-based violence and included a universal condemnation of all types of gender-based discrimination. Yet, in 2020, Malawi was ranked 116 out of 153 in the World Economic Forum Gender Gap Report and 172 out of 189 in UNDP’s Gender Inequality Index. An area of concern regards the high rates of child marriage, with 9 percent of girls already married at age 15 and 42 percent by the age of 18. Alongside these numbers, 31 percent of women report to have given birth by the age 18.
Another aspect potentially influencing gender equality is the prevalence of matrilinear land tenure systems, particularly in the southern and central parts of the country (as depicted in Figure 1). While previous research has shown that land ownership empowers women and suggested that property rights affect decision power over key decisions, fertility preferences, age of marriage etc., less research has been devoted to analysing the effects on women’s empowerment outcomes in a matrilinear kinship setting. Some recent literature however suggests women in matrilinear societies have greater say in household decisions – including financial ones – and are less accepting of, as well as exposed to, domestic violence (Lowes, 2021; Djurfeldt et al., 2018).
Figure 1. Intensity of matrilineal tenure in Malawi.
Methodology and Data
For the analysis, the authors make use of geo-coded data on aid projects from the Government of Malawi’s Aid Management Platform (AMP) and match it to household-level data from the Malawi Demographic and Health Survey (DHS). The country of Malawi and the period 2004-2010 were chosen in order to maximize data coverage on aid disbursement. Malawi’s AMP covers 80 percent of all aid entering the country during those years, which gives a much more complete picture compared to only focusing on one specific donor.
To identify causal impact, the authors apply a difference-in-differences specification on survey clusters in proximity to aid projects implemented between 2004 and 2010. Proximity was identified as within a 10-kilometer radius from an aid project. Among those, households interviewed in 2004, i.e., prior to the implementation date of any aid project, were considered the control group, and households interviewed in 2010 formed the treatment group. The underlying assumption of parallel pre-treatment trends was confirmed with the use of earlier DHS surveys. The model specification includes individual-level controls (age, ethnicity, household size, a Muslim dummy, years of education and literacy) and also a geographic fixed-effect based on a grid of coordinates.
The analysis distinguishes between the impact of aid in general, and the additional impact of gender-targeted aid. Gender-targeted projects are defined as projects that have any of the words woman, girl, bride, maternal, gender, genital or child, in the title, description or activity list. When estimating the effect of gender-targeted aid the authors control for overall aid intensity in the household’s vicinity. The estimated effect should therefore be interpreted as the additional effect from being exposed to a gender-targeted aid project while keeping the general number of aid projects in the area constant.
Figure 2. Map of aid projects and household clusters from 2004 and 2010 survey waves in Malawi.
To capture female empowerment, the authors make use of thousands of responses to DHS survey waves from 2004 and 2010. From these responses, the authors construct four different indices. Two of these are modelled on indices used in different contexts by Haushofer and Shapiro (2016) and Jayachandran et al. (2023). The former captures experiences of violence together with men’s and women’s attitudes towards violence, and some measures of decision making and control over household resources. The more recent index by Jayachandran et al. (2023) focuses on female agency and includes questions on women’s participation in decisions on large household purchases and daily expenditures, decisions on family visits, and decisions concerning their own healthcare.
To also capture questions related to sexual and fertility preferences, often regarded as measures of female empowerment, the authors construct two additional indices. The women’s attitudes index is based on responses to questions about whether the respondent is able to refuse sexual intercourse with her husband and ask him to use a condom, age at first marriage, and age at first childbirth, among others. The men’s attitudes index is based on questions about whether the respondent thinks it is justified to use violence to force intercourse, if a woman is justified to refuse intercourse, as well as fertility and child spacing preferences. In addition, all four indices are weighted and combined into an aggregated general index.
Considering all aid projects, the authors find that being exposed to an aid project in the 2004 to 2010 window has a significant positive impact on the agency index, the female attitude index and the combined general index (12, 11 and 31 percent of their respective means). When considering gender-targeted aid, the authors found the exposure to at least one such project to increase the women’s attitude index by 7 percent and the general index by 17 percent of their respective means. The impact is present for both a narrower and a wider exposure area, and quite persistent over time.
When breaking down the analysis for areas with matrilineal versus patrilineal land tenure systems the results diverge. In communities where the share of matrilineal ethnic groups exceeds the mean of 73 percent, the results are largely in line with those in the full sample. In patrilineal communities (< 73 percent matrilineal households), the results are however vastly different. Aid projects in general, and gender-targeted aid in particular, affect negatively the men’s attitudes index. In addition, gender-targeted aid seems to have no additional impact on the other indices.
