LSMS+ Program in Sub‑Saharan Africa Findings from individual‑level data collection on labor and asset ownership Section 1. Introduction and LSMS+ overview III Copyright © 2021 The World Bank. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/ by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following condition: Attribution—Please cite the work as follows: Hasanbasri, Ardina; Kilic, Talip; Koolwal, Gayatri; Moylan, Heather. 2021. LSMS+ Program in Sub-Saharan Africa: Findings from Individual‑Level Data Collection on Labor and Asset Ownership. Washington, D.C.: World Bank Group. Disclaimer The findings, interpretations, and conclusions expressed in this Guidebook are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Living Standards Measurement Study (LSMS) World Bank Development Data Group (DECDG) lsms@worldbank.org www.worldbank.org/lsms data.worldbank.org Cover images: © World Bank LSMS+ Program in Sub‑Saharan Africa Findings from individual‑level data collection on labor and asset ownership Ardina Hasanbasri Lecturer, University of Michigan and Independent Consultant, Development Data Group (DECDG), World Bank ardinah@umich.edu Talip Kilic Senior Economist, DECDG, World Bank tkilic@worldbank.org Gayatri Koolwal Founder, Development | Science and Independent Consultant, DECDG, World Bank gkoolwal@worldbank.org Heather Moylan Survey Specialist, DECDG, World Bank hmoylan@worldbank.org i Contents Executive Summary............................................................................. 1 Key findings..................................................................................................... 4 Section 1 Introduction and LSMS+ overview.........................7 1. How can we better capture men’s and women’s economic opportunities in surveys? ............................................................ 8 2. The LSMS+ Program: addressing key data gaps, and prioritizing individual interviews.................................................................... 9 3. LSMS+ implementation in SSA (2016‑2020).................................... 11 3.1. Share of eligible respondents interviewed for modules on asset ownership........................................................................................ 11 3.2. Extent of self‑reporting in modules on education, health, and labor........................................................................................ 14 3.3. Additional time spent in the field...................................................................... 15 4. Respondent characteristics across surveys supported by LSMS+ versus LSMS‑ISA..................................................... 16 Section 2 Labor outcomes...................................................................................... 23 1. Emphasizing self‑reporting of labor in surveys............................ 24 2. Labor outcomes covered in the LSMS+ supported surveys (2016‑2020) ............................................................................................. 26 2.1. Participation in different labor activities across LSMS+ supported surveys ........................................................................................... 26 2.2. Hours and earnings............................................................................................... 29 2.3. Non‑Market Activities........................................................................................... 30 3. Respondents’ engagement in multiple economic activities ................................................................................................ 32 4. Individual, household and geographic correlates of labor participation................................................................................................... 37 ii Section 3 Asset ownership and rights............................................. 43 1. The importance of individual‑level interviews in measuring asset ownership/rights........................................................ 44 2. LSMS+ survey modules on asset ownership.................................. 46 2.1. Asset classes............................................................................................................. 46 2.2. Interview approach................................................................................................ 47 3. Land ownership and rights, and livestock ownership, across the LSMS+ supported surveys....................................................... 48 3.1. What patterns emerge across men and women? ........................................ 48 3.2. Bundles of land ownership and rights............................................................ 53 3.3. Reporting discrepancies among couples....................................................... 56 3.4. Livestock ownership in Ethiopia....................................................................... 60 4. Financial account and mobile phone ownership ........................ 61 4.1. Individual‑level estimates...................................................................................... 61 4.2. Does individual‑level data collection affect household estimates of mobile and financial account ownership?.................................... 62 Annex 1............................................................................................................................ 66 Annex Table A1. Tanzania.................................................................................. 68 Annex Table A2. Malawi...................................................................................... 69 Annex Table A3. Ethiopia................................................................................... 70 iii © World Bank Executive Summary Monitoring progress towards several targets of the Sustainable Development Goals (SDGs), across poverty reduction, agriculture, gender, employment, and inequality, require sex‑disaggregated data on asset ownership, labor, time use, and roles in family enterprises. Expanding survey data collection on these topics is a priority for the World Bank, as expressed through the World Bank’s Household Survey Strategy, Gender Strategy, and the 18th replenishment of the International Development Association (IDA18), which committed to launch pilot data collection in at least six IDA countries to “gather direct respondent, intra‑household level information on employment and assets.”1 1 International Development Association (IDA), 2016. IDA18 Special Theme: Gender and Development. World Bank. 1 © World Bank Against this background, the World Bank Living In its current phase, the LSMS+ has built on Standards Measurement Study‑Plus (LSMS+) the multi‑topic survey questionnaire design program was established in 2016 to support survey that was pioneered by the LSMS and has data production and methodological research been supporting national statistical offices activities – to improve the availability and quality of in select IDA countries in operationalizing individual‑disaggregated survey data in low‑ and the latest international recommendations for middle‑income countries on key dimensions of individual‑disaggregated survey data collection men’s and women’s economic opportunities and on asset ownership and labor. The latter involves welfare. The LSMS+ is part of the Living Standards administration of individual‑level survey modules Measurement Study (LSMS), which over the last that are administered in private to household four decades, has provided technical assistance to members age 18 and older and that focus on national statistical offices globally on designing and work and employment, non‑farm enterprises, implementing high‑quality, multi‑topic household and ownership of and rights to selected physical surveys. These surveys have been used extensively and financial assets, including, at a minimum, in policy making and research on a wide range of dwelling and non‑dwelling land, financial assets topics, including poverty, consumption and income and mobile phones. The approach of surveying inequality, employment, non-farm enterprises, multiple individuals per household captures agriculture, education, and health, among others. intra‑household dynamics in labor and economic decision‑making — the understanding of which is critical for designing policy around improving economic opportunities, particularly for women. Besides data production, the LSMS+ also supports methodological survey research activities to improve the foundations of individual‑disaggregated data collection in household surveys. 2 © World Bank In Sub‑Saharan Africa, between 2016‑2020, the This report is targeted towards several LSMS+ program supported individual‑level data audiences interested in the relevance of collection in line with the aforementioned principles individual‑level, self‑reported data collection as part of Malawi Integrated Household Panel on highlighting economic constraints and Survey (IHPS) 2016, Tanzania National Panel Survey opportunities for men and women. This includes (NPS5) 2019‑20 and Ethiopia Socioeconomic national statistical offices (NSOs) interested Survey (ESS4) 2018‑19 ‑ all of which have been in adopting LSMS+ recommendations on supported by the LSMS‑Integrated Surveys on individual‑disaggregated survey data collection Agriculture Initiative (LSMS-ISA) since 2008. on asset ownership and labor market outcomes, This report presents the results from the as well as data users across researchers and analysis of the data from these surveys, with policymakers who are interested in the findings on a focus on differences across men and women gender inequalities that emerge from more granular in labor market outcomes and asset ownership data collection. Relevant for country NSOs and and rights. The report also highlights important survey practitioners as well, a companion report, areas of analysis that can be further explored LSMS+ Program Overview and recommendations with these surveys. for improving individual‑disaggregated data on asset ownership and labor outcomes, provides a more detailed overview of the LSMS+ program, including guidance and details on implementation for LSMS+ supported modules.   3 Key findings regarding labor and asset ownership and rights to assets include: Labor + + Gender gaps in labor market Regarding non‑market work, the outcomes can be better LSMS+ supported surveys reveal understood with multiple greater time spent in particular individuals reporting within by women, but also by men, households, as opposed to the for water and fuel collection standard survey practice that in comparison to other national allow for proxy respondents surveys conducted in these reporting on behalf of other countries under the World Bank household members. The Living Standards Measurement data from Ethiopia, Malawi and Study – Integrated Surveys on Tanzania reveal significant gender Agriculture (LSMS‑ISA) program. gaps across different areas of employment. Comparisons by Kilic et al. (2020a)2 of the Malawi IHPS/LSMS+ and the concurrently‑conducted Fourth Integrated Household Survey (IHS4), show that the IHS4 leads to significantly lower reporting of employment across a range of wage and self‑employment activities, with stronger effects for women and for a longer (12‑month) recall period. 2 Kilic, Talip, Goedele Van den Broeck, Gayatri Koolwal, and Heather Moylan. 2020. “Are You Being Asked? Impacts of Respondent Selection on Measuring Employment.” World Bank Policy Research Working Paper 9152. 4 Asset ownership and rights + + The findings from Malawi, In Ethiopia, an additional module Tanzania, and Ethiopia highlight on large livestock ownership the importance of separately also revealed significant gender eliciting information on these differences, particularly in rural constructs and identifying areas (about 58 percent of exclusive versus joint ownership. rural women reported owning large livestock, compared to 64 ‑ For example, even when percent of rural men). Livestock respondents report ownership in rural areas was also exclusive ownership of land, more likely to be joint. economic ownership and decision‑making over the proceeds from hypothetical + land sales is more likely to On mobile phones and financial be joint or distributed across accounts, ownership tends to multiple household members. be mainly exclusive as opposed ‑ There are significant gender to joint. Greater shares of differences in land ownership respondents across countries do and rights. In Tanzania, own a mobile phone, although women are significantly less they are mostly concentrated in likely than men to exclusively urban areas, and men are also own dwelling land. And while significantly more likely than a greater share of women women to own one. Gender are likely to exclusively own disparities in mobile phone dwelling and non‑dwelling land ownership also widen in rural in Malawi, as well as dwelling areas. land in Ethiopia, women are significantly less likely than men among landowners + to have rights to sell and When comparing with other bequeath. national surveys conducted in the countries, the individual‑level ‑ Across countries, the interview approach in LSMS+ individual interview approach supported surveys is associated also reveals substantial with a higher share of intra‑household agreement households reporting financial over ownership and rights account ownership across all constructs. countries, and mobile phone ownership in rural areas for Tanzania and Malawi. 5 © World Bank Section 1 Introduction and LSMS+ overview 1. How can we better capture men’s and women’s economic opportunities in surveys? 2. The LSMS+ Program: addressing key data gaps, and prioritizing individual interviews 3. LSMS+ implementation in SSA (2016‑2020) 3.1 Share of eligible respondents interviewed for modules on asset ownership 3.2 Extent of self‑reporting in modules on education, health, and labor 3.3 Additional time spent in the field 4. Respondent characteristics across surveys supported by LSMS+ versus LSMS‑ISA Section 1 / Summary 7 1. How can we better capture men’s and women’s economic opportunities in surveys? An improved understanding of men’s and women’s International momentum has built around economic opportunities, through within‑household, improving individual‑level survey data collection self‑reported, individual‑level data, is instrumental to better highlight economic opportunities and for the accurate targeting and design of economic constraints for men and women. Several targets policies — including the expansion of financial of the Sustainable Development Goals (SDGs), services, land reforms, as well as social protection across poverty reduction, agriculture, gender, policies aimed at reducing longstanding employment, and inequality, also hinge on economic inequalities between men and women. improved sex‑disaggregated data across asset Understanding the within‑household distribution ownership, labor, and time use, as well as roles of labor, as well as the ownership and rights of in family enterprises (Annex 1). Following this different assets such as land, financial accounts momentum, the World Bank Living Standards and mobile phones, is key for understanding Measurement Study‑Plus (LSMS+) program was how households make collective decisions in established in 2016 to enhance the availability and responding to policy interventions. quality of individual‑disaggregated survey data collected in low‑ and middle‑income countries on Nationally representative, multi‑topic household key dimensions of men’s and women’s economic surveys, while well positioned to examine opportunities and welfare.3 socioeconomic and demographic factors associated with economic outcomes, tend to use standard survey approaches that often mask true outcomes for men and women, as well as intra‑household differences. This includes the use of proxy reporting, as well as interviewing 3 For more information on LSMS+, please visit: https://www.worldbank. respondents in non‑private settings as opposed to org/lsmsplus. LSMS+ has been established with grants from the one‑on‑one, which — especially in lower‑income Umbrella Facility for Gender Equality Trust Fund, the World Bank Trust and more traditional contexts, as well as areas Fund for Statistical Capacity Building, and the International Fund for Agricultural Development, and is implemented by the World Bank where work is more seasonal — can lead to the Living Standards Measurement Study (LSMS) Team, in collaboration under‑ or misreporting of key economic roles with the World Bank Gender Group and partner national statistical offices. The program leveraged existing World Bank partnerships among household members, including women. with (1) United Nations Evidence and Data for Gender Equality Land ownership may also not necessarily equate (EDGE) Project on methodological experimentation and international guidelines on measuring asset ownership and control from a gender to rights over land, and these distinctions can perspective, and (2) the ILO, FAO, the Data2X Project and the Hewlett be obscured when individuals do not report for Foundation on methodological experimentation for operationalizing the 19th ICLS definitions of work and employment, with a focus on themselves. subsistence agriculture. 8 LSMS+ Program in Sub-Saharan Africa © World Bank 2. The LSMS+ Program: Individual addressing disaggregated data help reveal economic key data gaps, inequalities among and prioritizing men and women individual interviews Over the last decade, national statistical offices The LSMS+ program is the latest in this series of (NSOs) in Ethiopia, Malawi and Tanzania collaborations with the Central Statistical Agency have been receiving financial and technical of Ethiopia, Malawi National Statistical Office and assistance through the World Bank Living Tanzania National Bureau of Statistics on improving Standards Measurement Study – Integrated the quality of data collection in multi‑topic surveys. Surveys on Agriculture (LSMS‑ISA) initiative for Within these countries, LSMS+ supported the design and implementation of multi‑topic, individual‑level data collection as part of Malawi nationally‑representative, longitudinal Integrated Household Panel Survey (IHPS) 2016, household surveys and for testing and adopting Tanzania National Panel Survey (NPS5) 2019‑20 methodological innovations in survey data and Ethiopia Socioeconomic Survey (ESS4) collection. 