CAMBRIDGE – Economists have long argued that productivity is the foundation of prosperity. The only way a country can increase its standard of living sustainably is to produce more goods and services with fewer resources. Since the Industrial Revolution, this has been achieved through innovation, which is why productivity has become synonymous, in the public imagination, with technological progress and research and development.
Our intuition about how innovation promotes productivity is shaped by everyday experience in business. Firms that adopt new technologies tend to become more productive, allowing them to outcompete technological laggards. But a productive society is not the same as a productive firm. Something that promotes productivity in a business may not work, or may even backfire, at the level of a whole country or economy. Whereas firms have the luxury of focusing on the productivity of only those resources they choose to employ, a society needs to enhance the productivity of all of its people.
But many economists (and others) have failed to appreciate this distinction, owing to the assumption that technological progress will eventually trickle down to everyone, even if its immediate benefits accrue only to a small group of firms and investors. As economists Daron Acemoglu and Simon Johnson remind us in their useful new book, this belief has not quite been true historically. The Industrial Revolution may have inaugurated the period of modern economic growth, but it did not produce advances in well-being for most ordinary workers for the better part of a century.
Worse, the conventional narrative may have become even less true with the most recent wave of technological advances. New technologies may fail to lift all boats because their benefits can be overwhelmingly captured by a small group of players – be it a few firms or narrow segments of the workforce. One culprit is inappropriate institutions and regulations, which skew bargaining power in the economy or restrict entry by outsiders to modern sectors. Another is the nature of technology itself: innovation often empowers only specific groups, such as highly skilled workers and professionals.
Consider one of the paradoxes of the hyper-globalization era. After the 1990s, as trade costs fell and manufacturing production spread around the world, many firms in low- and middle-income countries became integrated into global supply chains and adopted state-of-the-art production techniques. As a result, these firms’ productivity increased by leaps and bounds. Yet the productivity of the economies in which they were domiciled stagnated in many cases, or even regressed.
Mexico provides a startling case study, since it was once a poster child for hyper-globalization. Thanks to the government’s liberalizing reforms in the 1980s and the North American Free Trade Agreement (NAFTA) in the 1990s, Mexico experienced a boom in manufactured exports and inward foreign direct investment. Yet the result was a spectacular failure where it really mattered. Along with many others in Latin America, Mexico experienced negative total factor productivity growth in subsequent decades.
As a recent analysis by the economists Oscar Fentanes and Santiago Levy demonstrates, Mexican manufacturing did indeed become more productive as it was forced to compete globally. While less productive firms that failed to adapt eventually shut down, many remaining firms adopted new technologies and became more productive.
The problem was twofold. First, manufacturing firms – especially formal ones – shrank in terms of employment, absorbing an ever-smaller share of the economy’s labor force. Then, the rest of the economy, which was dominated by small, informal firms, became less and less productive. The upshot was that productivity gains in the (shrinking) globally oriented manufacturing sector were more than offset by the poor performance in other activities, mostly informal services.
Fentanes and Levy attribute these consequences to Mexican labor and social-insurance regulations, which they claim encouraged informality and hampered the growth of formal-sector firms. Yet one can find the same pattern of productivity polarization in many other Latin American economies, as well as in Sub-Saharan countries.
An alternative explanation concerns the changing nature of manufacturing technology itself. So great are the skill and capital requirements of integrating into global value chains that countries poorly endowed with these resources face sharply rising cost curves, preventing their firms from expanding and absorbing much labor. Workers flocking to the cities from the countryside have little choice but to crowd into low-productivity petty services.
Whatever the underlying cause, this issue exemplifies why government strategies to boost productivity can miss the mark. Whether it comes in the form of plugging into global value chains, subsidizing R&D, or investment tax credits, conventional policies often target the wrong problem. In many cases, the binding constraint is not a lack of innovation in the most advanced firms, but rather the large productive gaps between them and the rest of the economy. Raising the bottom – by providing training, public inputs, and business services to smaller, service-oriented firms – can be more effective than lifting the top.
There are lessons here for the new age of artificial intelligence. Large language models’ potential to perform a wide range of tasks at greater speed has generated much excitement about significant future productivity growth. But, once again, the overall impact of this technology will depend on the extent to which its benefits can be disseminated throughout the economy.
As Arjun Ramani and Zhengdong Wang argue in a recent commentary, the productivity benefits of AI may be limited if important parts of the economy – construction, face-to-face services, human-dependent creative work – remain immune to it. This would be a version of the so-called Baumol cost disease, whereby the rising relative prices of certain activities choke off economy-wide improvements in living standards.
These considerations should not turn us into techno-pessimists or Luddites. But they do caution against equating productivity with technology, R&D, and innovation. Scientific and technological innovation may be necessary for the productivity growth that enriches societies, but it is not sufficient. Transforming technological progress into broad productivity growth requires policies specifically designed to encourage broad diffusion, avoid productive dualism, and ensure inclusivity.
Dani Rodrik, Professor of International Political Economy at Harvard Kennedy School, is President of the International Economic Association and the author of Straight Talk on Trade: Ideas for a Sane World Economy (Princeton University Press, 2017).
