A new paradigm in development economics?

Until very recently, if you spent anytime thinking about development policy, the chances are that you fell into one of three groups.  One group believes the problem with developing countries is lack of resources. So the solution is a vast increase in foreign aid.  A second group believes the real problem if lack of incentives. So the solution is more and better markets. The third group thinks the problem is lousy governments, so the answer lies with improved governance. I leave it to the reader to identify these positions with their most distinguished (or at least most vocal) representatives...

But there is something new afoot. Increasingly, some people are saying the right way to approach development policy is to start with the view that we actually don't know where the problems lie, to acknowledge that the key problems may differ from setting to setting, and to adopt an explicitly experimental attitude to policy selection and formulation so that you can learn about the environment in which you operate. In this approach, monitoring and evaluation are key, as you want to pull back from mistakes and improve policies over time. Indeed, you build the monitoring into the policy process itself so that learning becomes part and parcel of it--rather than something you leave to your researchers or economists.  This way of thinking about development policy is radically different from the three schools I summarized above, as it admits much greater diversity and heterodoxy. It is humble about the extent of our knowledge but optimistic about our ability to learn. 

This of course is the way I have been thinking about development policy for a while (having been influenced by the work of Roberto Unger and Chuck Sabel). It was striking to hear Esther Duflo articulate this afternoon a virtually identical perspective with respect to social policy. Esther was giving the Wellington-Burnham lecture at Tufts and I was her discussant. As I said in my comments following her presentation, this is nothing less than a paradigm change in development. It is extremely encouraging that people who are approaching development from rather different ends are converging on an identical perspective on how you think about development policy.

Now, Esther and I differ on what we mean by "evaluation": I think randomized field experiments have serious limitations and problems of generalizability ("external validity" to use the jargon) and that we can actually learn from a variety of other types of evidence. Sometimes it seems as if Esther thinks anything other than randomized evaluations is just "stories." But this is a discussion for another time.  For the moment, I rejoice that "macro" types like me and "micro" types like her are articulating a very similar line...

Comments

[jose carlos]

hi dani, liked very much your post. but i feel you tend to dismiss case studies as mostly 'just stories'. if this is the world of uncertainty and multiple equilibria, and self-discovery is the way to go, we need more and better 'development narratives' too. like the work of tendler or evans, which i find way more insightful than most of the narratives of the volume you edited.

[Greg]

Sachs, Easterly, Collier?

[Ron]

what's needed is a venture capitalist approach to development. Try a whole bunch of things with the understanding early on that 90% of those trials will fail (meaning you leave people no better off than they started). But when you hit upon the 10% that succeed, you uncover the Google of development

[David Bonbright]

I agree with your observation and see the same hopeful trend that you do. I work with a UK-based international citizen organization that seeks to maximize the developmental impact of citizen organizations, business and government, through designing and promoting innovative, practical methods of planning, doing and communicating that foster learning and responsiveness among all constituents. We believe that organizations function best when they listen and respond to those most affected by their activities. Our methodology helps civil society organizations to be inclusive and responsive in their engagement with constituents, systematic in their learning, and transparent in their public reporting. We have developed a suite of free-to-use prototype tools and offer advisory services that citizen organizations (as well as businesses that seek to maximize their social value) can use to plan and learn with their constituents, in an ecology of actors, for significant and lasting social change. We also seek to influence policy and practice for social impact through strategic research and publications. www.keystoneaccountability.org.

[haider A. Khan]

Indeed this humble approach is ultimately more rational than applying"rational expectations". A combination of a case-by-case approach where learning plays a key role can help us economists learn better where the pitfalls in policy making in specific contexts might lie. One of the challenges, of course, is to design appropriate institutional feed back processes. This approach can, I think, be applied to macro, meso and micro levels.Contrary to what some may think, this approach can also be formalized to some extent so that the charges of loose or unrigorous thinking can not be so easily levelled at this approach. Of course, your own work, Dani and some of your quasi-methodological observations on this blog illustrate this point. In my 2004 book on financial crises, I tried to show how banks in Taiwan (and perhaps elsewhere) could learn from past data on default rates and other relevant information but that this learning was a case of fuzzy logic and "bounded rationality".
The question was:Can banks learn to be realtively more rational in their lending activities? The answer depends on the institutionally bounded constraints to learning. From an evolutionary perspective the functionality (for survival) of "learning to be rational" creates strong incentives for such learning without, however, guaranteeing that each member of the particular economic species actually achieves increased fitness. I investigate this issue for a particular economic species, namely, commrercial banks in Taiwan. The purpose of this exercise was to illustrate the key issues related to learning in an economic model. I ended up by proposing a new screening model for bank commercial loans that uses the neuro fuzzy technique. The technical modeling aspect is integrally connected in a rigorous way to the key conceptual and theoretical aspects of the capabilities for learning to be rational in a broad but precise sense. This approach also compared the relative predictability of loan default among three methods of prediction--- discriminant analysis, logit type regression, and neuro fuzzy--- based on the real data obtained from one of the banks in Taiwan.The neuro fuzzy model, in contrast with the other two, incorporates recursive learning in a real world, imprecise linguistic environment. The empirical results show that in addition to its better screening ability, the neuro fuzzy model is superior in explaining the relationship among the variables as well. With further modifications,this model could be used by bank regulatory agencies for loan examination and by bank loan officers for loan review. The main theoretical conclusion to draw from this demonstration is that non-linear learning in a vague semantic world is both possible and usefulif prper attention is paid to contextuality. Therefore the search for alternatives to the full neoclassical rationality and its equivalent under uncertainty---rational expectations--- is a plausible and desirable search, especially when the probability for convergence to a rational expectations equilibrium is low.

