The role of public versus private social safety nets in mitigating the impact of shocks in rural Pakistan

The study addresses whether social safety nets (SSNs) in Pakistan protect vulnerable rural households from employing detrimental shock-coping strategies after adverse events, and whether public (government) SSNs are more effective than private (NGO) SSNs. SSNs are transfers (cash or in-kind) intended to maintain minimum well-being, food security, and risk management. Pakistan’s policy focus on SSNs increased after the 2008 food price shock and the launch of BISP, yet rigorous impact evidence—particularly for rural areas where poverty is concentrated—remains limited. Using PRPHS 2011–12 data collected soon after nationwide flooding, the study investigates the protective role of SSNs against reducing food consumption, switching to cheaper food, and distress sales of productive assets. The purpose is to provide comparative, evidence-based insights to inform the design and implementation of public and private SSNs in rural Pakistan.

Global evidence on SSNs is mixed but generally shows potential benefits when programs are well designed and consistently delivered. In developing countries, SSNs range from formal transfers to informal mechanisms (e.g., Hawala), with increased adoption since the 2007–08 food price crisis. Studies using rigorous methods (e.g., PSM, RDD) report: in Bangladesh, standalone SSNs often show limited protection against harmful coping but combinations can be more effective; in Ethiopia, PSNP increased food security; in Argentina, crisis-era income support reduced unemployment modestly; in South Africa, unconditional grants improved child nutrition and later labor outcomes; meta-analyses find both UCTs and CCTs increase school enrollment. In Pakistan, evaluations of Zakat and BISP show targeting challenges and often insignificant impacts on consumption and welfare indicators, though some studies find improvements in specific contexts. The literature lacks a comparative assessment of public versus private SSNs in rural Pakistan. The PRPHS dataset, with explicit modules on shocks, coping, and SSNs, provides an opportunity to fill this gap.

Design: Quasi-experimental impact evaluation using Propensity Score Matching (PSM) to estimate Average Treatment Effect on the Treated (ATT). Given cross-sectional data, PSM approximates a counterfactual by matching treated households to observationally similar untreated households. Data: Pakistan Rural Panel Household Survey (PRPHS) 2011–12 (Round 1) by IFPRI; 2,124 rural households from Punjab, Sindh, and Khyber Pakhtunkhwa. The survey includes household questionnaires (male and female respondents), with modules on shocks experienced and coping strategies, plus SSN receipt in the prior 12 months. Interventions: Two treatment definitions assessed separately: (1) Public SSNs (government)—BISP (UCT), Watan Card (flood cash), Zakat and Ushr, Bait-ul-Maal, pensions/retirement benefits, education stipends, madrassa/school feeding; (2) Private SSNs (NGOs)—services from organizations such as PPAF, NRSP, SRSP, Aurat Foundation, Edhi Foundation, INTERSOS, Save the Children, Plan International, NCHD, PIMN, Thardeep (healthcare, DRM, community development, training, microfinance). NGO programs often collaborate with government. Households receiving both types were excluded; those receiving informal assistance (community/relatives/Fitrana) were excluded due to unobservable determinants; only households reporting a shock were retained. Outcomes: Dummies for whether the household, in response to a shock, (i) reduced food consumption, (ii) switched to cheaper food, (iii) engaged in distress sales of productive assets. Effective SSNs are hypothesized to reduce the probability of these strategies (negative, significant ATT). PSM Specification: Binary logistic models estimated on pooled samples to predict treatment probability (propensity score). Covariates guided by program eligibility and prior studies, reflecting household welfare and demographics. Public SSN model covariates included: house ownership; electricity connection; attempted to obtain a loan; age and age squared of household head; gender of household head; household size; province dummies (Punjab, Sindh; KPK reference); education of household head (years); pucca house; numbers of large and small animals; migration. Private SSN model included similar covariates (migration not reported in final private model). Nearest Neighbour Matching with replacement; random sort order; common support enforced. Balancing assessed via standardized bias reduction, Rubin’s B and R, pseudo-R2, and LR tests. Diagnostics and Sample: Public SSN sample mean propensity score 0.242 (N=1,403); on common support, treated N=328, controls N=1,022 (5 treated off-support dropped). Balancing achieved with substantial bias reduction; Rubin’s B=24.4 and R=1.26 within recommended ranges; joint significance of covariates in treatment model p<0.001. Private SSN sample mean propensity score 0.084 (N=1,162); on common support, treated N=95, controls N=1,022 (3 treated off-support dropped). Balancing largely adequate; Rubin’s B=27.7 slightly above threshold but LR test indicates joint bias insignificant (p=0.979). ATT computed post-matching using psmatch2.

