Women’s Economic Empowerment: Indicators and survey design Amber Peterman UNICEF Office of Research—Innocenti Poppov Annual Conference: Methods Workshop, Addis Ababa June, 2015

Women’s empowerment: for what purpose? Women’s [and men’s] economic empowerment measures are important to measure and analyze to understand: ①  Overall levels and trends of gender economic relationships in a specific geographical setting/population of interest (program design) ②  Program impacts (change over program period) o  Direct (most proximate, linked to program participation) o  Intermediate o  Final (ultimate goal or objective of program) ③  Program moderators (baseline or initial conditions, understand program uptake and/or heterogeneous effects).

Commonly used indicators for measuring women’s economic empowerment in micro-surveys ¨ 

¨  ¨  ¨  ¨ 

¨ 

¨ 

¨ 

Labor force participation: Earnings, occupation type, time use (both productive and leisure), entrepreneurship, profits Agricultural productivity: Income outputs (on individual plots) Asset ownership: Land, productive assets, durable/household assets Financial inclusion: Savings, credit, bank account Consumption/expenditure: Individual or ‘gender-specific’ goods and services Decision-making and autonomy: Economic and other domains, ability to affect domains of one’s life, preferences over tasks and decisionmaking Combinations or aggregate measures: Of the above and others (indices) Subjective measures? Stress? Satisfaction and happiness?

Common themes & take away lessons ¨ 

There is no “right” measure–likely multiple measures are useful and warrant measurement within any given data collection, depending on: Objective of research (intervention type, feasibility of measuring outcomes that will change in time frame) ¤  Who the sample is (age, marital status?) ¤  Setting/cultural norms ¤ 

¨ 

¨ 

¨ 

Investigate indicators beforehand in any formative research or mixed methods data collection Entails significant additional cost to data collection (both logistical and time costs) – all ‘inputs’ and ‘outputs’ must be measured at individual level Measurement matters—different indicators present different concepts and dimensions of the ‘full picture’

“Good practices” for designing indicators

①  Keep it numerical (work “part time” or “full time” vs number of hours worked per day) ②  Keep it easy (use concepts appropriate for populations with low levels of formal education, use visual aids) ③  Keep it short ④  Keep it consistent (use same variations in response options) ⑤  Give a way out (allow non response and don’t know options) UN Foundation & Exxon Mobile: Measuring women’s economic empowerment “Road Map” http://www.womeneconroadmap.org/sites/default/files/Measuring%20Womens%20Econ %20Emp_FINAL_06_09_15.pdf

Structure of presentation: [quick] case studies ①  Gender and land ownership: Doss C, Kovarik C, Peterman A, Quisumbing A & M van den Bold (2015). “Gender inequalities in ownership and control of land in Africa: Myth and reality” Agricultural Economics, Special issue on gender and agriculture in sub-Saharan Africa, 46: 1-25.

②  Women’s decision-making: Peterman A, Roy S, Schwab B, Hidrobo M & DO Gilligan (2014). “Measuring women’s decision making: Indicator choice and survey design experiments from transfer evaluations in Ecuador, Uganda and Yemen (working paper).”

③  Women’s Empowerment in Agriculture Index (WEAI): Alkire S, Meinzen-Dick R, Peterman A, Quisumbing A, Seymour G & A Vaz (2013). “The Women’s Empowerment in Agriculture Index” World Development, 52: 71-91.

sub-Saharan Africa and 20 per cent in Latin America. •

Development projects do best when women’s ro

project design from the start. The IFAD-funded B Case study 1: Motivation ‘Sound bites’ In developing countries in Africa Project operates in the central and eastern Syria and Asia and the Pacific, women typically work 12 to 13 hours per week more than men.

• The value of men’s livestock holdings is about twice that of women’s. • Men’s landholdings average three times those of women. Women represent fewer than 5 per cent of agricultural landholders in North Africa and Western Asia, and an average of 15 per cent in sub-Saharan Africa.

on rehabilitating the badly degraded rangelands

Bedouin herders who live there. But the project r

lasting environmental and economic improvemen especially for women.