[Tim Worstall]

"Increasingly, some people are saying the right way to approach development policy is to start with the view that we actually don't know where the problems lie, to acknowledge that the key problems may differ from setting to setting, and to adopt an explicitly experimental attitude to policy selection and formulation so that you can learn about the environment in which you operate. In this approach, monitoring and evaluation are key, as you want to pull back from mistakes and improve policies over time."

Umm, isn't this just a restatement of the more markets approach? Markets are, in one formulation at least, the best method of discovery that we have.

[Jonathan Davies]

Dani

As a graduate student at UEA in the early 1980s a key text was Dennis Rondelli's "Development Projects as Policy Experiments" (1983). I have been fortunate to work for many years in developing countries working with development agencies, governments and civil society groups looking for solutions (usually seeking your second-best institutions). The 'pick-and-mix' approach is the only way forward. The difficulty often lies with donor's fixed ideas and development fashions.

[Barkley Rosser]

I am wondering if this posting is at least partly a reply by Dani to my request to him to comment on behavioral approaches. These tend to be closely linked to experimental economics, although they are not identical, and he here is, more or less favoring "field experiments," if with some caveats. I applaud.

Of course, within experimental economics there ia a large war going on now between advocates of lab and field experiments, with the former criticizing the lack of controls in the latter and the latter criticizing the potential unreality and framing that can go on in the former. Of course, most observers say "do both," but when limited funding is involved, people can get nasty.

In any case, I would guess that field experiments may have an edge in developing countries, although lab experiments may be useful in pinpointing specific behavioral differences that occur in specific societies, which seems to be part of the theme here, that there may be different strokes for different folks.

[bee]

I'm pleased to see that development economists are beginning to embrace the concept that lies behind contingency theory. Maybe we will finally begin to make progress in this field that translates into material insight into development.

[david]

"I think randomized field experiments have serious limitations and problems of generalizability ("external validity" to use the jargon) and that we can actually learn from a variety of other types of evidence."

Yep-- but I can't imagine anyone thinking differently and being taken seriously. I know they do and are; it just doesn't register as possible in a world of sensible people.

[jon]

Dani,

Interesting post. I'm fairly new to the world of development economics as an academic discipline, though having worked on some of the development "front lines" in the areas of HIV/AIDS and agriculture I'm certainly familiar with how those economics often play out in the real world. For a variety of reasons I'm trying to become more conversant in the somewhat more ephemeral world of academics/policy and so far my reading has taken me through what I'm fairly certain are your three most vocal proponents of the three most common schools of thought. However, I haven't found myself comfortable with any of those positions in and of themselves and I am now very intrigued by your so called "new paradigm in development economics" - for one reason because it sounds very similar to how good ag development happens - and I am wondering if you can point me in the direction to where I might read something from folks who are thinking and working in this direction - authors, journals, organizations - any and all would be welcome. Thanks, I always enjoy your blog and find it good food for thought.

[scott Bayley]

The work of UK evaluators Pawson and Tilley (they wrote the book 'Realistic Evaluation') helps to put Dani's comments into a broader evaluation context and also shows how to deal with the issue of 'external validity' in evaluation research. Pawson & Tilley like to advocate that outcome evaluations should focus on:
-What works?
-For who?
-In what circumstances? and
-How?

They believe that this approach offers much better 'contextually relevant' evidence for decision-makers compared to the traditional experimental approach (see the bottom of page 2 of the article below. The paragraph that starts off 'Realistic evaluation takes a different view of what constitutes experimentation from that which has come to prevail in orthodox evaluation circles').

http://www.danskevalueringsselskab.dk/pdf/Nick%20Tilley.pdf