From Wisconsin Economic Analysis Guidelines
It is occasionally necessary to deal with "sunk costs" in an economic analysis study. This term refers to costs related to the project which have already been incurred or irreversibly committed, and which cannot be recovered. A simple example is a set of unused conduits that were installed at an intersection for a possible future set of traffic signals. If the conduits are still in good condition and will be used for a new signal installation, their cost would be excluded from the economic analysis computations for the signal project, but if the conduits are damaged or in the wrong location they have no value to the project and the cost replacing them would need to be included in the computations.
Sunk Cost vs Economic Loss
It is important to make a distinction between a "sunk cost" and an "economic loss". This distinction relates to the potential to recover some of the costs that have already been incurred. For example, imagine that a young married couple purchases a matched pair of beautiful new Wisconsin-built motorcycles. Then, a few weeks later they discover that the woman is pregnant, and realize that they should have purchased a minivan instead. This is not a "sunk cost". While the couple may suffer some economic loss in selling the motorcycles for less than they paid, they can still recover a significant portion of their costs.
A sunk cost occurs when all or nearly all of the amount that was spent is unrecoverable. Often, these costs are literally sunk in the ground. For example, in the 1960s a substantial sum was spent on earthwork for a freeway-to-freeway interchange in Milwaukee County on I-894 west of 35th Street, but for various political and financial reasons the freeway that was supposed to intersect I-894 was never built. The resulting earthwork has no resale or salvage value; it is a sunk cost. Therefore, when I-894 is eventually rebuilt in the future, the costs that were incurred in the 1960s should not influence our decision in any way. If the earthwork is in just the right place for the new design, we should evaluate the project it as if that earthwork costs nothing. If the earthwork needs to be removed or modified to accommodate the new design, we should asses it based on the actual costs we will incur to remove or re-grade it.
The Psychology of Sunk Costs
Analysts need to be aware that decisionmakers often allow sunk costs to sway their decisions in irrational ways, especially if it is "high profile" project. An extreme example of this phenomenon occured when the Concorde supersonic passenger jet was developed by the British and French governments in the 1960s and 1970s. When the price of petroleum rapidly increased in the early 1970s, it soon became apparent to analysts that the costs of operating the Concorde would always exceed the amount that passengers were willing to pay to ride it. Meanwhile, several major technical obstacles delayed the project and greatly increased the cost and complexity of the design. Nevertheless, decisionmakers in both the British and French governments continued to pour money into the project, often arguing that "we've already spent a lot to develop the plane, so to avoid having the project look like a failure we need to find more money to finish it." This financially irrational thinking resulted in development costs that were nearly 6 times the original estimate. Only 14 planes were ever built, each one burdening British Airways and Air France with decades of operating losses. No other airlines ever bought any of the aircraft, in part because the intensive level of maintenance required for the planes cut into the amount of time they could actually be flown. In the end Concorde's initial technical success was overshadowed by the crash of one of the planes in 2000 (resulting the death of 100 passengers, 9 crew, and 4 people on the ground). Without doubt the project was also a financial disaster for British and French taxpayers.
Dealing with sunk costs rationally requires a forward-looking analysis. For example, suppose that a decision has been made to build a river bridge which costs $5.0 million and is expected to deliver $7.0 million of benefits during its service life (B/C =1.40). Construction begins and $1.5 million is spent to build the west bridge abutment. Then as construction of the east abutment begins, it is discovered that the soil on the east bank of the river is too weak to support the weight of the bridge (a problem that was not detected by the pre-construction soil borings). After some quick engineering work, three options are identified:
- Spend $6.0 million to remove the bad soils and drive deeper pilings, while keeping the bridge on its currently proposed alignment.
- Abandon the west abutment and spend $5.5 million to relocate the bridge a quarter mile further north where the soils are better.
- Cancel the project entirely.
In this situation the $1.5 million that has already been spent to build the west abutment is irrelevant to the decision. The bridge abutment is a sunk cost: we cannot sell it and it has no other use. Therefore, the project should be analyzed as if we are starting from scratch. Thus we can make the following observations:
- Option 1 provides $7.0 million of benefits at a cost of $6.0 million (B/C = 1.17)
- Option 2 provides $7.0 million of benefits at a cost of $5.5 million (B/C = 1.27)
- Option 3 provides no benefits and costs nothing (a B/C ratio cannot be calculated).
Conclusion: In this case, Option 2 is the most cost-effective choice: even though $1.5 million of infrastructure will be abandoned, relocating the bridge is still the cheapest way to get across the river. Nevertheless, in a totally rational decision-making environment we would also consider whether a project with a B/C of only 1.27 is still worthwhile compared to other projects that could be built with the available $5.5 million (in other words, if there is not too much political pressure to build the bridge we might consider walking away from the project and doing something more useful with the money).