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.
1. Social protection, entrepreneurship and labor market activation; International Seminar and Policy Forum.
The impact of linking conditional cash transfers
to agricultural credit on productive assets
accumulation of rural households in Peru
Cesar Del Pozo
Brasilia, September 10th – 11th 2014
2. Motivation and Research questions
CCT´s can have effects on agricultural outcomes (Todd et al. 2010; Gertler et al.
2006); CCT´s can reduce liquity constraints, CCT´s are a relevant, stable and
regular source of non-labor income, CCT´s can serves as a form of colateral for
credit.
Credit can increase income- generating activities and can improve assets position
of poor households (Karlan et al., 2007: Banerjee et al., 2009; Dong et al., 2010).
• Can CCT´s increase the stock of productive assets of rural and poor
households in Peru?
• Can the linking of CCT´s to agricultural credit improve assets position of rural
and poor households in Peru?
• Wich is the magnitude of these impacts?
• The linking of CCT´s to agricultural credit is a valid public policy option to
promote rural development?
3. Context background
Programa Juntos: relevant policy instrument to poverty alleviation in Peru:
• Start in 2005
• Operating at national level, mainly in rural areas
• Covers around 700000 households
• Fixed transfers UDS 71 bimonthly for at least 4 year.
• Conditionals: use of health services, school asistance.
• Targeting mechanism poor distritcs and poor households.
Agricultural credit
• Lack access to credit by rural households: 8% of total rural producers has credit
• Several types of credit lenders: informal, private banks, public bank (Agrobanco),
Microfinance Institutions
• Microfinace Institutions are the most relevant credit provider (66%)
4. Methodology: empirical challenges
• Juntos’ was not randomly assigned
• Credit access is a endogenous decision of households
• Programa Juntos is not formarly linked with any credit program at
national level
• The linking of CCT´s to agricultural credit is based on own decisions
of rural households.
• In households surveys does not exist enough information about
productive assets or agricultural credit.
• Census data is available.
5. Methodology: data and variables
Data:
• Agricultural Census: 1994 and 2012
• Around 2 million farming households.
Dependent variables:
• Agricultural assets: cultivated, land, rrrigated land, rate of cultivated land over total
land, rate of irrigated land over cultivated land, accumulation of productive
equipment, productive infraestructure.
• Livestock assets: number of cows, number of sheeps, small animals (guinea pigs
«cuyes»), poultry.
Independent variables:
• Household belong to CCTP in Peru: Programa Juntos.
• Household has agricultural credit by type of credit lender.
• Socioeconomics and geographical characteristics at district and household level.
6. Methodology: identification strategy
Explore targeting rules of Programa Juntos
0
.01 .02 .03
Distritos no coberturados Distritos coberturados
Target, Untreated Target, treated
10 20 30 40 50 60 70 80 90 100 10 20 30 40 50 60 70 80 90 100
Incidencia
Graphs by Distritos coberturados por el Programa Juntos
Districts with a Poverty headcount index > 40% is useful to identify poor
districts that are not yet incorporated by Juntos. It is caused by
unobservables (assumed exogenous).
7. Methodology: identification strategy
How deal with endogenous decisions of rural household about
agricultural credit? In this study, I categorizing rural households based
on they agricultural credit decisions:
Programa Juntos status Did not request
agricultural credit
Yes they request
agricultural credit (and
obtain it)
Target, treated Beneficiaries of CCT´s
without credit
Beneficiaries of CCT´s
with credit
Target, untreated Non-beneficiaries
without credit
Non-beneficiaries with
credit
8. Methodology: identification strategy
Final sample is aprox. 400.000 rural
households in 561 target districts:
• 450 target, treated districts: 300.000
households (treatment group)
• 111 target, untreated districts: 100.000
households (control group)
Baseline 1994, “Middleline” 2012
• Panel data at district level
• Pooled cross-sectional data at household
level
9. Methodology: quasi-experimental approach
Technical note: to improve comparison among districts and rural
household I apply a Propensity Score Matching in both districts and
household level to replicate targeting process and to reduce initial
observable differences.
Them, I apply a Differences in Differences model:
+ ߜ௧ + ߙ−௨௧௦−ௗ௧ሺ,
ܻ,,௧ = + ߛ,
∗ ௧ሻ + ܺ,,௧
′ ߚ + ,,௧
C is the households’ decision about agricultural credit:
C=0, Did not request agricultural credit
C=1, Yes they request agricultural credit (and obtain it)
C=2…5, credit by lender type (informal credit, private banks, public bank, Microfinance Institutions)
10. Dependent
variables
Mean
Average Treatment Effects on the treated (ATT)
• CCT increase cultivated land
• CCT+Credit increase cultivated land, increase poultry, and generate a type
of assets specialization
• CCT or CCT+credit no effects on productive equipment and/or productive
infraestructure
Results 1: Average Effects
C=0 C=1
Informal
credit
Private
banks
Public
banks
MFI
Cultivated land 2.06 0.33*** 0.64*** 0.57 -0.04 0.98*** 0.73**
Cow 2.5 0.04 -0.75*** 0.37 -0.66 -0.19 -1.20***
Sheep 5.97 -0.54 -1.37* -0.56 -1.89 -2.60* -0.12
Small animals 5.62 0.34 1.08 2.95 0.20 0.73 0.24
Poultry 7.52 1.32 3.92*** 7.49*** 5.41*** 4.98*** 1.98***
Obs 345931 22043 1095 2169 5126 11810
***, ** and *; denote significancy at 1%, 5% and 10%, respectively
Source: Own estimations
11. Results 1.1: Effects by gender
Average Treatment Effects on the treated (ATT) on
female household head
C=0 C=1
Public
banks
Dependent
variables
• CCT reduce cultivated land and the acumulation of small animals if
household head is female.
• CCT+Credit no effects if household head is female.
MFI
Cultivated land -0.12*** 0.11 0.03 -0.22
Small animals -0.77*** -0.82 1.76 0.53
Poultry -0.70 -1.93 -2.11 2.54
Obs 345931 22043 5126 11810
***, ** and *; denote signi ficancy at 1%, 5% and 10%, respectively
Source: Own estimations
12. Technical note: I apply an additional empirical approach to estimate the impact of
agricultural credit on beneficiaries of Programa Juntos. For deal with edogenous
decision of access to credit employ a IV approach.
First stage (Stiglitz y Weis, 1981; Carter y Olinto, 2000; Guirkinger et al., 2007; Cámara
et al., 2013): access to agricultural credit dependent of credit supply at local level (both
offices and «cajeros corresponsales»), own land, entrepreneurship trainning, technical
assistance (+). Educational level, age of household head, gender of household head
(female), isolation, population density and altitude (-).
Second stage:
Results 2: The impact of agricultural credit on Juntos households
Average Treatment Effects on the treated
(ATT) on Juntos households
Access to
agricultural credit
Access to agricultural
credit MFI
Variables
Cultivated land 1.28*** 1.10***
Small animals 13.18*** 22.08***
Poultry 5.82*** 6.92***
Obs 301368 301368
***, ** and *; denote significancy at 1%, 5% and 10%, respectively
Source: Own estimations
13. Initial conclusions
• Evidence of that the link of CCT´s to agricutlural credit improve assets
position of rural and poor households in Peru (cultivated land,
livestock accumulation.
• The magnitude of these impacts are relevant: increase 31% cultivated
land and 52% accumulation of poultry.
• The linking of CCT´s to agricultural credit can be a valid public policy
option to promote rural development, rural credit market have little
coverage and have few specific products, more dicuss about it.