Paris - 26 Oct 2007Bounie, François and Kiser 1 Debit Card Float and Consumption Patterns David Bounie Abel François Télécom Paris Elizabeth K. Kiser Federal.

Paris - 26 Oct 2007 Bounie, François and Kiser1

Debit Card Float and Consumption Patterns

David BounieAbel FrançoisTélécom Paris

Elizabeth K. Kiser Federal Reserve Board

The views expressed here are thoseof the authors and not necessarily ofthe Federal Reserve Board or theFederal Reserve System.

Paris - 26 Oct 2007 Bounie, François and Kiser2

Motivation

Rapid growth in card payment systems, especially debit cards– Displacing cash, checks

Float may affect ability to smooth consumption, if liquidity constrained– Payment cards as source of short-term credit

How do payment methods affect consumption behavior?

Paris - 26 Oct 2007 Bounie, François and Kiser3

Related Literature

1) Payment choice and behavioral motives– Ariely and Silva (2002)– Zinman (2006)– Fusaro (2007)– Soman (2003)

Paris - 26 Oct 2007 Bounie, François and Kiser4

Related Literature, cont.

2) Consumption behavior and the permanent-income hypothesis

– Browning and Lusardi (1996)– Gross and Souleles (2002) – Stephens (2003)– Soman (2003)

Paris - 26 Oct 2007 Bounie, François and Kiser5

Data

French consumer survey, March - May 2005– Nationally representative– Sponsor: Groupement des Cartes Bancaires

Two modules1. Interview to collect demographic/financial

characteristics: 1447 individuals

2. 8-day purchase diary: Completed by 1392 of these individuals We restrict our analysis to these individuals

Paris - 26 Oct 2007 Bounie, François and Kiser6

Socioeconomic/Financial Survey

Standard socioeconomic variables Week in which salary/pay is received

– 1st week of month, 2nd week of month, etc.

Payment card holdings Card type: deferred vs. immediate debit

Paris - 26 Oct 2007 Bounie, François and Kiser7

Socioeconomic/Financial Survey, cont.

Deferred debit:– Debit card purchase values remain in the account

until the first day of the following month– Like a charge card

24% of “primary” debit cards are deferred debit

Paris - 26 Oct 2007 Bounie, François and Kiser8

Purchase Diaries

Covers period of 8 days per individual– Dates vary within the sample

Information on– Value of purchase– Type of good/service– Store type– Payment method used

Paris - 26 Oct 2007 Bounie, François and Kiser9

Research Questions

Deferred debit cards as a source of short-term credit:– Do consumption patterns differ between

consumers with immediate vs. deferred cards?– If so, is this difference due to the availability of

short-term credit? Or to socioeconomic factors?– What do the differences, if any, mean for

consumption behavior?

Paris - 26 Oct 2007 Bounie, François and Kiser10

Empirical Approach

Retail purchases only (“discretionary”) Descriptive charts on purchase behavior Tests for simultaneity of purchase decision

and debit card type Regressions of purchase behavior as a

function of card type and socioeconomic factors

Paris - 26 Oct 2007 Bounie, François and Kiser11

Debit Card Holdings and Use

99% of individuals have a bank account Of bank account holders, 81% hold at least

one debit card Of debit card holders,

– 76% conduct debit transactions– 24% hold deferred debit cards

Paris - 26 Oct 2007 Bounie, François and Kiser12

Consumption Patterns:Immediate vs. Deferred Debit

 Average value of

daily purchases (€)

Average number of daily purchases

No card 27.2 1.2

Immediate debit

28.7 1.3

Deferred debit

35.7 1.3

Paris - 26 Oct 2007 Bounie, François and Kiser13

Purchase Patterns

We separately observe – Weeks to end of month– Weeks to next paycheck

Different predictions for consumption

Paris - 26 Oct 2007 Bounie, François and Kiser14

Predictions:Weeks to End of Month

Individual is liquidity constrained

Individual is not liquidity constrained

Have deferred debit:

Deferral is adequate to smooth cons.

No response No response/NA

Deferral is inadequate to smooth cons.

