1 IntroductionThe rapid rise of social protection up thedevelopment policy agenda has been startling, evenalarming. Part of the explanation is to be found in agrowing recognition that social protection can befunctional to the achievement of biggerdevelopment objectives, including even economicgrowth and the Millennium Development Goals(MDGs). Mounting evidence that well-designed socialtransfers can contribute to poverty reduction isappealing to development economists andpolicymakers who were previously sceptical aboutsocial protection’s unloved parents, ‘social safetynets’, which were disparaged during the 1990s aspolitically expedient, socially stigmatising and fiscallyunaffordable in poor countries. The ‘safety nets’agenda was vigorously attacked equally from the left(for its social residualism and political cynicism) asfrom the right (for displacing informal social securitymechanisms and creating dependency onunsustainable handouts) until it curled up and died,only to be triumphantly reborn as ‘social protection’around the turn of the millennium.

So what’s the difference? For one thing, the newsocial protection agenda comes with a fresh array ofconceptual frameworks, analytical tools, empiricalevidence, national policy processes, heavyweightagencies and big names in development studiesaligned behind it. Secondly, social protection appearsto be equally amenable to appropriation by the ‘right’(who are now inviting the poor to participate ineconomic growth opportunities with revitalisinginjections of targeted transfers) and the ‘left’ (who arehooking their ‘rights-based approaches’ onto thesocial protection bandwagon). Thirdly, the socialprotection agenda prioritises moving people fromdependency into productive livelihoods, whereverpossible. This concept of ‘graduation’ distinguishes

social protection from the narrower set of purelywelfarist measures, which are just a part of thetoolkit.

Advocates for social protection fall into two broadcamps, which we will label the ‘instrumentalists’ andthe ‘activists’. ‘Instrumentalist’ arguments point tothe dysfunctionality of extreme poverty, inequality,risk and vulnerability to the achievement ofdevelopment targets on which there is broadconsensus, currently embodied in the MDGs. For this‘social protection for efficient development’ camp,social protection is about putting in place riskmanagement mechanisms that will compensate forincomplete or missing insurance (and other) markets,until such time as poverty reduction and marketdeepening allow private insurance to play a moreprominent role.

‘Activist’ arguments view the persistence of extremepoverty, inequality and vulnerability as symptoms ofsocial injustice and structural inequity, and campaignfor social protection as an inviolable right ofcitizenship. For these ‘social protection for socialjustice’ placard-wavers, targeted welfarist handoutsare a necessary but perhaps uncomfortableintermediate step between ad hoc humanitarianismand the ideal of a guaranteed ‘universal socialminimum’, where entitlement extends far beyondcash or food transfers and is based on citizenship, notphilanthropy or enlightened self-interest.

Intriguingly, these ideological battles are beingfought out around basic and seemingly technicalchoices in social protection policies and programmes:choices of instruments (social pensions, public works,etc.) and choices in design (targeting vs. universalprovision, cash vs. food aid, conditional vs.unconditional transfers). Where a convergence

Editorial Introduction: DebatingSocial Protection

Stephen Devereux and Rachel Sabates-Wheeler

IDS Bulletin Volume 38 Number 3 May 2007 © Institute of Development Studies

1

between ‘economic’ and ‘social’ goals can bedemonstrated, consensus is possible – but theseopportunities are few and far between. Takeminimum wages: in discussions aroundmainstreaming social protection into Uganda’sPoverty Eradication Action Plan in 2003, the editorsof this IDS Bulletin argued for a binding statutoryminimum wage to be applied to exploitedagricultural workers in Uganda’s plantation sector.Although the argument was made on social justicegrounds, the Ministry of Finance was not convinceduntil empirical evidence from several countries wasproduced, confirming that raising wages forunderpaid workers does not necessarily result in‘disemployment’ effects and can even raiseproductivity and profits.

Anyone who believes that design choices in socialprotection programmes – whether to applyconditions, to transfer cash or food, who to target –are purely pragmatic technical issues, with decisionsbased on solid empirical evidence, is missing thepoint. These are ideological choices, reflecting the‘vision thing’ that Kabeer refers to in this IDS Bulletin.Which choices are made, and for what reasons,reflects the kind of society that policymakers andtechnocrats with power to direct social policy wishto promote. Compare the Government of India’srecent decision to enact an employment guaranteefor all rural households, with the attitude of theUnited States towards welfare provision for itscitizens. Social protection is self-evidently about avision of society.

