Performance Measurement for Social Protection Franziska Gassmann Bangkok, June 2012

Objectives Understand the importance of performance measurement as an essential tool for social protection governance. What is performance measurement? Where is it applied/can it be applied? Understand the basic principles underlying performance measurements in relation to social protection by discussing the type and characteristics of indicators and related issues of data collection and data quality. Link performance measurement to the policy making process and the dialogue with partner countries. Understand how to select indicators and set targets for EC programme management.

Social Protection Defined as public interventions to support poor and vulnerable households and individuals with the objectives to Smooth consumption Provide support during difficult times Reduce poverty and inequality Stimulate economic development The focus is on non-contributory programs financed from general government revenues, such as Cash transfers Subsidized goods and services Direct food support Public work programmes Social services Residential care provision

Outline Relevance of performance measurement for social protection Indicators for performance measurement Different types of indicators The characteristics of good indicators Performance indicators for social protection policies Measuring poverty Data requirements Wrap up

Performance measurement… …monitors at a programme level the cost and quality of services and outcomes for clients in order to account for public expenditures …is a tool for controlling and managing resources …requires the definition of outcome objectives for policies and programmes …provides regular and systematic information whether a program, policy or development strategy is implemented as planned, is achieving its objectives and whether and how it can be improved …moves the focus from inputs/outputs to outcomes and results

Why focus on outcomes? (EC Guidelines 2009) Results matter Encourage evidence-based policy making Protect political space of beneficiary countries Promote domestic accountability Stimulate demand for high quality data

Performance measurement for social protection Non-contributory social protection policies are financed from general government revenues Spending on social protection is heavily scrutinized and subject to constant pressure It is vital for the government to prove that public spending on social protection achieves its objectives effectively and efficiently The systematic and regular monitoring and evaluation of social protection policies is therefore crucial for any national government accountable to the public.

Regular performance measurement is especially important for social protection Social protection policies are developed at central government level, but service delivery takes place at local level Social programs rely on implementing agents The poor often live in remote areas or socially isolated surroundings Market forces are less prevalent in social service delivery

Guidelines for measuring progress towards intended outcomes in social protection Use information that is already available Monitor the extent to which social protection systems are pro-poor Agree on a common set of indicators Invest in building national capacity (EC, 2003) Performance indicators measuring achievements of objectives should be coherent with national and sector policies Coherence with MDGs A programme ensuring the quality of the indicators should be in place (EC, 2009)

Types of indicators

A good indicator is… (World Bank, 2004) A direct and clear measure of progress (specific) Varies across areas, groups and over time Sensitive to changes in policies Not easy to manipulate or blown off by unrelated developments Relevant for policy making Can be tracked in a cost-effective way (available) Consistent with policy-making cycle Easy to understand Reliable Consistent with data availability and data collection capacity

…and supported by intelligent benchmarks (Cichon et.al, 2004) Objective-based benchmarks Time-based benchmarks Comparative national benchmarks Comparative international benchmarks

Two dimensions of social protection performance Horizontal efficiency: Effectiveness of the programme in reaching the target group Under-coverage (exclusion) reduces horizontal efficiency Vertical efficiency: Efficiency of the programme in reaching the target group Inclusion errors reduce vertical efficiency

Targeting efficiency – evaluation matrix Target household Non-target household Household receives transfer Success Inclusion error Type II error Household does not receive transfer Exclusion error Type I error

Measuring exclusion and inclusion errors Exclusion error (under-coverage) = Number of poor w/o benefit Number of all poor Inclusion error = Number of non-poor with benefit Number of beneficiaries Leakage = Total amount of benefits received by non-poor Total amount of benefits paid

Distributional analysis – example

Measuring targeting effectiveness Share of transfers going to the poor = % of transfers received by poor % of poor in total population Example: Poor are 40% of the total population Benefits received by the poor = 20% Share received by the poor = 20/40 = 0.5 Interpretation: Value = 1: targeting is neutral Value < 1: targeting is regressive Value > 1: targeting is progressive

Measuring adequacy Share of benefits in total household income (consumption) = Benefit value Household income Change in poverty rates before and after transfer Absolute change = change in percentage points Relative change = change as % of pre-transfer poverty

Measuring Poverty (a very short intro to…) Why measure poverty? Assess living standard of population, identify the poor Identify needs/areas for policy support Evaluate effectiveness of national poverty-reduction and redistribution policies Assess impact and performance of development aid What do we need (besides data…)? Concept Indicators Measures Analysis and interpretation More reasons why we need to measure poverty: Advocacy purposes (= make the poor visible) Definitely since introduction of PRSPs is poverty reduction one of the main goal for development programs and projects; most projects have as their ‘impact objective’ or ‘outcome’ a goal such as ‘to reduce poverty’ or to increase the living standard of the population. The welfare of the population is the result of all national policies. If everybody would benefit equally, there would be no need for poverty and inequality measures, but some benefit more than others. Some policies, esp. social protection policies are explicitly targeted to reduce poverty, and, consequently, we have to measure poverty regularly.

