Draft # 4 A Laboratory Evaluation of Partner Selection in Group Lending Models

Houston Shaner Economics Honors Thesis Draft #4 A Laboratory Evaluation of Partner Selection in Group Lending Models This study attempts to simulate the group loan model employed by many Microfinance Institutions (MFIs). Specifically, an experiment using Vanderbilt University undergraduates sought to determine if positive assortative matching (PAM) occurred in the context of a pool of risky and safe borrowers and a type of joint liability group loan. The simulation manipulated two independent variables in a 2x2 factorial design with four conditions. The first independent variable tested a fixed distribution of the group loan between partners versus a participant-selected distribution of the loan. The second independent variable tested learning across iterations of the lending simulation by manipulating the display of simulation results. Direct evidence from the simulation provides only limited support for the PAM outcome, though indirect evidence provides both further support and further qualifications. A 2x2 ANOVA reveals a main effect of the results display factor and an interaction between both factors. Overall, heavily qualified empirical support is found for PAM, and learning effects and other potential confounds seem to be present.