Some of these visions are debated in the papers thatfollow this Introduction. This issue of the IDS Bulletinrevives a tradition of debate that was a feature of itsearly years, but disappeared some decades ago. Thefirst part of this IDS Bulletin is titled ‘Thinking’. Fiveconceptual frameworks or approaches to socialprotection are introduced by their protagonists (theWorld Bank, IDS, OECD, etc). These frameworks arecritiqued – commentators were encouraged to beprovocative and pithy – and the protagonists aregiven a ‘right to reply’ to their critics. Some stirringencounters are the result. In the second part, titled‘Practice’, this approach is pursued in some keydesign choices, including the cash vs. food debate,the conditionality debate, and targeting. The issuecloses with two reflective pieces on applications ofsocial protection to vulnerabilities in smallholderagriculture and health (HIV and AIDS).

2 Thinking: framing the debatesFive conceptual frameworks are presented anddissected in the ‘Thinking’ section of this IDS Bulletin:‘Social Risk Management’ (World Bank),‘Transformative Social Protection’ (IDS Sussex), ‘AssetThresholds’ (Michael Carter and Christopher Barrett),the POVNET approach (DAC/OECD), and the‘Universal Social Minimum’ (Koy Thomson/ActionAid).Two of these five frameworks are the product ofdonor agencies, two come from universities and oneoriginates from an NGO. This diversity of sources isencouraging for those who lament that socialprotection is an entirely donor-driven agenda.

We start with the Social Risk Management (SRM)framework, which has become the most influentialapproach to social protection in recent years, notleast because it has been heavily promoted by theWorld Bank through its financing, research, trainingand policy advice. A defining insight of SRM is thatpoverty and vulnerability are compounded byuninsured risk, so effective risk management will notonly stabilise income and consumption but is aninvestment in poverty reduction. Advocates of SRMmake a virtue of its focus on risk and efficiencyrather than equity and needs. According toHolzmann and Kozel (this IDS Bulletin), the WorldBank’s Social Protection Strategy ‘is innovative inplacing particular emphasis on risk and riskmanagement as a complement to social protection’smore traditional emphasis on equity and basic needs’.

To evaluate the SRM framework, we threw theweight of the IDS Director and three IDS studentsinto the fray. Lawrence Haddad opens with a gentlerebuke about SRM’s ‘failure to engage’ with politics,anthropology, geography, sociology and psychology:‘after all, the “S” in SRM stands for social’. AsGuenther, Huda and Macauslan then point out, theSRM focus on risk management ‘leads tointerventions that focus on transitory income shocksrather than on structural determinants of poverty’.Both sets of comments point to SRM’s narrowconceptualisation of risk, especially its neglect of‘social risk’ and ‘irreversibility’, which Haddadattributes to the World Bank’s ‘commitment tomono-disciplinarity’. In their ‘right to reply’,Holzmann and Kozel acknowledge the ‘bias towardeconomic interpretation’ but argue that SRM is opento drawing on other disciplines, and reassert thatSRM has shifted the social protection discourse from‘advocacy’ toward ‘evidence-based policy’.

Devereux and Sabates-Wheeler Editorial Introduction: Debating Social Protection2

The ‘Transformative Social Protection’ (TSP)framework was devised partly in response to aconcern that the increasingly hegemonic Social RiskManagement framework was really only about‘economic risk management’. TSP endorses theWorld Bank belief that risk matters and that socialprotection can be supportive of economic growth –through links to ‘prevention’ (insurance) and‘promotion’ (productivity-enhancing) interventions –but complements this by adding a ‘social’ strand toparallel the ‘economic’. Analytically, TSP argues forreconceptualising ‘vulnerability’ as ‘emerging fromand embedded in the socio-political context’(Sabates-Wheeler and Devereux, this IDS Bulletin),and requiring interventions that effectively tackle thestructural causes of vulnerability. ‘Transformativesocial protection’ extends beyond safety nets andwelfare handouts, towards supporting citizens toclaim social protection from the state as a basic right.

