Gender gaps in youth employment: a spatial approach Aslihan Arslan International Fund for Agricultural Development (IFAD) Joint work with Eva-Maria Egger (IFAD) and David E. Tschirley (MSU) Future of Work in Agriculture Conference World Bank, Washington, D.C. 20.03.2019
26
Embed
Gender gaps in youth employment: a spatial approach · - Employment transformation (Filmer and Fox 2014) • Wage employment important for youth employment challenge • Young rural
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Gender gaps in youth employment: a spatial approach
Aslihan ArslanInternational Fund for Agricultural Development (IFAD)
Joint work with Eva-Maria Egger (IFAD) and David E. Tschirley (MSU)
Future of Work in Agriculture ConferenceWorld Bank, Washington, D.C.
20.03.2019
Motivation
• Three transformations increase the importance of rural wage employment:- Structural transformation reaching rural areas (IFAD 2016)- Transformation of agri-food systems (Reardon et al. 2015)- Employment transformation (Filmer and Fox 2014)
• Wage employment important for youth employment challenge
• Young rural women face a triple burden:- Rural areas lag behind in the transformation. Thus, connectivity & mobility
important to access wage employment. (Christiaensen, et al. 2013)- younger women’s mobility constrained by social norms and domestic work
(Chakravarty, Das and Vaillant 2017)
• Wage employment of young rural women can contribute to - the empowerment of young women.- speed up the demographic transition hence contribute to rural
transformation (Stecklov & Menashe-Oren 2019)
Global Youth over the Rural Opportunity Space (Rural Development Report 2019)
3
Figure: Share of global rural youth within rural opportunity space (all low and middle income countries)
• Globally 67% of rural youth live in areas with high agricultural potential
• For welfare outcomes, commercialization potential (=connectivity) matters more
Individual employment data from 12 surveys
Total: ~420 K 4
Country Survey Name Year Sample Size (indiv.)
Geo-locationlevels
Sub-Saharan AfricaEthiopia Ethiopian Socioeconomic Survey 2015/2016 23,393 EAs
Malawi Fourth Integrated Household Survey 2016/2017 53,885 EAsNiger National Survey on Household Living
Conditions on Agriculture - Panel2014 22,671 EAs
Nigeria General Household Survey- Panel 2015/2016 24,807 EAsTanzania National Panel Survey 2014/2015 16,285 EAsUganda The Uganda National Panel Survey 2013/2014 9,376 EAs
Latin AmericaMexico Encuesta nacional de ingresos y gastos de
los hogares 2016 256,448 EAs
Nicaragua Encuesta nacional de hogares sobre medición de nivel de vida
2014 29,381 Municipality
Peru Encuesta nacional del hogares 2016 (Anual) – Condiciones de vida y pobreza
2016 134,235 EAs
AsiaCambodia Cambodia Socio-Economic Survey 2014 53,968 VillageIndonesia Indonesian Family Life Survey 2014 58,300 EAsNepal Nepal Living Standards Survey 2010 28,670 Village