↓ consump. before/at end of month

No response/NA

Have immediate debit No response No response

Paris - 26 Oct 2007 Bounie, François and Kiser15

Predictions:Weeks to End of Month

Individual is liquidity constrained

Individual is not liquidity constrained

Have deferred debit:

Deferral is adequate to smooth cons.

No response No response/NA

Deferral is inadequate to smooth cons.

↓ consump. before/at end of month

No response/NA

Have immediate debit No response No response

Note: Cannot observe…

Paris - 26 Oct 2007 Bounie, François and Kiser16

Inference:Weeks to End of Month

If we observe no response to “weeks to end of month” for those with deferred debit, we know either

– People are liquidity constrained but deferral amount is adequate to smooth consumption, or

– People are not liquidity constrained If we observe lower consumption around month end

for those with deferred debit, we infer people are liquidity constrained and deferral is inadequate to smooth consumption

We expect to find no response to “weeks to end of month” for individuals with immediate debit card

Paris - 26 Oct 2007 Bounie, François and Kiser17

Predictions:Weeks to Next Payday

Individual is liquidity constrained

Individual is not liquidity constrained

Have deferred debit:

Deferral is adequate to smooth cons.

No response No response/NA

Deferral is inadequate to smooth cons.

↓ consump. before payday No response/NA

Have immediate debit↓ consump. before payday No response

Paris - 26 Oct 2007 Bounie, François and Kiser18

Predictions:Weeks to Next Payday

Individual is liquidity constrained

Individual is not liquidity constrained

Have deferred debit:

Deferral is adequate to smooth cons.

No response No response/NA

Deferral is inadequate to smooth cons.

↓ consump. before payday No response/NA

Have immediate debit↓ consump. before payday No response

Again: Cannot observe…

Paris - 26 Oct 2007 Bounie, François and Kiser19

Inference:Weeks to Next Payday

If we observe no response to “weeks to payday” for those with deferred debit, we know either

– People are liquidity constrained but deferral amount is adequate to smooth consumption, or

– People are not liquidity constrained If we observe lower consumption before next payday

for those with deferred debit, we infer people are liquidity constrained and deferral is inadequate to smooth consumption

If we observe lower consumption before next payday for those with immediate debit, we infer people are liquidity constrained

Paris - 26 Oct 2007 Bounie, François and Kiser20

Unconditional Consumption Patterns:Weeks to Month End, Purchase Count

Mean Number of Daily Purchases: Weeks to End of Month

1.1

1.3

1.2

1.3 1.31.3

1.31.3

1.3

1.2

1.4

1.2

0.5

0.75

1

1.25

1.5

3 2 1 0

Weeks to end of month

Mean

nu

mb

er

per

day

No card

Immediate

Deferred

Paris - 26 Oct 2007 Bounie, François and Kiser21

Unconditional Consumption Patterns:Weeks to Payday, Purchase Count

Mean Number of Daily Purchases: Weeks to Next Payday

1.1

1.2

1.2

1.31.3 1.3

1.2

1.31.3 1.3 1.31.3

1

1.05

1.1

1.15

1.2

1.25

1.3

1.35

1.4

3 2 1 0

Weeks to next payday

Mean

nu

mb

er

per

day

No card

Immediate

Deferred

Paris - 26 Oct 2007 Bounie, François and Kiser22

Unconditional Consumption Patterns:Purchase Count Findings

No substantial differences across groups in weeks to end of month

Non-cardholders show modest (18% max) increases over weeks to payday

Look next at purchase values…

Paris - 26 Oct 2007 Bounie, François and Kiser23

Unconditional Consumption Patterns:Weeks to Month End, Purchase Value

Mean Value of Daily Purchases: Weeks to End of Month

29.8

32.5

26.2

22.6

30.8

26.5

28.7 29.1

35.837.1

38.7

30.3

15

20

25

30

35

40

3 2 1 0

Weeks to end of month

Mean

valu

e p

er

day (

€)

