Youth Migration and Labour Constraints in African Agrarian Households

Abstract

Using panel data from Ethiopia and Malawi, we investigate how youth migration affects household labour, hired labour demand, and income, and whether these effects vary by migrant sex and destination. Labour shortages arise from the migration of a head’s child. However, the migration of the head’s sons produces a greater burden, particularly on female heads/spouses (in Ethiopia) and brothers (in Malawi). Gains from migration in the form of increased total net income justify the increased labour efforts in Ethiopia. Weaker evidence suggests households in Malawi substitute hired for migrant family labour at the expense of total household net income.

Acknowledgements

We thank Xiaoya Dou and Mekamu Jedir Jamal for excellent research assistance. Our manuscript has benefitted from helpful discussions with Paul Christian and Emily Schmidt. We thank Fantu Bachewe and Bart Minten for sharing the Ethiopian wage data presented in this paper. Finally, this paper has benefitted from the support provided by the CGIAR Research Program on Policies, Institutions, and Markets (PIM) led by the International Food Policy Research Institute (IFPRI). Data and dofiles for the analysis are available upon request from the corresponding author.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary Materials

Supplementary Materials are available for this article which can be accessed via the online version of this journal at https://doi.org/10.1080/00220388.2018.1430770

Notes

1. The focus in this paper is the role of youth internal migration, but there are strong differences in the international migration patterns across contexts as well. For example, Ethiopians travel to the Middle East (de Brauw et al., 2017) and Malawians tend to gravitate toward South Africa and Zimbabwe (Lewin, Fisher, & Weber,  2012).

2. It should be noted that the evolution of migration patterns may ultimately have more dynamic impacts on structural transformation in developing countries which are left unexplored here due to data constraints. First, if educated sons move because of land scarcity or concerns over underemployment and if children who are left behind compensate for the shortfall in labour, a lack of human capital in rural areas may stymie advancements in agriculture. Second, remittance income may change who controls income within the household. While remittances have the potential to increase investment in agriculture, depending on the profile of the decision maker, they may also be spent on alternative physical or human capital investments. If women favour human capital investments (Mueller, Kovarik, Sproule, & Quisumbing, 2015), we might observe a disinvestment in agriculture rather than farm expansions and increased investments in modern technologies. Data covering a longer timeframe would be more fruitful to explore these long-term impacts of migration on human and physical capital investments and future farming systems.

3. Household attrition of the overall (rural and urban) sample is rather low (5% in Ethiopia, and 4% in Malawi) over the two year period (CSA and World Bank, 2015; NSO, 2014).

4. Table 1 shows that there are noticeable positive changes on average in the inherited land variable in both countries, driven by the death of a family member.

5. This information was collected using different protocols in each country. In Ethiopia, the information is constructed based on self-reported information by the proxy respondent (often the head of household). Despite the careful documentation of each household member in the baseline household roster, revealing or recalling the nature of the member’s absence is at the discretion of the household head, making the household member’s mobility status subject to measurement error. In Malawi, the migration definition is verified by the migrant at their destination at endline. While the tracking protocol offers precision in Malawi, a minority of migrants were unsuccessfully tracked (6%) and therefore are omitted from the analysis (National Statistical Office (NSO), 2014). These limitations in the measurement of migration are not unique to this study.

6. Migration distance is only available in Malawi, where migrants were tracked and georeferenced. Lee and Mueller (2016) find young (ages 15 to 24) rural-rural migrants tend to travel over one kilometre relative to rural-urban migrants who travel approximately 60 kilometres.

7. We may underreport the number of family members employed in the household’s nonfarm enterprise in both countries. The Ethiopia survey documents at most five people hired in the enterprise, while the Malawi survey asks for the identification of at most two household members who manage and two household members who own the enterprise. A total of four or less household members may be included in the Malawi data.

8. Values are winsorized at 1 per cent to reduce the influence of outliers.

9. The discrepancy in time frames is due to differences in the survey instrument. In Ethiopia, the labour questions are asked over the current season, while in Malawi, the labour questions are asked over the two seasons.

10. The migration of family members is dominated by the children of the heads in both Ethiopia and Malawi. Twenty-five (27) per cent of sons and 38 (37) per cent of daughters in Ethiopia (Malawi) moved by the follow up round. While the mobility of heads and their spouses in Malawi exceeds that of Ethiopia (13% compared to 2%), the younger generation is much more mobile.

11. Our final list of towns/cities is 25 in Ethiopia and four in Malawi.

12. Wage growth is converted into 2011 real terms in both Ethiopia and Malawi using consumer price indices (CSA, 2013b; NSO, 2016).

13. Busso, di Nardo, and McCary (2014) recommend the use of the nearest neighbour matching approach with bias-correction in lieu of other reweighting or matching estimators particularly when the overlap is poor. We estimate the distribution of the propensity scores based on our covariates of interest to find overlap may be imperfect (Figures A.1–A.2 in Supplementary Materials).

14. Tables A4 and A5 (see Supplementary Materials) model these migration decisions using probit regressions. The covariates are sufficiently strong determinants of the migration of sons in Malawi. Inherited land becomes a significant explanatory variable when evaluating the probability of someone in the pooled sample having a child of the household move to an urban area. This finding supports conventional wisdom regarding migration driven by land scarcity in Ethiopia (De Brauw & Mueller, 2012).

15. Few observations in our sample live in households that either have a head or spouse that migrates or an additional member that moves internationally. To assuage concerns that our estimates are driven by the inclusion of these few occurrences, we present the matching results for the employed on the farm outcomes dropping observations in households with head/spouse migrants and international migrants in Table A6 (Supplementary Materials).

16. Bias-corrected matching estimates are the same irrespective of the number of matches used (not shown here). Trimmed sample matching estimates are qualitatively similar as to those using the entire sample (Tables A7 and A8 in Supplementary Materials).

17. Interestingly, there is weaker evidence to suggest a greater percentage of daughters are tasked with working in the nonfarm enterprise when a head’s son migrates in Ethiopia. The order of magnitude far exceeds the percentage point increase in female adults called to work on the farm. Because our F tests do not support that these effects statistically differ from those observed among other family members, the figures are merely suggestive that daughters may also substitute for sons who spend more time on the farm during the harvesting period.

18. Agricultural and total profit are reported in 2012 US dollars and winsorized at 5 per cent. Total income includes agricultural revenue net input expenditures, non-farm enterprise revenue net input costs, wage income, and transfers.