Behavioral Design 101 Jonathan Zinman Dartmouth College and IPA’s U.S. Household Finance Initiative July 26, 2013 Ford Financial Products Innovation Fund Working Group Meeting.
Download ReportTranscript Behavioral Design 101 Jonathan Zinman Dartmouth College and IPA’s U.S. Household Finance Initiative July 26, 2013 Ford Financial Products Innovation Fund Working Group Meeting.
Behavioral Design 101 Jonathan Zinman Dartmouth College and IPA’s U.S. Household Finance Initiative July 26, 2013 Ford Financial Products Innovation Fund Working Group Meeting Outline • Symptoms (focus on household finance) • Diagnoses: what can go wrong (“behavioral factors”) • Treatments • Design principles for treatments Symptoms • Lots of borrowing (overborrowing?) – Credit card debt per household ~= $10,000 – Student loan debt per household ~= $10,000 – Mortgages, auto loans – Share of consumers with subprime credit: 56.4% 1 • Many without savings (undersaving?) – Households with insufficient liquid assets to subsist for three months at the poverty line in absence of income: 43.9% 2 – • Households that reported no saving in the previous year: 48% 3 Many pay premia for financial services (overpaying?) 4 – Assets – Loans – Advice Diagnosis: Struggles in Managing Desires • Factor: Self-Control problems (carpe diem!) – Treatments: commitment, support, on-ramping • Factor: Loss-aversion – Treatments: pre-commitment, change the frame • Factor: Projection bias – Treatment: help people visualize the future Diagnosis: Bandwidth Constraints • Factor: “limited attention” - Failing to consider contingencies - Tuning out/not engaging - Forgetting • Treatments: - Default options - Pre-commitment - Automation - Reminders Diagnosis: Biased Forecasting • Many types: - Factor: Belief in Law of Small Numbers - Factor: Non-belief in Law of Large Numbers - Factor: *Excessive optimism • Treatment: - Information/debiasing? Diagnosis: Getting the math wrong (even when there’s little uncertainty) • Factor: Exponential Growth Bias - People underestimate growth and decline when interest compounds Saving appears less renumerative “Low monthly payments” appear deceptively cheap • Treatments: - Saving: Show future values - Borrowing: APRs and other apples-to-apples comparisons • Factor: Low Numeracy • Treatments: - Simple, intuitive information? Diagnosis: Getting the facts wrong • Factor: Low Financial literacy - Lack of basic knowledge Of key concepts: diversification, inflation Of how products work: e.g., ARMs, term vs. whole life • Treatments: - Default options - Simple/intuitive disclosures Diagnosis: limited learning About finance, about oneself. Why? 1. Limited opportunities • • On high-stakes decisions (mortgage, job, auto financing) Even high-frequency decisions can have uncertain long-run implications - • Credit card use (what’s right debt load for me/my family)? Changing life circumstances creates moving targets 2. Difficult subject matter 3. Sensitive subject matter and motivated ignorance • Factor: Confirmation bias and other asymmetric learning Diagnosis: “Killer Apps” don’t work as well as in other markets Delegation: works pretty well in medicine, auto repair, etc. • But financial advice markets are a mess - Questionable quality: Lots of biased, even fraudulent advice - Limited scope: Who covers the household balance sheet? - Limited coverage: mass market? Competition: often brings prices down, but does not seem sufficient to eliminate yield-spread premia, $39 overdraft fees, etc. Developing Treatments: Design Principles • Simplify • Streamline (on-ramping) • Just-in-time (reach people at decision point) • Meet people where they’re at – Facilitating, nudging much more effective than felt-change • Diagnose and treat • Be humble: we still have a lot to learn – Mixed and limited evidence, especially outside the lab – A premise of behavioral social science is that context matters Thank You! Jonathan Zinman US Household Finance Initiative Innovations for Poverty Action www.poverty-action.org/ushouseholdfinance www.dartmouth.edu/~jzinman/ [email protected] Taxonomy of Behavioral Factors (see DellaVigna JEL) “Cross-cutting” biases/heuristics/ limitations: Anchoring, Limited attention, Innumeracy Biases in preferences • Time-inconsistency • Loss aversion Biases in price perceptions/valuation • Exponential growth bias • Anchoring DECISION DECISION Biases in expectations • Overconfidence • Over-optimism Alternate Taxonomy Biases/limitations in cognition that affect perceptions about how to maximize utility subject to constraints (Affect other key parameters we might model: expectations, prices, transaction costs…) Preferences DECISION