Dixit on costs

From Avinash Dixit's new book, Microeconomics: A Very Short Introduction, comes this discussion of costs and the supply curve. Dixit is discussing an example of calculating a short-run supply curve for crude oil.

Although this example makes the idea of a supply curve stand out vividly, it is merely illustrative. First, data are not available for all countries; for example Russia and China are omitted for want of cost data. Second, countries or regions are not the right unit of analysis. Ideally we should have data on lifting costs and capacities for all the thousands of individual wells. These differ greatly within a country; they would generate a smooth supply curve instead of the large steps shown. Finally and most importantly, lifting costs are not correct short-run marginal costs. Firms that operate the wells have the choice of leaving oil in the ground, thus producing at less than full capacity, if they believe that prices will rise in the future. In other words, opportunity costs are the correct measure. But these are based on expectations and calculations done within firms, and are not available in reported data. Therefore this example can be used to improve understanding of the ideas, but should not be taken literally (Emphasis added).

One implication of the bit in bold is that, at best, without detailed firm level data you cannot estimate a supply curve. But as expectations are subjective, at least to some degree, it may be impossible to get data on the opportunity costs at all. The expectations held by members of the firm will differ depending on the particular person being asked. Also even if that person can detail their expectations well enough to be part of a quantifiable data set, the data will change person by person. Thus the whole idea of an objective cost function that can be estimated from "the data" becomes untenable.

Now there's a thought that could ruin an econometrician's day.