The project has done much to improve condit

-- Oxfam, Aid, for women. LiteracyAction classes areetc. the foundation o

very high among Bedouin women. Training cours

food processing and sewing have eased women up income-earning opportunities.

Now that households are better off, there is le

early. And as women gain more economic auton

relations are shifting. Nofa Awad Al-Anad is marr

• Women receive only 5 per cent of Shaddade, in Hassekeh Province. She trained as Infographic from the Bill and the extension resources of UN men, and Bread SYP 500--(US$11) -- Variations found in Women, a day making dresses. She no Melinda Gates Foundation for thefewer world and and smaller others citing FAO gender loans. are granted money if she wants something for herself or her

database • Many land ruralrights women spend up to four hours a day collecting fuel for household use, sometimes travelling

“My father used to dominate in the household

I consult with each other when we want to do so

Why current statistics are flawed

¨  ¨ 

Single statistic: Contains some element of the truth? However, problematic because: 1)  2)  3)  4)  5) 

q 

Masks regional and within country variations No clear attention to how ‘women’s land ownership’ is defined – and how this relates to bundles of rights Treatment of joint ownership? No clear comparison group (assumed to be men in the same context?) Based on limited data

All these factors handicap policy recommendations, ultimately hurting efforts towards understanding gender-inequalities

Overall study objectives 1)  2)  3) 

Explore conceptual and methodological issues in estimating “women’s land ownership” Review existing evidence from large-scale studies in Africa (data collected post 2002) Provide new data-based estimates on women’s land ownership in Africa: q  q  q 

4) 

Food and Agriculture Organization (FAO) gender and land rights database Demographic and Health Surveys (DHS) Living Standard Measurement Survey-Integrated Surveys on Agriculture (LSMS-ISA)

Policy, advocacy and research implications of findings.

Gender and land conceptual issues

1) 

The definition of ownership: q  How

does this differ from access/effective ownership (management) or other bundles of rights? q  Self reported or documented (title, deed)?

How to treat jointness? 3)  Defining land area: 2) 

q  All

household land? q  Only farmable (agriculture) or rural land? q  Government/group (clan)/institutional land?

1)  2)  3)  4)  5) 

% of

This information can then be used to genera percentage of women who are landowners a Thisthis information beitused tobe gen For measure tocan be useful must a measure is calculated asthen follows, where the Operationalizingeither indicators percentage of women landowne the sum of the shares ofwho the are different cate individually or jointly, and the denom measure is calculated asafollows, further disaggregated into couples where and oth can compare them to produce measure of ineq either individually jointly, and the dend !"#$%!!"#$%&#'() percent of women owningorland would be!"#!!"# quite (1) , !"#$%!!"#$%&!!"!!"#$%!!! !"#$%!!"# owned land, compared with of a situation in which One of the limitations this approach is !"#$%!!"#$%&#'() !"# Indicator 2 does not tell us how widely land is o size or value data are available, additiona (1) , women who are landholders: !"#$%!!"#$%&!!"!!"#$%!!! !"#$%!! The measure (Indicator 2) is the perc landowners who are women. is thesecond mean size of plots.