In the paper underlying this brief, the authors study the effect of foreign aid on female empowerment, a frequent but understudied objective often set by donors. Looking at geo-coded aid projects in Malawi, the authors estimated such projects to positively impact girl’s and women’s empowerment across several indices. This is true for aid in general, and for some indices even more so when considering gender-targeted aid. Some of the positive results disappear or even change sign, though, in patrilineal communities, displaying the significance of pre-existing community norms for the effectiveness of development investments. Aid even generates a backlash when it comes to men’s attitudes towards women’s sexual and fertility preferences in these communities.
The takeaway from the study lies in foreign aid’s potential to empower women in targeted communities. This however hinges on pre-existing norms in recipient communities – something that aid donors should be aware of.
The authors emphasize the need for more research to better understand the role of pre-existing norms in the uptake of aid, to distinguish direct effects from aid from potential spillovers, and to understand what type of aid projects deliver the best outcomes in terms of female empowerment.
- Djurfeldt, A. A., E. Hillbom, W. O. Mulwafu, P. Mvula, and G. Djurfeldt. (2018). “The family farms together, the decisions, however are made by the man” -Matrilineal land tenure systems, welfare and decision making in rural Malawi. Land use policy 70, 601-610.
- Haushofer, J. and J. Shapiro. (2016). The short-term impact of unconditional cash transfers to the poor: experimental evidence from Kenya. The Quarterly Journal of Economics, 131(4), 1973-2042.
- Jayachandran, S., M. Biradavolu, and J. Cooper. (2023). Using machine learning and qualitative interviews to design a five-question survey module for women’s agency. World Development 161, 106076.
- Lowes, S. (2021). Kinship structure, stress, and the gender gap in competition. Journal of Economic Behavior & Organization 192, 36-57.
- Perrotta Berlin, M., Bonnier, E., and A. Olofsgård. (2023). Foreign Aid and Female Empowerment. SITE Working Paper Series, No. 62.
Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.
Recent economic research suggests that childhood environments in part determine success in life. So far, most of this evidence comes from rich countries. In a new paper, we use education data to measure intergenerational mobility across 26 African countries and find large differences across space. Results using data on migrants suggest that regions have causal effects on social mobility of Africans.
Why do people “make it” in life? Few of us can claim, as Robert Strauss, former U.S. ambassador to the Soviet Union and Russia, once quipped, that we were born in a log cabin we built ourselves. One chunk of individual success in climbing the social ladder is determined by our parents – be it through their genes (Sacerdote 2002, 2004), their parenting style (Doepke and Zilibotti 2017), or their income and connections. Another chunk is individual effort. Companies like Apple or Google were started in garages. That leaves our surroundings. Can the places we grow up in raise us up or pull us down? A growing body of research suggests that they can.
Growing evidence that “places matter” for individual mobility
At the forefront of these efforts, Chetty and Hendren (2018a, 2018b) have compared the incomes of American children to those of their parents. They link parents to kids through social security numbers in tax returns. Among families that moved, they find that children exposed to places with higher average social mobility for longer during childhood do better than children exposed to places with lower average mobility. Importantly, this holds when comparing the kids of parents with the same income and other observable characteristics, i.e. holding the “starting line” constant for everyone. Their findings have been reproduced for the U.S. (Chetty, Hendren, and Katz 2016), (Chyn 2018), Canada (Laliberte 2018), Australia (Deutscher 2018), and Denmark (Eriksen 2018). By contrast, studies identifying the causal effects of places on individual mobility in developing countries are still rare (a recent contribution by Asher, Novosad, and Rafkin (2018) on India is a notable exception).
New evidence from Africa
In a new paper (Alesina et al. 2018) we fill part of this gap by examining intergenerational mobility in Africa. After decades of stagnation, there is optimism about Africa’s future. Growth has returned (Young 2012), and some now see Africa as a continent of “1.2 billion opportunities” (Economist 2016). At the same time, anecdotal evidence suggests large inequalities, indicating that the recent aggregate gains may not be broadly shared, and that social mobility remains limited.
Measuring intergenerational mobility using education data
Measuring intergenerational mobility in Africa is difficult because of patchy data. Economists typically think of mobility in terms of income or wealth. In Africa, we lack tax records as well as administrative information linking children to parents. Instead we rely on censuses from 26 African countries and measure mobility using education data on children that share a household with their parents. [Card, Domnisoru, and Taylor 2018; Azam and Bhatt 2015; Narayan et al. 2018; Black and Devereux 2011 also study intergenerational mobility using education data.]
