Social Protection
and the
Poor in Bangladesh
Qaiser Khan, Kalanidhi Subbarao, Josh Al-Zayed
and Shaikh S. Ahmed, World Bank
Presented at Conference on What Works for the
Poorest: Knowledge, Policies and Practices
Dec 3-5, 2006, BRAC Center, Dhaka, Bangladesh
Social Protection Goes Beyond
Poverty alleviation


Social Protection includes poverty
alleviation and safety net programs but it
goes beyond to address risks and
vulnerabilities
Social protection is increasingly important
for countries like Bangladesh which are
transiting from a rural economy where
traditional safety nets work well to an
industrial economy where new safety net
tools are needed
Risk, poverty and vulnerability

Risk:
– uncertain event that may damage people’s wellbeing

Poverty:
– not having enough now of something valuable

Vulnerability:
– probability now of not having enough of
something valuable in the future
Why do we care about vulnerability?
1. Poverty  vulnerability in theory nor in practice
China
Côte d’Ivoire
Ethiopia
India
Indonesia
Pakistan
Russia
South Africa
Vietnam
Zimbabwe
1985-1990
1987-1988
1994-1997
1976/76-83/84
1997-98
1986-1991
1992-1993
1993-1998
1992/93-97/98
1992/93-1995/96
Percentage of Households who are:
Always
Sometimes
Never poor
poor
poor
6.2
47.8
46.0
25.0
22.0
53.0
24.8
30.1
45.1
21.9
65.9
12.4
8.6
19.8
71.6
3.0
55.3
41.7
12.6
30.2
57.2
22.7
31.5
45.8
28.7
32.1
39.2
10.6
59.6
29.8
 Chronic poor could be seen as the very vulnerable, temporary
poor as the vulnerable, and the non-poor as non-vulnerable to
poverty but vulnerable to income fluctuations nonetheless.
Social risk management matrix
Arrangements and
strategies
Risk reduction
Risk mitigation
-portfolio
Informal
Market-based
Public
*migration to less drought *food market
prone areas
integration
*disaster prevention
infrastructure
*drought resistant
*micro-finance
varieties; multiple activities
*research and
extension regarding
drought resistant
varieties
*mandating insurance,
including old age
pensions payments
-insurance
*marriage/family
*rainfall and commodity price based
insurance
Risk coping
*sale of assets
*consumption credits
*sending children to work
*public works
Type of Shocks – useful classifications
Examples of risky events by categories, classified by their degree of correlation
Idiosyncratic
events
Regional covariant
events
Rainfall
Landslides
Volcanic eruptions
Natural Risks
Health Risks
Life-cycle Risks
Social Risks
Illness
Injury / Accident
Disability
Birth / Maternity
Family break-up
Old-age
Death
Crime
Domestic violence
Economic Risks
Business Failure
Political Risks
Environmental Risks
Nation-wide and
international covariant
events
Earthquakes
Floods
Droughts
Strong Winds
Epidemic
Famines
Terrorism
Gangs
Unemployment
Harvest failure
Resettlement
Ethnic discrimination
Gender discrimination
Religious discrimination
Riots
Pollution
Deforestation
Civil strife
War
Social upheaval
Output collapse
Balance of payments shocks
Financial crisis
Currency crisis
Technology- or trade-induced
terms of trade shocks
Political default on social
programs
Coup d’état
Need for Social Protection
Some recent poverty thinking:
 Income distribution matters for
growth….
 Distribution of assets matters more.
 Growth helps the poor, but level of
investment is most important implication?: transfers, but not at cost
of investment.
 Transfers themselves as an investment,
in human capital – e.g. nutrition,
education
Three Constraints to Safety Nets in
Very-Low-Income Countries
 The
Information Constraint
 The
Administrative Constraint
 The
Fiscal Constraint
What is Possible? (I)
The Information Constraint