The secondsize measure (Indicator 2) Mean isofthe A third indicator uses the number of plots lan (2) (4) Mean of women’s plots; % of landholders who are women: !"#$%!!"#$%&!!"!!"#$%&#'() women (men) as the numerator. !"#$%!!"#$%&'() (2) The mean size alone is does not tell us anyth Again, the numerator the number of wom !"#$%!!"#$%&!!"!!"#$%&#'() !"#$%&!!"!!"#$%!!"#$%!!"!!"#$%! !"#$ % of plots owned by women:(3) denominator is a simple indicator that isofoften quite eas , wom is the number men and !"#$%!!"#$%&!!"!!"#$%! ! Again, the numerator is the number of w A more useful indicator ismust the distribution Indicators 1 and 2the arenumber often reported denominator is of men and w For this measure to be useful it be ainterch distrib Average size/value of women’s land calculate percentage of land that is ow former, wethe need to the percentages fo the sum of the shares ofknow the different categories Indicators 1 andinto 2 are often and reported further disaggregated couples other inte for !"#$!!"#!!!"#$%!!"!!"#$% !"#$!!" , former, we need to know the percentage % of land (area or value) owned7(5) !"#$%!!"#$!!"#! !"#a Again, it may be that each individual is asked One of the limitations of this approach is that ea ownership of everyone in the household. by women: size The or value data are available, additional meas first challenge with this indicator is h 7 is thedefined mean size of plots. Again, as it may be that each individual the total land area owned is byask m !"#$%!!"#$%&'()

ownership of everyone theland household. in which case the shareinof owned by

LSMS-Integrated Surveys on Agriculture

¨ 

¨ 

¨ 

¨ 

Joint effort led by the World Bank in 7 African countries starting in 2008 (Mali still in progress) Detailed plot level characteristics including, ownership, documentation, area (GPS) measures and value (self reported) measures Among total land area owned or accessed by households, women solely own a high of 31% in Malawi, followed by Uganda (16%), Tanzania (15%), Niger (8%) and Nigeria (3%) Comparatively, men solely own on average 21.8 times as much absolute land area in comparison with women in Nigeria, and between 1.1 (Malawi) to 6.9 (Niger) times as much land as women solely own in the other countries.

LSMS-ISA: Cross-country (area measures) Legend

Malawi

Ethiopia

Women's ownership, undocumented Women's ownership, documented Men's ownership, undocumented

6% 16% 19%

Men's ownership, documented

22%

0.6%

Joint ownership, undocumented

31%

Joint ownership, documented 41%

13%

17%

Accessed 0.2%

32%

Owned, undocumented (men and women)

0.4%

Uganda

Tanzania

Nigeria*

Niger

3% 5%

8%

14%

11%

1%

22%

13%

3%

10%

7%

31%

24% 36%

68%

32%

2.9%

15%

49% 5%

*In Nigeria “ownership is defined as the right to sell or use as collateral”

0.4%

22%

6% 5%

7%

Which countries have data?

¨ 

¨ 

¨  ¨ 

Lit review: 43 measures, 17 studies, 9 countries FAO Gender land data base: 9 countries DHS: 10 countries LSMS-ISA: 6 countries

Gender and land: Discussion points

¨ 

¨ 

¨ 

Individual and comparative levels and gaps in land measures differ based on indicator used and study sample Isolated statistics cannot be generalized and do not present an accurate picture of women’s landownership worldwide– yet very consistently women are disadvantaged and often the gender gap is large In future analysis it is crucial to: §  §  § 

Define definitions and indicators used Improve methodology, data collection and analysis efforts Recognize that country-specific statistics are the most relevant to drive policy and advocacy efforts in a given context and focus should be on developing strong in-country systems for data collection and dissemination.

Case study 2: Women’s decision-making

¨ 

¨ 

¨ 

Often preferred because they represent direct measures – rather than indirect (proxy) measures of women’s empowerment Despite a limited number of qualitative studies examining validation of questions, there is scant quantitative research examining robustness of quantitative measures. Many studies still conflate status (static) with empowerment (process) (Heckert and Fabric 2013). Although there is evidence that favorable outcomes are associated with bargaining power, empirical evidence cannot rigorously identify causality due to study design and data limitations -- it is therefore difficult to identify specific policies that increase women’s bargaining power in development settings (Doss 2013).

