Dynamic Lending in the Ghatak/Stiglitz/Weiss adverse selection model, with Comparison to Group Lending

We derive an optimal dynamic lending contract in a simple adverse selection model with limited commitment on the borrower side. An optimal contract involves “penalty” rates after a default, and favorable rates after a success. It also charges higher rates for first-time borrowers than for repeat borrowers, as in “relationship” lending. We compare the efficiency of a group lending contract (of the kind popularized by the micro-credit movement) to the dynamic, individualized contract. We find that the information revelation of both types of unconstrained contracts is similar, but the assumed constraints limit the usefulness of the information revealed in different ways. As a result, group lending can achieve full efficiency under some conditions where dynamic lending fails to. However, when neither type of lending achieves full efficiency, dynamic lending can outperform group lending, by charging high interest rates to borrowers who have defaulted even though this excludes unlucky safe borrowers. We discuss factors push toward one contract form or the other, and also characterize optimal dynamic group contracts. ∗We thank Abhijit Banerjee, Philip Bond, Maitreesh Ghatak, Rocco Machiavello, Alberto Motta, Quan Wen, and participants at NEUDC 2009 and ThReD Conferences in Mexico (2008) and Barcelona (2011). All errors are ours.