This presentation is part of the programme of the International Seminar "Social Protection, Entrepreneurship and Labour Market Activation: Evidence for Better Policies", organized by the International Policy Centre for Inclusive Growth (IPC-IG/UNDP) together with Canada’s International Development Research Centre (IDRC) and the Colombian Think Tank Fedesarrollo held on September 10-11 at the Ipea Auditorium in Brasilia.
Margarita Beneke & Amy Angel: Conditional cash transfers and rural development in latin america
1. Conditional Cash Transfers and
Rural Development in Latin America
Country Study: El Salvador
Margarita Beneke de Sanfeliú
Amy Angel
2. Please note:
• This is a LONG version of the actual
presentation.
• It is intended to provide our discussant with
more details about our methodology
• Actual presentation will be shortened to 10 –
12 minutes
• If more information is needed prior to the
event, please contact msanfeliu@fusades.org
3. Introduction
• This research will try to understand how households react to
the intervention of both Conditidonal Cash Transfers (CCT)
and Rural Development Projects (RD), like those promoted by
IFAD, compared with being exposed to only one of these
interventions.
• We aim to identify synergies and complementarities between
both types of interventions.
• If synergies are identified, development projects could be
more effective in reducing poverty, and CCT programs could
find better 'graduation strategies'.
4. Project objectives
• To describe and understand the mechanisms (at the household
and community levels) through which there exist or could exist
synergistic effects between rural development and CCT
programs.
• To inform policy makers at the national level and international
organizations that provide financing for CCTs and rural
development projects, about the potential for synergistic effects
between both types of interventions, and to suggest alternative
program designs to enhance and exploit these effects.
• To provide feedback and build capacity within IFAD's Country
Program Managers and country teams, to take greater
advantage of potential synergistic effects between IFAD projects
and CCT programs.
5. Project strategy
• COMPONENTS
– Technical component: evaluate the effect of having access to
CCT and RD, in terms of use of economic assets, poverty
reduction, gender effects and financial inclusion.
– Policy advocacy component : generate lessons and influence
policy decisions so that key aspects such as graduation from
CCTs or participation in RD projects take advantage of
multiplier effects that can enhance impact in reducing poverty
and improve resilience of poor rural women and men.
• COUNTRIES
– Group 1: Colombia, Peru, El Salvador: primary data collected
– Group 2: Mexico, Brazil and Chile: analytical studies with
secondary data, to provide lessons for group 1.
• SCIENTIFIC COMMITTEE
6. Hypothesis
• Households that are beneficiaries of CCT programs and are
involved in IFAD-RD projects compared with families that just
receive one type of intervention (CCT or IFAD-RD) and with
families that do not receive any kind of intervention
– Will be more successful the income level of the families
or in a given measure of poverty
– Will have greater access to the formal financial markets
financial inclusion indicators.
– Women will be more empowered role in household
decisions
– Will have better opportunities to allocate and enhance
economic assets productivity of production systems
within the household.
7. Methodology
• Literature review, sistematization of other impact evaluations
and surveys
• Document programs (CCT and DR)
• Mixed methods
– Quantitative: Household survey. One round (trying to get
“before” info from some other source).
– Qualitative: focus groups and in depth interviews with
households, project administrator, and communities
• Dialogue with Policy makers, program administrators,
politicians, other groups interested in RD
8. CCT in El Salvador:
Comunidades Solidarias Rurales (CSR)
• Cash transfers:
– Education bonus (children under 18 without complete
primary school): $15 per month
– Health bonus: children under 5 or pregnant women: $15
– Mixed bonus: if both conditions are met: $20
• Only one bonus per family
• The program is in 100 municipalities: rolling entry by level of
poverty
• Important: Af family could join the program only they met
qualifying characteristics at the moment of the “census”
• The program involves: “capacitaciones” and opportunities to
socialize
9. IFAD and MAG RD Programs
• Food Security (FS), for subsistence farmer families. Includes
field schools for traditional crop improvement, crop
diversification, natural resource management and home
health. Uses demonstration families who transfer knowledge
and technology to secondary families through demonstration
plots and direct training.
• Value Chains (VC), for small and medium commercial farmers
and entrepreneurs in handicrafts and rural tourism. Activities
include field schools to provide training to farmers and efforts
to combine purchases of inputs and marketing of products for
greater economies of scale.
10. Comparison groups
Proyectos DR
Beneficiarios Otros
A B
CCT+DR CCT
AC BC
CCT+DR CCT
C D
DR None
Salieron de
No recibe CSR
TMC
Comunidades
Solidarias Rurales
Recibe TMC
Nunca
recibieron
Possible
“graduation”
strategy
11. Possible comparison groups
No. Treatment Comparison Aspect to explore
1 A (CCT+RD) C (RD)
The additional effect of being exposed to the
two interventions, as opposed to only RD
2 A (CCT+RD) B (CCT)
The additional effect of being exposed to the
two interventions, as opposed to only CCT
3 A (CCT+RD) D (None)
The additional effect of being exposed to the
two interventions, as opposed to neither.
