Impact of employment on wellbeing: A gendered analysis from Cox’s Bazar Refugee Camps

Employment (paid work) can positively affect the financial stability and physical and mental health of the individual who is employed. There is also a large body of sociological evidence on how individual employment can positively affect others in the household. Yet, quantifying this potential impact has been understudied. We attempt to help fill the gap by measuring how an individual's employment affects their own wellbeing and that of their spouse and whether those effects differ depending on the gender of the employed. The question is particularly relevant as policymakers are increasingly looking at how best to target the provision of employment opportunities. 

Studying the impact of paid or unpaid part-time employment and unconditional cash transfers on wellbeing

Our study takes place in the refugee camps in Cox’s Bazar, Bangladesh. Close to 1 million refugees are confined in what has become the largest refugee encampments in the world. Life in the camps is grim. Refugees live in overcrowded areas, access to education is limited, and the threat of physical violence is ever-present. Armed groups and gangs pose serious security threats, leaving women and girls particularly vulnerable. Humanitarian aid is stretched thin. 

Our study took place over 6 weeks and involved 2,513 married couples. These were randomised into a control arm (that received no treatment intervention) and three treatment arms. The three treatment arms included the provision of part-time employment, unconditional cash transfer, and part-time employment (same as the first treatment arm) but without pay, respectively. Only one spouse (female or male) was engaged per couple, and the selection of the spouse who received the treatment was randomised. We interviewed both the treated individual (who received part-time employment (paid or unpaid) or unconditional cash transfer) and their spouse.

The treatments worked as intended. The impact on savings, debt reduction, leisure time, and productive activity varied across the treatment groups as expected. Employment and cash provision significantly increased savings and diminished debt irrespective of the participating spouse. Employed individuals and those engaged in unpaid work experienced less leisure time yet increased productive hours. 

Gendered impact of work and cash on wellbeing

We set out the seven main findings below. From this point, when we refer to an individual being “employed”, we mean they were assigned to the part-time employment arm. An individual who is “treated” was assigned to part-time employment, received cash, or worked without pay.

1. Employment provides similar psychosocial gains for women and men: Employed men and women experienced the largest gains across the treatment groups. They gained 0.11 standard deviation and 0.08 standard deviation, respectively, in their psychosocial index. This is an inverse covariance weighted index of depression severity, stress, life satisfaction, sociability, purposefulness, self-worth, locus of control, and stability.
2. Employment provides significant psychosocial benefits beyond those of cash for men. While women benefit four times more from work than cash, the difference is not statistically significant. We do not find that volunteering (part-time employment without pay) produces any psychosocial benefits.

These results suggest that employment is a powerful intervention, perhaps even more so for men, potentially because of the dynamics it creates within the household. We next examine the effects on the spouse of the individual engaged in employment, unconditional cash transfer, or unpaid part-time work. 

3. Husbands experience no detectable change in their psychological wellbeing when their wives are employed. In fact, there is no evidence that men are happier or upset when their wives receive any of the treatments.
4. Wives experience psychosocial gains when their husbands receive employment or cash. These gains are statistically equivalent to the benefits their employed husband derives and the gains experienced by other women in our sample who are treated.

These findings suggest that treatment impacts partner outcomes differentially depending on the gender of the person receiving the treatment. Given this evidence, we focus next on the relational health between partners in our sample. More specifically, we focus on two measures. The first is intimate partner violence (IPV), an inverse covariance weighted index summarising beliefs about and experience of verbal and physical abuse from one’s partner. The second is household bargaining. Taking each in turn, we find that: 

5. Women who work experienced no detectable changes in IPV. We also cannot reject a null effect for any of the other two treatment arms. In other words, we find no evidence of backlash.
6. However, we see very large and statistically significant reductions in IPV among the wives of men who work. The reduction in the IPV index experienced by the wives of employed husbands amounts to a 31% reduction in the likelihood of being physically threatened by their husbands over the previous month. We observe similar declines in IPV for the wives of men receiving unconditional cash or unpaid work. These reductions in IPV could help explain why women experience higher psychosocial wellbeing when their husbands are treated. What causes these reductions in IPV we observe when men are treated? While we cannot identify the exact mechanism, it could be because men spend less time at home or because treatment alleviates financial constraints. Alternatively, men may reduce IPV because they are less depressed or because their wives, who exhibit improved psychosocial wellbeing become more deferential, which makes them less abusive.

Finally, we consider the distinction between wellbeing and agency. It is possible that our treatments re-distributed power within the household in their favour, even if it does not translate to immediately improved wellbeing.

7. We do not see that women who are employed experience greater agency and empowerment. Using both revealed preference and self-reported measures of intra-household bargaining, we find no systematic evidence that women who are employed (or receive unconditional cash, or unpaid work) experience gains in their bargaining power. We also did not find that treated men exhibit a retrenchment of their power in the household, such that wives of employed husbands experience similar levels of empowerment.  

Broadening the scope of study of social protection programmes

Our findings confirm that employment opportunities impact both the individual and their household. Although labour supply decisions are often modelled at the individual level, our results demonstrate how such decisions also impact other household members. 

From a policy standpoint, our results underscore the importance of measuring the impact of social protection programmes on margins beyond material consumption (extending to psychosocial wellbeing, IPV, and bargaining power) and effects beyond the individuals (extending to one's partner, children, and beyond). Social protection programmes can have a wide range of impacts, and they can affect the intended participant and others in the household differently. 

While we find that employment provides meaningful and comparable psychosocial benefits to both women and men, there are additional factors to consider. We find that women are happier and safer when their husbands are employed. We also find that programmes like ours (that are similar to other social protection programmes like NREGA, cash transfers, among others) may not be able to redistribute power within households over shorter time horizons without direct norms engagement.

To date, many social protection policies such as cash transfer programmes, job training programmes, or cash-for-work programmes aim to be gender inclusive despite scarce evidence of the means through which female and male participants may derive value from the intervention. For example, a programme that is intended mostly for women may not necessarily advance the social protection outcomes as effectively as intended. Our study points to the need for more evidence-based approaches. We hope our findings will help in the design of social protection programmes. We also hope it encourages further efforts to probe their underlying value to individual and household wellbeing among vulnerable and low-resource populations.