Conceptualizing poverty, cq well-being Welfarist approach: Individual utility determines well-being (Lack of) well-being expressed in the level of income, consumption, assets, etc. Monetary measures Non-welfarist point of view: Other aspects determine well-being, such as ownership of commodities, fulfillment of basic needs, access to basic services, being socially included…. Lack of capabilities to function Rational consumer theory states that everybody makes the choice that provides the highest utility. The welfarist approach assumes that hhs have preferences and they are expressed in what they consume (food or health?) and that this choice gives them the highest utility they can possibly achieve. As such money is a good representation of welfare. No value attached to the choices made by the household. It also assumes that everything can be bought/consumed. Shows immediately shortcomings of purely welfarist approach. – choices made by household may not be wise; intra-household distribution? Accessibility of services? Only works if all choices are available. But also in non-welfarist approach and capability approach, command over commodities as an instrument to increase well-being still important.

Most common approaches Monetary poverty (income, consumption, expenditure, assets, …) Capability approach (coined by A. Sen) Social exclusion (think ‘EU’) Vulnerability (origin in disaster-risk-reduction) Monetary poverty: one-dimensional (although, implicitly multi-dimensional) Other approaches: multi-dimensional Monetary poverty: Still most common approach used for policy evaluation and poverty assessments Development away from simple income or consumption based concept to more sophisticated and complex concepts. Income may not enough to describe poverty. Poverty can materialize in other aspects, such as basic needs not covered, lack of access to education, health and other public services, relative deprivation, etc. Non-monetary measures gained ground over time. Why then, after all, are still most poverty assessments based on a monetary measure? All agree that poverty is multidimensional phenomenon, but many of the non-monetary concepts are difficult to be measured and valued. Monetary poverty: Concept/idea: welfarist approach where monetary status reflects a household’s utility level and preferences Economic welfare: goods and services consumed by household Straightforward, easy to interpret, may not capture well-being when markets do not function properly Capability approach: Term coined by Amartya Sen The capability approach argues that individuals are poor if they do not have the freedom to achieve their goals of well being and they are vulnerable when this lack of freedom persists over time (Sen 1999). ‘Utility’ and other commodity based formulations as a metric of welfare are rejected ‘Well-being’ means ‘being well’: ‘… value of living standards lies in the living and not in the possession of commodities…’ (Sen, 1987) Poverty is seen as a failure to achieve certain basic capabilities Basic capabilities: ‘the ability to satisfy certain crucially important functionings’ (Sen, 1993) An individual’s state of being can be viewed as a vector of functionings; i.e. achieved outcomes that determine life and living. These functionings are attained by utilising resources over which individuals have legal ownership rights, what Sen (1981) called entitlements. Measuring capabilities is difficult, since capabilities represent a set of potential outcomes In practice: focus on functionings (e.g. life expectancy, literacy, nutrition levels…)  very close basic needs approach

Poor ≠ Poor Each approach has different perspective on what constitutes a good life Each requires methodological assumptions Each has different policy implications MA: ensuring higher income solves problem CA: wider range of action needed (social provision of goods, improved allocation within family, efficient use of goods to achieve outcomes) SE: relative aspect important; just raising the level does not do the trick; redistribute policies Social exclusion 1980s and 90s: social exclusion emerged as concept in EU Marginalization and deprivation can exist even in rich societies with well developed welfare systems SE is ‘…a process whereby certain individuals are pushed to the edge of society and prevented from participating fully by virtue of their poverty, or lack of basic competencies and lifelong learning opportunities, or as a result of discrimination.’(EC, 2004) Goes beyond income poverty and includes social, cultural and political dimensions of exclusion  non-welfarist aspects of deprivation Close to Townsend’s concept of deprivation Acknowledges the fact that deprivation can have multiple faces Open method of coordination, Lisbon Agenda (2000) Laeken indicators Main characteristics of social exclusion approach: Relativity: Relative to a particular society Agency: SE as result of action of an agent Dynamics: Future is relevant as well, one disadvantage may lead to more disadvantages Multidimensional: Raises aggregation issues Problems with SE approach: Application to developing countries problematic as characteristics of SE differ Aggregation, causality between different dimensions What is ‘normal’? What are appropriate SE characteristics? Overall, least-defined and most difficult approach to poverty Vulnerability: No consensus on concept, how to define and measure it Originates in Disaster Risk Reduction General agreement that it differs from poverty and is a forward looking concept Most often: vulnerability as risk of falling into poverty, measured as expected poverty (Calvo&Dercon, 2005; Pritchett et.al. 2002; Haughton&Khandker 2009, and more)

In order to measure poverty, we need… Welfare indicator What indicator is best suited to determine the living standard of a household or individual in a given society? What indicator allows ranking households from poorest to richest? Different welfare concepts imply different indicators, and my identify different groups of poor Define a cut-off threshold, cq. poverty line Absolute versus relative poverty What constitutes a minimum living standard?

