The impact of unconditional cash transfers on enhancing household wellbeing in Pakistan: evidence from a quasi-experimental design

Cash transfers are widely used to reduce poverty and improve wellbeing in developing countries by providing direct monetary assistance that can meet immediate needs and enable longer-term investments in education, health, and productivity. They can also empower women and improve nutrition and schooling outcomes. Despite major progress, challenges remain: widespread food insecurity, child stunting, and out-of-school children persist globally and in Pakistan. In Pakistan, poverty has fallen over two decades but remains dynamic and uneven across regions, with high levels of food insecurity and malnutrition. Social safety nets (SSNs) aim to protect the poor, promote resilience, equity, and opportunities, and often include objectives related to women’s empowerment and human capital. Effectiveness of SSNs varies with targeting, coverage, and adequacy of benefits. In Pakistan, the Benazir Income Support Programme (BISP), launched in 2008 and recognized for its targeting and coverage, provides unconditional cash transfers (UCTs) to poor households (primarily to ever-married women), and conditional top-ups for schooling. Initially beneficiary identification involved parliamentarians, later replaced by a Proxy Means Test (PMT) via the National Socioeconomic Registry. Spending on SSNs increased substantially with BISP. Beyond routine UCTs and CCTs, funds have supported education initiatives, scholarships, and emergency relief, with a large budget approved for 2023–24 across multiple BISP sub-programs. The main objective of this study is to construct a socioeconomic wellbeing index and use it to evaluate the impact of BISP’s unconditional cash transfers on household wellbeing in Pakistan, including domains such as women’s empowerment, financial stability, living standards, and material resources, considering household and beneficiary characteristics. The study addresses a gap by integrating subjective and objective indicators into a composite socioeconomic wellbeing index and assessing differential impacts across domains and population subgroups, with implications for policy design, value-for-money, coverage, sustainability, and program impact.

Across developing and developed countries, 72–149 SSN/SA programs have been launched to alleviate poverty and protect against shocks. Global reviews report increasing commitment and contributions to poverty reduction but note persistent issues with low coverage and benefit adequacy. Targeting approaches include PMT, community-based selection, and self-targeting, each with trade-offs, including risks of exclusion/inclusion errors, rent-seeking, and reliance on limited indicators. Evidence is mixed: some programs are pro-poor but suffer from under-coverage; NGO delivery varies; local politics and corruption can hamper effectiveness. In Pakistan, BISP targets chronic and transient poor via a national poverty scorecard, primarily transferring to women. Studies report varied outcomes: improvements in women’s mobility and some living condition deprivations; mixed or null impacts on poverty reduction, child nutrition, and productive investment; and concerns over targeting performance and coverage. Some analyses suggest improvements in empowerment and schooling outcomes, while others find high exclusion and inclusion errors and limited effects on savings, indebtedness, food security, or child welfare. International comparisons show that program effects depend on transfer size, frequency, and design; smaller or irregular transfers often have limited impacts. The literature underscores the need for comprehensive measures of wellbeing beyond monetary indicators and robust evaluation designs. This study contributes by applying a holistic SES index and a quasi-experimental DiD approach to assess BISP’s UCT effects.

