2 shah dse2
TRANSCRIPT
The Dirty Business of Open Defecation: Lessons from a Sanitation Intervention
Manisha Shah
UCLA & NBER
Lisa Cameron, Monash
Paul Gertler, UC Berkeley & NBER
2 August 2013
WSP Asked “What works?”
• Evaluation of “at scale” interventions in 6 countries– 3 TSSM– 3 Hand Washing
• Coordinated– Same outcomes– Rigorous causal methods
• WSP learning agenda– Large team of IE experts & operational staff– BMGF funding
Child Health in the Developing World
• One child dies every 15 seconds from diarrheal diseases (WHO, 2000)
• Diarrhea and acute lower respiratory infections (ALRI) account for more than 40% of 10 million annual deaths young children (Black et al. 2003, Bryce et al. 2005)
• WHO and Unicef estimate 60% of poor (2.6 billion) lack access to improved sanitation (JMP 2006)
• 18.6 million people in Indonesia lacked access to proper sanitation last year
• Indonesia “not on track” for sanitation MDG
Social Marketing Events +
Communication Campaign
Demand sideSocial Marketing of Sanitation:
Supply side
Popularize improved sanitation
Sanitation choice catalogue
Training masons
3
Total Sanitation and Sanitation Marketing in Indonesia (SToPs)
Behavior Change Communications :
2
Community-led Total Sanitation:
Demand side
Stop OD by raising awareness
“map” the village
“walk of shame”
Triggers community action
Action plan & monitoring
1
4
Basic IE QuestionsWhat is the overall Impact of TSSM on• Sanitation improvement
and construction• Open Defecation• Health
– Diarrhea– Parasites– Anemia– Height and weight – Cognitive development
Advanced IE Questions
2. Decomposition of overall OD effect into
– Sanitation construction– Increased use of
sanitation (behavioral)
3. Liquidity constraints
4. Effects of stronger implementation
I. Theory of Change
II. IE Design
III. ResultsI. Sanitation
II. Open Defecation
III. Health Outcomes
IV. Implementation issues
V. India results
VI. Policy Messages
Today….
7
Conceptual Framework: Theory of Change
D = Open Defecation Rate
T = Share of households that have sanitation
DT = Open Defecation Rate of HHs with Sanitation
DNT = Open Defecation Rate of HHs without Sanitation
Decompose Open Defecation Rate into:
TSSM Pathways To Reduce OD
TD
TD
DDT
NT
T
NTT
1 = sanitation havenot do who thoseof use in .3
= sanitation have who thoseof use in .2
= onconstructi Sanitation .1
Indonesia and East Java
http://education.yahoo.com/reference/factbook/id/map.html
Randomly Sampled 160 communities (‘dusun’ or hamlet)
Randomly Assigned to
8 districts participated in study
Treatment80 dusuns
Random Sample 1046 HHs
East Java: 29 districts total10 districts in TSSM Phase 2
Control80 dusuns
Random Sample1041 HHs
Sampling & Experimental Design
Collected measures/outcomes
Community (160 dusuns):• Water supply • Sanitation facilities• Sanitation behavior• Existing programs
Household (2,087 hhs):• Basic demography• Welfare & labor market• Water supply facilities• Sanitation facilities• Sanitation behavior
Children <5 (2,353 children):• Anemia & anthropometry• Diarrhea & ALRI• Child development (ASQ)• Feeding & behavior
Longitudinal (2,087 hhs):• Child health measures• T/C compliance measures
Endline (2,500 hhs):• 2638 Children <5• Fecal samples • Everything else similar
All
Sanita
tion at Base
line
No Sanita
tion at Base
line
00.020.040.060.08
0.10.120.140.160.18
0.2
Sanitation Improvement/Construction Between Baseline & Endline
TreatmentControl
(1) (2) (3) (4) (5) (6) (7) No Sanitation
at BLNo Sanitation
at BL Full Sample -
No controlsFull Sample -
controlsPanel No sanitation
at BaselineSanitation at
BaselineNon-Poor Poor
Treatment 0.37*** 0.039*** 0.032*** 0.038** 0.007 0.044** 0.032 [0.01] [0.01] [0.01] [0.02] [0.02] [0.02] [0.03] Observations 2,500 2,500 1,908 939 969 596 333R-squared 0.11 0.11 0.12 0.21 0.16 0.22 0.43Means 0.128 0.128 0.128 0.081 0.171 0.105 0.042
Toilet Construction ITT Estimates
Open Defecation
Sanitation at
Baseline No Sanitation at Baseline
All Non-Poor Poor
Anyone
Treatment -0.06** -0.06*** -0.06** -0.06*
Control Mean 0.24 0.83 0.80 0.86
Women
Treatment -0.01 -0.06** -0.05* -0.07*
