Generation capacity expansion models can be interpreted in terms of equilibrium in a competitive environment.  The property remains valid in a risky world assuming that all agents in the economy are risk-neutral. The dual variables of the model play a crucial role as they represent prices and margins in these states of the world.  We show that the Lagrange multipliers associated to the nonanticipativity constraints are the profit margins of the different technologies. We give a sampling procedure for estimating the distribution of these profit margins (which allows one to compute accurately statistics of these distributions).
This reveals that plants have quite different risk exposures that investors maybe reluctant to value on the sole basis of expectation.  It is also  known that the NPVs of nuclear, coal or gas plants drastically change with the discount rate adopted.  In the second part of the talk, we rather use the modern approach of risk averse stochastic optimization based on risk measures.  We particularly focus on the good-deal risk measure (introduced by Cochrane and Saà-Requejo). We show its nice interpretation in corporate finance terms and that it is still amenable to solve large-scale problems.