returns to fertilizer and program efficiency: estimation techniques & results from crop...
DESCRIPTION
Presentation by Andrew Dorward (SOAS) at the National FISP Symposium, Lilongwe, MalawiTRANSCRIPT
1
Returns to fertilizer and program efficiency:
Estimation techniques & results from crop simulation
Andrew Dorward & Ephraim Chirwa,
School of Oriental and African Studies, University of LondonWadonda Consult
WACOL
Outline
Background – policy importanceReturns to fertiliser (kg grain yield per kg fertiliser
or nutrient N &/or P) are critical to the production benefits & economic returns to the programme
Therefore ….. What is the yield response?
Estimation techniques’ results & reliability
How can it be improved?Crop simulation modelling results & policy
implications2July 2014
Returns to fertiliser: yield response
3July 2014
-
0.50
1.00
1.50
2.00
2.50
-
50
100
150
200
250
300
30% 40% 50% 60% 70% 80% 90% 100% 110% 120%
US$mill
% of Simulation NUE (23.4 kg/kg hybrid, 18.0 local)
NPV (US$ mill)
BCR
Fiscal Efficiency
What is the yield response?
Difficult to obtain reliable information on smallholder yields and yield responses
Different data sources & methods with different reliability and bias
Variability across years (rainfall, population growth, changes in varieties & cropping methods)
Variation between areas & across farms (rainfall, population density, varieties, soils & cropping methods)
4July 2014
Data sources – methods & reliability
5
Farm surveys
• General random errors from farmer & enumerator errors, especially small plots
• (Omitted variables over-estimate yield response)
• Farmer estimates area
• Under-estimates of smaller plot areas & over-estimates of larger plot areas – average over-estimate
• Clustering in acre fractions• GPS area • Should be reliable• Farmer
estimates production
• Possible over- or under- estimates from farmer reporting & measurement units.
• Possible under estimates from omission of green maize & multiple respondents
• Differential bias effects on yield response?• Crop
cutting• Over estimates of yield: production & yield response affected by bias in area estimates
• Fertiliser use
• Under reporting of sales & purchases leads to under-estimates of yield response
On farm trials
• Common over-estimate of yield & yield response due to farmer selection & crop management
Crop simulations
• Model over- or under- estimates yield response
• Over estimate of yield response, no pest & disease
Yield response estimates
Study Response kg/kgN
Yield (kg/ha)
Local
Hybrid
LocalHybri
d
Chibwana et al (2010) (subs)
2008/9 12 131,31
21,510
Holden & Lunduka (2010,3)
2006,7,8
9? 14? 1,609
IHS2 – AISS2 (adjusted yield)
2004,6,8
6.6 – 13.9 682
NACAL (pure stand only) 2006/7 n.a. 928 1,803
Crop simulation (with subs)
2012/13 18 231,39
21,921
Makumba (2013) hybrid 2010/11 16.80 Fert 2,483
Fert 4,003
Kamanga (2013) CR hybrid
2003/4 29
1 weed 900
2 weed 1,300
SOAS (2008) Summary/review ‘median’
15 22
National food security: consumption,
7-1500
-1000
-500
0
500
1000
1500
2000
2500
3000
3500
400020
01/2
2002
/3
2003
/4
2004
/5
2005
/6
2006
/7
2007
/8
2008
/9
2009
/10
2010
/11
2011
/12
2012
/13
2013
/14
'000MT
Total consumption (MT)
National food security: consumption, production with subsidy
8-1500
-1000
-500
0
500
1000
1500
2000
2500
3000
3500
400020
01/2
2002
/3
2003
/4
2004
/5
2005
/6
2006
/7
2007
/8
2008
/9
2009
/10
2010
/11
2011
/12
2012
/13
2013
/14
'000MT
Total consumption (MT)
Production with subsidy (MT)
National food security: consumption, production & surplus/deficit with subsidy
9-1500
-1000
-500
0
500
1000
1500
2000
2500
3000
3500
400020
01/2
2002
/3
2003
/4
2004
/5
2005
/6
2006
/7
2007
/8
2008
/9
2009
/10
2010
/11
2011
/12
2012
/13
2013
/14
'000MT Domestic surplus (deficit) before subsidy (MT)Domestic surplus (deficit) with subsidy (MT)
Total consumption (MT)Production with subsidy (MT)
10July 2014-1500
-1000
-500
0
500
1000
1500
2000
2500
3000
3500
40002
00
1/2
20
02
/3
20
03
/4
20
04
/5
20
05
/6
20
06
/7
20
07
/8
20
08
/9
20
09
/10
20
10
/11
20
11
/12
20
12
/13
20
13
/14
'000MT Domestic surplus (deficit) before subsidy (MT)Domestic surplus (deficit) with subsidy (MT)Domestic surplus (deficit) without subsidy (MT)Total consumption (MT)Production with subsidy (MT)Production without subsidy
National food security: consumption, production & surplus/deficit without subsidy
Crop simulation
Commissioned maize simulation study under smallholder conditions (Anthony Whitbread et al, Goettingen University, now at ICRISAT)
Review & meta-analysis of response studies: limited / conflicting/ weak information Hybrid 30-80 kg grain/kg N @ 15-30 kg N/ha Local 10-30 kg grain/kg N @ 15-30 kg N/ha
APSIM model calibrated for smallholder conditions (eg weeds) historical (1927-2004) daily climate information for Kasungu 4 sandy or sandy clay loams (deep & shallow soils) Centre & South Range of management practices matching 2008 AISS2 observations
Planting date & density, N & P rates & times, good & poor weeding Used results to generate regression estimates of crop responses Evaluated crop responses using regression coefficients at observed
2012/13 crop management means
11July 2014
Crop simulation findings
Influences on yield & yield response Mean yield and fertiliser response estimates
12July 2014
Illustrative N Response, HYBRID by plant population & weeding, without & with P
13July 2014
0
1000
2000
3000
4000
5000
0 15 30 60 120
Gra
in y
ield
(kg/
ha)
Fertiliser N rate (kg/ha)
Low GoodLow PoorMod GoodMod PoorHigh GoodHigh Poor
Chitala 0PVariety: Hybrid
0
1000
2000
3000
4000
5000
0 15 30 60 120
Gra
in y
ield
(kg/
ha)
Fertiliser N rate (kg/ha)
Chitala 10PVariety: Hybrid
Source: Whitbread et al 2013
0
1000
2000
3000
4000
5000
0 15 30 60 120
Gra
in y
ield
(kg/
ha)
Fertiliser N rate (kg/ha)
Low GoodLow PoorMod GoodMod PoorHigh GoodHigh Poor
Chitala 0PVariety: Local
0
1000
2000
3000
4000
5000
0 15 30 60 120
Gra
in y
ield
(kg/
ha)
Fertiliser N rate (kg/ha)
Chitala 10PVariety: Local
Illustrative N Response, LOCAL by plant population & weeding, without & with P
14July 2014
Source: Whitbread et al 2013
Simulated yield response
Importance of hybrid seed early planting good agronomy potential for lower N application rates variable returns to N
Good potential returns to N and impact Nutrient responses with average smallholder
management Local 18 kg grain/kg N (@37 kg N/ha) Hybrid 22 kg grain/kg N (@47 kg N/ha) Hybrid without fertiliser + 600kg/ha
15July 2014
16
Returns to fertilizer and program efficiency:
Estimation techniques & results from crop simulation
Andrew Dorward & Ephraim Chirwa,
School of Oriental and African Studies, University of LondonWadonda Consult
WACOL