Social Protection Överview


Published on

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

Social Protection Överview

  1. 1. Please do not cite or quote without written permission of the authors UNITED NATIONS DEVELOPMENT PROGRAMMESequencing, Cost-Efficiency and Fiscal-Sustainability of Social Protection—An Overview By Yanchun Zhang, Nina Thelen and Nergis Gulasan Preliminary Draft 25 October 2012 Bureau for Development Policy United Nations Development Programme, New YorkNote: The views expressed in this paper are those of the authors and do not necessarily reflectthose of UNDP. The authors thank Anne-Isabelle Degryse-Blateau and Selim Jahan for supportto this project. The authors are grateful for helpful comments received from Artemy Izmestievand Claudia Vinay. Please send comments and suggestions to the following e-mail, and 1
  2. 2. Please do not cite or quote without written permission of the authors1. IntroductionSocial protection1 is gaining increasing support as an instrument to counter various types of riskand vulnerability in an ever globalizing world. The global economic and financial crisisreasserted the importance of protecting the livelihoods of all members of societies, especially themost vulnerable, through well-designed social protection systems.2 But while social protectionhas gained momentum in empowering people to weather an ever more insecure economicenvironment, social protection is more than just a buffer to the risks brought about by short termeconomic shocks like the recent financial crisis or fluctuations in the business cycle in general.More long term social pressures linked to population growth, demographic change andurbanization as well as environmental challenges through increasing droughts, storms and floodsaggravated by climate change have put government provision of social protection higher on thenational and global agendas. When well designed, social protection can contribute to economicgrowth.3 Social protection can also contribute to building more equal societies and reducing theeconomic costs and distorted opportunities (e.g. slower growth) brought about by large incomedifferences in societies. 4 Many social protection programmes in developing countriesincreasingly include transformative and productive elements which empower people to improvetheir lives and to graduate out of poverty. As a result social protection becomes an integrated partof development policy.Many questions, however, still remain to be answered on how to build a social protection systemthat protects the most vulnerable, contributes to sustainable growth and development andenhances resilience to shocks.Our paper focuses on three aspects of the social protection debate related to sequencing, cost-efficiency and fiscal-sustainability of building a well-functioning social protection system indeveloping countries. We aim to use some country experience to shed light on how developingcountries can proactively engage in an adaptable social protection agenda sequencing frominformal to more formal social protection arrangements, continually assess the evolution ofcountries‘ potential vulnerabilities and identify cost-efficient options to address thesevulnerabilities, and firmly embed fiscally sustainable social protection in national socio-economic development planning.The rest of the paper is structured as follows. Section 2 analyzes the sequencing of socialprotection to identify the limitations of informal and semi-formal social protection and pin downsome of the triggers for a switch to more formal social protection. Section 3 reviews cost-benefitanalysis to evaluate the economic affordability and cost efficiency of specific social protectionmeasures. Section 4 explores fiscal sustainability issues of a social protection system and1 The social protection term used in the proposed initiative is a broad concept which includes labour marketinterventions, social insurance programs and social assistance.2 G20 leaders and international agencies like the United Nations, the World Bank and the IMF are all supporting theimplementation of sound social protection systems and are calling on governments to step up measures to protecttheir peoples in order to restore growth, progress and confidence (see G20 2011).3 Dercon (2011); UNDP and ILO (2011).4 See, for instance, Stiglitz (2012). Soares et al. (2007) found that about one fifth of the 4.7 percentage point declinein the GINI coefficient during 1995-2004 could be attributed to Bolsa Famí lia. 2
  3. 3. Please do not cite or quote without written permission of the authorsdeveloping countries‘ options to create fiscal space for social protection provision. Section 5concludes.2. The Sequencing of Social ProtectionOver the past decade, social protection has received increasing attention in development policyas an instrument to reduce poverty and to enhance opportunities and human capabilities. 5Particularly since the mid-1990s, many governments of developing countries have successfullyexperimented with social protection which demonstrates the potential.6 In the absence of publicsocial protection arrangements or inclusive and complete credit, savings and insurance markets,many people in developing countries have historically relied much on traditional and informalrisk mitigation arrangements.Social protection arrangements seek to reduce poverty and vulnerability and to improve humanwelfare. Hence, the main motivation for the existence of social protection in societies lies in corehuman values such as responsibility, reciprocity, social solidarity, civility and self-help. 7Precisely because social protection is built on ethical values and often also rooted in traditions,friends and family (i.e. households more broadly) serve as actors in informal social protectionarrangements. In addition, a large number of people in developing countries rely on so-calledsemi-formal social protection arrangements. Examples include savings clubs, food cooperativesand church groups. These ―mutual aid societies‖ or ―member-based organizations‖ provide socialprotection where family networks have been eroded, like for instance in urban settings, or wheremore formal arrangements fail to deliver.8The types of arrangements of social protection, as well as social protection initiatives andprogrammes vary strongly across developing countries (and regions). While differences betweenlow-income countries (LICs) and middle-income countries (MICs) exist, generally, manydeveloping countries are now considering revising their social protection programmes andmoving towards more integrated, inclusive, modern and established social protection systems.One option for governments to take into consideration is the implementation of a basic set ofsocial protection arrangements at the national level, a so called ―Social Protection Floor‖. TheSocial Protection Floor initiative is a UN system-wide effort to ensuring a basic cover of socialprotection for all over the life cycle.9 The reasoning of such a floor is that all in need should haveaccess to essential health care and to basic income security.10In this section, we first explain what social protection is and, for the purpose of this paper, settleon a definition of the term. We then provide an overview and elaborate on the types of5 Cook and Kabeer (2009).6 Cook and Kabeer (2009).7 Chen, Jhabvala and Lund (2002); UN (2000).8 Tshoose (2010).9 3
  4. 4. Please do not cite or quote without written permission of the authorsarrangements in which social protection is provided (i.e. informal, semi-formal and formal).11Second, we discuss the limits of informal and semi-formal social protection arrangements.Third, we discuss evidence on triggers for a switch to more formal, modern, integrated andestablished social protection in different countries (world regions). The section concludes withfindings on enabling conditions that have helped developing countries to integrate socialprotection into their national development strategies and recommendations that developingcountries may with to consider when embedding existing non-formal social protectionarrangements in their national development strategies.2.1. Definition and Categorization of Social ProtectionThere is no unanimous definition of social protection 12 or clear categorization of socialprotection measures in the literature or in policy practice.13 For the purpose of this paper, wedefine social protection measures as interventions that are intended to reduce poverty andvulnerability (including transitory poverty and vulnerability owing to economic and othershocks) and to improve human welfare. Many social protection measures support poor andvulnerable groups of people, which includes groups like the (non-working) young, unemployed,elderly and special groups (children, sick, disabled and minorities) to cope with economichardships.14Social protection includes social insurance, social assistance and labour market interventions.15Contributory social insurance programmes under which people receive benefits or services basedon regular financial contributions to an insurance programme include publicly mandatedinsurance against ageing, disability, sickness, death or unemployment. Most social insuranceschemes are provided by the government and closely linked to the formal labour market whichoften limits their coverage to formal workers. However, we also count insurance schemes thatprotect against risks to livelihoods arranged by actors other than the government in ourdefinition.11 This paper is merely intended to provide an illustrative and informative overview of the topic. It is not intended toprovide a new analytical framework. For an alternative, deeper analysis of levels of social protection, formality andactors, please see Holzmann and Jø rgensen (2001).12 The term social protection is generally considered to encompass a broader array of arrangements and actors thansocial security. For instance, the former counts protection arranged through family members or members of a localcommunity (ILO 2010a). However, given the fact that the terms are often used interchangeably in the literature, weuse the terms social protection and social security interchangeably, unless noted otherwise.13 It should be noted that there is no generally recognized definition of social protection. In fact, variousinternational organization, scholars and even countries developed their own (Bonilla Garcí and Gruat 2003). For a acompilation of social protection definitions of selected international organizations, see Annex Box 1. Variousscholars have discussed social protection components and have developed typologies, see for instance Devereux andSabates-Wheeler (2004).14 We are aware that our definition may not be precise and can be broadly looked at as social policy measures, butdecided to use it for the purposes of this paper.15 Some narrower definitions of social protection used in other studies exclude interventions measures on publichealth, education, housing or labour market interventions (see Annex 1, Box 1). Our definition refers to the mostcommon areas of social protection as defined by the Governance and Social Development Resource Center (GSDRC2012). 4
