Credit Cards for the Poor Ronald Mann Discussion by Adair Morse University of Chicago Vissing-Jorgensen, 2007 Hypothesis • “…the products offered and taken up by LMI households… differ significantly from those used by middle-class households.” – Compare characteristics of LMI households to affluent ones for differences that support different [uses] – The paper contrasts profiles of use and users of cc by income quintile • Hard to get to endogenous provision of features by cc CC Companies Bank Outstandings ( $ millions) $134,700 # Active Accounts (1,000s) Citigroup 115,850 47,880 MBNA (B of A) 82,118 21,119 Bank America 61,093 18,773 Capital One 53,024 24,429 JP Morgan Chase 42,996 (transactions) (gimmicks + mix) (“subprime”) - Akers, Golter, Lamm, Solt 2005 FDIC Banking Review Frame Question: Supply Two types of Credit Cards 1. Transactions revenues • Still the lion share of income 2. Fee-based revenues • • • Would not offer card if depend on interest rate & transaction fee Relieves some credit constraints Gross Souleles (2002) $1000 added liquidity => $130 spending – Penalty fees (Hammer Consulting Survey 2004) • • • $13 billion 39% of cc income Increasing (Furletti Ody 2006) – Massoud Saunders Scholnick 2006 • • Interest rates and fees are substitutes (model) Cards differentiate based on demand preferences & default probabilities Frame question: Demand Why do LMI use credit card debt? 1. Consumption smoothing / income shocks 2. High (inconsistent?) discount rate: tilting life cycle consumption profile 3. Being manipulated (targeting by cc’s) • • (Angie Littwin – poor borrowers have few choices) – Does this belong in set? – Or do cc’s offer products that { profit maximize with respect to, market to, understand, all of these } groups (1), (2) or both I don’t think we even know (1) from (2), much less (3) Not clear (1) and (2) need/use same cc features e.g., Relating to profiles • Age: more borrowing by young (life cycle) • Education: – Greater slope in income = more borrowing early in career (life cycle) – Sophistication = less (consumption; targeting) or more borrowing (life cycle) – More educ = more availability (targeting) Empirical Thoughts • Agarwal, Chomsisengphat, Liu, Souleles 2005 • Consumer mistakes in not taking up cc offers is a small-dollar effect in most cases • Data: ideally a panel of individuals – (Ravina 2005 individual credit card balances) – Understand their credit use “type” – Observe patterns of spending and paying interest/fees – Policy: See if credit cards extract undue rents from LMI or if fee structure is optimal way around cc “fear” of offering high interest rate products SCF ideas • Characterizing income smoothers – Normal income vs current income • Group all those with x% lower current than normal income – Vulnerability • % of income that is used up in basic expenses – Repaying balances • Sometimes vs (always or never) – Shop around, use of CC for vacation, jewelry • Targeted & Temptation consumption types – Is it important to distinguish these? Yes if there is a distinction – Been turned down? – Use geography? Where do cc send most mailings? Things to know, Differencings – Relate (i) variance of normal to current income and (ii) vulnerability % to (a) interest rate paid & (b) credit limit – What is the size & rate for the balance for smoothers • Relative to same type & income individuals who sometimes pay off balances • Relative to same income individuals of temptation targeting type – Test: estimate a propensity to default. Do a diff-in-diff for matched group on the cost of debt. – Who are cc targeting? • Are lower quintile borrowers purely temptation/targeting types? – Evidence of purely fee play? – Want to know education levels, race, age, etc