Wednesday, February 24, 2010

But *WHY* is aid expertise not scalable?

My last post, "Aid Expertise is Not Scalable," drew a surprising amount of reaction.  Two former World Bank colleagues said "Amen!"   However, two people I respect a lot (they are leading experts in their field, in fact) replied along the lines of "Maybe, but why not?"
That is a fair question, and the full answer would be a great subject for a book.  
But here are a few thoughts that come to mind:
  1. First, expertise in some fields is inherently more scalable than in others.  Civil engineering, for example, is much more scalable, because the technologies to build a bridge are: (a) known with a high degree of confidence; (b) effective in a wide range of contexts, with relatively minor adaptions; and (c) transmittable to other potential experts using a commonly accepted language of mathematical symbols, standards, and procedures.  
  2. Economic development is more a social rather than civil engineering challenge.  And the problem is that we don't know very well how to effect the social changes we seek.  To the extent that we do know how to address a certain social challenge in one context, it is often much more difficult to transplant that solution to another context in another country. And even if the solution would work in other contexts, we usually don't have a good and commonly accepted "language" to enable that expertise to be transmitted quickly and effectively to other potential experts.
  3. The sheer number of topics on which we need expertise is enormous.  And if we multiply the number of topics by the number of different countries/contexts aid agencies work in...well, you get the picture.  A head hunter called me the other day to say that an international agriculture commission had hired her to find the best one or two "agricultural experts" in the world to act as advisors.  I explained, much to her disappointment, that agriculture covers a huge number of sub-specialities, ranging from seed breeding to soil science, to harvesting and processing.  She needed to pick a speciality, and I would try to deliver the expert.  (Believe it or not, in the late 1980s, I was one of the World Bank's experts on rubber processing in Indonesia, and I knew a lot about that topic but virtually nothing about seeds or soils.  And, to top it off, rubber processing in Asia was at that time different than rubber processing in other countries, so my expertise was of limited use in those places.)
The so-called bloated official aid agencies are not all that big anyway.  IBM has about 400,000 employees worldwide, and HP has about 325,000. What they do is complex, but not compared to the challenges of economic development.  The World Bank, by contrast, has roughly 10,000 employees and is constantly under pressure to shrink rather than grow, based at least partly on the perception that the marginal hire there is not adding value.  Even if the Bank were able to convince its stakeholders that it had figured out how to scale expertise, it is  highly unlikely that it would be able to grow to anywhere near the size of an IBM.
The only practical way to proceed is to abandon the idea that expertise is scalable within aid agencies,  and thus that aid agencies must have the top experts in every field. The current conceit of "expertise-ism" gives aid experts the power (and even responsibility) to decide on the allocation of very large amounts of resources.  This forces them to make judgements about many things about which they possess insufficient information and insufficient expertise.  Deep down, most conscientious experts at big aid agencies feel that this is a heavy and unmanageable burden for them personally, and naturally it leads to a lot of mistakes, even by smart, motivated, and caring experts.
Instead, the aid industry needs to embrace the idea that expertise is only scalable in a highly distributed and localized way within the countries that are trying to develop.  As part of this, we must abandon the idea that each expert must (a) be the creme-de-la-creme in terms of education and qualifications, and even ability; and (b) be correct every time.  Development will occur best if a lot of reasonable smart people experiment with a lot of different things in a decentralized way.  Many experiments will fail, and instead of piling on the criticism and critical evaluations, the system itself must focus its attention and efforts on the things that work, and take those to (the appropriate) scale for the contexts in which they are effective.
The information, feedback loops, and incentives needed for such a change to occur are a topic for another post.  But I will end here by observing that those needs will require a different skill set at aid agencies -  one that helps their experts become the equivalent of orchestra conductors rather than solo performers (or first-chair violins.)  And that is actually an exciting an energizing idea that some of the best aid agency leaders are starting to explore now.