Information - not known, expensive
(e.g. pensions elsewhere)
Proxy indicators e.g. demographic,
dwelling
Link to another program (e.g. PWP,
nutrition)
3 Ways around the Information
constraint:
Self-targeting
Community targeting
Universal programs
Self-targeting
 Two
examples –
 Public workfare programs with a
wage rate slightly below the market
wage – enables participants to selfselect themselves into the program
 Food subsidies – if restricted to types
of food consumed only by the poor,
self-targeting is possible
What is Possible? (II)
Community Targeting





Communities, or representatives
identify beneficiaries (subject to
criteria), deliver benefits
Information may be better, lower cost
Risks: favoritism, difficulties of
inclusion/exclusion; divisive.
Administrative support.
Remarkably little experience Rajasthan, Malawi, Uzbekistan
Generally limited (e.g. Armenia) , jury
still out.
What is Possible? (III)
The Administrative Constraint


Weak capacity, labor-intensive,
supervisory staff
Policy implications?
– Choose simple program designs
– Simple, repetitive steps, sustain over a
long period
– Choose a few simple, nationwide
programs
– Explore possibility of using existing
administrative systems.
What is Possible? (IV)
The Fiscal Constraint
- Typically VLIC total public expenditure
spending $50-75 p.c. p.a.
- Illustratively: $20 per annum
($1.67/month) to 40% below poverty line
= 21% of public spending; not affordable.
- No way of defining ‘right’ mix, but some
programs can reach the poorest groups if
carefully designed
Program Choices

Cash Transfers - Selected, Universal
(e.g. pensions, unemployment)
 Public Works: self-targeting, create
assets, counter-cyclical but Expensive
($2/$1 transferred); critical get wage
rate right; create ‘good’ assets; costs can
be reduced if quality assets created
 Food Programs - Free Dist’n, Foodfor-Work, Food Stamps, School Feeding.
Food Subsidies/Free
food/Conditional transfers
 Agricultural Inputs - Subsidies, Free

Packs

Nutrition Programs - Child nutrition,
micro-nutrient supplementation
Problems with Food distribution
programs
Seasonal Price/Supply Smoothing
Potentially attractive in VLICs. Options: NFR, buy
surplus, release stocks; controlled food grain
price
Risks: captured by non-poor (urban consumers,
middle-men for re-sale)
 Gov’t cannot afford to intervene on sufficient
scale (high cost, ultimately ineffective)
 discourages dev’t of private markets
Prefer: Arms length interventions, make use of
trade option as did Bangladesh; buy-sell at
commercial prices, influence aggregate supply,
check if market failure exists
Conditional food/cash transfers
 Where
possible, useful to link free
transfers (either food or cash) to
behavioral change from recipients:
for example, transfers could be
conditional upon children in the
family attending the school, or
getting vaccinated etc. Known as
“conditional cash transfers”, highly
successful in Bangladesh, Mexico,
Nicaragua, Turkey and now being
implemented in several countries
Effectiveness of Safety Net
Programs in Bangladesh
Targeting Effectiveness Measured as Ratio of Recipients to group size by Income Group (2005-06)
Lowest
10%
Lowest
Quintile
2nd
Quintile
3rd
Quintile
4th
Quintile
Top
Quintile
Total
VGD
1.61
1.59
1.46
0.96
0.73
0.27
100
Test Relief
1.80
1.89
1.10
0.91
0.74
0.36
100
VGF
1.80
1.89
1.10
0.91
0.74
0.36
100
Other Targeted
1.96
1.81
1.24
1.04
0.66
0.26
100
Total Targeted
1.79
1.73
1.29
0.98
0.72
0.29
100
Selection Criteria viewed by
Recipients in Bangladesh 2005-06
Reason of inclusion in targeted programs in percent 2005-06
Poor/ Landless
Divorced/
Widows
Disabled/ Other
Others
Total
Non-Targeted
50.4
12.7
15.7
21.2
100.0
Targeted
82.3
6.3
2.8
8.6
100.0
Total
70.2
8.7
7.7
13.4
100.0
Recipient’s Perceptions of
Bangladesh Programs 2005-2006
Participants Perception about assistance from targeted programs in percent
Lowest
10
%
Lowest
Quintile
2nd
Quintile
3rd
Quintile
4th
Quintile
Top
Quintile
Total
Very Helpful
60.5
56.1
56.7
59.2
63.6
50.4
57.6
Moderately Helpful
33.8
37.3
34.1
30.9
26.5
31.5
33.3
Slightly Helpful
5.8
6.6
8.3
9.1
8.9
18.1
8.5
Not Helpful
0.0
0.0
0.9
0.9
1.1
0.0
0.5
Estimates of Leakage from the VGD, VGF and FFE
Program
2000 HIESbased
survey
estimates
(metric tons)
95%
Confidence
Interval for
estimate
VGD
99,978
VGF
FFE
70,760
49,951
Program Offtake
for FY 19992000
(metric tons)
Survey estimate
as
% of total
program
allocation
(confidence
intervals)
[72,894,
127,061]
216,675
[34%–59%]
[44,251,
97,267]
149,138
[30%–65%]
[27,192,
72,710]
285,973
[10%–25%]
Comparative Losses from Different Programs
Leakage
VGD
PESP
RMP
A. Losses due to delivery of amount less than
stipulated and other unauthorized expenses (per
taka)
0.17
0.23
0.02
B(i)…at least one targeting criteria
0.00
0.11
0.00
B(ii)…at least three targeting criteria
0.61
0.98
0.00
…A+B(i)
0.17
0.34
0.02
…A+B(ii)
0.78
1.22
0.02
B. Losses due to beneficiaries not meeting…
Total Losses….
What the data shows
 Bangladesh
targeting criteria is good
at targeting the poorest
 Bangladesh programs are perceived
by the beneficiaries as helpful
specially the poorest
 Relatively high leakage from food
based programs and despite popular
assumptions they are not better
targeted
What the data does not show
Programs are very rural focused but the
country is getting rapidly urbanized with
increasing proportion of the poor living in
urban areas
 Programs have significant administrative
leakages
 Too many programs run by too many
Government departments and thus a large
administrative overhead which can be
reduced and used to provide benefits
 Too many layers of decision making in
selection of beneficiaries

