See fully formatted version here.
We all want practical solutions to rising global challenges. Universal social protection, a benefit system everyone has access to – also known as universal social security – is one of them. The World Bank is publicly committed to universal social protection, arguing it protects people when they are vulnerable or experiencing crises – including climate-related events and pandemics – and helps build economies and a ‘just transition’ to low-carbon growth.
The United Nations calls social protection a human right, yet more than half of the world’s population does not have access to it. Governments in the Global South are working to fill this gap, and support from the Global North is being led by the World Bank, which provides technical advice and resources. Currently, the Bank has a portfolio of almost $29.5 billion in social protection financing, covering an estimated 880 million people. But there is a problem. Notwithstanding its public commitment, for the part of social protection that is government-financed, as opposed to contributory, the World Bank promotes poverty-targeted social protection programmes, which have high targeting errors, may cause social unease and can set back the development of universal social protection systems for years.
Evidence suggests there is little relationship between income level and spending on social protection in low- and middle-income countries – it is a political choice.
The multiple costs of poverty-targeted social protection
There is a logic to poverty targeting, to give scarce government resources to those that need them most. But this is in practice difficult to do, especially when the documentation of incomes is limited, as it often is in developing countries. The World Bank has promoted a way around this – proxy means testing – which uses observable household information, such as the material a dwelling is made from, to estimate whether it is in poverty. It started in Augusto Pinochet’s Chile around 1980 and is now used in at least 60 countries.
But proxy means testing comes with costs. One is high targeting errors which have been a running debate between the World Bank and its critics. Stephen Kidd, CEO of Development Pathways, co-authored a paper in 2011 with Emily Wylde setting out estimates that around half of the intended target group for social protection schemes using proxy means testing in Bangladesh, Indonesia, Rwanda and Sri Lanka were excluded by programme design, sometimes less, sometimes a lot more. So, if you were within the target group of ‘the poor’, you had an even chance of not being selected for support. You might as well flip a coin. UK government guidance published in 2013 explained the high errors, by saying proxy means tests “are inevitably somewhat blunt instruments, particularly in poor countries where differences in income, assets and other household characteristics are very small, making it difficult to identify which households are above or below an eligibility cut-off point.” The guidance also pointed to some consequences of the high errors: “perceptions of [proxy mean tests] among non-beneficiaries in Mexico, Nicaragua, Indonesia and Lebanon led to tensions, unrest and even conflict.”
The World Bank published a paper in 2014 and a book in 2015 asserting proxy means testing “can accurately and cost-effectively target the chronic poor.” Two more papers from Stephen Kidd and colleagues, in 2017 and 2020, reported high targeting errors, in the second paper arising from both programme design and implementation.
Meanwhile, a new voice engaged in the debate. Martin Ravallion, then head of research at the World Bank, met Stephen Kidd in 2012 and discussed proxy means testing. After leaving the World Bank, in a paper in the 2016 Journal of Economic Perspectives he said it had turned poverty-targeting into a ‘fetish’. At the same time Ravallion wrote a paper with Caitlin Brown and Dominique Van de Walle, later published in the 2018 Journal of Development Economics, which estimated targeting errors for nine African countries. It concluded that universal programmes – where benefits go to groups in the population such as children, the elderly or persons with disabilities, regardless of income – may well be better than those using proxy means testing, even for poverty-targeted programmes’ primary objective, reducing poverty. Why? Because of high targeting errors; the relatively high costs of administering proxy means tests; the lack of transparency in terms of who is and is not eligible; and the potential to undermine political support for programmes. But the World Bank’s encouragement of proxy means testing continued, and even scaled up, through the promotion of social registries, which allow poverty targeting across a range of programmes, not just one. Social registries are now in more than 50 countries, most using proxy means testing. In 2021, Stephen Kidd, Diloa Athias and Idil Mohamud responded with a critique entitled Social registries: A short history of abject failure.
In 2022, the World Bank published a 539-page tome, Revisiting targeting in social assistance: A new look at old dilemmas, apparently aimed at settling arguments about targeting approaches. These include proxy means testing and community-based targeting, where community members have a say and which is often used in combination with proxy means testing, but which is vulnerable to corruption. The book argued all targeting options should be considered, because every country is different. It referenced the papers that Kidd co-authored and conceded proxy means testing is “difficult for people to understand”, adding it has “inbuilt statistical error”, and that universal schemes are “possibly pragmatic” in some circumstances. But it reverted to type saying, “To reduce poverty, it is usually more cost-effective to ensure that a greater share of benefits accrue to the poor than to expand coverage broadly.” Proxy means testing is described as “imperfect but realistic”.
Borrowing governments show a different way
The World Bank’s promotion of poverty targeting and proxy means testing has been opposed by a global coalition of highly motivated people and institutions. This includes the International Labour Organization (ILO), others in the United Nations and a wide range of academics, activists and civil society organisations. A campaign led by the ILO, including Michael Cichon and Isabel Ortiz among others, encouraged the World Bank’s public commitment to universal social protection. The Global Coalition for Social Protection Floors and more recently the Campaign for the Right to Social Security are also in the fight. This article is based on a book I am writing about one of these highly motivated people: Stephen Kidd – an ex-missionary who cut his teeth fighting for indigenous land rights in Latin America – who learned much on universality from economist Thandika Mkandawire, and who has been at the forefront of the fight for two decades.
