For society and industry alike, efficient allocation of resources is crucial. Numerous tools are available that in different ways rank available options and actions under the aim to minimize costs or maximize profit. One common definition of economic efficiency is least cost per unit supplied. A definition that becomes problematic if cost take negative values. One model, where negative costs are not uncommon, is expert based/bottom up [marginal abatement] cost curves. This model is used in many contexts for understanding the impact of economic policy as well as optimizing amongst potential actions. Within this context attention has been turned towards the ranking problem when costs are negative. This article contributes by widening the discussion on the ranking problem from the MACC context to the general definition of least cost per unit supplied. Further it discuss why a proposed solution to the ranking problem, Pareto optimization, is not a good solution when available options are interdependent. This has particular consequences for the context of energy systems, where strong interdependencies between available options and actions are common. The third contribution is a proposed solution to solve the ranking problem and thus how to define economic efficient when costs are negative. (C) 2016 Elsevier Ltd. All rights reserved.