MODELING COST-STRUCTURE - THE BELL SYSTEM REVISITED

This paper analyses a recent public policy decision and the crucial role played by the choice of the functional form. The break-up of the Bell System is indeed a classic case of antitrust settlement, and constitutes one of the most prominent and controversial applications of microeconomics to public policy. It is argued that demanding multi-output cost properties, such as economies of scope or subadditivity, should not be modelled with the translog functional form. The reason for this is the considerable robustness problem displayed by the translog cost function, which is the source underlying many of the ‘expert disagreements’. It is suggested that the non-robustness is due to what is termed the ‘Flip-Flop’ property, and it is shown that the available empirical evidence from the pre-divestiture Bell System supports this argument. To obtain robust results, we propose that the analysis is constrained to the proper region of the cost function. Using the same test as before, but imposing properness on the test region, we find that the pre-divestiture Bell Data are fully consistent with a natural monopoly. Furthermore, this result is robust. © 1990, Taylor & Francis Group, LLC. All rights reserved.