  5. 5. Please do not cite or quote without written permission of the authorsSocial assistance programmes are non-contributory transfer programmes targeted at the poor andthose vulnerable to poverty and shocks. There is no universal consensus on the types ofinterventions covered by the social assistance label. Common examples include cash transfers(conditional cash transfers (CCTs) which are transfers to poor households conditional on specificbehavior) and unconditional cash transfers (UCTs) like non-contributory pensions, etc.); in-kindtransfers (e.g. food transfers); fee waivers (health fees, school fees, scholarships); utilitysubsidies (e.g. electricity, housing and water) and so on.Labour market interventions are aimed at protecting people who are in the labour market or atpoor people who are able to work.16For the purpose of this paper, we differentiate between three types of social protectionarrangements17: (i) informal; (ii) semi-formal; and (iii) formal (public and private) arrangements.Informal arrangements are based on friends and family (―kinship‖).18 Semi-formal arrangementsare based on voluntary or membership associations, civil society organizations (CSOs) (e.g. non-governmental organizations (NGOs), trade unions). Formal arrangements are based on publicactors (i.e. the central or local government) and private actors (i.e. insurance companies (andbanks)).Social protection can be arranged informally, semi-formally or formally. Various actors areinvolved in these different types of arrangements. In the absence of formal social protectionmechanisms and—oftentimes—market mechanisms (for instance for credit and insuranceproducts), people in many developing countries have historically reverted to friends and familyfor help in the face of adverse conditions.19 This is what we define as traditional and informalsocial protection arrangements. The exact mechanisms of social protection provided in thisinformal way very much depend on the particular traditions of the society in question. Likewise,16 It should be noted that our definition of labour market interventions is broader than the definition by theGovernance and Social Development Resource Center that only counts ―[l]abour market interventions [that] provideprotection for poor people who are able to work‖ (GSDRC 2012). Labour market interventions include (vocational)training and skills development and changes to labour legislation. They also include labour market measures thatlikely fall under one of the other two areas of social protection (social insurance or social assistance) but are focusedon the labour market, like unemployment insurance, and part-time unemployment benefits. Public worksprogrammes (the poor working for food or cash), can be considered labour market interventions, but they are alsosometimes referred to as social assistance, since they basically function like conditional transfers (i.e. cash or food ishanded out in return for work on public infrastructure projects).17 It should be noted that this is the authors‘ own elaboration. We settled on this tentative typology to easeunderstanding of the landscape of possible arrangements of social protection. There is no generally accepteddefinition of ―formal‖ and ―informal‖ social protection arrangements in the literature. While Holzmann andJø rgensen (2001) identify three main arrangements of social protection (public, market based and informal) as doHoogeveen at al. (2004) (formal, market based and informal), Mohanty (2011) refers to CSOs as actors in semi-formal arrangements of social protection. Gentilini and Omamo (2009) consider both public and private actors to be―formal‖ actors in social protection arrangements. Our categorization includes three categories of social protectionarrangements: social protection through public and private actors (formal), social protection through community,groups, or member-based organizations (semi-formal) and social protection through households (informal).18 UN (2000).19 Mendola (2010). For a literature review on how people cope in the absence of publicly provided social protectionand often not accessible credit and insurance markets in low-income settings, please refer to Mendola (2010). For anoverview of how people cope with risk in an environment of non-inclusive markets, please refer to Mendoza andThelen (2008). 5
  6. 6. Please do not cite or quote without written permission of the authorsthe nature and number of events that are covered by informal social protection also varydepending on the cultural and country context.Much for the same reason as informal social protection provision, semi-formal social protectionarrangements are found to have developed. Mutual aid arrangements at community level as wellas money transfer between people that belong to the same community or neighborhood or familyare examples of semi-formal social protection arrangements. Instead of merely relying on friendsand family, many people have joined forces in mutual aid schemes like burial societies or othercommunity-based arrangements. Supporting one another, mutual aid arrangements are organizedthrough groups, associations and social networks.20 Mutual aid schemes are found to partly havedeveloped in response to the process of industrialization and urbanization in developingcountries, supporting individuals where the original extended family system had been eroded.21Burial societies and stokvels 22 in Africa, for example, are found to be more present in urbanareas.23 Burial societies help meet (material and non-material) obligations to organize a dignifiedfuneral according to traditions for a deceased family member which otherwise might stretch afamily‘s budget. Civil Society Organizations (CSOs) are found to be involved in the provision ofchild protection services in the Pacific Island Countries (PICs) while church groups are found toplay an important role in caring for the aged and disabled in Africa. 24 Some of these groupsspecifically target those in need. For instance, social protection arrangements in church groups inZimbabwe only cater to members who are unable to help themselves.25In the category formal social protection arrangements, the public sector (local and federalgovernment) is an important actor. Another actor in formal, modern social protectionarrangements is the private sector. Private actors play a role in the areas of microinsurance aswell as general insurance afforded by insurance companies (and, in some cases, banks). 26Microinsurance27 provides an affordable alternative for low-income populations who are eitherexcluded from insurance markets or without access to appropriate public social protection, whereit exists. It can help people with limited financial means to hedge themselves against risks suchas old age, death or illness.28 Microinsurance can usefully be attached to or linked to existingformal social protection arrangements, so that public and private instruments together cancontribute to broadening coverage against those risks that pose the greatest challenge for thepoor. 29 Microinsurance is regarded as ―a social protection instrument that can complementexisting social protection systems in a meaningful way‖.30 Regulations for private actors as well20 Mendola (2010).21 Tshoose (2010).22 Stokvels are common in Africa, are invitation only clubs and usually comprised of 12 or more members whocontribute on a regular basis, based on their income. Each month, a different member receives the money. Regularmeetings assure that members pay their contributions, a constitution regulates the functioning of the stokvel. Formore information, see Tshoose (2010).24 Dhemba, Gumbo and Nvamusara (2002); Mohanty (2011).25 Dhemba, Gumbo and Nvamusara (2002).26 We do acknowledge that not all microinsurance providers are formal private actors (please see note in Table 1 formore information).27 Per definition, microinsurance is targeted at low-income populations (Churchill 2006).28 BMZ (2011).29 BMZ (2011).30 BMZ (2011, p.2). 6
  7. 7. Please do not cite or quote without written permission of the authorsas close monitoring are needed to make sure that private actors can effectively complementgovernments in social protection.The private sector and the public sector can also usefully engage in partnerships in the areas ofpension. Many developing countries have started to develop ―hybrid pension systems‖, a mix ofpublic and private pensions. One example is Chile. Introduced in 1981, the private pension planhelped to accumulate a large pool of capital but had the downside of only benefiting those whopaid contributions (i.e. salaried workers in the formal sector).31 In 2008, the Chilean governmentexpanded public pensions to groups left out by the private pension system (i.e. the poor andinformal workers).32 About two-thirds of Chilean pension income will be paid from pre-fundedretirement accounts, one-third will be paid from tax-financed public benefits.33 A recent impactassessment of the 2008 reform‘s most important component, the new Basic Solidarity Pension(Pension Bá sica Solidaria) aimed at poor individuals aged 65 and older finds that targetedhouseholds (poor households with at least one person age 65 and older) received about 2.4percent more annual household income and were able to improve their welfare. 34 The studyfinds little evidence that the Basic Solidarity Pension led to a crowding-out of private transfers.35While the 2008 reforms have brought important improvements, a large share of the informalsector remains outside Chile‘s pension system.36Table 1 provides an illustrative overview of social protection arrangements (i.e. informal, semi-formal and formal), the actors involved in these arrangements, the three main areas of socialprotection interventions as well as examples of initiatives that fall under these main areas in therespective types of arrangement.37 For instance, utility subsidies are an example of a specificsocial protection initiative in the area of social assistance which is usually placed in the formal(public) arrangement category. A caveat regarding our overview that needs to be mentioned isthat borders between categories are often not clear cut. The line between informal and semi-formal social protection, for instance, is a fine one as is the line between semi-formal andformal.38 Thus, social protection initiatives can fall under more than one arrangement (sometimeseven with the involvement of external donors and other external actors).39While important, informal and semi-formal social protection arrangements face a number oflimitations which we will analyze in the next section.31 Gallardo (2008).32 Gallardo (2008).33 James, Cox Edwards and Iglesias (2010, p.25).34 Behrman et al. (2011, p.2) and Shelton (2012).35 Behrman et al. (2011).36 Shelton (2012).37 This compilation is merely meant to be illustrative and to provide an overview to the reader.38 There is no clear categorization of and/or distinction between semi-formal and informal arrangements of socialprotection (e.g. ADB (2010) even considers NGOs to be formal actors in social protection arrangements).39 The government can, in some cases, collaborate with the private sector and/or civil society actors in socialprotection arrangements. 7
