This presentation discusses using survey data to analyze elderly welfare and pension systems. It provides an overview of social protection programs and pensions frameworks. Survey data strengths include representing the entire population, while limitations include being a static snapshot. The presentation explores applying survey indicators like poverty, living arrangements, and elderly income generation to assess environment and performance of elderly welfare. Future work may focus on under-researched areas and exploiting existing survey data.
How to Troubleshoot Apps for the Modern Connected Worker
Pensions Core Course 2013: Assessing Elderly Welfare and Pension Performance with Survey Data
1. Assessing Elderly Welfare and Pension
Performance with Survey Data
April 3, 2013 Pensions Core Course
This presentation builds on the work of colleagues in HDN and PREM
2. Outline
• (1) Overview of survey data & social protection
– Role of social protection
– Application to elderly and pensions
– Pensions framework applied to survey data
– Understanding survey data strengths and weaknesses
• (2) Application: Survey data for elderly welfare and
pensions analysis –
– Environment
• Living arrangements
• Poverty
– Performance
• (3) Future work
2
4. Classification of programs
4
• Social assistance (Social
Safety Nets)SA
• Labor Market Programs
(active and passive)LM
• Social InsuranceSI
5. Objectives of pensions
• Elderly poverty protection
• Consumption smoothing
• Encourage savings
• Insurance against shocks
5
6. Understanding poverty
• No common definition for poverty exists
– General agreement: insufficient commodities
leading to constrained choices (Harold Watts)
– More narrow definition: lack of specific
consumptions (e.g. too little food energy intake)
– Less narrow definition: Poverty as lack of
“welfare” e.g., lack of “capabilitiy”: inability to
achieve certain “functionings” (“beings and
doings”) (Amartya Sen)
6 Based on DEC presentation
7. How poverty is commonly measured
• Individuals or households are ranked by income
or consumption
• The measure of income or consumption is
referred to as the ‘welfare aggregate’
• Poverty lines are then set either on a relative or
absolute basis, often based on an extreme and
basic standard of living
• Those with income or consumption (welfare
aggregate) below a given poverty line are
considered poor
7 Based on DEC presentation
8. Poverty measures
• Poverty headcount (FGT0) - % of individuals or households with welfare
below the poverty line
• Poverty gap (FGT1) - mean shortfall of poor from the poverty line,
expressed as a percentage of the poverty line
• Poverty severity (FGT2) – average of squared poverty gap ratio
8
Distance
squared
% Below line Avg distance below line
12. How is social protection applied to the
elderly?
• Equity - Elderly poverty protection
• Opportunity
– Consumption smoothing in retirement
– Encourage savings
• Resilience - Insurance against shocks
12
13. Pension Diagnostic Assessment –
Evaluation Process & Criteria
Enabling
environment
(Motivating reform,
framing & constraining
reform options)
Existing design
Demographic
profile
Macro-economic
environment
Institutional
Capacity
Financial market
status
Political economy
Initial
Conditions &
Inherited
System
Demand -
consumption
smoothing &
elderly poverty
protection
Supply -
mandatory &
voluntary
pension & social
security schemes
Family &
community
support
Reform
objectives
Primary:
improving
coverage,
adequacy, &
sustain-ability
for the long-
term
Secondary:
improving labor
markets,
macro/fiscal
position, &
contributing to
financial market
development.
Reform Design &
Implementation
Options
Design reforms -
introduce new
schemes,
parametric &
structural
reforms
Governance,
Institutional and
regulatory
reforms
Strengthening
institutions &
implementation
14. Diagnostic Assessment – Data,
Indicators and ToolsInformation/ Data
Elderly
incomes,
vulnerability &
poverty
Initial
conditions
Mandatory &
voluntary
pension
systems &
social security
schemes
Tools
Country HH survey data
ADEPT-SP x/country data
Environment –
UN Population Projections
Country admin data
Financial market data
Macro & fiscal data (country/IMF)
System design –
Admin data/country laws
WB database comparators
Performance –
Admin data/country laws
WB database comparators
HH survey data
Administrative data from social welfare
schemes, housing, health provision.
HH survey data.
Additional
state support
Indicators
Environment
Demographic
Economic
Financial
Informal Support ADEPT-SP
Apex
PROST
ASPIRE & ext.
x-country
data
Design
Structure of pension system
Qualifying conditions
Parameters
Performance
Coverage
Adequacy
Financial sustainability
15. Diagnostic Assessment – Tools
Tools
APEX
Evaluation of
individual level
benefits across
instruments + for
different income
groups.
Individual
replacement rates
Replacement of
average wage
Pension wealth
ADEPT-SP
Elderly welfare
Elderly poverty
Co-residence
Elderly income
generation
Comparisons of welfare,
poverty across elderly,
non-elderly & household
types.