Control Mean 0.072 0.77 0.73 0.83
Men
Treatment -0.03* -0.07** -0.05* -0.08*
Control Mean 0.12 0.79 0.77 0.83
Children
Treatment -0.04** -0.07** -0.07** -0.07*
Control Mean 0.18 0.79 0.75 0.84
Observations 967 939 596 333
Open Defecation
Sanitation at
Baseline No Sanitation at Baseline
All Non-Poor Poor
Anyone
Treatment -0.06** -0.06*** -0.06** -0.06*
Control Mean 0.24 0.83 0.80 0.86
Women
Treatment -0.01 -0.06** -0.05* -0.07*
Control Mean 0.072 0.77 0.73 0.83
Men
Treatment -0.03* -0.07** -0.05* -0.08*
Control Mean 0.12 0.79 0.77 0.83
Children
Treatment -0.04** -0.07** -0.07** -0.07*
Control Mean 0.18 0.79 0.75 0.84
Observations 967 939 596 333
Open Defecation
Sanitation at
Baseline No Sanitation at Baseline
All Non-Poor Poor
Anyone
Treatment -0.06** -0.06*** -0.06** -0.06*
Control Mean 0.24 0.83 0.80 0.86
Women
Treatment -0.01 -0.06** -0.05* -0.07*
Control Mean 0.072 0.77 0.73 0.83
Men
Treatment -0.03* -0.07** -0.05* -0.08*
Control Mean 0.12 0.79 0.77 0.83
Children
Treatment -0.04** -0.07** -0.07** -0.07*
Control Mean 0.18 0.79 0.75 0.84
Observations 967 939 596 333
Estimating Model Parameters from Decomposition
(1) (2) (3) (4) Any
Householder Women Men Child
Treatment -0.06** -0.05* -0.06* -0.06** [0.03] [0.03] [0.03] [0.03] Built Toilet -0.48*** -0.59*** -0.49*** -0.51*** [0.09] [0.07] [0.08] [0.08] Treatment*Built Toilet 0.08 0.15* 0.08 0.07 [0.11] [0.09] [0.10] [0.10] Constant 0.91*** 0.85*** 0.75*** 0.84*** [0.12] [0.14] [0.13] [0.13] Observations 939 939 939 939 R-squared 0.42 0.51 0.46 0.45 Means 0.827 0.765 0.789 0.785
Sample is No Sanitation at Baseline. Robust standard errors in brackets. *** p<0.01, ** p<0.05, * p<0.1 (two-sided test).
Decomposition of Δ in OD • Total estimated effect of TSSM on OD = -.06
• Components:– Δ in sanitation construction (infrastructure)
- .48*(.032) = -0.015– Δ in use of those who have sanitation (behavioral of those who
built) 0.08*.128 = 0.010
– Δ in use of those who do not have sanitation (behavioral of those who did not build)
-0.06*(1-.28) = -.052
• Note that they add up to -0.06
Messages
• TSSM reduced mostly through behavioral change Explained ≈70% of the reduction in OD
• Less successful through sanitation construction
• Big potential gains from sanitation constructionTSSM in Indonesia only increased sanitation by 3.7% At baseline only ≈ 45% had sanitation
Tenancy Issues
Permit Issues
Too Comples
Water not available
No one to build
Soil Conditions
No materials available
Satisfied with current
No Savings
Other
Space
High Cost
0 10 20 30 40 50 60 70 80
Obstacles to Building Sanitation
Households (IE)
Villages (IE)
Villages (Admin)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%
Implementation: % Triggered
Control Treatment
What If All Villages Were Triggered?
Results Summary
• TSSM was successful at– Reducing OD– Improving health outcomes
• Mostly worked through behavioral change• Less successful at motivating sanitation construction• Big potential gains through sanitation construction
– Cost and liquidity constraints biggest obstacles
• Full implementation increases effects by 40%
India Intervention (TSC)
• 80 rural villages in Madhya Pradesh (40T/40C)• Offered subsidies to poorer households and
resulted in a much greater increase in construction (toilet coverage: 22% v 41%; OD decreased 74% v 84%)
• BUT no consistent improvements in child health outcomes– Potential reason is endline happened >6months in
only 14 of 40 Treatment villages
Policy Messages
• TSSM (CLTS) model – Improves health primarily thru behavioral change– Less successful through sanitation construction
• Need to strengthen sanitation components– Subsidized prices– Credit– Community financing
• Need to Improve implementation
Next Steps
Seeking funding to re-visit households to:
• Evaluate households’ willingness to pay for sanitation. Offer microfinance to poorer households. Does this enable communities to become open defecation free?
• Examine the sustainability of the program impacts - whether the toilets are maintained and used in the longer term, and the consequent longer term health impacts.