  8. 8. Please do not cite or quote without written permission of the authorsTable 1: Social Protection Arrangements, Actors and Examples of InitiativesType ofarrangement Actor Specific initiative under each area Social insurance Social assistance Labour market interventionsFormal public sector  health insurance  cash transfers  training programmes (government) (conditional cash transfers (CCTs) and unconditional cash transfers (UCTs))  disability and  in-kind transfers  skills development for invalidity workers insurance  life insurance  fee waivers  part-time unemployment benefit  public pension  utility subsidies  reintegration in the job market  employment counseling  changes to labour legislation  public works programmes  employment guarantee programmes private sector  microinsurance* (e.g. insurance (e.g. health, life, companies, accident insurance banks) targeted at low- income populations)  insurance policy or contract (e.g. private pension, health, life, accident)Semi-formal civil society  in-kind transfers to  skill training and organizations children education for poor, youth (CSOs)**  social welfare and unemployed (e.g. non- services for governmental women, the elderly organizations and the disabled (NGOs)***, church groups) mutual aid  burial insurance arrangements  savings and credit often at through community community groups level (e.g. 8
  9. 9. Please do not cite or quote without written permission of the authors burial societies, Rotating Credit and Savings Associations (ROSCAs)) Informal**** family/kinship  remittances / (blood- direct money related), transfer friends  gift exchange / in- kind exchangeSources: Own elaboration based on Dekker (2008), Dhemba, Gumbo and Nvamusara (2002), GSDRC (2012),Holzmann and Jø rgensen (2001), Mohanty (2011).* Churchill and Matul (2012) differentiate between formal providers of microinsurance (e.g. insurance companies),and what we call ―semi-formal‖ providers (e.g. cooperatives, community-based organizations, mutuals, friendlysocieties).** Devereux (2010, p.13) broadly defines civil society to include ―[…] trade unions, rights-based NGOs,representatives of special interest groups (women, children, pensioners, people with disabilities, people affected byHIV and AIDS, homeless people, youth) community-based organisations (CBOs) and faith-based organisations(FBOs), as well as activist academics and the independent media.‖*** While we do recognize that NGO‘s can also be international actors (e.g. Save the Children), we mostlyconsidered nationally based NGOs.**** We consider informal social protection to be solely based on individuals and households. However, we dorecognize the argument that communities are sometimes extended families.2.2. Limitations of Informal and Semi-formal Social ProtectionA major part of the world‘s population still relies on informal arrangements as the main source ofsocial protection. 40 Informal and semi-formal social protection, however, is present in mostdeveloping countries: it has developed to fill gaps at the community and family/household levelthat government policies have not been able (or willing) to address 41 or that markets have notmanaged to (or have not been willing to) reach.4240 Holzmann and Jø rgensen (2001); Hoogeveen et al. (2004). Comprehensive social protection systems exist in onlyone-third of countries, where 28 percent of the global population lives; however, most of these systems cover onlyworkers in formal employment ILO (2010a, p.33). According to the ILO (2010a, p.33), only around 20 percent ofthe global working-age population and their families have access to comprehensive social protection. It is roughlyestimated that somewhere between 20 percent and 60 percent of the global population has access to basic socialprotection only ILO (2010a, p.33). According to the ILO, those enjoying only a basic level of income security(guaranteeing income at the level of the poverty line) at all stages of the life cycle as well as access to essentialhealth services are considered to benefit from basic social protection, or i.e., the social protection floor (ILO 2010a,p.22).41 Dercon (2002).42 For an analysis of how to make credit markets more inclusive for the poor, please refer to Mendoza and Thelen(2008). 9
  10. 10. Please do not cite or quote without written permission of the authorsWhile being vital for a large share of the population, especially the most vulnerable, theseinformal and semi-formal arrangements can have two important downsides: first, they mayprovide inadequate protection since they might collapse or malfunction under certaincircumstances or since pay-outs might be too small. Factors that can contribute to themalfunctioning of informal and semi-formal arrangements include inherent characteristics ofthese types of arrangements themselves (like the mere size of the mutual assistance group/pool)and the complex nature of the risks they are not equipped to cover (diversification of the mutualassistance group/pool). Second, they may be exclusive of the poorest or discriminate certainsocial groups.First, in some cases, informal and semi-formal social protection arrangements provideinadequate protection. For instance, when exposed to a risk that affects everyone in the risk pool.Risks can affect whole communities/regions (covariate risks) or they can affect a particularindividual (idiosyncratic risks). 43 By definition, insurance does not work for covariate risks, order for an insurance scheme to work the risk pool needs to be sufficiently diversified so thata particular adverse event only affects a limited number of members of the risk pool. Thus, forexample in the case of community-based insurance schemes, if an adverse event affects thewhole community this social protection set up does not work. 44 Given the size of typicalcommunities, there are many risks that qualify as covariate risks. They include financial crises,other macro-economic shocks, or natural disasters. Likewise, many diseases that afflictdeveloping countries are also covariate risks. For instance, HIV/AIDS diminishes the earningpotential of entire communities and is found to have hindered capacity and operability ofinformal social protection mechanisms in Eastern and Southern Africa.45Naturally, the likelihood of a covariate risk increases the smaller the size of and the morehomogeneous the risk pool.46 Since some informal and semi-formal risk-coping arrangements aresmall in size, chances are high that a shock may hit the whole group. 47 Also, payouts might notbe particularly high if the insured group is small.48 An informal and semi-formal arrangement isalso found to be less effective in case it is a ―horizontal‖ (between two equally poor parties)arrangement as opposed to a ―vertical‖ (between a rich and a poor party) arrangement since thepoor parties both have less resources at their disposal.49 Hence, unless informal and semi-formalarrangements are able to transfer part of the risk to insurance markets outside the community,they are likely to collapse when experiencing a covariate risk. 50 Formal social protection43 The counterpart of covariate risks are idiosyncratic risks, e.g. risks that affect a particular individual in acommunity. Examples of idiosyncratic risks are illness, disability, theft, etc. Informal risk coping arrangements arefound to work better in the context of risks that only affect an individual in a community instead of affecting thewhole community (Dercon 2002).44 Dercon (2002).45 ILO (2001); UNICEF (2008).46 It should be noted that small informal groups with a high level of trust have also been found to achieve risksharing and punishment in case of non-compliance (Mendola 2010). One should assume that the insured risk was ofidiosyncratic nature.47 Bhattamishra and Barrett (2010).48 Bhattamishra and Barrett (2010).49 Devereux (1999).50 Dercon (2002) notes that communities or individuals can alternatively also revert to intertemporal transfers, likeusing individual or community-level savings. However, the poorest might not possess savings. 10
  11. 11. Please do not cite or quote without written permission of the authorsarrangements can usefully complement or strengthen community-based arrangements to accesscommercial reinsurance markets.51On the other hand, some semi-formal social protection schemes have found to be overwhelmedby their membership size. While broad coverage means that there is strong demand for the semi-formal arrangement, it can also overburden its administrative and managerial capacity. Someburial societies in South Africa, for instance, have grown so big that they could possibly improvetheir service were they to be more regulated.52Second, informal and/or semi-formal social protection arrangements, in some cases, can beexclusive of the poorest or discriminative of certain social groups and minorities. For instance,they are exclusive of the very poor53 when they require contributions in cash or in kind which thepoorest are unable to pay. The poorest households in Fiji Islands and Vanuatu, themselves oftenpart of a family network of equally poor people, have been found to oftentimes be unable torevert to family support or to honor their contribution commitments to the community, church,etc. 54 Another barrier preventing people from accessing these arrangements can bediscrimination (for instance, based on ethnic background, migrant status, religion or gender).55More formal, more equitable social protection arrangements could help extend social protectionto the poorest and those excluded by discrimination.56While important, informal and semi-formal arrangements have many downsides as they canreinforce the poor‘s dependent status without making them resilient to (possible) frequentshocks.57 For instance, in the absence of formal insurance markets and social protection, farmersin Sub-Saharan Africa have been found to improvise measures like using mixed croppingsystems and planting multiple varieties to reduce the impact of shocks on their harvest. The pricethey pay is high: more work and lower yields.58 Governments may usefully strengthen informalarrangements in a given country context but may consider providing more formal solutions inothers.592.3. Potential Triggers for a Switch to More Formal Social ProtectionAs mentioned in the previous section, informal and semi-formal social protection arrangementsmay have a number of limitations that can lead to inadequate protection or non-inclusiveprovision. That may not in itself be a reason for a government to initiate a new formal socialprotection scheme or to broaden coverage of an existing social protection system. Our findingssuggest that a broad number of factors may potentially trigger a developing country to initiate amore formal social protection system by creating an enabling environment. These factors fall51 Bhattamishra and Barrett (2010).52 Bester et al. (2004); Olivier, Kaseke and Mpedi (2008).53 Bhattamishra and Barrett (2010); Dercon (2002).54 ADB (2010).55 Cook (2009).56 Bhattamishra and Barrett (2010).57 Cook and Kabeer (2009).58 UNDP (2012a).59 Cook and Kabeer (2009). 11
