On the long-run evolution of technological knowledge (Q852331)

Most models of endogenous technological change posit a relationship that links the change in aggregate technological knowledge to the level of existing knowledge and the amount of human capital employed in research activity. The focus on steady states has led many authors to depict this link by means of two particular differential equations. However, the proposed approaches, as the authors of the paper state, are not universally accepted as appropriate descriptions of the production of knowledge. For this reason, the authors revisit the debate about the appropriate differential equation that governs the evolution of knowledge in models of endogenous growth. They argue that the assessment of the appropriateness of an equation of motion should not only be based on its implications for the future, but it should also include its implications for the past. In this sense, they show that the analysis of the backward-looking properties, generates criteria that should be included in the overall assessment. Their analysis is based on the insight that once they stipulate an initial value for the level of knowledge, the solution to the chosen differential equation for knowledge determines its evolution for the time after and before the initial period. Therefore, they may look forward and backward at the implied evolution of knowledge. This way, the approach deviates from the often encountered interpretation of the initial condition, which is seen as a historically given constant. The approach proposed by the authors maintains moreover the view that the evolution of knowledge is plausible if it satisfies two asymptotic conditions: Looking forward, infinite knowledge in finite time should be excluded, and looking backward, knowledge should vanish towards the beginning of time (but not before). The key result of the analysis is that standard equations for the evolution of knowledge used in the modern literature on endogenous growth satisfy these requirements only under non-generic circumstances. In other words, their analysis showed that a plausible evolution is a probability-zero event. Moreover, the analysis showed that, generically, the behaviour of the processes under scrutiny is either implausible in the past and plausible in the future, or vice versa, or implausible at both ends of the time line.