Complexity and Development lecture at CGD

On February 5 I had the pleasure of hearing a lecture given by Owen Barder at the Center for Global Development, entitled “Complexity Theory and Development Policy.”  This blog comprises the notes I took during the lecture.

Overview: The lecture and presentation were a Complex Adaptive System (CAS) Theory  101 for the uninitiated and applies it to development at the high theory level (with hints to policy).  There were many book references throughout, shared below.  The audience was full of SAIS professors as well as the World Bank, former (retired) USAID staff, and some interns at various NGOs. CGD streamed the lecture live because there was so much interest in this topic.  Main point: we need to be able to fail and to document and share the anatomy of each failure, recognizing system-wide effects.

Intro: The first slide was the famous chart depicting GDP growth in South Korea and Ghana over the past 40 years or so.  Why did South Korea rise spectacularly and Ghana flatline? (Spoiler alert: EVERYONE can list a few reasons why this is so…and then he didn't return to this chart to tie it to CAS so…we’re still wondering about that.) He referenced Why Nations Fail by Acemoglu and Robinson who posit that corruption leads to bad policies; to change institutions, you need to change politics first.  Owen believes this book would explain everything if it included just one more chapter…and I’ll make you wait for that just like he did in the lecture.

Toaster project: a guy tries to build a toaster from scratch and succeeds somewhat; this experiment illustrates the benefits of economies of scale and the value of trial and failure.  “Development” has not followed this model.

The past 50 years of development saw: 1) fastest progress; 2) no explanation for difference; 3) questioned the existence of a missing ingredient; and 4) led to thinking that all is endogenous—a function of a system that cannot be affected from outside.

Adapt by Harris emphasizes learning from failure.  Testing through trial and error, adaptation and iteration is a better way to solve problems than trying to engineer a solution from the outset.  And an observant audience member pointed out that in that case, success cannot be separated from failure, to which Owen agreed.  Models and engineering solutions are not appropriate for development.

Industries, not firms, adapt.  Eighty percent of innovation results from firms going bust and new firms starting. Witness the fall of Barnes and Noble to Amazon, which then leads to Congress rethinking interstate commerce.  Firm goes bust and institutions adapt.  Adaptations affect each other – interlocking adaptive-ness yields a complex system.

Complexity overview: Origin of Wealth by Beinhocker describes an economy as a CAS.  Characteristics of a CAS:

1.       Butterfly Effect

2.       Predictable at Scale

3.       Emergent Properties – like thunderstorms; these are non-linear systems with system-wide properties

4.       Tend to complexity

5.       State of perpetual disequilibrium – periods of time look like they are relatively stable then there is seemingly a sudden change because all units are constantly adapting to each other

These are familiar traits in the study of economics.

Development: slight adaptation to Amartya Sen’s description—there is a need for a system that makes it likely for people to live out their life choices.

Discussion: Development is an emergent property of a CAS. What are the properties of the system that bring about this emergent property?

What can we do to accelerate the evolution of systems? To affect the rules that govern relationships of elements in the system?

1.       Resist engineering.  Instead, iterate, adapt and learn through trial and error rather than designing answers

2.       Resist fatalism

3.       Promote innovation and variation – equity promotes innovation, e.g. a social safety net gives people the freedom to innovate (leftist view).  The right side would say, eliminate regulation in order to get out of the way of entrepreneurs.

4.       Embrace creative destruction (Schumpeter’s term) – e.g. the evolution of the music industry

5.       Shape development – what is the fitness function.  Create selection pressures.

6.       Embrace experimentation

7.       Act global – open markets to poor countries, encourage migration, etc.

We want to accelerate adaptation.

 If development is an emergent property of a CAS then development policy should promote adaptation.  Does thinking about it this way help us?

How can we fail safely?  We should expect successes in portfolios, not individual projects (Remember—he’s an economist!).  We should package projects in portfolios (CSHGP, anyone?).  Take the implementation risk out of the public sector and put it in local groups, like “cash on delivery” aid.

Aside: DFID is an interdisciplinary agency and adopts interdisciplinary approaches.

Support of microfinance initiatives has blunted financial environmental pressures in some markets, creating a protected class.  Development should do the opposite—it should strengthen adaptive pressures.  And failure needs to generate feedback and learning.

Big question:  measurement (coming from the Bank, no less). Owen admits this is tricky and requires more thought: In CAS collect less data—focus on the part of the system you are working on.  Editorial note: I don’t agree with the second part.  Less data, of the right kind, is appropriate but to focus on measuring/documenting just the part of the system you’re working on is exactly where we are now.  I think we need to describe systemic effects as best we can.  That kind of documentation is significant for iterations and learning.

Ah, you’re wondering about his reference to the missing chapter in Why Nations Fail – it’s the identification of politics as an endogenous property that co-evolves with everything else.

Let's invite him to a CEDARS happy hour!