When do sunk costs matter for choices? An explanation of the sunk cost fallacy

A sunk cost is an unrecoverable past cost. Business school professors like to advise MBAs to ignore sunk costs when making strategic decisions. The lesson boils down to don't throw money at a bad investment when better investments are available. People have a tendency to get attached to projects or investments they have spent a lot of time with and a lot of money on. People hope to recover from bad decisions by dedicating more resources to the problems. The advice suggests that people should wipe their hands of bad decisions and strategize their investments on the margin. The logic is that one cannot go back and change the sunk expenses, so one should forget about them and choose the highest NPV (net present value) project available for new investments.

The sunk cost fallacy arises because the common definition of sunk cost is sometimes not fully appreciated. A sunk cost is not only a cost that is unrecoverable and possibly performed very poorly. A sunk cost requires that the value of the original investment can in no way be altered by current choices.

To help appreciate this nuance, consider the following example. If one half-builds a shopping mall, one has made plenty of unrecoverable investments. If one stops at any moment, then the value of the already made investments is very low, making the investments look bad. Perhaps one put in $100 million but the mall in the current condition is only worth $50 million.

Now lets think of what to do with the next million dollars. A marginal decision would examine only the return on the million dollars invested in different projects. Imagine that an investment other than the mall offers an astounding 100% return for sure - so that the one million becomes two million in a year. Assume that the investment in the mall yields 0% on the new one million. If the definition of a sunk cost is not fully appreciated, then one might be tempted to take the other project and make a 100% return.

However, this strategy ignores the possibility that the million dollar investment makes a terrible mall investment into a mediocre mall investment. The capital gain on the much larger sum invested in the mall could significantly outweigh the one million dollar gain on the other project. In order for the one million dollar new investment to beat the outside option of a 100% return, all the one million dollars has to achieve with the original mall investment is to raise the value from $51 (including new investment) to $52. In fact, the mall investment may rise from $51 to $75. Even though the investment in the mall is still negative since one put in $100 million, one is better off throwing money at the bad idea because doing so affects the value of the original investment.

 So the key element of sunk cost reasoning is that the value of a previous investment or sunk cost cannot be affected by new, marginal investments.