PROST
Baseline. Long-term projections
of financing gap for existing
schemes + replacement rates
for current and future
retirees
Reform scenarios. Long-term
projections financing gap +
replacement rates for
parametric and/or structural
reforms
Outputs to simulate other
instruments (social pensions,
voluntary savings)
WB Database & External
X-Country Data
Cross-country
comparisons
Demographics
Coverage
Adequacy
Affordability
Sustainability
16. What is survey data?
• Examples: HBS, LSMS, DHS
• Organization: Household or individual level
• Timing: Generally collected ever 2-3 years,
more frequent than census (~ 10 years)
• Information: Core demographics (eg age and
gender), expenditure/ income, employment
status, public and private transfers, etc
16
18. Household Input File
18
Household
Identification
Individual
Identification
STRATA PSU
Urban location
=1; Rural
location=2
Household
expansion
factor
Household
Size
Adult
equivalent
scale
Head of the
household
Age of the
household
member
Total
household
income
Poverty
line
Amount
received
from old
age
pensions
Participation in
scholarship
programs
Amount received
by the household
from
Oportunidades
Amount
received by the
household from
Pro-Campo
id_hh id_ind strata psu urban hhweight hhsize adul_eq head age hh_income pob_ing apos becas_ toport tprocam
20060150282 1 1 2 2 305 3 2 1 18 2459.34 938.61 0 180.49
20060150280 1 1 2 2 305 7 6 1 56 9094.69 938.61 1 334.24
20060150030 1 1 1 1 777 4 3 1 77 18183.37 938.61 1403.81 0
20060150040 1 1 1 1 777 1 1 1 92 4458.78 938.61 1604.35 0
20060150050 1 1 1 1 777 2 2 1 83 6397.05 938.61 1640.45 0
20060150060 1 1 1 1 859 5 2 1 41 12988.27 938.61 0
20060140410 1 1 7 1 638 10 6 1 56 10730.62 938.61 1411.48 0 514.18
19. Administrative vs Household Data
Administrative data
• - Limited population
coverage - only ‘covered’
included
• + Comprehensive data on
contributors, beneficiaries
• + Cumulative (over life
cycle)
• - Narrow variables (eg age,
gender, contribution)
Household data
• + Entire population
represented
• -/+ Generally lack data on
contributors, though extensive
info on recipients (and non-
recipients)
• - Static (singe year, usually not
panel)
• + Much more comprehensive
(demographic, poverty, public
& private transfers)
19
20. Uses of survey data (continued)
• Quality of data from design, implementation
and processing of surveys is critical
• Surveys do have limitations, such as
comparability across countries (eg income vs
consumption), recall, survey non-response,
comparability over time
• Nonetheless, survey data can provide a
wealth of information for welfare and
program assessment
20
21. ASPIRE Survey Initiative
• One of the largest datasets on social protection in the
world: the combined dataset contains information on
almost 5 million individuals in 1.3 mln. households.
• Reveals large information gaps in standard household
surveys, many of which do not include questions about
participation in social protection programs.
• Advocates for greater use of surveys for assessing
social protection policies, and provides tools for
enhanced data collection (survey modules, links to data
analysis software).
21
22. ASPIRE household survey components
1. Data collection, harmonization and validation
2. Tools for data analysis (ADePT and Simulators)
3. Training in tools (Bank and non-Bank clients)
4. Analytical notes and reports
5. External partnership and internal
collaboration
6. Dissemination
22
23. Types of Social Protection Programs
Category I Category II Type of program
Old age pension
Old age civil servant pension
Veteran's old age pension
Early retirement pension
Survivors pension
Survivors civil servant pension
Occupational injury benefits/ pension
Paid sick leave
Disability Disability pension
Unemployment compensation
Severance pay
Early retirement for labor market reasons
Labor market training
Youth measures
Subsided employment
Employment measures for disabled
Employment service and administration
Low income/ Last-resort program
Non-contributory/ Social pension
Family allowances*
Disability benefits
Housing allowances
Food stamps/ Vouchers
Conditional cash transfers Conditional cash transfers
Food rations
Supplementary feeding
School feeding
Energency food distribution
Fee waivers, education
Scholarships
Fee waivers, health
Food price subsidies
Public distribution systems
Energy and utility subsidies
Public works Public works
In-kind food transfers
Fee waivers and scholarships
General subsidies
Social safety net programs
Labor market programs Unemployment
Active labor market programs
Cash or near-cash transfers
Pensions and other social
insurance
Old age
Survivors
Occupational injury/sickness
benefits
23
24. ASPIRE includes the following 12 indicators for
measuring social protection:
• Coverage
• Beneficiary Incidence
• Benefit Incidence
• Generosity
• Leakage
• Targeting differential
• Impact on Poverty
• Impact on Inequality
• Coady-Grosh-Hoddinott indicator
• Program duplication
• Social protection overlap
• Cost benefit ratio
24
26. Motivations for work
• “Looking forward, this review suggests several concrete areas for
further work. First, investing in strategic knowledge products
particularly in areas related to coverage where there are significant
gaps including, but not limited to:
– Old age poverty, informal support and pension systems, especially in
countries with large agricultural and informal sectors such as in South
Asia and Sub-Saharan Africa. This requires efforts in several areas
including exploiting existing survey data, collecting administrative data
from country sources and micro-simulations with the APEX model.”