A second group of IDS students was invited to critiqueTransformative Social Protection. Ken Aoo and hiscolleagues identify a contradiction in TSP’s claims toempower the poor and its prescriptive agenda fortransformation, which could be interpreted aspatronising at the micro-level and violating nationalsovereignty at the macro-level. A related critique isthat the concept of TSP is too broad and holistic to beoperationally useful. Replying to their students,Sabates-Wheeler and Devereux counter that anydevelopment policy implies ‘meddling’ in the politicsand culture of other countries, and that TSP is not all-encompassing but favours interventions that maximisesynergies between (social) ‘protection’ and (economic)‘promotion’.

Carter and Barrett present an intriguing applicationof recent economic thinking and analysis to socialprotection. The basic idea of ‘asset thresholds’ is thata critical level of assets exists above which people can‘invest productively, accumulate and advance’, butbelow which people are in a ‘poverty trap’ fromwhich they have no prospects of escape. Empiricalevidence plus modelling reveals the profoundimplications for social protection policy of thisbifurcation – for instance, humanitarian assistance thatmaintains subsistence consumption does not preventasset depletion and only pushes increasing numbersof people into chronic poverty and a ‘relief trap’.

Stefan Dercon critiques the ‘unidimensional’ focus ofthe asset thresholds framework on assets, and its

neglect of institutional failures (e.g. caste structure) andmarket failures (e.g. inflexible land rights) that can alsoinduce ‘threshold effects’. One implication of allthreshold models, noted by Dercon, is that limitedsocial protection resources should rather be expendedon protecting people from sliding below the threshold– a genuine ‘safety net’ function – it is cheaper to keeppeople from falling down a ravine than helping themto climb out. The corollary is that chronic poverty isunlikely to be reduced at all by small, targeted cashtransfers; notwithstanding the current popularity ofthese social protection programmes. ‘Small is notbeautiful – it is useless for the poor.’ Replying toDercon, Carter and Barrett emphasise several pointsof agreement, and suggest a bifurcated policyresponse: ‘large-scale transfers to the very poor andthreshold-based social protection for the vulnerable’.

The next conceptual framework in this section ispresented by the chair of the POVNET Task Team onSocial Protection and Social Policy. Timo Voipioasserts that the OECD/DAC approach to povertyreduction has always been inflected by Nordic valuesof social development, which produced a multi-dimensional understanding of poverty and a focus onpro-poor growth that combines ‘economicopportunities, social protection andinclusion/empowerment’. In this context, Voipio seessocial protection performing two roles: ‘as a keyelement of pro-poor growth but also as a rights-based responsibility to care’.

Naila Kabeer is not convinced that POVNET thinkingon social protection represents a convergence ofviews, rather than an uneasy compromise between‘members of the club of the world’s richestcountries’, who bring ‘somewhat differentphilosophies and agendas’ to the table. Kabeer prisesopen a key area of disagreement, between ‘theWorld Bank’s residualist approach’ and the Nordiccommitment to universal basic social security, andargues that this reflects two radically different visionsof society: the former individualistic and market-based; the latter group-based and citizen-centred.Ultimately, Voipio appears to agree with Kabeer thatthe achievement of effective and equitable socialprotection in any society is a struggle for democraticrights, one that is more likely to be realised throughcollective action than donor projects.

This is precisely the keystone on which Koy Thomsonconstructs his justification for a ‘Universal Social

IDS Bulletin Volume 38 Number 3 May 2007 3

Minimum’, by which he means the ‘resources,opportunities, rights and power to lead anadequately decent and dignified life, and toparticipate and advance as a free and equal memberin society’. This is an explicitly ‘political programme’grounded in human rights and social justice theories,rather than an economic policy debate to be left to‘technocrats’. This approach to social protectionrequires a broader conceptualisation of poverty andof development, more reminiscent of Sen’s‘capabilities’ approach and his ‘development asfreedom’ argument.

