This presentation was given by Vanya Slavchevska (CIAT), as part of the Annual Scientific Conference hosted by the University of Canberra and co-sponsored by the University of Canberra, the Australian Centre for International Agricultural Research (ACIAR) and CGIAR Collaborative Platform for Gender Research. The event took place on April 2-4, 2019 in Canberra, Australia.
Read more: https://www.canberra.edu.au/research/faculty-research-centres/aisc/seeds-of-change and https://gender.cgiar.org/annual-conference-2019/
2. Rationale
Migration originating from rural areas is predominantly male
(Mueller et al. 2015) raising concerns about the consequences of
migration on sending rural communities in terms of
Women’s work and empowerment; traditional gender norms
Agricultural productivity and production, etc.
Household food security
In Nepal – >90% of international migrants are men; women
constitute around 60% of all agricultural workers
3. Research objectives
To examine the impacts of male-dominated rural outmigration on
sending communities:
In particular, the analysis aims to shed light on:
1. how outmigration influences women’s and men’s work in agriculture;
2. whether it also influences changes in decision-making about
agriculture; and
3. what the impacts on agricultural production and food security are.
4. Conceptual Framework
The New Economics of Labor Migration (NELM)
Migration as a household rather than individual decision
Migration affects sending communities mainly through 2 channels:
the loss of migrant’s labor, and
the remittance income.
5. Methodology – Individual-level analysis
𝑌𝑖ℎ = 𝛼 + 𝛽1 𝑀1ℎ + 𝛽2 𝑀2ℎ + 𝛽2 𝑅ℎ + 𝛾𝑋𝑖ℎ + 𝜀𝑖 (1)
𝑌𝑖ℎ is a set of different indicators for women’s and men’s work in
agriculture and outside of agriculture.
We model women’s and men’s labor allocation as a function of:
whether the individual lives in a household with an international migrant
(𝑀1ℎ);
internal migrant (𝑀2ℎ);
the (log) remittance income (𝑅ℎ) received by the household; and
the individual, household and community characteristics, 𝑋𝑖ℎ, and and 𝜀𝑖 is
the error term
6. Methodology – Household (farm)-level analysis
𝑌ℎ = 𝛼 + 𝛽1 𝑀1ℎ + 𝛽2 𝑀2ℎ + 𝛽2 𝑅ℎ + 𝛾𝑋ℎ + 𝜀𝑖 (2)
𝑌𝑖ℎ is a set of different indicators for decision-making on the
farm, farm production, productivity and food security
We model women’s and men’s labor allocation as a function of:
whether the individual lives in a household with an international
migrant (𝑀1ℎ);
internal migrant (𝑀2ℎ);
the (log) remittance income (𝑅ℎ) received by the household; and
the household and community characteristics, 𝑋ℎ:
7. Data
Source: “Technical Report on Survey of Migration and Women’s Empowerment in Agriculture” prepared by Nepa School
of Social Sciences and Humanities, September 2, 2017.
Primary survey data
collected August-September
2017
a sample of 1002
households
from 5 districts --
Achham, Rolpa,
Nawalparasi, Makwanpur
and Jhapa
representative at
district-level
Multi-topic survey
i) A-WEAI administered to
ONE individual from each
household.
ii) The Food Insecurity
Experience Scale (FIES).
8. Nepali Migration
International migration is an important HH livelihoods diversification strategy
Remittance share in GDP is 29.2% (WDI)
International migration has become more important than internal migration –
around 15% of working-age population in our sample are current international migrants
less than 3% of working-age individuals in our sample are classified as current internal migrants
Men dominate migration -- more than 93% of current migrants are men
Migrants tend to be:
younger than the average working-age population; and
better educated – only 9% of migrants, compared to 29% of the working-age population, have no
education.
