Gendered targeting of agricultural extension and weather variability in Africa south of the Sahara
Gendered targeting of
agricultural extension and
weather
variability in Africa south
of the Sahara
Carlo Azzarri (IFPRI)
Gianluigi Nico (FAO)
Background and motivation
• Agricultural extension and advisory services key in promoting technical change,
increasing productivity, improving livelihoods (Hulme 1989; Glendenning 2010; Benin et
al. 2011; Anaeto et al. 2012; Faure et al. 2012; OECD 2015); adapting agricultural practices,
and reducing the economic costs associated to climate and weather shocks (Asare-
Nuamah et al. 2019)
• Women’s participation in agriculture has positive effects on agriculture
production, productivity, investment in land, and additional outcomes including
poverty reduction and food security (Quisumbing et al. 1996; Quisumbing and Kumar 2014;
Doss et al. 2018)
• Gender-specific targeting and provision of extension services yield positive effects
on agricultural performance (Doss & Morris 2000; Doss 2001; World Bank 2012; Ragasa et al.
2013), although women have relatively lower access to extension and advisory
services than men (Peterman et al. 2014)
Research questions
1. Does gender-specific targeting and provision of agricultural extension
affect agricultural performance?
2. Does it mitigate the negative effects of weather variability and extreme
shocks?
Data
• LSMS-ISA household surveys: nationally-
representative, geo-referenced,
rural/urban, processed by RuLIS
Malawi (2010/11; 2013) N=1,082; 915
Nigeria (2012/13; 2015/16) N=2,740; 2,562
Tanzania (2012/13; 2014/2015) N=2,379;
2,016
Uganda (2010/11; 2011/12; 2013/2014)
N=1,939; 1,685; 1,028
for a pooled sample of 16,346 HHs
engaged in ag. production at any point
during the ag. season
• Remote sensing
Gridcell monthly time series data (1979–
2019) on rainfall and temperature from
ERA5 by the European Centre for Medium-
Range Weather Forecasts (ECMWF)
(Copernicus Climate Change Service 2020)
Econometric strategy
𝑌ℎ,𝑡,𝑐 = 𝛼0 + (𝛽1𝑝𝑟𝑒𝑐ℎ,𝑡,𝑐
𝑝𝑜𝑠
) + (𝛽2𝑝𝑟𝑒𝑐ℎ,𝑡,𝑐
𝑛𝑒𝑔
) + (𝛽3𝑡𝑒𝑚𝑝ℎ,𝑡,𝑐
𝑝𝑜𝑠
) +
(𝛽4𝑝𝑟𝑒𝑐ℎ,𝑡,𝑐
𝑙𝑡𝑎
) + (𝛽5𝑡𝑒𝑚𝑝ℎ,𝑡,𝑐
𝑙𝑡𝑎
) + 𝛽′𝑋ℎ,𝑡,𝑐 + 𝜂𝑐 + 𝛿𝑡 + (𝜀ℎ,𝑡,𝑐) estimated
through FE and RE
ℎ, 𝑡 and 𝑐 denote household, time, and country
𝑌ℎ,𝑡,𝑐: 1) agriculture income (log); 2) value of production over land cultivated (log); and 3) value of production over labor
worked (log)
𝑝𝑟𝑒𝑐ℎ,𝑡,𝑐
𝑝𝑜𝑠
and 𝑝𝑟𝑒𝑐ℎ,𝑡,𝑐
𝑛𝑒𝑔
: dummies for rainfall and drought shocks (over and under 1 SD above and below the long-term
averages)
𝑡𝑒𝑚𝑝ℎ,𝑡,𝑐
𝑝𝑜𝑠
: dummy for heat wave (over 1 SD above the long-term average)
𝑝𝑟𝑒𝑐ℎ,𝑡,𝑐
𝑙𝑡𝑎
and 𝑡𝑒𝑚𝑝ℎ,𝑡,𝑐
𝑙𝑡𝑎
: long-term rainfall and temperature averages (gridcell where household resides)
𝑋ℎ,𝑡,𝑐: socioeconomic factors associated to household farms, as controls
𝜂𝑐 and 𝛿𝑡: country- and time-specific effects; 𝜀ℎ,𝑡,𝑐 error i.i.d.
Descriptive results/1
Extension recipients show higher agricultural and crop income, value of harvest, land and labor productivity
Country
Extension/advisory
services:
Annual net agricultural
income
(PPP, 2011)
Annual crop
income (PPP, 2011)
Annual value of
crop production
(PPP, 2011)
Value of
production per
labor day
(PPP,2011)
Value of
production per
hectare of land
(PPP,2011)
Malawi
(2010-2013)
Not received 514.10*** 289.75*** 393.49*** 4.29*** 768.81***
Received 765.56*** 438.41*** 584.29*** 5.34*** 1027.80***
Overall 648.12 369.16 495.41 4.86 908.48
Nigeria
(2013-2016)
Not received 1638.89 1153.49 1665.58*** 7.5 3002.78
Received 1548.18 1163.91 1968.24*** 7.95 2907.79
Overall 1621.84 1155.49 1723.8 7.59 2984.5
Tanzania
(2013-2016)
Not received 869.04*** 618.00*** 740.23*** 5.00*** 612.18***
Received 1351.65*** 1127.92*** 1372.71*** 8.77*** 950.44***
Overall 925.41 677.64 814.2 5.44 651.6
Uganda
(2011-2014)
Not received 846.25*** 572.53*** 726.77*** 3.46*** 1061.77**
Received 1073.96*** 725.14*** 961.84*** 4.06*** 1150.23**
Overall 896.1 605.94 778.23 3.59 1081.22
* significant at 10%; ** significant at 5%; *** significant at 1%
Descriptive results/2
If women are added as recipients of extension ag. performance is consistently higher than in the
case of men-only extension targeting!
