Friday, April 12, 2013

Articulating time-to-build without committing the sunk cost fallacy

I'm worried I'm thinking/writing about this in the wrong way, so I wanted to run it by readers.

So I'm writing about undergraduate course choice right now, and one of the major issues I'm considering is the stock of courses by major that a student already has under their belt. A natural way to think about this (I think), is as Kydland-Prescott type "time-to-build" investment where the degree itself takes time to build.

I've got my own optimization problem that's different from Kydland and Prescott, and a macroeconomic law of motion for capital isn't really the right way of thinking about a microeconomic human capital investment anyway. But I want to make a reference and use the same sort of language as they do.

My concern is that when I try to articulate time-to-build I have this feeling like I'm saying it in a way that commits a sunk cost fallacy. The right way to put it - I think - is not really to talk about the optimization problem in terms of the investments that you've already made, but instead to look prospectively at how much building there is left to do (which of course is a decreasing function of investments you've already made). So I would say that when a student chooses whether to take a junior year course in engineering or English, one of the things that they consider is the amount of total investment in these fields that would be required to produce a degree in engineering or a degree in English. For any given student in any given semester, this amount of investment required is going to be inversely related to their stock of courses in engineering and English to date.

I think that's just fine. But the fact that it's still functionally tied to past investments makes me nervous. I have flash backs to a decade ago (good God - is it really a decade ago?) in my intro micro class getting it drilled into us that only dummies worry about sunk costs.

What do you all think?


  1. Meeeh... and now I'm thinking the classic example is always a factory you've started building.

    Which makes me wonder why no one raises this concern with Kydland and Prescott (unless the point with them is simply a lag in investments coming online - but I think it's more than just that).

    I suppose I could always just throw it into the discussion because even if its irrational people may still behave that way.

  2. Sunk cost fallacy tells us to evaluate only future costs and benefits when making decisions. But those future costs and benefits may depend on past choices, in which case past variables (such as accumulated capital stocks) are still relevant for decision-making (but not in the naive way the fallacy warns against).

    By the way, I think that K&P did use time-to-build mainly to incorporate investment lag, which they argued helped to better explain dynamics of aggregate investment. Curiously, this approach hasn't caught on much in later RBC/DSGE literature.

  3. This doesn't sound problematic to me, though I might be missing something.

  4. By looking at the cost to complete a student is not falling prey to a sunk cost fallacy.

    As I recall myself and classmates were constantly updating our academic and career targets according to our assessments of our abilities, our likes and what we thought the job market would be.

    One of the problems that students have is that generally politicians and educators are completely incompetent at predicting the job market and business lies about job prospects in the hope of generating an over supply of employees and so drive down wages. We see those factors at work in the talk about STEM programs.


All anonymous comments will be deleted. Consistent pseudonyms are fine.