the sunk cost fallacy

You’ve got to know when to fold ’em.

Yesterday at Indiana University we hosted Peter Frumkin, who talked about his new book (co-authored with Ana Kolendo) Building for the Arts: The Strategic Design of Cultural Facilities. The book contains a range of case studies of significant building or renovation of cultural facilities in the United States. A key finding, maybe not surprising, is the very high proportion of projects for which new revenues were highly overestimated, and costs of construction, and subsequent operations, were highly underestimated. City leaders, arts groups and donors come across as remarkably optimistic when it comes to projects.

On the analysis of their case studies, I asked the (potentially minor) question of whether the sample selection is biased. Suppose we randomly selected a group of major arts institutions, some of which did some major building in the past few decades and some which did not. Suppose that on average the decision-makers in these organizations are reasonably accurate at forecasting costs and benefits. But that is just the average; some are overly optimistic, and some overly pessimistic. The case studies we observe are likely skewed towards the optimist groups, since, after all, they are the ones who decided to build. The organizations not included in the case studies might be the pessimists who, in fact, should have built, but did not. I’m not sure how much of an effect this had.

But there is one issue that comes up in these projects-gone-wrong, where the net losses are far greater than anyone had anticipated at the approval stage of the project: as information came to light, why are organizations so resistant to change their minds, and back out. Property acquired for a project can be put back on the market. Projects that are too big to be sensible can be scaled back.

But arts leaders, like other humans, are susceptible to the sunk cost fallacy, which holds that once some money has been invested in a project, such that quitting the project would mean a loss that had to be written off, then the project must be followed through to completion. We have a bias against walking away from a bad idea once begun. There can be a number of reasons for this. Our society treats perseverance, even in the face of unfavorable odds, as a virtue. Nobody likes to admit that what they once favored, and advocated for, now looks unwise.

The way around the fallacy is this: on any project, on any given day, ask “what are the remaining costs of completing this project, and what are the benefits that will accrue to us when it is completed?” If the expected benefits exceed remaining costs, it is worth staying with the project. If the remaining costs exceed expected benefits, it is time to think about backing out. The costs that have been expended prior to today, that cannot be recovered, are sunk, gone, and should not be a part of the decision on whether to continue.

Let’s consider two examples.

One: suppose at the outset of a project to build a new wing on a museum the expected cost is $100 million, and the expected benefits are $130 million, where none of the benefits can be realized until the project is finally complete. Suppose one day the project is revisited. Total expenditures to this date are $40 million, but it now appears the final cost of the project will be $160 million. Furthermore, expected benefits, after some market research, have been scaled back to $90 million. It is time to walk away from it. Remaining costs are $120 million, which starting today would need to be spent to generate the $90 million in benefits, and that is not worth it. But what about the $40 million that has already been spent? True, that will be lost by walking away from the project. But completing the project would, in the end, result in an even greater loss, of $70 million ($160 million total cost minus $90 million in benefits). Walking away minimizes the losses.

Two: suppose at the outset of a movie pitch the expected cost is $80 million, and expected revenues are $120 million. The project gets the green light. When the film is in the can, such that all that remains are costs of marketing and distribution, it turns out the production of the film cost $110 million. Marketing and distribution costs are $50 million, so the total cost of getting the movie onto screens is $160 million. When the movie is finished being made, new estimates of box office have fallen to $60 million (the movie is not as good on screen as the idea was on paper). What to do? this movie is obviously a money loser. Yet, the studio would be sensible to release it. If remaining costs are $50 million, and expected box office is $60 million, the studio might as well take that net revenue. Overall, the movie will lose the studio $100 million ($160 million total costs minus $60 million box office). But if it is not released, the studio loses even more money: the $110 million it cost to make the film. It is remaining costs and benefits that determine whether to press onwards (Richard Caves analyses this problem in his great book, Creative Industries).

Beware the error of incorporating sunk costs in deciding future actions. Thinking about changing jobs? Look to expected costs and benefits arising from the move, in terms of increased satisfaction or pay in the new job, and the transition costs of relocating. Put aside what you have invested in the current job, that cannot be recovered. Thinking about selling your house? The relevant considerations are: what could you get for selling it, and what would it cost you to acquire a new place, and the benefits of the move, and the moving costs themselves. Whether you just spent $15,000 renovating a bathroom is beside the point – those funds are sunk, and what matters now is what your house could sell for and what you could buy elsewhere. Thinking about changing careers? Look to the future path in terms of your current field of work, and the alternatives. What you spent years ago getting your college degree is beside the point, those tuition dollars are sunk. Compare future paths, don’t be biased by always looking back in time.