Framing, Choice, and Household Finance: Results, Implications, and Related Work Jon Zinman Dartmouth College October 17,
Links to Papers Will mention several papers, all available (or coming soon) at
Marketing Experiment in a Consumer Loan Market “Pricing Psychology: A Field Experiment in the Consumer Credit Market” (with Bertrand-Karlan- Mullainathan-Shafir) Use randomized marketing “treatments”, adminstered via direct mail and phone, to: –Learn about consumer choice –Learn about how (social) marketing, can change financial decisions Simultaneously randomize price (interest rates)
Setup Consumer lender in South Africa High-interest loans to (working) poor Neoclassical model: –Loan terms drive decision: “economic content” –Describing offer in different ways, holding economic content constant, should not impact loan demand Test this: –Designed 10 treatments motivated by “behavioral” findings from lab experiments Frames: e.g., gain vs. loss Cues: e.g., “priming” phone call
Results 5 of the treatments affected loan demand to some extent –Effects very large relative to price 5 (or more) did not So results challenge both: –neoclassical economics (preference instability, bounded cognition important) –the value of modeling psychological “realism” When context matters, it’s hard to predict behavior
Potential Implications for Saving and Financial Education Content: –Strongest result is on “information overload” Applications: if frames and cues matter, “treatments” can be applied via other channels as well: –Financial ed –Sales pitches –Information packets –Product design
Extension: Marketing & Selling Savings Products Currently working with: –Bank in Peru –Major U.S. financial services company Designing frames and cues to test, e.g.: –Our borrowing study finds that showing a male client a picture of an attractive woman increases takeup –If this is an “arousal” effect, can we cue “sober” saving motives (photo of family, house, sickbed, etc.)?
Other Zinman Work Related to Financial Education How (well) do households make financial decisions? –High-frequency: Pretty well Consistent with neoclassical model (at least based on available evidence) See “Debit or Credit” and “A Behavioral Mirage”; also Miravete 2003 –Low-frequency: More problematic, “behavioral” Bounded cognition result: –Most households underestimate the true cost/yield of an interest rate: “Present-biased computation” –Bias is correlated with: using rule-of-thumb (focus on payments), higher borrowing costs, more borrowing, less saving –See “Fuzzy Math and Red Ink: Present-Biased Computation and Household Finance” (with Victor Stango)