Transcript 11 Robert Palacios_Engl

Design alternatives for
social pension programs
Robert Palacios, World Bank
International Conference – “Funded Systems: Their Role in
Solving the Pension Problem”
Varna, Bulgaria
May 31-June 1, 2007
Contents

The ‘coverage gap’ motivating social
pension policies

The role of social pensions today

Design and future role of social pensions
2
The ‘coverage gap’
 Almost all countries mandate pension coverage for
formal sector workers
 But after operating for decades, coverage of these
schemes has remained stagnant and for the
poorest, it may not make sense (James (1999)).
 There is little evidence that coverage is significantly
affected by type of scheme or whether it is funded
or not
 There is, however, a strong correlation to income
per capita which is a proxy for many factors
3
The long run ‘coverage gap’
The global coverage/income pattern suggests
that it will take a long time for coverage rates
to rise in developing countries
Fitted line including
TSE dummy
20000
90%
18000
80%
members/labor force
PPP Income per capita
100%
16000
14000
12000
10000
8000
6000
70%
60%
50%
40%
30%
4000
20%
2000
10%
0
0%
0%
20%
40%
60%
contributors/labor force %
80%
100%
0
5000
10000
15000
20000
25000
income per capita PPP-adjusted
4
The long run ‘coverage gap’
We also know that coverage rises with income within countries
80.0
Uruguay
% of employed contributing
70.0
60.0
El Salvador
50.0
40.0
30.0
20.0
India
10.0
0.0
1
2
3
4
5
Quintile
5
What are the policy options?
 Find ways to expand existing contributory schemes
 Statutory mandate can be expanded
 Better enforcement
 Improve schemes in terms of the ‘deal’ or rate of return
to contributors
 Reduce administrative costs
 Reduce non-pension incentives to be in the informal
sector
 Set up parallel schemes designed specifically to
increase voluntary participation by informal sector
workers
 Introduce or expand non-contributory or ‘social
pensions’
6
Social pensions – their role today
 Palacios-Sluchynsky (2006) document a wide
variety of SPs ranging from universal with large
benefits to narrow with small targeted benefits
 Proponents of large social pension schemes note
that because it is not tied to a previous
contribution history, it is the only program that
can address the problems of the current or soon
to be elderly
 However, introducing or expanding social pension
programs may imply important fiscal tradeoffs
and may lead to certain distortionary behavior
 Existing research provides limited guidance while
many countries are now increasing the role of SPs
7
Social pensions – their role today
0%
The bars for each
country show the ratio
of the social pension to
per capita income x the
ratio of the number of
recipients to the
number of elderly
5%
10%
15%
20%
25%
30%
South Africa
Mauritius
Bolivia
Namibia
Brazil
Botswana
Nepal
Egypt, Arab Rep.
Turkey
Costa Rica
Colombia
Chile
Bangladesh
Many countries (e.g.,
India, Chile, Kenya) are
introducing or
expanding SPs
India
Uruguay
Russian Federation
Argentina
Estonia
Algeria
Dominican Republic
8
Social pension design issues
 Key SP parameters are:
 Eligibility requirements
 Age
 Income
 Residency
 Benefit level
 Palacios, Sluchynsky and Biletsky (2007)
suggest the need to take into account initial
conditions and devise a dynamic SP policy
9
Parameters – eligibility age
22
20
life expectancy at age 60
Health and
productivity of the
elderly will be
lower in poor
countries and for
poor within the
country
18
16
14
12
10
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
PPP adjusted $ income per capita
10
Parameters – eligibility age
However, more
countries are
introducing
this link
22
20
life expectancy at NRA
While the SP
eligibility age
should be
coordinated with
the contributory
scheme age, the
latter are not often
tied to life
expectancy
18
16
14
12
y = 0.2634x + 0.831
2
R = 0.2698
10
45
50
55
60
65
70
Normal retirement age
11
Parameters – eligibility age



There is a case for poor countries to have lower
social pension eligibility ages than in richer
countries
The social pension age minimum can be higher
than the contributory pension age minimum when
the latter benefits are adjusted on an actuarially
fair basis and contributors have achieved the target
smoothing objectives (replacement rates)
Both types of eligibility ages can be structured
flexibly with a view towards maintaining the
‘pension wealth’ constant; this automatically
handles increasing longevity over time
12
Parameters – benefit levels

Generally subjective, but should compare to three
numbers





Poverty line
Other social assistance type benefits
Contributory scheme target benefits
Poor countries probably can focus more on absolute
poverty while middle and higher income countries
may tend to look at relative poverty
Ad hoc or discretionary benefit changes over time
should be avoided in favor of indexation based on
objective indices to avoid inter-cohort inequities,
make fiscal costs transparent and minimize populism
13
Parameters – benefit levels
Benefit level should be designed with both contributory and
social pension components integrated into objectives
Australia
Denmark
14
Parameters – income tests

Social pensions are not intended to supplement
the income of rich elderly, but universal plans
exist in several countries




New Zealand, Mauritius, Botswana, Brunei
Rationale often cited is that it is simple to
administer; but this comes with a fiscal cost and
less redistribution
Where income tax net is wide, this can be dealt
with through progressive income tax
But most LICs and MICs have narrow tax bases
15
Parameters – income tests



