Household Determinants and Impact of Soil and Water Conservation Practices in the Blue Nile
1. Household Determinants and Impact of
Soil and Water Conservation Practices
in the Blue Nile
Emily Schmidt (IFPRI)
Fanaye Tadesse (IFPRI)
Ministry of Agriculture
October 27th, 2011
2. The Blue Nile (Abbay) Basin
• Although Ethiopia’s biophysical potential is significant, land degradation and
poverty continue to challenge sustainable agricultural development
opportunities (studies on this include: Desta, et al. (2001); Shiferaw and
Holden (2001); Tefera, et al. (2002); Zeleke and Hurni (2002); Okumu et al.
(2002); Sonneveld (2002)).
• This is further aggravated by high population pressure in rural highlands,
climatic variability, limited use of sustainable land management practices,
and a high dependence on rain-fed agriculture.
• The on-site effects of land degradation (eg. erosion and loss of top soil),
measured in lost agricultural production is estimated to cost 2 to 6.75% of
AGDP per annum (Mahmud, et al. 2005)
3. Policies for addressing SLM strategies
• Previous 5 year plan (PASDEP): series of land and watershed
management activities with the goal of augmenting agricultural
production.
– piloting and implementing community-based approaches to SLM
– scaling up successful models for watershed management
– strengthening natural resource information management and M&E
• Recent 5 year plan (GTP): government outlines the need to promote and
invest in SWC moisture retaining taking into account varying agro
ecological zones (GTP, 2011).
• CAADP includes a major emphasis SLM as part of Pillar I.
– reversing fertility loss and resource degradation,
– supporting the rapid adoption of SLM practices;
– and improving management of water resources.
• Sustainable Land Management (SLM) Program in collaboration with MOA
and other donors
4. Study focus: Blue Nile (Abbay) Basin
• Assess the determinants of adoption of soil and
water conservation (SWC) structures
• Evaluate SWC private adoption impact on value of
production per hectare
• Understand time horizon of impact (how long
does it take to experience a benefit)
• Future: Explore policy options for incentivizing
local investment and up-scaling of sustainable land
watershed management activities
– Greater cost effectiveness, scalability, and
sustainability?
5. Sample Selection
• 9 woredas (1810 HHs) within the Blue Nile Basin
(*note: this survey is not nationally representative)
• Stratification: Random sample within woredas that have a recently
started or planned SLM program (SLMP – GIZ and World Bank)
– 3 sites (kebeles) per woreda (SLMP woredas)
• Past or Ongoing program
• GIZ Planned program (for 2011)
• No formal past program
• Nile Basin Development Challenge (NBDC) woredas: 4 sites within
each woreda / watershed
– 2 upstream from instrumentation
– 2 downstream from instrumentation
10. Ongoing SLM activities
Households Using SLM on Private Land
Yes No Total
Alefa 50% 50% 100%
Fogera 54% 46% 100%
Misrak Estie 54% 46% 100%
Gozamin 21% 79% 100%
Dega Damot 82% 18% 100%
Mene Sibu 7% 93% 100%
Diga 32% 68% 100%
Jeldu 2% 98% 100%
Toko Kutaye 79% 21% 100%
Average 40% 60% 100%
11. Ongoing SLM activities (2)
Percent of households reporting activities implemented in the village
25
20
15
10
5
0
12. Ongoing SLM activities (3)
Households who received assistance by type of support
Type of support Freq Percent.
Advice on how to construct bunds or terraces for soil conservation 1,107 61%
Advice on when to apply fertilizer 1,092 60%
Advice on how to apply fertilizer 1,086 60%
Advice on how to build drainage to reduce erosion 1,085 60%
Assistance in obtaining fertilizer 1,031 57%
Advice on the best time to plant crops 947 52%
Assistance in obtaining improved seeds 920 51%
Suggest new crops to grow 913 50%
Advise on procurement of livestock vaccines 783 43%
Advice or support of other veterinary services, including medicines 740 41%
Advice on the construction of irrigation or water harvesting
systems 705 39%
Advice on how best to deal with insect infestations 689 38%
13.
14. Perception of SLM activities
Most Important type of Infrastructure Built
(Percent of total households)
40
30
20
10
0
15. Perception of SLM activities (2)
Most Successful Sustainable Land Management activities (%)
40
35
30
25
20
15
10
5
0
stone soil bund check dam trees drainage grass strips
terrace planted ditch
16. Perception of SLM activities (3)
Household preference for future infrastructure (%)
25
20
15
10
5
0
17. Impact Analysis
• Nearest Neighbor Matching: measure the impact of
adopting specific SLWM technologies on value of
production and livestock holdings (match adopters vs.
non-adopters) based on observable characteristics
• Continuous Treatment Effect Estimation: estimate the
benefit (increase in value of prodution) given the number
of years adopters maintain SWC structures on their private
land (within adopters)
18. • Definition of SWC Adopters:
– Households that implemented
• Terraces
• Bunds
• Check dams
on at least 1/3 of private land within the last 15
years (1994-2009) = 24% of sample
• Definition of Value of Production:
– The total value of all crops harvested in Meher
2001 EC (2009-2010) per household
19. Determinants of Household Participation
Variable (only significant variables shown here) dy/dx Std. Err.
