Chris Blattman recently posted a powerful argument that as development policy, skills training is a bad investment. There is just very little evidence that it is effective.
From 2002 to 2012 the World Bank and its client governments invested $9 billion dollars across 93 skills training programs for the poor and unemployed. In lay terms, that is a hundred freaking million dollars per program.
Unfortunately, these skills probably did very little to create jobs or reduce poverty.
Virtually every program evaluation tells us the same thing: training only sometimes has a positive impact. Almost never for men. And the programs are so expensive—often $1000 or $2000 per person—that it’s hard to find one that passes a simple cost-benefit test.
In a much longer discussion paper, he and Laura Ralston call for more research, and for a greater focus on capital-based programs and the potential for complementarities with skill-based problems. In short, rather than train people to do something, either give them money to buy something, or build them something yourself. And then maybe see if skills training them helps even more.
Like any good paper by any self-respecting social scientist, one of their main recommendations is for more research that they and their students do. More evidence, more micro-pilots, more multi-country interventions to learn about context. I agree. But in the spirit of Chris’s invitation for “discussion, and comments and criticisms” let me suggest a role for more ethnography and more institutional analysis.
It always strikes me how different the view of (say) the World Bank is from that of the local entrepreneur, laborer, or mother who works at home. My immediate thought when I hear that any individually-targeted development intervention has failed is “well, could it have succeeded?” In other words, does the intervention manipulate a binding constraint for an individual or household? The point is not that I’m sure that all interventions fail to do so, but rather that I rarely have good reason to suspect that they will. The suggestion that follows is to learn more about how individuals and households make decisions in the local contexts in which they live their everyday lives. Find out what those constraints are, and then try to push on those. The people who know how to learn about those everyday constraints are trained in ethnography—and I mean serious ethnography, the kind that involves languages and staying outside of a hotel.
A focus on institutions implies a different direction. Everyone agrees that institutions are important, but the cutting edge in development economics and related parts of political science focuses elsewhere. Why? Because institutions aren’t manipulable, their features bundle lots of treatments, core concepts remain tremendously fuzzy (try defining governance, for example), we don’t seem to have learned a lot from decades of studying them, and the potential for disaster from bad institutional design is just enormous. Yes. But I don’t see a way around taking formal and informal institutions seriously in development policy. It is not true that variation in development outcomes is purely a function of individual characteristics aggregated to the region or country. Context is not reducible to individuals, and institutions are part of that context. Learn about these institutions, both formal structures and informal practices and conventions, even if they aren’t manipulable. Even descriptive knowledge about institutions can be immensely valuable.
So that’s my suggestion. Want to design better policy? Think both smaller and bigger. Understand local constraints (small) and institutions (big). Both of these things imply that there is knowledge out there that is useful, yet non-experimental in nature. By implication, effectiveness of interventions depends on factors that cannot be manipulated as part of an intervention’s design, but which should guide implementation and cost-benefit calculations anyway. I find it hard believe that anyone would disagree with these ideas—to me, my comments in this post are almost platitudes, they are so mild—but it is still worth saying them explicitly.