The Investment Case For Social Protection – Methodological Challenges and Lessons Learnt

The Investment Case For Social Protection –Methodological Challenges and Lessons Learnt

Franziska Gassmann Andrés MiderosCécile CherrierPierre Mohnen

UNICEF Office of Research, Florence, 18 March 2013

Maastricht Graduate School of Governance

Outline

• Why focusing on returns of SP investments?• Understanding social protection as economic

investment• Methodological options and challenges• An example: RoR study for Cambodia• Are there lessons to be learnt?• Conclusions


The usual arguments for extending social protection rely on…• Human rights • Empirical evidence on impacts• Pilot projects• Affordability studies


‘Social protection is a human right’

• Art.22 and 25 of Universal Declaration of Human Rights (UN 1948)

• Social Protection Floor: “a set of basic social rights, services and facilities that the global citizen should enjoy” (ILO&WHO, 2009:4)

• Social assistance as entitlement – state obligation (Cichon et al., 2011)

• But: – Insufficient domestic resources to ensure all basic human rights at

once– Governments set priorities, but not necessarily SP


‘We know it works’

• Large international evidence base on positive impacts of social transfers on poverty, inequality and access to basic sercives (e.g., EPRI, 2010; Barrientos & Nino-Zarazua, 2010; Arnold et al., 2011)

• But:– In the absence of a national social transfer program,

evidence needs to be drawn from other countries– Effects depend on program design and implementation – no

blue print


‘Let’s try and prove it’

• Social transfer pilot projects to generate country-specific evidence

• But:– Frequently donor-driven and not aligned with national policy

priorities/policy-making process– Weakly evaluated– Questions about scalability and financial sustainability


‘We know it’s affordable’

• Cost-feasibility studies have been mushrooming (e.g. ILO, HelpAge, Unicef)

• ‘Prove’ that the provision of a basic package of social protection is affordable

• But:– Does not take into account output, outcome or

impact measures


Rationale for investment case

• Prejudices frequently prevail, difficult to fight• Additional arguments are needed to move SP up the national

development agendas– Demonstrate value for money– Analyze fiscal sustainability– Prove cost-effectiveness, capture multidimensional effects– Compare with alternative investments

• Develop economic argument for social protection– Costs AND benefits– Short term AND long term– Direct AND indirect


Non-contributory social protection as economic investment• Paradigm shift: SP not just as a cost for the economy

– Source of resilience in tough times– Support for growth and productivity in good times– Mechanism for social inclusion

• SP and economic growth (Alderman&Yemtsov, 2012)– Building and protection human capital, productive assets– Enhancing community assets, infrastructure– Stabilizer of aggregate demand, improving social cohesion,

making reforms feasible• Supported by international evidence (Barrientos, 2012)

Households’ disposable income

Health, education, livelihoods

Consumption

Socio Economic Development

Poverty alleviation

Cash transfers

In-kind transfers

Non-contributorySocial Protection

Institutional framework, social cohesion and structural conditions

Long run

Equity

Environment

Public goods and services

Savings

Labour supply

Insurance

Taxes

Taxes

Taxes

Taxes

Human capital

Production

Labour productivity

Economic growth

Demand

InvestmentPhysical

capital


Methodological challenges


Quantifying costs – the easy part?

• Assumptions:– Number of beneficiaries– Transfer value– Administrative costs (hypothetical)

• Tools:– Pension Calculator (HelpAge) (online, many countries)– Basic Social Protection Costing Tool (ILO)– Basic Social Protection Floor Costing Tool (ILO&UNICEF)

• Used in Argentina, Madagascar, Senegal– Rapid Assessment Protocol (+) (ILO)

• Used in El Salvador, Indonesia, Burkina Faso• Outputs:

– Costs as % of GDP– Hypothetical impact on poverty headcount and gap (RAP+)


Quantifying benefits (1)

• Frequently focus on one specific impact dictated by policy objective, data availability, modeling skills

• Mathematical models:– Use international evidence on impact to estimate potential country-specific

impact given specific program characteristics– Examples:

• DFID business case: Nigeria, Pakistan, Gaza• MBB (WB and UNICEF): Ghana (edu), Nigeria (health)• AusAID: Bangladesh, Cambodia, PICs• Hodges: Mauritania

– Limitations: • Impact of social protection programs cannot be generalized; depend on

economic, social cultural context and capacity• Findings sensitive to underlying assumptions if context differs


Quantifying benefits (2)

