Many of us learned about sunk costs in a basic business or economics course -- sunk costs are the ones already in the past. They are those expenditures which, when the results of a particular choice are tallied up, will weigh in on the negative side, but don't have anything thing to do with continuing on with a particular investment.
For example, imagine you approve a thirty million dollar expenditure in a new production process, one which shows a nice return based on some huge productivity improvements. After spending the first five million, however, it becomes clear the original productivity rates will never be achieved. In fact, when you "run the numbers" on the remaining twenty-five million yet to be spent, the return is virtually zero. What do you do?
On a strictly economic basis, you punt. The original five million is a total loss, but you should stop the project because the twenty-five million yet to be spent is just good money being thrown after bad.
In reality, I've almost never seen this behavior.
Most people can't get the five million dollars already spent out of their mind. The want that investment to be productive, and so against all logic, continue to spend in the vain hope somehow the project will make its original projections. By that stage the five million is a sunk cost, and doesn't have any relevance to the decision to press on. Or does it?
In large corporations, the problem is larger than just "forgetting about" the initial five million outlay. It's all about admitting to mistakes or errors.
Stopping the project is tantamount to confessing to imperfect management ability. In most large corporations that comes with a political pricetag -- at the least a loss in confidence, at the worst a loss of job. If the decision maker admits to the error, they are likely to be punished, and the more the organization focuses on finding the guilty, the bigger those stakes become.
On the other hand, if the decision maker allows the project to continue, they will likely have time to devise a plan to lay-off blame elsewhere. Bad execution, deceptive suppliers, poor analysis -- the list goes on and on. I've even observed managers repositioning others to be the scapegoats for impending disasters, when what they should have done is called time out and stopped the spending.
The larger, more political, more punative and more complex the organization, the more likely they are to suffer the inefficiencies of the sunk cost fallacy.