This document summarizes two randomized controlled trials (RCTs) evaluating financial inclusion programs in Chile. The first RCT evaluated the Chile CUENTA program, which delivers conditional cash transfers through bank accounts rather than cash. It found that 55% of recipients opened bank accounts but impacts on savings were small. The second RCT evaluated a financial education program and found no effects on formal debt levels, interest rates, or credit access. Both evaluations highlight the importance of rigorously studying the impacts of programs aimed at improving financial inclusion and outcomes.
Claudia Martínez: Evaluating the impact of delivering ccts
1. EVALUATING
THE
IMPACT
OF
DELIVERING
CCTS
THROUGH
THE
BANK
AND
FINANCIAL
EDUCATION
Claudia
Mar<nez
Alvear
Catholic
University
of
Chile
September,
2014
2. MAIN
POINTS
§ Research
Agenda
§ Two
RCT
evaluaRons
in
Chile
on
financial
inclusion:
• Chile
CUENTA:
EvaluaRng
the
impact
of
delivering
CCTs
through
the
bank.
• EvaluaRng
the
impact
of
Financial
EducaRon
Program.
3. RESEARCH AGENDA
A
research
agenda
has
been
developed,
based
in
“Todas
Cuentan”
Research
plaYorm
that
promote
programs
evaluaRon
on
financial
inclusion.
The
aim
is
to
idenRfy
soluRons
for
the
local
problems
(resource
management,
women
empowerment).
The
plaYorm
has
incorporated
regional
research.
The
project
is
funded
for:
The
International Development Research Centre (IDRC).
Institute
for Peruvian Studies.
Fundación
Capital.
In
Chile,
the
plaYorm
has
financed:
Chile
CUENTA evaluation.
Financial
Education evaluation.
“
Saving accounts” evaluation.
4. CHILE
CUENTA
Abhijit
Banerjee
Massachuse]s
InsRtute
of
Technology
Claudia
Mar2nez
A.
Universidad
Católica
de
Chile
Esteban
Puentes
Universidad
de
Chile
5. MOTIVATION
Strong
increase
in
CondiRonal
Cash
Transfer
(CCT)
in
the
region
+
concerns
on
lack
of
financial
inclusion.
• Subsidies
payment
through
the
back
accounts.
However,
no
clear
the
evidence
about
the
addiRonal
benefits
that
the
nexus
between
the
CCT
program
and
this
policy
could
generate
(De
los
Ríos
&
Trivelli
2011).
•
Samaniego
y
Tejerina
(2010)
and
Maldonado
y
Tejerina
(2010)
detect
li]le
use
of
the
bank
accounts
between
the
CCTs
beneficiaries
(Oportunidades,
México).
• Seira
(2010)
detects
posiRve
impact
in
the
beneficiary’s
formal
savings
that
get
their
benefits
through
their
bank
accounts
(Oportunidades,
México).
6. CONTEXT
What
is
Chile
CUENTA?
• This
program
changes
the
way
in
which
the
social
benefits
are
delivered:
the
cash
subsidies
payment
changes
to
the
bank
account
subsidies
payment.
Who
implements
it?
• The
Ministry
of
Social
Development
(Chilean
Government).
What
is
the
goal?
• To
improve
poor
people’s
financial
access.
7. EVALUATION
DESIGN
§ Sample:
Five
municipaliRes
of
SanRago,
Chile
§ Time
frame:
program
implementaRon
started
in
October,
2012.
Follow
up:
Late
2014.
§ Treatment
arms:
§ Control
group:
• Puente
Program
beneficiaries
who
receive
their
social
benefits
in
cash.
• N=810
§ Treatment
group:
• Puente
Program
beneficiaries
who
receive
their
social
benefits
in
cash
and
get
the
chance
to
receive
the
subsidies
payment
trough
bank
account.
• N=2,422
8. EVALUATION
IMPLEMENTATION
Take up Chile CUENTA
Rechazados Aceptados No aplica
34%
55%
11%
[1]
[2]
[3]
[4]
Municipality
N Treated
N Rejected
Total accepted
Don’t apply
Take Puente Alto
736
275
333
128
45%
Peñaflor
224
71
144
9
64%
Maipú
370
112
227
31
61%
San Bernardo
811
279
456
76
56%
Conchalí
281
81
164
36
58%
TOTAL
2.422
818
1.324
280
55%
The
treatment’s
individuals
that
didn’t
get
Chile
CUENTA
offer
aren’t
consider.
