My earlier work on growth and technological change culminated in the book Economic Transformations: technological change and long term economic growth by myself, Kenneth Carlaw and Clifford Bekar. Currently, Carlaw and I are building on the basic growth model expounded in that book.

Most recently developed endogenous growth models follow Romer and Lucas in using a “flat” representation of technology that is expressed as a single variable or production function parameter. In contrast, empirical work on technology as found in authors such as Nathan Rosenberg and Paul David shows technology to be structured in many complex ways.

The literature on general purpose technologies (GPTs) contains one line of theorizing that has attempted to impose more structure by modeling technology in two tiers: a broadly applicable general purpose technology and the many derivative technologies that are required for the GPT to become effective.

Our latest model, “Sustained Endogenous Growth Driven by Structured and Evolving General Purpose Technologies", Journal of Evolutionary Economics, 2012”, models technologies in three tiers, each of which includes many agents. The first tier is a pure research sector in which agents produce new pure knowledge, using production functions that depend on both existing pure and applied knowledge. This new knowledge is accumulated over time and occasionally leads to the discovery of new technology that has the potential to evolve into a GPT — which it sometimes does. The second tier is an applied research sector in which agents employ the discoveries of the first tier in their production functions. They produce applications that are useful in both the consumption and the pure knowledge sectors. The third tier is a consumption sector in which agents use the technologies produced by the applied research sector to produce consumption goods.

This model incorporates at least ten important characteristics of general purpose and other widely used technologies that have been established by historical research, most of which have not previously been modeled.