2018‑19 ‑ all of which have been supported by the LSMS‑Integrated Surveys on Agriculture Initiative (LSMS-ISA) since 2008. Section 1. Introduction and LSMS+ overview 9 Box 1.1 presents the overview of LSMS+ supported interviews if they were 18 years and above.4 Along data collection in these three countries. Additional with a focus on self‑reporting, LSMS+ supported nationally representative surveys that have been surveys have worked towards (i) conducting or are being supported by the LSMS+ include individual interviews in private, and if possible the Cambodia Socio‑Economic Survey 2019‑20 simultaneously across different household and Sudan Labor Market Panel Survey 2021. The members, and (ii) ensuring as much as possible a findings from Cambodia are also available in a gender match between the interviewers and the companion report, LSMS+ Program in Cambodia: respondents (see Kilic and Moylan, 2016, for more Findings from individual‑level data collection on details about the approach).5 In each country, the labor and asset ownership. In its current phase, LSMS+ has financed the survey implementation costs LSMS+ has been supporting the implementation of for collecting additional information from individuals individual‑level survey modules (i.e. the individual that receive the new modules. The program has questionnaire) that is administered to adult also provided direct technical assistance to the household members in private to elicit information NSOs to ensure proper integration and successful on work and employment, non‑farm enterprises, implementation of the improved survey methods. and ownership of and rights to selected physical and financial assets, including, at a minimum, land, dwelling, financial assets and mobile phones. Any pre‑existing individual‑level survey modules on other topics, such as education and health, are 4 In case the head of household or his/her spouse is less than 18 years of also integrated into the individual questionnaire age, these individuals are still considered interview targets. 5 Talip Kilic and Heather Moylan. 2016. Methodological experiment on that is administered in each country. Respondents measuring asset ownership from a gender perspective (MEXA): technical are considered eligible for the LSMS+ individual report. World Bank. Box 1.1 LSMS+ supported national surveys in Sub‑Saharan Africa Malawi Tanzania Ethiopia Survey 2016 Integrated Household 2019‑20 Tanzania National 2018-19 Ethiopia Panel Survey Panel Survey Socioeconomic Survey (ESS 4) Implementing agency1 Malawi National Statistical Tanzania National Bureau of Ethiopia Central Statistical Office Statistics Agency Sample size for 2,508 households that had 1,184 households that had 6,770 households individual interviews been previously interviewed in been previously interviewed supported by LSMS+ 2013 and 2010 by the NPS in 2008‑09, 2010‑11, 2012‑13, and 2014‑15 Fieldwork period April 2016‑Jan 2017 Jan 2019‑Jan 2020 Sep 2018‑Aug 20192 Asset classes covered Dwelling and non‑dwelling Dwelling and non‑dwelling Dwelling and non‑dwelling in LSMS+ individual land (ownership and rights); land (ownership and rights); land (ownership and rights); questionnaire mobile phones and financial mobile phones and financial livestock; mobile phones and accounts (ownership) accounts (ownership) financial accounts (ownership) Additional modules Labor, education, health, food Labor, education, health Labor, education, health in LSMS+ individual insecurity questionnaire 1 Surveys also received technical and financial support from LSMS‑ISA and LSMS+; and using the World Bank Survey Solutions Computer‑Assisted Personal Interviewing (CAPI) platform. All surveys, as well as the sample size for individual interviews supported by LSMS+, are nationally representative. 2 For the ESS 4, the household questionnaire (which included the LSMS+ modules) was implemented between June‑August 2019. The agricultural questionnaire, which is separate from the LSMS+ modules, was implemented in rural areas from September 2018‑March 2019. 10 LSMS+ Program in Sub-Saharan Africa © Valentina Costa / World Bank 3. LSMS+ implementation in SSA (2016‑2020) 3.1. Share of eligible respondents interviewed for modules on asset ownership All respondents The asset modules in the surveys that have been self-reported in supported by the LSMS+ in Malawi, Ethiopia and the LSMS+ Tanzania covered ownership and rights of dwelling and non‑dwelling (primarily agricultural) land; assets modules mobile phones; and financial accounts. In Ethiopia, ownership of livestock was also covered. The assets modules were self‑reported, although there was some non‑response among eligible individuals, Table 1.1a shows that among households with as discussed below. The last section of this report non‑dwelling land, 94.7 percent of respondents discusses the assets questions and outcomes in Ethiopia were interviewed; these shares were across countries in detail. 78.5 percent in Tanzania and 82.9 percent in Malawi. The interview rates were somewhat higher At the individual‑level, Tables 1.1a and 1.1b present for dwelling land across countries. Table 1.1b also the share of eligible respondents that were shows that in Ethiopia, 95.8 percent of eligible successfully interviewed in the assets modules respondents were interviewed in the mobile — by comparing interviews conducted in these phone module; these shares were 80.2 percent in modules, where there was full self‑reporting, with Tanzania and 82.4 percent in the Malawi IHPS. the eligible respondent sample in the roster. In Malawi, because the number of adult interviews Overall, urban‑rural differences in interview rates was capped at four, the number of eligible were not very large, but there were substantial respondents in this table includes the household gender differences — particularly in Tanzania and head and/or spouse when available, as well as Malawi, where eligible women respondents were additional adult (age 18 and older) respondents up much more likely to be available for interviews than to four per household. eligible men. Section 1. Introduction and LSMS+ overview 11 Table 1.1a Share of eligible respondents interviewed in LSMS+ asset modules1 Malawi IHPS 2016 Tanzania NPS5 2019‑20 Ethiopia ESS4 2018‑19 Total Men Women Total Men Women Total Men Women (1) Among HH with non‑dwelling land: Number of eligible individuals Total 3,450 1,562 1,888 1,661 795 866 5,417 2,701 2,716 Rural 2,988 1,335 1,653 1,255 599 656 4,211 2,128 2,083 Urban 462 227 235 406 196 210 1,206 573 633 Share responding: Total 82.9 77.0 87.8 78.5 75.0 81.6 94.7 94.5 95.0 Rural 83.9 77.5 89.0 78.2 74.0 82.0 93.2 93.0 93.5 Urban 76.6 74.0 79.2 79.3 78.1 80.5 100.0 100.0 100.0 (2) Among HH with dwelling land: Number of eligible individuals Total 4,719 2,173 2,546 2,676 1,246 1,430 14,535 6,831 7,704 Rural 3,436 1,543 1,893 1,681 777 904 6,805 3,318 3,487 Urban 1,283 630 653 995 469 526 7,730 3,513 4,217 Share responding: Total 83.3 77.5 88.4 80.3 76.5 83.6 97.0 96.4 97.6 Rural 83.8 77.5 89.1 79.7 74.8 84.0 93.6 92.6 94.6 Urban 82.1 77.5 86.5 81.3 79.3 83.1 100.0 100.0 100.0 Table 1.1b Share of eligible respondents interviewed in LSMS+ asset modules1 Malawi IHPS 2016 Tanzania NPS5 2019‑20 Ethiopia ESS4 2018‑19 Total Men Women Total Men Women Total Men Women Number of eligible individuals Total 4,774 2,196 2,578 2,989 1,407 1,582 15,388 7,235 8,153 Rural 3,458 1,550 1,908 1,789 833 956 7,315 3,560 3,755 Urban 1,316 646 670 1,200 574 626 8,073 3,675 4,398 Share of eligible respondents interviewed across modules: (1) Assets: mobile phone ownership: Total 82.4 76.7 87.3 80.2 76.1 83.9 95.8 95.4 96.2 Rural 83.3 77.2 88.3 79.2 74.1 83.6 92.4 91.7 93.1 Urban 80.1 75.7 84.3 81.8 78.9 84.5 98.9 98.9 98.9 (2) Assets: financial account ownership: Total 82.4 76.7 87.3 * * * 96.9 96.2 97.4 Rural 83.3 77.2 88.3 * * * 93.4 92.4 94.4 Urban 80.1 75.7 84.3 * * * 100.0 100.0 100.0 1 For each country, the number of eligible adults from the roster (those 18 years and older, and the household head or spouse)) was compared with the interviews conducted in the assets modules. In Malawi, because the number of adult interviews was capped at four, the number of eligible respondents in this table includes the household head and/or spouse when available, as well as additional adult (age 18 and older) respondents up to four per household. 2 For the mobile phones module in Ethiopia, there were 120 respondents who refused to answer, and in this table were not counted as interviewed. 3 Within each of the assets modules, all respondents self‑reported. * In Tanzania, the financial accounts data file only includes details on those reporting owning a financial account. As a result, the share of eligible respondents who were interviewed cannot be constructed for this table, since eligible respondents who were interviewed but self‑reported that they did not own an account are not identifiable. Given the trends in Malawi and Ethiopia, however, the share of eligible respondents reporting for this module are likely to be similar to the mobile phone module. 12 LSMS+ Program in Sub-Saharan Africa © World Bank Focusing on the mobile phone module, at the household level, 98.7 percent of the 2,508 Malawi IHPS households completed at least one individual interview6; these shares were 99.8 percent for the Tanzania NPS 5 and 99.4 percent for the Ethiopia ESS4. Table 1.2 shows the within‑household interview success rate, breaking down the number of eligible adults versus the number of individual interviews completed, across countries. Across all households, regardless of the number of adults, 60 percent of the 4 randomly selected eligible adults per household were successfully interviewed in Malawi, 69 percent of all eligible adults in Tanzania, and 94 percent of all eligible adults in Ethiopia. The remaining share of households had more than one adult but failed to interview at least one of them, mostly concentrated in two‑person households where only one was successfully interviewed, or where two out of three eligible adults were interviewed. 6 For the remaining 1.3 percent of households, the reasons for non‑completion included (i) refusal due to the already lengthy household interview that had been completed; (ii) refusal due to the request to conduct the interviews in private, and (iii) loss of individual questionnaires due to Android tablet malfunction. Table 1.2 Distribution of LSMS+ households, by the number of adults interviewed in assets module1 Malawi IHPS 2016 Tanzania NPS5 2019‑20 Ethiopia ESS4 2018‑19 Total % Total % Total % Households Interviewed 2,476 1,182 6,727 All Eligible Adults 1,495 60% 812 69% 6,325 94% Interviewed 4 or more adults 115 5% 88 7% 672 10% 3 adults 209 8% 80 7% 979 15% 2 adults 872 35% 441 37% 3,340 50% 1 adult 299 12% 203 17% 1,334 20% Subset of Eligible Adults 981 40% 370 31% 402 6% Interviewed 3 out of 4 124 5% 78 7% 82 1% 2 out of 4 153 6% 43 4% 38 1% 1 out of 4 55 2% 7 1% 8 0% 2 out of 3 238 10% 83 7% 143 2% 1 out of 3 92 4% 19 2% 12 0% 1 out of 2 319 13% 140 12% 119 2% 1 For each country, the number of eligible adults from the roster (those 18 years and older, and the household head or spouse) was compared with the interviews conducted in the assets modules (specifically, the mobile phones module). Section 1. Introduction and LSMS+ overview 13 3.2. Surveying multiple Extent of self‑reporting individuals per in modules on education, household captures health, and labor intra-household dynamics in labor Table 1.3 shows that the share of eligible individuals interviewed across the labor, education and economic and health modules, and who self‑reported, varied across the LSMS+ supported surveys — although decision‑making within each country, the shares of self‑reporting eligible respondents were quite similar across the three modules. In Ethiopia, 68‑69 percent of men and 77‑78 percent of women self‑reported across there was therefore still some degree of proxy labor, education, and health; in Tanzania, these reporting among eligible respondents, and more so shares were 74‑75 percent of men and 84‑85 for men. However, as discussed in Section 2, proxy percent of women; and in Malawi, 77 percent of respondent reporting rates in the LSMS+ were men and 88 percent of women. Although the substantially lower than proxy reporting in similar LSMS+ implementation pushed for self‑reporting, (LSMS‑ISA) surveys in those countries. Table 1.3 Share of eligible respondents who self‑report in labor/health/education modules Malawi IHPS 2016 Tanzania NPS5 2019‑20 Ethiopia ESS4 2018‑19 Men Women Men Women Men Women (A) Number of eligible respondents covered in each module 1 Total 2,793 3,076 1,407 1,582 7,235 8,153 Rural 1,948 2,257 833 956 3,560 3,755 Urban 845 819 574 626 3,675 4,398 (B) Within (A): share of total eligible respondents self‑reporting2 Labor Total 0.77 0.88 0.75 0.85 0.69 0.78 Rural 0.78 0.89 0.74 0.85 0.71 0.74 Urban 0.76 0.86 0.77 0.85 0.67 0.82 Education Total 0.77 0.88 0.75 0.84 0.69 0.78 Rural 0.76 0.89 0.74 0.85 0.71 0.74 Urban 0.76 0.86 0.76 0.83 0.66 0.82 Health Total 0.77 0.88 0.74 0.85 0.68 0.77 Rural 0.78 0.89 0.73 0.85 0.71 0.73 Urban 0.75 0.85 0.75 0.84 0.65 0.81 1 For (A), the number of eligible respondents was the same across the labor, education and health modules (with the exception of Tanzania, where the health and labor modules had the same number of eligible respondents, but the education module had slightly fewer). For simplicity, the labor/health module numbers are presented for Tanzania. 2 (B) was calculated by separately merging the labor, education, and health modules with the household roster, calculating the number of eligible respondents who matched/were interviewed, and then calculating the share of those respondents who self‑reported. 14 LSMS+ Program in Sub-Saharan Africa © World Bank 3.3. Additional time spent in the field To get a better understanding of the additional For the IHPS, the field teams took, on average, costs of implementing individual interviews, the 4.51 days in an enumeration area that, on metadata extracted from the Survey Solutions average, contains 16 households. As mentioned CAPI application also allows for the calculation earlier, field teams for the IHPS tried to ensure of the amount of time enumerators spend on an as much as possible that each available adult interview. In the case of the Malawi, the IHPS household member was interviewed in private and the nationally representative 2016‑17 Malawi by an enumerator of the same sex, and that all Fourth Integrated Household Survey (IHS4) were private interviews for a given household were conducted concurrently using the same field conducted at the same time. For the IHS4, these teams. The IHS4 covered the same socioeconomic field teams took a total of 3.37 days to administer and demographic characteristics, as well as asset the IHS4 questionnaires to 16 households in each ownership and rights, as the IHPS but followed the enumeration area, with one enumerator visiting business‑as‑usual approach of interviewing the each household. As a result, in Malawi, field most knowledgeable household member(s) about teams needed approximately one extra day in an asset ownership and rights, as well as a greater enumeration area to conduct individual interviews. reliance on proxy reporting when the targeted respondent was not available. Because the surveys were conducted concurrently it allows for a true comparison of the additional time needed for LSMS+ individual interviews. Section 1. Introduction and LSMS+ overview 15 © World Bank 4. Respondent characteristics across surveys supported by LSMS+ versus LSMS‑ISA How do characteristics of respondents in the land ownership and rights as the IHPS, but only LSMS+ supported surveys compare with other asked these questions of one “most knowledgeable” recent surveys in these countries? In Table 1.4, respondent in each household on their and other respondent and household characteristics are members’ ownership and rights.7 presented spanning age, education, marital status, Comparisons of the two sets of surveys within household composition, household access to each country, however, do need to be conditioned infrastructure, as well as urban locality. on differences not only in survey design/ Characteristics of respondents in the earlier LSMS‑ISA implementation, but also time‑varying factors that surveys that were conducted these countries are can affect the interpretation of differences across presented for comparison and to provide additional the two sets of surveys within each country. This context (i.e., the Malawi IHS4 2016‑17, discussed is particularly the case for Ethiopia and Tanzania, earlier; the Tanzania National Panel Survey (NPS4) where the LSMS+ and LSMS‑ISA surveys for each 2014‑15; and the Ethiopia Socioeconomic Survey country were conducted 3‑4 years apart. (ESS3) 2015‑16). These three LSMS‑ISA surveys are Across countries/surveys, Table 1.4 shows that nationally representative and covered a similar range most of the respondent sample is rural, with about of socioeconomic and demographic characteristics 25‑30 percent of LSMS+ respondents living in urban as the LSMS+ supported surveys but followed areas. Some stark differences emerge across countries. more standard survey approaches including a greater reliance on proxy reporting when targeted 7 This comparison was the basis for Kilic et al. (2020b)’s examination, respondents were not available and conducting using both Malawi surveys, of the effects of individual interviews interviews in non‑private settings as opposed to vis‑à‑vis standard survey approaches in measuring asset ownership one‑on‑one.In the case of Malawi, the IHS4 2016‑17 and rights. also asked a similar line of questioning on agricultural 16 LSMS+ Program in Sub-Saharan Africa Table 1.4 Demographic and socioeconomic characteristics of individuals aged 18+, or household head/spouse: LSMS+ and comparison (LSMS‑ISA) surveys1 Ethiopia Tanzania Malawi LSMS+ (ESS4) ESS3 LSMS+ (NPS5) NPS4 LSMS+ (IHPS) IHS4 2018‑19 2015‑16 2019‑20 2014‑15 2016 2016‑17 Men Women Men Women Men Women Men Women Men Women Men Women HH head 0.66*** 0.22*** 0.64*** 0.21*** 0.64*** 0.20*** 0.64*** 0.19*** 0.67*** 0.21*** 0.70*** 0.25*** Age: 18‑24 0.25 0.25 0.24** 0.22** 0.22* 0.27* 0.31** 0.29** 0.30 0.29 0.28 0.28 Age: 25‑34 0.28** 0.30** 0.25*** 0.28*** 0.36*** 0.27*** 0.26 0.26 0.26 0.28 0.25*** 0.27*** Age: 45‑54 2 0.12** 0.11** 0.14 0.13 0.12 0.12 0.11* 0.15* 0.11 0.11 0.12*** 0.10*** Age: 55+ 0.16*** 0.13*** 0.18** 0.16** 0.15 0.18 0.15 0.14 0.13* 0.15* 0.14*** 0.16*** Ever attended 0.42*** 0.61*** 0.38*** 0.59*** 0.10*** 0.20*** 0.14*** 0.24*** 0.07*** 0.15*** 0.10*** 0.19*** school Years of 7.82 7.67 7.36 7.44 7.43 7.67 7.51 7.57 8.14*** 6.93*** 8.32*** 7.22*** school, if attended Married 0.63 0.62 0.64** 0.62** 0.53 0.49 0.51 0.50 0.67 0.67 0.68*** 0.62*** Separated/ 0.02*** 0.08*** 0.03*** 0.08*** 0.06*** 0.11*** 0.03*** 0.11*** 0.03*** 0.10*** 0.03*** 0.12*** divorced Widowed 0.01*** 0.10*** 0.02*** 0.13*** 0.01*** 0.10*** 0.01*** 0.11*** 0.01*** 0.11*** 0.01*** 0.05*** Number 0.38*** 0.32*** 0.20*** 0.12*** 1.00 0.91 0.66** 0.61** 0.67*** 0.47*** 0.56*** 0.36*** of months individual is away from HH HH size 5.42*** 5.19*** 6.72*** 6.33*** 6.29 6.13 5.88 5.94 5.47 5.45 5.03*** 4.93*** HH 0.68 0.72 1.12* 1.04* 0.81*** 0.95*** 0.81*** 0.96*** 0.82*** 0.98*** 0.86*** 1.03*** dependency ratio3 HH has 0.30*** 0.34*** 0.30*** 0.32*** 0.66 0.64 0.34 0.32 0.20** 0.17** 0.17*** 0.15*** electricity HH has piped 0.17*** 0.19*** 0.18** 0.20** 0.41 0.40 0.36 0.38 0.19** 0.16** 0.17*** 0.15*** water HH: walls 0.06*** 0.07*** 0.05 0.06 0.20 0.22 0.22 0.24 0.03 0.02 0.02 0.02 made of concrete Lives in urban 0.28*** 0.31*** 0.25*** 0.28*** 0.31 0.28 0.30 0.32 0.31*** 0.26*** 0.23*** 0.21*** area Observations 7,235 8,153 5,569 6,247 1,407 1,577 1,238 1,329 2,118 2,631 6,988 8,005 1 All estimates are weighted. Statistically significant differences between men and women, within each survey, are indicated by asterisks (***p<0.01, ***p<0.05, * p<0.10). 