As ‘friendly critics’ of the universal social minimum,Sabates-Wheeler and Devereux fully endorse itsaspirations but question its practicality, specifically its‘ideological egalitarianism’, which risks leaving all poorpeople poor but equal rather than ‘giving somepeople a hand-up out of poverty’. They also doubtwhether the fiscal resources and politicalconstituencies exist in poor countries to provide aminimum package of resource transfers and socialrights within national budgets and political processes,or whether this agenda will remain externally inducedand financed for years to come. Thomson concedesthat some of the most important constituencies forsocial protection – ‘Parliamentarians, civil societyorganisations, the media, the public’ – have beenalienated from current debates around socialprotection, but believes that the universal socialminimum could provide a platform for mobilisingthese constituencies.

3 Practice: what to do?Many issues in the design of social protectionprogrammes that might appear to be mere practicaldetails have provoked fierce theoretical and empiricaldebates, and ongoing controversy. Three of thesedesign issues are debated here: conditionality, cash vs.food, and targeting. These ‘downstream’ policychoices bear close scrutiny because they track thesame intellectual fault-line that runs through‘upstream’ thinking and conceptualisation of whatsocial protection is all about. Crudely, most‘instrumentalists’ would line up behind conditionaltransfers, in either food or cash depending onprogramme objectives, and fine targeting. Conversely,most ‘activists’ would argue for unconditionaltransfers, preferably in cash rather than food, anduniversal entitlements. For the instrumentalists, a keyconcern is how to maximise efficiency in resourcetransfers: conditionality achieves a double impact, and

pro-poor targeting minimises leakages to the non-poor. For the activists, two key considerations arehow to maximise freedom of choice (henceunconditional transfers, in cash) and how to entrenchsocial protection as a right (hence universal provision).1

We start this section with three very differentperspectives on conditionality. Armando Barrientosgives three cheers (or at least two-and-a-half) forconditional transfer programmes, citing empiricalevidence that requiring poor parents to send theirchildren to schools and clinics in return for resourcetransfers does indeed achieve the desired ‘doubleimpact’ on household consumption as well as pro-poor investment in human capital. Most ofBarrientos’ witnesses for the defence come fromLatin America, where the public services on whichtransfers are conditioned are more extensive andmore effective than in much of sub-Saharan Africa.This leaves unanswered the question of whetherconditional cash transfers (CCTs) are transferable topoorer and institutionally weaker countries in Africa,where CCTs have yet to be widely adopted.

Maxine Molyneux offers an ironic ‘Two Cheers forCCTs’ (rather than three), welcoming the evidencefor ‘positive outcomes in children’s welfare’ buturging caution over some ‘overblown claims’ madeon behalf of conditional cash transfers. Molyneux isespecially concerned about adverse gender impacts,agreeing that women are central to theseprogrammes but pointing out that their role islargely confined to ‘servicing the needs of others’,especially their children. Generally, it is mothers –not fathers – who are monitored for compliancewith the conditionalities imposed by CCTs, therebyreinforcing women’s role as carers of children, whilemen remain marginalised. Molyneux asks pointedly:in what sense does this empower women?

No cheers at all from Nicholas Freeland, whosedepiction of conditional cash transfers as ‘economicallysuperfluous, physically pernicious, morally atrocious andpolitically abominable’ is hardly a ringing endorsement.Freeland argues that the word ‘conditional’ isimprecise, since ‘all social transfers are conditional onsomething’; that the word is reminiscent of IMF andWorld Bank loan conditionalities and ‘smacks ofBretton Woods paternalism’; that conditions areunnecessary because the poor tend to spend some oftheir social transfers on services anyway (and theyshould be free to make these choices); and that

Devereux and Sabates-Wheeler Editorial Introduction: Debating Social Protection4

imposing any conditions on social transfers for the poorviolates the principle (which most countries have signedup to) that social protection is a ‘basic human right’.

