Gender Research for Agnostics: Methods and Findings from IFPRIs Gender and Intrahousehold Research P

1
Gender Research for AgnosticsMethods and
Findings from IFPRIs Gender and Intrahousehold
Research Program

• Agnes R. Quisumbing

2
What is gender research? (can I just put a
dummy variable and that will take care of it?)

• research that takes into account the socially
  constructed relationships/differences between men
  and women (not only biological differences)
  
• The dummy variable may work, but its
  interpretation will be very different, depending
  on culture and context (more on this later)
  

3
When does it make sense to pay attention to
gender in your research?

• When there are systematic gender differences in
• Outcomes (yield differentials, health and
  nutrition indicators, poverty rates, etc.)
  
• Determinants (effects of male and female
  schooling, male and female land ownership, male
  and female headship, etc.)
  
• Processes (when there are differences in the
  preferences, motivations, and behavior of men and
  women)

4
Gender will not always be the most important
variable in all contexts.

• But it is important to test it.
• Other factors which may be important
• Inter-household differences (e.g. between rich
  and poor, differences due to caste or ethnicity)
  
• Other sources of intrahousehold differences (age,
  birth order, relationship to head of household)
  

5
Four key aspects of IFPRIs gender and
intrahousehold research program

• Models of household behavior that allow for
  individual preferences
  
• Development of quantifiable indicators of
  bargaining power
  
• Methods to address endogeneity and measurement
  error in these indicators of bargaining power
  
• Qualitative and quasi-experimental methods for
  program and project evaluation

6
Country coverage

• Four high-concentration countries (Bangladesh,
  Guatemala, Ethiopia, South Africa)
  
• Eight supplemental studies countries
• Not in map Mexico

7
Qualitative and quantitative methods

• Combination of qualitative methods with
  quantitative household surveys
  
• Qualitative studies inform design of quantitative
  modules
  
• Different policy focus in each, but comparable
  modules across countries
  

8
Are there gender differences in outcomes? The
case of yields on male and female plots in
Burkina Faso (Udry, JPE 1996 Alderman et al.
1996 Udry et al. Food Policy 1997)
9
Are there gender differences in determinants?
Example 1. Determinants of input intensity at
the plot level, Burkina Faso
10
Are there gender differences in determinants?
Example 2. The case of cocoa yields in Western
Ghana (Quisumbing et al. EDCC 2001) (n391)
Regressors included characteristics of parcel at
acquisition, tree variety dummies, tree age varia
bles, and land tenure variables
11
Why does gender not seem to matter?

• Women get to cultivate cocoa on their own plots
  only after acquiring it as a giftin return for
  helping husbands to plant cocoa. So, conditional
  on planting cocoa, men and women have equal
  probabilities of cocoa planting
• Women do have lower yields, but only weakly
  significant (p0.11)
  
• Lower yields not due to differences in land
  tenure (controlled for by land tenure dummies)
  but possibly other constraints women face
  
• Need to understand the process by which men and
  women acquire land, and decisions to plant
  cocoatied up with shifts in the inheritance
  system

12
Does gender make a difference in terms of
processes? A more difficult question to answer

• What is the underlying model of household
  behavior?
  
• Tests of the unitary vs. the collective model of
  household behavior
  
• Unitary model households behave as one (common
  preference, or dictator) households pool
  resources
  
• Collective model different preferences within
  the household members do not pool resources
  
• Prediction of the collective model ones share
  of resources depends on bargaining power within
  household

13
What determines bargaining power within the
household?

• Control over resources (assets)
• Factors that can be used to influence bargaining
  process (legal rights, skills and knowledge,
  ability to acquire information, bargaining
  skills, use of violence)
• Mobilization of personal networks
• Basic attitudinal attributes (self-esteem,
  self-confidence, emotional satisfaction)
  

14
Proxies for bargaining power in empirical work

• Public provision of resources to particular hh
  member
  
• Shares of income earned by women
• Unearned income
• Inherited assets
• Assets at marriage
• Current assets

