Working Paper: CEPR ID: DP17157
Authors: Pascaline Dupas; Marcel Fafchamps; Deivy Houeix
Abstract: We investigate a method for eliciting relative poverty rankings that aggregates partial poverty rankings obtained from multiple individuals. We first demonstrate that the method works in principle, then apply it in urban Cˆote d’Ivoire. We find that constructed rankings are often incomplete, not always transitive and sometimes contain cycles. Pairwise rankings reported by respondents and constructed aggregaterankings are poorly correlated with measures of poverty obtained from survey data. Measuring relative poverty through peer rankings appears difficult in urban and periurban settings.
Keywords: relative poverty; peer rankings; network analysis; West Africa
JEL Codes: D31; O12
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Peer rankings (A14) | Constructed rankings (Y10) |
| Constructed rankings (Y10) | Observational measures of poverty (PMT and PPI) (I32) |
| Peer rankings (A14) | Predictive power of consumption differences (E21) |
| Constructed rankings (Y10) | Predictive power of consumption differences (E21) |
| Insufficient information about households (R20) | Poor correlation with poverty measures (I32) |
| Peer rankings (A14) | Noise in information about welfare (I38) |