Looking backward over my life, it has been a queer mixture. I have lived through a period of transition and therefore know what it is like at both en… - Arthur Lewis

My interest in overhead costs was the structure of prices in situations where average cost per unit exceeds marginal cost. The Pareto rule was that price should equal marginal cost, but to apply this rule would bankrupt the firm. In practice, such situations oscillate between bankruptcy and monopoly, as in the airline industry today. The general inclination of economists in those cases was to enforce marginal pricing and subsidize firms to the extent of the differences between marginal and average cost. This was hardly practical, as an industry-wide policy. Neither could it be justified, as many taxpayers would be forced to pay for services that they did not use. If one started from the premise that those who use the service should pay for it, the problem reduced to how to spread the fixed costs among the users. Here I started from the railway principle of “charging what the traffic will bear” and linked up with the new price discrimination theory, as elaborated by Joan Robinson.
Another aspect of overhead costs was the time dimension. Demand was not steady, but fluctuated. If the output could not be stored, there would be times of idle capacity, regular or irregular; how was the cost of this to be shared? I demonstrated that the correct approach to this problem was to treat the fixed investment as a producer in joint cost of different outputs at different times, each paying what it could bear, and subject to the sum of payments not exceeding total cost.

There are still people who discuss industrialization as... an alternative to agricultural improvement... this approach is without meaning in the West Indian Islands. There is no choice... between industry and agriculture. The islands need as large agriculture as possible... It is not ... that agriculture cannot continue to develop if industry is developed … the opposite is true: agriculture cannot... yield a reasonable standard of living unless new jobs are created off the land

A number of developing countries had been developing for a long time: Ceylon, for example, for a hundred years. Why was the standard of living of the masses still so low? One could understand this for much-exploited South Africa, but how for fairly enlightened Ceylon? The answer to both questions came by breaking an intellectual constraint. In all the general equilibrium models taught to me the elasticity of supply of labor was zero, so any increase in investment increases the demand for labor and raises wages. Instead, make the elasticity of supply of labor infinite, and my problems are solved. In this model growth raises profits because all of the benefits of advancing technology accrue to employers and to a small class of well-paid workers that emerges in an urban sea of a low-wage proletariat. In the commodities market an unlimited supply of tropical produce also gives the benefit of advancing technology to the industrial buyers, by the process already described.