Social Protection

Bangladesh Social Protection: A Large System with Three Structural Failures

Bottom Line

Bangladesh operates 115 safety net programs at a cost of BDT 1,262 billion (2.52% of GDP), covering 30.5 million households. The system is simultaneously over-extended and under-effective. Three failures compound each other: a catastrophic targeting system that excludes 71.0% of the poor; benefit levels so low (BDT 600/month for the flagship old age allowance, roughly $5.45 per month, far below the subsistence threshold in any district) that transfers cannot produce durable poverty exits even when they reach the right households; and fragmentation across 39 ministries that multiplies administrative costs while leaving whole lifecycle categories uncovered. The National Social Security Strategy (NSSS, 2015) laid out a credible lifecycle reform framework. A decade of partial implementation has not closed these gaps. Closing them requires three decisions: consolidate delivery infrastructure (single registry, unified MIS), raise benefit adequacy to a meaningful floor indexed to inflation, and fill the structural child coverage gap with a universal child benefit.

Program Architecture: Vast but Fragmented

115 registered programs span cash transfers, food distribution, employment generation, education stipends, and disaster relief, administered by at least 39 ministries with no common beneficiary registry, no shared MIS, and no unified accountability framework. The result is a system where the same household can receive four benefits while its equally poor neighbor receives none. Total spending of BDT 1,262 billion (2.52% of GDP) is not small in absolute terms. It is, however, distributed across too many channels with too much administrative overhead to generate commensurate impact. Regional peers illustrate the range: Nepal allocates approximately 2.8% of GDP to direct transfers, India approximately 1.5%, and Sri Lanka approximately 1.2%. Bangladesh's headline figure overstates net social protection spending because it includes pension civil service and national savings instruments; the net social safety net allocation is closer to 1.0% of GDP (CPD, 2024).

The NSSS envisions consolidation into five lifecycle clusters: children, working-age, elderly, disabled, and shock-affected. After a decade, the institutional prerequisite for that consolidation, a functioning National Household Database (NHD) with interoperable program MIS, remains incomplete. Until that infrastructure exists, program rationalization is structurally blocked.

Coverage Profile: Breadth Without Adequacy

Elderly. The old age allowance (Boysoko Bhata) reaches 5.8 million beneficiaries, covering 34.0% of the population aged 60 and above. Two in three elderly Bangladeshis receive nothing. Those who do receive BDT 600/month, roughly $5.45 per month, far below the subsistence threshold in any district. The nominal amount has not been meaningfully adjusted since 2016 and has lost approximately 35-40% of its real value to cumulative inflation. Nepal's universal old age pension covers all citizens aged 68+ at NPR 4,000/month (approximately $30); the gap with Bangladesh's coverage and adequacy is structural, not incidental.

Women. The widow allowance covers 2.46 million women. The Vulnerable Group Development (VGD) program reaches 1.05 million women with food rations and development training, the best-designed graduation-linked program in the portfolio.

Disability. The disability allowance reaches 2.05 million beneficiaries, again at benefit levels calibrated for administrative cost minimization rather than need.

Education. Primary stipends reach 13.0 million students; secondary stipends cover 4.2 million more. This is the system's strongest program: evidence links stipend coverage to Bangladesh's reversal of the female enrollment gap, where girls now outnumber boys at secondary level. The conditionality design and school-linked delivery make stipends the most administratively efficient transfer in the portfolio.

Food and emergency. VGF reaches 5.0 million households in lean seasons and post-disaster; the Food Friendly Program (OMS/TCB) serves 5.0 million cardholders at below-market rice prices. Both programs are politically popular and administratively entrenched but deliver value primarily as price stabilizers rather than poverty reduction instruments.

Employment. EGPP provides 0.8 million workers per cycle with 80 days of employment in rural infrastructure at BDT 200/day. Demand exceeds supply in every cycle; rationing occurs through the same union parishad selection process that produces high exclusion errors across all programs.

Targeting: The Central Failure

Exclusion error of 71.0% and inclusion error of 46.7% are not measurement noise -- they are the predictable output of a selection mechanism built on local political discretion. At the union parishad level, elected officials allocate program slots; poverty criteria exist on paper but are not enforceable. The World Bank estimates overall leakage at 24.0% of total spending (World Bank Social Protection Public Expenditure Review, 2023). Targeting performance is catastrophic.

Proxy means testing (PMT) performs poorly when consumption variance is small near the poverty line, which is the Bangladesh case: HIES 2022 shows a large mass of households clustered just above and below the upper poverty line of 18.7%, making binary classification inherently error-prone. The statistical argument for categorical or universal coverage of high-risk groups (elderly 60+, widows, disabled persons, children under 5) is stronger than the political economy argument against it: gross fiscal costs of universality are higher, but net costs fall when targeting overhead and leakage are subtracted.

Digitization: Meaningful Progress, Incomplete Delivery

42.0% of safety net payments now flow through mobile financial services (primarily bKash and Nagad) or bank transfer. Delivery is partially digitized but majority-cash. The shift has reduced ghost beneficiaries, intermediary skimming, and payment delays -- all documented in the COVID-era emergency transfer rollout, which reached 5 million households via mobile money in weeks. The remaining 58.0% of payments still flow through manual channels, where leakage risks are highest.