2 Excluded category is 35‑44. 3 Indicates dependency ratio of children and elderly. 70-75 of respondents % across countries live in rural areas Section 1. Introduction and LSMS+ overview 17 About 65 percent of respondents in the Tanzania LSMS+ supported surveys. Age distributions are LSMS+ live in electrified households, for example, quite similar across urban and rural areas, for all compared to only around 20 and 30 percent, three samples. Rural areas tend to have a greater respectively, in the Malawi and Ethiopia LSMS+. share of elderly, compared to urban areas. The Tanzania LSMS+ also has a larger share of Within each country, there were some notable households with piped water and with concrete differences across the LSMS+ and comparison survey construction. The Ethiopia LSMS+ has a larger samples. Some of these differences (in Ethiopia and share of respondents who never attended school — Tanzania, specifically) might be attributable to the time 61 percent for women and 42 percent for men, also a elapsed between these surveys. Within countries, all significant gap — compared to Tanzania and Malawi. three LSMS+ samples had greater access to electricity Among demographic characteristics, the Tanzania and piped water relative to their comparison surveys LSMS+ has a substantially lower share of married (with greater increases in the Tanzania LSMS+), as well men and women (53 and 49 percent, respectively), as the respondent spending greater time away from compared to the Ethiopia and Malawi LSMS+ the household. Within Malawi and Ethiopia, greater samples where more than 60 percent of men and shares of the LSMS+ sample lived in urban areas.8 women were married. Figure 1.1 below also presents the age distribution of respondents across the three 8 Average annual nonfood per capita consumption expenditure in Ethiopia LSMS+ households was also significantly higher relative to the ESS3 conducted in 2015/16. The differences were not significant in the Malawi LSMS+ compared to the Malawi IHS4 conducted in the same year. Figure 1.1 Age Distribution of LSMS+ Eligible Sample Ethiopia Malawi Tanzania Man − Rural Man − Urban 0.04 0.04 0.03 0.03 density density 0.02 0.02 0.01 0.01 0.00 0.00 0 25 50 75 100 0 25 50 75 100 age age Woman − Rural Woman − Urban 0.04 0.04 0.03 0.03 density density 0.02 0.02 0.01 0.01 0.00 0.00 0 25 50 75 100 0 25 50 75 100 age age 1 Cross-country comparison of ages by rural/urban and gender. Sample weights included. 18 LSMS+ Program in Sub-Saharan Africa © Valentina Costa / World Bank Respondents in the LSMS+ supported surveys in than men. Particularly in Tanzania and Malawi, all three countries were also less likely to report women were also more likely to live in households never having attended school. The demographic with a greater dependency ratio. breakdown of households (age distribution of Across the LSMS+ supported surveys, we also respondents, marital status, household size and observe that men and women in rural Malawi, as dependency ratio) was not very different across the well as Ethiopia, often reside in households with LSMS+ and comparison survey within each country. significantly different household consumption Table 1.4 also shows that differences between men (as reflected by log annual household non‑food and women within each sample are statistically consumption per capita, in Figure 1.2). In particular, significant, although the magnitude and direction Figure 1.2 presents trends for men and women, of differences vary by country. Within countries, by age groups and urban/rural, for each LSMS+ the direction of gender differences was more or sample, with 95 percent confidence intervals of log less consistent across the LSMS+ and comparison consumption per capita. In both rural Malawi and samples. In Tanzania and Malawi, women were Ethiopia (urban and rural), men of ages from about significantly more likely to never have attended 25 up to 40 years of age are significantly more school, whereas men were more likely in Ethiopia.9 likely to live in households with higher average There are some differences in the incidence of men non‑food consumption, compared to women in and women living in households with greater access the same age range. In the urban Ethiopia sample, to electricity and piped water, and in urban areas, significant differences persisted through much although these differences are small (greater share older age (up until about age 50) as well. of women in Ethiopia and lower in Malawi, with no difference in Tanzania). Men were significantly Among men and women more likely across countries to spend greater time away from the household. Among demographic aged 25-40 characteristics, men were much more likely to be designated as the household head. While the majority of individuals are married in all three samples, women in each sample were significantly more likely to be widowed or separated/divorced man are more likely to live in households 9 The difference in years of schooling, however, was not significantly with higher non-food consumption different across men and women generally, except in the case of Malawi. (in rural Malawi and urban/rural Ethiopia). Section 1. Introduction and LSMS+ overview 19 Figure 1.2 Annual household consumption per capita, by men and women Men Women Malawi − Rural Malawi − Urban 6.8 8.5 consumption per capita consumption per capita log annual HH log annual HH 6.6 8.0 6.4 7.5 20 30 40 50 60 70 20 30 40 50 60 70 age age Ethiopia − Rural Ethiopia − Urban 6.6 8.25 consumption per capita consumption per capita 6.5 8.00 log annual HH log annual HH 6.4 7.75 6.3 7.50 6.2 7.25 6.1 20 30 40 50 60 70 20 30 40 50 60 70 age age Tanzania − Rural Tanzania − Urban 12.0 consumption per capita consumption per capita 11.0 log annual HH log annual HH 11.5 10.5 11.0 20 30 40 50 60 70 20 30 40 50 60 70 age age 1 The figures show estimated lowess curves with 95% confidence intervals of log annual household consumption per capita for the LSMS+ sample. 20 LSMS+ Program in Sub-Saharan Africa © World Bank Section 1 / Summary + + + + The World Bank Living Within Sub‑Saharan Individual‑level survey Among eligible Standards Measurement Africa, the Malawi IHPS modules were administered respondents who were Study‑Plus (LSMS+) program (2016), Tanzania NPS5 on work and employment, interviewed (ranging from has been established (2019‑20), and Ethiopia non‑farm enterprises, and 73 percent of the randomly in 2016 to improve the ESS4 (2018‑19) have been ownership of and rights selected respondents availability and quality of supported under the LSMS+ to selected physical and in Malawi, 80 percent individual‑disaggregated program over the period of financial assets, including of all eligible adults in survey data collected in 2016‑2020. Respondents dwelling land, non‑dwelling Tanzania, and 96 percent low‑ and middle‑income were considered eligible land, financial assets of all eligible adults in countries on key for the LSMS+ individual and mobile phones. Ethiopia), all self‑reported dimensions of men’s interviews if they were 18 Pre‑existing individual‑level in the assets modules. In and women’s economic years and above. survey modules on other the labor, education and opportunities and welfare. topics, such as education health modules, there was and health, are also some degree of proxy integrated into the individual reporting, albeit lower questionnaire. compared to the previous rounds of the LSMS‑ISA surveys conducted in these countries, which followed more standard survey approaches including a greater reliance on proxy respondents. LSMS+ respondents that live in electrified households 65 20 30 Tanzania % Malawi % Ethiopia % Section 1. Introduction and LSMS+ overview 21 © Scott Wallace / World Bank 22 LSMS+ Program in Sub-Saharan Africa Section 2 Labor outcomes 1. Emphasizing self‑reporting of labor in surveys 2. Labor outcomes covered in the LSMS+ supported surveys (2016‑2020) 2.1. Participation in different labor activities across LSMS+ supported surveys 2.2. Hours and earnings 2.3. Non‑Market Activities 3. Respondents’ engagement in multiple economic activities 4. Individual, household and geographic correlates of labor participation Section 2 / Summary Section 2. Labor outcomes 23 1. Emphasizing self‑reporting of labor in surveys of eligible respondents who reported by proxy in the LSMS+ and comparison/LSMS‑ISA surveys in those countries, which did not necessarily rely on self‑reporting.12 Proxy reporting in the LSMS+ surveys tends to be lower than the comparable As discussed in Section 1, multi‑topic household LSMS‑ISA surveys in Malawi and Tanzania, and surveys with a focus on work and employment particularly the case in Malawi (the proxy reporting often rely on standard approaches such as proxy variable was not available in the Ethiopia LSMS‑ISA). reporting and non‑private interviews to construct In the Malawi LSMS+, the share of proxy reporting labor statistics for all members of the household. in labor for the full sample was 21 percent for men While often easier to implement, these approaches and 11 percent for women, compared with 48 and can lead to measurement error in understanding 29 percent of men and women, respectively, in both men’s and women’s labor outcomes across the Malawi IHS4. In Tanzania, 25 percent of men households. A lower incidence of accurate, and 15 percent of women in the LSMS+ sample self‑reported data can also mask intra‑household were reported by proxy, compared with 30 percent dynamics in labor and economic decision‑making, of men and 20 percent of women in the NPS4. an understanding of which is critical for designing Labor proxy reporting in Ethiopia was also similar policy around improving economic mobility, and to the Tanzania LSMS+; around 28 percent for notably for women who typically face poorer men and 19 percent for women. Table 2.1 also economic opportunities. These concerns have also presents the share reporting by proxy for the been reflected in international recommendations youth (up to age 24) age sample, which as seen in on the on improving the measurement of work and Section 1 is about one‑quarter to nearly one‑third employment in surveys — including changes to of the total sample, depending on the country. international definitions of work and employment Information on youth labor market outcomes is key under the 19th International Conference of Labour to understanding school‑to‑work transitions, as Statisticians (19th ICLS) to better capture total (paid well as other important policies around education, and unpaid) work burdens for men and women10 skills training and employment, but information can as well as ILO recommendations on improving the be more difficult to collect directly for this group.13 measurement of informal employment.11 While proxy reporting for youth goes up relative While full self‑reporting was not achievable in the to the overall sample, it is still lower in the LSMS+ LSMS+ supported surveys, enumerators focused compared in the comparable surveys for those on ensuring private, self‑reported interviews countries. This gap also widens considerably for the whenever possible. Table 2.1 presents the share youth sample in Malawi. 10 The ICLS, which meets every five years, affects how country labor 12 In Ethiopia and Malawi, the shares reporting by proxy in the labor force surveys — guided by the ILO — are designed and undertaken. module are presented (although in Ethiopia, the ESS3 did not have a Also see International Labour Organization (ILO). 2013. Resolution I self‑reporting/proxy question). In Tanzania, since the NPS4 only had this concerning statistics of work, employment and labour underutilization. question in the health module, reporting in the LSMS+ health module is 19th International Conference of Labour Statisticians, Geneva. presented — within the LSMS+, as seen as Section 1, the incidence of 11 International Labour Organization (ILO). 2018. Women and Men in the proxy reporting was similar across modules for each country. informal economy: A statistical picture (3rd edition). Geneva. https:// 13 See, for example, Desiere, Sam and Valentina Costa. 2019. www.ilo.org/wcmsp5/groups/public/‑‑‑dgreports/‑‑‑dcomm/documents/ “Employment Data in Household Surveys Taking Stock, Looking publication/wcms_626831.pdf Ahead.” World Bank Policy Research Working Paper No. 8882. 24 LSMS+ Program in Sub-Saharan Africa Table 2.1. Share of men and women reporting by proxy, in the LSMS+ and comparison (LSMS‑ISA) surveys Ethiopia Tanzania Malawi (labor module)2 (health module)2 (labor module) LSMS+ (ESS4) ESS3 LSMS+ (NPS5) NPS4 LSMS+ (IHPS) IHS4 2018‑19 2015‑16 2019‑20 2014‑15 2016 2016‑17 Men Women Men Women Men Women Men Women Men Women Men Women Share of eligible sample reporting by proxy (full sample): 1 Total 0.28 0.19 ‑ ‑ 0.25 0.15 0.30 0.20 0.21 0.11 0.48 0.29 Rural 0.23 0.21 ‑ ‑ 0.25 0.14 0.31 0.20 0.20 0.09 0.45 0.28 Urban 0.33 0.18 ‑ ‑ 0.25 0.15 0.29 0.20 0.22 0.13 0.61 0.34 Share of eligible sample reporting by proxy (aged 18‑24): 1 Total 0.43 0.27 ‑ ‑ 0.38 0.24 0.45 0.26 0.30 0.18 0.64 0.45 Rural 0.44 0.31 ‑ ‑ 0.38 0.25 0.47 0.26 0.30 0.18 0.62 0.43 Urban 0.42 0.25 ‑ ‑ 0.39 0.22 0.40 0.26 0.30 0.20 0.71 0.51 Number of eligible respondents (full sample): Total 6,967 7,945 5,745 6,462 1,386 1,565 1,216 1,319 2,723 3,027 12,103 14,063 Rural 3,292 3,547 3,871 4,151 814 945 714 759 1,898 2,224 9,594 11,350 Urban 3,675 4,398 1,874 2,311 572 620 413 461 825 803 2,509 2,713 1 The eligible sample included those aged 18 and older, or who were either the household head or spouse of the household head. 2 The self/reporting proxy variable was not available in the labor module for the Ethiopia ESS3. In Tanzania NPS4, only the health module had a self‑reporting/ proxy variable. Quantitatively assessing whether proxy versus with stronger effects for women and for a longer self‑reporting leads to significant differences (12‑month) recall period. The study also finds links in outcomes (and where proxy reporting may between underreporting and household wealth, have a greater effect on reporting) relies on an proxy reporting, as well as potential difficulties experimental setup with different survey associated with interpreting/answering questions approaches conducted in the same on household non‑farm enterprises. A key aim of timeframe and country context. In Malawi, this report, by providing a descriptive analysis of the concurrently‑conducted IHS4 alongside the survey sample, men’s and women’s responses the LSMS+ supported IHPS, and the broader to questions, is to motivate future work and approach in the IHS4 to surveying household research in this area across countries by users members on labor — allowing for proxy reporting interested in the data.  and non‑private interviews, for example, as The following discussion examines trends across well as not ensuring a gender match between the LSMS+ supported surveys in Sub‑Saharan respondents and enumerators — was the basis Africa more broadly, exploring participation for Kilic et al, (2020a)’s examination of how the across different age groups, multiple employment LSMS+ recommendations on individual‑level data activities, work in non‑market activities, and collection can better elicit labor outcomes for men regression analysis to understand what variables and women.14 Comparing the IHPS with the IHS4, correlates with participation. We also discuss, the study finds that the use of proxy respondents where possible and interpretable, differences and non‑private interviews in the IHS4 leads to in labor market outcomes between the LSMS+ significant underreporting of employment across survey and comparison surveys conducted under a range of wage and self‑employment activities, the LSMS‑ISA program (see Section 1). Again, the 14 Kilic, Talip, Goedele Van den Broeck, Gayatri Koolwal, and Heather Malawi survey provides a unique perspective since Moylan. 2020a. “Are You Being Asked? Impacts of Respondent the LSMS+ survey/IHPS and IHS4 were implemented Selection on Measuring Employment.” World Bank Policy Research concurrently, accounting for time‑varying factors Working Paper 9152. that could otherwise affect how differences across survey types are interpreted. Section 2. Labor outcomes 25 2. Labor outcomes covered in the LSMS+ supported surveys (2016‑2020) © World Bank 2.1. Participation in different The LSMS+ supported surveys included labor modules include questions on individual labor activities across activities over the last seven days and 12 months. LSMS+ supported surveys This section focuses on the sample of eligible individuals who self‑reported in the labor modules (as discussed in Section 1), and on three kinds of Table 2.2 presents the share of individuals working activities: across different activities. Certain questions are only available for the 12‑month period and, as a. any agricultural activities (production for own mentioned above, some activity categories vary consumption or sale); slightly across countries. In all three countries, the b. running or working in a non‑farm enterprise; gender gap is statistically significant for the share and of individuals in wage employment. In Tanzania, for example, where the share of men and women c. wage employment. Within wage employment, in wage employment is higher compared to the annual hours and earnings were also covered other two countries, 48 percent of men were across the three surveys. in wage employment in the last 12 months, There are some country‑specific activities — in compared to 22 percent of women. The shares Malawi, for example, LSMS modules asked of men and women in non‑farm enterprise (NFE) questions about ganyu labor which is short‑term work are not significantly different in Ethiopia and informal farming labor not on the household’s own Tanzania, although there are significant differences farm. In Ethiopia, LSMS+ modules also asked about in Malawi (particularly for running an NFE in the casual/temporary employment. last 12 months for urban and rural respondents, as well as rural respondents for the last 7 days). For Ethiopia, 83 percent of men also report being in any agricultural work compare to 55 percent for women. These overall country descriptive statistics highlight how gender gaps differ depending on the activity type but also by country. 