Specialization, Knowledge Dilution, and Scale Effects in an IO-Based Growth Model

We present a model where accumulation on non-rival knowledge drives growth but where the scale effect, which may be positive or negative, vanishes asymptotically. This result stems from the interaction between technological differentiation and market structure dynamics. Firms are linked to each other in networks of spillovers determined by the technological proximity of their activities. These spillovers-networks span only a fraction of the total economy and the average technological distance between firms increases with the size of the economy. When the economy expands, less related activities become profitable and specialization increases. As a result, the networks expand at a slower pace than the overall economy. In the limit, the networks cease to grow with the size of the economy. A larger economy, therefore, accumulates a larger tock of total knowledge but not necessarily a larger effective stock of knowledge that is useful to the individual firm. The reason is that the latter expands with the size of the network to which the firm belongs. The scale effect vanishes asymptotically because the stock of effective knowledge that each firm exploits is unrelated to the size of the economy when this is very large.