4 C (RD) D (None)
The effect of being exposed to RD interventions,
as opposed to no treatment
5 B (CCT) D (None)
The effect of being exposed to CCT
interventions, as opposed to no treatment
6 A (CCT+RD) AC (CT + RD)
The effect of receiving RD and continuing being
“active” in the CCT program vs not receiving CCT
7 AC (CT + RD) BD (CT)
The effect of being exposed to RD after exiting
from the CCT program (a “graduation strategy”)
as opposed to no treatment after exiting the CCT
program
8 AC (CT + RD) C (RD)
The effect having participated in CCR and
receiving DR as opposed to only particating in DR
12. Sample frame
Sample frame constructed combining several sources of info:
• Ministry of Agriculture (FS and VC 2010-2013)
Nombre de archivos
Base de Beneficiarios PRODEMOR CENTRAL PAF EP 3,177
Base de Beneficiarios PRODEMOR CENTRAL PAF SAN 14,136
CADENAS_PRODUCTIVAS_PRODEMORO 1,016
SEGURIDAD_ALIMENTARIA_PRODEMORO 14,421
Familias beneficiarias PAF SAN 25,357
Total 58,107
• FISDL:
n° de
registros
- Listings, different formats;
- Listings for some areas not
received on time
- No information about future
participants
– Census conducted prior to the start of CSR in each
community: 2005 to 2009, accordingly. Has info for ALL
households in 100 municipalities
– Complete CCT beneficiaries data base
We were able to match 19,342 individuals in 14,184 households
13. Sample frame (cont…)
• The unit of analysis is the household
• From the complete sample frame, we eliminated:
– Households with more than one CCT participant (5%), so
remaining households would have similar transfer values
– Households that exited CCT before 2012
– Households that entered a RD program before 2012
– Households without at least one child two years younger
or two years older than the limit for CRS at the time of
selection into the program (So treatment and control
households would be “almost” elegible to receive CCT)
– Households in municipalities without RD programs (so all
remaining could potentially participate).
14. • Each observation in the sample frame was classified into one of
the groups: A,B,C,D, AC, BD
– Comparisons among groups with RD beneficiaries can be
done directly A, C and AC
– Comparisons involving groups with and without RD
beneficiaries cannot be done directly (possible selection bias):
B, D and BD
• Beneficiaries of FS and VC programs appear to be different in
variables in Proxy means test used by CSR (according to
descriptive statistics using original “census” data) .
• We used discriminant analysis to classify non-RD beneficiaries
into one of the B, D and BD (separating by FS and VC) Using variables in
CCT Proxy means
test
Sample selection
15. Sample selection (cont…)
• To assign each observation to a treatment a control group for
each “strategy” and type (FS or VC)
- We used Propensity Score Matching (PSM) for each pair
of treatment and control. We left only those observations
in the area of common support
- We only found comparison-pairs that were sufficiently
similar for the following strategies:
6,176 hogares en 54 municipios
No. Treatment Comparison FS VC
1 A (CCT+RD) C (RD) YES YES
2 A (CCT+RD) B (CCT) NO NO
3 A (CCT+RD) D (None) NO NO
4 C (RD) D (None) NO NO
5 B (CCT) D (None) NO NO
6 A (CCT+RD) AC (CT + RD) YES NO
7 AC (CT + RD) BD (CT) YES NO
8 AC (CT + RD) C (RD) YES NO
16. Sample design
Available observations
FS VC
A CCT + RD 918 287 1,205
C RD 325 185 510
AC CCT + RD 643 643
BC CCT 3,818 3,818
Total 5,704 472 6,176
Randomly
selected from
all available in
each group
RD type
Group Total
RD type
FS VC
Group
Total
A CCT + RD 250 287 537
C RD 250 185 435
AC CCT + RD 250 250
BC CCT 250 250
Total 1,000 472 1,472
17. Geographical location of
final sample
Similar to:
Actual Intersection
of CCT and RD VC and FS
FS only
Field work:
27 Jan- 1 april
FS VC
A CCT + RD 241 239 480
C RD 213 152 365
AC CCT + RD 230 230
BC CCT 226 226
910 391 1,301
Group
RD type
Total
Total
Final sample
18. Survey Questionnaire Modules
A. Identification
B. Household composition (HC)
C. Education (HC)
D. Health (HC)
E. Ocupation and labor market (W)
e.1 Ocuations details (I) (D)
e.2 Job search
F. Information about land (PC)
G. Agricultural production
g.1 Crops (D)
g.2 Animales (D)
g.3 Equipamiento (CF)
H. Associations and social capital
h.1 Participation in associations
h2. Community relationship
I. Rural Development proyects
i.1 Food security
i.2 Value chains
J. Housing conditions and assets (PC)
K. Remittances and other income (I) (D)
L. Food security (Proxy, I)
M. Expectations, aspirations and
empowement
m.1 General perceived self-efficacy
m.2 Mood and self-esteem
m.3 Locus of control
m.4 Aspirations
m.5 Decision making (W)
N. Delinquency and other security issues
O. Financial services(FI)
m.1 Debts
m.2 Savings
P. Shocks
CH: Human capital
PC: Physical captal
D: Economic Diversification
I: Income
FI: Financial inclusion
W: Rol of women
19. Qualitative strategy
• Stage 1. Before survey (to inform data collection)
– Focus groups with:
• CSR regional staff
• IFAD program coordinators
• Field staff RD programs
– Interviews with CSR and RD administrators
• Stage 2. After survey (to explain and further explore findings
and possible RD program modifications)
– Feedback from survey fieldwork personnel
– Focus groups and semi-structured interviews (perceptions
of 106 individuals selected from survey sample)
• A, C, AC and BD, to explore the “why” of results)
– Interviews with community leaders
– Interviews with CSR and RD staff
20. Identification strategy
From survey results we noted:
- From the original CSR census (2005 to 2009) to the survey in
2014, there was some household restructuring (some
members left CCT households and “took” transfers with
them, or new members “brought” transfers to non-CCT
households).