• “Measurement of performance must come from systematic
evidence provided by household survey data. The surveys provide
information on the incidence of benefits and the impact of
programs on household poverty, among other things.”
Source: Dorfman and Palacios. 2012. “World Bank Support for Pensions
and Social Security”
26
27. Why use survey data for elderly
poverty and pensions?
• Ability to answer new and different policy
questions
– Environment – poverty, distribution of
income/consumption, living arrangements, key
demographics
– Design – N/A
– Performance – coverage (receipt), poverty impact,
adequacy, targeting, etc
• Cross-tabulate by key characteristics, eg geography, gender,
age, income
• More breadth of information on individuals and
households
27
28. Some practical uses of survey data
• Understand characteristics of elderly and non-
elderly population (e.g demographics, living
arrangements and welfare)
• Produce evidence- based findings on pension
and other programs, such as coverage,
adequacy, poverty impact, etc
• Provide snapshot of income by sources
(public, private) for different age groups
28
29. Pensions Survey Work
• East Asia and the Pacific (EAP) – 7 countries,
Eastern Europe and Central Asia (ECA)– 20
countries, Latin America and Caribbean (LAC)
– 20 countries, Middle East and Northern
Africa (MNA) – 5 countries, South Asia (SAR) –
7 countries. The surveys vary in size from
approximately 15,000 in Albania to 600,000
individuals in India.
29
30. Applications of Survey Data
• (1) Environment
– Living arrangements
– Poverty (elderly and non-elderly)
• Design – N/A
• (2) Performance
– Coverage
– Adequacy
– Poverty impact
– Program overlap
– Cost-benefit
– Targeting
30
32. Living Arrangements
• What – the structure of households by age,
gender, size
• Main indicator – Co-residence rate (elderly
living with non-elderly)
• Why – proxy for informal support by non-
elderly, ‘voluntary pillar’ of family support can
complement formal pension systems
32
33. Co-residence
• Correlates of co-residence in existing studies -
lower welfare, living in urban areas, widowers,
higher country level GDP per capita)
• Initial findings
– Co-residence more likely for: urban, not receiving
a pension, not receiving remittances, lower
welfare deciles, lower income countries
– Wide variation between, and within some regions
(eg ECA)
33
37. Coresidence rate by key
characteristics- country level
37
0.0% 10.0% 20.0% 30.0% 40.0% 50.0% 60.0% 70.0% 80.0% 90.0% 100.0%
Total
Urban
Rural
Male
Female
60-74
75+
Bottom Quintile
Top Quintile
Pension Receipt
No Pension Receipt
Remittance Receipt
No Remittance Receipt
Employed
Not Employed
38. Poverty of elderly
• Are elderly household more poor then non-
elderly households?
• Are elderly individuals more poor then non-
elderly?
• Preliminary findings (point estimates robust, though less for s.e.)
