This case study uses process tracing to estimate the contribution of LIWAY to a change in the structure of the labour market involving both domestic and overseas employment for low-income groups.

Evidence shows that LIWAY invested heavily in understanding the market system to obtain a unique perspective and vision of a more effective, sustainable labour system, which was valuable to a wide range of actors and potentially impactful in generating new jobs. The analysis led to a pilot which, while unsuccessful in itself, built the technical foundations and credibility for larger scale and more impactful iterations. Ultimately, the intervention resulted in the creation of a national database for the collection of data on both jobs and jobseekers with embedded functions for job matching domestically and overseas, verification and screening, training, and biometric identification among a range of other functions.

The platform and the system have ongoing challenges as part of a very complex settlement between different public, private, and civil society actors. Politics has played strong a role in both its development and its ongoing implementation, and the full potential impact is still far from being reached. Nevertheless, impact so far has been strong – assessed according to a range of indicators around registration and job matching as well as many other positive developmental indicators beyond LIWAY’s scope around worker protections and tracking. LIWAY’s contribution to this change is clear and validated through a wide range of data sources whose positions provide incentives to both validate and refute the contribution claims. It is not necessary to assess quantitative contribution as this stage as the number of people registered and jobs matched is increasing markedly on a monthly basis and the platform is being operated sustainably.

Three sets of reflections are provided in the case study as conclusions. Firstly, the reasons why it worked. These reasons include:

• Solid analysis, which was used as the common thread of the programme’s value addition.
• Proactive piloting which, while it wasn’t likely to be the eventual model, gave LIWAY a voice in the sector and some data to use.
• Generating buy-in and credibility among key actors by analysing change agents with the power to affect outcomes, including how these change agents shifted over time.
• Good fortune; there were events beyond the programme’s control, which could easily have gone the other way and prevented impact.
• Deploying the right human and financial resources at the right time, which was iterated as the intervention went on.

A second set of reflections is provided on why the intervention didn’t work more effectively (yet). Reasons here include:

• Bad luck; just as serendipity favoured LIWAY at certain points, COVID-19 and other political changes have significantly impeded progress.
• LIWAY aims for major changes in sticky and fragmented issues. With these potentially large and impactful changes come far greater risks of failure. The coordination of so many different actors and factors needed to make the intervention work has impeded the pace of change.
• Some problems are also so large as to be beyond the scope of LIWAY’s influence, including the major political changes that have taken place in Ethiopia over the intervention’s lifespan. For these changes which impede impact, the only appropriate strategy is to wait and ensure continued buy-in while they play out.

A final set of reflections is provided for interventions in exchange infrastructure or job matching in general, which has become an increasing focus in development due to the potential scale of impact:

• Some models require building out the access frontier from what exists, while others require inventing something entirely new designed for completely different socioeconomic and demographic characteristics. In LIWAY’s case, it was partly both. There were aspects of functional job-matching services aimed at higher-income segments (particularly for overseas work) and demonstrating the use case their generate surplus revenues and potential viability lower income market segments. At the same time, many aspects of that system were completely unviable for low-income, low-margin groups. These groups required different supply side behaviours (they weren’t proactively seeking work on their phones) and different business models (through bundled services) to reach economic viability. Many other actors with other types of incentives were required to make this use case viable.
• Diversification was key to economic viability. For low-income segments, there are many actor groups with many different reasons to want to access a labour market platform. This is perhaps more the case in these lower margin segments than for higher wage jobs; volumes are higher, access to other services is lower, and there are many more social and political actors who might have a reason to use this data. These different use cases were key to pushing the platform forward.

Recognising these different types of incentives was vital. While the bundling of services above was important, the primary driver of behaviour for private actors – be they technology providers, training companies, or banking services – is profit. However, when dealing in low-income segments, the value placed on the data by social actors, including those involved in economic development but also health, democracy, and migration, is also very high. This can be a more comprehensive and cost-effective solution for these actors too. These mixed incentives for viable business models are crucial in making the system work.