15
Considerations in choosing a proxy indicator

• Exogeneity with respect to labor force
  participation or savings/accumulation decisions
  
• Rules regarding assignment of asset ownership
  (legal, customary)
  
• Marriage market selection
• Cultural relevance of indicators

16
Focus on assets at marriage

• Assets at marriage important in many cultures
  start of the new social and economic unit
  
• Determined before decisions made within marriage
• Can be affected by personal and family
  characteristics
  

17
Women bring less to the marriage than men
18
Cross-country analysis (Quisumbing and Maluccio
OBES 2003)

• Do assets held by husband and wife affect
  household decisions on expenditures?
  

19
Do assets held by husband and wife affect
household decisions on expenditures?

• Expenditure categories
• Food budget share
• Health budget share
• Education budget share
• Child clothing budget share
• Cigarettes and alcohol budget

• Determinants
• total expenditure per capita
• household size
• demographic composition
• mens assets at marriage
• mens schooling
• womens assets at marriage
• womens schooling

Controlling for total expenditure, test equality
of coefficients of mens and womens assets at m
arriage/schooling Control for measurement error i
n assets using instrumental variables,
with family background characteristics as
instruments
20
Summary of expenditure shares regressions
21
Using a gender perspective in program and project
evaluations

• Quantitative methods
• With and without design (comparable groups)
  impact of new agricultural technologies in
  Bangladesh
  
• Matched groups design compare children of
  working mothers, those in Hogares Comunitarios
  program and those outside, matched by age and sex
  (Guatemala)
• Randomized assignment control and treatment
  communities in evaluation of PROGRESA in Mexico

• Qualitative methods
• Focus groups in adopting/nonadopting villages
• Focus groups of parents, madre ciudadoras, social
  workers, program staff
  
• Focus groups of promotoras, beneficiary mothers

22
General Research Findings
23
Households do not act as one when making decisions

• Men and women do not always have the same
  preferences nor pool their resources
  
• Mens and womens resources have different
  effects on household decisionmaking
  
• Who is targeted affects the outcome of policy

Source Quisumbing and Maluccio, OBES 2003
24
Share of resources depends on bargaining power,
but women control fewer resources than men
25
Local norms, not statutory laws, determine
womens rights

• Formal and informal legal and institutional
  frameworks are basis for womens rights
  
• In Ethiopia, local norms most important factor
  explaining distribution of assets upon divorce or
  death (Fafchamps and Quisumbing JDS 2002)
  
• Local custom may evolve to become more beneficial
  to women, if womens labor becomes more valuable
  (Quisumbing, Otsuka, Payongayong and Aidoo EDCC
  2001)

26
Relative to men, increasing womens resources
benefits families

• Equalizing resources held by women and men can
  increase agricultural yields by up to 20
  (Alderman et al. 1996)
  
• Increases in womens resources have the strongest
  effects on education, health, and nutrition
  (Hallman 2000 Smith et al. 2001)
  
• Womens social networks help families cope with
  income shocks (Maluccio et al. 2001)
  

27
Increasing womens resources improves
agricultural productivity

• Burkina Faso Output of womens plots, and total
  household output could be increased by 10-20 by
  reallocating resources from mens plots to
  womens plots (Alderman et al. 1996)
• Ghana strengthening womens property rights
  increases incentives to adopt agroforestry and is
  good for environmental management
  

28
Improving womens status and resources improves
child health and nutrition
Contributions to reductions in child
malnutrition, 1970-95
Source Smith and Haddad 2000
29
Innovative ways to increase womens resources
have made projects successful

• Examples include
• Credit and technologies targeted to women (Sharma
  2001 Bouis et al. 1998)
  
• Income transfers targeted to women (Skoufias and
  Mclafferty 2001 Adato et al. 2000)
  
• Community day care programs (Ruel et al. 2001)

30
Why does it make sense to pay attention to gender
in your research?

• Because it just might make your research better!