The largest welfare program in Pakistan, BISP has a stated objective of reducing gender inequalities and empowering women while combatting poverty. The transfer is specifically targeted at women. This feature is based on long documentation, both theoretical and empirical, that highlights that access to income, whether earned or through transfers, has positive effects on various women’s outcomes. The underlying mechanism at play is that such income provides women with outside options, improving their ‘threat point’ and thereby their status within the household. This could in turn tilt decision-making power in their favor allowing them to take up employment options which may have been previously denied by household heads. Besides, the extra income from cash transfers could relax credit constraints leading to greater productive investments and positive spillovers for work. Others could be better able to afford childcare and/or relief from household chores thereby freeing time for increased labor supply.
Yet, the expectation that transfer (and otherwise) income improves women’s options outside of marriage and changes household dynamics in their favor does not always hold. This is especially true when access to income does not necessarily mean control over it. In fact, the additional income from cash transfers may disincentivize working. Keeping the mixed evidence and theory in view, in a recent paper we estimate whether, and how, the BISP transfer has affected women’s labor supply.
We use the 2013-14 and 2015-16 rounds of the Pakistan Standard of Living Measurement (PSLM) surveys for our analysis. Unlike official evaluations, which have largely used Regression Discontinuity Design (RDD) methods, we use propensity score matching (PSM) techniques, which allows us to consider labor market effects for the entire intended recipient group instead of local treatment effects as in RDD. Matching is done purely on the basis of observable characteristics and may be prone to the issue of endogeneity arising from omitted variable bias. Keeping this criticism in mind, we augment our matching model to include covariates other than the poverty score which is the official criterion for BISP eligibility. Our matching yields approximately 2500 non-BISP and 700 BISP households for the 2013-14 round and about 3400 non-BISP and 800 BISP households for the 2015-16 round.
Our paper documents the effects on both labor supply as well as the sector of work. The latter allows us to more directly consider the nature of work that women do. This is relevant since not all types of work are equally empowering. Finally, we estimate effects on not just the recipient women themselves, but we also look to identify spillover effects on employment for other women in the household.
To capture effects on labor supply we consider the presence of at least one working woman within the household. We also calculate the ratio of the number of working women to the total number of women aged 10 and above in the household.
We find positive (and significant) effects on labor supply for women in beneficiary households relative to women in non-recipient households for both rounds. About 49 percent of recipient households as opposed to 44 percent in non-BISP households report at least one woman working in 2013-14 with the numbers standing at 44 and 32 percent respectively for 2015-16. Across all sectors and types of work, the ratio of working women to women aged 10 plus stands at an average of about 33 percent in BISP as opposed to an average of about 30 percent in non-BISP households in 2013-14, and at 29 and 22 percent respectively for the 2015-16 round of data. As expected though, men are significantly more likely than women to be working in both BISP and non-BISP households, with more than 90 percent of households seeing at least one male working, regardless of recipient status or round of data used.
Despite the wide gap in gendered employment, the treatment effect that we find is good news. It suggests that targeting the cash transfer to women is allowing more women to enter the labor force either because of changing incentives to work or due to changing household dynamics vis-à-vis women’s work in BISP relative to the non-BISP household.
To explore effects on the nature of work we look at sector of work and occupational skill level, average income earned, and employment status. In particular, we calculate the total number of women working in the primary sector in the house to the total number of working women in the house: similarly, for secondary and tertiary sectors. The primary sector covers agriculture, forestry, and mining. The secondary sector includes manufacturing and construction, and the tertiary sector includes services. To explore occupational skill levels, we consider the total number of women working in occupations classified as skill 1 i.e., elementary occupations to the total number of working women in the house: similarly, for higher-skill occupations. Compensation typically tends to rise as we climb occupation skill levels. Thus, skill 1 occupations are characterized by simple physical and manual tasks and are usually the lowest paid, skill 2 covers tasks such as operating machinery and equipment, skill 3 occupations have specialized skills and techniques and skill 4 occupations need complex problem-solving and creativity, and typically see the highest average wage levels.
Findings are less encouraging with regard to the nature of work that these working women are engaged in. The majority (75 percent in 2013-14 and 61 percent in 2015-16) of women are working in the primary sector, and nearly every third working woman in BISP households is working in low skilled jobs such as laborers, cleaners, vendors, etc. More than 40 percent of working women, irrespective of household recipient status, work as unpaid family help. Thus, we find that women remain concentrated in agriculture-related work earning little to no income, which then gets reflected in the very low average monthly income that we find within the data. This highlights their continued vulnerability.
At the same time, we observe greater resilience to macroeconomic shocks among recipient households from the first to the second round of the data. Between 2013-14 and 2015-16, and as shown in our paper, the agriculture sector saw a sharp decline, whereas manufacturing and services grew. We too find an overall dip in employment in 2015-16, which seems to be driven by this contraction in the agriculture sector. Yet, we also find that women in recipient households moved into the secondary sector at a rate unmatched by those in the non-BISP group. In particular, women in BISP households were more likely to be engaged as craft workers, suggesting a move towards micro-entrepreneurship. This change may explain why despite the low earnings overall, we see a rise in women’s income in the second round relative to the first.
All in all, we find that women in BISP recipient households gained relative to those in non-recipient households, both in terms of overall labor supply as well as the type of employment. Yet, women’s labor supply remained substantially below that of men, with the majority, regardless of BISP status, working in the primary sector with little to no pay. Perhaps of greater concern is the reduction in employment due to the poor growth of the agriculture sector in 2015-16 pointing to the high vulnerability of poor households in Pakistan. The existing policy narrative around employment, particularly for women, tends to focus largely on their improved participation in the labor market. Results such as those highlighted in our paper highlight that while overall supply is important to consider, we must also look at the nature of the work that women do, while examining how to build resilience to shocks. Changes to the BISP design that can cater to these issues, such as complementary arms focused on facilitating entrepreneurship and employment especially upskilling, are possible ways forward.
This post is based on Majid, H., & Riaz, S. W. (2022). Unconditional cash transfers and women’s labor supply in Pakistan. Journal of Development Effectiveness, 1-19.
Hadia Majid is Associate Professor Economics at LUMS.
Syeda Warda Riaz is a Ph.D. student at UC-Davis.