Destinations:
35% of international migration to India
>60% to Malaysia and the Gulf countries
Internal migration – primarily to Kathmandu
Main reasons for migration: economic (looking for better jobs)
9. Remittances
45% of all households in our sample receive remittances
87% of all households with a current international migrant
receive remittances
The median amount of the remittances sent by all migrants
over the past year was 160,000 Nepali rupees (or around
1,555 USD)
Almost 2/3 of remittance senders indicate how the
remittances should be used
12. Employed
(any)
Farm self-
employed
Farm
contributing
family
workers
Agricultural
(wage)
laborers
Processing
(agricultural
products)
Trading
(agricultural
products)
Nonagricultu
ral workers
Professional
(=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes)
(1) (2) (3) (4) (5) (6) (7) (8)
A. Working-age Women - no controls for remittances (N=1667) , OLS
International migrant in HH
-0.00508 0.167*** -0.177*** 0.00199 -0.0332** 0.00309 -0.00604 0.00298
(0.0174) (0.0241) (0.0274) (0.0118) (0.0168) (0.00382) (0.0124) (0.00952)
A. Working-age Men - no controls for remittances (N=1243) , OLS
International migrant in HH 0.0161 0.120*** -0.0428 0.00579 -0.00163 0.00703 -0.0935*** -0.0531**
(0.0225) (0.0290) (0.0402) (0.0182) (0.0238) (0.00926) (0.0307) (0.0236)
Note: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, * p<0.1
Migration and women’s work in Nepal,
OLS
13. Migration and Women’s Labor Supply
Working-age Women
Employed
(any)
Farm self-
employed
Farm
contributing
family
workers
Agricultural
(wage)
labourers
Processing
(agricultural
products)
Trading
(agricultural
products)
Nonagricultu
ral workers
Professional
(=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes)
(1) (2) (3) (4) (5) (6) (7) (8)
International migrant in HH -0.0479* -0.00138 -0.0648 0.0190 -0.0240 0.00255 -0.0140 0.00556
(0.0247) (0.0387) (0.0410) (0.0169) (0.0244) (0.00228) (0.0218) (0.00909)
Log total remittances in $ 0.00702** 0.0279*** -0.0188*** -0.00286 -0.00160 7.91e-05 0.00144 -0.000433
(0.00327) (0.00515) (0.00558) (0.00242) (0.00336) (0.000323) (0.00289) (0.000817)
Observations 1,667 1,667 1,667 1,667 1,667 1,667 1,667 1,667
14. Migration and Men’s Labor Supply
Working-age Men
Employed
(any)
Farm self-
employed
Farm
contributing
family
workers
Agricultural
(wage)
labourers
Processing
(agricultural
products)
Trading
(agricultural
products)
Nonagricultu
ral workers
Professional
(=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes) (=1 if yes)
(1) (2) (3) (4) (5) (6) (7) (8)
International migrant in HH 0.116** 0.165*** -0.0463 0.0291 -0.0486 -0.0152 -0.0216 -0.0530**
(0.0517) (0.0505) (0.0638) (0.0296) (0.0382) (0.0136) (0.0450) (0.0221)
Log total remittances in $ -0.0169** -0.00746 0.000591 -0.00392 0.00791 0.00375 -0.0121* -2.40e-05
(0.00692) (0.00693) (0.00842) (0.00353) (0.00510) (0.00246) (0.00626) (0.00343)
Observations 1,243 1,243 1,243 1,243 1,243 1,243 1,243 1,243
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
15. Migration, remittances and labour hours (per year) for non-
migrant women – Tobit estimates
Working-age women
Employed (any) Farming
Agricultural (wage)
laborers
Processing
(agricultural
products)
Non-agricultural
workers
Professional
hours/year hours/year hours/year hours/year hours/year hours/year
(1) (2) (3) (4) (5) (6)
International migrant in HH 34.82 -66.99 233.4* -204.6 -496.2 1,748***
(80.10) (75.16) (140.6) (278.2) (551.1) (340.0)
Log total remittances in $, raw 8.944 25.18*** -31.23 -33.86 53.65 -111.2**
(10.55) (9.760) (19.18) (38.88) (74.64) (47.27)
Observations 1,474 1,666 1,667 1,667 1,667 1,667
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
16. Migration, remittances and labour hours per year
for non-migrant men – Tobit estimates
Working-age men
Employed
(any)
Farming
Agricultural (wage)
laborers
Processing
(agricultural
products)
Nonagricultura
l workers
Professional
(1) (2) (3) (4) (5) (6)
International migrant in HH -4.667 417.0*** 227.2 -658.1*** -136.2 -2,708***
(147.4) (122.8) (292.4) (73.57) (371.4) (260.0)
Log total remittances in $, raw -21.93 -31.88* -36.86 124.4*** -95.11* 9.311
(20.17) (16.32) (38.87) (10.20) (52.95) (33.82)
Observations 1,081 1,241 1,243 1,243 1,243 1,243
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
