Working Paper: NBER ID: w29311
Authors: Gharad T. Bryan; Dean Karlan; Adam Osman
Abstract: We experimentally study the impact of relatively large enterprise loans in Egypt. Larger loans generate small average impacts, but machine learning using psychometric data reveals that ”top-performers” (those with the highest predicted treatment effects) substantially increase profits, while profits drop for poor-performers. The large differences imply that lender credit allocation decisions matter for aggregate income, yet we find that existing practice leads to substantial misallocation. We argue that some entrepreneurs are over-optimistic and squander the opportunities presented by larger loans by taking on too much risk, and show the promise of allocations based on entrepreneurial type relative to firm characteristics.
Keywords: entrepreneurial lending; credit allocation; psychometric data; machine learning; Egypt
JEL Codes: D22; D24; L26; M21; O12; O16
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Top performers (Y10) | Default risk perception (G33) |
| Default risk perception (G33) | Misallocation of loans (H81) |
| Altered loan officer incentives (G21) | Improved capital allocation (G31) |
| Poor performers (D29) | Riskier decisions (D91) |
| Riskier decisions (D91) | Diminished expected rewards (D80) |
| Larger loans (G51) | Entrepreneurial success (L26) |
| Larger loans (G51) | Capital efficiency (G31) |
| Larger loans (G51) | Aggregate productivity (E23) |
| Larger loans (G51) | Profit increase for top performers (L25) |
| Larger loans (G51) | Profit reduction for poor performers (D22) |
| Larger loans (G51) | Wage bill increase for top performers (J33) |
| Larger loans (G51) | Productivity increase for top performers (D29) |
| Larger loans (G51) | Household expenditures increase for top performers (D12) |