How should jobs policies in developing countries respond to the emerging pattern of structural change?"Kuznets' paradigm" for jobs and structural change no longer holds. Kuznets expected labor to move gradually from informal agriculture into better jobs with higher productivity and wages in well-organized, formal manufacturing and (eventually) services.

But in Sub-Saharan Africa, we see labor moving out of agriculture into informal jobs in both services and manufacturing. There are few formal jobs. Even in manufacturing, informality predominates. We can observe this clearly in Ethiopia and Tanzania. It contrasts with what happened a generation ago, in places such as Taiwan and Vietnam, where formal manufacturing jobs were the key to structural transformation. In Ethiopia, for example, manufacturing jobs are growing - they rose from 3% to 10% of jobs in the last ten years. But formal manufacturing jobs are stagnant. Manufacturing is experiencing exacerbated dualism. There's a cluster of large, formal firms with high productivity and few jobs - and thousands of small, informal, low productivity firms, with lots of jobs.

Conventional hypotheses would explain this by market failures that affect small firms - such as high labor costs and capital market failures. But there's little evidence of a "missing middle". And at the top end of the distribution of firms, the capital intensity of formal manufacturing in Tanzania and Ethiopia is higher than in the Czech Republic. This is inconsistent with their factor endowments, where low-skilled labor is abundant and capital is scarce and expensive.

The explanation may be that frontier technologies tend to be shared across economies. So, Ethiopia and Tanzania's formal manufacturing firms are importing technologies that predominate in global markets and reflect global factor endowments. As a result, the global collapse in demand for low skilled jobs in manufacturing is being replicated in Sub Saharan Africa. Figure 1 shows you what might be happening.

The starting point is that there are two sets of technologies - the capital-intensive technology on the left, and a more labor-intensive technology towards the right. At first, low-income countries use the latter. With the market price at Po, the labor-intensive technology is more profitable. But then, the system is shocked by a downshift in the cost function for capital intensive technologies (e.g. due to automation). The result is a new (lower) global market price (Pi), which lies below the entire (unshifted) cost function for the labor-intensive factories.

Figure 1: Analytics of technology choice in a world with skill/capital intensive technical change. Source: Diao, Ellis, McMillan, Rodrik (2020)

So, Low-Income Countries (LICs) no longer have a comparative advantage in labor intensive manufacturing. Their output falls (due to the loss of market share) and jobs are cut even more (because the lost output was relatively labor intensive). Note, also, that the new cost curve for the capital-intensive segment is steeper than the old one, reflecting the increased marginal importance of skills and other expensive complements, such as public infrastructure, to output growth. So, it's harder for skill-constrained, infrastructure-poor LICs to transition to this model.

What are the alternatives to manufacturing-led jobs growth? Can other sectors absorb large amounts of low skilled labor? Can they exploit global markets, so their growth isn't hampered by domestic demand constraints ? Unfortunately, it's hard to see how agriculture and tradeable services tick those boxes. Jobs in agriculture will have to shrink, as productivity rises faster than demand. High productivity, tradable services have few jobs and don't match the skill endowments of LIC economies. Even success stories like India and the Philippines have hit skill constraints.

I conclude that the key question facing economic development strategy is: "Where will the good jobs come from?". Good jobs need to be productive - but not necessarily the MOST productive - jobs.They need to be consistent with skill endowments. And they cannot be created by informal, low productivity firms. It seems inevitable that domestic services will become a crucial part of the answer.

To support the growth of good jobs, growth policy, jobs policy and social policy will need integrating, with complementary interventions for labor supply and demand.Education and skills will remain crucial. But we also need a new industrial policy. The matrix in figure 2 illustrates possible points of entry. Until now, we've focused industrial policies on high productivity sectors, supporting innovation and trade agreements for global market access. Meanwhile, social policies have focused on the low productivity population, including education and training programs, and redistributive transfers.

Figure 2: Reconsidering development policy. Source: Rodrik

Instead, we should focus on the middle productivity activities that can produce lots of better jobs.As we do so, the potential for creating good jobs in domestic services will loom large. Supporting this process means addressing complex problems. Uncertainties preclude the likely effectiveness of simple Pigouvian employment subsidies. We should consider, instead, customized business incentives, explicitly designed to create jobs, e.g. by using soft conditionalities, and not necessarily focused on export champions.

Across the board, governments should integrate the large externalities that are linked to better jobs into their funding decisions for training and the incentives for investment and the choice of technology.This is a "structuralist" or "productivist" approach, which merges the equality/inclusion agenda with the jobs and economic growth agenda.

This blog is based on a Keynote Speech at the 5th Jobs and Development Conference on September 2nd, 2021.

Topics
Jobs & Development
Social Protection

Attachments

  • Original document
  • Permalink

Disclaimer

World Bank Group published this content on 05 October 2021 and is solely responsible for the information contained therein. Distributed by Public, unedited and unaltered, on 05 October 2021 16:18:07 UTC.