Knowledge, growth without scale effects, and the product life cycle

Résumé du document

For several reasons, knowledge cannot be treated like any other commodity. One of these reasons is the non-confrontational nature of knowledge, which means that one person's use of certain knowledge does not diminish the relevance of another person's use of the same knowledge (at the same time). This important property of knowledge was used in several early models of R&D-based growth such as the researches of Romer (1990), Grossman and Helpman (1991), and Aghion and Howitt (1992). In these models this property leads to a scale effect, which boils down to larger economies growing faster than smaller economies (with the measure of size suitably defined (cf. Backus, Kehoe and Kehoe 1992)). In an influential paper, Jones (1995a) pointed out that growth with scale effects, as predicted by the early models of R&D-based growth, is inconsistent with empirical facts. Over the last 40 years the OECD countries have experienced a tremendous rise in the number of people involved in R&D activities whereas the growth rates of per-capita income have shown no corresponding increase.

Extraits

[...] These authors find that knowledge is diluted when it is spread over more subjects, even when these subjects are ‘close', i.e. technologically similar to the subject that originally created the knowledge, part of the same firm, and/or (iii) geographically near to the subject that originally created the knowledge. The dilution of knowledge is the natural way to ultimately explain growth without scale effects, for it signifies that it is not the absolute amount of resources spent on the creation of new knowledge but rather the amount spent per economic subject (e.g. [...]

[...] Besides as the result of productivity may increase through learning by doing (i.e. the accumulation of tacit knowledge) which by its very nature is closely related to a single product and cannot be expected to be very useful in the production of another product without some adaptation in the form of doing and consequently learning being required. Whether or not the broad dilution effect creates enough of a productivity difference between sectors to lead to growth without scale effects subsequently depends on other features of the model, namely the speed at which demand shifts between sectors and the relative timing of the introduction of new products (sectors). [...]

[...] The idea of this hierarchy is that some human desires are more pressing than others: the very basic needs such as water, food, sleep, etc. (physiological needs) are the most pressing needs, while self-development and the thriving to exploit ones potential to the fullest (self-actualization needs) are the least pressing needs (Maslow 1970). In the hierarchy, the physiological needs have the highest priority before needs like self-actualization and hence people only consume products that satisfy lower priority needs if the more basic needs have already been satisfied. [...]

[...] That is, in industries that comprise relatively large firms productivity grows relatively fast. The extent of the effect is such that when average firm size increases with ten percent, the rate of growth of productivity increases with roughly The positive relation holds for several tests of robustness, e.g. the use of different measures of size and repeating the analysis for samples of different country-composition, but loses its statistical significance when country dummies are included. This last result is in line with their finding of large cross- country differences in the size distribution of firms. [...]

[...] With structural change taking place, a sector can exhibit growth scale effects but because demand shifts to a (new) less productive sector relatively fast, the scale effects simply vanish at the aggregate level. As the model is very simple, the - obviously complex - interaction among the variables underlying this mechanism of structural change is not explicitly modelled. Instead, to keep the intuitive nature of the mechanism just described, the variables that are of interest are straightforwardly defined as a function of time in a way that is exogenously given. [...]