26 LSMS+ Program in Sub-Saharan Africa How does participation vary by age? Figure 2.1 men in wage or salaried employment in Tanzania reports locally weighted regressions of labor and Ethiopia. In urban Malawi, older men and participation in the last 7 days, across activities women are also more likely to be engaged in NFE by age in rural and urban areas.15 The shares are activity. Further research is needed to understand relatively stable across ages with some groups these dynamics in how age‑associated gender gaps exhibiting an inverse‑U pattern — in particular, urban evolve for salary work. 15 Trends were also similar for labor participation in the last 12 months. Table 2.2. Share of Activities Across LSMS+ Supported Surveys, by Reference Periods Last 7 days Last 12 months Agr. Agr. NFE NFE Wage Agr. Agr. NFE NFE Wage (any) (CFW) (ran/ma‑ (suppor‑ employ‑ (any) (CFW) (ran/ma‑ (suppor‑ employ‑ naged) ting) ment naged) ting) ment Malawi IHPS 2016/2017 Urban Men 0.12 0.05 0.18 0.05 0.34 0.32 - 0.23 0.12 0.39 Women 0.15 0.06 0.20 0.06 0.15 0.38 - 0.31 0.10 0.18 T‑test 0.21 0.45 0.30 0.61 0.00 0.03 - 0.00 0.42 0.00 Rural Men 0.47 0.15 0.12 0.03 0.11 0.90 - 0.19 0.08 0.14 Women 0.48 0.21 0.09 0.03 0.03 0.90 - 0.14 0.06 0.03 T‑test 0.95 0.00 0.00 0.91 0.00 0.79 - 0.00 0.08 0.00 Ethiopia ESS4 2018/2019 Urban Men 0.13 - - - 0.27 - - 0.28 0.02 0.37 Women 0.07 - - - 0.13 - - 0.24 0.03 0.18 T‑test 0.00 - - - 0.00 - - 0.08 0.31 0.00 Rural Men 0.83 - - - 0.03 - - 0.09 0.01 0.06 Women 0.55 - - - 0.01 - - 0.10 0.01 0.02 T‑test 0.00 - - - 0.00 - - 0.70 0.52 0.00 Tanzania NPS5 2019/2020 Urban Men 0.22 - - - 0.43 0.33 0.08 0.29 0.01 0.55 Women 0.20 - - - 0.20 0.31 0.10 0.24 0.02 0.32 T‑test 0.64 - - - 0.00 0.58 0.54 0.27 0.09 0.00 Rural Men 0.63 - - - 0.27 0.83 0.22 0.13 0.01 0.48 Women 0.60 - - - 0.11 0.81 0.32 0.10 0.02 0.22 T‑test 0.50 - - - 0.00 0.59 0.01 0.34 0.83 0.00 1 All estimates use household sampling weights; only self‑reporting sample included. “-” indicates data was not available for that variable. 2 Ethiopia does report participation in NFE for the past 7 days, however, it did not divide this measurement into manager and supporting role thus not included in this table. Section 2. Labor outcomes 27 Figure 2.1 Participation Rate for Selected Activities in the Last 7 days Agri NFE Salary Agri NFE Salary Men Men Men Women Women Women Malawi − Urban Malawi − Rural 0.75 0.8 Share of respondents Share of respondents 0.6 0.50 0.4 0.25 0.2 0.00 0.0 20 30 40 50 60 70 20 30 40 50 60 70 age age Tanzania − Urban Tanzania − Rural 0.75 Share of respondents Share of respondents 0.8 0.50 0.25 0.4 0.00 0.0 -0.25 20 30 40 50 60 70 20 30 40 50 60 70 age age Ethiopia − Urban Ethiopia − Rural 0.4 Share of respondents Share of respondents 0.75 0.3 0.50 0.2 0.1 0.25 0.0 0.00 20 30 40 50 60 70 20 30 40 50 60 70 age age 1 All estimates use household sampling weights; only self-reporting sample included. 95% confidence intervals also presented for each curve. 2 For Malawi, individuals working in an NFE either as a manager/owner or supporting worker were aggregated for NFE work. 28 LSMS+ Program in Sub-Saharan Africa 2.2. Figure 2.2 Annual hours in main wage employment (conditional on working), Hours and earnings men and women Men Rural Men Urban On wage employment, Figures 2.2 and 2.3 present Women Rural Women Urban the distribution of yearly hours and earnings across men and women, as well as in rural/urban areas. Ethiopia Figure 2.2 shows that urban workers work longer hours on average annually. In Ethiopia, gender 0.00080 differences are not statistically significant across rural and urban areas. In rural Malawi, and in 0.00060 Tanzania in particular, men are significantly likely to density work more in wage employment. 0.00040 Figure 2.4 also presents the log annual earnings for Malawi and Ethiopia. As expected, within each 0.00020 country, urban men have higher earnings than urban women; these differences were statistically 0.00000 significant in Malawi. The similarity in earnings 0 1000 2000 3000 4000 5000 distribution for some groups (rural Malawi, for annual hours example) is quite interesting since the annual hours distributions are different. In Kilic et al. (2020a), Malawi when comparing the annual hours and earnings distributions of the IHPS and IHS4 samples in 0.00080 Malawi, women’s conditional annual hours worked in wage employment are higher in the IHS4 0.00060 compared to the IHPS, and conditional hours density and earnings in ganyu for both men and women 0.00040 are also higher in the IHS4. However, with the exception of men’s hours in ganyu, the differences 0.00020 are weakly significant. The choice of survey approach may also matter more with longer recall 0.00000 periods as well as seasonality of different activities 0 1000 2000 3000 4000 5000 among this population. annual hours Tanzania 0.00080 0.00060 density 0.00040 0.00020 0.00000 0 1000 2000 3000 4000 5000 annual hours 1 All estimates use household sampling weights; only self-reporting sample included. 2 Distribution of annual hours work in main salary occupation. Only answered for those with a salary job. 3 Gender differences in distribution across all groups (country and rural/ urban) are significant at p<0.01 level. Section 2. Labor outcomes 29 Figure 2.3 Annual earnings in main wage employment (conditional on working), men and women Men Rural Men Urban Women Rural Women Urban Ethiopia Malawi 0.40 0.40 0.30 0.30 density density 0.20 0.20 0.10 0.10 0.00 0.00 0 5 10 15 20 0 5 10 15 20 log annual earnings in local currency log annual earnings in local currency 1 All estimates use household sampling weights; only self-reporting sample included. 2 Annual earnings for Tanzania were not yet available. Earnings include cash and in-kind transfers combined. 3 Gender differences in distribution across are significant at the p<0.01 level for urban Ethiopia, and at the p<0.05 level for urban Malawi. © World Bank 2.3. Non‑Market Activities In Tanzania as well as Ethiopia, women spend disproportionately more time in non‑market work, with the exception of agriculture in own‑use production (reflected in Table 2.3 among those who are engaged in agricultural work). In Ethiopia, the gap between men and women is also much larger in rural areas. The average hours spent in water collection of women is more than twice that of men in rural areas, and about 1.5 times in urban areas. Table 2.3 also compares the data on non‑market work as elicited in LSMS+ Unpaid work: supported surveys in Malawi and Ethiopia the LSMS+ highlights versus their LSMS‑ISA counterparts (the IHS4 2016/17 in Malawi and the ESS3 2018/19 in greater time spent by Ethiopia).16 The LSMS+ supported surveys, in both men and women in particular, reveal greater time spent by men and women in water and fuel collection than unpaid work than other the comparison/LSMS‑ISA surveys. recent national surveys, 16 In the Tanzania NPS4, only fuel and water collection time in particularly in Ethiopia the last 24 hours was asked, and so couldn’t be compared one‑to‑one with the LSMS+/NPS5. 30 LSMS+ Program in Sub-Saharan Africa Table 2.3 Work in Non‑Market Activities Malawi Rural Urban (A) LSMS+ round (IHPS 2016/17) Men Women T‑test Men Women T‑test Hours spent in own‑use production: last 24 hours Water collection 0.27 0.33 0.00 0.15 0.20 0.05 Fuel collection 0.14 0.17 0.02 0.04 0.06 0.15 Engages in agriculture for own use (Y=1 N=0): Those in agr: mainly for own production 0.71 0.78 0.01 0.95 0.92 0.30 (as opposed to market)‡ (B) IHS4 2016/17 Hours spent in own‑use production: last 24 hours Water collection 0.13 0.24 0.00 0.09 0.12 0.00 Fuel collection 0.11 0.15 0.00 0.03 0.05 0.05 Engages in agriculture for own use (Y=1 N=0): Those in agr: mainly for own production 0.74 0.75 0.21 0.82 0.88 0.02 (as opposed to market)‡ Ethiopia Rural Urban (A) LSMS+ round (ESS4 2018/19) Men Women T‑test Men Women T‑test Hours spent in own‑use production: last 7 days Water collection 0.25 0.65 0.00 0.28 0.45 0.00 Fuel collection 0.21 0.61 0.00 0.13 0.26 0.00 Engages in agriculture for own use (Y=1 N=0): Those in agr: mainly for own production 0.90 0.92 0.14 0.84 0.83 0.89 (as opposed to market)‡ (B) ESS3 2015/16 (where data are available) Hours spent in own‑use production: last 7 days Water collection 0.08 0.40 0.00 0.04 0.12 0.00 Fuel collection 0.13 0.37 0.00 0.04 0.13 0.00 Engages in agriculture for own use (Y=1 N=0): Those in agr: mainly for own production (as opposed to market)‡ Tanzania Rural Urban (A) LSMS+ round only (NPS5 2019/20) Men Women T‑test Men Women T‑test Hours spent in own‑use production: last 7 days Food‑related activities 0.49 1.67 0.00 0.27 1.98 0.06 Nonfood‑related activities 0.36 0.23 0.13 0.12 0.18 0.40 Water collection 0.79 2.63 0.00 0.69 1.62 0.00 Fuel collection 0.50 1.46 0.00 0.33 0.42 0.46 Engages in agriculture for own use (Y=1 N=0): Those in agr: mainly for own production 0.85 0.90 0.80 0.87 0.92 0.49 (as opposed to market) 1 All estimates use household sampling weights; only self‑reporting sample included. Section 2. Labor outcomes 31 3. Respondents’ engagement in multiple economic activities Tables 2.4‑2.7 examine whether the individual any agricultural work (for own‑production and for interview approach in the LSMS+ supported surveys market); supporting or owning non‑farm enterprises also reveals information about multiple economic (NFE); and work as an employee — either wage, activities that men and women are involved in, casual (included in Ethiopia), and ganyu (included both over the last 7 days as well as last 12 months17. in Malawi). For each country, the share of men Table 2.4 below examines men’s and women’s and women reporting multiple activities in the participation in more than one activity across comparison/LSMS‑ISA surveys are also presented. Looking across surveys, the share of individuals 17 The variable on multiple activities was constructed based on the engaged in multiple activities is quite high, introductory yes/no questions in the labor module on participation in the last 7 days and 12 months in agriculture, wage, and non‑farm particularly in Malawi and Tanzania. In the Malawi enterprise activity. Specifically, a respondent was considered to be LSMS+, for example, the majority of rural workers involved in multiple activities if they reported “yes” to more than one of these yes/no questions (with separate variables constructed for (71 percent for men and 55 percent for women) engagement in multiple activities over the last 7 days and 12 months). Table 2.4 Share of men and women engaged in multiple activities, conditional on working (LSMS+ supported survey versus comparator LSMS‑ISA survey in each country) Malawi Ethiopia Tanzania IHPS 2016/17 IHS4 2016/17 ESS4 2018/19 ESS3 2015/16 NPS5 2019/20 NPS4 2014/15 Last Last Last Last Last Last Last Last Last Last Last Last 7 12 7 12 7 12 7 12 7 12 7 12 days months days months days months days months days months days months Urban (a) Men 0.15 0.41 0.18 0.41 0.11 0.20 0.12 0.17 0.15 0.30 0.12 0.26 (b) Women 0.19 0.40 0.19 0.40 0.10 0.16 0.09 0.13 0.09 0.21 0.16 0.29 T‑test: 0.09 0.73 0.75 0.85 0.26 0.10 0.30 0.00 0.21 0.08 0.82 0.18 (a)‑(b) (p‑val) Rural (a) Men 0.28 0.71 0.30 0.67 0.08 0.24 0.13 0.27 0.23 0.49 0.27 0.45 (b) Women 0.24 0.55 0.23 0.54 0.07 0.14 0.08 0.15 0.16 0.27 0.20 0.33 T‑test: 0.05 0.00 0.00 0.00 0.30 0.00 0.00 0.00 0.06 0.00 0.01 0.00 (a)‑(b) (p‑val) 1 All estimates use household sampling weights; only self‑reporting sample included. 2 Multiple activities include any agricultural work, non‑farm enterprise, and wage employment. For Ethiopia, 12 months of agricultural work was not reported. 7‑day agricultural work questions were used instead. 32 LSMS+ Program in Sub-Saharan Africa © World Bank were engaged in multiple activities over the last In urban Tanzania, a greater share of LSMS+ 12 months. At this more aggregate level, large respondents report work in agriculture along with differences do not emerge when comparing the wage work, compared to the NPS4 where more were LSMS+ supported surveys with their LSMS‑ISA concentrated completely in agriculture. There are counterparts; further investigation can also fewer differences across the IHPS and IHS4 in Malawi. shed light as to whether differences emerge by Some of these changes in the composition of work respondents’ age or other individual/household in Ethiopia and Tanzania may be due to time‑varying characteristics. factors, but again could be explored in more detail (particularly in Tanzania, where the LSMS+ is part of an Men also typically show a higher proportion ongoing panel survey in the country). of working in multiple activities across these categories, especially in rural areas. Other forms Within the LSMS+ supported surveys, there are not of unpaid work that women are engaged in are many gender differences in Ethiopia among those not included in this table, which likely explains this working in multiple activities. In Tanzania, a much pattern. In most urban groups, the gender gap is greater share of rural men (36 percent) report not statistically significant. working in agriculture as well as wage work, compared to rural women (16 percent). A greater For each of the three countries, Tables 2.5‑2.7 share of urban women in the Malawi IHPS (about below present a more detailed activity‑wise 13 percent) also report working in agriculture as breakdown for the individuals are engaged in well as NFE work, compared to men (5 percent). (across main and secondary activities in the last 12 months). Comparisons are presented as well across Apart from multiple activities, one interesting the supported by the LSMS+ versus the LSMS‑ISA. similarity across countries is that urban women The share of respondents in multiple areas of who are in non‑agricultural work are much more work can be calculated from the shares in the likely to be engaged in non‑farm enterprises, off‑diagonal of each matrix in the tables. whereas men are more likely to be in wage work. In Ethiopia, for example, 46 percent of urban Similar patterns do emerge across the individual‑level women working in non‑agriculture were in NFEs, approach in the LSMS+ and the LSMS‑ISA, with a compared to 31 percent of urban men, with few exceptions. In Ethiopia, a greater concentration almost all of these men and women working of rural men and women are engaged solely in solely in this area of work. agriculture in the LSMS+ as opposed to the ESS3. Section 2. Labor outcomes 33 Table 2.5 Ethiopia: share of employment activities by survey year (LSMS+ and comparison LSMS‑ISA survey)versus comparator LSMS‑ISA survey in each country) Table A. Among those working: share across Table B. Among those working: share across activities (last 12 months; LSMS+ ESS4 2018/19) activities (last 12 months; ESS3 2015/16 ) Table A1. Men. LSMS+ Table B1. Men Urban Urban Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.13 0.02 0.00 0.02 0.00 Agriculture 0.07 0.02 0.01 0.02 0.00 NFE (run/ NFE (run/ manage) 0.28 0.05 0.00 manage) 0.30 0.05 0.01 NFE NFE (supporting 0.02 0.01 0.00 (supporting 0.04 0.01 0.00 role) role) Wage (salaried) 0.35 0.07 Wage (salaried) 0.42 0.02 Wage (casual) 0.00 Wage (casual) 0.03 Rural Rural Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.82 0.07 0.01 0.02 0.00 Agriculture 0.69 0.09 0.01 0.02 0.05 NFE (run/ NFE (run/ manage) 0.02 0.00 0.00 manage) 0.06 0.00 0.00 NFE NFE (supporting 0.00 0.00 0.00 (supporting 0.01 0.00 0.00 role) role) Wage (salaried) 0.01 0.00 Wage (salaried) 0.03 0.00 Wage (casual) 0.00 Wage (casual) 0.02 Table A2. Women. LSMS+ Table B2. Women Urban Urban Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.12 0.03 0.00 0.00 0.00 Agriculture 0.06 0.04 0.01 0.01 0.00 NFE (run/ NFE (run/ manage) 0.43 0.03 0.00 manage) 0.43 0.03 0.01 NFE NFE (supporting 0.05 0.00 0.00 (supporting 0.06 0.00 0.00 role) role) Wage (salaried) 0.27 0.03 Wage (salaried) 0.30 0.01 Wage (casual) 0.00 Wage (casual) 0.03 Rural Rural Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.80 0.07 0.01 0.00 0.00 Agriculture 0.68 0.08 0.01 0.00 0.02 NFE (run/ NFE (run/ manage) 0.08 0.00 0.00 manage) 0.13 0.00 0.00 NFE NFE (supporting 0.01 0.00 0.00 (supporting 0.02 0.00 0.00 role) role) Wage (salaried) 0.01 0.00 Wage (salaried) 0.02 0.00 Wage (casual) 0.00 Wage (casual) 0.02 1 All estimates use household sampling weights; only self‑reporting sample included. Blank cells = no observations. 2 12‑day agricultural activity for Ethiopia is not reported. A 7‑day measure is used instead. 3 2.2 percent of those working across men and women has more than 2 activities and not counted in the table above. For ESS3, 1.6% of working individuals had more than 2 activities. 34 LSMS+ Program in Sub-Saharan Africa Table 2.6 Share of Employment Activities by Survey Year (Tanzania) Table A. Among those working: share across Table B. Among those working: share across activities (last 12 months; LSMS+ NPS5 2019/20) activities (last 12 months; NPS4 2014/15) Table A1. Men. LSMS+ Table B1. Men Urban Urban Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.11 0.10 0.00 0.15 0.00 Agriculture 0.23 0.07 0.00 0.14 0.00 NFE (run/ NFE (run/ manage) 0.18 0.08 0.00 manage) 0.17 0.03 0.00 NFE NFE (supporting 0.00 0.00 0.00 (supporting 0.00 0.00 0.00 role) role) Wage (salaried) 0.37 0.00 Wage (salaried) 0.34 0.00 Wage (casual) 0.00 Wage (casual) 0.00 Rural Rural Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.32 0.13 0.01 0.36 0.00 Agriculture 0.41 0.14 0.00 0.27 0.00 NFE (run/ NFE (run/ manage) 0.03 0.01 0.00 manage) 0.03 0.00 0.00 NFE NFE (supporting 0.00 0.00 0.00 (supporting 0.00 0.00 0.00 role) role) Wage (salaried) 0.09 0.00 Wage (salaried) 0.07 0.00 Wage (casual) 0.00 Wage (casual) 0.00 Table A2. Women. LSMS+ Table B2. Women Urban Urban Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.20 0.08 0.00 0.12 0.00 Agriculture 0.27 0.17 0.00 0.08 0.00 NFE (run/ NFE (run/ manage) 0.29 0.03 0.00 manage) 0.19 0.03 0.00 NFE NFE (supporting 0.00 0.00 0.00 (supporting 0.00 0.00 0.00 role) role) Wage (salaried) 0.26 0.00 Wage (salaried) 0.24 0.00 Wage (casual) 0.00 Wage (casual) 0.00 Rural Rural Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.58 0.12 0.01 0.16 0.00 Agriculture 0.55 0.14 0.02 0.18 0.00 NFE (run/ NFE (run/ manage) 0.04 0.01 0.00 manage) 0.04 0.00 0.00 NFE NFE (supporting 0.00 0.00 0.00 (supporting 0.00 0.00 0.00 role) role) Wage (salaried) 0.04 0.00 Wage (salaried) 0.03 0.00 Wage (casual) 0.00 Wage (casual) 0.00 1 All estimates use household sampling weights; only self‑reporting sample included. Blank cells = no observations. 