Technology and Development
I’ll return to our old friend Schumacher today, but with a different focus and more recent insights from Acemoglu and Johnson (2024 winners of The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel, along with James A. Robinson). The topic for today is what Schumacher terms intermediate technology in the context of development economics. The problem to be addressed is the uplift of economic and living conditions for those living in developing nations, principally those outside major cities – i.e., in villages and small towns. An important factor for these residents is the lack of financial and institutional supports for industrial development using the same technology as employed in developed nations. Schumacher identifies a negative effect on employment if development focuses solely on industrial enterprise in major cities, which he sees as leading to the “migration of destitute people into towns that cannot absorb them” (SIB, p. 183). While I have a rosier perspective than that of Schumacher (he was not a fan of cities as a general rule), I agree with his notion that maximizing output per person will not be as effective as maximizing the available work opportunities. Schumacher calls this principle intermediate technology for its focus on providing an intermediate technological solution – one that improves working conditions beyond their current levels but does not automate production to the point of eliminating low skill labour. Without an extensive tangent, the term low skill (or unskilled) has always bothered me. It seems a snobbish way of describing skilled labour obtained over many years of practice, though perhaps not in a formal educational setting. It is also the case the agricultural work in the developed world (again, a bit of an imprecise term) requires much formal education in agricultural science, mechanical engineering, and other fields.
Schumacher makes it clear that modern industry cannot meet the needs of developing nations, which are rich in labour and short on capital, given that the processes are tailored to nations short on labour and rich in capital. He defines an intermediate technology using the reference of developing nation technology as a $1 technology and developed nation technology as a $1000 technology. What is needed is a $100 technology, one which improves the efficiency of production without a significant reduction in labour requirements and at a capital cost. Without restating Schumacher verbatim, we can consider two ways of approaching intermediate technology: 1) a small-scale deployment of a current technology. He gives the example of a small-scale petroleum refinery that has a lower capital cost but with similar levels of efficiency of their larger counterparts; 2) the use of a technology from the recent past, during a time when developed nations relied more heavily on labour rather than capital. A couple scenarios I have observed in my own life ring true with the approach of Schumacher. As a high school student, I had the opportunity to travel to the Peruvian Andes on a school trip. We worked with a local village to construct a preschool. One of the key features of the construction that I recall was the use of local materials. Having heard of other projects using modern materials (e.g., a steel roof, plywood walls, etc.) leading to issues for the community when the structured failed or required repairs, I thought the use of local materials was an excellent idea. While it did not result in a shiny modern building, the school fit into the context of the village and employed methods and materials known to the residents. I was also reminded of a program a friend took part in during our time as undergraduate students. She worked with an NGO to repair medical equipment in a central American hospital (I believe it was in Guatemala). The hospital received donations of medical equipment, but it often lacked the local expertise to make repairs to the equipment. This case reminds me of the way transit management often works, with government being excited about a new train line but more apathetic about operations and maintenance.
Let me turn now to the work of Acemoglu and Johnson. In their book “Power and Progress”, they provide an interesting analysis of technology and its relationship with prosperity. A motivation for their work is the effect of artificial intelligence on progress and its equitable distribution. They highlight the cases of South Korea, Taiwan, Mauritius, and China as seeing rapid growth (of greater than 5 percent per year) due to their integration into the technology economy, in conjunction with exports to global markets. Acemoglu and Johnson provide examples that parallel the arguments of Schumacher, though they do not reference his work. For example, high- and middle-income nations invest heavily in research into crop pest and pathogen solutions. Corn strains resistant to the European Maize Borer and Western Corn Rootworm have been developed over the last few decades. However, most of these crop strains and chemicals are not useful to African or Asian farmers facing infestations of African Maize Stalk Borer or Desert Locust. They cite estimates that global agricultural productivity could be raised by 42 percent if research was redirected towards pests and pathogens affecting crops in low-income nations. They also highlight the negative effects on equality associated with capital-intensive solutions adopted from high-income nations in development work. Such solutions lead to a small and well-paid workforce operating alongside a larger and low paid workforce using existing capital.
We now come to a central thesis of Acemoglu and Johnson that the AI revolution risks producing an economy that replaces low-skilled labour without a viable alternative outlet for that labour. AI requires highly skilled production and service labour, resources severely lacking in most developing nations. A key point here is that technology should raise the marginal productivity of labour, not replace it. Such a pattern was associated with the Industrial Revolution of the 18th century. Development began with the power in the hands of a small number of capital owners, before shifting to workers through changes in social power. Acemoglu and Johnson chart this history over 1000 years and make the argument that AI is not following the same path of facilitating marginal labour productivity growth and distribution of social power to the population. They advocate for machine usefulness over machine intelligence. It is an interesting discussion, but one that goes beyond the scope of this article.