The Lean Startup and Poor Economics may seem like very different boks. While the first is a business book for startups, the second one is about development economics. Yet after reading them back to back, I realized they are not only about the same thing (learning), but that together they suggest a very slick method to think about economic development.
In Poor Economics, Esther Duflo and Abhijit Banerjee (D&B) argue that “so much of anti-poverty policy has failed over the years because of an inadequate understanding of poverty.” They then go on to share how hundreds of rigorous randomized controlled trials have helped us learn about poverty, and explain how using that knowledge is key in the fight against it.
In The Lean Startup, Eric Ries tells us that startups fail because they have an inadequate understanding of who their customers are, or even of what their product should be. He then goes on to argue that in that uncertain environment startups exist to learn, and that the learning process holds the key to a startup’s success.
Both books are excellent reads alone, but together they were a real source of inspiration. Combining rigorous testing methods to learn about poverty with practical startup management methods could help development practitioners reach some transformational outcomes. Imagine how much great stuff could come out of a crossover between the World Bank and Dropbox!
But how would that actually work? One way would be to apply “lean methods” to development projects such as the build-measure-learn loop as described by Ries:
Build – Create a minimum viable product (i.e. enough of a product to allow you to learn)
Measure – Test your assumptions against an appropriate accounting system
Learn – Decide how to proceed or change course
But are randomized control trials anything other than a way to build, measure and learn from evidence? I think they are pretty close.
There are two lean startup concepts, however, which are not obvious in D&B’s work and which I think we could stress more in development today.
The first is the use of a minimum viable product (MVP), which is the idea that early products should focus on collecting the maximum amount of learning from a minimum amount of effort. A minimum viable development project (MVDP) would be a ruthlessly cheap development experiment that would allow us to learn about development outcomes D&B-style. For example, simple mobile classrooms or clinics could test a community’s needs before the decision to build a full-fledged school or hospital. And while I acknowledge that pilot projects are not something new, and are the basis of much of the learning in Poor Economics, the emphasis here is on the words “maximum” and “minimum.”
The second is the idea of acceleration, which is the quick iteration through the build-measure-learn cycle. Learn quickly. Decide quickly. Avoid waste. While institutions like the World Bank are indeed focused on learning and their project cycle does include evaluation, the timeframe for most projects is very long. And while in some instances development outcomes do require a much longer timeframe of evaluation, I also think there are projects out there that could be launched, tested and changed in a matter of days, creating a great deal of learning and improving development effectiveness.
I acknowledge that policy-makers often operate under tighter constraints (including political ones) and have to solve for many more variables than business leaders, but still there should be some startup lessons learned that we can apply to policy-making and development.
BACK in 2008 I wrote a piece where I suggest a framework for thinking about development called freedom diagnostics. The idea is that policy-makers can focus their efforts on those policies which give the greatest bang for the buck in terms of human capabilities (or as Amartya Sen calls them, freedoms). The way to do it is to focus on the interactions amongst these freedoms. At the time I was taking classes with Rodrik, Hausmann and Sen, so it was just natural for me to merge the idea of growth diagnostics with a more broader-based idea of what constitutes development. For the first post in this blog I will post that piece. Look forward to your comments.
ON THE NEED FOR FREEDOM DIAGNOSTICS
A Tailored Approach to Economic Policy
Even before the global financial crisis, the failure of the one-size-fits-all approach to economic growth was already clear. The standard laundry-list policy prescriptions with neo-classical flavor (so called Washington consensus) had been discredited amongst development economists and growth diagnostics was slowly taking its place. The new approach not only promised to be more effective, but also came with a significant jump in modesty – it was full of humbling words such as ‘contingent’, ‘constraints’ and ‘priorities’. Today, growth diagnostics has gone mainstream and the crisis has forced even developed countries to accept ‘unconventional’ policies – the mantra now is a tailored approach.
However, while recognizing the importance of a tailored approach, the growth diagnostics framework also rejected a nihilist attitude where ‘anything goes’. Although the right thing to do may change from country to country and from time to time, there is such a thing as ‘the right thing to do’: one must find and relax the binding constraint to growth. The implication of growth diagnostics is that a small concentrated effort can have a huge impact on the economy, allowing politically and administratively constrained policy-makers to focus their limited resources where they matter the most.
But is it Just the Economy, Stupid?