Example: Review of programming and impacts on women’s empowerment (van den Bold and colleagues (2013)) Type of intervention

Quantitative evidence

Qualitative evidence

1

Conditional Cash Transfers (CCTs)

Mixed

+

2

Unconditional Cash Transfers (UCTs)

Mixed

More evidence needed

3

Microfinance

Mixed

Mixed

4

Agricultural interventions

Mixed/More evidence needed

Mixed/More evidence needed

“While many development initiatives seem to target women specifically, or have women’s empowerment as one of their objectives, no sufficient body of evidence overwhelmingly points to success in terms of improving women’s empowerment, or improving nutrition through women’s empowerment (pp. 29)”

Objectives: Transfers and decision-making

① 

② 

③ 

Whether relative rankings of decision-making are sensitive to differences in indicator construction and survey design experiments Tests correlation between the various decision-making indices and other proxy (indirect) measures of women’s status (women’s education and age) or development outcomes (household dietary diversity and food consumption) Whether the transfer programs in Ecuador, Uganda and Yemen had a measurable impact on decision-making outcomes for women 

Standard decision-making questions in quantitative household surveys “Who in your household usually has the final say” … Own health ¤  Own earnings ¤  Children’s health ¤  Children’s education ¤  Small daily household (food) purchases ¤  Large household (asset) purchases ¤  Use of family planning ¤ 

¨ 

¨ 

Collected in the Demographic and Health Surveys and other large multi-topic surveys Typically asked only to women

Standard response options

¨  ¨  ¨  ¨ 

¨ 

¨ 

Respondent herself Her partner Respondent and partner jointly Respondent and others in the household jointly Others in the household Could also enter one or more IDs of household members

Indicator constructs: Decision-making indicators

¨ 

¨ 

¨ 

Does jointness matter?: Create standard measures (sum and factor analysis index): 1) Sole, 2) sole and joint decisions Underlying threat points: Who makes the decision, or who would make the decision in the case of a disagreement or dispute? (=1 if woman still can make decision under dispute) Division of tasks/preferences: “In an ideal situation, who in your household would make the decision?” ( =1 if actual = ideal)

Percentage of women's reported sole decision-making accross domains 0.8

0.72

0.67

0.7 0.6 0.5

0.61 0.54

0.51 0.44

0.47

0.46 0.44

0.4 0.42

0.4

0.56

0.34

0.3

Ecuador

0.4 0.34

0.23

0.37

Yemen 0.19

0.2 0.1 0 Own work for pay

Own health

Child's education

Child's health

Daily food purchases

Women report making: ¤  4.5 (out of 9) sole decisions in Ecuador (50%) ¤  2.5 (out of 6) sole decisions in Uganda and Yemen (42%) ¤  Highest: Own health, daily food, purchases ¤  Lowest: Child’s education and large asset purchases

Large asset purchases

Uganda

Comparison of women's decision-making indicators 10

9

9 7.52

8 7

6

6

6 5

6

5.31 4.47

Ecuador

4.41

Yemen

4 2.54 2.51

3

3.1 2.43

2.28

Sole after disagreement

Ideal decisionmaking

Uganda

2 1 0 Sole

Sole or joint

Total possible

¤ 

Including jointness increases decision-making in all countries (more in Ecuador and Uganda)

¤ 

Women report higher decision-making in Ecuador after disagreement, lower in Yemen Ideal decision-making is not markedly higher than actual in Ecuador and Yemen

¤ 

Relative rankings of decision-making using factor analysis

Ecuador (N = 1,174) Sole Sole or joint Ideal After disagreement alpha statistic

¤ 

¤ 

Sole 1.00 0.31 0.52 0.65 0.91

Sole or joint 1.00 0.47 0.36 0.86

Ideal

1.00 0.43 0.89

After disagreement

1.00 0.90

Households in Ecuador and Uganda show low correlation between different constructions of indicators (majority do not exceed 0.50) Households in Yemen show higher correlations, however differences still exist (0.74 – 0.89)

Summary: What do we know, what can we do better?

