Working Paper: NBER ID: w19069
Authors: Victor Stango; Jonathan Zinman
Abstract: We document cross-individual variation in U.S. credit card borrowing costs (APRs) that is large enough to explain substantial differences in household saving rates. Borrower default risk and card characteristics explain roughly 40% of APRs. The remaining dispersion exists because a borrower can receive offers and hold cards with wide-ranging APRs, as different issuers price the same observable risk metrics quite differently. Borrower debt (mis)allocation across cards explains little dispersion. But self-reported borrower search/shopping (along with instruments for shopping implied by Fair Lending law) can explain APR differences comparable to moving someone from the worst credit score decile to the best.
Keywords: No keywords provided
JEL Codes: D14; D22; D4; D83; G21; G23
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| Observable borrower default risk (G33) | APR variation (F31) |
| Product differentiation (L15) | APR variation (F31) |
| Substantial dispersion in borrowing costs (G19) | Household savings rates (D14) |
| Borrowing costs (G32) | Household savings rate (D14) |
| Consumer search intensity (D12) | Borrowing costs (G32) |
| Consumer shopping behavior (D19) | Borrowing costs (G32) |