The legacy of conflict: aggregate evidence from Sierra Leone
This paper studies the general equilibrium impact of civil war in Sierra Leone. I first use an instrumental variable (IV) strategy and geographic conflict variation to estimate reduced-form effects. I show that civil war leads to affected areas having a higher share of workers in agriculture, fewer educated workers, and lower worker income.
In order to explicitly take into account general equilibrium effects such as selective migration in response to the war, I then develop an economic geography model. The model sheds light on different mechanisms through which conflict affects aggregate income: changes in education, firm productivity, and individual productivity have both direct effects on income and indirect effects by changing the allocation of labour across sectors and locations. Changes in amenities also affect the spatial allocation of labour.
Next, while education outcomes are observed, I structurally estimate all unobserved parameters. In particular, I leverage the structure of the model along with observed income information and migration flows to identify firm productivities, amenities, and average individual productivities.
Finally, I use the model to perform counterfactual simulations. Aggregate income in Sierra Leone is estimated to be 32 per cent lower today as a result of the civil war. Importantly, firm productivity losses can explain most of the decrease, while human capital reductions alone can only account for a small part of the effect. Selective migration in response to the war also seems to play an important role and implies that local reduced-form effects are misleading when trying to estimate aggregate effects.