16
1 Consumer Homing on Payment Cards From Theory to Measuremen Christopher M. Snyd Dartmouth Colle Jonathan Zinm Dartmouth Colle

1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

Embed Size (px)

Citation preview

Page 1: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

1

Consumer Homing on Payment Cards:From Theory to Measurement

Christopher M. SnyderDartmouth College

Jonathan ZinmanDartmouth College

Page 2: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

2

Plan for Today

1. Overview of our project Bridge theory and empirics on card platforms More project than paper at this point!

2. Review of theory models

3. From Theory to Empirics: Measurement Issues

4. Review of existing empirical evidence

5. How our work fits in: New stylized facts Bridge-building: questions and approaches we find interesting Next steps

Page 3: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

3

Payment cards as two-sided markets Payment card is a platform intermediating transactions between merchants and customers.

“No Surcharge Rule”, externalities on both sides mean that equilibrium will depend on prices platform charges to both sides (cardholder fee f, which may be negative, and merchant discount, m)

Homing often crucial modeling issue

• Customers may carry (and use) one or more credit cards

• Merchants may accept one or more credit cards

Policy questions

• Regulation of interchange fee a

• “No Surcharge Rule”

• “Honor All Cards Rule”

Customer

Issuing Bank

Merchant

Acquiring Bank

good sold at price p

paysp + f

paysp - m

pays p - a

Platform

Page 4: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

4

Homing in the theoretical literature

General lessons:

• If consumers multihome (carry multiple cards) and merchants singlehome (accept only one card), networks will try to steer merchants toward their card, putting downward pressure on interchange fee and merchant discount.

• Perhaps more realistically, if merchants multihome (accept all cards), then the extent of consumer singlehoming will determine how hard networks compete to steer consumers toward their card, resulting in low with a high interchange fee and low cardholder fee (bonus?).

Page 5: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

5

Homing in the theoretical literature (con’t)

Rochet and Tirole (2003 JEEA “Platform Competition”)

Rochet and Tirole (2006 RNE “Card Payment Systems”)

• Cardholder multihoming increases merchant surplus• Effect dampened larger fraction of informed purchasers (know merchant acceptance policy before shopping).

Guthrie and Wright (2007 JIE “Competing Payment Systems”)• Large set of equilibria. Homing behavior on and off equilibrium pathselects the equilibrium less favorable for multihoming side.

Rochet and Tirole (2006 mimeo “Honor All Cards”)• Assumes credit and debit separate markets• One credit card• Two debit cards; all debit cardholders multihome• Shows “Honor All Cards” increases social welfare• Extension in Section 5.4 to imperfect substitution. HAC maylower welfare if credit a good enough substitute for debit.

1

2211 d

Ddd singlehoming index

/S

Bo

m

f duopolyequilibriumfee structure

Page 6: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

6

Measurement Issues:Categories of homing in payment cards

“We need to figure out empirically whether the case of single-homing or multi-homing is more descriptive”- (Rochet and Tirole 2006 RNE)

How define multihoming? Multihoming among credit cards

Multihoming among debit cards

Multihoming between credit and debit

Multihoming according to holding vs. use

Whose multihoming matters? Multihoming across whole population or restricted to subpopulation of cardholders?• Who is marginal consumer?

Page 7: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

7

Empirical literature

Rysman (2007 JIE “Payment Card Usage”)Credit card multi-homing prevalent in membership but not in usageProprietary data (Visa PSPS)

Klee (2006 mimeo “…Decade of Change…”)Prevalence of various types of payment multihomingCorrelations with household characteristics

Zinman (2007 mimeo “Debit or Credit?”)Debit and credit substitutes, and increasingly strong ones

Oversights? Please let me know.

Page 8: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

8

Summary of Empirical literature

Accumulating stylized facts

Links to theory of card platforms? Weak so far.