• Microsimulation models:– Ex-ante analysis of potential impact at micro-level– Based on national household survey data– Examples:

• Many single country studies• ADePT: e.g., Senegal

– Limitations:• Static simulation (only few dynamic models)• Focus on distributional effects (poverty, inequality)• Strong assumptions, such as perfect targeting• Exclude second-round effects

• Other models:– Macro models

• Focus at aggregate level • Cross-country analysis

– Comupatble General Equilibrium (CGE) models


Comparing costs with benefits

• Challenge: assign monetary value to benefits, impacts• Express benefits in terms of impact on economic growth

– Neglects distributional issues• Economic rate of return:

– Benefits: sum of distributional and behavioral income effects– Costs: costs of proposed SP package– Dynamic: aggregate effects over time to estimate long-term

impact


RoR study Cambodia

What are the economic returns of social protection in the mid- and long-term?

• Partial approaches:• Cross country analysis.• Computable general equilibrium model.• Cost-(effectiveness/benefit) analysis.• Treatment effects.• Microsimulation (can be extended).


RoR study Cambodia – Model• RoR: relation between net benefits and costs• Data: CSES (2004 and 2009) Cambodia• Microsimulation (steps):

• Static (cost-effectiveness): changes on poverty and inequality (direct distributional effect)

• Returns of human capital (education) at the household level• Behavioural (income) effects (probabilistic models)

• School attendance (education)• Nutrition (health)• Labour (participation and supply)

• Dynamic: 20 periods


RoR study Cambodia – Model

Household consumption

Poverty and inequality

Education(school attendance)

Health(underweight)

Labour participation

Human capital

Labour productivity

Social protection

Economic performance

Return

Direct (distributional) effects

Behavioural (income) effects


RoR study Cambodia – Policy

Social Protection Instrument

Target Population Benefit

Transfer Total Cost

KHR billion

% of GDP

KHR billion

% of GDP

Cash transferPoor children 0-6 years old in rural areas, up to two per household

USD 12 per month (60% rural food poverty line) 391 0.9 430 1.0

Social pension

Poor persons 65+ in rural areas

USD 20 per month (100% rural food poverty line) 139 0.3 153 0.4

Scholarship Poor children at lower secondary in rural areas

USD 50 per year (20% rural food poverty line) 25 0.1 28 0.1

Public works

Poor persons 18-64 years old in rural areas, up to 1 per household (80 days per year)

USD 2.3 per day 50 0.1 75 0.2

• Total costs is around 1.6% of GDP (USD 166 million, 2009).


RoR study Cambodia – Behavioural effects I

Model: 2SLS for all households

Dependent variable (independent variable) Urban(non-poor)

Urban (poor)

Rural(non-poor)

Rural(poor)

log of household consumption per capita (maximum level of education within the household)

0.042*** 0.016** 0.026*** 0.018***(0.005) (0.007) (0.003) (0.002)

Model: Probit model for individuals 6-25 years old

Dependent variable (independent variable)Education level (rural-poor)

Primary Lower secondary

Upper secondary

School attendance (log of household consumption per capita)

0.226** 0.560** 0.373(0.089) (0.262) (0.516)

Model: Probit model for children under 5 years old

Dependent variable (independent variable) National National (poor)

Rural(poor)

Underweight (log of household consumption per capita) -0.043*** -0.048 -0.038(0.015) (0.038) (0.041)

Underweight (no toilet facility in the house = 1) 0.026 0.062** 0.081***(0.016) (0.029) (0.029)


RoR study Cambodia – Behavioural effects II

(5-14) (15-17) (18-30) (31-45) (46-64)Unpaid work

Urban (Poor) -1.533* -4.076*** -1.605* -0.499 -0.369 (0.815) (1.004) (0.832) (0.544) (0.806)

N 785 381 1,011 639 369Log Pseudo Likelihood -483 -296 -844 -382 -231 Rural (Poor) -1.520*** -1.442** -1.236*** -0.547 -0.119

(0.440) (0.568) (0.395) (0.344) (0.349)N 6,997 3,091 7,474 4,896 2,519Log Pseudo Likelihood -5,400 -2,807 -6,580 -2,942 -1,617

Paid workUrban (Poor) -0.353 1.940*** 0.936 -0.921 -0.025 (0.338) (0.745) (0.803) (0.674) (0.916) Rural (Poor) -0.042 -0.256 1.004** 0.374 -0.605 (0.192) (0.448) (0.395) (0.388) (0.452)