¨ 

¨ 

¨ 

¨ 

Intent matters: Explore the wording of questions which most reflect local perceptions of how decisions are made, as well as program goals— particularly in local languages, which often have limited vocabulary for nuances. Formative research is particularly helpful for both these points. Ask about the right domains: Specific to the level of influence one might expect the program to change or depend on for leveraging benefits – specific to context. Analysis: Pay attention to response options which reflect possible decisionmaking arrangements. Do not assume that sole decision-making is preferable to joint decision-making, depending on household structure and power dimensions within the household. More research!: We need to continue advancing the frontier of how to most accurately capture and analyze decision-making and women’s empowerment.

Case study 3: WEAI ¨ 

¨ 

Objective: Design, develop, and test an index to measure the greater inclusion of women in agricultural sector growth that has occurred as a result of US Government intervention under the Feed the Future (FTF) Initiative (Bangladesh, Guatemala and Uganda) What is “greater inclusion”? The concept of Inclusive Agricultural Sector Growth is broad and multi-dimensional, Feed the Future defines it as: “the empowerment of women in their roles and engagement throughout the various areas of the agriculture sector, as it grows, in both quantity and quality”

Nuts and bolts An aggregate index in two parts: ¨ 

Five domains of empowerment (5DE): assesses whether women are empowered in the 5 domains of empowerment in agriculture. Based on the Alkire Foster methodology: ¤  ¤ 

¨ 

¨ 

¨ 

Incidence of Empowerment - The percentage of women who are empowered Adequacy among the Disempowered - The weighted share of indicators in which disempowered women are empowered

Gender Parity Index (GPI): reflects the percentage of women who are as empowered as the men in their households Constructed from data of the primary male and primary female adults in the same household 5DE, GPI range from zero to one; higher values = greater empowerment

Five domains of empowerment

An individual is considered to be ‘empowered’ if he/she achieves adequacy in 80% of the weighted indicators

Bangladesh Pilot results

q 

Southern part of the country: q  25 villages from 5 rural districts q  Sample: 450 households (800 individuals)

q 

39% of women are empowered

q 

Disempowered women have adequate achievements in 58.4% of domains

  q 

59.8% of women enjoy gender parity

q 

Households without gender parity have a 25.2% empowerment gap between the woman and man

Overall, the WEAI score is 0.762

Bangladesh: How to increase empowerment?

Moving forward: Preliminary results from FTF WEAI Scores baselines, WEAI scores Country  

Region  

WEAI  Score  

Ranking  

Bangladesh  

Asia  

0.66  

Low  

Cambodia  

Asia  

0.98  

High  

Nepal  

Asia

0.80

Medium

Tajikistan

Asia

0.69  

Low  

Haiti  

Latin America & Caribbean  

0.85  

High  

Honduras  

Latin America & Caribbean  

0.75  

Medium  

Kenya  

East Africa  

0.72  

Low  

Rwanda  

East Africa  

0.91  

High  

Uganda  

East Africa  

0.86  

High  

Ghana  

West Africa  

0.71  

Low  

Liberia  

West Africa  

0.69  

Low  

Malawi  

Southern Africa  

0.84  

Medium  

Zambia  

Southern Africa  

0.80  

Medium  

Disempowerment scores Disempowerment scores for women (1-5DE) for women (1 – 5DE)

0.00  

0.05  

Disempowerment  Index  (1  -­‐  5DE)  

0.10  

0.15  

0.20  

0.25  

0.30  

0.35  

0.40  

WEAI Summary ¨ 

¨ 

¨ 

¨ 

¨ 

¨ 

WEAI seeks to measure empowerment [in agriculture] directly using a measure which is comparative (within households) Three ways to change WEAI: 1.  Empower women 2.  Increase the adequacy among disempowered women 3.  Increase gender parity As a monitoring indicator for FTF to evaluate whether programs are having intended effect on women's empowerment As a diagnostic tool to help identify areas in which women and men are disempowered, so that programs and policies can be targeted to those areas Conducting more research: testing new indicators/assessing validity in different contexts, modifications, WEAI 2.0. WEAI Resource Center: http://www.ifpri.org/book-9075/ourwork/program/weai-resource-center

