Infrastruture Access Pricing and Lump Investments

Abstract

Consider the problem of financing and providing an incentive for an infrastructure provider to enhance the infrastructure. Much of the theoretical literature on access pricing is not applicable for two reasons, Firstly, it ignores the difference between two cost concepts. One, which determines revenue requirements, is the accounting cost of the existing infrastructure, a sunk cost which reflects past conditions. The other, which is relevant to optimal resource allocation, is the forward-looking cost of enhancement, which depends upon current and expected future prices and technolog. Secondly, it assumes that the capacity of the infrastructure is continuously variable, so that there is such a thing as a strictly marginal cost of capacity. This paper deals with the contrasted case where capacity can be increased only in large chunks and where the forward-looking cost of increases amy differ considerably from the sunk accounting cost of existing capacity. Capacity additions, that is to say enhancements of the existing infrastructure required to meet a growing demand, are often indivisible, entailing lumpy investment. When this is the case capacity cannot be gradually increased pari passu with demand but only in large amounts at considerable intervals. Before an increase, the pressure of demand on capacity grows, with the result that congestion costs are increasingly incurred. This justifies higher and higher access charges for use of the infrastructure, though, in practice, non-price rationing is frequently applied instead. Once a capacity increase occurs, congestion costs and the appropriate level of access charges fall drastically. Thus the problem arises that the investment may not be remunerative. Section I describes the nature of congestion costs and the lumpiness of capacity increments in three industries, airports, railways and electricity transmission. Then the problem of paying for investment is posed with the aid of some simple static models in section II. Finally, section III examines some complications ignored in these simple models and some possible solutions to the problem.