Sunday, January 15, 2012

a string of zeroes

Loss aversion is the aspect of prospect theory that gets the most attention. A kink in the value function produces lots of neatly testable implications, so tons of empirical work studies it. And all of this collected wisdom is summarized as "losses matter a lot, but gains don't matter very much."

Loss aversion is very important: when comparing losses and gains, losses do indeed dominate individuals' decisions. But reference dependence itself is also very important, and overlooked, since an important consequence of it is a negative (lack-of) outcome: our minute-by-minute experiential utility is mostly a long string of zeroes.

We're really good at predicting what will happen to us. We plan and execute, leading to intended predictable results, and we have plenty of experience with random chance and external factors to not be so surprised by those either. How frequently are we significantly surprised? Not often. And that's how often our utility functions differ from zero.

This implies that, actually, both losses and gains are VERY important. That meshes with experience. It's true that, if I am expecting to sit in a middle seat and get moved to the window on a flight, the bump in utility is smaller than if I've been counting on a window seat and get moved to the middle. But either of those experiences has an enormous effect on my short-term happiness*, and either one is much more likely than a direct comparison between losses and gains such as is the focus of the prospect theory literature.

[Let me be more precise. There are two technical details that can interfere with this interpretation. First of all, it matters what the reference point is, and second of all, it matters whether the gain-loss value function is the same thing as the utility function (or is there is a separate standard consumption utility component.)

As to the first question, I'll jump on board with the Koszegi-Rabin version of prospect theory, which endogenizes the reference point as recent expectations. (I suppose that sounds like I'm choosing that variety in order to support my conclusion, but actually the way I'm thinking about it is that the conclusion is self-evident and that this supports their particular reference point theory.)

As to the second question, this is why I qualify the original statement with "minute-by-minute experiential". I'd rather be a millionaire (who earned their wealth in a deliberate, predictable method, never experiencing gain-loss utility in the process) who never wins anything than a homeless person who sometimes surprisedly picks up ten dollar bills on the sidewalk, but the short-term experiential utility of both the millionaire and the homeless person is dictated not by bank account balance but by small surprises.)]

*I recently had a flight in which I was assigned a middle seat, and tried to change it, but was told I couldn't. When the plane had completely finished boarding, the window seat next to me was still open. It made me so happy that I'm still excited to write about it a week later...

No comments: