Studies 'financial mistakes' in credit card usage high-cos

1
Comments on the PowerPoint Presentation ofWho
Makes Credit Card Mistakes?

• Katherine Samolyk
• FDIC
• Presented at the FDIC 2006 Fall Workshop,
  Washington D.C.
• These comments reflect my views not those of the
  FDIC or its Board

2
Overview of the Research Presented

• Studies financial mistakes in credit card
  usagehigh-cost borrowing while holding
  low-interest-bearing deposits.
• Looks at of who makes these credit card
  mistakes? (Income, education, unemployed,
  immigrant status, credit score)
• Uses proprietary data for a single depository
  institution on credit card balances and deposit
  account balances to examine how demographic
  characteristics of bank customers are related to
  borrowing high-lending low (BHLL) behavior
  associated with simultaneously having interest
  bearing credit card debt and low yielding bank
  deposits).

3
Conceptual Backdrop

• What is the explicit cost of holding
  low-interest-bearing but liquid bank deposits
  while you are financing expenditures using
  relatively expensive credit card debt?
• Zinman (2006)
• Unadjusted Wedge min Credit Card Debt, Demand
  Deposits
• Unadjusted COSTmax0, Unadjusted
  Wedge(r_card-r_dep)

4
Conceptual Backdrop

• Discuss why people might make these mistakes
  --Distinction between nonstandard behavior vs
  nonstandard preferences
• Low Education causing mistakes is example of
  nonstandard behaviordont know better
• Different from nonstandard preferences (i.e.
  behavioral-psych explanations)
• Very important question in consumer finance
  because the answer has very different policy
  implications---
• Mistakes because of nonstandard preferences --
  policies regarding the provision of financial
  services that elicit optimal behavior (behavioral
  economics)
• Mistakes because of not knowing something can be
  corrected by financial education.

5
Conceptual Backdrop

• But is the behavior really a mistake?
• Portfolio management problem --asset and debt
  management under liquidity constraints
• Transactions managementincoming and outgoing
  payments
• BHLL may not always represent either a mistake or
  weird preferences if there are costs and
  frictions in household portfolio and payments
  decisions
• Rational - Precautionary Balancesto avoid costs
• Zinman , 2006
• Telyukova and Wright, 2006

6
What are the costs of not having deposit
balances?

• Various deposit account fees associated with not
  having some minimum balance
• Costs of not making payments NSF fees, late
  payment fees returned check fees other
  pecuniary and nonpecuniary costs of nonpayment
• Other forms of short-term borrowingpayday loans
  fee- based overdraft protection
• the costs / benefits depends on
• Customer expenditure/payment patterns and the
  means of payment used to make transactions
• Customer wealth the composition of wealth
  (liquidity) investment opportunities other
  sources of credit (home equity)

7
Data and Methodology

• Match (1) Credit Card Monthly Statement Data with
  (2) Deposit Account Monthly Statement Data
  100,000 customers--19 months of data (not
  independent observations for a given customer)
• Use proxies for customer characteristics
  Canadian census data on Income Education
  Unemployment House Ownership Immigrant Status
• Match each consumers postal code (50 or so
  households) with appropriate dissemination area
  in the Canadian Census Data (Average from 200
  Households)
• Ask whether/how economic and demographic
  variables in the DA where customers live are
  related to the simultaneous use of credit card
  debt and deposit holdings

8
Data and Methodology

• Dependent variables
• Proportion of Months that a consumer BHLLs ---20
  possible proportions (mostly 0s or 1s)
• Proportion of Months that each consumer BHLLs
  measured net of Precautionary Balancese.g.
  holds balances in excess of 1 std deviation of
  balances
• 20 possible outcomes (mostly 0s or 1s)
• How is this measuredfluctuations in each
  customers daily balances or in end-of-month
  statement balance?
• Consumer is Delinquent or Over limit on Credit
  Card but still has deposits that could pay
  outstanding Credit Card debt (two outcomes 0 or
  1)

9
Results

• Less BHLL BEHAVIOR
• University Grad (All 3 models) what
  is the omitted group? fewer less than
  high school?
• Own House (All 3 models)
  fewer renters
• Immigrant-Dev Countries (2 models) fewer
  nonimmigrants ?
• Higher Income (1 model)
  
• Higher credit score (all three models)
• Greater BHLL BEHAVIOR
• High School Only (All 3 models) what is
  the omitted group?
• Some Post Sec (All 3 models) fewer
  less than high school?
• Renters (All 3 models)
  fewer homeowners
• Nonimmigrants (2 models) fewer
  DC immigrants
• Lower income (1 model)
  
• Lower credit score (all three models)

10
Issues/Questions

• Questions about how mistakes are quantified
• Can you observe all of the credit card debt owed
  by each depositor (cards from other issuers?)
• How is debt measured from monthly statement ?
• Do you exclude customers that dont have a credit
  card with bank from the empirical tests since
  they cant BHLL?
• Are you including saving accounts and checking
  accounts?
• How are deposits measured from the monthly
  statementminimum balance, average balance, month
  end balance?
• Are deposit balances and credit card debt
  measured on the same day?

11
Suggestions

• Could do more to try to quantify the costs of
  BHLL behavior
• Dependent variables not capturing magnitudes and
  attendant costs of BHLL behavior
• If balance are so low not worth using them to
  pay off debt
• Can you estimate the costs from your data?
• Would make dependent variable more continuous
• Interesting to try to do more to understand BHLL
  behavior
• Are there seasonal patterns (Christmas time)
• Interesting information about customer deposit
  and transaction/expenditure patterns in monthly
  statements
• Does your data have details on the timing and
  composition of deposits and expenditures ?
• Could be used to further understand this type of
  consumer financing behavior---e.g. are the
  mistakes satisfying minimum balance
  requirements?
• Look forward to reading the paper!