This paper analyzes decisions on emissions of a stock pollutant under uncertainty in a two period model. Decisions are based on a weighted average of expected utility (EU) and the MaxiMin criterion. I first show that more weight on the worst case (less weight on EU) may lead to increased first period emissions. The effect of learning possibilities on emissions is not clear in general, but depends qualitatively on the weight given to MaxiMin: For the quadratic utility case, considering prospective learning increases today’s abatement effort, i.e. the “irreversibility effect” holds, if the weight on EU is small. This contrasts standard results on the irreversibility effect for EU which translates to small weights on MaxiMin. There is, however, the possibility of a negative value of learning. It is shown that the irreversibility effect holds if and only if the value of learning is negative. Consequences for the applicability of generalized EU-MaxiMin are discussed.


Lange, Andreas


uncertainty, MaxiMin, irreversibility, learning