Working Paper: NBER ID: w31819
Authors: Fanghua Li; Chenyang Ji; Moshe Buchinsky
Abstract: In this paper we examine the causal relationship between formal social insurance and individuals’ migration decisions. We exploit a quasi experimental design in rural China, under which county officials were assigned to a group of villages (i.e., treated villages) to serve as village supervisors (VSs) for the local leaders. We show that this led to reduced favoritism in welfare allocation by the local leaders, thereby increasing the efficacy in the formal social insurance in the treated villages. We use detailed geo-referenced administrative household-level data suited for a spatial regression discontinuity design (RDD) to obtain an average treatment effect (LATE) of the improved social insurance on migration. The apparent variation in the implementation of the reform across treated villages and heterogeneous impacts on different family clans make it possible to directly link changes in the efficacy of the insurance to migration choices. We find a large positive migration effect, of about 19%, for the young males and females. In turn, this led to a large boost in the average household’s income in just two years.
Keywords: social insurance; migration; China; quasi-experimental design; poverty alleviation
JEL Codes: J01; O10
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| assignment of village supervisors (VSS) to treated villages (P32) | reduced favoritism in welfare allocation (I38) |
| reduced favoritism in welfare allocation (I38) | increased efficacy of formal social insurance (H55) |
| increased efficacy of formal social insurance (H55) | migration decisions (F22) |
| migration decisions (F22) | increase in household income (D19) |