The
take
up
55%
rate
of
all
treatment
group
allows
to
find
0,22
sd
effect
(US$8
monthly
average
saving
last
three
month).
9. PROCESS
EVALUATION
§ Process
EvaluaRon
in
parallel
to
the
implementaRon.
§ EvaluaRon
AcRviRes:
Focus
Group
and
Interviews.
The
scheme
shows
the
most
important
results.
Posi>ve
Points
Comfort.
OrganizaRon.
Speed.
Bank
account
use.
Security.
Nega>ve
Points
Date
change.
Statement
accounts
are
received.
Li]le
informaRon
about
bank
account
balance
and
charges.
Other
Points
Chile
CUENTA
as
a
good
mechanism
for
money
management
but
not
as
a
saving
incenRve.
Management
problems.
Extra
work
to
“Family
support”.
Ø The
main
obstacles
are
management
problems.
Ø People
appreciate
the
possibility
administrate
money
through
bank
account.
10. NEXT
STEPS
§ Follow-‐up
survey:
December
2014
§ Analysis:
May
2014
11. FINANCIAL
EDUCATION
Claudia
Mar2nez
A.
Universidad
Católica
de
Chile
Esteban
Puentes
Universidad
de
Chile
Alejandro
Drexler
University
of
Texas
at
AusRn
12. MOTIVATION
§ No
clear
evidence
of
what
works
(Xu
and
Zia
2012).
• Rules
of
thumbs
are
more
effecRve
than
regular
training
in
affecRng
business
pracRces
(Drexler,
et
al.
2012).
• In
general,
financial
educaRon
programs
don’t
have
effects
on
sales
or
business
profits
(Karlan
and
Valdivia
2012).
• In
Chile
Mar<nez,
Puentes
and
Ruiz-‐Tagle
(2013)
studied
the
gather
impact
of
financial
educaRon
and
assets
transference.
They
detected
a
posiRve
effect
on
beneficiaries’
income.
13. CONTEXT
§ ¿What
is
the
program
about?
• 20
hours
of
group
workshop.
• Oriented
to
“Microempresa”.
• Four
modules:
1.Budget;
2.Saving;
3.Household
wealth
and
indebtedness;
4.
Investment.
§ ¿Who
implements
it?
•
FOSIS
(MDS).
§ ¿What
is
the
goal?
•
Provide
financial
tools
and
informaRon.
14. EVALUATION
DESIGN
§ Sample:
Entrepreneurship
Fosis
2012
Program
Beneficiaries
(YES,YEI,
YEII).
Regions:
V,
VII
and
RM.
§ Time
frame:
between
December
2012
and
May
2013.
§
Treatment
arms:
The
invitaRon
to
parRcipate
was
randomized.
o Control
group:
• No
invitaRon
to
parRcipate.
o Treatment
group:
• InvitaRon
to
parRcipate.
15. EVALUATION
DESIGN
Dates
and
number
of
evalua>on
par>cipants
by
region
Region
Total Evaluation Participants
Initial Month
End Month
V
976
January 2013
March 2013
VII
470
December 2012
December 2012
RM
3.124
January 2013
May 2013
Total
4.570
Source: self generated with reports FOSIS .
Participants of the evaluation includes invited to participate in the program, the reserve group and the
control group.
16. EVALUATION
DESIGN
§ Impact
variables:
level
of
formal
debt
of
individuals
and
access
to
formal
credit.
(Only
variables
available
in
administraRve
data)
§ Hypothesis:
o Rate
credit
reducRon.
o Debt
level,
two
opposing
effect.
o Be]er
credit
condiRons.
17. EVALUATION
DESIGN
§ Using
administraRve
data:
from
the
Superintendency
of
Banks
and
Financial
InsRtuRons
(SBIF)
and
FOSIS.
§ SBIF
Data:
from
January
2012
to
August
2013.