– By household type ( any elderly, only elderly, some
elderly, no elderly)
– By individual – youth, working age, elderly
38
39. Welfare distribution by age group –
individual level
39
0
.2.4.6.8
Density
8 10 12 14
Log of percapita welfare
Youth Working Age
Elderly
Poverty line
41. Regional comparison – poverty % by
age group
41
0.000
0.050
0.100
0.150
0.200
0.250
Chart Title
Overall Youth Working Age Elderly
42. Country level - Poverty Headcount by
Household Type
42
0%
5%
10%
15%
20%
25%
30%
Average 1) Elderly:
lone
2) Elderly:
2+
5) Elderly
with
Working
Age
7) Elderly
with
Working
Age and
Youth
6) Elderly
with Youth
3) Working
age only
8) Working
age and/or
Youth
HH Only
Elderly
HH Some
Elderly
HH No
Elderly
44. Select performance indicators
– Coverage - receipt
– Adequacy – transfer amount/ poverty line or
/welfare
– Poverty impact – reduction in poverty due to
transfer
– Program overlap – receipt of 2+ programs
– Cost-benefit – reduction in poverty gap for each $
spent
– Targeting - % intended beneficiaries receiving
transfers
44
45. Simulated poverty impact
45
Poverty Rate (FGT0)* Poverty Gap (FGT1)* Poverty Severity (FGT2)*
All Income No pension transfer All Income No pension transfer All Income No pension transfer
Individual level
Population 0.5077 0.5878 0.2312 0.3265 0.1487 0.2435
Youth (0-14) 0.6196 0.6651 0.2873 0.3617 0.1860 0.2642
Working Age (15-60) 0.4718 0.5248 0.2157 0.2789 0.1405 0.2024
Elderly (60+) 0.3502 0.7611 0.1330 0.5455 0.0687 0.4685
Elderly (60-74) 0.3562 0.7380 0.1279 0.5224 0.0656 0.4485
Elderly (75+) 0.3331 0.8279 0.1476 0.6124 0.0777 0.5262
Non-Elderly (0-59) 0.5218 0.5723 0.2400 0.3069 0.1559 0.2233
Household level
Elderly-only households 0.0410 0.8512 0.0177 0.7742 0.0126 0.7337
Some elderly households 0.4817 0.7283 0.1817 0.4697 0.0917 0.3767
Non-Elderly Households 0.4325 0.4550 0.2088 0.2335 0.1452 0.1690
Total 0.4197 0.5321 0.1924 0.3116 0.1269 0.2429
Notes
The poverty line is set at total median income (deflated) – Note that for
illustration purposes as typical poverty line is 50% median
The simulated welfare subtracts pension income (deflated)
Weights are applied for statistical accuracy
47. Pipeline work
• Using survey data for country work,
particularly environment and performance
diagnostics – poverty profile, living
arrangements, coverage, adequacy, etc
• Regional and cross-regional comparison
• Multi-country indicators and research
examining factors such as correlates to
coresidence & elderly poverty
47
48. ADePT Training
• There will be a hands-on computer training
next week for those interested.
• Please sign-up
48
49. Select References
• Cameron. 2000. “The Residency Decision of Elderly Indonesians: A Nested Logit Analysis”. Demography. Volume 37-No. 1.
• Costa 1998.The Evolution of Retirement: An American Economic History, 1980-1990. University of Chicago Press. Chicago.
• Deaton. 2000. The Analysis of Household Surveys: A Microeconmetric Approach to Development. World Bank.
• Dorfman et al. 2012. “Malaysia Elderly Protection Study”. World Bank. Washington DC.
• Evans, M. 2012. “Profiling Elderly Populations and Elderly Poverty – Practice Note”.
• Giles et al. 2010. “Can China’s Rural Elderly Count on Support from Adult Children?”. World Bank. Washington DC.
• Holzman, R. and Hinz, R. “Old Age Income Support in the 21st Century: An International Perspective on Pension Systems and Reform”, World Bank,
2005.
• OECD. 2011. Pensions at a Glance 2011. Paris.
• Rofman, R. et al. 2008. Pension Systems in Latin America: Concepts and Measurements of Coverage. World Bank.
• Whitehouse, E. 2007. Pensions Panaroma. World Bank. Washington DC.
• World Bank. 2009. Closing the Coverage Gap. Washington DC.
• World Bank. “International Patterns of Pension Provision II: A Worldwide Overview of Facts and Figures”. Washington DC.
• World Bank. “Pension Indicators: reliable statistics to improve pension policymaking”, World Bank Pension Indicators and Database, Briefing 1,
Washington DC.
• World Bank. “Coverage: How much of the labor force is covered by the pension system?”, World Bank Pension Indicators and Database, Briefing 2,
Washington DC.
• World Bank. “Adequacy (1): Pension entitlements, replacement rates and pension wealth”, World Bank Pension Indicators and Database, Briefing 3,
Washington DC.
• World Bank. “Adequacy (2): Pension entitlements of recent retirees”, World Bank Pension Indicators and Database, Briefing 4, Washington DC.
• World Bank. “Financial sustainability: Assessing the finances of pension systems over the long term”, World Bank Pension Indicators and Database,
Briefing 6, Washington DC.
• World Bank. “Economic efficiency: Minimizing the pension system’s distortions of individual choices”, World Bank Pension Indicators and Database,
Briefing 7, Washington DC.
• World Bank. “Administrative efficiency: assessing the cost of running public pension systems”, World Bank Pension Indicators and Database, Briefing
8, Washington DC.
• World Bank. World Bank, “Security: Risk and uncertainty in retirement-income systems” , World Bank Pension Indicators and Database, Briefing 9,
Washington DC.
• World Bank. 2012. “Pensions Database”. Washington DC.
49
50. Thank you!
50
• If your country is interested in survey training on Social
Protection and Poverty (1/2 day to 3 day courses):
– Please contact Mr. Ruslan Yemtsov, ryemstov@worldbank.org
Mr. Robert Palacions rpalacios@worldbank.org or Mr. Brooks
Evans bevans2@worldbank.org