This document discusses using crop modeling software (DSSAT) to simulate maize yield patterns in Zambia during an El Niño event in 2015-2016 that caused delayed rainfall. The study found that modeling supply shocks from climate events can help estimate their impacts on household incomes and food security to better guide policy responses. Policymakers often focus on ensuring food availability through trade barriers or stockpiling, but availability alone does not address the access problems and income shocks that small farmers face from poor harvests.

1. Poverty impacts of maize export
bans in Zambia
Jawoo Koo, Abdullah Mamun & Will Martin
IFPRI
22 September 2021

2. Policy Context
▪ Many countries worry about supply shocks to food staples
oBut policy makers often continue to focus on availability
▪ Often use trade barriers to ensure availability
oThough Amartya Sen pointed out this may exacerbate access problems
▪ We now have household & geospatial data to assess impacts
oCan we use this to better guide policies?

3. ▪ Supply shocks often major source of income shocks
▪ Modeling supply shocks using DSSAT
▪ Price changes may compound or offset income effects of supply shocks
▪ Policy needs to focus on dealing with income losses & food insecurity
Issues

4. Supply shocks and income shocks

5. Supply shocks often large for staples
▪ Standard deviation of maize yield changes in Zambia 22%
ovs 7% for global maize yields
oAnd 29% in Malawi
oShocks larger at farm level
▪ Income effects determined by output
oRather than net sales for price changes
oBut how diversified are households?
▪ Climate change likely exacerbating these shocks

6. Focus on availability
▪ Policy makers frequently focus on food availability
▪ A tempting response is to restrict exports
oMake sure the food doesn’t “get away”
▪ Or, especially in countries that are net importers, make state purchases
oTo ensure availability
oAccompanied with export restrictions
▪ Often, availability data go down to localities

7. What matters is income effects of shocks
▪ If producers are large corporations, probably no need for policy intervention
▪ But if small, poor, specialized producers hit, the resulting income shocks
may be drastic
▪ Farm households often diversified
o Need to know who produces what
o And impacts on household incomes

8. Zambia an interesting case study
Household type % of popn Pov rate %
Overall 100.0 64.4
Urban 41.9 39.8
Rural 58.1 82.1
Mostly farm Inc 43.1 86.3

9. 10% Maize Yield Fall
Hhold type
Del poverty rate
% pts
Overall 1.0
Urban 0.5
Rural 1.3
Mostly farm income 1.8

10. Zambia supply shock recap
▪ Extremely high poverty rate, especially among farmers
▪ Poverty rises when maize yields fall
▪ Large maize yield declines increase poverty among farmers sharply

11. Modeling Supply Shocks using DSSAT

12. Delayed rainfall onset
July – December 2015 July 2015 – February 2016

13. Model-estimated maize yield pattern
Filling the data gap
The DSSAT maize model was calibrated to simulate
three previous non-El Niño seasons and used to
estimate yield under 2015/16 El Niño.
oGrid-based weather data from NASA-POWER
oMaize geography based on MapSPAM
oSubnational production statistics data
oAdjusted planting dates to capture the delayed
onset of rainfall during the 2015/2016 season.

14. DSSAT
Decision Support System
for Agrotechnology Transfer
▪ Research tool for
crop production analyses
▪ Incorporates
o Crop-Soil-Weather-Management
models
o Utilities to help users integrate
data with models
o Data: Weather, Soil,
Experiments
o Analysis: Evaluation,
Risk/Uncertainty, Economics
o Support: Graphics, Weather
Generator, Parameter
Estimator