The fight is important, but what matters most, of course, is what governments do themselves. New research shows 52 low- and middle-income countries have introduced 88 tax-financed schemes, including child benefits, pensions and disability benefits which are universal, or near-universal where they are complementing contributory programmes. And it is not just wealthier countries. For example, Nepal introduced universal schemes when it was a low-income country. In fact, evidence suggests there is little relationship between income level and spending on social protection in low- and middle-income countries – it is a political choice. These schemes have mostly been implemented without the support of donors, and even in the face of World Bank opposition, for example in Kenya, Lesotho, Mauritius, Mongolia, Namibia, Nepal, Thailand and Zanzibar.
There are governments that have not chosen universal programmes of course – often encouraged by donors – and a combination of poverty-targeted and universal schemes is common, as it is in the Global North. But Kidd and his collaborators do not argue that all programmes should be universal. What they object to is the World Bank pre-judging the issue. Not least because poverty-targeted programmes often create a ‘missing middle’, stuck between them and contributory programmes at the high-income end of the scale, which are restricted by high and persistent labour market informality. Hence the need for universal schemes, which may draw broader political support and so be more sustainable.
Why does the World Bank do it?
The World Bank promotes poverty targeting and proxy means testing for several reasons. Firstly, it is a bank that arranges loans. Ravallion has said, “The pressure to lend influences the Bank’s ability to deliver objective policy advice to client countries.” Loans are given for established approaches like poverty-targeted social protection schemes, and sometimes conditions on using proxy means testing are attached to loan disbursements, such as in Jordan. These are agreed with governments, but governments need the money.
Secondly, the World Bank, with other donors, uses an overly technical approach to supporting governments, likely stemming from the highly academic background of its staff, and from their need to retain control. Marion Ouma from the Nordic Africa Institute in Sweden and Professor Jimi Adesina from the University of South Africa describe this need to keep control in Kenya, where the depoliticisation of the social protection policy agenda by donors “weakens democratic processes, compromises ownership and limits popular discussions on policies.” It also weakens the connection between spending and revenue raising, which is core to the government-citizen social contract.
Thirdly, a short-term ‘results agenda’ exists within the World Bank and other donors, which can conflict with long-term national development. This stems from a need to show impacts within 3-5 year donor programme timeframes; from the World Bank’s unenviable remit to ‘end extreme poverty’; and from the need to achieve the Sustainable Development Goals by 2030. The upshot can be hard-pressed government officials grappling with over-complicated poverty-targeted schemes instead of relatively simple universal programmes government systems can deliver at scale, and a perhaps years long delay in developing universal social protection.
Fourthly, there is a lack of lesson learning. Ravallion argues that what is needed is “a quite fundamental change in the [World] Bank’s culture…reoriented from lending to learning.” Too often the World Bank uses evidence selectively to make its case, instead of keeping an open mind. This is ironic given its ambition to be a Knowledge Bank. And it is not just proxy means testing: There are similar issues in the promotion of conditional cash transfers – benefits in exchange for behaviours such as children attending school – and workfare – poverty-targeted social protection given in exchange for manual work.
Why do major World Bank shareholders such as the US, Japan, the UK, Germany and France not insist on a different approach? Leaving aside ideology and geopolitical relationships, funding countries have a complicated and interdependent relationship with the World Bank: They need to show results from aid spending too, which are often delivered through World Bank programmes. They – we – are all implicated. It is also hard for other World Bank member countries to step into the breach given skewed voting power within the World Bank.
All is not lost
All is not lost, however. The World Bank has also done good things on social protection. It has advocated for growth in spending for many years, significantly increasing its own funding commitments. It has supported important elements of system building such as better public financial management and digital payment systems. More recently it has recognised both the importance of human rights and the stigma associated with programmes for ‘the poor’, and it has begun to support the delivery of some universal schemes, for example in Kosovo, Nepal and Sudan.
But pressure on the World Bank must continue now until a tipping point is reached in its approach to social protection. The Bretton Woods Project and others have called for an independent evaluation of the World Bank’s strategic direction, set out in its recent ‘roadmap’ for reform. The World Bank and other donors should also remind themselves of the Paris Declaration, which more than 100 countries signed up to in 2005. Its principles include ownership – meaning low- and middle-income countries set their own strategies – and alignment – meaning donors get behind those strategies. This can be challenging where systems of accountability and democracy are still under development. But it will, in the end, be the difference between supporting the brief blossoming of a donor-funded programme and real, lasting change.
There is an urgency to this. Not only does more than half of the world’s population not have access to social protection, but geopolitical, socioeconomic and climatic risks have been added to ongoing poverty and food insecurity, all of which lands hardest on women and children. Lessons from the response to Covid-19, “the largest scale-up in history”, are that relatively simple social protection programmes that are large and quickly scalable, such as the universal pension in Bolivia and near-universal child benefit in Mongolia, respond best to big shocks. Furthermore, evidence from the ILO states “today’s developing countries have a similar level of GDP per capita to high-income countries when they established their social security systems”. So, as well as urgency there is opportunity.
Matthew Greenslade is an economist who has worked on the value for money and financing of social protection systems for over twenty years. He has worked with the UK government, in the Department of Work and Pensions and DFID, including in the Nepal and Uganda country offices. Most recently he has been supporting governments in East Africa as an associate consultant with Development Pathways. Matthew’s book, Beyond the World Bank: The Fight for Universal Social Protection in the Global South, on which this article is based, will be published by Bloomsbury later this year.
