Urban Agglomerations and Wage and Self-employment Jobs in Ethiopia

Abstract

Agglomeration effects on the workers’ and firms’ productivity in developed countries are widely confirmed by previous studies, while empirical evidence for developing countries is still limited. This paper sheds light on another urban productivity factor that is particularly important for developing countries: self-employment and wage jobs. The transition from the economy dominated by the former to the one by the latter is a key to economic development. This study examines how urban agglomerations are linked to the mix of self-employment and wage jobs by focusing on one of the least urbanized low-income countries: Ethiopia. Applying an instrumental variable approach to worker-level cross-sectional data, the analysis finds a negative link between town population size and the share of self-employment workers: a log increase in town population size lowers the probability of engaging in self-employment by 7.9 percentage points. Particularly female, young, and less-educated workers are more likely to work with wage jobs in larger towns. The findings suggest a crucial link between urban agglomerations and employment modes in the developing world.

Keywords:

• Urbanization
• labor market
• urban agglomerations
• East Africa
• Ethiopia

Acknowledgments

We would like to thank David Newhouse, Tom Bundervoet, and anonymous referees for their useful comments. The findings, interpretations, and conclusions expressed in this paper 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. The code developed by the authors can be shared upon request. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Disclosure statement

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

Notes

1 Consumption and wage/self-employment are also inter-temporally associated at the city level. The UEUS and HCES data provide statistics representative of selected major cities and towns (Mekele, Asayta, Gonder, Dessie, Bahir Dar, Debre Zeite, Adama, Jimma, Jigjga, Assosa, Gambela, Harari, Addis Ababa, and Dire Dawa). Major cities and towns with an increasing share of wage employment have experienced stronger consumption growth during the period. The positive association between consumption and wage employment shares (and the negative association between consumption and self-employment shares) remain even after econometrically controlling for the changes in employment shares in economic sectors, as well as town fixed effects. Results available upon request.

2 The UEUS was conducted only in urban areas based on a stratified two- or three-stage cluster sample design. The CSA defines urban areas as localities with 2,000 or more inhabitants or all administrative capitals of regions, zones, and woredas. In addition, urban areas include localities with at least 1,000 people who are primarily engaged in nonagricultural activities and/or areas where the administrative official declares the locality to be urban. The urban area was stratified into 16 major urban centers and 8 other urban centers. For the former, the census enumeration areas (EAs) were randomly picked up with their population sizes considered. For the latter, urban centers and EAs were considered as the primary sampling units (PSUs) and the secondary sampling unit (SSU). The two data sets are both created by the Ethiopian CSA with a similar sampling framework. The labor module of the two surveys is similar as well. For the main outcome of the employment type, they both ask the same question of ‘What was the status in employment on your main job?’, and they use the same options for response. The CSA has launched the LFS program to be conducted every five years, while the UEUS is conducted every year. The 2013 survey is the third in the series after the 1999 and 2005 LFS. The survey covered both urban and rural areas of all the regions in particular and the country in general. The recent survey collected data from selected major towns with population size 100,000 and above, including region capital cities (International Household Survey Network 2013).

3 Ethiopia is administratively divided into regions and further subdivided into zones and woredas (districts). In this paper, we focus on towns or cities, which are often woreda capitals. We also consider Addis Ababa, Dire Dawa, and Harari as independent cities for our analysis.

4 This formula is used to calculate the town population each year i (=2012, 2013): (4) $$po{p}_i=\frac{2014-i}{2014-2007}\times po{p}_{2007}+(1-\frac{2014-i}{2014-2007})\times po{p}_{2014}$$(4)

5 In the UEUS/LFS, an individual is classified as unemployed if she or he is not engaged in productive activities and is available for work in the next month if the opportunity arises.

6 Our analysis indicates that the city size is negatively associated with employment in the public sector, meaning the share of public sector employment is large for smaller cities. Nevertheless, estimating the OLS and IV models on both private and public sector workers results in the same conclusion as presented (results available upon request).

7 Fixed effects for the data source and year of data collection, UEUS (2012), and LFS (2013) are included as well.

8 To provide a comprehensive range of analysis, the supplemental document provides the discussion and estimation results of other labor outcomes (formality of employment; working hours; and underemployment, unemployment, and nominal wage).

9 Some literature (for example, McCullough (2017)) highlights that wage differentials between farming and off-farm employment disappear once hours worked are controlled for.