  12. 12. Please do not cite or quote without written permission of the authorsinto two broad categories: internal triggers and external triggers. Internal triggers can be relatedto the limitations of existing informal and semi-formal social protection schemes andarrangements themselves or to a change in a country‘s political and economic context which inturn can influence the development path of a country (i.e. change in wealth level, populationrelated change), or boost public support and political will. External triggers include change inthe nature of shocks (i.e. from idiosyncratic to covariate), economic crises, natural disasters, etc.Many governments have moved in the direction of scaling up existing or implementing newformal social protection schemes when such a potential trigger is present. If not addressed in atimely fashion and with the right responses, vulnerabilities that are exposed in economic crisesand through other triggers may threaten growth prospects, hard-earned development gains andsocial stability. In this section we explore which factors appear to potentially trigger countries tomove towards a more modern, more efficient (and ultimately towards a more formalized) socialprotection system.2.3.1. Potential Internal Triggers for a Switch to More Formal Social ProtectionA change of a country‘s political or economic context may provide a trigger for broadeningformal social protection coverage. For instance, many formerly planned economies startedbroadening public social assistance coverage to previously excluded groups when switching tomarket economy systems. 60 Apart from the regime change itself, liberalization, increasedexternal competition and pressure on public accounts are factors that are found to have pushedtowards an extension of social protection.61 The Republic of Korea, for instance, experienced aperiod of extremely rapid growth brought about by industrialization. One of the least developedcountries in the 1960s, by 1995, the Republic of Korea was classified as an upper middle-incomecountry and today is classified as a high income country.62With increasing wealth, the Republic of Korea started to gradually implement social protectioninitially following the model of Western welfare states using the social insurance model coveringformal sector workers. The Republic of Korea‘s four main social insurance programmes havebeen progressively built up since the mid-1960s: Industrial Accident Compensation Insurance(1964), Medical Insurance (1977), National Pension Insurance (1988) and EmploymentInsurance (1995).63 Increasingly, the Republic of Korea‘s the development path on which theRepublic of Korea embarked, started to challenge the country‘s existing social protection system.As the Republic of Korea moved up the development ladder, it started facing different andevolving challenges like expanding and/or aging population, fast urban migration, and increasingdemand for more equal access to quality social services. The fast aging of the Republic ofKorea‘s population poses challenges to the country‘s pension system. By 2040, fewer than twopeople of working age in the Republic of Korea are expected to support every person age 65 andolder.64 One innovative approach the Republic of Korea has taken to address this problem is to60 Cook and Kabeer (2009).61 Cook and Kabeer (2009).62 Jung and Shin (2002, p.270); World Bank (2012).63 Jung and Shin (2002, p.270).64 The Economist (2012). 12
  13. 13. Please do not cite or quote without written permission of the authorsprovide subsidies for the employment of the elderly so they continue working and providing forthemselves for longer. 65 As of 2008, elderly people in the Republic of Korea also have auniversal basic pension, earned-income tax credit and an insurance scheme providing long-termcare at their disposal, amongst others.66 Similar challenges are also faced in the developing Asiaregion more broadly where informal social protection mechanisms are eroding due to changes inworking habits, cultural values, family structures and urbanization.67In Latin America, the change in the political context was marked by the democratization in thelate 1980s, early 1990s of many countries in the region. Democratization is said to havecontributed to scaling up bottom up public support for social protection in Latin America.68 Astudy cites ―strong popular demand for social protection‖ 69 as an enabling environment for thebirth of a generation of ―highly innovative, domestically designed poverty and vulnerabilityreduction programmes‖ 70 in Latin America (e.g. Bolsa Escola/Famí lia (Brazil),Progresa/Oportunidades (Mexico), and Chile Solidario (Chile)). Brazil is an example of acountry where the government, in collaboration and pushed by civil society, has shown strongsupport for social programmes. The end of the military dictatorship in Brazil in 1985 wasaccompanied by an increasing occupation of the public arena by civil society and various grass-roots movements which brought the concerns of the people to the country‘s agenda. Thesemovements played a key role in the establishment of a new Constitution. With the approval ofthe Constitution in 1988 ―a new landmark point was established as a universalized social securitymodel came to life that was grounded in citizenship rights.‖71While bottom up support by the citizens was a key trigger for Brazil‘s move towards moreformal social protection, political will at the federal level was the key driver to broadencoverage, efficiency and areas of social protection interventions. The outspoken support of socialprogrammes by a new government put social protection high on the agenda of the federalgovernment in the early 2000s; the current government continued and deepened this course.Investing heavily in social as well as economic development, Brazil‘s approach to socialprotection demonstrates that the government is treating poverty as a multi-dimensional problemthat goes well beyond the lack of income.72 Since the early 2000s, the government has extended,designed and implemented a number of integrated programmes for social protection, extensionof basic services, and food security that have helped break vicious circles of social exclusion,lack of opportunity, low incomes, and poor health. 73Bottom up support through civil society, citizens and even the media are also said to have servedas a driving force in some successful African country cases. 74 In South Africa, civil society,through media campaigns, advocacy and street protests and even judicial proceedings, has65 The Economist (2012).66 The Economist (2012).67 AusAid (2012).68 Barrientos and Hulme (2008).69 Barrientos and Hulme (2008, p.9).70 Barrientos and Hulme (2008, p.9).71 UNDP and ILO (2011, p.66).72 UNDP (2012b).73 UNDP (2012b).74 Devereux (2010). 13
  14. 14. Please do not cite or quote without written permission of the authorsdemanded changes towards more equal and inclusive social protection policies. 75 Part of thesuccess of civil society in South Africa was due to the fact that it used jurisdiction to claim theright to social assistance (i.e. this right is written in the South African Constitution). 76Whilesocial protection systems in LICs in Eastern and Southern Africa are still in the initial stages ofdevelopment, they are more advanced in MICs in the region.77 Successful upscaling of socialprotection in African countries is found to be linked to strong political will at the federal level (asopposed to an agenda of donors).78 South Africa is said to avail of the most comprehensive socialprotection system in Sub-Saharan Africa, also to a great extent thanks to political will.79Lastly, transition from conflict toward a more stable, accountable regime can be a potentialtrigger for a government to implement or broaden social protection schemes. For instance, someLICs, after exiting from conflict, broadened their social protection coverage. 80 In Liberia andSierra Leone the governments invested in labour market interventions aimed at the young. 81Social protection in post-conflict countries helped the government to build trust though theprovision of support and opportunities to its citizens. Social protection was also a key instrumentin building a more peaceful and equal nation after the end of the Apartheid regime in SouthAfrica.82 The South African government broadened its formal social protection system primarilyfocusing on child grants and non-contributory pensions and with the help of labour marketpolicies bringing more people into jobs.832.3.2. Potential External Triggers for a Switch to More Formal Social ProtectionAs discussed above, the change in the nature of shocks that households experience has an impacton the functioning of informal social protection arrangements. For example, while householdsand individuals in the South-East and East Asia region have historically experienced shocks thatare mostly of idiosyncratic nature, the increasing integration of the regional economy impliesthat economic shocks increasingly affect a broader range of people, challenging informalarrangements.84 This situation can trigger the need for more formal, more stable social protectionsystems.Economic crises have proved to be a testing moment for the adequacy of a country‘s socialprotection system as they tend to expose the system‘s flaws and weaknesses. In the case of LatinAmerica, the economic crisis in the early 1980s is often cited as a trigger to a switch to a moreformal and systematic approach to social protection.85 Prior to the crisis, like in many developingworld regions, social protection in Latin America was mostly limited to workers in the formal75 Devereux (2010).76 Devereux (2010).77 UNICEF (2008).78 Devereux (2010).79 Devereux (2010).80 Cook and Kabeer (2009).81 Alderman andYemtsov (2012).82 Alderman andYemtsov (2012).83 Du Toit and Neves (2009).84 Cook (2009); Sumarto and Bazzi (2011).85 Barrientos and Hulme (2008). 14
