"In this paper we argue that endogenous product selection provides a neglected source of bias in measured productivity in addition to the conventional biases of exit self-selection, endogeneity, and misspecification of the production technology and demand. When firms choose their products at a more disaggregated level than is observed in plant- and firm-level datasets, measured productivity reflects both characteristics of the firm and attributes of the products that are non-randomly chosen by the firm. To characterize the resulting bias in measured productivity, we develop a model of industry equilibrium in which firms endogenously sort across products. Following the standard "revenue production function" estimation approach, we use the model to derive the bias in measured firm and aggregate productivity. Calibrating the model's parameters, we show that the bias in measured firm and industry productivity can be quantitatively large and influences the response of both productivity measures to changes in parameter values.Our analysis points to a number of areas for further research. On the one hand, the estimation of structural models of industry equilibrium that feature endogenous product choice is a promising line of inquiry. On the other hand, the development of census of production datasets containing detailed information on establishments' output, inputs, and prices by product is clearly a priority. While census datasets typically allocate establishments to a "main industry" based on their largest good, our research points to the additional insights to be gained from gathering more detailed information on the set of goods establishments supply. It is important to remember, however, that even when this information is available, the industrial classification used is typically coarse compared to the level at which firms make decisions about products. As a result, the bias in measured productivity induced by endogenous product selection is likely to remain a concern."
Recordei agora um postal, redigido à porta de uma moradia em Guimarães com uma garagem onde ouvia uma máquina de bordar a trabalhar, sobre a produtividade e o numerador como variável e não como uma constante. Escrito em Agosto de 2011 explicou melhor um outro escrito em Outubro de 2006.
Alterar o tipo de produtos produzidos é uma forma de alterar o numerador da equação da produtividade. E de que maneira.
Trecho retirado de "Products and Productivity" publicado pelo The Scandinavian Journal of Economics 111(4), 681-709, 2009.
Sem comentários:
Enviar um comentário