Introduction

Our earlier discussion has on several occasions alluded to the impact of uncertainty on allocations involving exhaustible resources. But we have not gone into such issues in any detail. Instead, we have supposed an absence of uncertainty, or at the very least, that one would not go far wrong if all random variables were replaced by sure values (e.g. their expected values). It is time to remedy this. The need for such an exploration is plain. Uncertainty about future possibilities looms rather large, and this may have a telling impact on the rates at which we ought today to deplete our stocks of exhaustible resources. Thus, for example, any appraisal of the rates at which fossil fuels are depleted today must surely be based on an assessment of the likelihood of major technological discoveries in energy generation in the future—such as ‘clean’ breeder reactors or controlled nuclear fusion. Likewise, an appraisal of the depletion rate for iron ore reserves must depend inter alia on a view of the likelihood of alternatives to steel being found in many uses. Then again, new reserves of extractive resources are continually being discovered. But such discoveries cannot be predicted with certainty.

A discussion of uncertainty and its effects on natural resource allocation readily invites one to consider the production and transmission of information. Seismic surveys and laboratory experiments designed to help control nuclear fusion are both aimed at producing information.