Poor economics: a review

Picking up this book, written by two of the world's leading 'randomistas', I looked forward to endless analysis of randomised controlled trials, coupled with a self-satisfied belief that policy design should always, and only, derive from statistical analysis of carefully designed experiments. Up to a point, I was right. Abhijit Banerjee and Esther Duflo are, of course, enthusiasts for RCTs: that is the unique selling point of the Poverty Action Lab they founded at MIT. They obviously like statistics. I do too, and because I read this on my tablet, can report that the word 'experiment' occurs 84 times in the book, 'random' 76 times, and that 16 times there are references to Banerjee and Duflo's eighteen-country data set. On the other hand, there are 43 results for 'visit' and 24 for 'talk', so it is clear that Banerjee and Duflo don't spend all their time 'staring at data'.

In fact, the book is endearing for its rich anecdotal detail, from countries as far apart as Morocco, Mexico, Kenya, India and Indonesia, among others. It is often Banerjee and Duflo's conversations, sometimes with each other, more often with poor people, that inspire new ideas and lines of enquiry. There is no shortage of evidence from RCTs. However, the key methodological lesson of the book is the need for unblinking tenacity in unravelling the tangled ball of wool that presents itself to anyone trying to understand the livelihood strategies of the poor, and the policies which help or hinder. Banerjee and Duflo describe this as 'unravelling mysteries'. Sometimes, the right way to unravel a mystery is to run an RCT. But sometimes, the right way forward may be to sit down with poor women, men and children, or with Government officials, or NGONon-governmental organisation workers, and simply talk it through. Banerjee and Duflo don't appear to have much time for participation or participatory methods (zero references), but they do acknowledge the value of qualitiative methods (1 reference). RCT: Rigour, Creativity, Talking!

The book consists of ten chapters, plus a kind of postscipt, entitled 'In place of a sweeping conclusion', which nicely summarises the key substantive argument: that careful policy design can reshape institutions, re-order politics, and change lives. 'We may not have much to say about macroeconomic policy or institutional reform', they say, 'but don't let the apparent modesty of the enterprise fool you: small changes can have big effects'. De-worming children in Kenyan schools, sprinkling micronutrients onto children's meals, nudging people with incentives, all these change lives.

The ten chapters consist of an Introduction, four chapters which deal with food, health, education and family planning, and five which tackle 'institutions': insurance, credit, saving, entrepreneurship and politics. The text touches on more topics than this list might suggest - there is quite a bit on migration, for example, and on agricultural innovation - but still this is a slightly selective list, which may say something about the scope of RCTs. There is not much on labour markets, for example, or industrial policy, or transport. It is easier to experiment with de-worming tablets than with motorways or bridges. Is the focus on 'small changes', one is left wondering, a methodologial necessity?

The four thematic chapters set out to unravel mysteries, tell stories and also demonstrate Banerjee and Duflos' willingness to take on big names and big debates: Sachs and Easterly feature prominently, on such topics as poverty traps and the role of the state.  An important, but unstated theme is the relative importance of culture and economics. For two self-confessed economists, Banerjee and Duflo opt surprisingly often for the side of culture.

When it comes to food, for example, there is apparently no poverty trap, because the poor could mostly afford to eat properly if they wanted to, but instead choose to spend on small luxuries (which in Morocco at least include satellite television and mobile phones): investment in de-worming and micronutrient supplementation is a better investment than subsidised food. That seems to me to underplay the existence of food poverty and to ignore the role of poor sanitation in causing under-nutrition. It is also not quite consistent with Banerjee and Duflo's later interest in the devastating impact of shocks on family welfare, and their enthusiasm for social protection.

In the next chapter, on health, they describe an RCT in which people are so enthused by an incentive of only 2 lbs of lentils that immunisation rates increase sevenfold. We are told that the reason is not because they are hungry. Instead, the reason (hypothesised not tested) is 'time inconsistency': 2 lbs of lentils now acts as a nudge, providing a short-term incentive to mothers to accept the cost and inconvenience of carrying out an action (immunising their children) which has only long-term benefits and is easily put off. Banerjee and Duflo argue that it is rather like incentivising people to keep New Year's Resolutions, such as going to the gymn. A pity that this idea is not subject to more rigorous scrutiny and testing. It might not be possible to devise an RCT, but qualitative or participatory approaches might help. For example, groups of women could be asked about the actual costs of participating in immunisation programmes, which are probably not trivial.

