Rural Youth Welfare along the Rural-urban Gradient: An Empirical Analysis across the Developing World

Abstract

We use survey data on 170,000 households from Asia, Latin America and Africa, global geo-spatial data, and an economic geography framework to highlight five findings about rural youth in developing countries. First, the youth share in population is falling rapidly, and youth numbers are stable or falling slowly everywhere, except in Africa. In Africa, youth share is rising very slowly, but numbers are set to double in 40 years. Second, large majorities of rural youth live in spaces that are not inherently limiting: two-thirds live in zones with highest agricultural potential, and one-quarter combine this with highest commercialisation potential. The 4% that do live in inherently challenging spaces are concentrated in pockets of persistent poverty in middle-income countries. Third, rural spaces’ commercial potential has large impacts on welfare outcomes, but their agricultural potential has no detectable impact. Fourth, households with young members face income- and poverty ‘penalties’ in all regions and spaces within them, compared to households without young members. The poverty penalty declines sharply over space as commercial potential rises, but the income penalty shows ambiguous patterns. Fifth, households with young members earn lower relative returns to education, with varying patterns over space.

Acknowledgements

Funding for this research was provided by the International Fund for Agricultural Development (IFAD). The authors are thankful to all the participants of the multiple workshops held throughout the content development phase of the Rural Development Report (RDR) 2019: Creating Opportunities for Rural Youth. Special thanks to Paul Winters for leading and contributing to the analytical content of the RDR, based on which the research ideas for this paper were developed. The authors also thank to Margherita Squarcina, Michael Dolislager and Fabian Löw for excellent research assistance. The data and code for the paper can be made available upon request from the corresponding author.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary materials

Supplementary Materials are available for this article which can be accessed via the online version of this journal available at https://doi.org/10.1080/00220388.2020.1808197

Notes

1. Although different regions and countries may use different age ranges to define youth, we use the UN definition to ensure comparability across all countries covered by our data.

2. We use ‘developing world’ to refer to low and middle income countries, as defined by World Bank. We interchangeably use ‘developing world’ and ‘low and middle income countries’ to refer to the same set of countries.

3. Personal aptitudes and orientation of a young person also matter, but are largely unobservable in existing data sets. Conceptual and practical challenges also make it problematical to estimate young people’s welfare separately from that of the households in which they reside. We therefore focus on questions that can be addressed with existing data, and call in the final section for increased research to address these more challenging, individual-level aspects.

4. The thresholds to define high and low categories are based on the average value of the ST indicator and the median value of the RT indicator. The choice is driven by the fact that RT indicator is a continuous variable, hence has a wide distribution, and the ST indicator is a percentage, hence bounded from both ends.

5. The opportunities available to rural youth in their localities and how they compare to other localities potentially influence migration decisions. To the extent that migration is driven by push factors the ROS can capture part of these incentives. Capturing pull factors by accounting for the ROS in potential migration destinations is impossible with our data and beyond the scope of this paper. Future research can address this question, though relying on a more limited set of countries with needed migration information.

6. http://www.worldpop.org.uk/.

7. We expect population density to correlate also with infrastructure and public service provision in general (for example education, health), which are other components of rural opportunity. We continue using ‘commercialisation potential’ for the dimension of rural opportunity, however, to keep parallels with the Wiggins and Proctor (2001) framework.

8. The production of the WorldPop datasets follows Tatem et al. (2007), Gaughan et al. (2013), Alegana et al. (2015) and Stevens et al. (2015).

9. As defined by the World Bank in 2018. Note that we exclude the small island states and resource rich countries, which tend to be outliers on the scales of structural and rural transformation indicators we use for the national setting.

10. Table S.1 in the supplementary material shows the population density threshold to define each quartile and the average population density within each quartile.

11. Full Time Equivalents (FTE) were calculated using individual level data on time allocation across different activities in our 12 country data set. See Dolislager et al. (2020) for more details.

12. EVI is an improvement over the so far more widely used NDVI, which ‘utilises only the red and infrared bands and is subject to noise caused by underlying soil reflectance, especially in low-density vegetation canopies, and also to noise from atmospheric absorption. EVI utilises the blue band for correcting for atmospheric aerosols’ (Jaafar & Ahmad, 2015).

13. A detailed list of all data sets and levels and sources of geolocations are presented in the supplementary material, table S.2.

14. Geo-referencing levels vary from survey to survey: enumeration areas in SSA, and municipalities/villages in LAC/APR. See methodological appendix B in the supplementary material for a detailed explanation of the merging procedure.

15. Note that, in all cases, we proxy income with per capita expenditure. All references to ‘income’ should be interpreted in this way.

16. Appendix Table A1. includes detailed descriptions of all variables used in the analysis. Table S.3 in the supplementary material presents the summary statistics separately for each country.

17. Although all differences in means (except the share of sales in farm income) are statistically significant given the very large sample size, the values of differences in these variables are too small to be meaningful in practice.

18. Abay et al. (2020) use a similar definition of mostly young households to discuss how household structure affects landscapes of rural youth opportunities.

19. Sub-regional analysis shows that the Near East and North Africa sub-region exhibits similar patterns to SSA (IFAD, 2019, Figure 9.1).

20. Total number of rural youth mapped in this figure is 778 million, which is higher than the official UN figure of 494 million rural youth due to the fact that most youth administratively categorised as urban live in non-urban spaces as defined by the rural-urban gradient.

21. Authors’ calculations from WorldPop data as explained in the methods section.

22. (0.67/0.08)/(0.33/0.92) = 23.3.

23. Table S4 in Supplementary Material presents youth population and distribution across the ROS for each country in our national level analysis. Grouping of these countries by transformation category shows the prevalence figures quoted here. See also IFAD (2019) Figure F, page 28 and Figure 2.5, page 81.

24. Full regression results of are provided in the Appendix Tables A2.

25. See supplementary table S3; Note that, because the youth poverty penalty is defined relative to non-youth, its decline over the rural-urban gradient cannot be explained by a simple correlation between average incomes of countries and their level of urbanisation, nor by a general pattern of better-off households moving to urban areas.

26. We also expect these returns to be higher in more transformed economies, for the same reason. Our results are presented by region, however, which roughly corresponds to levels of transformation with LAC being most transformed, followed by Asia, and Africa the least transformed.

27. Note that the relative sizes of these percentages across regions should not be taken to indicate differences in the level of monetary returns, since for example incomes in LAC are far higher than in Africa and thus equal percentage rises would be much higher in monetary terms in LAC.

28. Note that the average share with secondary education in Asia is also very low in our sample. This is because the data from Bangladesh and Cambodia are the oldest data points in our sample (both from 2010) and that both countries have made great strides in secondary education completion rates after the survey year (World Bank 2019. World Development Indicators, Lower secondary completion rate, total (% of relevant age group), accessed in October 2019).