Gabriel Leite Mariante is a PhD candidate in Economics at the LSE. He works in Development Economics, focussing on social protection, labour markets and gender inequality in low- and middle-income countries.
Cash transfers are the world’s most common anti-poverty policy in low- and middle-income countries. A key question for design and evaluation of these policies is: what is their impact on recipients’ employment decisions?
Economic theory suggests it is ambiguous. Under the neo-classical model, where people like leisure and dislike working, it is negative: unearned cash will decrease your propensity to work (Vivalt et al, 2024). However, people might face constraints keeping them out of employment, which a transfer can help overcome. Such constraints could be: childcare (Berlinski et al, 2024), the need for an initial investment (Balboni et al, 2022), or costs of job search (Caria et al, 2024). If this is the case, then a transfer can increase employment. Empirically, it is also ambiguous: recent meta-analyses have found both positive and negative effects across different contexts (Crosta et al, 2024; Diaz-Pardo and Rao, 2024).
This ambiguity suggests that the answer is: it depends. In my job market paper, I look at what mechanisms, policies and contextual factors can shape this effect for men and women in Brazil.
Who? Women work more, men don’t
The Brazilian government implements the world’s largest cash transfer programme: Bolsa Familia. It has many components, the largest of which is an unconditional transfer aimed at households in extreme poverty that increases recipients’ income by around 40%. The unconditional transfer is targeted at all households considered by the government to be living in extreme poverty, defined as having an income per capita below a specified threshold.
In 2014, the government enacted a reform increasing this threshold of extreme poverty, hence increasing the benefit coverage. I use this rule change to create a “treatment” and a “control” group. The treatment group consists of people who, before the reform announcement, had an income just below the future unannounced threshold, and hence became eligible for the benefit when the reform kicked in. The control group consists of people who just missed out by being a little bit above the threshold.
I look at the difference in employment between treatment and control groups around the time of the reform. I find thatwomen who receive the benefit are 7% more likely to be formally employed over two years, compared to women who don’t. Men, on the other hand, see no significant effect.
Why? Spending in education frees up time dedicated to care work
When exploring relevant dimensions of heterogeneity, I show that the effect on women is concentrated on mothers with kids between the ages of 3 and 6. Turning to data on the children themselves, I also see that the transfer increases enrollment in education of children of the same age group (3 to 6), but has no effect on the others.
I complement the analysis with data from two household surveys: first, the 2014 to 2016 waves of the National Household Survey (PNAD), that includes a labour force module that asks respondents who are not employed if they “would be available to work if offered a job”. Those who answer “no” are then asked the reason why. The survey shows that women are about three times more likely to report being unavailable to work than men, a difference entirely driven by the reason “has caring responsibilities to another household member”. This is illustrated in Figure 2 below. When measuring the effect of the transfer, I find that recipients are around 9% less likely to report being unavailable to work than non-recipients.
Second, I use data from the 2009 Household Budget Survey (POF – a nationally representative survey on household expenditure) to look at differences in spending patterns between recipients and non-recipients of the cash transfer. I find that recipient households are 22% more likely to have any expenditure related to children’s education. This difference is driven by spending in complementary costs to schooling, such as school material, uniforms and after-school activities.
Overall, the evidence shows that women are using the benefit to increase their propensity to enrol their young children in education, and freeing up time previously dedicated to care work at home.
Where? Effects are higher in poorer areas with better provision of education
This mechanism suggests a potential role for local economic factors, as well as for complementary policies that ease access to schooling.
To investigate this, I take advantage of the fact that Bolsa Familia is a centralised programme. It is implemented with the same design across a vastly diverse and unequal country, which allows me to focus on the role of local context and keep programme design constant.
I separately estimate the effect of the transfer on employment across Brazil’s 557 micro-regions (the usual definition of a local market – almost everyone lives and works within the same micro-region). Using Bayesian techniques to aggregate evidence from different contexts (Meager, 2019; Angrist and Meager, 2023), I show that the effect is larger in areas that are poorer (i.e. lower GDP per capita), but that have more local provision of public goods in education and health.
This suggests that the provision of local public services by the government allows constraints to employment to be relaxed. To reinforce this point, I take advantage of the rules governing the allocation of public funds across Brazilian municipalities. These rules generate changes in municipal education budgets around arbitrary population thresholds that are as good as random (Corbi et al, 2019). I compare the effect of the transfer on employment in municipalities that exogenously receive more funds – a share of which is earmarked for education – with those who receive less funds. I find that the effect of the transfer on employment in municipalities that receive more funds is more than twice the national average. In those which receive less funds, it is not significantly different from zero.
Summing up
Overall, my empirical evidence has two main findings. First, a large unconditional cash transfer helps women send children to school, and hence helps overcome care work constraints that were keeping them out of employment. Second, there is a role for policy in complementing this effect via provision of local services, particularly education, which enable the relaxation of women’s constraints to employment.
Feature Image shared by the author, sourced from Google Images.
