'poverty targeting' and the 'proxy means test' - why anthropologists know more than economists about households and should share what they know.


A few days ago I listened to a great talk by Stephen Kidd, an anthropologist by training, who now works as a consultant for AUSAID and DFID the big Australian and British government agencies tasked with deciding on foreign aid. Stephen took us non-economics minded anthropologists and others through the ideology of international aid policy and the evidence that is used to back it up. Part of what he has been doing recently is to take a close look at the methodology and actual mathematics behind what is taken as hard fact by the big players including, of course, the World Bank.

In the paper attached here, Stephen and colleagues take down the Proxy Means Test as an central aspect of development policy. Proxy Means Testing goes hand in hand with a view that development aid should 'target the poorest' in society. It is taken for granted that this poverty targeting is intrinsically better than more generalised schemes of societal improvement in part because if 'the poor' can be accurately discovered the cost of development aid will be reduced and the right people will have their lives affected. A key method of finding 'the poor' in the societies in question is the Proxy Means Test. Proxy Means Testing involves very large surveys where a combination of 'proxies', for example, does the family own a television?, are used to decide whether the families are 'poor'. If they are poor they receive a grant with conditions attached.

The point of comparison for 'targeted poverty reduction' approaches are systems which are universally available under a certain condition - e.g. if you are old you get a pension. In many societies certain kinds of universal benefits coexist with poverty targeted benefits. However, there is a strong ideological bias in institutions like the World Bank that money should be targeted at 'the poor' and should have strings attached. Notice that the most socially developed countries in the world with the lowest levels of poverty - Sweden, Denmark, Germany, France etc. - use universal systems of benefit, while countries with some of the highest levels of social inequality/income disparity in the world - e.g. the U.S., England - focus their policy-making on means testing and targeting (isolating) 'the poor'.

In the paper attached here Stephen points to features of Proxy Means Tests that should be obvious to any practising anthropologist. The means test surveys are massive in scale but are based on extremely brief exposure to households and household assets (assuming that people are foolish enough to put their assets on display for the surveyors. Despite the claim that these surveys are extremely accurate there is a great deal of evidence that there are very large exclusion and inclusion errors involved in the surveys - as Stephen Kidd memorably put in his talk in some cases  you 'might as well throw the money out of a helicopter'.

A whole slew of issues are brought up. Amongst these is that, if a methodology is divisive (defining some as 'poor' and others as undeserving or 'not poor'), and if it is highly inaccurate 'you might as well throw the money out of a helicopter', then its relative cheapness is illusory. For anthropologists there is an important bridge here to include qualitative knowledge as part of a critique of development ideologies.

Here is a link to Stephen Kidd's website: http://www.developmentpathways.co.uk/category/pubs/

Views: 483

Reply to This

Replies to This Discussion

Huon, thanks so much. This paper is very interesting, indeed. Having, however, briefly scanned the short papers posted on the Development Pathways website, I am led to a conclusion that suggests that your title is misleading. As I read them, these papers are a strong and sustained example of advocacy for universal benefits conceived as entitlements, as opposed to programs that target only those deemed to be in special need. The issue is not whether aid programs could be better targeted by incorporating more sophisticated understandings of need (a project to which your title suggests anthropologists might have a lot to contribute). It is, on the contrary, whether targeting per se is an inherently flawed approach that empirically works less well than a universal-benefit approach. See, for example, the discussion of child poverty rates in Scandinavia as opposed, for example, the USA. 

You are right about the general trend of the argument, but my point is that proxy means testing, the method on which the poverty targeting view turns, could and should be a focus of critique by anthropologists who have far more knowledge of actual households and about local notions of poverty. Sorry for not making myself clear.

Hi, Huon

I've been holding my tongue (fingers?) still, hoping that someone else would join this thread. But the topic is fascinating, so let me ask a question.  Let's take it as given that anthropologists will know more about actual households and local notions of poverty. Given, however, that what we know is typically based on only a small number of cases, how do we work with people who need larger numbers to justify policy decisions? My guess is that they will be well aware of the dangers of using proxy measures (a standard topic in basic intro statistical methods courses) but use them anyway since anecdotal data alone will not be seen as justifying  a policy that affects tens or hundreds of thousands of households, and we cannot claim to know them all as intimately as we know a few. 

Very interesting paper, and you are right, Huon, the methods look familiar to an anthropologist, but nice to see them picked up more formally in such development contexts. Such proxies, for one, can be more sensitive to wealth (and not just income).  All the same, the sticky point remains where to draw the cut off lines, it seems to me.  All of our solid poverty measurement methodologies ultimately fall back on this subjective determination, or . . . .