2 3.7 percent of those working across men and women has more than 2 activities and not counted in the table above. For ESS3, 4.4 percent of working individuals had more than 2 activities. Section 2. Labor outcomes 35 Table 2.7 Share of Employment Activities by Survey Year (Malawi) Table A. Among those working: share across Table B. Among those working: share across activities (last 12 months; LSMS+ IHPS 2016/17) activities (last 12 months; IHS4 2016/17) Table A1. Men. LSMS+ Table B1. Men Urban Urban Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.08 0.04 0.02 0.07 0.08 Agriculture 0.11 0.05 0.00 0.10 0.09 NFE (run/ NFE (run/ manage) 0.13 0.02 0.01 manage) 0.13 0.02 0.01 NFE NFE (supporting 0.03 0.01 0.02 (supporting 0.02 0.00 0.01 role) role) Wage (salaried) 0.25 0.04 Wage (salaried) 0.25 0.03 Wage (ganyu) 0.12 Wage (ganyu) 0.10 Rural Rural Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.22 0.10 0.02 0.07 0.38 Agriculture 0.26 0.07 0.01 0.06 0.42 NFE (run/ NFE (run/ manage) 0.01 0.00 0.00 manage) 0.01 0.00 0.00 NFE NFE (supporting 0.00 0.00 0.00 (supporting 0.00 0.00 0.00 role) role) Wage (salaried) 0.02 0.00 Wage (salaried) 0.03 0.00 Wage (ganyu) 0.03 Wage (ganyu) 0.03 Table A2. Women. LSMS+ Table B2. Women Urban Urban Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.18 0.14 0.04 0.04 0.05 Agriculture 0.20 0.12 0.03 0.05 0.08 NFE (run/ NFE (run/ manage) 0.18 0.02 0.02 manage) 0.14 0.01 0.02 NFE NFE (supporting 0.04 0.00 0.00 (supporting 0.05 0.01 0.00 role) role) Wage (salaried) 0.15 0.01 Wage (salaried) 0.17 0.01 Wage (ganyu) 0.07 Wage (ganyu) 0.05 Rural Rural Agr. NFE NFE Wage Wage Agr. NFE NFE Wage Wage (run/ (support‑ (salaried) (casual) (run/ (support‑ (salaried) (casual) manage) ing role) manage) ing role) Agriculture 0.41 0.08 0.02 0.02 0.35 Agriculture 0.42 0.06 0.02 0.02 0.36 NFE (run/ NFE (run/ manage) 0.01 0.00 0.00 manage) 0.01 0.00 0.00 NFE NFE (supporting 0.00 0.00 0.00 (supporting 0.00 0.00 0.00 role) role) Wage (salaried) 0.01 0.00 Wage (salaried) 0.01 0.00 Wage (ganyu) 0.02 Wage (ganyu) 0.02 1 All estimates use household sampling weights; only self‑reporting sample included. Blank cells = no observations. 2 10.8 percent of those working across men and women has more than 2 activities and not counted in the table above. For ESS3, 9.1 percent of working individuals had more than 2 activities. The most common multiple activity in Malawi among the 10.8 percent is agriculture, NFE, and Wage (ganyu) that accounts for almost half of the 10.8 percent 36 LSMS+ Program in Sub-Saharan Africa © Valentina Costa / World Bank 4. Individual, household and geographic Table 2.8 below, specifically, also examines the correlates correlation of individual‑level ownership of assets (mobile phones, financial accounts, and reported land ownership) with participation in agricultural or of labor non‑agricultural work for women. Full regressions, controlling for individual, household and geographic participation (urban/rural, as well as interview month fixed effects) are available in the Annex. The regressions, while not causal, reflect a strong link across countries with financial account ownership and women’s labor participation in non‑agricultural activities, In this section, we investigate what variables even after controlling for individual and household correlate with participation in agricultural or socioeconomic status, as well as geography. Mobile non‑agricultural activities in the last 7 days. Non‑farm phone ownership is also linked with women’s enterprises and wage employment is part of the non‑agricultural work in Tanzania. Women’s non‑agricultural activities in the analysis below. The individual ownership of land, as expected, is strongly main goal is to understand how the correlates may associated with agricultural activity across countries, differ across gender, urban/rural, and survey rounds. and particularly so in Tanzania. Section 2. Labor outcomes 37 Table 2.8 Women’s labor market participation: association with individual‑level asset variables Tanzania NPS5 Ethiopia ESS4 Malawi IHPS Dependent variable: Individual Agr. Non‑agr. Agr. Non‑agr. Agr. Non‑agr. in the past 7 days participated in agr./non‑agr. activities (Y=1, N=0) Individual owns a mobile phone 0.064 0.113** ‑0.046* 0.032 ‑0.002 0.009 (Y=1 N=0) [1.12] [2.22] [‑1.79] [1.27] [‑0.08] [0.34] Individual owns a financial account ‑0.063 0.165** 0.007 0.073*** 0.034* 0.083*** (Y=1 N=0) [‑0.89] [2.14] [0.28] [3.30] [1.80] [4.02] Individual owns any land 0.244*** ‑0.031 0.087*** ‑0.012 0.070*** ‑0.025 (reported ownership; Y=1 N=0) [4.47] [‑0.61] [2.93] [‑0.75] [3.09] [‑1.16] Observations 760 760 6,293 6,293 2,596 2,596 R‑squared 0.503 0.390 0.324 0.221 0.218 0.098 1 All regressions weighted by the household sampling weight. Only the self‑reporting sample included. Regressions also control for geographic, and interview month fixed effects. 2 Dependent variable was =1 if the woman respondent reported participating in the last 7 days in any agr. (non‑agr.) activity. 3 Annex Tables A1‑A3 provide full results with all right‑hand side variables, across individual, household and geographic controls. Among other associations with individual and Specifically, Table 2.9 shows that the increase in the household characteristics, Annex Tables A1‑A3 share of variation in outcomes for women’s work is show that women at the extremes of the age much greater in the Tanzania and Ethiopia LSMS+ distribution (18‑24, and 55+) are less likely to report when controlling for geographic/enumeration being in non‑agricultural work, whereas in Malawi area fixed effects, and in particular compared to and Tanzania, women ages 55+ are more likely to their LSMS‑ISA counterparts. In separate results in be involved in agriculture. There is some positive Table 2.10, this pattern is consistent for men’s work association with women’s non‑agricultural work across agriculture and non‑agriculture in Tanzania and household access to piped water in Tanzania, and Ethiopia. For Malawi, the same differences did as well as electricity in Ethiopia, and the reverse not emerge. relationship with agricultural work. Generally, Annex Tables A1‑A3 also show that similar patterns emerge with the same regressions for the LSMS‑ISA surveys conducted in these countries. One pattern that does stand out Across all countries, when comparing the LSMS+ with their LSMS‑ISA counterparts is the importance of geographic/ individual ownership enumeration‑area fixed effects in the LSMS+ supported surveys. Annex Tables A1‑A3 reflect a of financial accounts higher R‑squared in general, particularly in Tanzania is strongly linked and Ethiopia, for the LSMS+ supported surveys when controlling for the same right‑hand‑side with women’s variables. Table 2.9 below examines the extent non‑agricultural to which these geographic effects, that typically help account for local unobserved effects, may be employment contributing to this difference. 38 LSMS+ Program in Sub-Saharan Africa Table 2.9 Role of enumeration‑area fixed effects in explaining women’s labor market participation, LSMS+ versus comparison LSMS‑ISA surveys in each country Work in non‑agriculture (Y=1 N=0): Work in agriculture (Y=1 N=0) EA fixed effects EA fixed effects EA fixed effects EA fixed effects included not included included not included Tanzania LSMS+ (NPS 5, 2019‑20) R‑squared 0.318 0.206 0.401 0.287 Observations 1,004 1,004 1,004 1,004 LSMS‑ISA (NPS 4, 2014‑15) R‑squared 0.286 0.152 0.482 0.339 Observations 866 866 866 866 Ethiopia LSMS+ (ESS4, 2018‑19) R‑squared 0.203 0.177 0.281 0.271 Observations 6,293 6,293 6,293 6293 LSMS‑ISA (ESS3, 2015‑16) R‑squared 0.172 0.148 0.157 0.144 Observations 6,198 6,198 6,198 6,198 Malawi LSMS+ (IHPS, 2016) R‑squared 0.128 0.064 0.271 0.164 Observations 2,596 2,596 2,596 2,596 LSMS‑ISA (IHS4, 2016‑17) R‑squared 0.181 0.101 0.326 0.164 Observations 8,002 8,002 8,002 8,002 1 All regressions weighted by the household sampling weight. Only the self‑reporting sample included. The same individual and household characteristics in Annex Tables A1‑A3 are included in the regressions, along with interview month fixed effects. 2 Dependent variable was =1 if the woman respondent reported participating in the last seven days in any agr. (non‑agr.) activity. © World Bank Section 2. Labor outcomes 39 Table 2.10 Role of enumeration‑area fixed effects in explaining men’s labor market participation, LSMS+ versus comparison LSMS‑ISA surveys in each country Work in non‑agriculture (Y=1 N=0): Work in agriculture (Y=1 N=0) EA fixed effects EA fixed effects EA fixed effects EA fixed effects included not included included not included Tanzania LSMS+ (NPS 5, 2019‑20) R‑squared 0.403 0.278 0.442 0.355 Observations 867 867 867 867 LSMS‑ISA (NPS 4, 2014‑15) R‑squared 0.426 0.297 0.494 0.399 Observations 875 875 875 875 Ethiopia LSMS+ (ESS4, 2018‑19) R‑squared 0.301 0.276 0.509 0.495 Observations 4,937 4,937 4,937 4,937 LSMS‑ISA (ESS3, 2015‑16) R‑squared 0.254 0.216 0.253 0.254 Observations 5,534 5,534 5,534 5,534 Malawi LSMS+ (IHPS, 2016) R‑squared 0.233 0.190 0.295 0.129 Observations 2,075 2,075 2,075 2,075 LSMS‑ISA (IHS4, 2016‑17) R‑squared 0.218 0.136 0.320 0.166 Observations 6,988 6,988 6,988 6,988 1 All regressions weighted by the household sampling weight. Only the self‑reporting sample included. The same individual and household characteristics in Annex Tables A1‑A3 are included in the regressions, along with interview month fixed effects. 2 Dependent variable was =1 if the woman respondent reported participating in the last seven days in any agr. (non‑agr.) activity. In general, more precise, self‑reported data on individual labor outcomes can allow for clearer policy links on how work is associated with individual and household demographics, education, access to infrastructure, and local factors. Multiple areas of work, as well as intra‑household dynamics can also be explored in greater detail. All of this presents an important opportunity to improve assessments of economic constraints and choices that men and women make, and pathways to improving economic mobility. 40 LSMS+ Program in Sub-Saharan Africa @ photo credit © World Bank Section 2 / Summary + + + + Improved self‑reporting There are significant Women spend In regressions examining in labor modules can gender gaps in participation disproportionately more effects on women’s reduce measurement in wage employment across time in non‑market work labor participation, error in understanding the LSMS+ supported compared to men, including individual‑level ownership key outcomes across paid surveys in Malawi, activities such as fuel and of assets collected in and unpaid work, as well Tanzania and Ethiopia. firewood collection. The the LSMS+ supported as shed greater light on In general, gender gaps LSMS+ supported surveys, surveys (land, mobile intra‑household variation in participation across in particular, reveal greater phones and financial in men’s and women’s activities differ depending time spent by respondents in accounts) are strongly economic opportunities. on the activity type but also water and fuel collection than correlated with women’s by country. Comparisons the comparison LSMS‑ISA labor participation, and + by Kilic et al. (2020a) of the surveys conducted in these particularly ownership of While full self‑reporting was Malawi IHPS/LSMS+ and countries. financial accounts with not achievable in the LSMS+, the concurrently‑conducted women’s participation in enumerators focused on IHS4, show that the + non‑agricultural (wage ensuring one‑on‑one, IHS4 leads to significant Across surveys, the share and NFE) work. Geographic self‑reported interviews underreporting of of working individuals and enumeration‑area whenever possible. Proxy employment across engaged in multiple fixed effects also appear reporting of labor and a range of wage and activiies (main and to have a stronger role health outcomes was self‑employment activities, secondary activities as in explaining variation in significantly lower in the with stronger effects for elicited in the labor module women’s labor participation, Malawi and Tanzania women and for a longer questionnaire) is quite across agriculture and LSMS+ compared to (12‑month) recall period. high, particularly in Malawi non‑agriculture, in the LSMS‑ISA surveys and Tanzania. Men also LSMS+ supported surveys. conducted in these typically show a higher countries (i.e., the Malawi proportion of working in IHS4 and the Tanzania multiple activities across NPS4). these categories, especially in rural areas. Other forms of unpaid work that women are engaged in are not included in these estimates, which likely explains this pattern. Section 2. Labor outcomes 41 © Valentina Costa / World Bank 42 LSMS+ Program in Sub-Saharan Africa Section 3 Asset ownership and rights 1. The importance of individual‑level interviews in measuring asset ownership/rights 2. LSMS+ survey modules on asset ownership 2.1. Asset classes 2.2. Interview approach 3. Land ownership and rights, and livestock ownership, across the LSMS+ supported surveys 3.1. What patterns emerge across men and women? 3.2. Bundles of land ownership and rights 3.3. Reporting discrepancies among couples 3.4. Livestock ownership in Ethiopia 4. Financial account and mobile phone ownership 4.1. Individual‑level estimates 4.2. Does individual‑level data collection affect household estimates of mobile and financial account ownership? Section 3 / Summary Section 3. Asset ownership and rights 43 © World Bank 1. The importance International momentum behind improving the availability and quality of of individual‑level individual‑disaggregated survey data on asset ownership and control has accelerated, in large part to the United Nations Evidence and interviews Data for Gender Equality (EDGE) initiative. Specifically, the UN EDGE initiative supported in measuring survey experiments and pilots across countries between 2014‑16, including the Methodological Experiment on Measuring asset ownership/ Asset Ownership from a Gender Perspective (MEXA) in Uganda (see Box 3.1; Kilic and rights Moylan, 2016),18 aimed at understanding how to better collect individual‑level data on asset ownership and rights. This work ultimately resulted in the United Nations Guidelines for Producing Statistics on Asset Ownership from a Gender Perspective (UNSD, 2019).19 18 Additional country pilots supported by UN EDGE were implemented by the national statistical offices across Georgia, Maldives, Mexico, Mongolia, the Philippines and South Africa. 19 The guidelines can be accessed at https://unstats.un.org/ edge/publications/docs/Guidelines_final.pdf. @ photo credit 44 LSMS+ Program in Sub-Saharan Africa Box 3.1 Methodological Experiment on Measuring Asset Ownership from a Gender Perspective (MEXA) in Uganda (Kilic and Moylan, 2016) MEXA is a randomized household survey experiment that In Arms 1‑4, the respondents reported on all assets owned, was implemented by the Uganda Bureau of Statistics in either exclusively or jointly, by members of the household. Arm 2014, in collaboration with the UN EDGE Initiative and the 5 was identical to Arm 4, except that respondents reported only World Bank Living Standards Measurement Study (LSMS), on assets they themselves owned, either exclusively or jointly. providing a unique opportunity for more in‑depth analysis of The asset types included: dwelling, agricultural land, livestock, gender disparities in asset ownership, with a focus on land agricultural equipment, other real estate, non‑farm enterprises/ ownership. enterprise assets, financial assets and liabilities, and valuables. Differentiation across legal, reported, and economic ownership The experiment targeted 140 enumeration areas (EAs) across and the bundle of rights (sell, rent out, use as collateral, Uganda, and randomly allocated four households in each bequeath, and make investments) at the asset level was key. EA to each of five treatments/arms that differed in terms Individuals associated with each of these constructs were of respondent selection. Regardless of the treatment, the uniquely identified. respondent(s) were interviewed alone. On the design, implementation and analysis of MEXA, please The first four treatments/arms included interviewing refer to: Kilic, T., and Moylan, H. (2016). “Methodological 1. the self‑identified most knowledgeable household experiment on measuring asset ownership from a gender member; perspective: technical report.” Washington, DC: World Bank. 2. a randomly selected member of the principal couple; 3. the principal couple together; 4. all adult household members, simultaneously. As seen in the MEXA experiment The 2019 UN guidelines provide empirical evidence in Uganda, that support a number of methodological shifts in how survey questions on asset ownership following these and rights are designed and administered. This recommendations includes moving away from the common practice of relying on a “most knowledgeable” household provides a more member who also reports for other household members; emphasizing interviewing multiple adults complete picture per household; and private interviews regarding of ownership of respondents’ personal ownership of and rights to assets, either exclusively or jointly with someone and rights to assets else.20 As seen in the MEXA experiment in Uganda, within households, following these recommendations provides a more complete picture of ownership of and rights particularly among to assets within households, particularly among women; minimizes distortionary proxy respondent women; minimizes effects and intra‑household discrepancies in distortionary proxy reporting; and reveals hidden assets. respondent effects 20 Also see related work by Grown et al. (2005) and Doss et al. (2011): and intra‑household Grown, C., Rao Gupta, G., and Kes, A. (2005). Taking action: achieving gender equality and the millennium development goals. London: discrepancies in Earthscan Publications.; and Doss C., Deere, C. D., Oduro, A., Swaminathan, H., Suchitra, J., Lahoti, reporting; and reveals hidden assets R., Baah‑Boateng, W., Boakye‑Yiadom, L., Twyman, J., Catanzarite, Z., Grown, C., and Hillesland, M. (2011). “The gender asset and wealth gaps: evidence from Ecuador, Ghana, and Karnataka, India.” Bangalore: Indian Institute of Management. Section 3. Asset ownership and rights 45 © World Bank 2.1. Asset classes On land, both dwelling and non‑dwelling land are covered in the LSMS+. Parcels were first identified and rostered through the household questionnaire are carried forward to individual interviews. For each parcel, respondents are asked about different types of ownership (reported, economic, and documented); rights (to sell, bequeath, use as 2. collateral, rent out, and make improvements/ invest); as well as decision‑making in the case of agricultural parcels (Box 3.2).23 Respondents are also asked about perceived tenure security. The questions on rights were not asked of the respondent if he/she LSMS+ survey did not name himself/herself as a reported owner for a given parcel. 