Sunday, November 7, 2010

Irrationality or uncertainty

Decided to return to work full-time about a month ago and as a new employee I had to attend an employee benefits seminar. The major items in these seminars are health insurance (medical, dental and vision) as well as flexible spending accounts. I was befuddled by the array of choices and I wonder how many people are as well.

For instance if I had data on individual insurance plans of enrollees as well as their medical spending over the year as well as their alternative insurance plans, can I, based on the data, decide whether they are irrational or uncertain?

This is similar to the the observation that when faced with a decision on how to allocate their money in a 401(k), many people default to equal proportions and that many individuals (including myself) pick funds that are highly correlated and under-diversified.

Is this irrationality? It is certainly a large deviation from the plan of a perfect foresight infinite horizon agent but does the fact that we deviate from what dynamic programming tells us we should do (dang if I know since formulating Bellman's equation isn't exactly something I excel at) evidence of irrationality per se or computational limitations.

If it is the latter then this can be addressed by having asset allocation tools made available as many mutual fund companies now do. Likewise, tools to select health insurance can also be built based on projected spending and income - so does deviation from such recommendations reflect irrationality, hidden information or a combination of many factors?

1 comment:

Ashwin said...

Many "irrational" actions are really entirely rational when we take into account the irreducible uncertainty under which the actions are chosen. I'd especially recommend anything written by Gerd Gigerenzer but especially his aptly named book called "Rationality for mortals: how people cope with uncertainty".

I discussed the same theme and Gigerenzer's work in the context of asset allocation/investment in a blog post a while ago here .