Be Careful! A Short Note on a Possible Bias in (Trade) Structural Change Analysys

I was accustomed to think that the world, in this of "modern economic growth", is becoming less specialized: the invention of new goods is, only partially, a "Schumpeterian" process, in the sense that new goods sometimes replace old ones, but it can also happens that these new goods simply are added to the old ones. Also in modern theoretical literature emerges (at least) the idea that producers use an increasing variety of intermediate goods and that consumers are likely better with more variety of goods in their hands. The process of change in the produced/consumed goods is one of the aspects of the broader spectrum of economic structural change that accompanies economic growth, "structural change" being one of the basic stylized facts of growth according to the Nobel Lecture of Simon Kuznets. In this paper, I will suggest that sector disaggregated data, necessary to study structural change, contain a bias that "hides" this process and causes a drift. By using trade data (because of their higher sector disaggregation richness) I'll first show that there is a tendency for a steady increase in sector concentration. Next, I will argue that this is due to the impossibility to properly register product innovation and finally, through a very rough model and an empirical example of two countries, I will also suggest that this reflects differently in developing and developed countries.