Turning next to the thorny issue of cash transfers vs.food aid, we invited Paul Harvey (who is based at theOverseas Development Institute (ODI), which hasrecently taken up the case for cash transfers inemergencies) to debate with Ugo Gentilini (who isbased at the World Food Programme (WFP), whereposters in the corridors proclaim: ‘We Feed People’).Harvey lists several advantages of cash over foodtransfers (e.g. cost-effectiveness, choice and dignity),and asserts that many alleged negative features ofcash (higher security and corruption risks, lowernutritional impacts, etc.) have been exaggerated.Gentilini argues that the choice of cash or foodshould not be elevated to a principle – ‘cash and foodare just instruments and not strategies’ – but shouldbe determined by the context and objectives of eachsocial protection intervention. A particular context-specific concern is how local markets function andhow traders, producers and prices will respond tocash and/or food transfers. More generally, Gentilinicites recent empirical evidence suggesting that the‘anti-food movement’ has overstated thedependency and disincentive effects of food aid.

Harvey agrees with Gentilini’s conclusion that ‘cashand food can be complementary’ in theory, butpoints out that in practice food aid continues todrive the global humanitarian system. Untilinternational aid is untied from donor food surplusesand national governments play the lead role inhumanitarian response, agency decisions will not bebased on evidence and food will continue to beprovided even when cash would be moreappropriate. Gentilini’s riposte focuses on Harvey’sbelief that cash transfers will inevitably increase andultimately displace food aid in most emergencies.Gentilini argues that the evidence base on theimpacts of cash in emergencies is mixed, that marketfailures pose a real threat to cash transferprogrammes, and that beneficiaries often express apreference for food rather than cash, which it is‘inappropriate’ to ignore.

John Hoddinott ends our policy debates by debatingtargeting with himself – ‘Social protection: to targetor not to target?’ – but reveals his bias immediatelyby offering ‘two vignettes’ that illuminate howtargeted social transfers ‘do indeed deliver a greater

share of programme benefits to poor households’.Canada’s Old Age Security pension is amusing aswell as enlightening: attempting to reconcile equity(universalism) with efficiency (pro-poor targeting),the Canadian government delivers this pension to allits citizens, but taxes it at 100 per cent for the non-poor! Hoddinott concedes that the various costs oftargeting – to administrators, beneficiaries and theircommunities – can be significant, but he maintainsthat the savings in reduced leakages to the non-poorusually outweigh these costs. Finally, although this isessentially an economic analysis that refutesarguments for universal provision grounded in socialjustice, Hoddinott is emphatically opposed toexploiting social stigma for efficient self-targeting:‘this is wrong wrong wrong’.

The closing pair of papers in this compilation applycurrent thinking on social protection to two keysectors for poor people – agriculture and health –but speak to a bigger and urgent need for socialprotection interventions to be informed by improvedanalysis. Rebecca Holmes, John Farrington andRachel Slater argue against the tendency to draw anartificial distinction between ‘consumption-smoothing’ social protection interventions designedto address vulnerability in the ‘domestic’ domain, and‘productivity-enhancing’ interventions designed toachieve pro-poor growth in the ‘productive’ domain.Especially in sectors like smallholder agriculture,where the homestead is also the workplace and theworkforce is dominated by family labour, thesynergies between the two domains are so strongthat an integrated rather than dichotomisedapproach to social protection and other pro-poorpolicies is clearly called for.

The final paper rounds off this IDS Bulletin bybringing together fresh ideas for improved analysisand response. Jerker Edström builds a case forreconceptualising ‘vulnerability’, the defining conceptthat motivates all social protection, but one that isoften inadequately understood and weakly carriedthrough into policy design and implementation. Bydisaggregating vulnerability into ‘embodied andpersonal biological and psychological dimensions ofsusceptibility/resistance and sensitivity/responsiveness,with contextual inter-personal and environmentaldimensions’, Edström implicitly lays out a range ofentry points for social protection interventions. Howmany policymakers consider whether their socialprotection programmes are aimed at strengthening

IDS Bulletin Volume 38 Number 3 May 2007 5

resistance, reducing sensitivity, or addressing theenvironmental determinants of vulnerability?Edström also questions the usefulness of thehousehold as a unit of analysis and response, whenso much social protection is a response to thebreakdown of families (e.g. street children) andaccess to (or exclusion from) support from extendedfamilies and communities is a critical determinant ofindividual vulnerability or resilience. AlthoughEdström is preoccupied with vulnerabilities related toHIV and AIDS, these arguments apply equally to allsocial protection thinking and practice.