When it comes to education, incentives again work, though not as expected, allegedly for cultural reasons. Conditional cash transfers have been much-lauded, but Banerjee and Duflo argue that the cash has the impact, not the condition: school enrolments rise equally when people are given cash without the requirement to enrol. They hypothesise (again without testing) that this is because people quite want to send their childen to school but can't afford the costs, both direct (uniforms, books) and indirect (lost wages). They don't quite bring themselves to say so  - in fact they come close to arguing the opposite - but this suggests that parents do recognise the benefits of education and take the long view. Banerjee and Duflo rather think, from anecdotal evidence, that poor parents, abetted by the school system, fall victim to a culture of low expectations. It would have been interesting to ask parents systematically.

The chapter on family planning, again, veers between economics (people need many children because they will depend on surviving children in old age) and culture (grandparents play caring roles). This is a familiar see-saw, for example in the debate about falling fertility in Bangladesh, and it is a pity it is not addressed as such. Naila Kabeer is one who has worked extensively on the issue.

If poverty traps receive short shrift in the first part of the book, they are treated more kindly in the second. The poor suffer greater risk than the rich and are harder hit by shocks - and this makes them depressed, as it would. Livelihood diversification and mutual aid provide some insurance, but not enough, especially on the health side (and, Banerjee and Duflo might have added, are not much help in the case of co-variate shocks, for example when whole regions are hit by drought). The mystery to be unravelled is why cheap insurance does not catch on. It is also a mystery why people don't save more. Cost-effectiveness and lack of trust in institutions seems to play a role, but we come back again to a hypothesis about 'time inconsistency' - crudely, people can't be bothered. Is that consistent with all the evidence about diversification? When it comes to insurance, Banerjee and Duflo see a case for state subsidy - interestingly not because of externalities (as in the case for immunising children to protect the herd), but on pure welfare grounds. On savings, they are in classic interventionist mode, exploring ways in which traditional savings groups can be strengthened, in order to bolster people's enthusiasm and put steel in their otherwise weak-willed backbones: for example, developing new savings accounts, persuading people to drink less tea and save the money instead,  or giving people locked cash boxes and taking away the key! For the record, the last option does not work.

A similar enthusiasm for programme adjustment is evident in the chapter on micro-credit, which is not as critical (or wide-ranging in its analysis) as some recent literature, but does make important points about the inflexibility of many programmes. Microcredit is of modest help, but does not help escape a poverty trap. Partly, as we discover in a chapter on the poor as entrepeneurs, this is because the poor don't actually want to be self-employed business people. For many, running businesses, for example small shops, is a livelihood strategy of last resort. They would much rather have a job, preferably in the Government or otherwise in the private sector. Banerjee and Duflo conclude: 'Microcredit and other ways to help tiny businesses still have an important role to play in the lives of the poor, because these tiny businesses will remain, for the foreseeable future, the only way many of the poor can manage to survive. But we are kidding ourselves if we think they can pave the way for a mass exit from poverty'.

The last substantive chapter of the book continues the theme of making small programme adjustments. Banerje and Duflo acknowledge the importance given to institutions in the current debate about development, and note particularly the pessimism that many feel about the scope for improving accountability and performance.  They argue, however, that progress can be made by leaving aside the big debate about 'INSTITUTIONS', written in capital letters, and focusing instead on marginal improvements to 'institutions', written in small letters. Examples include publishing information about school budgets, using 'mystery shoppers' to monitor the efficiency of police stations, improving the design of ballot papers, and introducing quotas for female leadership of local government councils. All this appears consistent with research on real world politics, in Africa at least, such as that led at ODIOverseas Development Institute (London) by David Booth, which has the theme of ‘working with the grain’.

Banerjee and Duflo draw breath at the end of the book, and propose five conclusions drawn from their wide-ranging review of the lives of the poor:

1. The poor often lack critical pieces of information and believe things that are not true.
2. The poor bear responsibility for too many aspects of their lives.
3. There are good reasons that some markets are missing for the poor, or that the poor  face unfavourable prices in them.
4. Poor countries are not doomed to failure because they are poor, or because they have had an unfortunate history.
5. Expectations about what people are able or unable to do often turn into self-fulfilling prophecies.

These provide the platform for their final plea, reported above, to design and implement small changes, nudges that work.

Poor Economics is readable and grounded. I certainly admire Banerjee and Duflo's painstaking commitment to unravelling the mysteries, and they have many insightful anecdotes. I fear they are on dangerous terrain and sometimes wrong on the great culture versus economics debate and could wish that they had brought a sociologist, anthropologist, political scientist or psycholgist onto their writing team - preferably one of each. And on the other great debate, between quantitative and qualitative methods, especially the role of RCTs,the evidence here is convincing that they have a place. However, I note Banerjee and Duflo’s opening acknowledgement that a plurality of methods is needed, and wish they had obeyed their own injunction.

________

‘Poor Economics: a radical rethinking of the way to fight global poverty’, by Abhijit Banerjee and Esther Duflo, is published by Public Affairs, New York, 2011