Peterman et al gender differences in non land assetsPresentation Transcript
A review of empirical evidence on gender differences in non-land agricultural inputs, technology and services in developing countries Amber Peterman Julia Behrman Agnes Quisumbing Poverty, Hunger and Nutrition Division (PHND), IFPRI FAO SOFA Writers Workshop – September 16, 2009
… for implementation, effectiveness and evaluation of agricultural programs.
“ Failure to recognize the roles, differences and inequities [between men and women] poses a serious threat to the effectiveness of the agricultural development agenda ”
~ Gender and Agricultural Sourcebook (World Bank, FAO, IFAD 2009 [p2]).
Despite this consensus, still mixed evidence on:
Gender differences in, magnitudes and effects of agricultural inputs.
Generally research has focused on land.
Dated and regional specific studies.
In response to these gaps, this review:
Focuses strictly on empirical household or plot-level data analyzed in program evaluations, agricultural or socio-economic research.
Sufficient sample sizes, attention to measurement, econometric evaluation techniques.
Recent papers in the last 10 years (1999 to 2009).
Inclusion of both published and gray literature (forthcoming, technical reports).
Four key areas: 1) technological resources, 2) natural resources, 3) human resources and 4) social and political capital.
Attempt to make contrasts and comparisons between regions to identify how women farmers face similar or different constraints (Asia, SSA, Middle East and Latin/South America).
Some notes on measurement and organization
Starting point: seminal research on gender and agriculture, followed by online database searches, google scholar, website searches of agriculture organizations, emails and inquires to researchers in the field.
Include studies which provide mean values (descriptive statistics) as well as those which analyze use as an outcome (and use gender as an explanatory variable).
Generalizations and specifics assigned to terms women’s ‘use’, ‘access’, ‘adoption’, and ‘participation.’
Collect key information from each study to input to a “matrix of findings.”
Example of summary matrix Page authors (year) country(crop) N Input type Stats on use Outcome Gender indicator Effect size Other comments Pub?
1. Technological resources
Advancements in technological resources provide means to improve soil fertility, increase land productivity and overall crop yields.
Marginal benefit may be especially significant for women farmers who are more likely to be asset poor.
Inorganic fertilizers (including
Improved seed varieties/seeds
Doss and Morris (2001) examines fertilizer and seed adoption among 420 Maize farmers in Ghana.
Women farmers have lower mean inputs for both fertilizer and modern seed varieties.
Uses two stage probit models no significant differences in adoption after controlling for complementary inputs.
Sensitivity analysis on gender (female farmers within female headed versus male headed households).
Horrell and Krishnan (2007) examine fertilizer, seed and machinery among 300 farmers in Zimbabwe (primarily maize).
Female headed households have lower mean values for all inputs.
Distinguishes between de jure and de facto (widowed) female heads.
Using tobit models finds no significant differences between male and female headed households in usage, intensity of usage or productivity once controlling for other inputs.
Further analysis suggests de facto female headed households receive lower prices for output and have lack of access to selling/marketing consortia.
1. Key studies in technological resources Page
1. Summary of results for technological resources Page
21 studies reviewed (15 published in peer reviewed journals).
19 measures of inorganic fertilizer use, 11 seed varieties, 9 machinery/tools and 5 pesticide/insecticide.
Descriptive statistics (24 indicators)
19 (79 percent) find men have higher mean inputs
5 (21 percent) find women have higher mean inputs
Bivariate or multivariate analysis (34 indicators)
19 (56 percent) find gender is insignificant
14 (41 percent) find men have significantly higher inputs
1 (3 percent) find women have significantly higher inputs
2. Natural resources
Importance of natural resources increasing owing to growing concern with increasing population pressure and stress on environmental resources.
Women may have lower endowments of natural resources ; these are not always visible or easily measurable.
Natural soil improvement
(manure/composting, fallow periods,
alley/hedgerow cropping, intercropping).
2. Water (for agricultural use, irrigation).
Jagger and Pender (2003) examines adoption of natural resource management techniques (manure, crop residue and mulching) among 451 Ugandan hhlds.
Impact of programs on NRM technology adoption using two stage probit models.
Female headship insignificant in adoption across techniques.
Number of males in household significantly associated with adoption of crop residues and manure.
Pender and Gebremedhin (2006) examine manure/composting and burning to prepare fields among 500 hhlds in Ethiopia.
Uses probit models.
Female headed households (22 percent of the sample) less likely to use manure/composting, equally as likely to burn fields.
2. Key studies in natural resources Page
12 studies reviewed (9 published in peer reviewed journals).
12 measures of soil fertility, 3 water.
Descriptive statistics (11 indicators)
8 (72 percent) find men have higher mean inputs
3 (27 percent) find women have higher mean inputs
Bivariate or multivariate analysis (12 indicators)
9 (75 percent) find gender is insignificant
3 (25 percent) find men have significantly higher inputs
0 (0 percent) find women have significantly higher inputs
2. Summary of results for natural resources Page
Human capital endowments and investments supremely important and broad topic, with spillover effects to other social and economic sectors.
