What is the history and contribution of decentralisation to Ethiopia’s development? Jean-Paul Faguet draws on recent research to show that decentralisation helps explain the country’s extraordinary economic performance since the country’s civil war.

Ethiopia has been in the news for terrible reasons lately. The horrific violence now consuming the northwest of the country (see ‘They have destroyed Tigray, literally‘) is tied up in a regional conflict with roots in the country’s deep ethnic divides. These are the same cleavages that shaped the country’s ethnic federalism – that is, the very structure of the Ethiopian state (explained succinctly in Appendix C of this book). Paradoxically, this federal design was until recently judged highly successful in stabilising this enormously diverse country and facilitating its rapid development. This makes the current, vicious spasm all the more tragic.

‘What went wrong?’ is a long, complex tale of imperial conquest, in the past, and of powerful and headstrong leaders today, and beyond the scope of this post. What I do here, instead, is to rewind before the current conflict to examine decentralisation and development during the two decades following the end of Ethiopia’s civil war in 1991. This story is based on a paper I wrote with Qaiser Khan and Priyanka Kanth, published recently in Publius: The Journal of Federalism.

Beginning in the mid-1990s, Ethiopia adopted a federal structure, featuring nine regional states divided along ethno-linguistic lines. In 2002, it decentralised government further, devolving authority and resources for key social services down to woreda (district) level. During this period, Ethiopia enjoyed a golden age as one of the world’s most striking success stories, with economic growth and human development progress amongst the best in the world. Hence our question here: does decentralisation help explain Ethiopia’s extraordinary development performance over the past generation?

Ethiopia’s development miracle

With a population of 115 million divided amongst 93 mother tongues and 98 ethnicities, Ethiopia is a big, important, highly diverse country. It is also a remarkable case of development, suffering war, famine and chaos in the 1970s and 1980s before transforming itself into one of the fastest growing economies in the world. For 10 of the past 15 years, Ethiopia sustained growth rates above 10%, with no year lower than 6.8%. Throughout this period the economy diversified rapidly. Ethiopia is also important for the policy experiment it represents: a low-income country with terrible human development indicators that both federalised and decentralised, and then saw dramatic improvements in human development.

These human development advances have been both broad and sustained. Child mortality fell from 123 per thousand in 2005 to 67 per thousand in 2016. Ethiopia achieved the Millennium Development Goal 4 (Child Mortality) target ahead of time, and primary net enrolment rates rose from 68% in 2004 to 85% in 2015. These development milestones are coupled with a drastic reduction in poverty. Based on official data, the population below the poverty line fell from 39% in 2004 to 27% in 2015.

Did decentralisation contribute to these successes, or was it incidental? Our evidence implies that decentralisation did indeed improve public education and health sector performance in Ethiopia. Specifically, decentralisation to regional and woreda (district) governments led to higher net enrolment rates (NER) and improved provision of antenatal care (ANC) to women. Our methodology allows us to distinguish between the effects of greater expenditures vs. decentralisation per se on NER and ANC. The expenditure effect is positive, as one would expect. But there’s a separate, statistically significant and substantively important effect of decentralisation on these outcomes that dominates the pure expenditure effect.

The magnitudes are significant: the incremental effect of decentralising health and education service provision to the average woreda is an estimated 13% for ANC and 18% for NER. The main channel for these improvements is institutional, related to local control over education and health services, as opposed to local expenditures per se. Such effects might be supply-side, such as greater efficiency in public management, or better-informed decisions; demand-side, such as higher citizen demand for education and health services; or both, such as improved accountability of officials to citizens. With currently available data, it’s unfortunately not yet possible to disentangle these effects. Our results are robust to different data types, estimation techniques (time series, panel, diff-in-diff), and model specifications.

Future research could probe further into the institutional mechanisms by which decentralisation improves service provision, and whether these vary by sector or region. But what cannot be doubted is that Ethiopia made remarkable progress in health and education during the past generation, and decentralisation is an important part of that success. That this could happen in an environment of relatively low state capacity implies that decentralisation is a powerful reform and hints at its potential to improve governance more broadly.

Lastly, it’s difficult to overstate the empirical difficulties we encountered. Creating the database required years of work and a huge amount of improvisation on the part of a well-qualified research team. We’re happy to make the full database of Ethiopian subnational expenditures, social development outcomes and other characteristics freely available. We hope future researchers will use it to investigate interesting, difficult questions on subnational development in Ethiopia.

This post was first published on the LSE International Development blog.
Photo: John Iglar. Licensed under the Creative Commons Attribution-ShareAlike 2.0 Generic | Flickr