Design: Quasi-experimental Difference-in-Differences (DiD) using panel data from three waves of the BISP Impact Evaluation Survey (2011 baseline; follow-ups in 2016 and 2019) conducted by Oxford Policy Management. The 2011 baseline identified eligible households (PMT threshold 16.17) as treatment and ineligible (above threshold) as controls. The analysis uses a balanced panel of 5,265 households tracked across all three waves (treatment: 3,969; control: 1,296). Data: Household- and individual-level indicators were drawn to construct a composite socioeconomic status (SES) index, the primary outcome. Covariates included household remittances and employment status; DiD regressors captured time effects, group effects, and their interaction. SES construction: Principal Component Analysis (PCA) on standardized indicators (correlation-matrix approach due to differing units). The first principal component (PC1) provided weights; standardized scores per indicator were multiplied by PC1 loadings and summed by domain to obtain domain scores; domain scores were then summed for household SES. Indicators (15) across four domains: - Material sources: ownership of agricultural land (rural), livestock (rural), and household assets (e.g., refrigerator, AC, fan, geyser, washing machine). - Living standard: access to electricity, clean drinking water, toilet facilities, non-dirty roof. - Financial hardship: presence of cash savings; no household member took a loan in last 12 months. - Women’s empowerment/mobility: ability to go to local market, visit health facilities/doctors, visit friends’ homes, visit shrines/mosques, and vote without restrictions. Indicators coded 1 if non-deprived, 0 otherwise. DiD model: Y = β0 + β1[Time] + β2[Intervention] + β3[Time×Intervention] + β4[Covariates] + ε. Key assumptions: allocation not based on baseline outcomes; parallel trends; stable composition in repeated panels; no spillovers. Estimation covered three intervals: 2011–2016, 2016–2019, and 2011–2019. Sample sizes by wave: 1,755 panel observations per wave (control 432; treatment 1,323) totaling 5,265 panel observations. Statistical inference flagged 1% significance where applicable. The study did not apply propensity score matching due to small control group size.

- SES distributions widened over time: PCA score range expanded from approximately −2 to 2 (2011) to −5 to 5 (2019). - Overall SES means (Table 4): Beneficiaries: 57.1 (2011), 48.3 (2016), 47.9 (2019) – net decline of 9.2 points (2011–2019). Non-beneficiaries: 58.8 (2011), 49.7 (2016), 48.7 (2019) – net decline of 10.1 points. - Urban–rural (beneficiaries): Urban SES: 60.6 (2011) → 57.7 (2016) → 54.9 (2019) [−5.7 overall]; Rural SES: 56.2 (2011) → 46.0 (2016) → 46.1 (2019) [−10.1 overall]. Urban non-beneficiaries fell by 9.1 points; rural non-beneficiaries by 10.7 points. - Quintiles: Larger SES reductions in lower quintiles than upper ones between 2011 and 2019. For non-beneficiaries, average SES declines: Q1 −14.4, Q2 −12.7, Q3 −9.8, Q4 −2.7. For beneficiaries, declines were relatively smaller: Q1 −12.2, Q2 −12.8, Q3 −9.6, Q4 −2.5, suggesting slightly better preservation of wellbeing among lower-quintile beneficiaries compared to similar non-beneficiaries. - Domain-wise trends (2011→2019): • Material sources: slight improvement to 2016, then decline by 2019. Non-beneficiaries: ~0.20→0.22→0.18; beneficiaries: ~0.19→0.21→0.17 (beneficiaries’ reduction marginally smaller). • Living standard: improved for both groups: non-beneficiaries ~0.53→0.60→0.63; beneficiaries ~0.50→0.60→0.63. • Financial hardship wellbeing: modest declines over time; 2011 values ~0.62 (non-beneficiaries) and 0.61 (beneficiaries); by 2019 decreased roughly 5 points (non-beneficiaries) and 4 points (beneficiaries). • Women’s empowerment/mobility: remained high; non-beneficiaries ~0.90 (2011), 0.90 (2016), 0.83 (2019); beneficiaries ~0.89 (2011), 0.91 (2016), 0.89 (2019), indicating better maintenance among beneficiaries by 2019. - Difference-in-Differences (Table 5, dependent variable: SES index): • 2011–2016: Time −9.38 (p<0.01); Treatment −2.12 (p<0.01); DiD +0.16 (ns); Household size +0.51 (p<0.01); Employment status +4.35 (p<0.01); Remittances +1.67 (p<0.01). • 2016–2019: Time −0.26 (ns); Treatment −2.24 (p<0.01); DiD +0.68 (ns); Household size +0.79 (ns); Employment status +5.64 (p<0.01); Remittances +3.97 (ns). • 2011–2019: Time −9.61 (p<0.01); Treatment −2.23 (p<0.01); DiD +0.43 (ns); Household size +0.69 (p<0.01); Employment status +4.93 (p<0.01); Remittances +2.59 (p<0.01). - Interpretation: SES declined significantly over 2011–2016 and 2011–2019 across both groups. The DiD interaction (average treatment effect on the treated) was positive but statistically insignificant in all periods, indicating limited measurable impact of UCTs on SES. Employment is strongly and consistently associated with higher SES; remittances also show positive associations (significant in two of three periods). - Overall: BISP UCTs slightly mitigated declines relative to controls in some subgroups/domains but did not produce statistically significant improvements in the composite SES.