  15. 15. Please do not cite or quote without written permission of the authorssector.86 The crisis brought about structural adjustment and economic liberalization which in turnwere followed by a rise in poverty and inequality. Initially, no broad social protection reformsfollowed (only social insurance institutions for the formally employed were reformed andfragmented social assistance programmes and social funds were mounted). The step for morecomprehensive and permanent public responses only started in the mid 1990s.87Another example is the Mexican economic crisis of 1994 (known as the ―Tequila Crisis‖) thatmotivated a change of the social protection system in Mexico. 88 The crisis exposed theinsufficiency of the existing mechanisms to protect the poor, so the new incoming administrationembarked on a two-pronged approach between (modestly) increasing support under existingmechanisms and slowly replacing the old system with new programmes. 89 Mexico‘s well-knownconditional cash-transfer programme Progresa-Oportunidades has its roots in the change createdby the Mexican economic crisis of 1994.90 And while the recent global economic and financialcrisis triggered the expansion of existing social protection arrangements (including of theOportunidades programme) it also led to the creation of a more comprehensive approach tosocial protection, the Vivir Mejor (Live Better) strategy, which is essentially a Social ProtectionFloor.91The Asian crisis of 1997-1998 is another example. Despite earlier reforms, the Asian crisis hitthe Republic of Korea hard; growth plummeted, unemployment rates soared and poverty levelsincreased. 92 Exposing the limitations of informal arrangements in those East Asian countrieswhose systems were based on ‗Confucian familism‘ (characterized by a ‗smaller government‘that provided less social protection for its citizens, relying on the family as a provider instead)the Asian crisis turned out to be strong trigger for more formal social protection systems.93 Priorto the Asian crisis, in the Republic of Korea, for instance, welfare services (e.g. health care andpensions) were delivered by state-owned corporations rather than the government. Furthermore,people between 18 and 65 years of age were not entitled to receive public assistance. 94 After theAsian crisis social assistance programmes like a cash transfer programme to the poor, even ifthey were able to work, were introduced.95 Initial evidence suggests that the social protectionreforms implemented after the Asian crisis have helped to shield the Korean people from thedetrimental impact of the recent global economic and financial crisis.96The global economic and financial crisis of 2008 may also have pushed many Latin Americancountries to start a second round of reforms of their pension systems. After the first round ofreforms in the 1980s, the second round of reforms is found to pay more attention to tackling86 Barrientos and Hulme (2008).87 Barrientos and Hulme (2008).88 Levy (2006).89 Levy (2006).90 Levy (2006).91 UNDP and ILO (2011).92 World Bank (2012) and Cook (2009).94 The Economist (2012).95 The Economist (2012).96 15
  16. 16. Please do not cite or quote without written permission of the authorspoverty of the elderly, to expand coverage and equity to shield people from market risks. Hence,non-contributory schemes like social pension and more universal pensions have been placedhigher on the agenda.97The case of Indonesia serves as a striking example of a country whose steady path to economicgrowth and poverty reduction was interrupted by an external trigger, the Asian crisis of 1997-1998 (see Box 1).Box 1: Time to Act When a Trigger Exposes Vulnerabilities and Limitations: The Case ofIndonesiaThe Asian crisis of 1997-1998 hit Indonesia at a time when it had experiences around 30 years ofstrong economic growth and poverty reduction. When the crisis hit, GDP growth plummeted andpoverty rates increased, showing that the poor were hard hit by the crisis. The financial andeconomic crisis soon became a political crisis, forcing the head of government at the time to stepdown. By early 1998 the country was suffering from the combined effects of financial,economic, and political crises. Exposing the insufficient protection of a large share of thepopulation to such a shock, Indonesia‘s government had to react swiftly, implementing socialassistance programmes to shield the most vulnerable. What started as formally provided socialassistance has developed into the beginning of a social protection floor for all which in Indonesiaincludes components in the areas of health care (JAMKESMAS scheme targeted at the poor andnear-poor (76.4 million people), universal health insurance coverage is envisioned by 2014),food security (subsidized ―rice for the poor‖ programme), access to education (scholarships forstudents from poor families), UCTs, CCTs (Programme Keluarga Harapan or PKH) and aCommunity Empowerment Programme (PNPM).Sources: World Bank (2012) and social protection should usefully be in place once a country is hit by a shock (ex ante),governments have increasingly been encouraged by the international community and by exampleof their peers (e.g. Brazil, South Africa are role models in social protection for many otherdeveloping countries) to invest in social protection even in the absence of shocks and othertriggers.2.4. Embedding Social Protection in a National Development StrategyWe previously identified limitations of non-formal social protection. Vital for a large share of thepopulation in developing countries and a useful source of information on the local context, theseexisting arrangements might usefully be made part of a country‘s social protection strategy.98 Wealso identified a number of potential triggers that may contribute to creating an enablingenvironment for a country to implement a more formal system and a more broad-based coverageof social protection. Building on this information, we will present our findings on enablingconditions that have helped developing countries to integrate social protection into their nationaldevelopment strategies and on recommendations that developing countries may with to consider97 Calvo, Bertranou and Bertranou (2010).98 Hoogeveen et al. (2004); Olivier, Kaseke and Mpedi (2008). 16
  17. 17. Please do not cite or quote without written permission of the authorswhen embedding existing non-formal social protection arrangements into their developmentstrategies.Public support has been found to be a strong driving force behind the successful social protectioncountry cases Brazil and South Africa.99 Both cases demonstrate that the formalization of socialprotection initiatives, through writing them into law, makes it more likely that they will be takento scale or institutionalized.100 According to Pero and Szerman (2005): ―The New [Brazilian] Constitution was ambitious: it settled social-democrat guidelines for social policy, stressing the universality of coverage and benefits, thus opposing the patterns prevailing until the 1970s. […] the use of selectivity criteria to distribute benefits to the most needy was also introduced. Furthermore, the Constitution deepened the ongoing decentralization process, strengthening the fiscal and administrative autonomy of sub-national governments.‖101One of the aspects pushed by Brazil‘s new Constitution was the decentralization of spending andbetter targeting of social expenditure for those who needed it most. 102 Despite high socialspending, social indicators in Brazil deteriorated further throughout the 1980s 103 and firstdetermined steps to breaking the inability of social spending to reduce poverty and inequality inBrazil and towards implementing a new social development strategy were only adopted by a newgovernment as of 1995. 104 Improvements in social protection spending, policy design andimplementation in Brazil owe much to partnerships between the federal government, sub-national governments, civil society and the private sector. An important building block ofBrazil‘s Bolsa Famí (the Bolsa Escola programme) was first developed and implemented at liathe municipal level before being scaled up to the national level.105 The partnership with civilsociety has helped the Brazilian government to improve the accuracy of Bolsa Famí lia’s 106beneficiary registry, and hence its targeting accuracy over time.Strong political will at the federal level is also a key prerequisite for the successful expansion ofsocial protection. A review 107 of social protection in Southern Africa concludes that socialprotection interventions have higher chances of succeeding if they are driven by political will,i.e. if they are government-led from the beginning than if they are donor-driven. For instance,successful social pension schemes for all older citizens were introduced in Lesotho (2004) andSwaziland (2005).108 These schemes were designed and implemented without donor support.109In general, in Southern Africa, government-led SP systems in South Africa, Botswana, Namibia,99 Devereux (2011); UNDP (2012a).100 Devereux (2010).101 Pero and Szerman (2005, p.5).102 Pero and Szerman (2005).103 Brazil‘s GINI index peaked in 1989 at 63, making Brazil one of the most unequal societies in the world (WorldBank 2012).104 World Bank (1988) as quoted in Pero and Szerman (2005). For detailed information on Brazil‘s socialdevelopment strategy under Cardoso, please refer to Faria (2002).105 De Janvry (2005); Pero and Szerman (2005).106 Lindert et al.(2007).107 Devereux (2010).108 Devereux (2010).109 Devereux (2010). 17
  18. 18. Please do not cite or quote without written permission of the authorsand (until recently) Zimbabwe are found to be more successful than donor-led SP systems inLesotho, Malawi, Mozambique, Swaziland and Zambia.110In some cases, government-donor or public-private partnerships are usefully implemented tocomplement capacities. For instance, in Ethiopia and Malawi, index-based weather insurancescheme pilots have been set up for farmers. While donors pay the insurance premiums,transferring the cost of droughts to the international insurance markets, national governmentsprovide weather station infrastructure, foster an appropriate legal and regulatory environmentand educate farmers on insurance matters. 111 In Nepal, external actors like internationalorganizations and bilateral aid agencies are found to play a strong role in the country‘s socialprotection context. Over the past 20 years, these agents are found to have contributed withknowledge and expertise to promotion, design and implementation of social protection in Nepalwhile the government has been primarily responsible for financing and administering socialprotection initiatives.112The public sector might usefully engage existing schemes when considering scaling up formalsocial protection systems. This is also the approach that the Social Protection Floor promoted bythe UN-system is taking. According to UNDP and ILO (2011): ―A key strength of the social protection floor approach is that it does not start from scratch but with a careful analysis and stocktaking of existing structures and strengths and weaknesses of schemes and programmes in place. Building on the national social protection system by improving coordination of different activities, exploring synergies and increasing efficiency will free resources for extending social protection to those currently not covered.‖113Governments are well-advised to carefully analyze existing non-formal social protectionarrangements when considering to implement new or to scale up existing formal socialprotection schemes. 114 This is due to the fact that a randomly implemented formal socialprotection scheme might do more harm than good. For instance, the implementation of a formalfood for work programme may incentivize able bodied individuals to drop out of informalinsurance arrangements. This might leave the informal arrangement with a less diversified riskpool of less able bodied individuals (i.e. the elderly).115Integration should ultimately aim at serving the purpose of increasing the extension of coverageand/or providing a minimum level of protection.116 Prior to setting up a new formal system orextending an existing one, the reasons behind the existence of informal social securityarrangements as well as the nature of the relationship between current informal and formalarrangements need to be analyzed and understood. In order to be considered worthy candidates110 Devereux (2010).111 UNDP (2012a).112 Upreti et al. (2012, p.39).113 UNDP and ILO (2011, p.15).114 Hoogeveen et al. (2004); Olivier, Kaseke and Mpedi (2008).115 Hoogeveen et al. (2004).116 This paragraph is based on Olivier, Kaseke and Mpedi (2008). It should be noted that the original study talksabout ―social security‖, not social protection. 18
