Cost functions, subaddivity and natural monopoly: review, extension and application to Australian telecommunications

Abstract

This study applies a test for subadditivity (natural monopoly) to Australian telecommunications industry data for the period 1954 to 1990. If an industry exhibits subadditive cost, a monopoly can provide total industry output at a lower cost than multiple firms. The test for subadditivity is dependent on econometric estimation of a theoretically valid cost function. The cost function employed in this study is a multiple output variation of the symmetric generalised McFadden cost function. The main advantage of this specification is the ability to impose concavity on the cost function with respect to the input prices without imposing a priori restrictions on the input substitution elasticities.While there have been numerous previous subadditivity studies, this study is novel in two respects. First, this study contains the results of a direct test involving the provision of data carriage services provided by Australias monopoly carrier from 1970 to 1990. Thus, the test for subadditivity is applied to a relatively new service at a time when demand is in its infancy. Second, the approach to modelling makes explicit allowance for radical technological changes and lags in adjustment.The results indicate cost complementarity between data-aggregate output and large economies of scale. However, these effects are not strong enough to guarantee subadditivity. Analysis suggests that the most likely cause of subadditivity is the extent of network duplication between competitors. Evidence of subadditivity is found for firms that duplicate more than 30% of the networks fixed cost. This implies that at the national level, competition policy is the right choice. This suggests that regulated competition is likely to be no more costly than monopoly.