15. MANAGEMENT
• Planting window
• Planting density
• Irrigation
• Inorganic fertilizer
• Organic manure
• Tillage
• Residue
CULTIVAR
• Phenology
• Max # of kernels
• Kernel filling rate
*DSSAT Cropping System Model Ver. 4.0.2.000 May 21, 2009; 16:32:33
*RUN 1 : RAINFED LOW NITROGEN
MODEL : MZCER040 - MAIZE
EXPERIMENT : UFGA8201 MZ NIT X IRR, GAINESVILLE 2N*3I
TREATMENT 1 : RAINFED LOW NITROGEN
CROP : MAIZE CULTIVAR : McCurdy 84aa ECOTYPE :IB0002
STARTING DATE : FEB 25 1982
PLANTING DATE : FEB 26 1982 PLANTS/m2 : 7.2 ROW SPACING : 61.cm
WEATHER : UFGA 1982
SOIL : IBMZ910014 TEXTURE : - Millhopper Fine Sand
SOIL INITIAL C : DEPTH:180cm EXTR. H2O:160.9mm NO3: 2.5kg/ha NH4: 12.9kg/ha
WATER BALANCE : IRRIGATE ON REPORTED DATE(S)
IRRIGATION : 13 mm IN 1 APPLICATIONS
NITROGEN BAL. : SOIL-N & N-UPTAKE SIMULATION; NO N-FIXATION
N-FERTILIZER : 116 kg/ha IN 3 APPLICATIONS
RESIDUE/MANURE : INITIAL : 1000 kg/ha ; 0 kg/ha IN 0 APPLICATIONS
ENVIRONM. OPT. : DAYL= 0.00 SRAD= 0.00 TMAX= 0.00 TMIN= 0.00
RAIN= 0.00 CO2 = R330.00 DEW = 0.00 WIND= 0.00
SIMULATION OPT : WATER :Y NITROGEN:Y N-FIX:N PHOSPH :N PESTS :N
PHOTO :C ET :R INFIL:S HYDROL :R SOM :G
MANAGEMENT OPT : PLANTING:R IRRIG :R FERT :R RESIDUE:N HARVEST:M WTH:M
*SUMMARY OF SOIL AND GENETIC INPUT PARAMETERS
SOIL LOWER UPPER SAT EXTR INIT ROOT BULK pH NO3 NH4 ORG
DEPTH LIMIT LIMIT SW SW SW DIST DENS C
cm cm3/cm3 cm3/cm3 cm3/cm3 g/cm3 ugN/g ugN/g %
-------------------------------------------------------------------------------
0- 5 0.026 0.096 0.230 0.070 0.086 1.00 1.30 7.00 0.10 0.50 2.00
5- 15 0.025 0.086 0.230 0.061 0.086 1.00 1.30 7.00 0.10 0.50 1.00
15- 30 0.025 0.086 0.230 0.061 0.086 0.70 1.40 7.00 0.10 0.50 1.00
30- 45 0.025 0.086 0.230 0.061 0.086 0.30 1.40 7.00 0.10 0.50 0.50
45- 60 0.025 0.086 0.230 0.061 0.086 0.30 1.40 7.00 0.10 0.50 0.50
60- 90 0.028 0.090 0.230 0.062 0.076 0.05 1.45 7.00 0.10 0.60 0.10
90-120 0.028 0.090 0.230 0.062 0.076 0.03 1.45 7.00 0.10 0.50 0.10
120-150 0.029 0.130 0.230 0.101 0.130 0.00 1.45 7.00 0.10 0.50 0.04
150-180 0.070 0.258 0.360 0.188 0.258 0.00 1.20 7.00 0.10 0.50 0.24
TOT-180 6.2 22.2 45.3 16.1 21.4 <--cm - kg/ha--> 2.5 12.9 87080