Labour participation – Multinomial probit


RoR study Cambodia – ResultsBenefit Scenario Period 1 Period 5 Period 10 Period 15 Period 20

Average years of education (18-64 years old)

With social protection 6.52 7.67 9.00 10.40 11.62

Without social protection 6.52 7.65 8.89 10.22 11.41

Benefit (difference) 0.00 0.02 0.11 0.19 0.21

Total household consumption average annual

growth rate (%)



Benefit (difference) 1.55 0.26 0.12 0.07 0.04

Poverty headcount (%)



Benefit (difference) -6.0 -6.0 -4.2 -3.8 -2.6

Inequality (Gini of consumption)



Benefit (difference) -0.016 -0.014 -0.013 -0.012 -0.010

Cost Policy Period 1 Period 5 Period 10 Period 15 Period 20

Cost (% of GDP) Social protection package 1.6 1.4 1.2 0.9 0.8

RoR Discount rate Period 1 Period 5 Period 10 Period 15 Period 20

Rate of Return (Absolute benefit on total

household consumption / absolute cost) (%)

2% -11.6 -10.0 -4.1 5.8 14.7

3% -11.6 -10.1 -4.3 5.0 13.3

4% -11.6 -10.1 -4.6 4.3 11.9


RoR study Cambodia – Results

14.711.9

-15.0

-10.0

-5.0

0.0

5.0

10.0

15.0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

RoRC (d=2%) RoRC (d=4%) RoRC (d=6%)

Period

• Rate of returnEffect on total household consumption through human capital accumulation, due to higher school attendance thanks to social transfers.


RoR study Cambodia – Final remarks

• Any model is always a simplification of real life.

• Effects, benefits and returns may be higher if complementary policies are also implemented.

• Improving health and education coverage and quality.• Enhancing sanitation conditions.• Fostering economic productivity, formal labour market,

industrialization, innovation and technical change.


RoR study Cambodia – Final remarks

• Additional effects may increase benefits and RoR. • Behavioural (non-economic) effects due to SPI design.• Spillover effects and regional multiplier.• Institutional change and social cohesion.• Health status improvements (e.g. nutrition).

• Financing aspects (taxation), administrative issues (inefficiency) and targeting errors may reduce RoR.

• Specific SPI design (e.g. targeting, conditionality, payment mechanism) may affect RoR.


RoR study Cambodia – Final remarks• Investing in social protection promotes equitable

economic growth.• Solving human capital constraints, and fostering economic development at

the micro level.• A basic package of SPI for poor rural individuals in Cambodia has a RoR of

between 12% and 15%, after 20 periods (years). It becomes positive after 12 periods (years).

• Dynamic microsimulation provides a novel approach to analyse economic returns of social protection.

• Modelling options depends on data constrains.• Costs and specific impacts has been estimated. The model can be used

to study a gap regarding economic returns in the mid- and long- term.


The role of CBA in policy making

• Instrument to support evidence-based policy making and mobilize financial resources

• Enable dialogue with ‘money keepers’– Speek the language they speek– Build confidence of social ministries– Examples: Mozambique, Senegal


Conditions for success

• Analysis as integral part of policy dialogue• Needs to be followed up with in-country

support• Timing: a crisis may trigger interest• Policy options to be defined in consultation

with national stakeholders• Robust and transparent methodology


But…

• Evidence on costs versus benefits or rates of return alone does not do the trick– Policy makers are often prejudiced and myopic– Extending social protection is path dependent, depends on local

preferences and capacities– Political will and strong leadership remain pivotal (South Africa, Lesotho,

Brazil)– Strong impact evidence is crucial to build political support (Latin America in

general)• CBA needs to be used with caution, aware of the inherent risks

– Narrow focus on poverty reduction– Fragmented approach to social protection– Less attention to development process– Distraction from political economy


Conclusions

• Despite compelling evidence on the positive impacts of social protection, many governments are yet to be convinced

• Treating social protection as investment helps establishing the link between costs and benefits, taking into account the time needed to earn the money back

• Economic analyses such as CBA/RoR are only one component of policy decision-making– Positive social outcomes and human rights remain valid arguments

• Political will remains decisive for the expansion of social protection It all comes down to politics!– More research needed on the underlying motives/incentives/processes of

countries that successfully introduced/expanded social protection

The Investment Case For Social Protection – Methodological Challenges and Lessons Learnt

Documents

Transcript of The Investment Case For Social Protection – Methodological Challenges and Lessons Learnt