Most of us fall victim of the sunk cost fallacy. In short, a sunk cost is an expenditure made in the past that cannot be modified. Neither now nor in the future. A sunk cost is a cost that cannot be recovered. The English language has a nice expression to describe it in plain words: "There is no use crying over spilled milk". The milk won't flow back into the saucer pan.
Many people nevertheless behave as if there was a hope that the milk would go back where it was.
A prime example is that of capital budgeting. Capital budgeting is the field that deals with the evaluation of investment decisions, that is whether one should buy a new piece of equipment, invest in a new factory etc... The usual treatment of such a decision is to carefully identify ingoing cash-flows and outgoing cash-flows over the investment lifespan. But not any kind of cash-flows though. Only those that are unambiguously triggered by the new investment. The economists call them marginal cash-flows.
This is generally where the sunk cost fallacy strikes hard. People are for instance tempted to include past expenses and, for instance, to allocate existing overheads. Including past cash-outlays is somehow a way to try to recoup the cash lost from past mistakes. This is the best way to loose money twice! It does not make any sense to overburden a project with costs that have nothing to do with them. Cut your losses first and then figure out whether some new venture is worth investigating.
Allocating existing overheads to forthcoming projects is bad and, sadly enough, quite common practice. Anybody who has taught capital budgeting rules knows that his or her audience insists on including a percentage of existing overheads into the outgoing cash-flows of the project. The usual argument goes on like this: "Well, if you don't do it who gonna cover them?".
There are different ways of addressing this objection. First, assume that an ideal world where everything can be processed just-in-time. Such a question would not arise. No fixed costs would have to be pre-committed. Only when a project would require a given amount of costs to be implemented would these costs be taken into account as they are indeed triggered by the project.
Now, we know that just-in-time is a fiction and that some costs have to be pre-committed. If every new project is charged by some overhead allocation, one ends up with the rather unfortunate result that "the fatter the better". Again, this is not because one may have too fat a legal department within the corporation that one must try to recoup this legal expense through wrong overhead allocation decisions. The truth is that the wrong decision was made at the time when corporate money was allocated to legal. It would have been wiser to wait, namely not to exercise one's allocation option to soon.
If it turns out that legal is too fat, the best way out is not to try to justify its existence by allocating the fat to new investment projects but to take the brave decision to downsize it!
There are for sure a host of good reasons why some costs must be pre-committed (signaling, contract obligations etc...). Nobody denies them. Hopefully all the options have been carefully looked at before leaping. Once these costs are incurred they won't change whether or not some new projects happen or not.
The key step then is obviously to go hunting for good projects and create new wealth opportunities. This hunt is not like horse racing: If you find a fast horse, ride it as fast as possible and don't let anybody make you believe that your horse needs handicapping!