  19. 19. Please do not cite or quote without written permission of the authorsfor integration, informal arrangement should be able to meet a number of criteria that includeprudent management, capacity financial viability and sustainability. Once an informalarrangement is found suitable to be scaled up and/ or integrated, governments can avail of anarray of government interventions to integrate them into, or link them to formal provision. Theseinterventions include training, subsidies, technical assistance, etc.For instance, an assessment in the Southern African Development Community (SADC) region117concludes that self-organized mutual support systems (at community level) in the SADC regionlend themselves better to being incorporated into social security systems than traditional supportsystems which are often rooted in African traditional values.118 Burial societies in Ethiopia andTanzania, may also be suitable candidates for linking them to additional types of insurance sincethey are often relatively formal in nature.119 However, evidence on how, why and when informaland semi-formal arrangements have been taken into account in the process of broadeningcoverage of social protection systems is scarce. More information could contribute to more andbetter informed decisions.Finally, social protection systems should be judged and measured against its efficiency ofcontributing to more equal societies, to more security for people (especially the poor and themost vulnerable), more stable and inclusive growth. The government is the actor that has theauthority to set the legislative and regulatory framework for other social protection providers tothrive and to assure that social protection is equitable and inclusive. Governments can andshould, usefully engage in close cooperation with actors that are well-informed about localconditions and public needs (like NGOs) and actors that might be able to improve efficiency(like the private sector). 120 Governments in many developing countries have partnered withprivate insurance companies to extend, assure (and provide) pension.1213. Cost Efficiency of Social ProtectionThe affordability of social protection schemes is a major concern for many developing countries,in particular the LICs. For a long time, the prevailing view was that social protection was notaffordable in developing countries. It was mostly believed that social expenditures wereunproductive measures that would crowd out private investment and create large fiscal deficitsthat would not be manageable.122In this section we will explore cost-efficiency of formal social protection programmes. In thiscontext, it is important to highlight differences between cost-efficiency and cost-effectiveness asthese two similar concepts can often lead to confusion and mistakenly can be used117 Current members of SADC are: Angola, Botswana, Democratic Republic of Congo (DRC), Lesotho,Madagascar, Malawi, Mauritius, Mozambique, Namibia, Seychelles, South Africa, Swaziland, United Republic ofTanzania, Zambia and Zimbabwe. For more information, see Olivier, Kaseke and Mpedi (2008).119 Dercon et al. (2004).120 For instance, building the link between the government and the people, NGOs in the Pacific Island Countries arefound to contribute to provide advocacy services, counseling, education and training (Mohanty 2011).121 Tapia (2008).122 Cook and Kabeer (2010, p. 7); Ortiz and Yablonski (2011, p. 51). 19
  20. 20. Please do not cite or quote without written permission of the authorsinterchangeably. Whereas the main objective for cost-effectiveness is to achieve results at alower cost compared with alternatives, for cost-efficiency the main objective is ―to achieve themaximum possible outputs, outcomes, and impacts with the minimum possible inputs.‖123Therefore, cost-efficiency of a particular social protection programme will not depend solely onthe level of financial sources allocated to it but also, and most importantly, on achieving thedesired impact (e.g. improvements in human development outcomes, mitigation of risks andvulnerabilities) based on country-specific development priorities. The concept of cost-efficiencyplaces more emphasis on impacts than the concept of cost-effectiveness does and in turn takesinto account a broader set of costs and benefits.There are various costs associated with social protection that can be broadly grouped into directand indirect costs. Direct costs include those associated with setting up, implementing, andrevising (as appropriate) a social protection programme. Indirect costs include both opportunitycosts of not allocating resources to other programmes and negative externalities on non-beneficiaries and/or local economy.Besides the costs, affordability of a social protection scheme needs to be evaluated withconsideration of both the direct and indirect benefits 124 and impacts on people and on thesustainability of the economy as well as the contribution to building resilience to shocks of sucha scheme. Since it is not feasible to quantify all costs and benefits involved, qualitative analysesoffer an important complementary role to quantitative analyses.A careful needs assessment would enable a country to identify the most prevailing and urgentvulnerabilities that need to be addressed by a social protection programme.125 After determiningthe priority areas for action, the main challenges lie in ensuring that social protectionprogrammes are well designed, help achieve desired outcomes based on country priorities, areaffordable, and do not lead to costly entitlements that a country cannot sustain over time.Therefore, a comprehensive cost-benefit analysis for a social protection programme based oncountry priorities would be the one that would explore not only current affordability but alsofiscal sustainability.However, comprehensive cost-benefit analyses of social protection programmes are rare indeveloping countries mainly because of time and capacity constraints to assess social protectionprogrammes. Another reason why comprehensive cost-benefit analyses are not common indeveloping countries could be political if those in power aim at reaping political benefits ofsocial protection programmes within limited timeframes (e.g. before elections).It should be noted that with regard to certain social protection schemes, absence of cost-benefitanalyses (both ex-ante and ex-post) are not unusual even in advanced countries. For instance, a123 IEG (2007, p. 65). For more comprehensive definitions of and discussions on efficiency and cost-effectiveness,see Chapter 11 of IEG (2007).124 In this paper, the concept ―benefit‖ refers to all positive impacts of a social protection program and should not beconfused with the ―benefit amount‖ of a particular scheme.125 In this context, the ILO‘s Rapid Assessment Protocol (RAP) provides a useful tool for developing countries totake stock of and map existing social protection measures and identify priority areas for intervention (UNICEF andILO, 2011, p. 11). 20
  21. 21. Please do not cite or quote without written permission of the authors2007 inventory of youth employment interventions in 84 countries from all regions in the worlddemonstrated that among the OECD countries, only Canada, the U.K. and the U.S conductedimpact evaluations and cost-benefit analyses of such interventions.126 Outside these countries andother than studies sponsored by international organizations, rigorous evaluations of youthemployment interventions were found to be very rare. 127 The ILO has recently adopted aresolution to call on governments, social partners and the multilateral system to take urgent andrenewed action to address the crisis of youth employment. 128 This resolution also stressed thatthe ILO‘s own activities promoting youth employment ―should be subject to rigorous monitoringand evaluation to ensure approaches are cost-effective and provide a positive impact.‖129One example of a developing country that has been undertaking social protection cost-benefitanalyses is Cambodia. The Government of Cambodia recently developed a social protectionstrategy based on a vulnerability and gap analysis; consultations with development partners andother stakeholders; as well as technical assistance from various International Organizationsincluding for costing exercises. Cambodia‘s strategy stresses that ―financing of the socialprotection programme must be seen as an investment rather than as an expenditure.‖130 Box 2summarizes how Cambodia‘s social protection strategy has been evolving in order to provide thereader an overview of possible issues that developing countries can encounter while expandingsocial protection and exploring affordability of priority programmes.Box 2: Cambodia’s National Social Protection Strategy for the Poor and VulnerableIn 2011, the Government of Cambodia adopted its National Social Protection Strategy for thePoor and Vulnerable (NSPS). Before developing this Strategy, the Government first identifiedgaps and constraints with regard to the effective and efficient provision of social protection suchas lack of longer-term vision for social assistance development; low local capacity; limitedcoordination among social protection interventions; problems with collecting and monitoringdata and assessing existing interventions; and inadequate budget for implementation.The Government of Cambodia stresses that ―limited fiscal space and implementation capacitiescall for prioritisation of options for social protection development in the short term.‖ The NSPSgives priority to addressing major sources of vulnerability (such as chronic and transient poverty,hunger, shocks, and social exclusion) by taking short- and medium-term measures including cashand in-kind transfers and fee exemptions; public works programmes; and social welfare services.The Strategy also sets the long-term framework for sustainable and comprehensive socialprotection for all in accordance with the Social Protection Floor Initiative. The aim is toestablish both contributory social security mechanisms for the formal sector and improved socialassistance for the informal sector.While preparing the NSPS, the Government held technical consultations with developmentpartners and national stakeholders, such as civil society organizations. Several International126 Betcherman, et al. (2007, p. 31).127 Betcherman, et al. (2007, p. ii).128 ILO (2012a).129 ILO (2012a, p. 14).130 UNDP and ILO (2011, p. 156). 21
