Sunk Cost

A "sunk cost" is a cost that you have already incurred and that you cannot recover.

People often have an irrational desire to use products for which they have paid a lot of money, or to continue following a plan that has required a great investment. They think that if they abandon the product or change their approach, they will be throwing money or time away. That's not true - the money and time have already been thrown away. Continuing to use a bad product or to follow a bad plan is only increasing the amount being thrown away.

Whenever you make a decision, it has to be made according to what you know now and upon reasonable expectations of the future. Hoping that bad results from a past decision will eventually "turn around" if you stick with that decision is wishful thinking of the worst sort. It always takes courage to admit you made a bad choice and that you need to change your mind, but it is the only thing to do. It takes even more courage to try to convince others that they made bad choices and need to accept the sunk costs, but that's something you sometimes have to do.

Programmers are faced with these kinds of choices all the time. They are always under pressure from superiors and customers to reuse existing software and hardware and to avoid buying new things when old things are available. Programmers make BuyDontBuild decisions, and are expected to stand by such decisions until the end of time.

A similar problem manifests itself among inexperienced investors--people will hold onto an underperforming stock on the theory that if they sell now; they will lose their money. If the price of a stock is less than what you paid for it; you have already lost money. A rational decision on whether or not to buy, sell, or hold a stock is only based on what you think the stock will do in the future; and not on the need to recoup any losses which have been incurred. (Of course, this advice is abstract--in the RealWorld there may be tax and other implications which affect the decision to sell now or hold out. In particular (AmericanCulturalAssumption) you cannot write off a loss on a stock until the stock is actually sold--the loss is "recognized". Of course, IamNotaLawyer...

The problem is that technology changes and new ideas come so fast that new stuff is often going to be cheaper than reusing old stuff, but that concept is counter-intuitive to most decision makers. They think in terms like "ReturnOnInvestment". So start using business terms like "sunk cost", and maybe you will be more successful at getting through to them.

One of the fundamentals you learn in introductory management accounting is that when comparing alternatives (ROI, etc.), sunk costs are irrelevant and not considered or factored into the calculations.

That's the point: when a manager says "We should use X so that we can get some return on our investment", you reply "No, that doesn't make sense. X is a sunk cost, and is not providing value as we continue to use it." By recategorizing X as a sunk cost in the manager's mind, instead of as an asset, you can trigger a desirable business-school-mandated reaction.

Compare also operating costs versus capital expenditures. It appears that operating costs are not accounted with nearly much weight as capital expenditures. A programmer or team of programmers may know and be able to demonstrate that the operating costs of one year spent maintaining and updating a creaking, dusty, cranky dog of a system are some multiple of the cost of replacing it wholesale and could be recovered in 90 days, but from at least one programmer's experience, some accountants consider that an unimportant factor in their calculations. Someone with management/accounting background may be able to shed more light on that.

some accountants consider that an unimportant factor in their calculations

Good ones don't (good? heck, ones that know their butt from a hole in the ground). Obviously you don't need to be an accountant to recognize the problem with those calculations. Other than dumb accountants, the real problem may be a dysfunctional capital budgeting process which may be a result of the political issues inherent in setting budgets (whether capital or operating) or which may just be a badly designed process (which itself could also be for political reasons).

See if you can wrap you minds around this one:

Amateur investors will, say, buy a stock at $20. They'll lose $5, leaving them at $15. Then, rather than say, "I wonder if that $15 could be put to better use", they'll say, "I've got to hang onto this dog until it gets back to $20, where I can drop it."

Go ahead. If you try hard enough to understand it, your head will implode with a gentle violence. Not unlike their money.

There's an application of this in PokerGame: Once you've put money into the pot, it's no longer yours. Beginning players will often make an early mistake, then think "Well, I've already put this much money into the pot, I might as well see it through to the end." That's incorrect. How much of the pot's money used to be yours is completely unrelated to whether or not you should bet now, this time around.

Or staying to the end of a film or play you're not enjoying, because you've paid for it and you want to get your money's worth. The mistake being that you've ultimately paid to enjoy yourself, not to be an observer of a performance. As in the previous examples, there is always the chance that things will improve, but 'YesterdaysWeather'.

People often have an irrational desire...

BuddyProgramming seems to be another example of this fallacy. Is there a name for this fallacy?

I see this lots of other places also. We spent <some large amount of money> on <some tool>., therefore we must use <some tool> as much as possible.

Like all FallaciousArguments, sometimes the conclusion happens to be true. If a car-repair shop bought one of those gadgets that plug into and talk to the car computer, they might as well spend a couple of seconds plugging it into every car that rolls in.

However, other times the conclusion is false.


View edit of August 18, 2012 or FindPage with title or text search