Old Age Income support in
Bangladesh- a rapid evolution
Allowance Per Beneficiary (Tk.)
250
200
150
100
50
0
1997- 1998- 1999- 2000- 2001- 2002- 2003- 2004- 2005- 200698
99
00
01
02
03
04
05
06
07
Number of Beneficiary
1,800,000
1,600,000
1,400,000
1,200,000
1,000,000
800,000
600,000
400,000
200,000
0
1997- 1998- 1999- 2000- 2001- 2002- 2003- 2004- 2005- 200698
99
00
01
02
03
04
05
06
07
Grameen Bank’s Pension Saving
Scheme Growing rapidly
Way to the future



Rationalize programs by using the one player that
gets involved in all programs i.e. union parishads.
Expand the better safety net programs such as
old age pensions, Public works and widow/
disability payments using UPs to select
beneficiaries and distribute benefits
Build on Bangladesh’s massive micro-finance
network to provide social protection for those just
above the poorest – e.g. Grameen Bank already
provides pensions and insurance (life and health
to over 6.5 million members.
Illustration of new urban poverty
and vulnerability challenge
Bangladesh
2005 - Percent enrolled by quintile and region
90.0
85.0
80.0
75.0
70.0
rural
65.0
urban
metropolitan
60.0
55.0
50.0
45.0
40.0
Low est 40%
Middle 20%
Highest 40%
rural
57.1
67.0
77.4
urban
60.6
71.9
85.5
metropolitan
42.9
57.8
80.5
Addressing the Urban Poverty
Challenge
 In
an ironic reversal of history,
enrollment rates in metro areas for
the poorest quintiles are worse than
rural areas – this represents both
past successes in rural areas and
new challenges in urban areas.
 New
approaches to urban safety nets
and safety ladders are needed
Move to ex-ante risk management
from ex-poste risk coping
Bangladesh current programs are
characterized by ex-poste risk coping
programs with the possible exception of
Grameen’s pension scheme
 New programs need to address ex-ante
risk management. They could include
expanded pensions, health insurance,
unemployment insurance, catastrophic
risk insurance i.e. floods.