A taper or partial offset with income is one
option and is used in several countries, but it
may be difficult and costly to assess hh incomes
A more practical and cheaper approach is to
apply a “pension test”
There are two problems with the pension test:


It will result in regressive errors by ignoring nonpension income and,
It would reduce incentives to participate in the
contributory scheme
16
Parameters – income tests




How convincing are these arguments?
First, there will be errors, but some of this can be dealt
with through income tax and some can be dealt with
through joint annuity provisions
Second, there is little evidence that rates of return
affect participation in contributory schemes (except in
the case of retirement behavior) – requires long term
view that is rare.
Third, as explained by Valdes-Prieto (2000), optimal
income tax theory (Slemrod et. al., 1994) is consistent
with a declining marginal tax on income for more
productive workers and targeting makes this possible.
17
Chilean SP reform
PILAR
VOLUNTARIO APV
PMG
PILAR
PILAR
CONTRIBUTIVO
CONTRIBUTIVO
PILAR
SOLIDARIO
PASIS
18
Initial conditions and dynamic SP policy



Most high income OECD countries are in a
steady state policy - high coverage and not
likely to significantly change the role of SPs
Low and middle income countries (LICs and
MICs) in contrast may still choose a different
long term SP policy (e.g., Bolivia, Chile)
We argue that initial conditions matter and
that the SP policy should be dynamic and
adapt as the contributory schemes expand
19
Initial conditions and dynamic SP policy
We focus on three stylized cases to see general
policy implications of different SP policies
Country PPP$YCAP
Coverage ratio Ratio 20-59/60+
types
population
LIC
>4500
17%
7.6
MIC
4500-15,000
51%
6.3
HIC
15,000+
90%
3.4
TSE
2000-20,000
66%
3.7
Source: authors’ calculations based on World Bank pension database and SIMA.
20
Simulation of dynamic SP policy





Case 1: shift to a universal SP at 40% of YCAP
Case 2: allow 100% pension test and introduce DC
scheme with 40% target RR at Lifetime Average
Income per capita (LIA)
Case 3: allow 50% pension test and introduce DC
scheme with 40% RR target at LAI
Case 4: CPI index the SP with 40% RR target at LAI
and introduce DC scheme
Coverage of the DC scheme allowed to expand and
mature along with growth in income per capita
growth over a 60 year period
21
Results for LIC
In the first year, there is only the universal pension…in
case 1, this remains the long term policy
Contributory
Social
Contributory
Social
2.5
Gross relative pension level
1.25
1
.75
.5
.25
0
0
.5
1
1.5
2
2.5
3
Individual income, proportion of average income per capita
2
1.5
1
.5
0
0
.5
1
1.5
2
2.5
3
Individual income, proportion of average income per capita
22
Results for LIC
In case 2, we apply a 100% pension test and introduce
the DC scheme so that at maturation it looks like this…
Contributory
Social
2.5
Gross relative pension level
1.25
1
.75
.5
.25
0
Contributory
Social
2
1.5
1
.5
0
0
.5
1
1.5
2
2.5
3
Individual income, proportion of average income per capita
0
.5
1
1.5
2
2.5
3
Individual income, proportion of average income per capita
23
Results for LIC
2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065
Real GDP Growth, %
NPV of Social, %GDP
Social, %GDP
2
2
3
3
4
4
5
5
%
%
6
6
7
7
8
8
9
9
The long run cost of the SP is
significantly lower due to DC scheme
2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065
Social, %GDP
NPV of Social, %GDP
24
Results for LIC
In case 4, the objective is absolute poverty, so
the CPI indexed SP declines relative to incomes
Contributory
Social
Gross relative pension level
1.25
1
.75
.5
.25
0
Contributory
2
1.5
1
.5
0
0
.5
1
1.5
2
2.5
3
Individual income, proportion of average income per capita
Social
2.5
0
.5
1
1.5
2
2.5
3
Individual income, proportion of average income per capita
25
Results for LIC
2
3
2
4
3
4
5
%
%
5
6
6
7
7
8
8
9
9
Case 4 has lowest long run fiscal cost
2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065
Social, %GDP
Case 2
NPV of Social, %GDP
2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065
Social, %GDP
NPV of Social, %GDP
Case 4
26
Results for MICs and TSEs




The tradeoffs differ for MICs because (a) they are
already older but (b) they have higher coverage – (a)
raises SP costs while (b) reduces them
Same is true for TSEs, except that there is a special
pattern of temporarily declining coverage that affects
the trajectory of the SP costs.
TSEs are also older than MICs generally.
In TSEs, the SP can be seen as a bridge between a
period of high coverage before and after the
transition
27
Conclusions





There is increasing pressure in many countries to
introduce or expand SPs in light of persistently low
coverage of contributory schemes, funded or not
Social pension parameters should be carefully
considered and viewed in conjunction with
contributory schemes
Pension tests may be the most practical mechanism
for controlling costs through targeting
A dynamic SP policy can reduce the long term fiscal
costs as contributory schemes expand and mature
Initial conditions will result in different
combinations of social vs contributory pensions and
different time paths of convergence to steady state
28
Thank you
29