If Part 1 asked whether the Family Card is
universal or targeted, Part 2 asks a harder question:
If it is targeted — even partially — how would
eligibility actually be determined?
In public debate, targeting sounds simple:
“Give it to the poor.” In practice, identifying who is poor in a
lower-middle-income country like Bangladesh, with a large informal sector, is
one of the most complex tasks in public policy — and often the point where
well-intentioned programs falter.
Bangladesh is not an economy where most
households file verifiable income tax returns. Earnings come from daily wage
labor, small trade, agriculture, remittances, and micro-enterprises. Income
fluctuates by season and by shock. A household that appears stable today may
struggle tomorrow.
Poverty is dynamic. Any targeting system must
be dynamic too. So the real question is not whether targeting is desirable. The
real question is: what tools are available — and how accurate can they
realistically be at national scale?
Why This
Matters for Bangladesh
Bangladesh has made remarkable progress in
reducing extreme poverty. Yet vulnerability remains widespread. Millions of
households live just above the poverty line and can fall back due to illness,
job loss, climate shocks, or remittance disruption.
A targeted Family Card would need to identify
not only the currently poor, but also those at risk of becoming poor. That is
far harder than drawing a line using yesterday’s data.
Targeting accuracy will shape both cost and
credibility. If too many poor households are excluded, the program fails its
purpose. If too many non-poor households are included, resources are diluted
and public trust erodes.
How
Targeting Usually Works
No country relies on a single method. Most
systems combine approaches, each with strengths and limits.
One common method is proxy means testing
(PMT). Instead of verifying income directly — often impossible in informal
economies — PMT uses observable characteristics such as housing quality, asset
ownership, education, and family composition to estimate the likelihood that a
household is poor.
Bangladesh’s National ID system and household
databases could, in principle, support such a model.
PMT can be applied nationally using
standardized criteria and reduces reliance on local discretion. But it is
imperfect. Assets do not always reflect vulnerability. A household may own a
modest home yet face heavy medical costs. A migrant family may appear stable in
survey data while living with volatile income. And unless models are regularly
updated, they quickly lose accuracy.
All PMT systems produce two types of error:
some poor households are left out, and some non-poor households are included.
Perfect precision is unattainable. What matters is whether errors are
transparent, correctable, and kept within acceptable bounds.
Geographic targeting is another approach —
prioritizing districts or upazilas with higher poverty rates. It allows faster
rollout but can create inequities. A poor household in a relatively prosperous
district may receive nothing, while a better-off household in a poorer district
qualifies automatically. Poverty does not follow administrative boundaries, and
poverty maps can quickly become outdated.
Community-based selection adds local
knowledge. Local representatives often know which families are struggling due
to illness or recent shocks. But discretion introduces risks — favoritism,
capture, inconsistent standards. Digitizing such lists does not eliminate bias
if the underlying process is flawed.
For this reason, effective systems layer
methods: statistical screening to generate an initial list, local validation to
correct omissions, and grievance mechanisms to address mistakes. Targeting is
not a formula. It is an institutional system.
What the Pilot
Reveals
The announced pilot spans selected upazilas across
multiple divisions, covering areas described as high-poverty,
climate-vulnerable, urban, and remote. That selection already reflects
geographic prioritisation.
However, within those pilot areas, the transfer is not
being extended to all households. Instead, beneficiaries are to be selected
from among households identified as poor or vulnerable. The specific
eligibility criteria have not yet been detailed publicly.
This structure indicates that the pilot is testing a
layered model: first selecting areas, then identifying eligible households
within those areas. It is therefore examining identification mechanisms as much
as delivery logistics.
That distinction is important. A pilot designed to
test a universal entitlement would typically provide the benefit to all
households within selected locations, focusing mainly on payment systems and
administrative coordination. A pilot that screens households is, by design,
testing targeting systems.
This does not imply that targeting is inappropriate.
Many successful social transfer programs began with narrower coverage and
expanded gradually as administrative systems strengthened.
What matters is clarity. If the long-term intention is
universal coverage, the transition path should be explained. If the long-term
model is targeted, then eligibility rules and updating mechanisms must be
clearly articulated.
Ambiguity about what the pilot represents — a test of
logistics or a test of targeting — risks creating confusion about the program’s
eventual design.
Capacity
Matters More Than Formulas
Designing targeting rules is easier than
implementing them across millions of households.
Administrative capacity requires updated data,
functioning appeals systems, verification audits, coordination across agencies,
and the ability to detect manipulation. Bangladesh has important strengths —
high national ID coverage and widespread mobile financial services — that
provide a solid foundation.
But constraints remain. Digital literacy gaps,
connectivity challenges, delays in updating data, and limited grievance redress
capacity all affect how targeting works in practice.
A credible targeted program requires more than
an algorithm. It requires a transparent correction system. If households
perceive the process as opaque or arbitrary, trust erodes quickly — and
restoring trust is far harder than building it.
Poverty
Moves. Lists Must Move Too.
Poverty is not static. Households fall into
hardship due to illness, job loss, climate shocks, remittance declines, or
business failure. Others move out of poverty through new opportunities. A
static beneficiary list quickly becomes outdated.
Here lies a structural difference between
universal and targeted models. A universal program automatically covers
households when shocks occur. A targeted program must continually update
eligibility to remain fair. The more dynamic the economy, the heavier the
administrative burden.
There is no perfect solution — only trade-offs
between precision, cost, and timeliness.
Technology
Helps — But It Is Not the Solution
Digital transfers reduce leakage and increase
traceability. Direct transfers to mobile wallets such as bKash or Nagad, or to
bank accounts, are a major improvement over cash distribution.
But technology does not eliminate governance
challenges. Incorrect data, identity mismatches, account control within
households, and overwhelmed grievance systems can still undermine trust. If the
beneficiary list is wrong, a flawless payment system simply pays the wrong
person efficiently.
Digital infrastructure strengthens targeting.
Institutional integrity sustains it.
The Core
Trade-Off
Targeting promises precision. Universality
promises simplicity.
Precision requires strong institutions,
regular data updates, and political discipline. Simplicity requires fiscal
space.
The choice is not ideological. It is
institutional.
If the Family Card is targeted or hybrid, its
long-term success will depend less on the Tk 2,500 amount and more on the
credibility of the identification system.
Cash matters. But institutional trust matters
more.
In Part 3, we move from identification to
scale — examining what different coverage paths would mean for Bangladesh’s
overall budget and the trade-offs they imply for health, education,
infrastructure, and climate resilience.