We measure upward mobility as the likelihood that kids of parents with less than primary education complete at least primary school. Similarly, we call an individual downwardly mobile if her parents have completed at least primary education and she has failed to do so. We compute these measures among children aged 14-18. This gives them enough time to complete primary school if they were ever going to do so. At the same time, most children at that age still live with their parents, which limits potential bias from co-residence selection.
Using education to measure social mobility has five advantages. First, education is a broad measure of living standards, reflecting not just income, but also aspirations and capabilities. Second, unlike income, much of which is informal and therefore under-reported in poor countries, schooling can be easily measured. Third, education, once completed, remains fixed and so intergenerational mobility can be assessed early in life. Fourth, “Mincerian returns” – how much extra income one more year of schooling commands in the labor market – seem to be especially high in Africa (Young 2012; Psacharopoulos 1994; Caselli, Ponticelli, and Rossi 2014), suggesting that education is a meaningful proxy of income. Finally, more schooling is correlated with many positive outcomes: household asset ownership, lower fertility, and even support for democracy. These correlations hold strongly comparing two individuals living in the same place, which means that education “quantity” is a useful stand-in-measure of living standards, even if the quality of schooling differs from place to place.
Main data patterns
The census data give us millions of individual observations to accurately measure intergenerational mobility over time (birth-cohorts) and in small geographic areas. First and most prominently, the descriptive analysis reveals differences in mobility both across and within countries. Figure 1 shows the geography of upward mobility across the 26 countries. Darker regions indicate places with lower mobility – children of illiterate parents are less likely to finish primary school.
Figure 1. Upward mobility across Africa
Source: Alesina et al. 2018
Country-differences are clearly important – South Africa is more mobile than Mozambique. Still, even within countries, there are vast differences as figure 2, which zooms in on Ghana, illustrates.
Figure 2. Upward mobility in Ghana
Source: (Alesina et al. 2018)
In some regions in Northern Ghana, average mobility is below .2 while it exceeds .8 in Accra, the capital. Second, while mobility does increase over time, these increases are modest and most pronounced in the most recent decades. This is still consistent with overall rising education, since average schooling in Africa was low until recently. Taking patterns one and two together, the persistent variation in mobility between places is more important than changes in mobility over time.
What accounts for differences in mobility across space? By far the strongest correlate of intergenerational mobility is the average literacy in the same place in the generation of the parents. This means that, comparing two individuals that grew up as children of illiterate parents in different regions, the individual that grew up in the region that has higher literacy in her parents’ generation has a greater chance of completing at least primary school. Several explanations might account for this pattern. Most simply, some regions have more schools than others, and can educate more individuals “per period”. One alternative story are peer effects: even though my parents are uneducated, I learn by example from the people around me that going to school is possible and desirable.
Beyond the correlation with initial education, we find that geography, colonial history, and at-independence development matter for intergenerational mobility. There are two important caveats to these results. First, pinning down the mechanism of why initial literacy and mobility are related remains a challenge. Second, these results represent correlations and not causally identified effects.
Causal effects of regions
To make causal inferences, we use data on families that have moved between two regions within a country in two ways. First, we compare siblings from migrant households, one child born in the origin of migration, the other in the destination. Figure 3 shows a (binned) scatter plot of the association between average birth-region upward mobility (computed among non-migrants) on the horizontal and individual likelihood of upward mobility on the vertical axis, conditional on household as well as birth-cohort effects. The slope indicates that kids born in a region with a ten percent higher mobility are 2.65 percent more likely to complete primary school compared to their siblings born in a different region with lower mobility.
Figure 3. Migrant vs non-migrant siblings
Source: (Alesina et al. 2018)
Second, we compare migrants that moved at different ages during childhood. Figure 4 plots the effects on individual outcomes of moving from a place of on average zero mobility to a place where all children of uneducated parents become educated against the age of the child at which the move occurred, once again comparing individuals within the same household. As intuition would suggest, earlier moves to better regions have larger positive effects than later moves, and effects turn insignificant towards the end of the period relevant for primary school.
For both empirical strategies, the sibling comparisons (enabled by household fixed effects) are crucial to separate treatment effects of regions from sorting whereby illiterate parents that may be more motivated/capable in educating their children move to regions with greater average opportunities.