  22. 22. Please do not cite or quote without written permission of the authorsOrganizations including the World Food Programme (WFP) and the World Bank assisted theGovernment in undertaking a scoping and mapping exercise on existing safety-net programmes.The ILO also applied a diagnostic tool called Social Protection Expenditure and PerformanceReview (SPER) in order to assess system financing, to identify coverage gaps, and to discusspolicy issues for consideration by national policy makers. The ILO estimated that the totalexisting social expenditure for the year 2010, including ODA-funded programmes and subsidiesfor the health sector was at about 5.5 percent of GDP. The majority of social spending wasallocated for health, corresponding to about 60 percent of total social expenditure.As part of the technical consultations the World Bank conducted a study that estimated that aconditional cash transfer programme for poor families with pregnant mothers or children underfive would cost 0.21 percent of GDP (excluding certain administrative costs and supply-side costof providing nutrition services). Moreover, the NSPS provides preliminary cost estimates foreach short-term (up to 2013) priority programmes and actions but exact resource requirementswill be determined through further analyses and assessments. The Strategy states that ―a costingexercise for the medium- and long-term implementation of the NSPS will be developed as apriority activity during the first year of implementation (including a detailed costing of existingand planned interventions and a fiscal space analysis). Financing arrangements, including jointpool arrangements for certain tasks, will be discussed with development partners to embark on aprogramme-based approach for social protection in Cambodia and to align and harmonise donorsupport for the NSPS.‖ It is also stressed that during this process financing of the socialprotection programme must be seen as an investment rather than as an expenditure.Sources: ILO (2012b), Kingdom of Cambodia (2010), Royal Government of Cambodia (2011), UNDP and ILO(2011).In the following sub-sections, we will discuss cost-efficiency of social protection by taking stockof several costing exercises and analyzing potential short- and long-term benefits of socialprotection.3.1. Cost-Benefit AnalysesInstitutions such as HelpAge, International Labour Organization (ILO), Overseas DevelopmentInstitute (ODI), UNICEF, and World Health Organization (WHO) developed social protectioncosting models and/or undertook costing exercises. Some of these models have been applied byindividual countries in order to examine the feasibility of certain social protection programmes.At the same time, a number of institutions conducted broader studies looking into various sets ofcountries in order to contribute to the affordability debate.One widely cited example is the costing studies that the ILO undertook in seven sub-SaharanAfrican131 and five Asian countries132 estimating the annual costs of a basic social protection131 Burkina Faso, Cameroon, Ethiopia, Guinea, Kenya, Senegal, and the United Republic of Tanzania132 Bangladesh, India, Nepal, Pakistan, and Viet Nam 22
  23. 23. Please do not cite or quote without written permission of the authorspackage that includes universal basic old-age and disability pensions, 133 basic universal childbenefits, 134 universal access to essential health care, 135 and a 100 day employment guaranteescheme.136 For the countries considered, the annual cost of a basic social protection package isestimated to be in the range of 3.7 to 10.6 percent of GDP in 2010.137 The estimated annual costsof each element are as follows:138 Universal basic old-age and disability pensions: between 0.6 and 1.5 percent of GDP Basic universal child benefits: between 1.2 and 3.6 percent of GDP Universal access to essential health care: between 1.5 and 5.5 percent of GDP 100 day employment guarantee scheme: between 0.3 and 0.8 percent of GDPThe cost of providing universal health care is noticeably higher than the cost of the remainingelements of the package. The results of this ILO study have been cited in the literature139 with thepurpose of demonstrating that a basic social protection package excluding universal health careprovision would be affordable even in low-income countries (the selected group includes bothlow-income and lower middle-income countries).Nevertheless, affordability is not evident merely from these numbers. According to this ILOstudy, even when governments increase the share of public spending attributed to socialprotection to 20 percent of their total budget, seven out of twelve countries analyzed will still notbe able to fill the financing gap from domestic resources by 2030. 140 Moreover, these costs werecalculated before the global financial and economic crisis hit, therefore the crisis impact on fiscalspace available to developing countries was not taken into account.141Other costing exercises have focused on specific elements of a basic social protection packagelooking at different or a greater number of countries. Some of these exercises estimate only133 ―It was assumed that the simulated universal old-age and disability pension would be set at 30 percent of GDPper capita, with a maximum of one US dollar (PPP) per day (increased in line with inflation) and would be paid toall men and women aged 65 and older; and to persons with serious disabilities in working age (the eligibility ratiowas assumed to be 1 percent of the working-age population, which reflects a very conservative estimate of the rateof disability).‖ ILO (2008, p. 6).134 The level of the child benefit is assumed to be ―15 percent of GDP per capita with a maximum of half of one USdollar (PPP) per day (increased in line with inflation) and paid for up to two children under the age of 14 per womanwho has given birth. The rationale behind this assumption is to tackle claims that universal child benefits wouldprovide an incentive to increase fertility.‖ ILO (2008, p. 7).135 It was assumed that basic health care costs would be based on a ratio of 300 medical staff to 100,000 population,with medical staff wages indexed in line with GDP per capital growth (health staff wages were assumed at aminimum of three times GDP per capita) and overhead costs of 67 percent of staff costs. ILO (2008, p. 21).136 The assumed beneficiary group of the employment guarantee scheme constitutes 10 percent of the working-agepopulation in each country. ―The benefit is only available to households not benefiting from any other form of cashtransfer. It was assumed that the simulated employment scheme would provide a benefit set at 30 percent of GDPper capita, with a maximum of one US dollar (PPP) per day (increased in line with inflation). The benefit would bepaid for a total of 100 days in the year.‖ ILO (2008, p. 9).137 ILO (2008, p. 10).138 ILO (2008, pp. 6-9).139 ILO (2011); Ortiz and Yablonski (2011).140 ILO (2008, pp. 13-14). These seven countries are Burkina Faso, Cameroon, Ethiopia, Kenya, United Republic ofTanzania, Bangladesh, and Nepal.141 UN NGLS (2010, p. 18). 23
  24. 24. Please do not cite or quote without written permission of the authorscurrent costs, while others also provide estimates of future costs to discuss sustainability. Thefollowing sub-sections give an overview of these studies. It is worth highlighting that the aim isnot necessarily to provide comparisons (as methodologies, assumptions, and countries analyzeddiffer from one study to another) but instead to offer the reader a wide range of existing analyses.3.1.1. Universal Old-Age PensionsA HelpAge International study built on previous costing exercises by organizations such as theILO and estimated the cost of a universal old-age pension in 50 low- and middle-incomecountries.142 The study argues that universal old-age pensions in the countries analyzed would becurrently affordable (in 2010). It is estimated that a universal pension for everyone over 65143would cost less than 1.8 percent of GDP in all 50 countries (exceeding 1.5 percent of GDP onlyin China, Jamaica, Sri Lanka, and Thailand).144 Such scheme would cost around 1 percent ofGDP or less in most sub-Saharan African countries.145 Moreover, these costs would not surpass 8percent of current government expenditure in any of the 50 countries. 146 In 15 countries, thesecosts correspond to around or less than 2 percent of government expenditure (e.g. Burkina Faso,Malawi, Senegal, Mongolia, Ghana).147Moreover, in order to assess sustainability, this study projected the future costs of a universalold-age pension for everyone over 60 in Rwanda, Paraguay, and Thailand under differentscenarios. For example, when the pension is indexed to average income, the costs would riseover time in all three countries as populations age. As a percent of GDP, the costs wouldcorrespond to 1.4 in Rwanda, 3 in Paraguay, and 5.2 in Thailand by 2040.148 The reason behindthe higher cost for Thailand is that by 2040 a quarter of its population is projected to be over60.149 The study argues that governments can contain these costs by indexing the value of thetransfers to inflation and/or by increasing eligibility age as populations age and healthy lifeexpectancy increases.1503.1.2. Universal Health CareA WHO study estimated that providing key health services in 49 low-income countries wouldcost around USD 44 per capita on average in 2009, increasing to around USD 60 per capita by2015.151 This estimate includes the cost of interventions to achieve the health-related MDGs as142 Knox-Vydmanov (2011). HelpAge International calculations assume a transfer level of 20 percent of GDP percapita and set administrative costs as 5 percent of the total cost of transfers. (pp. 2-3).143 The study also provided separate cost estimates of pensions covering everyone over 60 and 70.144 Knox-Vydmanov (2011, Figure 1, p. 3).145 Knox-Vydmanov (2011, Figure 1, p. 3).146 Knox-Vydmanov (2011, Figure 2, p. 4).147 Knox-Vydmanov (2011, Figure 2, p. 4).148 Knox-Vydmanov (2011, p. 7).149 Knox-Vydmanov (2011, p. 7).150 Knox-Vydmanov (2011, p. 8).151 WHO (2010, p. 22). 24