24 Respondents modules on are also asked to identify, in the case of joint ownership or where permission is needed to exercise rights, up to three asset ownership household members. In Malawi they also recorded two non‑household members who share ownership/give permission through a network roster capturing the name and The LSMS+ modules delve into intra‑household relationship to the household. In Tanzania ownership across different types of assets — and and Ethiopia, the numbers of male and the in turn highlight important patterns of ownership number of female non‑household owners and decision‑making that can inform policy efforts were collected. to expand access to financial services, land, and For other asset classes, questions on property rights in general. For all countries where the household dwelling follow the same the LSMS+ has been implemented — including structure as the parcel module (following Malawi, Tanzania, and Ethiopia —individual‑level household‑level questions on the dwelling modules on assets span (a) ownership and rights to and then asking the same individual‑level land parcels21, as well as (b) respondents’ ownership questions on different types of ownership (exclusive or joint) over other assets including and rights). For financial accounts, mobile financial accounts, and mobile phones. Ownership phones, and livestock, respondents were of livestock was also included in Ethiopia. asked about whether they owned these The module on land, specifically, addresses the assets exclusively or jointly with others data needs for both SDG indicators 1.4.2 and 5.a.1 – (and, if jointly, with which other household covering all land owned or accessed via use rights, members). and following recent recommendations in 2019 by the Food and Agricultural Organization (FAO), the 23 Along with rights/ownership, respondents reported on World Bank, and UN Habitat.22 The modules on how each parcel was acquired; identified the individuals from whom the asset was inherited or received as a gift, as mobile phones and financial accounts cover SDG applicable; and provided the current hypothetical sales value indicators 5.b.1 and 8.10.2. for each asset (and the construction costs specifically for the dwelling) and limited information on their knowledge of asset transactions in their communities. 21 A parcel is defined as a continuous piece of land which can have more 24 The scope of rights included in the questionnaire was than one plot. influenced by Schlager and Ostrom’s (1992) theoretical 22 The World Bank, FAO and UN Habitat (2019). Measuring Individuals’ framework which focuses, in the context of natural resources, Rights to Land: An Integrated Approach to Data Collection for SDG on issues related to access, withdrawal, management, Indicators 1.4.2 and 5.a.1. exclusion and alienation while defining a bundle of rights. 46 LSMS+ Program in Sub-Saharan Africa 2.2. interview targets. Within‑household interviews were always administered in private and, depending on the Interview approach fieldwork set‑up in the country, were attempted to be administered simultaneously, as well as, to the best extent possible, with gender matching between the The LSMS+ modules on asset ownership and enumerator and respondent.26 27 rights attempted to carry out personal interviews of adult household members, inquiring about their Regarding agricultural land, following the creation personal ownership of and rights to assets in the of a roster of all owned and/or cultivated agricultural aforementioned asset classes ‑ corresponding to parcels and the identification of those that are Treatment 5 (“T5”) of MEXA described in Box 3.1, and “owned” by at least one household member, this leveraging the contextualized and improved versions common list of owned parcels that is generated as of the MEXA T5 questionnaire instrument (Kilic and part of the household interview was fed forward to Moylan, 2016). Appendix I includes the protocol each individual interview in that household. 28,29 for administering the individual questionnaire. The individual interviews were capped at four 26 For more information on the organization and implementation of per household in Malawi and it was ensured that the individual‑disaggregated data collection as part of the IHPS, please consult the survey’s basic information document, which the head of household and his/her spouse (if one can be accessed here: https://microdata.worldbank.org/index.php/ exists) were among the individuals interviewed.25 catalog/2939/download/47216. 27 See Operational Guidance. In Tanzania and Ethiopia all eligible adults were 28 Parcel is defined as a continuous piece of land which can have more than one plot; in the IHPS this is referred to as “Garden”. 25 This was an upper limit that only applied to 1 percent of the sampled 29 In this process, the enumerator for each individual interview in household population that had more than four adults. If a sampled each household copied the garden roster from the tablet of the household had more than four adult household members, following the primary enumerator assigned to the household into his/her tablet preference given to the head of the household, and his/her spouse if that generated a new questionnaire (under Survey Solutions census applicable, the remaining interview targets (2 or 3 depending on the mode) for each interview target. To better facilitate the process, the presence of a spouse) were selected at random from the remaining enumerators also had paper booklets of household, garden and plot pool of adult household members. rosters to ensure unique identification of household members and parcels across the individual interviews in the same household. Box 3.2 LSMS+ questions over parcel ownership and rights, as well as decision-making Reported Economic Documented Rights1 Decision‑making owner owner owner (agricultural With regard to this [PARCEL], parcels) Do you own this If this [PARCEL] were Does your household are you among the [PARCEL], either to be sold today, have a document for individuals who have this Are you among the alone or jointly with would you be among this [PARCEL], such right, even if you need to decision‑maker(s) on someone else? the individuals to as an application obtain consent or permission this [PARCEL] regarding (If jointly, list others – decide how the receipt, land from someone else? the timing of crop up to three members money is used? investigation paper, activities, crop choice, If yes, do you need from HH roster, and (if others also certificate (title) from and input use? permission or consent from number of non‑HH involved, list up the government, (if others also involved, anyone else (name those members) to three members paper from local list up to three members members; and number of from HH roster, and authority, lease or from HH roster, and non‑HH members)? number of non‑HH rental contract? number of non‑HH members) With regard to this [PARCEL], members) Are you listed on the who else has this right, even title or ownership if they needed to obtain document as owner consent or permission from of this parcel? someone else? Does the (If others also listed, person need permission or name up to three consent? From whom does members from HH the person need permission roster, and number of or consent? non‑HH members) 1 Questions on rights are asked separately for rights to sell, bequeath, use as collateral, rent out, and make improvements/invest in it. 2 For non-dwelling land in Ethiopia, the right to sell was not asked because land is State-owned. Section 3. Asset ownership and rights 47 3. © Valentina Costa / World Bank Land ownership and rights, and livestock ownership, across the LSMS+ supported surveys 3.1. What patterns emerge across men and women? What do the LSMS+ supported surveys reveal on men’s and women’s land ownership and rights? Figures 3.1 and 3.2 present summary statistics on variables capturing exclusive versus joint ownership and rights of non‑dwelling and dwelling land, respectively, along with a dichotomous variable, “SDG owner”, based on the definition of the SDG indicator 5.a.1 (if the individual is a documented owner of a parcel, has the right to sell, or has the right to bequeath). Shares of ownership/ rights are broken out by women and men overall, as well as for women non‑heads of household.30 Adjusted Wald tests for equality of means were conducted across these three groups, with significant differences (p<0.05) indicated by darker‑colored bars. 30 As compared to women where there was greater diversity in household status by head, spouse, or other household members, nearly all men reporting ownership were household heads. 48 LSMS+ Program in Sub-Saharan Africa Apart from gender, the distribution of ownership In rural Tanzania, Malawi and Ethiopia 25-30 % and rights varies by country, rural/urban residence, as well as type of land. As expected, a greater share of rural respondents claim ownership and rights over land (dwelling or non‑dwelling) than urban respondents. In Tanzania and Malawi, exclusive economic ownership for both types of of non-dwelling land land also tends to be much lower than exclusive reported ownership across both urban and rural is jointly owned areas. As a result, even when reported land ownership is exclusive, decision making over the proceeds from selling land is more likely to be In Malawi, on the other hand, where the majority distributed across multiple household members. of ownership and rights over land follow Interesting gender differences also emerge matrilineal traditions, the shares of women with across the three countries, when comparing exclusive reported and economic ownership over (a) all men with all women, and (b) all men with non‑dwelling and dwelling land are significantly non‑household head women.31 In Tanzania, women higher than that for men. Country context therefore are significantly less likely to have exclusive matters. A recent study by Kilic et. al (2020b)32 ownership over both types of land compared to also compared the Malawi LSMS+/IHPS with men, with wider disparities for non‑household head the concurrent IHS4 that asked only one “most women. In Ethiopia, non-household head women knowledgeable” respondent about household are also significantly less likely to have exclusive members’ agricultural land ownership and rights— ownership over land, although women overall finding that the IHS4 resulted in higher rates of have higher exclusive dwelling land ownership exclusive reported and economic ownership of than men (a closer look at the data shows this is agricultural land among men, and lower rates of driven by women heads of household, who are joint reported and economic ownership among less likely to have a spouse living with them). women. Malawi was a unique case where this Women are also significantly less likely than men comparison could be made, since the IHS4 was in Ethiopia to have joint ownership of non‑dwelling conducted at the same time, and with the same and dwelling land. In Tanzania and Ethiopia, women questionnaire format as the non‑dwelling land are also significantly less likely to be SDG owners assets module in the LSMS+, but with a different of non‑dwelling and dwelling land, with wider interview approach. disparities for Tanzania. Figure 3.3 also presents, by dwelling and non‑dwelling land, how land ownership and rights vary by age. The higher share of respondents 31 As seen in Section 1, Table 1.4, most men respondents were household owning dwelling as opposed to non‑dwelling land heads, whereas women’s relationship to the household head varied more. This is why the additional category of women respondents is also reflected in the figure. With the exception (non‑household head) was broken out. of dwelling land in Malawi, the gender gap in ownership and rights tends to widen among men and women above 50 years of age. Trends in With the exception of dwelling land in reported, economic and SDG ownership, by age, Malawi, the gender gap in ownership also tend to be similar in most contexts, except for and rights tends to widen among men Tanzania where there is a much larger gap between and women above reported ownership and economic as well as SDG 50 ownership for respondents below 50 years of age, with this gap narrowing somewhat (more so for years of men) among the elderly. age 32 Kilic, Talip, Heather Moylan, and Gayatri Koolwal. 2020b. “Getting the (Gender‑Disaggregated) Lay of the Land: Impact of Survey Respondent Selection on Measuring Land Ownership and Rights.” World Bank Policy Research Working Paper 9151. Section 3. Asset ownership and rights 49 Figure 3.1 Shares of men and women with different ownership and rights, non-dwelling land (Darkened bars = significant gender differences at p<0.05)1 All men All women Non - HH head women Tanzania - Urban Tanzania - Rural 0.4 0.4 ** 0.3 0.3 0.2 0.2 ** ** ** ** 0.1 0.1 ** ** ** ** ** ** 0 ** 0 Exclusive Joint Exclusive Joint Exclusive Joint Exclusive Joint Reported Economic SDG Reported Economic SDG Owner Owner Malawi - Urban Malawi - Rural 0.4 0.4 ** 0.3 0.3 ** ** 0.2 ** ** 0.2 ** ** ** ** ** 0.1 0.1 ** 0 0 Exclusive Joint Exclusive Joint Exclusive Joint Exclusive Joint Reported Economic SDG Reported Economic SDG Owner Owner Ethiopia - Urban Ethiopia - Rural 0.4 0.4 ** 0.3 0.3 ** ** ** ** ** ** 0.2 0.2 0.1 0.1 ** ** ** ** ** 0 ** 0 Exclusive Joint Exclusive Joint Exclusive Joint Exclusive Joint Reported Economic SDG Reported Economic SDG Owner Owner 1 Dark blue bars with ** reflect cases where men’s outcomes are significantly different (p<0.05) from (a) the full sample of women (when significant, indicated by dark orange bars with**), or (b) from non-household head women (when significant, indicated by black bars with **). In all other cases (light colored bars) there were no significant differences across these two sets of groups. All estimates are weighted using household sampling weights. 2 “SDG owner” is based on the definition of the SDG indicator 5.a.1. (takes the value one if the individual is a documented owner, has the right to sell, or has the right to bequeath). 50 LSMS+ Program in Sub-Saharan Africa Figure 3.2 Shares of men and women with different ownership and rights, dwelling land (Darkened bars = significant gender differences at p<0.05)1 All men All women Non - HH head women Tanzania - Urban Tanzania - Rural 0.6 0.6 0.5 0.5 0.4 0.4 ** ** 0.3 0.3 ** ** ** 0.2 ** 0.2 ** ** ** ** ** 0.1 ** 0.1 ** ** ** 0 0 Exclusive Joint Exclusive Joint Exclusive Joint Exclusive Joint Reported Economic SDG Reported Economic SDG Owner Owner Malawi - Urban Malawi - Rural 0.6 0.6 ** 0.5 0.5 ** ** ** ** 0.4 0.4 ** ** 0.3 ** 0.3 ** ** 0.2 0.2 ** 0.1 ** 0.1 ** 0 0 Exclusive Joint Exclusive Joint Exclusive Joint Exclusive Joint Reported Economic SDG Reported Economic SDG Owner Owner Ethiopia - Urban Ethiopia - Rural 0.6 0.6 0.5 0.5 ** ** ** ** 0.4 0.4 ** ** 0.3 0.3 ** ** 0.2 ** ** ** 0.2 ** 0.1 0.1 ** ** ** ** ** ** ** ** ** ** ** ** 0 0 Exclusive Joint Exclusive Joint Exclusive Joint Exclusive Joint Reported Economic SDG Reported Economic SDG Owner Owner 1 Dark blue bars with ** reflect cases where men’s outcomes are significantly different (p<0.05) from (a) the full sample of women (when significant, indicated by dark orange bars with**), or (b) from non-household head women (when significant, indicated by black bars with **). In all other cases (light colored bars) there were no significant differences across these two sets of groups. All estimates are weighted using household sampling weights. 2 “SDG owner” is based on the definition of the SDG indicator 5.a.1. (takes the value one if the individual is a documented owner, has the right to sell, or has the right to bequeath). Section 3. Asset ownership and rights 51 Figure 3.3 Land ownership and rights, by age of respondents Reported Economic SDG owner Reported Economic SDG owner owner Men owner Men Men owner Women owner Women Women (1a) Ethiopia: Non-dwelling land (1b) Ethiopia: Dwelling land 0.6 0.6 Share of respondents Share of respondents 0.4 0.4 0.2 0.2 0 0 10 20 30 40 50 60 70 10 20 30 40 50 60 70 age age (2a) Tanzania: Non-dwelling land (2b) Tanzania: Dwelling land 0.8 0.8 Share of respondents Share of respondents 0.6 0.6 0.4 0.4 0.2 0.2 0 0 10 20 30 40 50 60 70 10 20 30 40 50 60 70 age age (3a) Malawi: Non-dwelling land (3b) Malawi: Dwelling land 0.8 0.8 Share of respondents Share of respondents 0.6 0.6 0.4 0.4 0.2 0.2 0 0 10 20 30 40 50 60 70 10 20 30 40 50 60 70 age age 1 In Ethiopia, right to sell was not asked for non-dwelling land. 52 LSMS+ Program in Sub-Saharan Africa © World Bank 3.2. Bundles of land ownership Reported and economic ownership are also generally tightly linked — either with respondents and rights claiming neither reported nor economic ownership, or both. In Tanzania, there was some variation, albeit small — about 9 percent of women claimed Understanding how ownership is linked with reported but no economic ownership over specific rights is also important for policy design. non‑dwelling land (and 16 percent of women for Figure 3.4 presents the share of women and men dwelling land); for men, these shares were about 6 with different bundles of ownership and rights, for and 16 percent, respectively.34 Similarly, among men dwelling as well as non‑dwelling land.33 Generally, and women landowners, rights to sell and bequeath only small shares of men and women claim all overlapped substantially. The main exception was ownership and rights to land. Within Ethiopia — and for dwelling land in Ethiopia, and for non‑dwelling Tanzania in particular — wider gender inequalities land in Malawi, where a little more than 10 percent are also apparent, where men were more likely of men and women claimed they had rights to to claim they had all ownership and rights over bequeath, but not sell, land. different types of land compared to women. In Tanzania, for example, only about 10 percent of Interestingly, as compared to ownership, gender women claimed they had all ownership and rights disparities tended to widen among landowners in to non‑dwelling land (compared to 21 percent rights to sell and bequeath, with the exception of of men); for dwelling land, these shares were 15 Ethiopia where gender gaps were roughly similar percent of women and 27 percent of men. Within across landownership and rights. In Tanzania, for Ethiopia, 18 and 23 percent of women and men, example, the share of women landowners reporting respectively, claimed all ownership and rights neither the right to sell nor bequeath their land to non‑dwelling land, compared to 25 and 31 (51 percent for non‑dwelling land, and 56 percent percent of women and men for dwelling land. In for dwelling land) was more than twice that of men Malawi, relative shares of men and women with landowners. This was the case even in Malawi, all ownership and rights were similar – about 23 where a greater share of women had both reported percent for non‑dwelling land, and 30 percent for and economic ownership of non‑dwelling land dwelling land. As compared to the other countries, compared to men. The findings therefore point to a greater share of women in Malawi also claimed a need to better understand and compare specific both reported and economic ownership of dwelling nuances of asset ownership and rights across men and non‑dwelling land. and women. 