The irony is kind of cool. An anthropologist sets out to use the tools of his trade to demystify the the world of economics…
What follows is thus my attempt at an ethnography of the world—imaginative, discursive, but also technical and action oriented—of Paul Collier and, by extension, of the broader genre of popular economics, from Freakonomics to Dead Aid.
…but draws on too smaller sample size, i.e. Paul Collier, and and ends up mistakenly concluding that quantitative social science may be able to tell us a lot about correlations but has little worthwhile to offer on causality.
I think there are definitely limitations to quant work, but in the era of natural experiments, instrumental variables, RCTs and so on, quantitative researchers are better placed than ever to highlight causal links. Paul Collier mightn’t be, but that just goes to show the dangers of relying on a dataset of one.
McGovern’s article is still worth reading though. Despite the shortcomings of his approach he nevertheless renders ‘War, Guns and Votes’ a pile of smoking discredited rubble. And he does an excellent job of de-constructing Collier’s rhetoric.
FWIW I think some of his critique, where it’s right, isn’t really to do with extrapolating causation from correlation at all. Rather it’s more to do with the following. In social science a good R2 (r-squared) might be 0.4. That is the variance in your independent variable explains 40% of the variance in your dependent variable. That still leaves 60% potentially unexplained, and perhaps unexplainable through quantitative techniques. Here, I think, is where qualitative work has the most to offer. A certain geographical trait might leave countries more prone to civil war, for example. And it’s worth identifying that through regression work. But the relationship is unlikely to be so strong that all countries with the trait will end up war torn. Some will defy the trend. And why they do is a really interesting question. Quite possibly there are too many variables, or too many variables which can’t be quantified at least, to analyse this using the tools of statistics. Yet careful process tracing, or something similar may well be able to give us a good idea of what has happened and provide policy relevant insights.
As I said, the review may not be perfect. But it does have its strengths. This quote, for example, ought to be emblazoned on the walls of every economics department, aid agency, and think tank currently doing development work (with the word economists being replaced with “people like us” where appropriate:
“[T]he self-assuredness of economists such as Collier may be a part of the problem and not, as they suppose, a part of the solution. For an outsider, the strangest thing about the field of economics is the fact that although it appears to be wrong much of the time, rather than becoming chastened and introverted, most of its practitioners seem to become bolder, drawing strength from their failures as the mythological Antaeus did by touching the ground.”