A M B E R P E T E R M A N ,
A G N E S Q U I S U M B I N G ,
J U L I A B E H R M A N , A N D
E P H R A I M N K O N Y A
IFPRI
AFRICA GROWTH FORUM
JANUARY 19-20, 2011
Understanding the Complexities Surrounding Gender Differences in Agricultural Productivity in Nigeria and Uganda
Harvesting in Nigeria, Credit: Yosef Hadar
Outline of presentation
Page 2
Framing the issue
Methods
Findings
Policy implications
This paper:
Page 5
Provides new estimates of gender differences in agricultural productivity using IFPRI household survey data from Nigeria (2005) and Uganda (2003)
Address some complexities by looking at:
Crop choice
Sensitivity of productivity estimates to choice of stratifying ‘gender’ variable (sex of hh head, sex of plot owner, mixed ownership)
Heterogeneity within agro-ecological zones
Controlling (where possible) for hh-level unobservables
Controlling (where possible) for biophysical characteristics of plot
Nigeria 2005 Uganda 2003
Page 6
Collected to evaluate Fadama II, a national agricultural welfare program
Household level data: 3,750 hhs
Gender variable: Sex of hh head
Collected to study natural resource management and poverty
Plot level data: 3,625 plots in 851 hhs
Gender variable: Sex of crop ownership for plot, also allows for mixed ownership; sex of hh head also collected
Biophysical plot characteristics
Methods: Data
Both countries: Large agricultural sectors, diversity in agro- ecological zones, crop choice, ethnic variation and low women’s status and property rights.
Methods: Empirics, tobit model
Page 7
ln Yi = α0 + α1ln Li + α2 ln Ti + ß ln Ei + γ EXTi+ δ Genderi + ε
Yi ith hh or plot value of crop yield per unit area
L i labor input (hired or family)
T i vector of land, capital, and other conventional inputs
E i educational attainment
EXT i index of extension services
Gender i dummy variable for the sex or gender of the farm manager or household head
ε error term
Methods: More on empirics
Page 8
Allow for mass point at zero using tobit
Treatment of zero as either fallow or no output
Crop choice modeled using probit and Cragg’s two-tiered unconditional tobit
Uganda: explore robustness to inclusion of fixed effects using Honoré’s fixed effects tobit estimator
All regressions control for age, education of head, hh size; land, irrigation, fertilizer and seeds, extension, labor (previous season inputs);
All full sample regressions control for primary crop indicators (results are robust to inclusion of secondary crop indicators).
Findings: Plotting productivity
Page 9
Findings: Summary of tobit estimation results
Page 10
Variable
Nigeria Full Maize Rice Cassava Tomato Leafy veg
Cowpea
FHH=1 -0.32*** -0.25 -0.03 -0.49 -2.08** -0.34 -0.06
Uganda Full Banana Beans &peas
Maize Sweet potato
Cassava Sorghum
Femalecrop owner=1
-0.27** 0.23 0.07 -0.06 -0.80* -0.27 -0.93**
Mixed owners=1
-0.29* 0.21 -0.82** -0.65 -0.98*** -0.66 -0.38
Findings
Page 11
Productivity significantly lower on plots owned or managed by females; results hold taking into account farm and hh characteristics and crop choice
Results vary across crops, agro-ecological zones, and with inclusion of biophysical characteristics
Type of gender indicator matters: extent of productivity differential diluted when headship is used as stratifying variable
Productivity lowest on mixed ownership plots, but not robust to inclusion of hh fixed effects. Indicates bargaining difficulty with mixed ownership/decision making?
Policy implications—part 1
Page 12
Headship as a stratifying variable underestimates productivity differences => need to pay attention to level of aggregation in collecting sex-disaggregated data
Productivity lowest on female-owned plots =>pay attention to gender differences in control of resources in research and program implementation
Policy implications--2
Variation by region, crop, biophysical characteristics => address gender in context of regional ecological and biophysical needs, cultural context
Avoid extrapolation of policy findings from very localized studies; increase geographical representativeness of data collection and analytical efforts
Credit: ILRI
Questions, Comments?
Page 14
Paper funded by the FAO as a background paper for the State of Food and Agriculture (2010) and we gratefully acknowledge funding. Thanks to Edward Kato for assistance with data and understanding of local context and to Andre Croppenstedt and two anonymous reviewers for helpful comments on an earlier draft.
Paper is forthcoming in the Journal of Development Studies (2011)
Top Related