The trickiest thing about use of material possessions as proxies for wealth is the way in which their value changes over time. Consider the TV set. In Japan, the first commercial TV broadcast was in 1953. In the 1950s ownership of a TV signaled substantial affluence. Now Over 90% of Japanese households own at least one TV set. It has been decades since Yale anthropologist Bill Kelly wrote a brilliant article on the introduction of TV in rural Japan, where traditionally the head of the household sat with his back to the _tokonoma_ the alcove in which a scroll was hung or a vase placed. That alcove turned out to be a good place to put a TV, but to see it the head of household had to move to the side of the table, losing his place of honor in the family hierarchy. By the mid-1990s, a study I quoted in my book on Japanese consumer behavior began with the nostalgic question, "Do you remember when there was just one TV and it was in the living room?" and then went on to remark that everyone now had their own TV sets and were watching them in their own rooms.

This is, of course, just one case of an issue that is endlessly debated by economists—what should go in the standard shopping cart used to define a basic standard of living.

Thanks to both of you. Your point about change is clearly crucial, John, as you show with that nice example. I was also thinking, Ted, about your paper on The Good Life when I posted this: the ideological assumptions behind poverty targeting and proxy means tests seem grotesquely far from local concerns about what would constitute a good life. Of course we can argue that the debate can take place entirely without anthropologists' involvement merely by looking at the statistics, but it strikes me that anthropologists should have a lot to say constructively about this.

One possibility for something constructive is the distinction between patina and fashion as sources of value that Grant McCracken draws in Culture and Consumption.

Where patina is a source of value, things become more valuable over time. Malinowski mentions, for example, famous pieces that have circulated widely through the kula, accumulating stories about dramatic encounters between the individuals through whose hands they pass.  Modern equivalents include famous works of art. One might also note the prevalence in fantasy fiction of magical swords and other objects with magical powers, the Ring that Rules them All in Tolkien, for example.

Where fashion is a source of value, things lose value over time. The essence of fashion is novelty, the newer the better. The whole fashion system turns on seasons, when what was once the height of fashion slips into obsolescence. Fashion has, of course, become the driving force in consumer electronics marketing as well as apparel. I think here of the dissatisfactions that I am beginning to feel with my first-generation iPad. I bought my wife a second-generation iPad as a Christmas present and now find myself looking longingly at the new fourth-generation device. 

One very interesting question is how fashion, prized for its novelty, can be elevated to art and then back to fashion again. The dress worn by Audrey Hepburn in Roman Holiday, for example. Collectors would kill to own the original. Not so for the Peterman knock-off my daughter once wore to a ball. 

On a larger scale, of course, there are landscapes venerated by eco-purists, who attribute to them the quality that Kant called sublime. Are they destroyed by the presence of tourists? What about strip mines? How far can change intrude while the ancient retains its value? 

I'm rambling now, but, returning to the topic at hand, this distinction may go to the heart of many complexities associated with treating things as proxies for something called the good life.

Along the lines of John's previous comments, in parts of Yucatan, Mexico, where proxies for poverty have long been used, there is talk now of discarding telephones and t.v.s as signs on not being poor (since virtually everyone has a cell phone and a t.v.) and using air conditioning instead. This points to the moving target of proxies (probably a good thing) and also to what would be a fascinating study of changes over time. There is a nice piece in the current issue of The Economist about what poor Kenyans will give up to keep minutes on their cell phones, including meat at a meal or a bus ride (see http://www.economist.com/news/middle-east-and-africa/21566022-repor...).
As a note, among Maya subsistence farmers, "poor" is locally defined most often as not having land. Poor indeed. And, as Huon suggests, such local definitions probably vary significantly and in ways that we should heed.

Thanks John, Ted. Again, John is pointing to issues of value that extend beyond and cross over the apparently simple point-of-sale-money values we tend to think of as a baseline and which you critiqued in your paper, Ted.

The mobile phone versus square meal sounds, as presented in the article, like a fairly simple issue of marginal utility; but whatever specific meanings the decisions involved have to people are surely best assessed on the ground: from my experience in the Caribbean I can vouch that, in addition to expanding the scope for social exchange, and sociability generally, having a mobile phone opens chances for a lucky hit on some unpredictable resource; for example a god sibling is coming to town with bags of coffee and may drop one off etc. Mobile phones have attracted quite a bit of attention from anthropologists including from OAC regular Erin Taylor who has worked in Haiti on phone credit/money.

Another point where anthropologists could decide to involve themselves is in exploring the knock on consequences of introducing different kinds of benefits. Poverty targeted benefits are often trumpeted even though older universal benefits are doing a better job at improving social conditions, he argues. But any kind of benefit once introduced may have further effects on social conditions which were never anticipated. Though some of the invidious effects of isolating a group of people in a community as 'the poor' can be fairly easily be anticipated; as can conflicts when people realise that some people have been included or excluded from benefits.