33 For both non‑dwelling and dwelling land, rural/urban patterns for 34 Because of the way the individual interviews were conducted, Tanzania and Ethiopia matched the patterns for the total sample fairly “economic but not reported ownership” was not a possible response closely. option. Section 3. Asset ownership and rights 53 Figure 3.4 Share of women and men with different bundles of ownership and rights (1/2) Men Women Ethiopia: Non-dwelling land ALL OWNERSHIP RIGHTS among OWNERSHIP * Reported, no economic Neither reported nor economic Reported and economic owners, RIGHTS Bequeath No bequeath 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Ethiopia: Dwelling land ALL OWNERSHIP RIGHTS among owners, OWNERSHIP * Reported, no economic Neither reported nor economic Reported and economic Neither sell nor bequeath Both sell and bequeath RIGHTS Bequeath, no sell Sell, no bequeath 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Tanzania: Non-dwelling land ALL OWNERSHIP RIGHTS among owners, OWNERSHIP * Reported, no economic Neither reported nor economic Reported and economic Neither sell nor bequeath RIGHTS Both sell and bequeath Bequeath, no sell Sell, no bequeath 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Tanzania: Dwelling land ALL OWNERSHIP RIGHTS among owners, OWNERSHIP * Reported, no economic Neither reported nor economic Reported and economic Neither sell nor bequeath RIGHTS Both sell and bequeath Bequeath, no sell Sell, no bequeath 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 54 LSMS+ Program in Sub-Saharan Africa Figure 3.4 Share of women and men with different bundles of ownership and rights (2/2) Men Women Malawi: Non-dwelling land ALL OWNERSHIP RIGHTS * OWNERSHIP Reported, no economic Neither reported nor economic Reported and economic Neither sell nor bequeath owners, Among RIGHTS Both sell and bequeath Bequeath, no sell Sell, no bequeath 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 Malawi: Dwelling land ALL OWNERSHIP RIGHTS * OWNERSHIP Reported, no economic Neither reported nor economic Reported and economic Neither sell nor bequeath owners, Among RIGHTS Both sell and bequeath Bequeath, no sell Sell, no bequeath 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 All estimates are weighted using household sampling weights. 2 Because of the order in which questions were administered, no respondents across countries had economic but not reported ownership. 3 In Ethiopia, right to sell was not asked for non-dwelling land. Section 3. Asset ownership and rights 55 © Valentina Costa / World Bank 3.3. Reporting discrepancies among couples Along with the insight that individual‑level interviews can provide on gender differences in ownership and rights, as well as intra‑household inequalities, there are risks that individual reporting can also lead to discrepancies among household members. This includes differences in household members’ views on who in fact owns/has rights over certain assets (or whether assets are owned at all). Figure 3.5 examines reports from married couples across the three countries, and finds that the share of parcels where couples agree on (a) rights to bequeath, (b) rights to sell, (c) Across countries, couples that disagree are not economic ownership, and (d) reported ownership necessarily concentrated in any one particular are relatively high, although there are specific scenario or category of ownership/rights. In cases where disagreement exists. In Ethiopia, for Ethiopia, for example, there is more substantial more than 85 percent of parcels, couples agree disagreement over the right to bequeath — about on economic and reported ownership of both 29 percent for non‑dwelling land, and 36 percent categories of land (and among owners, most for dwelling land, with most cases of disagreement agree that ownership is joint). Agreement over being concentrated where the husband says he economic and reported ownership is also high in has sole rights, whereas the wife says she does. Tanzania (couples agree on reported ownership In Tanzania, disagreement occurs mainly over for about 70 percent of parcels, with higher reported ownership, and where the wife claims shares of agreement on economic ownership ownership (either sole or joint) when the husband and rights to sell and bequeath). Among owners, says he is the sole owner. And in Malawi, most most agree that ownership is joint, with a smaller disagreement occurs for reported or economic share agreeing that ownership is solely the ownership, but mostly where the husband says husband’s—and almost all agree that the husband ownership is joint, but the wife says she does not maintains rights to sell and bequeath. In Malawi, own land. couples agree on reported ownership for about Country context therefore matters in interpreting 70 percent of non‑dwelling parcels, and agree on disagreement, and in particular Figure 3.5 — along economic ownership for more than 80 percent with earlier results in Figures 3.1‑3.2 — show that of dwelling parcels.35 About 80 percent or more prevailing gender differences in ownership and of non‑dwelling and dwelling parcels also have rights claimed by men and women also tend to agreement on rights to sell and bequeath, and be associated with the direction of disagreement among owners that these rights lie either with the (whether the husband or wife claims a greater role, wife or the husband. for example). 35 In Malawi, for reported ownership of dwelling land, there was an issue with how other joint owners were coded, so the discrepancy results for reported ownership of dwelling land are not reported in Figure 3.5. 56 LSMS+ Program in Sub-Saharan Africa Figure 3.5 Spousal agreement and disagreement over ownership and rights to land parcels (1/3) Bequeath Sell Economic Reported Ethiopia: Non-dwelling land Total (disagreement) (most common scenarios) Husb: J, Wife: W Disagreement Husb: J, Wife: Doesn’t own Husb: H, Wife: W Husb: H, Wife: J Total (agreement) No ownership Agreement Wife Joint Husband 0.0 0.2 0.4 0.6 0.8 1.0 Ethiopia: Dwelling land Total (disagreement) (most common scenarios) Husb: J, Wife: W Disagreement Husb: J, Wife: Doesn’t own Husb: H, Wife: W Husb: H, Wife: J Total (agreement) No ownership Agreement Wife Joint Husband 0.0 0.2 0.4 0.6 0.8 1.0 Section 3. Asset ownership and rights 57 Figure 3.5 Spousal agreement and disagreement over ownership and rights to land parcels (2/3) Bequeath Sell Economic Reported Tanzania: Non-dwelling land (most common scenarios) Total (disagreement) Disagreement Husb: J, Wife: Doesn’t own Husb: H, Wife: W Husb: H, Wife: J Total (agreement) No ownership Agreement Wife Joint Husband 0.0 0.2 0.4 0.6 0.8 1.0 Tanzania: Dwelling land Total (disagreement) Disagreement (most common scenarios) Husb: J, Wife: Doesn’t own Husb: H, Wife: W Husb: H, Wife: J Total (agreement) No ownership Agreement Wife Joint Husband 0.0 0.2 0.4 0.6 0.8 1.0 58 LSMS+ Program in Sub-Saharan Africa Figure 3.5 Spousal agreement and disagreement over ownership and rights to land parcels (3/3) Bequeath Sell Economic Reported Malawi: Non-dwelling land (most common scenarios) Total (disagreement) Disagreement Husb: J, Wife: Doesn’t own Husb: H, Wife: W Husb: H, Wife: J Total (agreement) No ownership Agreement Wife Joint Husband 0.0 0.2 0.4 0.6 0.8 1.0 Malawi: Dwelling land (most common scenarios) Total (disagreement) Disagreement Husb: J, Wife: Doesn’t own Husb: H, Wife: W Husb: H, Wife: J Total (agreement) No ownership Agreement Wife Joint Husband 0.0 0.2 0.4 0.6 0.8 1.0 1 All estimates are weighted using household sampling weights. H = husband, W = wife, J = joint. 2 Because of the order in which questions were administered, no respondents across countries had economic but not reported ownership. 3 In Ethiopia, right to sell was not asked for non-dwelling land. Section 3. Asset ownership and rights 59 Rural livestock ownership in Ethiopia is also more likely to be joint as opposed to exclusive 3.4. Table 3.1 Share of men and women owning large livestock, in rural and urban areas: Livestock ownership Ethiopia LSMS+ in Ethiopia Women Men Rural The Ethiopia LSMS+ also included a module on Overall 0.577*** 0.636*** exclusive/joint ownership of large livestock. (0.494) (0.481) Table 3.1 presents differences in ownership Exclusive 0.260 0.268 across men and women, as well as by rural and urban areas. The share of respondents owning (0.439) (0.443) large livestock was, as might be expected, much Joint 0.367*** 0.444*** higher in rural areas, and with significant gender (0.482) (0.497) differences (about 58 and 64 percent of rural Observations 3,755 3,560 women and men, respectively, with significant differences across the two groups arising from Urban joint ownership). However, sizeable shares of Overall 0.166** 0.197** urban respondents also reported ownership (0.372) (0.398) (17 percent of urban women, and 20 percent of Exclusive 0.083*** 0.101*** urban men, with significant gender differences in (0.276) (0.302) urban areas stemming mainly from a very small increase in exclusive ownership among urban Joint 0.093 0.110 men). Similar to land ownership in Ethiopia, Table (0.290) (0.313) 3.1 also shows that rural livestock ownership Observations 4,398 3,675 is also more likely to be joint as opposed to exclusive—for example, 37 percent of rural 1 All estimates are weighted using household sampling weights. Standard deviations in parentheses. Statistically significant differences women and 44 percent of rural men reported between men and women indicated by asterisks (***p<0.01, ***p<0.05, joint ownership of large livestock, compared * p<0.10). to 26‑27 percent that claimed exclusive 2 Livestock categories included the following: bulls, oxen, cows, steers, heifers, calves, goats, sheep, camels, horses, mules, and donkeys. ownership. 3 Because a household can own multiple livestock, a respondent could have exclusive as well as joint ownership of different animals. 60 LSMS+ Program in Sub-Saharan Africa 4. Financial account and mobile phone ownership 4.1. Individual‑level estimates Table 3.2 presents reporting on financial accounts financial account was roughly similar (around and mobile phone ownership among men and 24‑25 percent). Greater shares of respondents women, and Figure 3.6 breaks this down by urban across countries own a mobile phone, although and rural areas. Overall, the share of individuals they are mostly concentrated in urban areas owning a financial account is quite low, and there (Figure 3.6), and men are also significantly more are significant gender inequalities, particularly in likely than women to own one. Gender disparities in Ethiopia and Tanzania. In Ethiopia, 18 percent of mobile phone ownership also widen in rural areas. women and 31 percent of men owned a financial For both financial accounts and mobile phones, account; these shares were 9 and 15 percent, nearly all respondents reported exclusive as respectively, in Tanzania; and in Malawi the opposed to joint ownership. share of men and women reporting owning a Table 3.2 Share of women and men owning a mobile phone and financial account Ethiopia Tanzania Malawi Women Men Diff. Women Men Diff. Women Men Diff. (p‑value) (p‑value) (p‑value) Mobile phone Overall 0.27 0.50 0.00 0.58 0.78 0.00 0.36 0.56 0.00 Exclusive 0.25 0.48 0.00 0.56 0.76 0.00 0.35 0.55 0.00 Joint 0.01 0.03 0.00 0.02 0.02 0.63 0.01 0.01 0.39 Number of 7,846 6,899 1,325 1,070 2,595 2,075 respondents2 Financial account Overall 0.18 0.31 0.00 0.09 0.15 0.00 0.25 0.24 0.90 Exclusive 0.16 0.27 0.00 0.09 0.15 0.00 0.23 0.22 0.83 Joint 0.03 0.06 0.00 0.001 ‑ 0.62 0.02 0.03 0.16 Number of 7,945 6,965 1,557 1,407 2,595 2,075 respondents2 1 All estimates are weighted using household sampling weights. 2 Because of missing responses among the eligible sample in either the mobile phones or financial accounts module, the number of observations can vary. Section 3. Asset ownership and rights 61 Figure 3.6 Mobile phone and financial account ownership across countries Men Women Share owning a mobile phone Share owning a financial account 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Ethiopia Tanzania Malawi Ethiopia Tanzania Malawi 2018/2019 2019 2016/2017 2018/2019 2019 2016/2017 1 All estimates are weighted using household sampling weights. 2 All differences between men and women were statistically significant, with the exception of financial account ownership (urban and rural) in Malawi, as well as mobile phone ownership (urban) in Tanzania. © World Bank 4.2. Does individual‑level data collection affect household estimates of mobile and financial account ownership? Another perspective on the value of individual‑level interviews arises from understanding how this data might change household‑level estimates of financial account and mobile phone ownership, compared to the traditional survey approach of asking one individual per household. Table 3.3 compares aggregated household‑level statistics on financial account and mobile phone ownership for the LSMS+ supported surveys, along with national surveys from the same countries under the LSMS‑ISA program. As discussed in Section 1, comparisons of the two sets of surveys within each country, however, do need to be conditioned on differences not only in survey design/implementation, but also time‑varying factors that can affect the 62 LSMS+ Program in Sub-Saharan Africa interpretation of differences across the two sets of The comparisons do show that the LSMS+ surveys within each country. This is particularly the supported surveys reveal a higher share of case for Ethiopia and Tanzania, where the LSMS+ households reporting financial account ownership, and LSMS‑ISA surveys for each country were and mobile phone ownership in rural areas for conducted 3‑4 years apart, and where the trends Tanzania and Malawi — although the Malawi in mobile phone and financial account ownership comparison is the only one to address external should be linked to changes in access over time as time‑varying factors, such as country‑level well as, potentially, survey design. improvements in infrastructure and economic growth, that could affect mobile phone and financial Table 3.3 Comparing household‑level estimates of financial account and mobile phone ownership, by LSMS+ and LSMS‑ISA surveys conducted in the same countries1 Ethiopia Tanzania Malawi ESS3 LSMS+ Diff. NPS4 LSMS+ Diff. IHS4 LSMS+ Diff. 2015-16 (ESS4) (p‑value) 2014‑15 (NPS5) (p‑value) 2016‑17 (IHPS) (p‑value) 2018‑19 2018‑19 2016 (1) (2) (1) ‑ (2) (3) (4) (1) ‑ (2) (5) (6) (1) ‑ (2) Full sample Owns financial 0.35 0.44 0.00 0.21 0.26 0.14 0.26 0.36 0.00 account (Y=1 N=0) Number of financial 0.58 0.85 0.00 0.28 1.46 0.00 ‑ ‑ ‑ accounts Owns mobile phone 0.54 0.55 0.71 0.79 0.88 0.00 0.53 0.60 0.00 (Y=1 N=0) Number of 4,953 6,770 989 1,184 6,894 2,457 households Urban Owns financial 0.70 0.76 0.04 0.40 0.46 0.30 0.50 0.58 0.09 account (Y=1 N=0) Number of financial 1.37 1.71 0.00 0.50 2.78 0.00 ‑ ‑ ‑ accounts Owns mobile phone 0.86 0.87 0.78 0.96 0.96 0.78 0.83 0.84 0.56 (Y=1 N=0) Number of 1,681 3,655 1,557 1,407 2,595 2,075 households Rural Owns financial 0.22 0.28 0.02 0.12 0.17 0.14 0.20 0.28 0.00 account (Y=1 N=0) Number of financial 0.29 0.43 0.00 0.14 0.87 0.00 - - - accounts Owns mobile phone 0.43 0.40 0.40 0.72 0.84 0.00 0.45 0.51 0.01 (Y=1 N=0) Number of 3,272 3,115 550 682 5,509 1,804 households 1 All estimates are weighted using household sampling weights. “-” indicates data was not available for that variable. 2 Because of missing responses among the eligible sample in either the mobile phones or financial accounts module, the number of observations can vary. Section 3. Asset ownership and rights 63 account ownership. Figure 3.7 also shows, across Figure 3.7 Share of households with a financial the distribution of household nonfood per capita account in LSMS+ and comparison survey, by expenditure, that higher reporting of financial percentile of per capita expenditure (Malawi) account ownership in Malawi under individual‑level interviews occurs among households at relatively LSMS+ (IHPS 2016) IHS4, 2016-17 higher percentiles (above the 50th percentile). As a result — along with the advantages of having 0.8 individual‑level data — from a basic household welfare perspective, understanding the extent to Share of HH with a financial account 0.6 which households overall have access to certain services and technology is also critical, and how more nuanced collection of this information within 0.4 households contributes to this understanding. 