17. Decision-making about agriculture and land
ownership in Nepal
Male land
manager(s)
only
Female land
manager(s)
only
Joint land
manager
Male land
owner(s)
only
Female land
owner(s)
only
Joint land
owner
(1) (2) (3) (4) (5) (6)
International migrant in HH -0.0577*** 0.184*** -0.127*** -0.00981 0.0445 -0.0347
(0.0206) (0.0354) (0.0368) (0.0436) (0.0424) (0.0315)
Observations 876 876 876 691 691 691
Robust standard errors in
parentheses
*** p<0.01, ** p<0.05, * p<0.1
18. Migration and agricultural incomes
(Log) total
harvest value
Log harvest
value GRAINS
Log harvest
value VEGGIE
Total net crop
income
(npr)
Total net
income from
livestock (npr)
(1) (2) (3) (4) (5)
International migrant in HH -0.365*** -0.333*** -0.311 -17,632 -433,608
(0.0960) (0.0948) (0.226) (10,675) (446,549)
Log total remittances in $, raw 0.0299** 0.0295** 0.0250 87.61 -118,950
(0.0126) (0.0144) (0.0262) (1,058) (116,878)
Observations 963 927 765 946 946
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
19. Conclusions
The study finds that male outmigration in Nepal is associated with significant changes in
women’s work and roles on the farm
Women in migrant HH are more likely to be the primary farmer, rather than contributing family
member on the farm
They see improvements in decision-making about agricultural production
The effects of migration on women’s and men’s work are mediated by the receipt of
remittances
Men’s increased employment on the farm is in response to the migration of a family member,
rather than remittance income
Women’s increased employment on the farm seems to be linked to the receipt of remittances
Yet, there is little evidence that male outmigration significantly strengthens women’s
economic empowerment
The majority of employment is on the farm; no increased participation in higher value
nods of agricultural value chains
20. Preliminary Policy Implications
Gender-sensitive agricultural extension services and services
tailored to contexts with the changing agricultural production
modes
Enabling environment and incentives for women and men to
mobilize remittances for productive purposes, including more
investments in agriculture or small businesses
Strengthen women’s access to higher-earning activities in
agricultural value chains and food systems
21. Next steps:
Isolate the causal effects of migration on the labor and empowerment outcomes
of non-migrant women and men in sending communities and on agricultural
production and food security
Instrumental variable approaches
Building a panel
Explore the heterogeneity of impacts depending on the characteristics of migrants
(e.g. destination, length of migration, etc.) and the characteristics of the
individuals and households who stay behind (e.g. age, etc.)
Editor's Notes
Pic by Neil Palmer (CIAT)
Outmigration from rural areas is a main component of the structural transformation process in developing countries.
In response to the absent migrant labor, women may have to increase their labor allocation to the family farm to keep agricultural production at the same level. (Alternatively, migrant households may change or reduce agricultural production.)
Remittances have a separate effect on women’s labor supply – they may raise women’s reservation wages, resulting in reduced time in remunerated employment; or they may relax growth constraints for family farming, making family farming more attractive than other paid or unpaid activities. These hypotheses have been tested in various studies, though with little attention to the types of paid and unpaid work performed by women.
The New Economics of Migration approach (Stark and Levhari, 1982; Lucas and Stark, 1985; Stark and Bloom, 1985; Katz and Stark, 1986)
migration as a collective and not an individual decision >>> the unit of observation is a household, and the principal independent variable is a dummy variable for household with at least one migrant
In Nepal, we only control for whether the household has at least one international migrant; the base category include both non-migrant household and households with domestic migrants only since the latter are very few (only about 55 households in the whole sample). In Senegal, however, we include controls for both international and internal migration since both types of migrations are important in the country.