Country
Extension/advisory
services:
Annual net agricultural
income
(PPP, 2011)
Annual crop
income (PPP, 2011)
Annual value of crop
production
(PPP, 2011)
Value of production
per labor day
(PPP,2011)
Value of production
per hectare of land
(PPP,2011)
Malawi
(2010-2013)
Not received 514.10*** 289.75*** 393.49*** 4.29*** 768.81***
Only males 728.76*** 393.5 534.10* 5.40*** 923.37
Only females 604.62 370.1 488.45 4.82 969.85
Jointly received 945.27*** 545.11*** 720.68*** 5.74*** 1187.24***
Total 648.12 369.16 495.41 4.86 908.48
Nigeria
(2013-2016)
Not received 1638.89 1153.49 1665.58*** 7.5 3002.78
Only males 1480.84** 1216.3 2052.93*** 8.07* 2955.53
Only females 1301.61 528.87*** 923.53*** 6.4 3018.19
Jointly received 1925.94** 1162.57 1972.68** 7.98 2656.93
Total 1621.84 1155.49 1723.8 7.59 2984.5
Tanzania
(2013-2016)
Not received 869.04*** 618.00*** 740.23*** 5.00*** 612.18***
Only males 1473.27*** 1225.61*** 1434.23*** 8.87*** 879.94***
Only females 632.71** 499.09 765.09 6.63* 1044.57***
Jointly received 1652.03*** 1418.89*** 1760.75*** 10.51*** 1090.45***
Total 925.41 677.64 814.2 5.44 651.6
Uganda
(2010-2014)
Not received 846.25*** 572.53*** 726.77*** 3.46*** 1061.77**
Only males 1087.23*** 783.66*** 1020.57*** 4.10*** 1126.63
Only females 987.66* 686.00** 869.19** 3.97** 1206.25**
Jointly received 1140.10*** 709.22*** 993.94*** 4.11*** 1119.98
Total 896.1 605.94 778.23 3.59 1081.22
* significant at 10%; ** significant at 5%; *** significant at 1%
Descriptive results/3
Targeting also the opposite sex when farms are managed just by one-sex household member is associated to
higher ag. output
Country Extension/advisory services:
Annual net agricultural
income
(PPP, 2011)
Annual crop
income (PPP,
2011)
Annual value of
crop production
(PPP, 2011)
Value of
production per
labor day
(PPP,2011)
Value of
production per
hectare of land
(PPP,2011)
Malawi
(2010-2013)
Not received 514.10*** 289.75*** 393.49*** 4.29*** 768.81***
Received by males and land managed by males 720.70** 386.59 538.95* 5.58*** 913.48
Received by females and land managed by females/jointly 678.06 373.9 490.37 5.09 998.63*
Received by males and managed by female/jointly or
862.28*** 524.32*** 685.87*** 5.34*** 1136.75***
Received by females/jointly and managed by males
Total 648.12 369.16 495.41 4.86 908.48
Nigeria
(2013-2016)
Not received 1638.89 1153.49 1665.58*** 7.5 3002.78
Received by males and land managed by males 1466.16** 1194.06 2027.66*** 8.23** 2968.86
Received by females and land managed by females/jointly 1264.61* 500.97*** 1313.25** 5.46** 2372.45**
Received by males and managed by female/jointly or
2047.60*** 1412.02** 2090.67*** 8.21 2956.06
Received by females/jointly and managed by males
Total 1621.84 1155.49 1723.8 7.59 2984.5
Tanzania
(2010-2014)
Not received 869.04*** 618.00*** 740.23*** 5.00*** 612.18***
Received by males and land managed by males 1283.52*** 1029.86*** 1210.72*** 6.98** 808.20**
Received by females and land managed by females/jointly 1130.34** 960.46*** 1266.23*** 8.57*** 1082.21***
Received by males and managed by female/jointly or
1582.80*** 1329.06*** 1556.84*** 9.95*** 917.49***
Received by females/jointly and managed by males
Total 925.41 677.64 814.2 5.44 651.6
Uganda
(2011-2014)
Not received 846.25*** 572.53*** 726.77*** 3.46*** 1061.77**
Received by males and land managed by males 1217.52*** 790.62*** 1099.23*** 4.33** 1004.48
Received by females and land managed by females/jointly 1083.42*** 695.81*** 929.58*** 4.08*** 1184.73**
Received by males and managed by female/jointly or
1011.32** 764.00*** 984.69*** 3.94** 1125.86
Received by females/jointly and managed by males
Total 896.1 605.94 778.23 3.59 1081.22
VARIABLE
Agricultural
income
Value of
production
(per hectare)
Value of
production
(per labour
day)
Specification 1
ES Received by males and land managed by
males; or ES Received by females and land
managed by females 0.0648** 0.0790*** 0.598***
(0.0284) (0.0295) (0.150)
ES Received by males and females and land
managed by males and females 0.156*** 0.0651 0.732***
(0.0432) (0.0453) (0.230)
Rainfall long term average (mm) -0.00172*** -0.00185***
-
0.0199***
(0.000236) (0.000235) (0.00121)