SOIL ALBEDO : 0.18 EVAPORATION LIMIT : 2.00 MIN. FACTOR : 1.00
RUNOFF CURVE # :60.00 DRAINAGE RATE : 0.65 FERT. FACTOR : 0.80
MAIZE CULTIVAR :IB0035-McCurdy 84aa ECOTYPE :IB0002
P1 : 265.00 P2 : 0.3000 P5 : 920.00
G2 : 990.00 G3 : 8.500 PHINT : 39.000
*SIMULATED CROP AND SOIL STATUS AT MAIN DEVELOPMENT STAGES
RUN NO. 1 RAINFED LOW NITROGEN
CROP GROWTH BIOMASS CROP N STRESS
DATE AGE STAGE kg/ha LAI kg/ha % H2O N
------ --- ---------- ----- ----- --- --- ---- ----
25 FEB 0 Start Sim 0 0.00 0 0.0 0.00 0.00
26 FEB 0 Sowing 0 0.00 0 0.0 0.00 0.00
27 FEB 1 Germinate 0 0.00 0 0.0 0.00 0.00
9 MAR 11 Emergence 29 0.00 1 4.4 0.00 0.00
27 MAR 29 End Juveni 251 0.43 4 1.6 0.00 0.09
1 APR 34 Floral Ini 304 0.44 4 1.5 0.00 0.50
*DSSAT Cropping System Model Ver. 4.0.2.000 May 21, 2009; 16:32:33
*RUN 1 : RAINFED LOW NITROGEN
MODEL : MZCER040 - MAIZE
EXPERIMENT : UFGA8201 MZ NIT X IRR, GAINESVILLE 2N*3I
TREATMENT 1 : RAINFED LOW NITROGEN
CROP : MAIZE CULTIVAR : McCurdy 84aa ECOTYPE :IB0002
STARTING DATE : FEB 25 1982
PLANTING DATE : FEB 26 1982 PLANTS/m2 : 7.2 ROW SPACING : 61.cm
WEATHER : UFGA 1982
SOIL : IBMZ910014 TEXTURE : - Millhopper Fine Sand
SOIL INITIAL C : DEPTH:180cm EXTR. H2O:160.9mm NO3: 2.5kg/ha NH4: 12.9kg/ha
WATER BALANCE : IRRIGATE ON REPORTED DATE(S)
IRRIGATION : 13 mm IN 1 APPLICATIONS
NITROGEN BAL. : SOIL-N & N-UPTAKE SIMULATION; NO N-FIXATION
N-FERTILIZER : 116 kg/ha IN 3 APPLICATIONS
RESIDUE/MANURE : INITIAL : 1000 kg/ha ; 0 kg/ha IN 0 APPLICATIONS
ENVIRONM. OPT. : DAYL= 0.00 SRAD= 0.00 TMAX= 0.00 TMIN= 0.00
RAIN= 0.00 CO2 = R330.00 DEW = 0.00 WIND= 0.00
SIMULATION OPT : WATER :Y NITROGEN:Y N-FIX:N PHOSPH :N PESTS :N
PHOTO :C ET :R INFIL:S HYDROL :R SOM :G
MANAGEMENT OPT : PLANTING:R IRRIG :R FERT :R RESIDUE:N HARVEST:M WTH:M
*SUMMARY OF SOIL AND GENETIC INPUT PARAMETERS
SOIL LOWER UPPER SAT EXTR INIT ROOT BULK pH NO3 NH4 ORG
DEPTH LIMIT LIMIT SW SW SW DIST DENS C
cm cm3/cm3 cm3/cm3 cm3/cm3 g/cm3 ugN/g ugN/g %
-------------------------------------------------------------------------------
0- 5 0.026 0.096 0.230 0.070 0.086 1.00 1.30 7.00 0.10 0.50 2.00