No card

Immediate

Deferred

Paris - 26 Oct 2007 Bounie, François and Kiser24

Unconditional Consumption Patterns:Weeks to Payday, Purchase Value

Mean Value of Daily Purchases: Weeks to Next Payday

26.5

21.1

26.7

34.7

29.0

25.9 26.6

33.3

37.9 37.7

32.931.0

15

20

25

30

35

40

3 2 1 0

Weeks to next payday

Mean

valu

e p

er

day (

€)

No card

Immediate

Deferred

Paris - 26 Oct 2007 Bounie, François and Kiser25

Unconditional Consumption Patterns:Purchase Value Findings

Across weeks to end of month:– Non-cardholders – decreasing over month– Immediate cardholders – flat– Deferred cardholders – increasing over month

Across weeks to payday:– Non-cardholders – lower before payday– Immediate cardholders – lower before payday– Deferred cardholders – higher before payday

Paris - 26 Oct 2007 Bounie, François and Kiser26

Ceteris Paribus?

Do patterns remain, holding all else equal?

Paris - 26 Oct 2007 Bounie, François and Kiser27

Regressions

Predict value of purchases per day Condition on socioeconomic variables Allow response to weeks to end of month

and weeks to payday to vary with whether individual holds deferred debit

Paris - 26 Oct 2007 Bounie, François and Kiser28

Testing for Consumption Effects

First, test for simultaneity Do unobserved factors that affect

consumption also affect whether the consumer holds a deferred debit card?– Durbin-Wu-Hausman test– We cannot reject the hypothesis that the

prediction errors are mean-independent

Paris - 26 Oct 2007 Bounie, François and Kiser29

Regression to Test for Consumption Effects

Level of observation is week segment– Each individual appears in two segments

Include only individuals with debit cards Regressed mean value of purchases per day on

– Income, household size, education, age, gender, city size, employment status, housing type

– Dummies for days of week in segment– Weeks to end of month, weeks to next payday variables

Paris - 26 Oct 2007 Bounie, François and Kiser30

Interactions: Deferred Debit and Time Variables

Interacted “weeks to end of month” and “weeks to next paycheck” with indicator for having deferred debit– Time effect may vary by subgroup – Allows us to test our predictions

Paris - 26 Oct 2007 Bounie, François and Kiser31

Results

“Weeks to end of month” not significantly different between deferred and immediate debit cardholders

– If liquidity constraints are present for deferred debit holders, deferred debit amount is adequate to smooth consumption

Even after controlling for socioeconomic variables, “weeks to payday” differs significantly between types of cardholders

– If liquidity constraints are present for deferred debit holders, deferral amount is adequate to smooth consumption

– Liquidity constraints are present for immediate debit holders

Paris - 26 Oct 2007 Bounie, François and Kiser32

Regression Predictions:Weeks to End of Month

Predicted Value of Daily Purchases: Weeks to End of Month

30

2629 30

3236

33

39

0

5

10

15

20

25

30

35

40

45

3 2 1 0

Weeks to end of month

Mean

valu

e p

er

day (

€)

immediate deferred

Paris - 26 Oct 2007 Bounie, François and Kiser33

Regression Predictions:Weeks to Next Payday

Predicted Value of Daily Purchases: Weeks to Next Payday

3027 26

33

32

39 38

32

0

5

10

15

20

25

30

35

40

45

3 2 1 0

Weeks to next payday

Mean

valu

e p

er

day (

€)

immediate deferred

Paris - 26 Oct 2007 Bounie, François and Kiser34

Conclusions

Results suggest that deferred debit may allow cardholders to smooth consumption over short periods

Charge card float may have real impact on consumption patterns

Similar products in economies where deferral is unavailable could improve consumer welfare

– U.S.: Many low- and moderate-income households use “payday loans” with extremely high finance charges

– Low-value, short-term loans Caveat – These may not be the customers to whom

banks would want to offer such a card (risk)

Paris - 26 Oct 2007 Bounie, François and Kiser35

Next Steps

Try to proxy for liquidity constraints– Spending relative to income

See whether spending patterns differ for durables vs. nondurables– Expect large, lumpy expenditures to be more

strongly affected by liquidity constraints