4 ConclusionThis selective overview of conceptual frameworksand practical dilemmas in social protection hashighlighted how rapidly thinking and practice havemoved forward in a few short years, but it has alsorevealed that a range of conceptual, empirical andpolicy issues remain unresolved. Some of these issueswill resolve themselves, as evidence and experienceaccumulate, lessons are learned and best practice isinstitutionalised within agencies responsible fordelivering social protection. Other issues pose biggerquestions, however. At every level – conceptualframeworks, evidence-building, and policy processes– there are basic disagreements and debatesbetween alternative approaches that are notamenable to consensus or compromise.

4.1 Conceptual challengesAll of the conceptual frameworks discussed in thisIDS Bulletin are preoccupied with social protection asa response to the economic and social vulnerabilitiesthat poor people face, yet none of these frameworksseems up to the task of comprehensively describingthe risk environment and proposing a focused andappropriate set of policy options to mitigate orreduce vulnerabilities. For instance, a fundamentaldistinction has emerged between conceptualapproaches that incorporate structural vulnerability intheir understanding of the risk environment, andmake social inclusion an explicit objective of socialprotection programming, and those that do not. Canthese two approaches complement each other, orare they inevitably in opposition? More broadly, isthere any scope for bringing together thecontrasting visions of the ‘instrumentalists’ and the‘activists’, or is this an ideological fault-line thatdivides the two camps as sharply as that dividing theinterests of capital and labour?

4.2 Evidence gapsThe evidence base on what works in socialprotection and what doesn’t is patchy, but studiesare accumulating and the gaps are rapidly beingfilled. The ‘cash vs. food’ debate, for instance, revealsthe importance of basing policy choices on a soundanalysis of local markets and the articulatedpreferences of (female and male) beneficiaries, ratherthan on resource availability or an untested beliefthat one form of transfer is always best. A similarcase for context-specific programming is presentedby the debate on conditionalities in social protection,which must take account of the quality andavailability of local services on which transfers areconditioned, and should also assess much morecarefully the implications for women.

A very different challenge is presented by theevidence on ‘asset thresholds’. If these do exist inreality, such that poor people need large ‘lumpy’transfers to escape from poverty and ‘graduate’ fromdependence on social assistance, then tiny transfersare indeed ‘useless for the poor’, and the value incampaigning for a ‘universal social minimum’ isseriously compromised. The underlying issue here isfamiliar from social welfare debates: does a rights-based approach to redistributive transfers inevitablyput the brakes on economic growth, and if so, towhat extent is each society willing to trade offeconomic growth against social equity?

4.3 Policy processesA final set of concerns asks the question: where arethe drivers of social protection coming from? Toomuch of the current social protection agenda isdesigned and financed by external actors – bilateraland multilateral donors, international NGOs,academics and consultants – and not enough isdriven by domestic constituencies – nationalgovernments, local civil society, citizens. This hasunsatisfactory and potentially ominous implicationsfor the ownership of these processes, accountabilityfor delivery and impacts, and political and financialsustainability of social protection programmes. Thereare simply ‘too many pilots and not enough drivers’of social protection, especially in Africa, which hasyet again become a laboratory for experimentation,this time with dozens of ‘boutique’ cash transfer pilotprojects that are weakly grounded in domestic policyprocesses and are rarely institutionalised at scale bynational governments.

Devereux and Sabates-Wheeler Editorial Introduction: Debating Social Protection6

The ultimate goal, surely, should be nationally ownedsocial protection policies, underpinned by a ‘socialcontract’ between the state and its citizens, wheregovernments acknowledge that social protection is aright for which they are the duty-bearers, andcitizens mobilise to demand that this right is

effectively delivered to them. The risk of not movingrapidly in this direction is that social protection willfall as quickly as it has risen, and become in a fewyears’ time just another shooting star in the historyof development policy.

IDS Bulletin Volume 38 Number 3 May 2007 7

Note1 Of course, this distinction is deliberately drawn

excessively sharply here. Many social protection‘activists’ are (or should be) also preoccupied withefficiency, and most ‘instrumentalists’ would claim

to be deeply concerned with equity – indeed,they would argue that the efficient, pro-poorallocation of scarce public resources is aprerequisite for social justice.