Temperature long term average (C) -0.0915*** -0.144*** -0.682***
(0.00639) (0.00671) (0.0340)
Econometric results -baseline
Beneficial effects of extension services on
agricultural income and profitability per
unit of land and labor…
Both more humid and hotter areas are
negatively correlated with agricultural
performance
…especially when both sexes are
involved in management and targeting
VARIABLES
Agricultural
income
Value of
production
(per hectare)
Value of
production
(per labour
day)
Specification 2
ES Received by males and land managed by males; or ES Received by
females and land managed by females 0.0673** 0.0797*** 0.633***
(0.0284) (0.0295) (0.149)
ES Received by males and females and land managed by males and females 0.152*** 0.0638 0.726***
(0.0432) (0.0453) (0.230)
Rainfall long term average (mm) -0.00186*** -0.00192*** -0.0192***
(0.000256) (0.000259) (0.00133)
Temperature long term average (C) -0.0893*** -0.144*** -0.654***
(0.00643) (0.00677) (0.0343)
Rainfall flood shock dummy (1 sd) -0.0909*** -0.00785 -0.931***
(0.0248) (0.0260) (0.132)
Rainfall drought shock dummy (1 sd) 0.143*** 0.0307 0.712***
(0.0266) (0.0277) (0.140)
Temperature heat shock dummy (1 sd) -0.100*** -0.0892*** -0.186
(0.0269) (0.0286) (0.145)
Econometric results –with weather shocks
Overall, flood and heat waves
negatively affect ag.
performance; drought has a
surprisingly positive effect
Econometric results –with interaction terms
VARIABLE
Agricultural
income
Value of
production (per
hectare)
Value of
production (per
labour day)
Specification 3
ES Received by males and land managed by males; or ES Received by females and land
managed by females 0.104* 0.182*** 0.965***
(0.0598) (0.0630) (0.319)
ES Received by males and females and land managed by males and females 0.211** 0.133 1.947***
(0.0855) (0.0928) (0.470)
Rainfall long term average (mm) -0.00184*** -0.00197*** -0.0189***
(0.000258) (0.000260) (0.00133)
Temperature long term average (C) -0.0895*** -0.144*** -0.660***
(0.00644) (0.00678) (0.0343)
Rainfall flood shock dummy (1 sd) -0.107*** 0.0114 -1.034***
(0.0274) (0.0288) (0.146)
Rainfall drought shock dummy (1 sd) 0.166*** 0.0323 0.964***
(0.0284) (0.0295) (0.149)
Temperature heat shock dummy (1 sd) -0.0950*** -0.0713** -0.149
(0.0293) (0.0312) (0.158)
ES received by females only or by males only#Rainfall flood shock dummy (1 sd) 0.0361 -0.110* 0.470
(0.0578) (0.0600) (0.303)
ES jointly received#Rainfall flood shock dummy 0.191** 0.0205 0.150
(0.0901) (0.0940) (0.476)
ES received by females only or by males onlyc#Rainfall drought shock dummy -0.0604 -0.0339 -1.380***
(0.0576) (0.0599) (0.303)
ES jointly received#Rainfall drought shock dummy -0.261*** 0.0340 -1.244***
(0.0908) (0.0944) (0.478)
ES received by females only or by males only#Temperature heat shock dummy -0.0414 -0.0387 0.0324
(0.0585) (0.0616) (0.312)
ES jointly received#Temperature heat shock dummy -0.0308 -0.140 -0.999**
(0.0877) (0.0936) (0.474)
Observations 15,060 16,346 16,361
Beneficial effects of extension services on
agricultural income and profitability per
unit of land and labor if both sexes are
involved
Gendered effect in case of a flood
shock: when females contribute to farm
decision making (and are also extension
beneficiaries) the negative impact of
flood on agricultural income is
mitigated by 19%
Overall, flood negatively affects ag.
performance; drought has a
surprisingly positive effect
Opposite effect in case of drought
Conclusions
• The study takes advantage of high-resolution spatial data and panel data for four countries
in SSA
• Novel focus on gendered targeting effects of extension services, combined with weather
shocks
• If females are also farm managers (and targeted by extension) ag. performance is higher
relative to the traditional case when only males are farm managers (and beneficiaries of
extension)
• Securing women with either access to extension/advisory or control over land significantly
increases agricultural income by nearly 16%.
• Including also females among beneficiaries is essential in mitigating the negative effects of
flooding associated to devastating effects in most sub-Saharan African countries