Sunk costs have been incurred already and cannot be reversed (The Economist, 2004). Joint costs are incurred in connection with multiple projects.
Two pitfalls which may affect benefit-cost analyses of public projects are:
According to standard engineering economics, the first step should be to identify all cost components which are "separable" (or "incremental") with respect to different associated projects or different phases of one large project. What is left over are the non-separable or "joint" costs. There is no theoretically correct basis for allocating non-separable costs. However whatever method is used should stand the tests of reasonableness and fairness. It is especially important to avoid loading up one project which may be heavy on benefits with more than its fair share of joint costs in order to subsidize a second related project which might not otherwise be economically justified. Reasonable approaches include allocating joint costs in proportion to other costs, in proportion to benefits, or in proportion to some combination of these. It is advisable to perform sensitivity analysis to measure the extent to which varying the allocation rule for joint costs affects the selection among the alternatives being considered.
A related pitfall involves excessive bundling of too many different features into one "take it or leave it" alternative containing weaknesses which if removed would significantly improve the overall payoff of the project. A key principle is "Insofar as practicable, separable decisions should be made separately" (Grant et al, 1990, p. 10). Alternatives should be loaded up with additional features only to the extent that each increment stands on its own merits, and makes the entire project more attractive.
Sunk costs are defined as money that has been spent or irrevocably committed. The only relevant costs to be considered are those occurring in the present and future. Although past actions can certainly influence the magnitudes of costs and benefits for future alternative actions, those past commitments must not be permitted to otherwise influence the decision. For example, in private investment analysis, a common mistake is to evaluate an asset which might be committed to a new project at its original acquisition cost, rather than at its current asset value. In public investment analysis, past commitments to particular transportation technologies often make it politically difficult to consider other technologies objectively.
The Economist. Economics A-Z (webpage). 2004. Available at: http://www.economist.com/research/Economics/alphabetic.cfm?letter=S#sunkcosts.
Grant, E. L., W. G. Ireson, R. S. Leavenworth. Principles of Engineering Economy. Eighth Edition. John Wiley & Sons. New York. 1990.
Newnan, D. G., J. P. Lavalle, T. G. Eschenbach. Engineering Economic Analysis. Eighth Edition. Engineering Press. Austin TX. 2000.
Eschenbach, T. G. Engineering Economy: Applying Theory to Practice. Second Edition. Oxford University Press. New York. 2003.