The Perils of the Learning Model for Modeling Endogenous Technological Change

摘要: Learning or experience curves are widely used to estimate cost functions in manufacturing modeling. They have recently been introduced policy models of energy and global warming economics make the process technological change endogenous. It is not appreciated that this a dangerous modeling strategy. The present note has three points. First, it shows there fundamental statistical identification problem trying separate learning from exogenous estimated coefficient will generally be biased upwards. Second, we two empirical tests illustrate potential bias practice show parameters robust alternative specifications. Finally, an overestimate provide incorrect estimates total marginal output therefore optimization tilt toward technologies incorrectly specified as having high coefficients.