Gender gaps in LFP and in wage employment
5
Research questions
1. To what extent can connectivity explain the gender gap in:- youth LFP? - non-farm wage participation (NFWP)?
2. Does this differ for the AFS vs. non-AFS sector, and by region?
Two spatial aspects of connectivity matter:• Opportunities available where you live: Rural Opportunity
Space population density• How long does it take to get to nearest city? travel time
Hypotheses: If there was no mobility constraint, …1. young women would be equally likely to access non-farm wage
employment as young men with the same connectivity𝑃𝑃(𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 = 1)𝑖𝑖
2. longer travel time would affect young women’s likelihood to access wage jobs within the same location as much as for young men 𝑃𝑃(𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 = 1)𝑖𝑖
𝐽𝐽
= 𝛼𝛼 + 𝛽𝛽1𝐹𝐹𝑊𝑊𝐹𝐹𝑖𝑖𝐽𝐽
+ 𝛽𝛽2𝑇𝑇𝑇𝑇𝐹𝐹𝑊𝑊𝑖𝑖𝐽𝐽
+ 𝛽𝛽3𝐹𝐹𝑊𝑊𝐹𝐹 ∗ 𝑇𝑇𝑇𝑇𝐹𝐹𝑊𝑊𝑖𝑖𝐽𝐽
+ 𝛽𝛽4𝑋𝑋𝑖𝑖𝐽𝐽 + 𝑊𝑊𝑖𝑖𝐽𝐽
𝑯𝑯𝟎𝟎: �𝜷𝜷𝟏𝟏 + �𝜷𝜷𝟑𝟑𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑻𝑱𝑱 = 𝟎𝟎
7
Data Sources
• Household survey data from 12 low and middle income countries (Mexico, Nicaragua, Peru, Cambodia, Indonesia, Nepal, Ethiopia, Malawi, Niger, Nigeria, Tanzania, Uganda)
• Merged with geo-spatial data on - population density (WorldPop project)- travel time to cities/towns (Open Street Map, Google
roads database, Global Human Settlement Layer)- greenness (Enhanced Vegetation Index )
8
Spatial data: Rural-urban gradient based on population density• 1km x 1km resolution population density maps for each country
globally (age and gender disaggregated)• order global sample of grids from least to most dense and define
quartiles of rural-urban gradient• Match to enumeration areas (or admin units) with available geo-codes
9
Table 1: Population density based rural-urban gradient vs. admin urbanization rates: Population shares by region.
Population density based rural-urban gradient Administrative
Regions Rural Semi-Rural Peri-Urban Urban urbanization rate
Summary statistics of sample (15 to 24 year old men and women)
11
Note: Point estimates are population means. Level of statistical significance of Wald-test of difference in means .01 - ***; .05 - **; .1 - *;
Table 1: Summary statistics of main variables for women and men aged 15 to 24 yearsMean Difference in means
Female MaleNo. of observations 61 899 61 411In labor force 0.51 0.65 -0.14***Of those in labor force, work in:Off-farm wage employment 0.24 0.28 -0.03***
Age 19.08 18.94 0.14***Currently in school 0.44 0.47 -0.04***Secondary school completed 0.39 0.39 -0.00Married 0.20 0.06 0.14***Household characteristics:Household size 6.24 6.63 -0.40***Dependency ratio 0.63 0.58 0.05***
Gender gap in LFP: marriage matters (IV for first step)
12
Table: Marginal effects from Probit estimation of LFP, separately for men and women, by age groups
Married Population density
Travel time to cities
15-24 years M 0.164*** -0.026***F -0.088*** -0.016M 0.048*** 0.023*F -0.070** 0.018
25-64 years M 0.169*** -0.007***F -0.084*** -0.008M 0.049*** 0.005F -0.068** 0.012
LFP (First step): Marginal effect of being a young female along the rural-urban gradient and by travel time (vs. young males)
13
By region:
Non-farm wage participation (NFWP, Second step): Marginal effect of being a young female – by sector (compared to young males)
14
NFWP (Second step): Marginal effect of being a young female by region & sector (compared to young males)
15
Rural-urban gradient
Travel time
16
How does the mobility constraint affect young women’s NFWP within R-U categories? (compared to young males)
Conclusion
• With innovative combination of household and geo-spatial data, we show that the gender gap in NFWP significantly changes over the rural-urban gradient
• Young women’s mobility constraint exists mostly for non-AFS wage employment, presumably because more AFS jobs are available in more remote areas
• Positive spill-overs of young women in employment: empowerment, faster demographic transition & rural transformation
Policy:» invest in connectivity!» Spatially differentiated approach to young women’s
• Chakravarty, S., Das, S. and J. Vaillant (2017), Gender and Youth Employment in Sub-Saharan Africa. A Review of Constraints and Effective Interventions, Policy Research Working Paper Series, 8245, Washington, DC: Worldbank.