Furthermore, to understand whether the labor effect of migration or the income effect from the receipt of remittances is more important for women’s outcomes, in some models we also control for whether the household received any remittances in the last year, 𝑅 ℎ , as well as an interaction term between having a migrant in the household and having received remittances (M*R). Not all migrant households receive remittances and also some non-migrant household receive remittances, perhaps from more distant relatives. An indicator variable for remittance receipts is likely less subject to measurement or reporting errors as it is more likely that the respondent remembers whether someone in the household received remittances but may not remember the exact amount received over the whole year. Respondents may also have apprehensions about reporting the correct amount of remittances received.
Vector X includes individual characteristics (age, age squared, marital status, education, ethnic and religious background), household demographic characteristics, household wealth and asset characteristics (quality of the construction materials of the dwelling, quality of sanitary facilities, source of drinking water, access to electricity, household ownership of land and land area owned and cultivated, and livestock ownership expressed in Tropical Livestock Units - TLU) and household non-wage income sources (a dummy for whether the household received any social assistance).
The New Economics of Migration approach (Stark and Levhari, 1982; Lucas and Stark, 1985; Stark and Bloom, 1985; Katz and Stark, 1986)
migration as a collective and not an individual decision >>> the unit of observation is a household, and the principal independent variable is a dummy variable for household with at least one migrant
In Nepal, we only control for whether the household has at least one international migrant; the base category include both non-migrant household and households with domestic migrants only since the latter are very few (only about 55 households in the whole sample). In Senegal, however, we include controls for both international and internal migration since both types of migrations are important in the country.
Furthermore, to understand whether the labor effect of migration or the income effect from the receipt of remittances is more important for women’s outcomes, in some models we also control for whether the household received any remittances in the last year, 𝑅 ℎ , as well as an interaction term between having a migrant in the household and having received remittances (M*R). Not all migrant households receive remittances and also some non-migrant household receive remittances, perhaps from more distant relatives. An indicator variable for remittance receipts is likely less subject to measurement or reporting errors as it is more likely that the respondent remembers whether someone in the household received remittances but may not remember the exact amount received over the whole year. Respondents may also have apprehensions about reporting the correct amount of remittances received.
Vector X includes individual characteristics (age, age squared, marital status, education, ethnic and religious background), household demographic characteristics, household wealth and asset characteristics (quality of the construction materials of the dwelling, quality of sanitary facilities, source of drinking water, access to electricity, household ownership of land and land area owned and cultivated, and livestock ownership expressed in Tropical Livestock Units - TLU) and household non-wage income sources (a dummy for whether the household received any social assistance).
a sample of 1002 households
from 5 districts (Achham, Rolpa, Nawalparasi, Makwanpur and Jhapa), distributed across two ecological zones (the hills and the Terai) and the five former developmental regions
representative at district-level
Some of the remittances in Nepal are almost always used to purchase food. In addition, the remittances are used for clothing, education fees, payment of debts and health. A non-negligible share of households uses the remittances for household farming activities including for the purchase of land.
Food is by far the most often stated use of remittances in Senegal as well. Similar to Nepal clothing, education fees, payment of debts and health comprise an important share of use of remittances. Unlike Nepal, farming activities are rarely listed as a use of remittances.
Figure 4: Use of remittances, Nepal
Multiple choices were possible
Figure 4: Use of remittances, Nepal
Multiple choices were possible
‡ For greater clarity, women in households that receive remittances but do not have an international migrant are excluded from the estimation in Panel B (these women constitute around 3 percent of the finale female sample). In Panel B the base category includes households with no internal or international migrants that do not receive remittances either.
All models also include the following controls: age; age squared; marital status; educational attainment; whether the woman is high caste or low caste; whether she is Muslim; household demographic structure (the number of children under 5, children 5-10 years old, male and female children 11-14 years old, males and females 15-17 years old, number of adult men and adult women in the household); wealth variables (including material of walls, roof, and floor, the type of toilet, access to electricity, access to piped water, whether the drinking water source is on the household grounds, whether the household owns land and area of land owned, livestock ownership measured in Tropical Livestock Units (TLU)); and district dummies.
This is at the household level and takes into account all women and men; reported by the respondent to the HH questionnaires.
Some other impacts: reduction in income from agriculture; but no significant effect on food security