1. Social Pensions
World Bank Core Course
Robert Palacios
April 2013

2. Motivation
• Coverage in contribution-based pension schemes has
remained low for decades in developing countries
• Social pensions are increasingly proposed as a way to
address the ‘coverage gap’
• Many countries have recently introduced new schemes
or expanded existing ones and others are considering
• Some of the policy questions:
• Are they affordable and what are the tradeoffs?
• Do they create distortions for savings and labor supply?
• How do they affect the contributory scheme?
• How do they fit with social assistance? 2

3. Contributory pension coverage
3

4. Large social pension programs in 1990

5. Maldives
Mauritius
Cape Verde
Timor
Large social pension programs in 2013
Has risen from 6 to 23

6. Options for non-contributory
support to the elderly
• Universal social pension
• Targeted social pension
• Inclusion in general social assistance
• Note: Each of these options can separate
redistribution/poverty alleviation from
insurance/savings as suggested in “Averting
the Old Age Crisis”, (World Bank 1994)

7. Universal Social Pensions
• Advantages:
• Eliminates need for targeting
• Fewer issues of disincentives for labor
supply and savings, especially pension
contributions
• Political economy favorable
7

8. Universal Social Pensions
• Disadvantages
• High cost for reasonable benefit adequacy
• Kakwani and Subbarao 2004: 70% of poverty threshold
to those above 65 in 15 African countries – ranges from
0.7% of GDP in Madagascar to 2.4% in Ethiopia
• Most countries spend 0.5-2% of GDP on ALL targeted
transfers
• Projected spending can rise much higher with aging
• Given fixed budget envelope, targeting can allow
lower eligibility age and/or higher benefit
• Administration still requires key processes
including identification, enrolment and
transactions/payments (Nepal example)
8

9. 9
Some empirical evidence supports the intuition that the life
expectancy differential by income level is greater in LICs
(Bannerjee and Duflo (2005) and Pal and Palacios (2010))

10. 10

11. Targeted Social Pensions
• Advantages:
• Significantly reduces overall cost (even taking into
account targeting cost)
• At same cost, benefits can be higher or eligibility
age lower
• Disadvantages
• will result in errors of inclusion and exclusion
• Increased distortions, in particular to save for
retirement in voluntary or mandatory schemes
11

12. Integration with General Social
Assistance
• Advantages:
• Compared to separate safety net scheme, minimizes
administrative costs, avoids duplication of functions
• Likely to maximize poverty reduction impact for given
budget envelope
• In high co-residence situations, the two targeted
approaches should converge
• Disadvantages, other considerations
• Concerns over disincentives for labor supply and
savings may be different
• Re-certification/graduation issues may differ
• Intra-household distribution may not be desirable 12

13. Are elderly households poorer?
13
Some
elderly

14. Old
Not Old
Poverty Line
Even where the old are poorer than average, there are
many poor who are not old, and old who are not poor.

15. Simulated poverty impact
15
Based on HH survey data in
two regions of Zambia
(Watkins 2008)

16. Co-residence higher in LICs
16

17. Other impacts of social pensions
• Large schemes with no or limited targeting shown
to reduce elderly poverty significantly (but
question is whether that is best potential poverty
impact)
• Targeted schemes vary widely in targeting
outcomes
• There is evidence of indirect behavioral effects of
larger schemes including:
• Reduction of labor supply of coresident workers
• Reduction of private intergenerational transfers
• Permanent income increased due to investment
• Better health indicators for children in pensioner
households
• Practically all of this based on handful of studies of
Bolivia, Brazil and above all, South Africa
17

18. 18
“it has been held
that the prospect of
a pension for their
closing years will
disincline the poor
to make or continue
the exertions that
many of them make
at present for their
own support and
that the
considerations
which induce to
industry and thrift
will cease to operate
in future.”

19. Contributory pension incentives
• All of the options may lead to lower savings if
people feel that they have a minimum old age
income guaranteed – in some models, this is the
justification for the mandate (Kotlikoff 1987)
• Means-testing should discourage savings most
and this includes contributions to pension
schemes, but with low contributory coverage,
ability to ‘game’ the system will be negligible
• In medium coverage countries, design is much
more important – examples of Mexico vs Chile 19

20. 20
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
1 2 3 4 5
Quintile
%ofemployedcontributing
Uruguay
El Salvador
India
Contributory pension incentives

21. Some criteria
Introduction
or expansion
of SPs
Contributory
scheme
coverage
Social
Assistance
Poverty ratio
elderly/non-
elderly HH
Other social
indicators*
Supporting Low Limited or
non-existent
High Better
Detracting High Broad; high
spending ratio
Low Worse

22. Concluding Remarks
• Social pensions may be part of the answer to the
coverage gap in pensions but scarce resources mean
there are tradeoffs
• Large SP schemes may be redundant if there already
exist broad social assistance programs or if coverage is
high in the contributory scheme
• Universal vs targeted can be considered as a
continuum – the tradeoff is between targeting errors
and the ability to pay more to the poor. This is an
empirical question but failure to deliver benefits is not
only due to poor targeting but also the other processes
involved
• Incentive issues with contributory scheme greater to
the extent there is an overlap of households covered
which is more likely in middle income countries
22

23. Thank you