5- 15 0.025 0.086 0.230 0.061 0.086 1.00 1.30 7.00 0.10 0.50 1.00
15- 30 0.025 0.086 0.230 0.061 0.086 0.70 1.40 7.00 0.10 0.50 1.00
30- 45 0.025 0.086 0.230 0.061 0.086 0.30 1.40 7.00 0.10 0.50 0.50
45- 60 0.025 0.086 0.230 0.061 0.086 0.30 1.40 7.00 0.10 0.50 0.50
60- 90 0.028 0.090 0.230 0.062 0.076 0.05 1.45 7.00 0.10 0.60 0.10
90-120 0.028 0.090 0.230 0.062 0.076 0.03 1.45 7.00 0.10 0.50 0.10
120-150 0.029 0.130 0.230 0.101 0.130 0.00 1.45 7.00 0.10 0.50 0.04
150-180 0.070 0.258 0.360 0.188 0.258 0.00 1.20 7.00 0.10 0.50 0.24
TOT-180 6.2 22.2 45.3 16.1 21.4 <--cm - kg/ha--> 2.5 12.9 87080
SOIL ALBEDO : 0.18 EVAPORATION LIMIT : 2.00 MIN. FACTOR : 1.00
RUNOFF CURVE # :60.00 DRAINAGE RATE : 0.65 FERT. FACTOR : 0.80
MAIZE CULTIVAR :IB0035-McCurdy 84aa ECOTYPE :IB0002
P1 : 265.00 P2 : 0.3000 P5 : 920.00
G2 : 990.00 G3 : 8.500 PHINT : 39.000
*SIMULATED CROP AND SOIL STATUS AT MAIN DEVELOPMENT STAGES
RUN NO. 1 RAINFED LOW NITROGEN
CROP GROWTH BIOMASS CROP N STRESS
DATE AGE STAGE kg/ha LAI kg/ha % H2O N
------ --- ---------- ----- ----- --- --- ---- ----
25 FEB 0 Start Sim 0 0.00 0 0.0 0.00 0.00
26 FEB 0 Sowing 0 0.00 0 0.0 0.00 0.00
27 FEB 1 Germinate 0 0.00 0 0.0 0.00 0.00
9 MAR 11 Emergence 29 0.00 1 4.4 0.00 0.00
27 MAR 29 End Juveni 251 0.43 4 1.6 0.00 0.09
1 APR 34 Floral Ini 304 0.44 4 1.5 0.00 0.50
*DSSAT Cropping System Model Ver. 4.0.2.000 May 21, 2009; 16:32:33
*RUN 1 : RAINFED LOW NITROGEN
MODEL : MZCER040 - MAIZE
EXPERIMENT : UFGA8201 MZ NIT X IRR, GAINESVILLE 2N*3I
TREATMENT 1 : RAINFED LOW NITROGEN
CROP : MAIZE CULTIVAR : McCurdy 84aa ECOTYPE :IB0002
STARTING DATE : FEB 25 1982
PLANTING DATE : FEB 26 1982 PLANTS/m2 : 7.2 ROW SPACING : 61.cm
WEATHER : UFGA 1982
SOIL : IBMZ910014 TEXTURE : - Millhopper Fine Sand
SOIL INITIAL C : DEPTH:180cm EXTR. H2O:160.9mm NO3: 2.5kg/ha NH4: 12.9kg/ha
WATER BALANCE : IRRIGATE ON REPORTED DATE(S)
IRRIGATION : 13 mm IN 1 APPLICATIONS
NITROGEN BAL. : SOIL-N & N-UPTAKE SIMULATION; NO N-FIXATION
N-FERTILIZER : 116 kg/ha IN 3 APPLICATIONS
RESIDUE/MANURE : INITIAL : 1000 kg/ha ; 0 kg/ha IN 0 APPLICATIONS
ENVIRONM. OPT. : DAYL= 0.00 SRAD= 0.00 TMAX= 0.00 TMIN= 0.00
