Ethiopian Development Research Institute (EDRI) and International Food Policy Research Institute (IFPRI), Seventh International Conference on Ethiopian Economy, June 24, 2010
Insurance Motives to Remin: Evidence from a Matched Sample of Ethiopian Internal Migrants
1. Insurance Motives to Remit: Evidence from a Matched Sample of Ethiopian Internal Migrants Alan de Brauw (IFPRI) Valerie Mueller (IFPRI) Tassew Woldehanna (Addis Ababa University) Ethiopian Economic Association Meetings June 25, 2010
6. Migrant’s Characteristics % % Male 62.0 Amhara 30.4 Ages 19-40 yrs 83.40 Oromo 20.0 Single 57.2 Tigrayan 12.5 <5 years ed 17.9 Orthodox christian 53.0 5-8 years ed. 30.4 Protestant 26.8 9+ years ed. 35.1 Muslim 15.3
11. Conceptual Framework Source: Amuedo-Dorantes and Pozo 2006. Self-insurance (Precautionary Saving) Responsive to the immigrant’s rising exposure to risk in the host community Family-provided Insurance Responsive to the immigrant’s rising exposure to risk in the host community Remittances Sent to finance family’s consumption in home community Sent to accumulate assets in home community Altruism Non-responsive to the immigrant’s rising exposure to risk in the host community
Point 2: Mention that were times when entire households had left and we followed those who had moved for economic reasons according to a PA official. Also mention that we had a 77 percent success rate excluding international migrants. 16 percent of our households with migrants had migrants abroad.
Point 1: Every village has at least one migrant household, but there’s quite a bit of heterogeneity in migration rates which cannot entirely be explained by access to roads or travel time to the nearest city. Point 1: This makes especially sense when you break the rates by location type. The predominant for of migration is rural to rural (64 percent), then rural-urban (31 percent), so it may not necessarily be important to have access to the nearest city if networks matter more. We find in another paper that networks are a strong determinant of migration in this sample.
Demographic composition is consistent with who was targeted in the tracking study; individuals motivated to leave to primarily seek employment elsewhere.
Very few differences. Remitters are wealthier, larger household size, also primarily dependent on farming.
Remitters appear wealthier and older than non-migrants. They tend to have more people in the household (vulnerability?) However, less literate. If you also compare non-remitting migrants to non-migrants, you confirm that there seems to be a positive selection among remitters, and negative selection of migrant non-remitters.
Two Dependent Variables (Remittance amount and Remit or Not). Four models (OLS, Tobit, LPM, Probit). RHS: Individual migrant characteristics X, Source household Characteristics H, Destination household Characteristics D, Migrant shocks, Source household Shocks. Dummies for regional origin, location, time of move. Standard errors clustered by the origin neighborhood.
-Males give more but no evidence that they remit more or less -Occupation matters, pretty much everyone gives more than farmers except for civil servants. -We also control for age and education but there is not significance on these parameters, likely due to the lack of variation in our sample at least on the age variable.
Potential Precautionary Savings story when look at parameters on land per daughter variable. Competition for resources may increase incentives to remit. Destination household characteristics don’t have significant effect on remittance behavior.
Migrant shock consistent with precautionary savings and self-insurance motive. No evidence of trying to insure the family.
Risk aversion: could be picking up shocks matter to the risk averse rather than to remittance behavior.