Poverty and violence in Africa
A recent article on VoxEU discusses the link between poverty and violence in Africa. It’s a tricky topic because there are good arguments to be made for causation in both directions and the direction of causation matters hugely for aid policy.
Given the endogeneity in the data I was surprised to see that it took until 2004 for the first paper to be published that employed an instrumental variable approach to estimation of the relationship. Reassuringly, a number papers have been published since then. The authors of the Vox article use rainfall as their instrument, since it is highly correlated with poverty in Africa but uncorrelated with violence.
Using these econometric techniques they find that increased poverty leads to a large increase in the chances of violence erupting within a country. As a baseline they find that the average probability of conflict in a year when there is no change in GDP is 20%. From this, already high, baseline they find that a 5% decrease in GDP increases the chances of civil conflict in the next year to 30%. It appears that they use a linear model, so each 1% decrease in GDP leads to a 2% increase in the risk of conflict in the following year.
As a consequence, the authors recommend that African countries which experience drought or poverty in any particular year should be eligible for short term poverty relief. This is intended to reduce the likelihood of conflict ensuing from the drop in incomes.
It is particularly salient advice right now as the world sees commodity prices, upon which African countries often rely for their incomes, come down from their recent highs. Hopefully the international community listens to researchers and moves to help the commodity exporting nations in Africa before they start to feel the bite of the falling prices.
Copyright tvhe.wordpress.com ©
While it is nice to see some attention being paid to these issues, it is also frustrating that effort is being squandered on research into what appear quite mundane questions.
The link between crime (including violent crime) and negative shocks to income is well established for Western countries. There is also evidence that negative shocks to income inequality are particularly destructive. Where income and measures of income inequality ‘bounce’ back following such a shock there is usually a considerable lag before crime drops away correspondingly (actually quite interesting).
Is there any reason why things should be different for commodity based economies in Africa? I suppose that the article cited shows that this is not the case.
However I’m not really sure that this research is ‘salient advice’ that the world has been waiting for to allow it to start helping in Africa during times of (relative) poverty. Poverty in itself is a problem, and if it was considered not important enough by those who control aid to begin with, evidence of the added incidence of violence probably won’t do it either.
At a more basic level though, is this ‘new’ evidence anyway, or just another development paper that, true, does contribute to the literature, but doesn’t practically help deal with the institutional problems that are really the problem?
Well, I think looking at the direction of causality helps to clarify what one of the problems is. Knowing that the marginal cost of reduced income is higher may correspondingly increase aid.
I’m not sure that the link between crime and income in the West sheds much light on the link between income shocks and civil war in entirely different cultures. If such generalisations worked then that would substantially reduce the workload of researchers everywhere 😛