Acknowledgements: Gender and land study

Funding for this work was provided by the Consultative Group on International Agriculture Research (CGIAR), Program on Policies, Institutions, and Markets and by an anonymous donor. We are grateful to IFPRI colleagues Maha Ashour, Zhe Guo, Caitlin Kieran, Hazel Malapit, Ruth Meinzen-Dick, and Wahid Quabili for helpful comments and assistance. Hosaena Ghebru Hagos of IFPRI for allowed access to unpublished Trabalho de Inquerito Agricola (TIA) survey collected by Ministry of Agriculture data from Mozambique; Perrine Burnod of Centre de coopération internationale en recherche agronomique pour le développement (CIRAD) for allowing access to unpublished data from Madagascar. This paper benefited from helpful comments from the participants at the Gender and Agricultural Productivity in Sub-Saharan Africa workshop at the International Fund for Agricultural and Development in Rome, and the International Conference on Agricultural Statistics 2013 in Rio de Janeiro, and to the Living Standards Measurement Study— Integrated Surveys on Agriculture (LSMS-ISA) team at the World Bank for helpful comments.

Acknowledgements: Decision-making study

In country partners for data collection and survey management: Centro de Estudios de Población y Desarrollo Social (CEPAR), Yemen Polling Company and Makerere University. IFPRI colleagues including John Hoddinott, Nancy Johnson, Amy Margolies, Hazel Malprit, Vanessa Moreira and Agnes Quisumbing for helpful discussions at study conception and contributions through work on the larger food and cash transfer evaluation. Caroline Guiriec for assistance in administration of the grant. WFP (Rome, Quito, Kampala and Sana’a) for excellent collaboration and program implementation. Funding from the Government of Spain for the impact evaluations and to the CGIAR Research Program on Agriculture for Nutrition and Health (ANH) led by IFPRI for the analysis and writing of this paper.

Acknowledgements: WEAI Funding for the was provided by the US government’s Feed the Future Initiative. We thank Caren Grown and Emily Hogue for their guidance and dedication in conceptualizing and realizing this index; Joanne Tomkinson, John Hammock, Hazel Malapit, Amy Margolies, Chiara Kovarik, Betsy Pereira, Katie Sproule, Elisabeth Becker for their valuable input to design, fieldwork, and administrative and communication components of this effort. We also thank the participants of the initial and final methods workshops and others who have provided helpful feedback and guidance. Finally, we thank our collaborators, Data Analysis and Technical Assistance, Ltd., in Bangladesh, Vox Latina in Guatemala, and Associates Research Uganda Limited in Uganda. Md. Zahidul Hassan, Monica Dardon, and Herbert Kamusiime, who led the fieldwork in Bangladesh, Guatemala, and Uganda, and all the individuals who worked as enumerators and data entry and logistics specialists on the pilot survey and case studies.

Works cited

¨ 

¨ 

¨ 

¨ 

¨ 

Doss C. 2013. Intrahousehold Bargaining and Resource Allocation in Developing Countries. World Bank Research Observer, 28(1). Heckert J and MS Fabric. 2013. Improving Data Concerning Women’s Empowerment in Sub-Saharan Africa. Studies in Family Planning, 44(3): 319-344. Kabeer N. 2001. “Reflections on the measurement of women’s empowerment.” In Discussing Women’s Empowerment – Theory and Practice. Sida Studies No. 3. Novum Grafiska AB: Stockholm. Malhortra A, Schuler SR and C Boender. 2002. Measuring Women’s Empowerment as a Variable in International Development. Background paper prepared for the World Bank Workshop on Poverty and Gender: New Perspectives. van den Bold M, Quisumbing A and S Gillespie. 2013. Women’s Empowerment and Nutrition: An Evidence Review. International Food Policy Research Institute (IFPRI) Discussion Paper #01294. Washington DC.