by Jurgen Brauer, October 2006
Copyright: J. Brauer. No reproduction without permission.

Earlier this month Edmund Phelps, an economics professor at Columbia University in New York, got a surprise – and so did his colleagues worldwide – when the Nobel committee called to say that it had awarded him the 2006 economics prize. He had been tipped for a number of years but as the call never came, other professors began to be championed. Among them were Paul Romer of Stanford University, widely regarded as the inventor of modern economic growth theory, and Jagdish Bhagwati and Paul Krugman, both eminent international trade theorists. And if it wasn't specific professors that were tipped, speculation ran as to the fields of specialty that might be ripe for an award. Among those, surely, one would need to count the economics of technology and the diffusion of technology within and across economies. Technology, after all, improves productivity – at home and in the workplace – and makes for healthier, longer, and more convenient lives.

One view of how technology spreads, once a new invention comes along, is borrowed from medicine. Technology spreads like a contagious disease, an epidemic!

One view of how technology spreads, once a new invention comes along, is borrowed from medicine. Technology spreads like a contagious disease, an epidemic! A central source, the inventor, transmits knowledge. The inventor is in touch with, say, 100 people but the 100 people in turn are in touch with another 100 people each. That amounts to 10,000 people already and accounts for explosive growth in the spread or diffusion on a new technology. Of course since there are a limited number of people on earth, the growth must stop. At some point, people who are unaware of the new technology are harder to find (everyone assumes that everyone already knows!) and so, after an explosive phase of knowledge diffusion, a phase of very slow growth follows as the market is saturated with general knowledge about the invention. Relatively simple technology will spread faster than more complex technology; likewise technology will spread faster where there is greater density of population so that "infection rates" take place faster. Technology will also spread faster in environments that consist of fairly homogenous populations in terms of background knowledge and experience, and less fast when it has to "jump" from one community to another. For example, certain technologies will spread very fast among "techies" in the computer world but more slowly, maybe much more slowly, among non-techies.

Often people need to be persuaded to adopt new technology because adoption can be costly and needs to be set against the likely benefits to be gained.

The epidemic model is useful but has its limitations. It assumes that people are willing and able to receive information and then simply adopt the technology. In practice, people often need to be persuaded to do so because adopting new technology can be costly and needs to be set against benefits to be reaped. Models that focus on the adoption decision of an individual firm or person are called probit models. "Probit" refers to the probability that a firm or person either adopts or does not adopt a given new technology, and they trace the determinants by which people arrive at a "no" or "yes" adoption decision as well as the speed with which people switch from one to the other.

Understanding the determinants of the decision is of obvious importance for government policy. If technology improves lives, then we will want to know how to mitigate the "no" determinants and, conversely, how to make the "yes" determinants obtain more weight in the decision making process. Larger firms, for instance, can spread the cost of a new technology over larger production runs or larger numbers of customers to which it supplies services. But in many contrary instances it has also been found that smaller firms adopt new technology more quickly either because they are more nimble in their decision making or because the new technology is the brainchild of an inventor who then starts up a new, and necessarily small, firm. For example, there were no big companies that wanted to deal with the new-fangled personal computer technology of the 1980s!

Changing over from one technology to another involves search, learning, and switching costs that may be considerable. Potential suppliers should be glad to ease those costs – but they also need to beware of inadvertently undercutting old technology from which they can still earn a revenue stream. Why push the envelope if there is still money to be made from business-as-usual? One policy implication of this potential problem is clear: encourage innovators to enter an industry by reducing whatever barriers-to-entry may exist.

Epidemic models tell us about the spread of technology over time (the "when"); probit models are better at informing us about the "who" and the "why".

Whereas epidemic models are good at telling us something about the cumulative spread of certain technology over time (the "when"), probit models are better at informing us about individual adoption decisions (the "who" and "why"). Neither model is very good at the "where," that is the spread of technology across geographical space, either within or across countries. There are other models available that help economists think about technology and its spread. Clearly this is important research and some lucky, and no doubt deserving, fellow may one day obtain a Nobel prize for it as well.