663 / 2024-04-30 18:44:37

Data-Driven Robust Allocation in Cash Transfer Programs

cash transfer program,imperfect targeting,distributionally robust optimization

Cash Transfer Programs (CTPs) are widely used in different regions of the world as an effective approach to fighting extreme poverty. A key issue is how to use the limited beneficiaries' information to establish an allocation policy that balances efficiency and equity under the condition that potential beneficiaries cannot be accurately targeted. In this paper, we consider minimizing a generalized poverty measurement function FGT that can capture multiple poverty measures indexes, and construct a distributionally robust distributional framework based on event-wise ambiguity set. A linear decision rule approximation can be employed to approximate the nonlinear term in the objective function, which does not lead to a loss of optimality.  For the objective with the PGI index, to deal with an infinite number of constraints in the nonlinear model G-SDRO, we use the dual theory to reformulate the constraints and derive a tractable SOCP. For the PSI objective that balances equity and efficiency, to deal with an infinite number of constraints in the S-SDRO model, we design a Column-Constraint Generation(CCG) algorithm that can solve the nonlinear problem efficiently. Using real data from the China Family Panel Studies (CFPS), we show the advantage of the robust allocation policy constructed from the S-SDRO model. Finally, we also report that the S-SDRO model is able to mitigate negative spillovers resulting from the allocation of funds to a greater extent than other allocation policies.