Insights from Behavioral Economics to Small Business

Insights from Behavioral Economics
to Small Business Banking
Antoinette Schoar
Michael Koerner '49 Professor of Entrepreneurial Finance
MIT Sloan School of Management
CEPR-EBRD: “Understanding Bank in Emerging Markets”
September 5-6, 2013
What is Behavioral Economics?
What is Behavioral Economics?
The Marshmallow Experiment
The Marshmallow Experiment
Giving in …
Traditional – Mapping Intention to Action
Decision
Actions
Outcome
Yes
A
B
Yes
No
C
D
No
Behavioral – Obstacles to the Translation
Decision
Actions
People misjudge how the
will behave
Outcome
Ex post need to justify actions
(or just confused)
Yes
A
Yes
No
B
No
Why is this Relevant for Finance?
• Small changes in how financial products are
set up significantly affect customer outcomes
– Credit risk is not a fixed type but endogenous to
the state of the credit market
– The sophistication of financial institutions causally
affects the structure of credit risk in the economy
• Banks have embraced retail financial tools for
SME banking to reduce transaction costs
– But opportunities to use psychological insights to
improve repayment behavior
Getting to the top of mind: testing the
effectiveness of reminders
• Banks in Bolivia, Peru,
the Philippine and
Uganda
• Customers with loans
of savings accounts
• Monthly text message
reminders
– Estimate size of effect
– Generic text message
reminder or highlight a
particular goal
Source: Karlan, McConnell, Mullainathan, and Zinman (2010) and Cadenas and Schoar (2011).
Text message have similar effects as 20%
reduction in interest rate
18%
% reduction in late payment
16%
* Heterogeneity who responds to
each treatment: the young are
more like to respond to SMS
14%
12%
10%
8%
6%
4%
2%
0%
No reminder
Generic
reminder
SMS
Reminder
Cash incentives
for
Goal-specific
reminder
on time payments
Text message reminders increase savings
Percentage increase in savings
18%
16%
14%
12%
10%
8%
6%
4%
2%
0%
No reminder
Generic reminder
Goal-specific reminder
Information Avoidance: Foreclosure Resolution
• Distress homeowners do not return
calls and mailing about mortgage
reduction offers
• >85% of borrowers in distress do
not call back
• Lack of information seeking to
avoid dealing with pain
• Type of outreach matters
• Highly distressed borrowers
respond only to “soft touch”,
constructive messaging
• Casually distressed borrowers
respond to harsh messages
(strategic default)
1
3
Mullainathan, Schoar and Shafir (2011)
Take Away
• People suffer from inattentiveness even toward
goals that are important to them
– Not following through on goals results in sub-optimal
outcomes
• Reminders can help people stick to a plan
– Text message reminders are very cost-effective
– Content and type of message matters
Loyalty as Alternative Collateral
• Idea: Many SMEs do not have collateral to get a bank
loan even if they have cash flow to pay it. Test a
collateral free loan product and use relationships as
alternative collateral?
Different Hypotheses for Effectiveness
of Outreach
• Challenge: Test if relation building can ensure
repayment and increase client loyalty
• Can the services that bank provides substitute for the
lack of financial infrastructure in the market?
• Depends on the reasons behind defaults
• Strategic default: Reduction of moral hazard?
• Loyalty: Build personal relationship between loan
officer and clients?
• Behavioral factors: SMEs have poor planning skills
Set Up
• Only uncollateralized lending facility to SMEs in India
• Credit assessment is based on a score card model
• Centralized risk team makes credit decisions based
on observable information, e.g. tax filing, bank
statements
•
Loans are structured as a one year overdraft facilities
• Payment modality like a credit card: monthly interest
payments and 5% of balance has to be paid
• Loan size between $10k-$50K
• Penalty interest rate starts after 30 days late
Implementation
• Hire six relationship managers to follow up with clients in
treatment groups A and B
• Very clear separation from credit assessment team
• Convey to borrowers that relationship managers will
not be involved in loan renewal
•
Relationship managers have set scripts to call clients
• Check in every two weeks independent of loan
status. Solve problems with accounts, remind
customers of delays in payment if necessary etc
• No cross selling (!)
Experiment Set-up
• Group A: No monitoring treatment
– Control Group
• Group B: Reminder treatment
– Send SMS with interest and principal due every month. Follow
up with phone call if clients have outstanding balances
• Group C: Medium touch treatment
– Random loan officers follow up with clients regularly to solve
problems, understand the nature of the business
• Group D: Personal touch treatment
– Assign individual loan officer to create “ongoing relationship”
with client. Treatment as in Group C.
Default Rates Differ Drastically by Treatment
Delinquent Customers per Treatment Group
20%
19%
18%
16 %
15 %
14 %
14%
12%
12 %
11 %
10 %
10%
10 %
9%
9%
8%
8%
6%
6%
6%
4%
3%
2%
0%
0%
SBL, Loans 5+ Lakh, CPA Cities
[N=774]
High
SBL Power, Loans
- Touch (Relationship)
5 + Lakh, CPA
Cities
[N= 311 ]
Medium
SBL, Loans
- Touch (Monitoring)
5 + Lakh, Non
Cities
[N= 158 ]
Low
-
- CPA
- Touch (Reminders)
SBL Power, Loans <
5 Lakh, Non
CPA Cities
[N= 505 ]
Control
-
Differences in Account Behavior
Results I
• Significant reduction in late payments for
borrowers in groups A and B
• Almost 20% reduction in late payments
• Onset of late payment and number of late payment
spells are reduced for treatment groups A and B
• Ultimate default seems to converge: Accounts in
default are handled by separate department
• Improved outcomes at renewal stage for
borrowers in groups A and B
Reachability
Complaint Behavior
Results II
•
Compare treatment groups A versus B
• Other treatment groups did not have outreach data
• Significant higher likelihood of reaching clients in A
group than B; especially on second follow up call
• Likelihood of complaints are lower for treatment
group A but conditional on complaint the nature of the
complaint seems more serious
•
Find similar results for ability to reach customers for
follow up calls by ICICI bank and cross selling
Take Away
• Personalized attention by loan officers matters
for repayment behavior and loan renewal
decisions
• Groups A & B have significant change in repayment
• Relationship lending affects client’s willingness
to engage in moral hazard behavior and loyalty
to the bank
• Relationships constitute “alternative collateral”
Conclusion
• Rethink model of credit risk in loan markets
• Default decisions are situation contingent
• Screening versus moral hazard?
• Stability of prediction models?
• Advances in lending products helps borrowers
improve their credit type
• Virtuous cycle of credit access
• Need more work on equilibrium effects and
competitive responses of banks
What is Behavioral Economics?
Clocky helps you get out of bed.
Hit the snooze button and it rolls
off the nightstand, around the
room and hides.
Next time it goes off you must
get up just to find it.