reframing behavior: the impact of the card act on · pdf file · 2017-02-23(neil...

Reframing Behavior:The Impact of the CARD Act on Cardholder

Repayment Rates

Dennis Campbell (Harvard Business School)Claudine Gartenberg (Harvard Business School)P t T f (H d B i S h l D2D F d NBERPeter Tufano (Harvard Business School, D2D Fund, NBER,and Dean Elect‐University of Oxford Saïd Business School)

© 2011 Campbell, Gartenberg and Tufano

H A R V A R D B U S I N E S S S C H O O L 1

Summary of preliminary findings

1. We examine four main questions: did disclosures change behavior? Of whom? Persistent? Impact on cardholder debt?

2. Preliminary findings – good newsi. Fewer payments at minimum amount, on aggregateii Some consumers reframed to the 36 month suggestionii. Some consumers reframed to the 36 month suggestioniii. Reframing consumers appear to be most credit constrainediv. They are somewhat persistent

3. Less positive newsi. They are only somewhat persistentii C d t i il l h did t d t l thii. Compared to similar people who did not adopt plan, they

seem to be increasing their amount of debtiii. Moving target

© 2011 Campbell, Gartenberg and Tufano2

Research setting and data

• Affinity Plus Federal Credit Union• 132,000 members,• Portfolio of about 30,000 credit cards• 1.3 billion in assets

23 b h i hi Mi• 23 branches within Minnesota

• Going forward: Replication and extension of study with data from aGoing forward: Replication and extension of study with data from a large national bank

C t N t t ll d i t!• Caveat: Not a controlled experiment!


CARD Act: Plain language disclosures, Financial consequences of decisionsq


Empirical evidence on framing effects

• “Framing” affects decision‐making in a variety of contexts• 401k choices (Madrian and Shea 2001)401k choices (Madrian and Shea 2001)• Cooperation in experimental settings (Andreoni 1995)• Numeric judgments (Mussweiler and Strack 2000)

• Lab experiments suggest that reframing effects could be substantial for credit cardholders• “Minimum payment” line is excluded for some participants• For revolvers, when a suggested minimum payment was given the average

repayment was £99 (23% of the balance). When no required minimum payment was stated the repayment averaged £175 (40% of the balancepayment was stated, the repayment averaged £175 (40% of the balance (Neil Stewart, “The Cost of Anchoring on Credit Card Minimum Payments”, Psychological Science 20, 39‐41 (2009) )


Empirical predictions

• With the 36‐month disclosure, cardholders have an additional frame

Research questions:

.3y

q1. What if anything are the

sizes of these first order impacts?

.1.2

Rel

ativ

e Fr

eque

ncy

2. Who changes behavior?• Demographic

0

0 .2 .4 .6 .8 1Payment/Balance

characteristics• Previous payment and

card usage patterns

• We would expect to see more payments at 36 month level

3. Are these results persistent?

h h


• Fewer at minimum level(?) 4. What is the impact on indebtedness?

Overall, trends in balance per capita and median payoff duration did not significantly change…

500

30Number of payment periods and statement balance

2400

2523

00ce

am

ount20

er p

erio

ds

0022

00Ba

lanc

10Num

b

2000

210

0


2

Jan 09 July 09 Jan 10 July 10

Median # periods until payoff Median balance 7

…But there is a noticeable uptake in the fraction of cardholders paying at the 36 month level…

08Fraction of customer population at different payment levels

36 month versus placebo amounts

.06

.0dh

olde

rs.0

4of

tota

l car

d.0

2Fr

actio

n o

0


Payment at 36 month amt Payment at 24 month amt


y yPayment at 12 month amt Payment at 6 month amtPayment at 3 month amt

What do we know about the customers who are following the new disclosure guidelines?

1 Prior implied months to pay off balance1. Prior implied months to pay off balance

2. Credit utilization

3. Credit score

4. Balance


1. Implied months to pay off balance: Cardholders who paid more slowly prior to Act adopted 36 month payment more often

.236-month uptake by pre-Act implied payment durations

.15

dhol

ders

.1

of to

tal c

ard

.05

Frac

tion

0


45+ months 33-44 months


15-32 months 8-14 months<7 months

2. Credit score: Cardholders with lower credit scores prior to Act adopted 36 month payment more often

.15

36-month uptake by pre-Act credit score quintile.1dh

olde

rs5

.of

tota

l car

d.0

5Fr

actio

n 0


Bottom quintile Second quintile


q qThird quintile Fourth quintileTop quintile

3. Credit utilization: Customers with high balance/limit ratios prior to Act adopted 36 month payment more often

.15

36-month uptake by pre-Act credit constraint quintile1dh

olde

rs5

.of

tota

l car

d.0

5Fr

actio

n 0


Bottom quintile, least constrained Fourth quintile


Bottom quintile, least constrained Fourth quintileThird quintile Second quintileTop quintile, most constrained