Mainstream development theory has not been this exciting ever since Amartya Sen came out with Development as Freedom. Sen reminded us that poverty is not just about money, it is about much more than that. It is a lack of capabilities to do those things which we have reason to value. It is a lack of freedoms. Sen’s work went exactly where growth diagnostics doesn’t go – economic growth is not the only aspect of development. Back when Sen raised a flag, we nodded and agreed to think more broadly about development. But the way we approached the problem of broad-based development was the exact same: we wrote laundry lists just like the Washington consensus: “Sure, give me a bit of free speech. Oh yes, literacy please. Can I have some higher life expectancy to go with that education? And since I am at it, why don’t I take some free and fair elections too.”
Should we trash those ‘freedom lists’ as quickly as we got rid of the Washington consensus? I think not. For one, there is an obvious difference between these freedom lists and the standard policy reform lists which failed to promote economic development. Freedoms have intrinsic value in and of themselves. Higher literacy, for example, is a goal of its own and is therefore constitutive of development. Meanwhile, there is nothing intrinsically valuable about fiscal discipline – we only value it to the extent that it helps us achieve other development goals.
Nevertheless, the case for a better alternative to freedom lists and cookie-cutter governance reform remains strong. For policy- makers who need to decide what to do next, the problem remains the same. How can we prioritize? Which freedoms should we focus on first? Of course we would like to have all of them at once – that is technically correct – but in practice that is just not feasible. Is there perhaps a way to apply the lessons of growth diagnostics to these freedoms lists? Can we pursue something like Freedom Diagnostics?
Let’s consider…
Freedoms beget Freedoms
First, as Sen points out, we should recognize that freedoms are not only constitutive of development, but are also instrumental to development. There are important interlinkages between freedoms. Political freedoms can help achieve economic security, while social opportunities facilitate economic participation and so on. Sen focuses heavily on the complementarities between freedoms, but there may be situations where a clear positive linkage is not so clear. While at times, such complementarities may be very strong, at others they may be rather subdued. It is more or less widely accepted that simply introducing elections on a country may not lead to additional political freedoms or better social or economic opportunities. The post-independence African experience is a case in point. Conversely, increased economic opportunities in Asia have often failed to generate increases in other substantive freedoms.
It’s the Interactions, Stupid
A key lesson of growth diagnostic is that in an under-performing economy, distortions and imperfections are rampant. Therefore the effect of a policy which eliminates or introduces one or more distortions in a market will have interaction effects across all markets and deliver mixed results. We must take these interactions in consideration when evaluating the overall improvement of our policy, which can often be small and at times even negative (a second-best complication). The astute decision-maker must then focus on the distortions which will have the greatest overall effect. Those are the binding constraints to the economy. The lesson is that we can have the most impact by crafting policies which are mindful of interactions across markets.
Conceptually, it is not difficult to merge the two ideas by focusing on interactions to imagine the possibility of Freedom Diagnostics. By being mindful of the connections among different freedoms, we should be able to focus on promoting those capabilities which will have the greatest overall impact. Here one may say that the intrinsic value of freedom poses a moral difficulty with choosing among freedoms. It would be at a minimum uncomfortable to say that an increase in a specific freedom can have an overall negative effect on welfare. (for example when people say that a country is not ‘ready’ for democracy.) But the approach is not dependent on the existence of overall negative effects, only on differential effects when a particular freedom can give an additional bang for your buck.
Most importantly, a freedom diagnostics approach is only meant to serve as a practical way of approaching policy-making in an imperfect world. If there are multiple freedoms lacking and if there are interactions between them, there must be a freedom which, if introduced, will have the greatest impact on others. That would be the binding constraint to freedom. Once we appreciate the existence of such a constraint we should strive to find it and relax it. That way we can focus our efforts where they can have the most impact.
Back to Reality
Of course, although freedom diagnostics is conceptually possible, in practice its application would be a challenge. Compared to growth diagnostics, freedom diagnostics would probably involve an even longer time-horizon of analysis. (How long does it take for political freedoms to have a significant impact on social outcomes?) Moreover, much learning is needed to establish all of the interactions between freedoms. However, the existence of difficulties or even the impossibility of coming up with a perfect freedoms diagnostics framework should not lead us into disregarding the implications of this line of thinking. We cannot but decide on our policy priorities.
A Way to Focus Broad-based Development Efforts
To conclude, we cannot ignore Sen’s call for a broader informational base in evaluating development and focus solely on growth. And to be fair to growth diagnostics, it does not claim to be a theory of long- run broad-based development. It simply focuses on growth, which is an important part of a broader picture. Even the most enthusiastic proponent of growth diagnostics would agree that the effects of growth alone do not constitute development. Growth diagnostics could be but a part of an overall freedom diagnostics framework – a framework which could provide us with a way to focus our broad-based development efforts.