HH head age (years) -0.013 ** (0.005)
HH head age sq. 0.000 * (0.000)
Land size in hectares 0.019 ** (0.009)
Land size sq. -0.001 (0.000)
Household experienced flood and erosion (yes=1) 0.081 ** (0.034)
Slope (omitted=flat slope)
Steep slope (percentage of plots with steep slope) 0.159 *** (0.055)
Mixed slope (percentage of plots with mixed slope) 0.056 (0.077)
Fertilizer use (yes=1) 0.061 ** (0.028)
Soil Quality (Omitted=fertile land)
Semi-fertile land (percentage of plots that are semi-fertile) 0.066 *** (0.041)
Non fertile land (percentage of plots that are not fertile) 0.149 * (0.050)
Agroecological Zone (Omitted=Dega: mid-highlands 2,300-3,200m.)
Kolla (lowlands: 500 – 1,500m.) -0.181 *** (0.026)
Woina Dega (low-highlands: 1,500 – 2,300m.) -0.176 *** (0.052)
Wurch (mid-highlands: 3,200 – 3,700m.) 0.282 * (0.157)
Kilometer distance from city of at least 20,000 people -0.010 *** (0.003)
Number of observations=1256
Prob > chi2 =0
Pseudo R2 =0.2480
20. Percent of total plots under SLWM on private land
(1944-2009 )
20
18
16
14
12
10
8
6
4
2
0
21. Nearest Neighbor Matching – split sample
Outcome Variable ATT Observations
1992-2002 (1985 – 1995 E.C.)
Value of Agricultural Production 0.152 ** 1373
(0.071)
Livestock value (in Birr) -0.429 1318
(.100)
2003-2009 (1996 – 2002 E.C.)
Value of Agricultural Production -0.015 1397
(0.062)
Livestock Value (in Birr) -0.158 1327
(0.095)
• Households that adopted SLWM on their private land in the first
10 years of analysis have 15.2% (2,329 birr avg.) greater value of
production in 2010 than non-adopters.
22. Early adopters show significant increases in
value of production, but how long does it take?
• We want to understand when households experience a
significant increase in value of production, and how long
they need to maintain this infrastructure in order to do so
•We also want to understand how much benefit does a
household gain by maintaining a structure another year
23. Estimated Treatment Level Function
E[lnvalueprod(t+1) - E[lnvalueprod(t)] 0.2
Level of
0.15 treatment Marginal
0.1 (years) effect
0.05 7 0.03
8 0.04
0
Treatment range 9 0.05
-0.05 with statistically 10 0.07
significant impact 11 0.08
-0.1
12 0.09
-0.15
13 0.11
1 3 5 7 9 11 13 15 17 14 0.12
Treatment Level 15 0.14
• On average, a household must maintain SWC structures for at least 7 years in
order to experience a significant increase in value of production.
• Well maintained SWC structures may slow ongoing degradation in the initial years
of maintenance, but nutrient build-up may take time to show significant impact on
value of production
24. Conclusions
• Households with poorer quality of soil, steeper slopes, and
experience erosion or flooding are more likely to invest in
SLWM
– Agricultural extension and SWC program targeting may be well placed
• Remoteness has a significant but small correlation with
household probability of adopting SLWM.
– If farmers do not see a marketable outlet for increased
production, they may be less willing to implement yield increasing
strategies
– SLM extension programs in remote areas may have lagged behind
• Those that invest in fertilizer are also more likely to invest
– Proxy willingness to invest in technologies / innovation to increase
output.
– In fact, those that put fertilizer on their plots are 6 percent more likely
to adopt SLWM technology as well.
25. Conclusions (2)
• Households that construct and sustain SWC for at least 7
years experience higher value of production in the
medium term
– Unlike technologies such as fertilizer or improved
seeds, benefits realized from constructing SLWM structures may
accrue over longer time horizons.
• A mixture of strategies may reap quicker benefits
– Although soil bund, stone terraces, and check dams were
identified as the three most important conservation
measures, they may not give desired results by themselves in
the short run
– Physical SWC measures may need to be integrated with soil
fertility management (composting, cover crops, fertilizer
application etc.) and moisture management (mulching, deep
ploughing, etc)
26. Conclusions (3)
• The longer one sustains SWC, the higher the marginal
benefit of sustaining an extra year of activity.
– Well maintained SWC structures would begin to slow ongoing
degradation in the initial years of maintenance, but nutrient build-
up may take more time to show significant impact on value of
production.
• Although the marginal benefit increases with each
additional year that the structure is maintained, we
assume that these benefits may plateau at a certain
treatment level.
– As nutrient repletion and erosion control is successful, we would
expect to see diminishing returns as the necessary biophysical
components are replaced.
27. Next Steps.
Further collaboration:
• Nile Basin Development Challenge (N2 and N3)
– Development trajectories in Ethiopia: analysis of
socio-economic impacts and land cover change
• Lisa Rebello, Katherine Snyder, Fanaye Tadesse, Emily
Schmidt
• Possible second round of survey with GIZ
(June/July 2012)
Why were the richest sites not using SLM? Initial analysis – income doesn’t correlate with SLM, so wanted to look at this.Mean total expenditure (column 4 varies by site). Think of sources of income: do farmers have larger farm size, higher value of agricultural production, or higher non-agricultural production. So, lets look at production – substantial variation in production: Fogera and Jeldu has very high value of production – Fogera – lot of teff and fairly large farms, and so mean value of production is very high, and the mean value is quite a bit higher than the expenditure value that we have. Possibly unusually good harvest, and invested. Diga also high incomes – it has very large farms – largest farms in sample and growing a lot of maize. Ag incomes are very high relative to expenditures. (bumper harvest?) Further cleaning on yield data which not showing here. MisrakEstie and DegaDamot – same agro-ecological zone, rather small farm sizes and rather large shares of non-farm income.
This is an important slide!
Percentage of all households? Can we get percentage of households who actually implemented these activities?