One of the points that Stephen Kidd is making which is worth looking into in detail is that there is a clear ideological mindset behind poverty targeting approach which goes something like this - "poor people are as specific group in society who represent a problem for the majority: that problem can be solved by targeting special moneys at that group, this is a better approach than universal provision because it is cheaper": the self-fulfilling 'proof' of this view is that poverty targeted programmes cost less money than the universal ones. As Kidd argues:

"As noted above, the authors assume that the  poverty targeted and universal programmes have the same budget. Yet an analysis of the political  economy of targeting would indicate that this is an entirely unrealistic proposition. Indeed, Pritchett (2005) – in an excellent World Bank paper regards the assumption of fixed budgets as “naïve.”  In practice, universal programmes usually have  larger budgets than poverty-targeted options, and frequently significantly more.

In part, this is because they generate greater political support than poverty-targeted schemes. By ensuring that everyone can benefit from the programme – both the poor and non-poor – they build alliances among the majority of citizens. In contrast, richer taxpayers tend to resent financing benefits that are only for the poor – and from which they are excluded – leading to poverty-targeted programmes receiving small budgets. The evidence can be seen in Latin America. In Brazil, while the Bolsa Familia programme costs only 0.4% of GDP, the Rural Pension – which is almost universal in coverage, though only in rural areas – costs around 1.3% of GDP. While Bolivia’s universal pension costs 1.1% of GDP, the budget of Argentina’s poverty targeted pension is only 0.2% of GDP.

Consequently, the assumption of equal budgets for both universal and poverty-targeted programmes leads to further considerable distortions in the analysis. It ensures, however, that poverty targeted programmes appear to perform significantly better than they would in real life."


Thanks for launching this thread, Huon. The problem is to know where to begin, whether to laugh or cry. I am sure you are right to argue that targetting "the poor" is a socially divisive strategy. The World Bank was founded with that explicit purpose, since when it has promoted PSD (more than PMT -- didn't they get the acronym's common meaning?), Private Sector Development or how to enrich the corporations. Cowen and Shenton in their excellent book, Doctrines of Development, suggest that development is one of two things: how to promote capitalism or how to put sticking plaster on the damage it does. Since the 80s, it has been mainly the second. There is a list of books as long as your arm showing that aid does more for the donors than the recipients. I recommend Judith Tendler's Inside Foreign Aid more than Dambisa Moyo's more recent Dead Aid. No-one in their right mind wakes up in the morning and asks "What can I do for the poor?". I recall a sociologist who found it incomprehensible that the inhabitants of Brazil's favelas were converting to pentecostalism in droves so that they could give their money to Atlanta televangelists. But the alternative was to be on the receiving end of good works orchestrated by the Vatican. I am sure no-one here would find it remarkable that my grandmother refused to join the "free library" or receive National Health spectacles because gifts were an instrument of feudal domination in her folk memory. Or, in the immortal words of Mike Todd, gambler and an Elizabeth Taylor ex, "I have been broke several times, but poor never". The very word reeks of inequality. Excuse all the aphorisms. This topic is too big for a short post.

But what I don't get is why you should think it is worth anthropologists using their concrete knowledge of households to point out that this kind of thing is a sham? The post-structuralists, such as Jim Ferguson in his excellent Anti-Politics Machine, have long ago shown that "development" aid is a fraud in any real terms. The Third World debt crisis saw a bigger transfer from poor to rich countries than any other in the history of imperialism, while aid budgets shrank. Under such circumstances it was just talk, or discourse as they prefer to say. The post-development anthropologists like Arturo Escobar have been banging this ideological drum for ages.

It is clear that you favour universal rights within social democracies of the kind that Europe once had and the BRICS are trying to build now, on the back of their capitalist booms. The best hope for people everywhere is to be part of such a state, not the recipient of western aid. Or, as my grandmother used to say, "If you can't pay, don't go".

But what I don't get is why you should think it is worth anthropologists using their concrete knowledge of households to point out that this kind of thing is a sham?

Fair enough - I admitted to John that you can do the debate without any reference to actual people or their interests; you only have to know the facts in general statistical and ideological terms. Your story rang a bell; my granny was very frightened, in the 1980s, of being put into a 'workhouse' for much the same reasons yours didn't use the 'free' library - albeit workhouses no longer existed by then.

But is your question, 'why would we engage in the specifics when we have known for a long time that in general terms that this is a con?' or is it 'why would anthropologists engage in these debates when their method of study and their results are irrelevant and won't change anything?' Presumably, one part of an answer would be to provide a completely different model - of the good life or the human economy.

Reply to Discussion




© 2014   Created by Keith Hart.   Powered by

Badges  |  Report an Issue  |  Terms of Service