This paper uses micro data from the 2000/01 Enquete de Condition de Vie des Menages in
Rwanda to study the livelihood of agricultural subsistence households and a resettlement policy in
post-conflict rural Rwanda. Arguably exogenous conflict-induced migration is employed to account
for the incidence of conflict at the household level. A post-conflict resettlement policy, the villagization
or imidugudu policy, is also assessed in the pilot years of its implementation. We also exploit this
programme as a source of exogenous variations in patterns of resettlement to measure differentials in
skill spill-overs across returnees in policy and non-policy areas. Controlling for (unobserved) prefecture
heterogeneity and input consumption, we find that the (unobserved) heterogeneity associated
to conflict-induced migration has a negative, large, and significant effect on agricultural output in all
areas. Our results also suggest that, although there is evidence that returnees are more motivated in
achieving better livelihoods than stayers. We also find some support to the notion that displacement
decreased returnees’ level of agricultural know-how, and that villagization reduced skill spill-over
flows from stayers to returnees. However, those results are not preserved when instrumental-variable
estimation is used to correct for endogeneity issues in the input-output specification. Overall, we find
that returnees should be the target of reconstruction and welfare-improving intervention programmes
in post-war Rwanda, and that extending the programme to the whole of Rwanda is unlikely to prove
production-enhancing.