A maximum entropy approach to the identification of productive technology spillovers

R&D activities by one industry often have positive effects on the productivity performance of other industries, as a consequence of technology spillovers. Econometric problems (such as multicollinearity), however, have prevented researchers from identifying the industries that have been responsible for the most important technology spillovers. This paper proposes an alternative estimation approach (Generalized Maximum Entropy econometrics), which can cope with datasets characterized by a high degree of multicollinearity. For a number of industries, rates of return to R&D expenditures by other industries are estimated on a bilateral basis. Furthermore, productivity effects of spillovers from the foreign counterparts of the industry are estimated. The analysis is done for eighteen industries in twelve OECD countries in the period 1976-1999