- Some households that were identified as beneficiaries of RD
programs in the official listings, responded that they were
not actual beneficiaries
Identification strategy: Intent to Treat Effect, using
single difference with PSM (variables in original census)
21. First Quantitative Results:
In El Salvador, what generates synergy between RD and CCT
programs, “cash” or “something else”? Is adding RD a good
“graduation strategy”?
Given that:
The $ transferred is very small (avg. $15.18 per month, currently
payable $60 every 4 months); amount is the same since 2005
Previous evaluations of CSR showed that women´s empowerment
(mainly in the domestic domain) increased due to “capacitaciones”
and opportunities for social interaction provided by the program
We test the (additional) hypothesis that the $ amount is NOT what
generates the synergy:
Results [A (CCT + RD)] = Results [AC (CCT + RD)]
If this is the case, then we can test if adding RD after “exiting” CCT
program would be a good “graduation strategy”
Results [AC (CCT + RD)] > Results [BD (CCT)]
22. In El Salvador, “something else” generates synergy
between RD and CCT programs… and adding RD is a
good “graduation strategy” …
• We did not find a significant difference in outcomes for
those still receiving CCT and those who exited the program
Group Differences
A (TMC+DR) AC (TMC+DR) BD (TMC) A vs AC
Característica
AC vs BD
Empowerment
Empowerment Index 66.3% 68.4% 64.8% 3.6% **
Domain: Production 71.6% 77.2% 66.6% 10.6% ***
Domain: Resources 54.8% 56.8% 50.6% 6.2% **
Domain: Income 87.3% 88.9% 80.3% 8.6% ***
Financial Inclusion:
Accounts or formal credit formal inst. 22.4% 21.7% 15.0% 6.7% **
Income proxy:
Asset index (productive or household) 7.70 8.47 7.63 0.84 ***
Reduced assets because of food insecurity 22.8% 22.6% 28.3% -5.7% *
N 241 230 226
• Adding RD produces “gains”…. even in empowerment
23. … especially for women
Empowerment
increased, even
in non “domestic”
domains
Group Differences
A (TMC+DR) AC (TMC+DR) BD (TMC) A vs AC
Selected variables
WOMEN
Empowerment
AC vs BD
Empowerment Index 58.5% 62.4% 59.0% 3.4% *
Domain: Production 50.0% 66.3% 53.9% 12.5% **
Domain: Resources 41.4% 48.9% 42.7% 6.2% *
Domain: Income 82.3% 87.5% 77.1% 10.4% **
Financial Inclusion:
Accounts or formal credit formal inst. 20.2% 16.7% 14.1%
Income proxy:
Asset index (productive or household) 7.69 8.14 7.39 0.75 **
Reduced assets because of food insecurity 22.6% 20.0% 32.4% -12.4% **
N 124 120 142
MEN
Empowerment
Empowerment Index 74.6% 74.8% 74.6%
Domain: Production 94.4% 92.3% 88.0%
Domain: Resources 68.9% 65.5% 63.9%
Domain: Income 92.7% 90.5% 85.7%
Financial Inclusion:
Accounts or formal credit formal inst. 24.8% 27.3% 16.7% 10.6% *
Income proxy:
Asset index (productive or household) 7.71 8.83 8.06 0.77 **
Reduced assets because of food insecurity 23.1% 25.5% 21.4%
N 117 110 84
24. Results
• We could identify some positive synergies (Intent to Treat
Effects) in all domains evaluated (income, empowerment,
financial inclusion)
• There is evidence that the “other” activities of the CCT
program (“capacitaciones” and opportunities of social
interaction) combined with RD, produce gains, especially for
women.
• Results in terms of empowerment in areas other than
“domestic”, are larger for women.
• There is evidence that RD programs could be good
“graduation strategies” for CCT beneficiaries
• Preliminary qualitative results suggest that there is
substantial potential for improving results we will be able
to suggest modification to program design and
implementation, including further inter-agency coordination