armenia: water-to-market farmer training/media/publications/pdfs/... · 2014. 5. 4. · investing...
TRANSCRIPT
Investing in Agricultural Training:
Effects of the Water-to-Market
Farmer Training in Armenia
January 2013
Ken Fortson, Anu Rangarajan, Randall Blair, Joanne Lee, Val Gilbert
Many donors fund agricultural training programs
– As standalone programs or as part of broader programs
– World Bank alone has spent $400 million to fund agriculture
education and training over past 20 years
Yet, meager evidence these programs are effective
– A 2011 World Bank study considered 271 agriculture
evaluations (training and otherwise); only 3 used randomized
controlled trials (RCTs)
– Many “evaluations” simply present select success stories
– Nonexperimental / quasi-experimental studies also less
compelling in this setting
• Selection mechanism is unclear
• Weak correlation of key variables over time
Background
2
Water-to-Market Program and Evaluation
Millennium Challenge Corporation (MCC) funded a
“Compact with Armenia” that included infrastructure
investments in irrigation, roads, and agricultural training
– MCC is a relatively new US development agency; among its
goals, MCC aims to rigorously evaluate its investments
whenever feasible
Today: Findings from evaluation of the “Water-to-Market”
(WtM) training program
– Funded by MCC
– Strongest completed evaluation of this type of program (as far as
we are aware!)
• Features a cluster-randomized controlled trial
• Three-year follow-up period
Two types of training: On-farm Water Management
and High-Value Agriculture
Activities: Classroom and practical training, and the
establishment of demonstration plots to display
technologies in practice
WtM Farmer Training
4
4
Objectives: Teach farmers to adopt… – New and more efficient irrigation techniques (OFWM)
– New cropping techniques and high-value crops (HVA)
Expected results: increased and more cost-effective
agricultural production, higher sales, improved
household well-being
Training targets – OFWM: 45,000 farmers; HVA: 36,000 farmers
– Estimate that about 47,800 unique households participated
Program costs: $14.3 million from 2007 to 2011 – About $310 per trained household
Program model is not universal but is fairly common
for agricultural training
WtM Farmer Training (continued)
5
5
Communities were randomly assigned into a treatment
group or a control group (geographically stratified)
Treatment group communities (112) were offered training;
control group communities (77) were not offered training
during the evaluation period of 3 years – Examine outcomes as of 2010 agricultural season, 2-3 years after
the treatment group communities were offered training
Evaluation Approach
6
6
Communities in the WtM Study
7
Farming Practices Survey developed to provide
longitudinal data for the evaluation – All treatment and control communities included
– 3,547 households included in analysis, 75% of baseline
respondents
• 2,133 treatment group households, 59% of whom completed
training in evaluation timeframe
• 1,414 control group households, 10% of whom completed
training in evaluation timeframe
– Estimates presented are regression-adjusted estimated effects
of the intention to treat (ITT)
Qualitative interviews and reports used to document
implementation (not presented here)
Data Used for the Farmer Training Evaluation
8
Training Did Not Affect Irrigation Practices
Treatment Effects of WtM Training on Irrigation Practices Used by Respondent Households
(Percentages)
Sources: 2007–2008 and 2010–2011 Farming Practices Surveys.
*Denotes statistically significant treatment-control differences at the 10 percent level.
0
10
20
30
40
50
60
70
80
90
Simple Medium Advanced Irrigation Scheduling
Organizational
OFWM Practices
Treatment Group
Control Group
0
-3
0.2 0.3* 0.0
45 45
76 79
Percentage
9
Small Differences in HVA Practices
Treatment Effects of WtM Training on HVA Practices Used by Respondent Households
(Percentages)
Sources: 2007–2008 and 2010–2011 Farming Practices Surveys.
Notes: Estimates may not appear to equal difference in means due to rounding.
10
Training Did Not Affect Crop Cultivation
Treatment Effects of WtM Training on Crops Cultivated by Respondent Households
Sources: 2007–2008 and 2010–2011 Farming Practices Surveys.
Notes: High outliers censored at the 98th percentile for each variable. Estimates may not appear to
equal difference in means due to rounding.
11
Training Did Not Affect Household Income
Treatment Effects of WtM Training on Annual Household Income (USD)
Sources: 2007–2008 and 2010–2011 Farming Practices Surveys.
Notes: High outliers censored at the 98th percentile for each variable. Estimates may
not appear to equal difference in means due to rounding.
12
No effect on types of crops cultivated or on
agricultural expenses
Some statistically significant estimated effects on
production of specific crops, but mixed signs
– Positive effect on tomato and potato market value
– Negative effect on grape market value
No effects on consumption or poverty reduction
Snapshot of Other Findings
13
No evidence that training substantially improved
farmers’ well-being measured by income, avoidance
of poverty, or consumption
Limited evidence of HVA practice adoption, and no
evidence of irrigation practice adoption
Why might treatment effects have been limited? – Too soon? Only 2-3 years after training
– Light touch intervention (one time training, 4-5 days each)
– Large numbers to train, no “targeting” of farmers
– Practices may have been too advanced or expensive
– Some practices not aligned with incentives, e.g., water savings
– Reliable irrigation not available until late in Compact
– Farmer knowledge may not be the constraint it’s thought to be
– Farmer behavior may be harder to change than was thought
Conclusions
14
Mathematica® is a registered trademark of Mathematica Policy Research.
Ken Fortson:
For More Information
15