The study set out to assess whether BISP’s unconditional cash transfers improved household socioeconomic wellbeing using a multidimensional SES index and a quasi-experimental DiD design. Findings indicate broad declines in SES between 2011 and 2019 for both beneficiary and non-beneficiary households, with beneficiaries experiencing slightly smaller declines in some domains and lower quintiles. However, the DiD estimates were consistently positive yet statistically insignificant, implying limited causal impact of UCTs on overall SES. The results align with mixed evidence from prior BISP evaluations and international literature suggesting that the effectiveness of cash transfers depends on transfer size, real value, frequency, program coverage, and implementation quality. Potential reasons for limited impact include erosion of real transfer value due to inflation and economic slowdown, irregular payment schedules, and insufficient funding relative to household needs and frequency gaps. At the same time, specific domains such as women’s empowerment/mobility and living standards show relative improvements or maintenance among beneficiaries, suggesting domain-specific benefits. Employment and remittances emerged as key correlates of SES, underscoring the importance of labor market opportunities and migrant/family support in household welfare trajectories. Overall, while UCTs may provide short-term cushioning, their standalone effect appears insufficient to shift multidimensional wellbeing without complementary interventions that enhance income generation, asset accumulation, and resilience.

This study constructed a multidimensional SES index and applied a DiD quasi-experimental design to three waves of BISP impact evaluation data (2011, 2016, 2019) to assess the effect of UCTs on household wellbeing in Pakistan. Across the period, SES declined for both treatment and control households; DiD estimates were positive but insignificant, indicating limited measurable impact of UCTs on overall SES. Nevertheless, beneficiaries showed relatively better maintenance in certain domains (e.g., women’s empowerment) and among lower quintiles compared to non-beneficiaries. The study highlights inflation-driven erosion of real transfer value, irregular disbursements, and funding adequacy as likely contributors to limited impacts. Policy recommendations include: shifting part of expenditures toward income generation, capital asset development, microbusiness support, and climate-resilient agriculture/livestock; promoting inclusive growth policies to reduce gender and employment disparities; leveraging remittances through facilitative policies; and instituting rigorous, continuous program evaluation. Future research should integrate higher-frequency administrative and survey data, examine heterogeneity by transfer size/frequency, and evaluate combined interventions (UCT plus livelihoods/skill-building/insurance) to determine effective pathways to sustained improvements in multidimensional wellbeing.

- Data constraints: No impact evaluation data beyond 2019; reliance on three waves limits assessment of more recent program changes. - Design constraints: Quasi-experimental DiD relies on parallel trends and absence of spillovers; pure experimental assumptions not met. - Small control group: The relatively small control sample precluded the use of propensity score matching and raises concerns about baseline imbalances and power. - Measurement: SES index construction depends on selected domains/indicators and PCA weights; alternative specifications might yield different sensitivities. - Program implementation factors: Potential irregular payment intervals, funding adequacy, and inflation may confound realized impacts; data on timing and real value of transfers at household level were limited. - External validity: Findings pertain to BISP’s design and period studied; generalizability to other settings or later program phases may be limited.