- ATT estimates for coping strategies (t-statistics in parentheses): • Public SSNs: Reducing food consumption ATT = -0.0396 (t=-1.31, ns); Switching to cheaper food ATT = -0.0915 (t=-2.09, p<0.05, significant); Distress sales of productive assets ATT = -0.0213 (t=-0.75, ns). • Private SSNs: Reducing food consumption ATT = 0.0421 (t=0.64, ns); Switching to cheaper food ATT = -0.1158 (t=-1.43, ns); Distress sales ATT = -0.0105 (t=-0.28, ns). - Overall, receipt of either type of SSN did not significantly protect households from employing most detrimental coping strategies in response to shocks. - Public SSNs showed a statistically significant protective effect specifically on the likelihood of switching to cheaper food, suggesting better maintenance of food quality among treated households, potentially due to in-kind food assistance (e.g., wheat distribution, school feeding). - Private SSNs did not show significant protective effects on the examined short-term coping outcomes; in one case (reducing food consumption) the ATT was positive but insignificant.

The analysis indicates that, for rural Pakistani households facing shocks, SSNs generally did not prevent the use of detrimental coping strategies, suggesting limited short-term protective effectiveness. The exception is public SSNs, which significantly reduced switching to cheaper food, likely reflecting in-kind food transfers and the use of cash transfers predominantly for food. These findings answer the research question by showing that while both public and private SSNs underperformed on most immediate coping outcomes, public programs provided modest protection for food quality. Potential reasons for weak effects include small transfer sizes (e.g., BISP monthly amount), delivery inefficiencies (irregularity, leakage, duplication, red tape), lack of coordination among providers, and program politicization. Private SSN impacts may materialize in longer-term outcomes (e.g., skills, livelihood changes) not captured by the selected short-term coping indicators. The results underscore the importance of program design, delivery regularity, and alignment of benefit type (cash vs in-kind) with pressing household needs during shocks.

The study contributes a comparative, evidence-based assessment of public versus private SSNs in rural Pakistan using PRPHS 2011–12 and PSM. It finds largely insignificant impacts of both SSN types on harmful coping strategies, with a notable exception: public SSNs significantly reduced switching to cheaper food, indicating better protection of food quality for beneficiaries. Policy recommendations include: considering in-kind food transfers alongside cash to bolster food security during shocks; conducting rigorous needs assessments and leveraging local social risk management capacities; enhancing coordination and synergy between government and NGOs in targeting, program design, and service delivery; and instituting strong, continuous monitoring and evaluation systems. In the broader disaster-risk context, strengthening national disaster management (NDMA) implementation capacity, early warning systems, and infrastructure (dams, levees, floodwalls) and exploring social insurance for high-risk areas can mitigate shock impacts. Future research should examine longer-term outcomes (assets, livelihoods, risk-taking) and the combined effects of multiple SSNs.

- Data and timing: PRPHS Round 1 (2011–12) queried shocks over the prior five years; the analysis assumes sufficient overlap between shock timing and SSN receipt in the prior 12 months, supported by widespread flood-related shocks roughly 12 months before the survey. - Coverage: Balochistan was excluded due to security constraints; results generalize only to rural Punjab, Sindh, and Khyber Pakhtunkhwa. - Design: Cross-sectional PSM cannot fully eliminate unobserved selection; causal inference is limited compared to panel DD or RDD. - Treatment definition: Households receiving both public and private SSNs, and those receiving informal assistance (community/relatives/Fitrana), were excluded; effects of multi-program participation and informal protection are not assessed. - Covariates: Some welfare indicators (income, employment, consumption expenditure) were excluded due to volatility and price differentials, potentially omitting relevant predictors. - Sample size: Few SSN recipients reduced the ability to analyze individual programs and may limit statistical power. - Measurement: Lack of data on the age of the specific recipient within the household for private SSNs; outcomes focus on short-term coping and may miss longer-term effects.