• Christiaensen, L., De Weerdt, J., and Todo, Y. (2013). Urbanization and poverty reduction: The role of rural diversification and secondary towns. Agr. Econ., 44(4–5), pp. 435-447.
• Filmer, D. and Louise Fox (2014).Youth employment in sub-Saharan Africa: The World Bank.
• Stecklov G and A Menashe-Oren (2018), The Demography of Rural Youth in Developing Countries, Background paper for Rural Development Report 2019, Rome: IFAD.
• Tschirley DL, Snyder J, Dolislager M, Reardon T, Haggblade S, Goeb J, Traub L, Ejobi F, Meyer F (2015), Africa ' s unfolding diet transformation: implications for agrifood system employment, Journal of Agribusiness in Developing and Emerging Economies, 5(2): 102-136.
19
LFP along the rural-urban gradient
20
Travel time over the rural-urban gradient
Quartiles mean min max sdRural 120.81 3.73 1994.47 174.46Semi-Rural 66.92 3.25 584.15 68.37Peri-Urban 38.75 1.82 1050.22 39.39Urban 36.83 1.73 307.29 26.01
21
Wage income and poverty
22
Table: Share of total income from different sources along the rural urban gradient, by poverty status of household
Income source Rural-urban gradient Poor Non-poor Difference
Age 34.24 33.36 0.88***Currently in school 0.15 0.19 -0.04***Secondary school completed 0.36 0.43 -0.06***Married 0.51 0.45 0.06***Household characteristics:Household size 5.75 5.92 -0.17***Dependency ratio 0.80 0.70 0.10***Any income from farming 0.55 0.56 -0.00
Hypotheses to test: If there was no mobility constraint, …
1. young women would be equally likely to access non-farm wage employment as the average person in the labor force with the same connectivity
𝑃𝑃(𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 = 1)𝑖𝑖𝐽𝐽 = 𝛼𝛼 + 𝛽𝛽1𝐹𝐹𝑊𝑊𝐹𝐹𝑖𝑖
𝐽𝐽 + 𝛽𝛽2𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑖𝑖𝐽𝐽 + 𝛽𝛽3𝐹𝐹𝑊𝑊𝐹𝐹 ∗ 𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑖𝑖
𝐽𝐽 + 𝛽𝛽4𝑋𝑋𝑖𝑖𝐽𝐽 + 𝑊𝑊𝑖𝑖𝐽𝐽
J=connectivity𝐻𝐻0: �̂�𝛽3 = 0
2. longer travel time affects the likelihood to access wage jobs in the same location for young women as much as for the average person𝑃𝑃(𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊 = 1)𝑖𝑖
𝐽𝐽
= 𝛼𝛼 + 𝛽𝛽1𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑖𝑖𝐽𝐽
+ 𝛽𝛽2𝑇𝑇𝑇𝑇𝐹𝐹𝑊𝑊𝑖𝑖𝐽𝐽
+ 𝛽𝛽3𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌𝑌 ∗ 𝑇𝑇𝑇𝑇𝐹𝐹𝑊𝑊𝑖𝑖𝐽𝐽
+ 𝛽𝛽4𝑋𝑋𝑖𝑖𝐽𝐽 + 𝑊𝑊𝑖𝑖𝐽𝐽
J=location𝐻𝐻0: �̂�𝛽3 = 0
For both: holding everything else constant and adjusting for selection into labor force participation
25
Methodology: Blinder-Oaxaca decomposition of gender gap in wage participation
𝛽𝛽∗ : coefficients from pooled modelY: Participation in off-farm wage employmenti: individual; k: enumeration area; c: countryX: spatial (population density in EA, population density in 50km radius, travel time to nearest city); individual (age, education) and householdcharacteristics (dependency ratio, gender of head, income diversification)C: country dummies
Control for selection into labor force participation (Heckman)