RAIN= 0.00 CO2 = R330.00 DEW = 0.00 WIND= 0.00
SIMULATION OPT : WATER :Y NITROGEN:Y N-FIX:N PHOSPH :N PESTS :N
PHOTO :C ET :R INFIL:S HYDROL :R SOM :G
MANAGEMENT OPT : PLANTING:R IRRIG :R FERT :R RESIDUE:N HARVEST:M WTH:M
*SUMMARY OF SOIL AND GENETIC INPUT PARAMETERS
SOIL LOWER UPPER SAT EXTR INIT ROOT BULK pH NO3 NH4 ORG
DEPTH LIMIT LIMIT SW SW SW DIST DENS C
cm cm3/cm3 cm3/cm3 cm3/cm3 g/cm3 ugN/g ugN/g %
-------------------------------------------------------------------------------
0- 5 0.026 0.096 0.230 0.070 0.086 1.00 1.30 7.00 0.10 0.50 2.00
5- 15 0.025 0.086 0.230 0.061 0.086 1.00 1.30 7.00 0.10 0.50 1.00
15- 30 0.025 0.086 0.230 0.061 0.086 0.70 1.40 7.00 0.10 0.50 1.00
30- 45 0.025 0.086 0.230 0.061 0.086 0.30 1.40 7.00 0.10 0.50 0.50
45- 60 0.025 0.086 0.230 0.061 0.086 0.30 1.40 7.00 0.10 0.50 0.50
60- 90 0.028 0.090 0.230 0.062 0.076 0.05 1.45 7.00 0.10 0.60 0.10
90-120 0.028 0.090 0.230 0.062 0.076 0.03 1.45 7.00 0.10 0.50 0.10
120-150 0.029 0.130 0.230 0.101 0.130 0.00 1.45 7.00 0.10 0.50 0.04
150-180 0.070 0.258 0.360 0.188 0.258 0.00 1.20 7.00 0.10 0.50 0.24
TOT-180 6.2 22.2 45.3 16.1 21.4 <--cm - kg/ha--> 2.5 12.9 87080
SOIL ALBEDO : 0.18 EVAPORATION LIMIT : 2.00 MIN. FACTOR : 1.00
RUNOFF CURVE # :60.00 DRAINAGE RATE : 0.65 FERT. FACTOR : 0.80
MAIZE CULTIVAR :IB0035-McCurdy 84aa ECOTYPE :IB0002
P1 : 265.00 P2 : 0.3000 P5 : 920.00
G2 : 990.00 G3 : 8.500 PHINT : 39.000
*SIMULATED CROP AND SOIL STATUS AT MAIN DEVELOPMENT STAGES
RUN NO. 1 RAINFED LOW NITROGEN
CROP GROWTH BIOMASS CROP N STRESS
DATE AGE STAGE kg/ha LAI kg/ha % H2O N
------ --- ---------- ----- ----- --- --- ---- ----
25 FEB 0 Start Sim 0 0.00 0 0.0 0.00 0.00
26 FEB 0 Sowing 0 0.00 0 0.0 0.00 0.00
27 FEB 1 Germinate 0 0.00 0 0.0 0.00 0.00
9 MAR 11 Emergence 29 0.00 1 4.4 0.00 0.00
27 MAR 29 End Juveni 251 0.43 4 1.6 0.00 0.09
1 APR 34 Floral Ini 304 0.44 4 1.5 0.00 0.50
OUTPUT
Phenology
flowering, grain/seed/tuber,
maturity
Yield component
grain/seed/tuber, biomass, LAI
Growth
grain/seed/tuber, biomass, LAI
Soil
nitrogen balance, water
balance,
carbon balance
0
1
2
3
4
5
6
7
8
9
10
0 50 100 150 200
Yield
(t/ha)
Fertilizer (kg[N]/ha)