Snyder and Zinman project: Start by adding to stylized facts

o Especially re: multi-homing on credit and debito Comparison between SCF and Visa PSPS

Working on tightening links to theory.o Just questions, no answers, so far

Page 9: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

9

Some Facts to Focus on

Multihoming prevalence in Survey of Consumer Finances (SCF)

Strengths: Complements Visa PSPS (used in Rysman) SCF has non- payment card holders And is publicly available Nationally representative? Can look at trends over relatively long timeframe

Weaknesses: Measures holding and use at level of card type (e.g. general purpose)

o not at network- or card-level No transaction-level data Nothing on debit intensity

Page 10: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

10

Results

Table 1. Multihoming in Holding: Credit Cards

1992 1998 2004

U.S. Pop % MH % U.S. Pop % MH % U.S. Pop % MH %

U. S. population 100 39 100 45 100 48

Has checking or payment card 88 44 92 49 94 51

Has payment card 79 49 86 53 90 54

Has credit/charge card 63 61 68 66 72 67

Uses payment card — — 67 62 81 57

Uses credit/charge card 48 66 54 69 59 70

Page 11: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

11

Results (con’t)

Table 2A. Multihoming in Holding: Between Credit and Debit Cards

1992 1998 2004

U.S. Pop % MH % U.S. Pop % MH % U.S. Pop % MH %

U. S. population 100 43 100 52 100 58

Has checking or payment card 88 48 92 56 94 62

Has payment card 79 54 86 60 90 65

Has credit/charge and ATM card 43 100 52 100 58 100

Uses payment card — — 67 71 81 70

Uses credit/charge card 48 71 54 77 59 81

Uses debit card — — 34 78 59 77

Page 12: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

12

Results (con’t)

Table 2B. Multihoming in Use: Between Credit and Debit Cards

1992 1998 2004

U.S. Pop % MH % U.S. Pop % MH % U.S. Pop % MH %

U. S. population 100 — 100 21 100 37

Has checking or payment card 88 — 92 23 94 39

Has payment card 79 — 86 25 90 41

Has credit/charge and ATM card 43 — 52 43 58 62

Uses payment card — — 67 32 81 45

Uses credit/charge card 48 — 54 39 59 63

Uses debit card — — 34 63 59 62

Page 13: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

13

Results (con’t)

Table 3. Multihoming in Holding: Any Type of Payment Card

1992 1998 2004

U.S. Pop % MH % U.S. Pop % MH % U.S. Pop % MH %

U. S. population 100 53 100 61 100 66

Has checking or payment card 88 60 92 66 94 70

Has payment card 79 67 86 71 90 75

Has credit/charge and ATM card 43 100 52 100 58 100

Uses payment card — — 67 82 81 78

Uses credit/charge card 48 — 54 91 59 92

Uses debit card — — 34 79 59 78

Page 14: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

14

Conclusions from Data so far re: Prevalence

“We need to figure out empirically whether the case of single-homing or multi-homing is more descriptive”-(Rochet and Tirole 2006 RNE)

Multihoming more descriptive. e.g., 75% of payment card holders potential multihomers in 2004

Multihoming description becomes more apt: In recent years (upward trend in most measures) If count credit and debit as well as credit and credit If focused on holding vs. usage

o But multihoming in credit and debit prevalent in usage as wello e.g., 41% of payment card users in 2004 (up from 25% in 1998)

Other findings: In some large segments all consumers are (potential) multi-homers Multihoming in credit card holding SCF << Visa PSPS

Page 15: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

15

From Prevalence to Key Parameters from Theory?

But can we use homing patterns to put magnitudes on theoreticalpatterns of interest? (cross-)elasticity of platform demand in Rochet and Tirole

Possibly.• multihomers in use: most substitutable?• multihomers in holding: moderate substitutability• singlehomers in holding: least substitutable

Some models focus explicitly on membership, not usage (RT 2006 RAND)• realistic if it’s the elasticities that really matter?

Page 16: 1 Consumer Homing on Payment Cards: From Theory to Measurement Christopher M. Snyder Dartmouth College Jonathan Zinman Dartmouth College

16

Next Steps (?)Theory that helps focus on empirically identifying substitutability parameters move beyond description of homing patterns…. to estimating elasticities

Important to capture distinct features of debit platforms? more platforms (regional networks) fixed investment (terminals) consumer substitutability between PIN and signature

New data: Large panel of linked checking and credit card account data

o Transaction characteristicso Fixed and transaction costso Borrowing and spending behavior

Does it matter if homing behavior is driven by neoclassical and/or“behavioral” (e.g., mental accounting) motives? perhaps, if behavioral patterns permit price discrimination and/or dull price sensitivity