4. High balance: Customers with higher balances prior to Act adopted 36 month payment more often

.15

36-month uptake by pre-Act statement balance quintile1dh

olde

rs5

.of

tota

l car

d.0

5Fr

actio

n o

0


Highest balance quintile Second quintile


Highest balance quintile Second quintileThird quintile Fourth quintileLowest balance quintile

Using regression analysis, we see that all four of these factors drive uptake of 36 month amount

D d t i blPaid 36 amount 4+ tiDependent variable: 4+ times

Age 0.0085Gender ‐0.1374Duration customer 0.0026

(1) Prior duration of payment amt 0.7559***( ) C dit 0 1568**(2) Credit score ‐0.1568**(3) Credit utilization 0.2056**

(4) Balance 0.2465***(4) alance 0. 465Constant ‐7.2543***Observations 7512


How does adopting this 36‐month guideline impact customer credit behavior and overall indebtedness?

Current balance = P i b l h fPrior balance + new purchases – payments + fees

How did adoption affect:1 New purchases?1. New purchases?2. Payments?3 Overall total balance?3. Overall total balance?


Compared to whom?15

The good news: Customers following 36 month guidelines for 4+ months made fewer new purchases than a matched cohort

Purchase utilization for cohort of customers choosing 36 month 4+ timesversus propensity-scored matched sample

06.0

8t l

imit

.04

.as

es /

cred

it.0

2P

urch

a0


4+ payments at 36 month amount


4 payments at 36 month amountMatched sampleAll others

The bad news: Customers following 36 month guidelines for 4+ months made smaller payments than a matched cohort

5Payment utilization for cohort of customers choosing 36 month 4+ times

versus propensity-scored matched sample

.4.5

bala

nce

2.3

stat

emen

t .1

.P

aym

ents

/ 0

P




4+ payments at 36 month amountMatched sampleAll others

And more bad news: Customers following 36 month guidelines for 4+ months had higher credit balances than a matched cohort

8t

Credit utilization for cohort of customers choosing 36 month 4+ timesversus propensity-scored matched sample

.7.8

l cre

dit l

imit

.5.6

lanc

e / t

ota

.4at

emen

t ba

.3Sta




4 payments at 36 month amountMatched sampleAll others

The new disclosure rules, as written, are a moving target – will they help consumers out of debt?

• If someone makes no new purchases and pays exactly the 36If someone makes no new purchases and pays exactly the 36 month number, they will NOT get out of debt in 36 months:

Th 36 h i l l d h h h ff• The 36 month amount is recalculated each month ‐ the payoff period will always be 36 months away

• With 15.32% APR (non‐reward rate), it will take a member with the average balance ($3900) paying her 36 month amount, followed by “minimum‐minimum” amount, 150 months to getfollowed by minimum minimum amount, 150 months to get out of debt


To summarize: more research, good news—and less positive newsp

1. Research should inform regulation ex ante and ex post• This research is preliminary, to be corroborated on second, p y, ,

larger card issuer’s data• Need to study overall credit consequences for consumer

2 Preliminary findings good news2. Preliminary findings – good news• Fewer payments at minimum amount, on aggregate• Some consumers reframed to the 36 month suggestion• Reframing consumers appear to be most credit constrained• They are somewhat persistent

3 L iti3. Less positive news• They are only somewhat persistent• Compared to people who didn’t adopt the plan, they seem to


p p p p p , ybe increasing their amount of debt

• Moving target phrasing 20

Reframing Behavior:The Impact of the CARD Act on Cardholder

Repayment Rates

Dennis Campbell (Harvard Business School)Claudine Gartenberg (Harvard Business School)P t T f (H d B i S h l D2D F d NBERPeter Tufano (Harvard Business School, D2D Fund, NBER,and Dean Elect‐University of Oxford Saïd Business School)


H A R V A R D B U S I N E S S S C H O O L 21

reframing behavior: the impact of the card act on · pdf file · 2017-02-23(neil...

Documents