The binding constraint on further digitization is not technology: Bangladesh has 180+ million mobile money accounts. The constraint is MIS interoperability. Separate program databases (DSS, MoPME, Ministry of Food) cannot exchange beneficiary data, so a single registry cannot be populated without a cross-ministry data governance agreement that has been proposed but not enacted. Until the NHD is live with mandatory program linkage, digitization advances at the payment layer without addressing the targeting layer where the core failures reside.

Lifecycle Reform: Progress and the Critical Gap

The universal pension scheme (Probash), launched August 2023, is the most structurally significant reform since the NSSS: it offers contributory pension products to all citizens aged 18-50, with government co-contribution for informal sector workers below an income threshold. Early enrollment of 0.3 million is modest relative to the informal workforce (85%+ of total employment). A voluntary contributory scheme cannot achieve meaningful coverage of informal workers without stronger incentives. The design needs either a mandatory component or a substantially larger co-contribution to make enrollment financially compelling at low income levels.

The lifecycle gap that has no program response at all is early childhood. Bangladesh allocates approximately 8% of its social protection budget to children despite children constituting over 30% of the population. Returns to investment in the first 1,000 days exceed returns at any subsequent life stage (World Bank, 2018). A universal child benefit of BDT 800/month for children under 5 would cost approximately 0.4% of GDP, achievable through reallocation from lower-impact programs and modest fiscal expansion.

BRAC's Ultra-Poor Graduation model, developed in Bangladesh and replicated in 40+ countries, produces lasting income gains 7+ years after program completion (Bandiera et al., 2017). The government's "My House, My Farm" adaptation has reached millions of households but with diluted asset transfer and coaching intensity. Restoring program fidelity would improve impact without increasing headline beneficiary counts.

Scenarios: Base Case and Risk Case

Base case (2024-2028). Probash enrollment grows to 2-3 million, G2P digitization reaches 60-65%, and the NHD is partially operationalized for 3-4 major programs. Poverty rate holds near current levels (18.7%) with modest reduction driven by GDP growth rather than safety net reform. Targeting errors improve marginally as NHD coverage expands, but exclusion error remains above 50%. Benefit adequacy continues to erode in real terms without indexation.

Risk case. Fiscal space tightens further as debt service costs rise and tax revenue growth stalls (tax-to-GDP ratio 7.5%, lowest quartile globally). Budget allocation to social protection falls in real terms. Climate shocks -- one major cyclone or extended flood season -- push extreme poverty (currently 5.6%) up by 1-2 percentage points, exceeding the shock-absorption capacity of a system with no pre-agreed horizontal expansion protocol. The demographic transition (60+ population projected at 22% of total by 2050) makes the pension gap increasingly visible politically, generating pressure for poorly designed expansions of the existing allowance rather than a structural contributory system.

Prioritized Recommendations

1. Complete the NHD and mandate single-registry linkage for all 115 programs within 24 months. Assign institutional authority to the Cabinet Division's Central Management Committee with a funded secretariat and binding data-sharing agreements. This is the prerequisite for every other reform: without it, deduplication, targeting improvement, and lifecycle tracking are impossible.

2. Index all allowances to the CPI with annual automatic adjustment. The BDT 600/month old age allowance has lost 35-40% of its real value since 2016. Indexation prevents further erosion at near-zero administrative cost. Pass a legislative floor at the same time: the old age allowance should be no less than 10% of the urban lower poverty line, rising to 20% within five years.

3. Introduce a universal child benefit (UCB) of BDT 800/month for children under 5, financed through reallocation from the least effective food subsidy components and a 0.1-0.2 percentage point GDP increase in social protection spending. Estimated cost: 0.4% of GDP. The targeting mechanism is categorical (age-based), eliminating exclusion error by construction and removing union parishad discretion from enrollment.

4. Convert EGPP into a demand-driven rural employment guarantee with a legal 80-day minimum, modeled on MGNREGA's right-to-work architecture. Link employment activities to climate adaptation: flood embankment maintenance, drainage improvement, and coastal afforestation. Counter-cyclical design means program cost rises precisely when agricultural income falls, providing automatic fiscal stabilization.

5. Establish a pre-positioned shock-responsive social protection protocol with automatic triggers (flood alert level, cyclone category, food price threshold), pre-agreed benefit expansion rules (vertical: +20% to existing beneficiaries; horizontal: 50,000 emergency enrollments per trigger event), and a 0.2% of GDP contingency reserve. The current post-disaster ad hoc distribution through local disaster management committees is slow, inequitable, and disconnected from the regular system. Forecast-based financing, already piloted by WFP in Bangladesh, provides the operational model.

Sources: Ministry of Finance Budget Documents FY2023-24; World Bank Bangladesh Social Protection Public Expenditure Review 2023; HIES 2022 (BBS); DSS Annual Report 2023; NSSS Mid-Term Review 2021; CPD Social Protection Review 2024.