Figure 4. Migration exposure effects
Source: (Alesina et al. 2018)
New research points to the importance of “places” in shaping individual social mobility. Complementing several recent works on developed economies, we document that opportunities for educational advancement vary widely within and across African countries. The strongest correlate of differences in mobility between places are differences in the initial education level in the generation of the parents, with more educated places showing higher mobility. Using information on migrants, we find that regions have a causal impact on individual outcomes. Taken together, our results suggest that initial conditions have persistent effects on the transmission of human capital between generations and that overall regional differences in human capital transmission in turn matter for who “makes it” in Africa.
- Alesina, Alberto, Sebastian Hohmann, Stelios Michalopoulos, and Elias Papaioannou. 2018. “Intergenerational Mobility in Africa.” Centre for Economic Policy Research Discussion Paper 13497 https://cepr.org/active/publications/discussion_papers/dp.php?dpno=13497
- Asher, Sam, Paul Novosad, and Charlie Rafkin. 2018. “Intergenerational Mobility in India: Estimates from New Methods and Administrative Data.” Mimeo, Dartmouth College.
- Azam, Mehtabul, and Vipul Bhatt. 2015. “Like Father, Like Son? Intergenerational Educational Mobility in India.” Demography 52 (6): 1929–59. https://doi.org/10.1007/s13524-015-0428-8.
- Black, Sandra E., and Paul J. Devereux. 2011. “Recent Developments in Intergenerational Mobility.” In Handbook of Labor Economics, 4B:1487–1541. Elsevier.
- Card, David, Ciprian Domnisoru, and Lowell Taylor. 2018. “The Intergenerational Transmission of Human Capital: Evidence from the Golden Age of Upward Mobility,” 102.
- Caselli, Francesco, Jacopo Ponticelli, and Federico Rossi. 2014. “A New Dataset on Mincerian Returns.” Unpublished.
- Chetty, Raj, and Nathaniel Hendren. 2018a. “The Impacts of Neighborhoods on Intergenerational Mobility I: Childhood Exposure Effects.” The Quarterly Journal of Economics 133 (3): 1107–62. https://doi.org/10.1093/qje/qjy007.
- ———. 2018b. “The Impacts of Neighborhoods on Intergenerational Mobility II: County-Level Estimates.” The Quarterly Journal of Economics 133 (3): 1163–1228. https://doi.org/10.1093/qje/qjy006.
- Chetty, Raj, Nathaniel Hendren, and Lawrence F. Katz. 2016. “The Effects of Exposure to Better Neighborhoods on Children: New Evidence from the Moving to Opportunity Experiment.” American Economic Review 106 (4): 855–902. https://doi.org/10.1257/aer.20150572.
- Chyn, Eric. 2018. “Moved to Opportunity: The Long-Run Effects of Public Housing Demolition on Children.” American Economic Review 108 (10): 3028–56. https://doi.org/10.1257/aer.20161352.
- Deutscher, Nathan. 2018. “Place, Jobs, Peers and the Teenage Years: Exposure Effects and Intergenerational Mobility.” Mimeo.
- Doepke, Matthias, and Fabrizio Zilibotti. 2017. “Parenting With Style: Altruism and Paternalism in Intergenerational Preference Transmission.” Econometrica 85 (5): 1331–71. https://doi.org/10.3982/ECTA14634.
- Economist, The. 2016. “1.2 Billion Opportunities.” The Economist.
- Eriksen, Jesper. 2018. “Finding the Land of Opportunity Intergenerational Mobility in Denmark.” Mimeo.
- Laliberte, Jean-William. 2018. “Long-Term Contextual Effects in Education: Schools and Neighborhoods.” Mimeo.
- Narayan, Ambar, Roy Van der Weide, Alexandru Cojocaru, Silvia Redaelli, Christoph Lakner, Daniel Gerszon Mahler, Rakesh Ramasubbaiah, and Stefan Thewissen. 2018. Fair Progress?: Economic Mobility Across Generations Around the World. World Bank Publications.
- Psacharopoulos, George. 1994. “Returns to Investment in Education: A Global Update.” World Development 22 (9): 1325–43. https://doi.org/10.1016/0305-750X(94)90007-8.
- Sacerdote, Bruce. 2002. “The Nature and Nurture of Economic Outcomes.” American Economic Review 92 (2): 344–48. https://doi.org/10.1257/000282802320191589.
- ———. 2004. “What Happens When We Randomly Assign Children to Families?” NBER Working Paper 10894. https://www.nber.org/papers/w10894.
- Young, Alwyn. 2012. “The African Growth Miracle.” Journal of Political Economy 120 (4): 696–739. https://doi.org/10.1086/668501.
Disclaimer: Opinions expressed in policy briefs and other publications are those of the authors; they do not necessarily reflect those of the FREE Network and its research institutes.