As African countries look into various forms of social protection, one of the key debates revolves around the issue of targeting – who should benefit from cash transfers or other forms of support? A debate has arisen between two positions – those who believe in poverty targeting and those who favour categorical targeting.
Poverty targeting means that programmes should attempt to identify the poorest and most deprived members of society, and provide benefits only to them.
Categorical targeting, on the other hand, means that benefits go to people who fall into a specific category that is closely associated with poverty – such as old age pensioners, or children under a certain age.
The question of which form of targeting is better, is one of the biggest dilemmas facing policy makers who are thinking about implementing large-scale, countrywide social protection programmes.
A new brief prepared for the Regional Hunger and Vulnerability Programme (RHVP) by Frank Ellis and Francesca Marchetta, weighs in on this debate with some new evidence.
The brief looks at a common approach to poverty targeting in southern Africa. In this approach, benefits are given only to the ‘ultra poor’ – households that can’t even get enough food to eat, and who also lack active adult labour in their households – in other words, nobody in the home who is able to go out and work. In Malawi and Zambia, the proportion of households that meet these criteria is commonly understood to be about 10 per cent (based on national living standards surveys).
In pilot social cash transfers in these two countries, this national proportion of 10 percent has also been used at community, sub-district and district level – so within each community or district, only about 10 percent of people are selected to benefit from cash transfers.
There are two objections to this:
Firstly, it excludes many ultra-poor households who are just as badly off but happen to have labour in their household (even though they may still lack land, skills, tools and job opportunities).
Secondly, while the proportion of the ultra poor may be roughly 10 percent overall nationally, this figure is likely to vary widely across geographical areas within a country – so in some areas you may find that many more than 10% of the local population is ultra poor, while in other areas there will be less than 10%.
Ellis and Marchetta investigate the validity of these objections by looking at evidence from two countries: Ghana and Malawi. They chose these two countries among other reasons, because they offer some interesting contrasts: socio-economic conditions in Ghana differ widely across geographical regions, while Malawi is relatively more homogenous.
The authors worked with national survey data in the two countries, and used these figures to test the impacts of different approaches to poverty targeting.
The findings are interesting. Firstly, they found that without a doubt, it’s a very bad idea to apply a uniform percentage across the entire country, as a cut-off point for selecting beneficiaries. This is because some regions within a country simply have more poor people than others – in Ghana for example 70% of those who should be beneficiaries are concentrated in just three of the ten regions in the country, so targeting 10% in each region would be highly inequitable. Even in countries such as Malawi which is relatively homogenous,if a uniform proportion (10% for example) is applied, there is very high risk of leaving out people in deprived places who should benefit from transfers and including people in less deprived places who should not be eligible.
Secondly, the experiments found that ‘leapfrogging’ is a definite problem in places where income distribution is relatively equal to start with – such as in Malawi. Because poor people’s incomes do not differ all that much, when benefits are given to the 10% of the so-called ‘ultra poor’, many of these people suddenly become better off than large numbers of non-beneficiaries.
So what does this mean for the targeting debate?
The authors point out that both categorical and poverty targeting have strengths and weaknesses.
One advantage of a categorical approach is that the choice of category usually involves a single rule (like an age range or a minimum age) and payments are made to all those who obey the rule. Because everyone who falls into that category will benefit, such a scheme is seen as fair, and so is likely to be more socially and politically acceptable than a scheme based on poverty targeting. In the case of old age pensions, for example, all citizens know that when they reach the required age, they will receive the pension.
But categorical transfers do have weaknesses. Some people who are not necessarily poor will benefit, and poor people who do not fit the criteria may not benefit – such as a poor household without a pensioner in it. Also, because everyone within a specific category will be covered, categorical social protection schemes can be expensive – and so difficult for low-income countries to afford.
Poverty targeted transfers also have strengths and weaknesses. Their strengths are that they can reach diverse types of people desperately in need of state support, and – at least in theory – it is possible to minimise errors of including or excluding the wrong people. Also, by limiting benefits to a specific percentage of the population, governments can keep such transfers affordable.
But while this looks good in theory, in practice it is very difficult and expensive to minimise inclusion and exclusion errors.
What the study by Ellis and Marchetta shows, is that if such poverty targeting does not take account of regional differences in poverty within a country, it may also leave out many of the poorest and benefit many who are relatively better off.
Finally, in many countries, it is difficult to ensure that the transfer amount is big enough to make a difference, but small enough so that it doesn’t leapfrog beneficiaries over non-beneficiaries. Such leapfrogging is inherently unfair, and can undermine social acceptance of the grant and create political difficulties.