How I Became a Manufacturing Skeptic

May 16 , 2026
By Dani Rodrik

---

At a gathering of academics and policymakers at Harvard this month, a participant reminded me that I had published a column 15 years ago on "The Manufacturing Imperative". The piece emphasised the importance of industrialisation in driving economic growth, creating good jobs, and building a middle class.

"This is one of my all-time favourite articles," the policymaker from Africa told the audience.

There is scarcely a greater reward for a scholar than having his ideas resonate strongly with the people for whom he writes. But in this case, a gentle rebuke came along with the praise. What I had written in that column and many other places at the time seemed to conflict sharply with the arguments I was making at this conference about the limits of manufacturing.

The contradiction was real. In recent years, I have become sceptical about the viability of the traditional industrialisation-led growth model. I have argued for a different model of economic growth, emphasising the development of productive capabilities in labour-absorbing, mostly non-tradable services. I have warned policymakers in Africa and other developing regions that trying to emulate the East Asian model would produce, at best, manufacturing enclaves, with a tiny sliver of productive firms integrated into global value chains while the bulk of the labour force remains stuck in low-productivity activities.

Mexico exemplifies this outcome. As Santiago Levy, a former Mexican deputy minister of Finance, pointed out at the same conference, Mexico's exports of manufactured goods have increased more than tenfold since the country joined the United States and Canada to form the North American Free Trade Agreement (NAFTA) in 1994. At the doorstep of a giant market and with policymakers determined to promote foreign trade and inward investment, few countries were blessed with better circumstances for export-oriented industrialisation.

Yet Mexico's overall economic performance has been dismal, even by undemanding Latin American standards, with a declining productivity trajectory.

What made manufacturing the powerful economic escalator it once was was that it could employ large numbers of low-skilled workers while placing limited demands on the governance and infrastructure of low-income countries. Today's manufacturing is different. Competing successfully on world markets and with China at home requires skills, technologies, and other capabilities that are in short supply in poor countries, precisely because they are poor. Manufacturing no longer offers a shortcut that sidesteps these fundamental constraints.

The result is that even when countries manage to pull more workers into manufacturing, this occurs through the expansion of small-scale, mostly informal enterprises and at the expense of productivity. This is the story of industrialisation in Ethiopia, which once represented the hope that the East Asian model could be transplanted in Africa. Expansion of manufacturing employment and an increase in manufacturing productivity used to go hand in hand in early industrialisers such as Japan, South Korea, and Taiwan. They now move in opposite directions in Ethiopia, Bangladesh, India, and even Vietnam.

I became a manufacturing sceptic reluctantly. The evidence was hard to ignore as manufacturing technologies became more sophisticated and the failure of countries outside East Asia to industrialise successfully became increasingly apparent. I began to consider alternative growth strategies not because I came to think of broad-based industrialisation as less desirable, but because I became convinced it was less feasible. As John Maynard Keynes reputedly said, "When the facts change, I change my mind; what do you do, sir?"

Here is a sobering calculation. Of the two billion workers in the developing world today, I estimate that roughly three-quarters (1.5 billion) are in occupations that neither require university education nor are exposed to the international economy through trade or offshoring. These are subsistence farmers, street vendors, retail and food service workers, casual workers, and others in non-traded occupations. Their numbers will only increase in the years ahead, even if their share of the total declines somewhat.

The critical question facing policymakers is how to enhance these workers' economic opportunities. The numbers make it painfully clear that neither industrialisation nor education can be the answer, as desirable as these may be. Finding ways to increase worker productivity in labour-absorbing services will be crucial. Otherwise, gains in living standards cannot be sustained.

Non-traded services have traditionally been a drag on economic growth. Many policymakers are accordingly pessimistic about their potential. But this may be changing. Something akin to a revolution in service productivity has been underway, most visibly in advanced economies, through organisational innovations, digital platforms, and other new technologies.

For developing economies as a whole, the last three decades have been a rare period of rapid economic growth and convergence with advanced economies. Remarkably, it is services, not manufacturing, that are responsible for this outcome.

As the economists Tianyu Fan, Michael Peters, and Fabrizio Zilibotti show in detailed empirical work, India's remarkable economic growth has been driven by productivity gains in consumer services such as retail and hospitality produced for local markets, not in skill-intensive, exportable services such as ICT and BPO for which the country is well known. These authors have documented a similar mechanism at work in sub-Saharan Africa's rapidly growing economies.

The evidence suggests a virtuous cycle of economic growth built on middle-class services. Expansion of the middle class shifts consumer demand toward higher-quality and more productive services, which in turn enables the rise in workers' incomes that underpins the middle class. But the process is not automatic. It requires an important role for the government in facilitating the requisite productivity enhancements.

As Rohan Sandhu of Harvard Kennedy School and I have argued, many successful experiments around the world already provide proof of concept. They include initiatives that encourage platform companies to employ local inputs and workers, assist micro enterprises with training and certification, and provide customised AI and other technological tools adapted to developing-country circumstances.

Dedicated efforts can build a more reliable, inclusive growth model. Without them, the vast majority of workers in the developing world will be left in precarity, isolated from the high-productivity enclaves linked to the global economy.

---

PUBLISHED ON May 16,2026 [ VOL 27 , NO 1359]

---

[ssba-buttons]