(own labor, hired labor).
Extension and agricultural knowledge
(advisory services, farmer field
schools, trainings etc.).
3. Lifecycle challenges
(marriage, reproductive health,
3. Human resources Page
3. Key studies for human resources
Davis et al. (2009) examines field farmer school (FFS) participation in Kenya, Tanzania and Uganda among 267 – 300 farmers.
FFS participation equally accessible for male and female headed farmers in Kenya and Tanzania.
Female headed households in Uganda less likely to participate, due to lack of time, distance and information about FFS.
Results suggest FFS have a higher impact on productivity, crop and livestock income for female headed households as compared to male headed households.
IFPRI Gender and Governance team (2009) examine extension services in Ethiopia, Ghana and India among 676 – 1753 households.
Large mean differences in contact with extension services (e.g. in India 1 percent versus 27 percent).
However, multivariate analysis indicates these differences are largely accounted for by background factors (regional variation and assets).
3. Summary of results for human resources
16 studies reviewed (12 published in peer reviewed journals).
14 measures of extension, 11 labor, 1 lifecycle.
Descriptive statistics (25 indicators)
14 (56 percent) find men have higher mean inputs
11 (44 percent) find women have higher mean inputs
Bivariate or multivariate analysis (13 indicators)
7 (54 percent) find gender is insignificant
5 (38 percent) find men have significantly higher inputs
1 (8 percent) find women have significantly higher inputs
Provide context for informal learning, creation of social safety nets, organization for regulation, protection, change and challenge of agricultural and development related factors.
(local level agricultural focused
co-ops, user groups, committees).
Non-group informal information exchange
(via social networks).
3. Political representation
4. Social and political capital Page
4. Summary of results for social and political capital
10 studies reviewed (3 published in peer reviewed journals).
18 measures of group participation, 1 non-formal information exchange,
0 political representation.
Descriptive statistics (6 indicators)
4 (67 percent) find men have higher mean inputs
2 (33 percent) find women have higher mean inputs
Bivariate or multivariate analysis (19 indicators)
Godquin and Quisumbing (2008) examine participation in general and production groups among 304 households in the Philippines.
Gender does not affect group participation overall, however men more likely to be in production oriented groups as compared to women.
IFPRI Gender and Governance team (2009) examine participation in CBOs, farmer based organizations and agricultural cooperatives in India, Ghana and Ethiopia respectively (966 – 1761 households).
In India, female household head not significantly associated with CBO participation, however women participate mainly in self help/woman’s groups while men participate in forest groups, cooperative societies and caste associations.
In Ghana, male headed households significantly more likely to participate in farmer based orgs using probit regression.
In Ethiopia, descriptive and bivariate analysis shows men headed housholds significantly more likely to participate in co-ops (4 versus 24 percent).
4. Key studies for social and political capital Page
What can we say about those cases where we do, indeed, find differences? Do they matter?
Division of labor.
What can we say about regional evidence?
Approximately 80 percent of studies from SSA.
Studies from Asia have tended to focus on men’s and women’s labor inputs, rather than productivity on male and female farms, because of joint farming.
Lacking studies on Middle East and South/Latin America.
What can we say about diversity of input evidence?
Most evidence for technology (especially fertilizer, seeds) and human resources (extension).
Least evidence for machinery, lifecycle factors and non-formal information exchange.
What can we say about gender measures?
Majority disaggregate at the household head level.
Few do sensitivity analysis – however when done, evidence indicates they seem to matter.
Summary of Key findings and recommendations
Women are almost always disadvantaged in mean use indicators.
However, these differences do not necessarily translate to differences when other factors are controlled for, depending on study design, evaluation framework, etc.
Factors are likely to differ based on geographic area, cultural context.
Main question for policy: when and where do gender disparities in inputs matter?
Gender indicator matters . Recommendations for plot specific and sensitivity analyses.
Lack of regional diverse studies . Recommendation for studies on Latin/South America and Middle East.
Lack of attention to some inputs . Recommendation for studies which examine or include measures of machinery, lifecycle challenges and non-formal information exchange.
Understanding gender differences in agricultural productivity in Nigeria and Uganda (Peterman, Quisumbing, Behrman & Nkonya)
IFPRI collected Household and plot-level data from Nigeria (2005; N = 3707) and Uganda (2003; N = 2536).
Female headed households in Nigeria and female owned plots in Uganda have significantly lower productivity.
Productivity differences persist in both counties after controlling for complementary inputs, however this varies within primary crops and within agro-ecological zones.
In Uganda, lowest productivity among ‘mixed gender ownership’ plots ,which may be suggestive of bargaining issues within households
Gender indicator matters in Uganda – use of alternate indicators at the household level (female headship, percent female managed) do not produce same results.