I've Got a Sinking Feeling About This

Mike Munger brings to our attention a paper that purports to "measure" whether sunk costs matter. In the abstract, the authors claim:

"Behavioral economics implies that teams favor players chosen in the lottery and first round of the draft because of the greater financial and psychic commitment to them. Neoclassical economics implies that only current performance matters."

Now, I am not an expert on the literature on sunk costs, but I have taught the topic, and thought about it some. The mainstream view appears to me to fluctuate between:

1) a priori, only future costs can be considered, a proposition which can be made true by defining future costs appropriately; or
2) people do worry about sunk costs, but they really ought to stop doing so.

If we adopt viewpoint one, then "empirical" studies are irrelevant: the proposition is true a priori.

If we adopt viewpoint two, all this study would tell us is that NBA management has gotten this message. It certainly would not decide any question dividing behavioral and neoclassical economics. (And it is kind of weird to say that behavioral economics "implies" that people pay attention to sunk costs: it is an empirical discipline! Perhaps it finds that they do so, but it certainly doesn't assume this.)

15 comments:

  1. I'm in agreement with the substance of your post, Gene, but I think you misfired when you wrote: "And it is kind of weird to say that behavioral economics "implies" that people pay attention to sunk costs: it is an empirical discipline!"

    Yes, it would be weird if they wrote that, but they didn't. According to you, what they actually wrote was, "Behavioral economics implies that teams favor players chosen in the lottery and first round of the draft because of the greater financial and psychic commitment to them."

    So that makes perfect sense: Behavioral economics has found a bunch of situations where sunk costs affect current actions (apparently), and in the setting of professional sports this approach would imply blah blah blah.

    ReplyDelete
  2. I Think my straightforward paraphrase of what the authors said is more likely than your convoluted paraphrase.

    ReplyDelete
    Replies
    1. Man you are one of the top-5 most stubborn people I regularly talk to...

      How's this? Of the following two options, I claim they are saying (B) more than (A), and further that there is nothing weird about (B):

      (A) "Behavioral economics implies that people pay attention to sunk costs."

      (B) "Behavioral economics implies that people running professional sports teams pay attention to sunk costs."


      Since this paper is the first (?) to measure this, it obviously can't be the case that behavioral economics has already investigated and rendered an opinion on (B).

      Delete
    2. "Since this paper is the first (?) to measure this, it obviously can't be the case that behavioral economics has already investigated and rendered an opinion on (B)."

      Which was exactly my point, and why I was saying that sentence is weird!

      I am not saying your interpretation of what they wrote is impossible. Maybe it is what they meant. Then they should have written what they meant, and I wouldn't have said that sentence was weird!

      Delete
    3. OK Gene, the Apollo astronauts tested whether a hammer and feather fell at the same rate on the Moon. If they had said, "Newton's theory implies that this would happen, so we're going to test it," would you object that Newton used empirical means to come up with his view, and that this sentence was therefore weird?

      The reason I'm digging in here is that you are totally 100% unequivocally wrong, in both letter and spirit. When you test two competing theories, you tease out what each *implies* (or predicts) for a novel situation to which neither has been calibrated. There is nothing wrong with that particular sentence from the Abstract of this paper.

      Delete
    4. 1) "The reason I'm digging in here is that you are totally 100% unequivocally wrong..."

      Hmm, i have readily admitted a possibility that your interpretation is correct. But you are *absolutely certain* mine is wrong! And just who is supposed to be extremely stubborn?!

      2) And it doesn't even bother you that Keshav agrees with you?! (Just joshing, Keshav!)

      Delete
    5. And: behavioral econ may have found that most people worry about sunk costs. That does not imply anything about whether NBA execs, very smart people with lots of money on the line, will also worry about sunk costs!

      Delete
    6. Never mind Keshav. It will bother Bob more that *I* agree with him.
      The claim in this case is that BE has empirically established that people usually do not ignore sunk costs. This is an extrapolation of that finding.

      Delete
  3. Gene, I agree with Bob. If someone said "Physics implies that this rocket doesn't have enough fuel to achieve orbit", that wouldn't be strange at all. Would you really reply to that by saying "Physics is an empirical discipline, so it's weird to say that physics implies the rocket equation. Perhaps physics finds that the rocket equation is true, but it doesn't assume it."?

    ReplyDelete
  4. Is "If we adopt viewpoint two, all this study would tell us is that NBA management has gotten this message."

    meant to say

    "If we adopt viewpoint two, all this study would tell us is that NBA management has NOT gotten this message."?

    ReplyDelete
    Replies
    1. No: the study shows they ignore sunk costs. They HAVE gotten the message.

      Delete
    2. OK, thanks.

      I since clicked on the link to the study and see "Second, we use regression discontinuity to capture changes when a player's draft position crosses thresholds. We find that teams allocate no more time to highly drafted players. "

      That wasn't clear from the post.

      BTW: I am missing Bob's objection to 'And it is kind of weird to say that behavioral economics "implies" that people pay attention to sunk costs: it is an empirical discipline!'

      I assume its in the job description of behavioral economics is to tell us whether they pay attention to sunk costs or not.

      Delete
    3. Correct, rob: and if behavioral econ happens to find this is true for the general populace, that does not imply anything at all for any special subset of that populace.

      Delete
  5. "The problem is that the discussions of sunk cost often seem to become normative: the reader is being told he ought not to worry about sunk costs. But how does one fit that in with subjectivism?"

    Is this common in the field of economics? Because it seems to happen a lot. And is it something economic analysis should avoid doing (i.e., concluding that because privatizing a forest will increase wealth it should therefore be done)?

    ReplyDelete
    Replies
    1. This is a confusion. It is NOT normative to say that you "should" ignore sunk costs because should is not expressing a value judgment. It is just a linguistic short cut. It menas "if you want to maximize x, do y." In calculus we might say "you should use integration by parts" or "you must substitute for ..." These are not normative.

      Delete