0.2 0 0 20 40 60 80 100 Percentile of per capita exp. (nonfood) Share owning a financial account: in Ethiopia 18 31 women % men % in Tanzania 9 women % 15 men % © Valentina Costa / World Bank in Malawi 24-25 men and women % 64 LSMS+ Program in Sub-Saharan Africa Section 3 / Summary + Findings on land ownership/ Findings on financial account/ Individual interviews following rights, and livestock ownership: mobile phone ownership: recommendations under 2019 UN guidelines on self‑reported data over different types of + + ownership and rights, and Even among exclusive reported landowners, The share of individuals owning a financial exclusive and joint roles, (a) economic ownership/decision making over account is quite low, and there are significant provide a clearer picture of the proceeds from selling land is more gender inequalities, particularly in Ethiopia ownership of and rights to likely to be joint/distributed across multiple and Tanzania. assets within households, household members. particularly among women; (b) minimize distortionary + + proxy respondent effects and Greater shares of respondents across In Malawi, matrilineal traditions over land intra‑household discrepancies countries do own a mobile phone, although ownership and inheritance lead to a more in reporting; and (c) can reveal they are mostly concentrated in urban areas, equitable distribution of reported and hidden assets. and men are also significantly more likely economic ownership across men and women, than women to own one. Gender disparities compared to Ethiopia and Tanzania. Gender also widen in rural areas. + disparities tended to widen, however, among Within the LSMS+ supported landowners in rights to sell and bequeath surveys, individual‑level modules across all countries. A separate study (Kilic et al., 2020b) compared the Malawi LSMS+/ + on land span exclusive and joint ownership (reported, economic, IHPS with the concurrent IHS4 that asked only For both financial accounts and mobile documented) and rights (sell and one “most knowledgeable” respondent about phones, nearly all respondents reported bequeath) over dwelling and household members’ land ownership and exclusive as opposed to joint ownership. non‑dwelling land. Individual rights—finding that the IHS4 resulted in higher modules on financial accounts rates of exclusive reported and economic and mobile phone ownership ownership of agricultural land among + also ask about exclusive versus men, and lower rates of joint reported and The individual‑interview approach in LSMS+ joint roles. economic ownership among women. supported surveys also leads to a higher share of households reporting financial account ownership across all countries, and + mobile phone ownership in rural areas for Across countries, agreement among couples Tanzania and Malawi. over ownership and rights is relatively high, with agreement spanning a range of 70‑85 percent of parcels depending on the country and type of ownership/rights. There are still substantial areas of disagreement, although disagreement is not necessarily concentrated in any one particular scenario or category of ownership/rights, and country context matters in interpreting disagreement. + In Ethiopia, large livestock ownership was high in rural areas, and significantly higher for men (about 58 percent of rural women and 64 percent of rural men). There was, however, also substantial ownership in urban areas (about 17 and 20 percent of urban women and men, respectively). Livestock ownership in rural areas was also much more likely to be joint. Section 3. Asset ownership and rights 65 Annex 1. SDG targets and indicators requiring individual‑level data on economic outcomes Goal 1. End poverty in all its Goal 5. Achieve gender small‑ and medium‑sized enterprises, forms everywhere equality and empower all including through access to financial women and girls services. Target 1.2 / By 2030, reduce at least by 8.3.1 / Proportion of informal employment half the proportion of men, women and Target 5.4 / Recognize and value unpaid in non‑agriculture employment, by sex. children of all ages living in poverty in care and domestic work through the all its dimensions according to national provision of public services, infrastructure Target 8.5 / By 2030, achieve full and definitions (also Target 1.1 on eradicating and social protection policies and the productive employment and decent work extreme poverty for all). promotion of shared responsibility within for all women and men, including for young 1.2.1 / Proportion of population living below the household and the family as nationally people and persons with disabilities, and the national poverty line, by sex and age. appropriate. equal pay for work of equal value. 5.4.1 / Proportion of time spent on unpaid 8.5.1 / Average hourly earnings of female Target 1.4 / By 2030, ensure that all men and male employees, by occupation, age domestic and care work, by sex, age and and women, in particular the poor and the and persons with disabilities. location. vulnerable, have equal rights to economic 8.5.2 / Unemployment rate, by sex, age resources, as well as access to basic services, Target 5.a / Undertake reforms to and persons with disabilities. ownership and control over land and other give women equal rights to economic forms of property, inheritance, natural resources, as well as access to ownership Target 8.6 / By 2020, substantially reduce resources, appropriate new technology and and control over land and other forms of the proportion of youth not in employment, ¬financial services, including micro¬finance. property, financial services, inheritance education or training. 1.4.2 / Proportion of total adult population and natural resources, in accordance with 8.6.1 / Proportion of youth (aged 15–24 with secure tenure rights to land, (a) with national laws. years) not in education, employment or legally recognized documentation, and (b) 5.a.1 / (a) Proportion of total agricultural training. who perceive their rights to land as secure, population with ownership or secure rights Target 8.10 / Strengthen the capacity by sex and type of tenure. over agricultural land, by sex; and (b) share of domestic financial institutions to of women among owners or rights‑bearers encourage and expand access to banking, of agricultural land, by type of tenure. Goal 2. End hunger, achieve insurance and financial services for all. food security and improved Target 5.b / Enhance the use of enabling 8.10.2 / Proportion of adults (15 years nutrition and promote technology, in particular information and and older) with an account at a bank sustainable agriculture communications technology, to promote or other financial institution or with a the empowerment of women. mobile‑money‑service provider. Target 2.3 / By 2030, double the 5.b.1 Proportion of individuals who own a agricultural productivity and incomes of mobile telephone, by sex. Goal 10. Reduce inequality small‑scale food producers, in particular women, indigenous peoples, family within and among countries farmers, pastoralists and fishers, including Goal 8. Promote sustained, Target 10.2 / By 2030, empower and through secure and equal access to land, inclusive and sustainable promote the social, economic and political other productive resources and inputs, economic growth, full and inclusion of all, irrespective of age, sex, knowledge, financial services, markets productive employment and disability, race, ethnicity, origin, religion or and opportunities for value addition and decent work for all economic or other status. non‑farm employment. 10.2.1 / Proportion of people living below Target 8.3 / Promote 2.3.1 / Volume of production per labor 50 percent of median income, by sex, age development‑oriented policies that unit (day) by classes of farming/pastoral/ and persons with disabilities. support productive activities, decent job forestry enterprise size. creation, entrepreneurship, creativity 2.3.2 / Average income of small‑scale and innovation, and encourage the food producers, by sex and indigenous formalization and growth of micro‑, status. 66 LSMS+ Program in Sub-Saharan Africa © Valentina Costa / World Bank Section 3. Asset ownership and rights 67 Annex Table A1. Tanzania. OLS Regressions: Women’s labor market participation LSMS+ (NPS5, 2019‑20) NPS4 (2014‑15) Dependent variable: Baseline Baseline w/asset Individual in the past 7 days participated in … ownership vars activities (Y=1, N=0) Agr. Non‑agr. Agr. Non‑agr. Agr. Non‑agr. HH head ‑0.119 0.072 ‑0.094 0.072 ‑0.077 0.109 [‑1.56] [0.98] [‑1.36] [1.05] [‑1.04] [1.25] Age: 18‑24 ‑0.109 ‑0.231*** ‑0.017 ‑0.212*** ‑0.132*** ‑0.150 [‑1.37] [‑3.40] [‑0.23] [‑3.13] [‑2.95] [‑1.64] Age: 25‑34 ‑0.010 ‑0.082 0.022 ‑0.078 ‑0.035 ‑0.047 [‑0.16] [‑1.23] [0.34] [‑1.19] [‑0.61] [‑0.50] Age: 45‑54* 0.066 ‑0.106 0.052 ‑0.112 0.067 ‑0.146* [0.95] [‑1.36] [0.80] [‑1.51] [1.05] [‑1.76] Age: 55+ 0.184** ‑0.268*** 0.185** ‑0.263*** 0.018 ‑0.326*** [2.09] [‑3.12] [2.23] [‑3.11] [0.16] [‑3.70] Years of school, if attended 0.016 ‑0.088*** 0.014 ‑0.097*** 0.020 ‑0.013 [0.53] [‑2.91] [0.45] [‑3.14] [0.58] [‑0.46] Years of school sq. 0.000 0.008*** 0.000 0.007*** ‑0.001 0.001 [0.01] [4.16] [0.04] [3.95] [‑0.51] [0.77] Married ‑0.017 ‑0.040 ‑0.043 ‑0.035 0.125*** ‑0.113 [‑0.18] [‑0.80] [‑0.50] [‑0.70] [2.73] [‑1.63] Separated/divorced ‑0.111 0.092 ‑0.116 0.088 ‑0.008 ‑0.007 [‑1.23] [1.14] [‑1.43] [1.18] [‑0.10] [‑0.07] Widowed 0.047 ‑0.052 0.074 ‑0.062 0.141 ‑0.040 [0.45] [‑0.62] [0.74] [‑0.74] [1.59] [‑0.40] Months away from HH ‑0.008 ‑0.015* ‑0.010 ‑0.016* 0.007 0.001 [‑0.42] [‑1.77] [‑0.52] [‑1.79] [0.54] [0.07] Log HH size 0.133*** ‑0.092** 0.140*** ‑0.067* 0.051 ‑0.072 [2.70] [‑2.34] [2.93] [‑1.73] [1.42] [‑1.54] HH dependency ratio† ‑0.060* 0.011 ‑0.070** 0.009 0.005 ‑0.033 [‑1.95] [0.51] [‑2.24] [0.40] [0.23] [‑1.51] HH has electricity ‡ ‑0.095* ‑0.037 ‑0.065 ‑0.058 ‑0.125*** 0.118** [‑1.88] [‑0.72] [‑1.32] [‑1.13] [‑2.64] [2.61] HH has piped water ‡ ‑0.014 0.127** 0.002 0.125** ‑0.034 0.066 [‑0.26] [2.11] [0.03] [2.14] [‑0.75] [1.36] HH: walls made of concrete ‡ ‑0.121 ‑0.009 ‑0.093 ‑0.032 ‑0.191*** 0.077 [‑1.60] [‑0.11] [‑1.19] [‑0.39] [‑3.42] [1.06] Individual owns a mobile phone 0.064 0.113** [1.12] [2.22] Individual owns a financial asset ‑0.063 0.165** [‑0.89] [2.14] Individual reported to own land 0.244*** ‑0.031 [4.47] [‑0.61] Rural 0.073 ‑0.046 0.115* ‑0.016 0.314*** ‑0.024 [0.96] [‑0.76] [1.69] [‑0.26] [4.96] [‑0.38] Constant 0.754** 1.164*** 0.507 1.106*** 0.407* 0.535** [2.09] [2.81] [1.54] [2.73] [1.74] [2.18] Observations 760 760 760 760 734 734 R‑squared 0.469 0.376 0.503 0.390 0.376 0.204 1 All estimates weighted by household sample weight. Standard errors in brackets, ***p<0.01, **p<0.05, *p<0.10. 68 LSMS+ Program in Sub-Saharan Africa Annex Table A2. Malawi. OLS Regressions: Women’s labor market participation LSMS+ (IHPS, 2016) IHS4 (2016‑17) Dependent variable: Baseline Baseline w/asset Individual in the past 7 days participated in … ownership vars activities (Y=1, N=0) Agr. Non‑agr. Agr. Non‑agr. Agr. Non‑agr. HH head 0.019 0.074*** 0.005 0.076*** 0.024 0.217*** [0.62] [2.72] [0.16] [2.78] [1.45] [12.82] Age: 18‑24 ‑0.054 ‑0.121*** ‑0.040 ‑0.106*** ‑0.081*** ‑0.115*** [‑1.62] [‑4.10] [‑1.19] [‑3.64] [‑4.58] [‑6.22] Age: 25‑34 ‑0.025 ‑0.008 ‑0.022 ‑0.004 ‑0.022 ‑0.012 [‑0.91] [‑0.33] [‑0.78] [‑0.17] [‑1.34] [‑0.76] Age: 45‑54* 0.027 ‑0.043 0.024 ‑0.044 0.024 ‑0.089*** [0.86] [‑1.21] [0.75] [‑1.24] [1.20] [‑4.43] Age: 55+ 0.096** ‑0.236*** 0.091** ‑0.231*** ‑0.028 ‑0.247*** [2.53] [‑7.28] [2.39] [‑7.06] [‑1.42] [‑13.55] Years of school, if attended 0.005 ‑0.001 0.005 ‑0.004 0.008** ‑0.007** [0.98] [‑0.22] [0.84] [‑0.65] [2.57] [‑2.53] Years of school sq. ‑0.000 0.001** ‑0.000 0.001** ‑0.000*** 0.001*** [‑0.78] [2.57] [‑0.82] [2.37] [‑2.73] [4.80] Married 0.065* 0.025 0.045 0.017 0.109*** 0.096*** [1.67] [0.70] [1.14] [0.46] [7.64] [7.02] Separated/divorced ‑0.055 0.046 ‑0.063 0.039 0.068*** 0.107*** [‑1.08] [0.96] [‑1.23] [0.82] [3.05] [4.40] Widowed ‑0.139** ‑0.006 ‑0.150*** ‑0.010 ‑0.017 ‑0.019 [‑2.47] [‑0.11] [‑2.69] [‑0.19] [‑0.56] [‑0.63] Months away from HH 0.005 ‑0.007 0.006 ‑0.007 ‑0.009*** 0.002 [0.96] [‑1.24] [1.05] [‑1.20] [‑3.21] [0.66] Log HH size 0.052** 0.027 0.050* 0.029 0.033** ‑0.008 [1.99] [1.13] [1.88] [1.21] [2.39] [‑0.62] HH dependency ratio† ‑0.021 ‑0.010 ‑0.022* ‑0.008 ‑0.011 0.006 [‑1.65] [‑0.73] [‑1.72] [‑0.60] [‑1.63] [0.99] HH has electricity ‡ ‑0.090*** 0.011 ‑0.094*** 0.004 ‑0.065*** 0.030 [‑2.71] [0.28] [‑2.69] [0.12] [‑3.22] [1.32] HH has piped water ‡ ‑0.056 0.019 ‑0.054 0.005 ‑0.074*** 0.023 [‑1.56] [0.54] [‑1.48] [0.14] [‑2.98] [0.99] HH: walls made of concrete ‡ ‑0.103* ‑0.001 ‑0.098* ‑0.000 0.057 0.043 [‑1.93] [‑0.01] [‑1.85] [‑0.01] [1.46] [1.10] Individual owns a mobile phone ‑0.002 0.009 [‑0.08] [0.34] Individual owns a financial asset 0.034* 0.083*** [1.80] [4.02] Individual reported to own land 0.070*** ‑0.025 [3.09] [‑1.16] Rural 0.266*** ‑0.082** 0.249*** ‑0.081** 0.114*** ‑0.100*** [8.13] [‑2.14] [7.45] [‑2.15] [2.82] [‑3.50] Constant 0.427* 0.201* 0.441* 0.215** 0.700*** 0.260*** [1.92] [1.96] [1.92] [2.09] [8.86] [4.04] Observations 2,596 2,596 2,596 2,596 8,002 8,002 R‑squared 0.213 0.094 0.218 0.098 0.210 0.113 1 All estimates weighted by household sample weight. Standard errors in brackets, ***p<0.01, **p<0.05, *p<0.10. Annex 69 Annex Table A3. Ethiopia. OLS Regressions: Women’s labor market participation /01 LSMS+ (ESS4, 2018‑19) ESS3 (2015‑16) Dependent variable: Baseline Baseline w/asset Individual in the past 7 days participated in … ownership vars activities (Y=1, N=0) Agr. Non‑agr. Agr. Non‑agr. Agr. Non‑agr. HH head 0.024 0.053*** 0.025 0.045** 0.033 0.081*** [0.86] [2.64] [0.91] [2.20] [1.41] [3.60] Age: 18‑24 ‑0.083*** ‑0.057** ‑0.075*** ‑0.048** ‑0.030 ‑0.092*** [‑3.00] [‑2.56] [‑2.70] [‑2.15] [‑1.06] [‑4.98] Age: 25‑34 ‑0.037 0.018 ‑0.033 0.020 0.013 ‑0.035** [‑1.56] [1.01] [‑1.39] [1.10] [0.48] [‑2.45] Age: 45‑54* ‑0.002 ‑0.040** ‑0.004 ‑0.036* ‑0.012 ‑0.036* [‑0.08] [‑1.97] [‑0.13] [‑1.81] [‑0.39] [‑1.81] Age: 55+ ‑0.101** ‑0.085*** ‑0.099** ‑0.085*** ‑0.015 ‑0.104*** [‑2.42] [‑3.15] [‑2.27] [‑3.10] [‑0.42] [‑4.18] Never attended school ‑0.015 0.041 ‑0.004 0.035 ‑0.075* 0.061** [‑0.28] [1.24] [‑0.07] [1.01] [‑1.74] [2.05] Years of school, if attended ‑0.002 ‑0.014 ‑0.002 ‑0.014 0.016 ‑0.016 [‑0.13] [‑1.47] [‑0.19] [‑1.46] [1.61] [‑1.64] Years of school sq. ‑0.000 0.002*** ‑0.000 0.002*** ‑0.001** 0.002*** [‑0.61] [3.06] [‑0.46] [2.74] [‑2.55] [2.60] Married 0.006 ‑0.022 ‑0.005 ‑0.024 0.026 0.005 [0.22] [‑1.05] [‑0.17] [‑1.14] [0.92] [0.29] Separated/divorced ‑0.078* 0.107*** ‑0.076* 0.107*** ‑0.032 0.042 [‑1.87] [3.25] [‑1.83] [3.31] [‑0.81] [1.58] Widowed ‑0.042 ‑0.013 ‑0.045 ‑0.010 ‑0.034 ‑0.021 [‑0.92] [‑0.45] [‑1.00] [‑0.37] [‑0.84] [‑0.71] Months away from HH ‑0.010 ‑0.005 ‑0.009 ‑0.007 0.003 ‑0.006 [‑1.34] [‑1.19] [‑1.29] [‑1.38] [0.32] [‑1.10] Log HH size 0.040* ‑0.003 0.036* 0.003 0.004 ‑0.025** [1.94] [‑0.16] [1.75] [0.15] [0.20] [‑2.06] HH dependency ratio† 0.047 ‑0.021 0.040 ‑0.007 ‑0.087* ‑0.031 [0.85] [‑0.90] [0.70] [‑0.30] [‑1.76] [‑1.10] HH has electricity ‡ ‑0.107* 0.156*** ‑0.104* 0.144*** ‑0.093** 0.085*** [‑1.89] [3.40] [‑1.85] [3.07] [‑2.56] [3.53] HH has piped water ‡ ‑0.101*** ‑0.033 ‑0.092*** ‑0.040 0.023 0.018 [‑3.36] [‑0.79] [‑3.03] [‑0.90] [0.81] [0.60] HH: walls made of concrete ‡ ‑0.041* ‑0.005 ‑0.033 ‑0.016 0.002 0.004 [‑1.65] [‑0.17] [‑1.35] [‑0.50] [0.07] [0.12] 70 LSMS+ Program in Sub-Saharan Africa Annex Table A3. Ethiopia. OLS Regressions: Women’s labor market participation /02 LSMS+ (ESS4, 2018‑19) ESS3 (2015‑16) Dependent variable: Baseline Baseline w/asset Individual in the past 7 days participated in … ownership vars activities (Y=1, N=0) Agr. Non‑agr. Agr. Non‑agr. Agr. Non‑agr. Religion: Catholic* 0.211** ‑0.024 0.238*** ‑0.020 0.317*** 0.034 [2.50] [‑0.78] [2.77] [‑0.67] [3.45] [0.65] Religion: Protestant ‑0.017 0.003 ‑0.006 0.003 0.091** 0.005 [‑0.44] [0.08] [‑0.16] [0.11] [2.33] [0.29] Religion: Muslim ‑0.101*** ‑0.055*** ‑0.094*** ‑0.049*** 0.023 0.008 [‑3.28] [‑3.27] [‑3.00] [‑2.92] [0.59] [0.48] Religion: Traditional ‑0.140 0.175** ‑0.119 0.180** 0.149 ‑0.006 [‑0.77] [2.22] [‑0.68] [2.36] [1.49] [‑0.09] Religion: Pegan 0.193** ‑0.080 0.155** ‑0.076 0.153 0.152 [2.55] [‑1.44] [2.13] [‑1.35] [0.79] [0.89] Religion: Wakefta 0.144 0.043 0.152 0.048 0.370* ‑0.080** [1.13] [0.38] [1.18] [0.44] [1.78] [‑2.48] Religion: Other ‑0.128 ‑0.254*** ‑0.102 ‑0.319*** ‑0.108 ‑0.046 [‑1.10] [‑3.48] [‑0.93] [‑4.10] [‑1.30] [‑1.58] HH faced shock affecting income/assets ‑0.044 0.005 ‑0.044 0.009 0.005 ‑0.005 [‑0.63] [0.09] [‑0.60] [0.16] [0.18] [‑0.36] Individual owns a mobile phone ‑0.046* 0.032 [‑1.79] [1.27] Individual owns a financial asset 0.007 0.073*** [0.28] [3.30] Individual reported to own land 0.087*** ‑0.012 [2.93] [‑0.75] Rural 0.279*** ‑0.059 0.254*** ‑0.048 ‑0.137*** 0.056*** [5.26] [‑1.37] [4.88] [‑1.09] [‑5.48] [3.16] Constant 0.094 0.072 0.103 0.052 0.629*** 0.171** [0.84] [1.09] [0.95] [0.80] [4.85] [2.01] Observations 6,341 6,341 6,293 6,293 6,198 6,198 R‑squared 0.318 0.213 0.324 0.221 0.189 0.180 1 All estimates weighted by household sample weight. Standard errors in brackets, ***p<0.01, **p<0.05, *p<0.10. Annex 71 Acknowledgments This work was made possible by funding from the Umbrella Facility for Gender Equality Trust Fund, the World Bank Trust Fund for Statistical Capacity Building, and the International Fund for Agricultural Development. The authors would like to thank Caren Grown, Kathleen Beegle, and Calogero Carletto for their valuable inputs into the design and implementation of Living Standards Measurement Study – Plus (LSMS+) program activities. We are grateful to Amparo Palacios-Lopez, Isis Gaddis, and Sydney Gourlay for their inputs into the design of the questionnaires for the surveys that have been supported by the LSMS+ program over the period 2016-2020. We would like to thank (i) Alemayehu Ambel, Aphichoke Kotikula, and Manex Bule Yonis for their comments on the earlier version of this report, (ii) Pietro Bartoleschi and Cristina Vitelli for creating the visual identity of this report and transitioning the content into the publication template, (iii) Ilaria Lanzoni and Giulia Altomare for their support throughout the publication process, and (iv) Sile O’Broin for her editorial review of the earlier version of this report. Our heartfelt thanks and gratitude are due to the survey management and field staff at the partner National Statistical Offices that have been supported by the LSMS+, along with the World Bank Living Standards Measurement Study (LSMS) team members that contributed to each country program. Talip Kilic and Heather Moylan oversaw the provision of the World Bank technical assistance to each NSO on the design and implementation of the survey supported by the LSMS+ program. From Ethiopia, we would like to thank (i) the Central Statistics Agency (CSA) Ethiopia Socioeconomic Survey (ESS) management team members: Biratu Yigezu, Tesfaye Kebede Amare, and Habekiristos Beyene Haile, (ii) the CSA ESS 2018/19 field staff, and (iii) the World LSMS team members: Alemayehu Ambel, Asmelash Haile Tsegay, Manex Bule Yonis, Wondu Yemanebirhan Kassa, and Seblewangel Ayalew Woreta. From Malawi, we would like to thank (i) the Malawi National Statistical Office (NSO) Integrated Household Survey (IHS) management team members: Mercy Kanyuka, Jameson Ndawala, Lizzie Chikoti, Bright Mvula, Lameck Million, Imran Chiosa, Twikaleghe Mwalwanda, Sautso Wachepa, Glory Mshali, Dama Kaipa, Charles Chakanza, Charles Mbewe, Steve Pakundikana and Henderson Chilenje, (ii) the NSO Integrated Household Panel Survey (IHPS) 2016 field staff, and (iii) the World LSMS team members: Wilbert Vundru Drazi, Fiona Nattembo, and John Ilukor. From Tanzania we would like to thank (i) the Tanzania National Bureau of Statistics (NBS) National Panel Survey (NPS) management team that is led by Mlemba Abassy, (ii) the NBS NPS 2019/20 field staff, and (iii) the World Bank LSMS team members: Jonathan G. Kastelic, Harriet Kasidi Mugera, and Darcey Johnson. Photo credits go to Valentina Costa of World Bank (for Malawi), Midas Touch Communication PLC (for Ethiopia), and Khalfan Mlulu (for Tanzania). 72 Living Standards Measurement Study – Plus (LSMS+) www.worldbank.org/lsmsplus II LSMS+ Program in Sub-Saharan Africa