One of the most stubborn obstacles to the promotion and adoption of comprehensive social transfers in southern Africa is the difficulty of designing and implementing affordable and effective targeting mechanisms. The main reason given for spending on targeting is that it improves the efficiency (saves money) and effectiveness (increases impact, primarily in terms of reducing poverty and inequality) of social transfers. Many would argue that, without targeting, the cost of a comprehensive social transfer programme would be prohibitive.

While it is relatively easy to target on the basis of specific characteristics such as age – sometime known as "categorical targeting" – targeting becomes more problematic when one wants to select the poorest or the most vulnerable. Targeting on the basis of poverty or vulnerability can either be done by applying a set of "means testing" criteria, often using community-based mechanisms, or through "self-selecting" mechanisms such as public works programmes which sets wage rates at levels intended to attract only those in need.

Few people would dispute the value of targeting if it did indeed make social transfer programmes more efficient and effective. But the evidence does not always support this.

At the most fundamental level, there is ample evidence to suggest that targeting simply does not work in an African context, and that where it is applied or imposed it is simply circumvented. Many NGOs will attest to the fact that, even when the poorest members of a community are earmarked for food aid, the intended beneficiaries routinely share the distributed food with the wider community. This practice appears to be so common that it begs the question as to why intra-community targeting persists.

As Ryszard Kapuscinski so eloquently expressed it in his book "The Shadow of the Sun":

"African tradition is collective, for only in a harmonious group could one face the obstacles thrown up by nature. And one of the conditions of collective survival is the sharing of the smallest thing."¹

Opposition to targeting within communities is only partly cultural. In southern Africa, poverty is endemic, and attempts to separate the destitute from the ultra poor from the poor can create largely artificial distinctions between people. A familiar refrain when communities are asked to identify their poorest members is "but we are all poor".

In practice, targeting is often invasive, divisive and unpopular. For example in Malawi, there were cases (in the context of the Targeted Inputs Programme) where communities rebelled against being required to select one household over another, and rejected outright the entire benefit, despite the desperate need for resources, because of the divisions it brought.

Even where (or rather, if) targeting is supported and popular, there are still significant challenges in making it work.

The preoccupation with the need to target has been the driving force behind the considerable investment, primarily by donors, in establishing more and more sophisticated vulnerability assessment and analysis systems. However, while increasingly elaborate methodologies might tell you, down to the last detail, the distinguishing characteristics of the most vulnerable, no system will identify or pinpoint the specific households or individuals. Ultimately, the onus and responsibility for this is on the members of the community themselves.

In efficiency terms, there is considerable evidence to raise serious doubts about the worthiness of targeting social transfers: A UN review (Targeting and Universalism in Poverty Reduction²) concluded that even well-regarded targeted social transfer programmes record high levels of under-coverage (exclusion error): the much lauded Bolsa Familia family income grant programme in Brazil, for example, excludes 73% of its intended beneficiaries. Similarly, leakage rates (inclusion error, where resources go to those they are not intended for) from evaluated targeted social transfer programmes are also high. The UN review quotes leakage figures of up to 61 percent of total programme costs. Targeting also heightens the possibility of clientelism and patronage because not everyone receives the benefit equally, thus leaving room for bias in selection and for corruption. This applies irrespective of whether targeting occurs at national or community level.

Similarly, the argument that targeting is more pro-poor than universal provision can be questioned.

In terms of income redistribution, the UN review noted that "(average) targeted programmes in sub-Saharan Africa transferred 8 per cent less to poor households than (untargeted) programmes". The report also points out that "societies that lean towards (untargeted) social policies have less inequality than those that prefer targeting". Other evidence for this is provided in The Universal Welfare State as a Social Dilemma³ and The Paradox of Redistribution and Strategies of Equality3) . Part of the reason why targeted programmes are not necessarily more pro-poor than untargeted ones is that targeting tends to lead to a reduction of the budget devoted to poverty and welfare. Efficiency gains are not used to increase the coverage of social protection but are converted into reduced spending by funding agents. The same number of people are covered with less money, rather than more people being covered with the same amount of money.

And then of course there is the question of efficiency: does targeting save money?

The evidence cited above indicates that there are high costs involved in both over-coverage and under-coverage. An Asian Development Bank study concluded that "in practice most targeting measures have been high cost measures of transferring benefits to the poor". The administrative costs of implementing social transfer programmes vary considerably. The operating cost of the on-going Mchinji Cash Transfer Scheme in Malawi is roughly 20 per cent of the overall cost. Although the breakdown between the specific costs associated with targeting and the normal administrative costs of a programme is not always transparent, it is clear that the cost implications of under- and over-coverage greatly exceed the direct administrative costs of a targeted programme. These larger costs are eliminated in a universal programme.

So, if targeting on the basis of poverty doesn’t work, what is the alternative?

To be continued . . .

Footnotes

1. Kapuscinski goes on to give the following example: "One day a group of children surrounded me. I had a single piece of candy, which I placed in my open palm. The children stood motionless, staring. Finally, the oldest girl took the candy, bit it into pieces, and equitably distributed the bits."
2. http://www.unrisd.org
3. http://www.russellsage.org
   http://www.socsci.auc.dk