• Credit
requested
with
informaRon
about
interest
rate,
amount
and
term
in
months.
• The
average
debt
and
the
credit
unions
unpaid
debts.
• Bank
debt
and
average
unpaid
debts
separated
by
type:
commercial,
consumer
and
mortgage.
§ With
Financial
Household
Survey
(EFH),
the
appropriateness
of
using
SBIF
data
the
evaluaRon
is
analyzed:
• 23.8%
of
households
with
the
lowest
income
(1
quinRle)
have
data
that
might
be
found
in
SBIF.
• 64.9%
of
total
debt
corresponds
to
a
formal
debt.
18. EVALUATION
DESIGN
§ FOSIS
data:
• PAM
debts
and
the
insRtuRon
requested
(the
most
common
Fondo
Esperanza
FoundaRon).
• Rates
and
subsidies
amount.
• InformaRon
about
entrepreneurship:
economic
acRvity,
formalizaRon,
sales
levels,
among
others.
§ The
crossing
with
FOSIS
data
allows
to
know
how
many
individuals
applied
funding
through
PAM:
• Between
March
2012
and
December
2013,
408
parRcipants
have
requested
funding
through
PAM,
with
a
total
of
586
credits.
19. EVALUATION
IMPLEMENTATION
Acceptance
rate
of
the
invita>on,
total
and
by
Region
V
VII
RM
Total
Total of Evaluations Participants
976
470
3124
4570
Invited
446
260
1253
1959
Trained
153
96
232
481
Acceptance rate
34.3%
36.9%
18.5%
24.6%
§ High
regional
variaRon.
§ Given
this
acceptance
rate
and
the
number
of
individuals,
the
minimum
detectable
size
is
0.34
standard
deviaRons
(with
80%
power
and
5%
significance).
20. EMPIRICAL
SPECIFICATION
§ The
empirical
specificaRon
used
was:
풚↓풊 =휶+휷 푫↓풊 +휸 푿↓풊 + 풖↓풊 (ퟏ)
풚↓풕 =휶+휷푫+휸푿+! 풚↓풕−ퟏퟐ +풖
(ퟐ)
Where:
o 푦:
The
studied
result
(monthly
level
of
debt,
by
type
of
debt).
o 퐷↓푖 :
Dummy
variable
that
equals
1
if
the
individual
was
invited
to
par>cipate
in
the
workshops
and
0
otherwise.
o 푋:
Control
variables(region,
above
or
under
the
50
income
percen>l
ande
age).
o 푦↓푡−12 :
Monthly
level
of
total
debt/
debt
dummy/
unpaid
debt
dummy,
12
months
before
the
measurment.
§
훽
represents
the
impact
of
the
offer
of
parRcipaRon
in
the
program
21. RESULTS
§ No
effect
on
studied
variables
such
as
interest
rate,
commercial
debt
level,
consumpRon
credit,
mortgage
credit,
cooperaRve
debt,
total
debt
and
probability
of
having
debt
or
unpaid
debt three
months
axer
the
treatment
ended.
§ Similar
results
are
obtained
axer
removing
the
1%
of
higher
debt
levels
(outliers).
§ AddiRonally,
a
regional
analysis
is
done.
22. CONCLUSIONS
No
effect
was
found
over
the
formal
debt
and
the
PAM
credit
taking.
The
low
acceptance
rate
makes
difficult
to
find
an
impact
and
invites
us
to
rethinking
the
program
design.
Effects
in
saving?
The
implementers
and
users
indicate
that
the
topic
has
a
great
relevance
and
the
effort
should
be
conRnued
through
the
FOSIS.
Importance
of
evaluaRon
of
process,
and
impact
evaluaRon.
23. GENERAL
CONCLUSIONS
Important
topic
lot
to
be
learned
Importance
of
process
and
impact
evaluaRon
Growing
literature
Consider
the
heterogeneity
of
individuals
and
the
right
treatment
Rmes
24. EVALUATING
THE
IMPACT
OF
DELIVERING
CCTS
THROUGH
THE
BANK
AND
FINANCIAL
EDUCATION
Claudia
Mar<nez
Alvear
Catholic
University
of
Chile
September,
2014