  25. 25. Please do not cite or quote without written permission of the authorswell as those targeting noncommunicable diseases. 152 Cost estimates were made for eachcountry, and then aggregated; hence they are ―simply an (unweighted) average across the 49countries at the two points in time.‖153 Obviously, cost estimates vary by country. For instance,while five of the countries analyzed would need to spend more than USD 80 per capita in 2015,six countries would need to spend less than USD 40.154Current health spending varies substantially from one country to another. For instance, annualhealth spending in the US and Norway surpasses USD 7,000 per capita and OECD members as agroup spend on average around USD 3,600 per capita. 155 On the other hand, among WHO‘sMember States, 31 countries spend less than USD 35 per capita annually and four countriesspend less than USD 10, even when external aid is taken into account. 156 The WHO stresses thatthe poorest countries would need assistance from the international community to expand accessto health services since the Organization argues that ―even with relatively high levels of domesticgrowth, and national budgets that prioritize health, only eight of the 49 countries have anychance of financing the required level of services from domestic resources in 2015.‖1573.1.3. Child BenefitsA Save the Children UK study estimated the likely current costs of providing different types ofunconditional child benefits for a large sample that includes 57 developing countries. 158 Theaverage cost159 of providing universal child benefits for children under 5 is estimated to be 2.08percent of GDP. 160 When children under 5 who are below the poverty line are targeted, theaverage cost decreases to 1.28 percent of GDP.161While the average costs look relatively modest, the results of this study exhibit significantvariations across countries and regions. On a positive note, in many poor and middle-incomecountries in Asia the cost of a universal cash transfer for children under 5 would be less than 1.5percent of GDP.162 However, one exception is Nepal, where this cost is estimated to exceed 2.5152 More specifically, the study ―included interventions proven to reduce mortality among mothers, newborns andchildren under five; childbirth care; reproductive health services; prevention and treatment of the main infectiousdiseases; diagnosis, information, referral, and palliative care for any presenting conditions; and health promotion.‖WHO (2010, p. 38).153 WHO (2010, p. 23).154 WHO (2010, p. 23).155 WHO (2010, p. 21).156 WHO (2010, p. 21).157 WHO (2010, p. 23).158 Yablonski and O‘Donnell note that this is a static analysis of the likely current cost of child benefits andestimates will change over time according to the particular combination of changes in each country arising from:population growth, changes in poverty headcount, changes in average poverty gap, economic growth, potentialchanges in administrative costs over time (p. 44).159 The average cost has been calculated for 54 countries as Burundi, Liberia, and Democratic Republic of Congo aretreated as outliers.160 Yablonski and O‘Donnell (2009, Table 2, p. 26).161 Yablonski and O‘Donnell (2009, Table 2, p. 26).162 Yablonski and O‘Donnell (2009, Figure 6, p. 27). These countries include Bangladesh, China, Cambodia, India,Indonesia, Laos, and Pakistan. 25
  26. 26. Please do not cite or quote without written permission of the authorspercent of GDP. 163 Moreover, for LICs in Africa, a universal cash transfer for children under 5is found to be unaffordable in most cases. It is argued that some of these countries would needconsiderable external assistance to fill the gap. For instance, Liberia and Tanzania would needdonor funding equal to approximately 90–95 percent and 70-85 percent of costs, respectively.164Countries such as Sierra Leone, Niger and Mozambique currently are not able to afford even themore narrowly targeted options at national scale out of domestic resources.1653.1.4. Social Pensions and Child BenefitsA joint UNICEF and ODI study estimated the possible costs of social pensions and child benefits(universal and/or selective) in five West African countries. These simulations as a percentage ofboth GDP and recurrent expenditure are presented in Table 2 and vary significantly across thesefive countries. For instance, while the cost of a universal child benefit and social pensionprovision is estimated at 1.1 percent of GDP in Equatorial Guinea, the same provision for Ghanais estimated to cost 11.3 percent of GDP (which is considerably higher than the upper range ofrelated ILO estimates) corresponding to more than 60 percent of recurrent expenditure. Theestimated costs for Equatorial Guinea are substantially lower than the ones for Ghana becausethe former‘s per capita GDP is much higher due to its oil exports.166However, the study highlights that affordability in simple aggregate terms does not necessarilyimply feasibility of a programme. For instance, whereas the oil-rich Equatorial Guinea seems tohave necessary fiscal space to finance additional social protection expenditures, it may facepolitical and institutional challenges. In Section 4, we will discuss in more detail a range ofchallenges that countries may encounter in addition to financial constraints.163 Yablonski and O‘Donnell (2009, Figure 6, p. 27).164 Yablonski and O‘Donnell (2009, p. 27).165 Yablonski and O‘Donnell (2009, pp. 26-28). In such cases, the authors recommend rolling out a universalprogramme geographically in areas with the highest poverty rates and argue that ―gradual expansion by age orgeography will help to keep costs manageable, and allow time for building the systems and capacity necessary todeliver programmes at scale.‖ (p. 28).166 UNICEF and ODI (2009, p. 25). 26
  27. 27. Please do not cite or quote without written permission of the authorsTable 2: Annual Programme Expenditure Cost Estimates of Child Benefit and SocialPension Options: Simulations for Congo, Mali, Senegal, Equatorial Guinea and Ghana Congo, EquatorialCosts Republic Mali Senegal Guinea Ghana Pension Pension Pension Social Social Social UCB UCB UCB UCB UCB SCB SCB SCB% of GDP 2.0 1.2 1.0 5.9 3.2 6.4 3.7 0.9 0.2 8.7 2.6% of 16.7 9.9 8.3 42.8 23.5 30.0 17.6 20.8 5.0 46.3 13.9recurrentexpenditureNote: UCB: Universal Child Benefit, SCB: Selective Child Benefit.Source: UNICEF and ODI (2009, Table 2, p. 26).3.1.5. The Social Protection Floor Costing ToolIn addition to undertaking costing exercises that covered a selected group of countries, someinternational organizations have built costing tools that can be applied by individual countries.For example, following the adoption of the Social Protection Floor (SPF) Initiative, UNICEF andILO jointly developed the SPF Costing Tool in 2010. The objective of the tool is to help supportpolicy decisions regarding selection, revision and investment in social protection programmes byproviding an initial basic assessment of the potential costs of both new schemes andmodifications to existing ones.167 It should be noted that the tool is only adapted to estimating thecost and impact of cash transfers.168Whereas the SPF Costing Tool was developed in order to support the Social Protection FloorInitiative, the tool can be applied to other social protection contexts. The cost of a specific socialprotection scheme would of course vary across countries depending on demographic, labour, andmacroeconomic conditions. While the SPF tool allows users to enter these relevant data points,the tool has certain built-in assumptions about how these different parameters interact. Countriescan choose to conduct a simple application of the tool by using its built-in assumptions only (e.g.Senegal), or can carry out more elaborated costing exercises by modifying these assumptionsbased on their national context (e.g. Argentina and Egypt—See Box 3). 169 Some countries167 UNICEF and ILO (2011).168 The SPF Costing Tool can provide cost estimates for old-age pensions, child benefits, disability benefits, orphanbenefits, education stipends, birth lump-sum benefits, youth labour market programmes, and unemploymentprogrammes (UNICEF and ILO (2011, p. 5)). For a discussion of the SPF Costing Tool‘s limitations, see UNICEFand ILO (2011, pp. 9-10).169 UNICEF and ILO (2011, pp. 3-4). The countries that have conducted social protection costing exercises using theSPF Costing Tool include Argentina, Egypt, Madagascar, Mozambique, and Senegal. 27
  28. 28. Please do not cite or quote without written permission of the authorschoose to utilize multiple tools. For instance, Senegal‘s simple application of the SPF CostingTool was complemented by the World Bank‘s ADePT tool for estimating poverty impacts.170Box 3: Application of the SPF Costing Tool in EgyptA recent study applied the SPF Costing Tool for a proposed system of child cash transfers inEgypt, in order to estimate not only its costs but also poverty impacts. The study simulationsshow that with an overall cost of 0.88 percent of GDP, the proposed scheme can lift 19.3 percentof poor people in Egypt out of poverty according to the national poverty line. The potentialimpact on children is expected to be greater with an estimated poverty reduction of 28.2 percentamong poor children up to 14 years of age.While examining the sustainability of such a scheme, the study took population dynamics intoaccount. Egypt‘s demographic profile appears to be favourable because of the growing ratio ofworking age population in the overall population. Hence, over the period 2012-2020 the cost ofthe scheme is projected to fall as a percentage of GDP benefiting from the declining dependencyratio—regardless of whether the scheme‘s benefit amount is held constant in real terms or as apercentage of GDP per capita.Source: Rabi (2012).3.1.6. Rapid Assessment Protocol (RAP)On the basis of the SPF Costing Tool, the ILO (in close collaboration with UNICEF) hasrecently developed a new costing tool called the Rapid Assessment Protocol (RAP), whichprovides a useful method for developing countries to take stock of and map existing socialprotection measures and identify priority areas for intervention.171 The tool also provides ―a basisto discuss and simulate alternative financing options and fiscal space.‖172The first step in this exercise is to construct an SPF Rapid Assessment Matrix (see Figure 1) inorder to analyze the present and planned future social protection provisions according to thebenchmarks set by the four guarantees of the SPF and to identify gaps in policy design andimplementation.173170 UNICEF and ILO (2011).171 UNICEF and ILO (2011, p. 11).172 ILO and IMF (2012, p. 3).173 Bonnet et al. (2012, p. 7). 28