Proportion of health deprived children by income quintile, Vietnam 2006 24

Relative vs. absolute Relative poverty line Absolute poverty line Set as a share of average or median income or expenditures Moving concept; there will ‘always’ remain a class of poor people Closely linked with income distribution Examples: EU social indicator (60% of median income) Absolute poverty line Based on minimum consumption basket (food plus?) Survey-based or derived from other standards (e.g. national legislation, WHO) Examples: 1$ per day line, national poverty lines Issue: Relative poverty line allows comparison between countries Median or mean? What share? absolute poverty line has also relative aspects who determines what needs to be covered? Goods/services in minimum basket differ per country, ev. Even per region- think for example of Russia with very warm and very cold regions – needs for heating ) Choice of poverty line also depends on purpose of poverty analysis international lines not useful for national policy making International lines can be close or far off national lines

Example: Poverty in Country X Absolute poverty is defined regardless of the general welfare level or distribution in a given society. An individual is considered poor if he/she is not able to obtain the minimum necessities to maintain a purely physical existence . Relative poverty identifies those people that are poorest within the overall distribution in a given society. The concept changes with the ideas and values prevailing in a society.

How to determine the minimum living standard? In monetary terms: Objectively defined minimum, e.g. based on a basket of goods and services – absolute level As a percentage of the average living standard – relative level In non-monetary terms: Minimum years of schooling Access to primary health care Possession of x household appliances Having one hot meal per day … Many different ways to define minimum standards. Think of MDGs, or CRC. Regarding the latter: formulated rather generally, minimum still has to be defined; varies per society.

Examples of poverty lines International absolute poverty lines: 1.25 USD PPP per capita per day Useful for international comparison, but not necessarily for national policy making International relative poverty lines: 60% of median income per adult equivalent World Bank methodology based on actual consumption for national poverty lines: Extreme poverty: costs to obtain a minimum food basket Absolute poverty: Minimum food basket plus allowance for non-food goods and services Subjective poverty lines Derived from minimum income question Frequently higher than ‘objective’ poverty lines Guaranteed minimum income (budget based) Other methods: e.g. food share (show engel curve); this can also be applied to other expenditure items, e.g. housing (to determine housing allowance), but has its own problems in latter case – preference of hh where and how to live. No general acceptable standard. Minimum subsistence level – very popular in former SU countries! Not based on hbs, but on scientifically determined consumption standards.

y Poverty measures HC z PG q n

Poverty incidence (headcount) The 3 most common poverty indicators n=total population; q=poor; yi=welfare level; z=poverty line Poverty incidence (headcount) P0 = q/n = number of poor total population Poverty gap = average gap as a percentage of the poverty line P1 = 1/n ∑ q [(z – yi) / z]; non-poor: gap is zero! Poverty severity = weighted average gap P2= 1/n ∑q [(z – yi) / z]2 INDEX!

Example: Kyrgyz Republic 2005 Source: WB (2010) Poverty indicators by region (absolute poverty line) Total country: 43.1% 10.5% 3.6

Scorecard Kyrgyz Republic (WB, 2009)

Data requirements – Types of data Administrative data Provide information on inputs and outputs Collected by line ministries or special agencies Less suitable for analysis of policy outcomes Household survey data Representative sample of population Essential for poverty analysis Essential for assessing policy outcomes Population census Basic information on all inhabitants (limited indicators) Very costly, at most every 10 years Needed for poverty maps Usually not suitable for assessing policy outcomes

Data requirements – household surveys Essential for analysis of policy outcomes Availability improved considerably over the past decade Many variations, but not all equally suitable HBS, LSMS, MICS, DHS, etc. Have their own shortcomings Reliability of information (e.g. income from transfers) Questionnaires not always in line with latest policies Target groups cannot always be identified (e.g. disabled, war veterans) How frequent should they be available? No general rule At least every 2-3 years, if financially affordable

Data requirements – qualitative data Provides information not provided in surveys Subjective dimensions of well-being Intra-household inequality Barriers to access (benefits, services) Cultural, political or social factors that determine policy outcomes and impacts In-depth interviews, focus group discussions To inform survey design or to validate findings from survey

Wrap up Performance measurement ensures the accountability of the policy maker and the transparency of the decisions taken Availability and measurement of performance indicators is essential for social protection policy making Ensure regular and systematic collection of reliable information Analyse the collected information from various perspectives Data requirements need to be carefully assessed as resources are often scarce Better have a limited set of core indicators which are regularly collected and analysed Performance measurement is part of the policy process and the interpretation of indicators needs to be placed in the political and economic context