16. Impact of El Niño on Yields

17. Price Shocks and Poverty

18. Capturing food price impacts on welfare
▪ Consider welfare of a household as a function of prices and wages
▪ 𝐵 = 𝜋 𝒑, 𝑤 − 𝑒 𝒑, 𝑤, 𝑢 = z 𝒑, 𝑤, 𝑢
o𝜋 𝒑, 𝑤 represents profits from household firm(s)
o𝑒 𝒑, 𝑤, 𝑢 a “full” cost function representing the cost of expenditure less
wage earnings
oRepresent the behavior of the household as consumer & factor supplier

19. dB = (𝜋𝑝 −𝑒𝑝)∆𝑝
+ (𝑒𝑤− 𝜋𝑤)∆𝑤
1st order impacts of change in p
Net sales*
Price change
Net Labor Sales*
Wage change

20. Focus on direct price impacts
▪ Wage changes tend to be somewhat sticky
▪ Hired farm labor relatively minor in rural Zambia
▪ Deaton net sales criterion

21. Maize price decline of 10%
Del Poverty
Rate
% pts
Overall 0.75
Urban 0.17
Rural 1.16
Male-headed 0.94
Female-headed 0.56
Mostly farm income 1.66

22. Net sellers dominate
▪ Lower prices raise poverty
oEven in urban areas
▪ Effects stronger for rural areas
▪ And for male-headed households
▪ Strongest for farm households
oWho lose most from yield declines

23. Policy Responses to 2015-16 El Niño

24. The 2015-16 El Niño severe
▪ During the growing season, governments became very
concerned by hot, dry weather
▪ Zambian government introduced a precautionary export
ban
oOther countries very adversely affected so large loss of
export opportunities
▪ Late rains raised average yields
oPrimarily in the north
oBut the export ban stayed in place

25. Impacts weather shock on rural hholds
% pt change in poverty rate
Household type Net change
Into
poverty
Out of
poverty
All rural farm 1.4 3.3 1.9
Male-headed 1.7 3.5 1.8
Female-headed 1.1 3.0 1.9
Most income farm 1.4 3.4 1.9

26. Price protection in Zambia
▪ Maize trade policy since 2001 characterized as ad hoc and has seen
frequent use of export bans and export permits
▪ Updated nominal rate of protection (NRP) for Maize (1961-2017), we see
negative protection in all years, except 2005 and 2008.
▪ The average rate of protection was -39 percent over full period and -31
percent over 2005 to 2017

27. Price protection in Zambia
-100
-80
-60
-40
-20
0
20
40

28. Price protection in Zambia
▪ Trade and storage intervention didn’t stabilize domestic prices relative to
external prices over 2005-2017
▪ With NRP of -31 percent, domestic price is only 69 percent of the export price
▪ As Zambia has been a net exporter in recent years, the depression of domestic
prices has been the result of export restrictions
oCountries like Malawi sometimes need to subsidize imports
▪ This “cheap food” policy creates a serious risk to food availability

29. Export ban depressed prices by 24%
Household type Change in poverty rate % pts
Net
change
Into
poverty
Out of
poverty
Overall 1.36 1.63 0.26
Urban 0.17 0.67 0.50
Rural 2.23 2.31 0.09
Most income farming 2.42 2.48 0.06

30. Impacts of shock & response
▪ El Nino had little effect on overall yields
o But sharply reduced yield in the southern belt
o Pushing vulnerable farm households into poverty
o But yields increased in the north
▪ Policy response didn’t solve income problem
o Lowering domestic prices substantially raised poverty
o Some households adversely impacted by both shocks

31. Need better policy responses
▪ Cheap food policies create food availability challenges in most years
▪ Availability-based policy can exacerbate the income effects of weather shocks
o Export bans hurt net sellers– many poor net sellers in Zambia
▪ “No regrets” policies such as adaptive social safety nets & risk insurance
needed to deal with the income effects of supply shocks
o DSSAT modeling can help identify those who lose & target beneficiaries

32. Based on
Koo, J., Mamun, A. and Martin, W. (2021), ‘From Bad to Worse: Poverty
Impacts of Food Availability Responses to Weather Shocks’ Agricultural
Economics, https://doi.org/10.1111/agec.12657
Abdullah Al Mamun, Antony Chapoto, Brian Chisanga, Stephen
D’Alessandro, Jawoo Koo, Will Martin and Paul Samboko (2018),
‘Assessment of the Impacts of El Niño and Grain Trade Policy
Responses in East and Southern Africa to the 2015–16 Event’ IFPRI,
IAPRI and World Bank
oIncludes an analysis of Malawi