It's been fashionable to point out that Homo economicus is not a fully rationally self-interested animal. Kahneman and Tversky were instrumental in waking us up to the reality that the human brain is not a universal well-rounded problem-solving machine - and of course, as a product of the accumulated exaptations and legacy systems and local optima and ad hoc functionalities of evolution, why would we expect it to be? Rather than serving as cause for hand-wringing, appreciating this fact lets us either do something to address and correct it, or at least to call it out and create hacks and workarounds. To me this is the promise of the Late Enlightenment.
I recently moved from the Bay Area to San Diego. Looking at gasbuddy.com, I noticed that the spread on gas prices in San Diego seemed greater. My experience of actually price-comparing gas stations has borne this out. There is greater variation over smaller areas in San Diego than there is in San Francisco. Sometimes there are ten cent differences at service stations across the street from each other. What's more amazing is that there are quite often cars filling up at the more expensive one. I've been tempted to walk up and ask people why. Clearly, this bothers me.
This suggests a way to compare the degree of rational self-interest between two geographic areas - take the price on 87 for all the gas stations within two predetermined square miles. Find the standard deviation for each square mile. Take repeated readings over some period of time and average, if you're worried about changes in supply cost rippling through the market and driving up the SD out of instability, rather than irrationality (non-100%-efficient markets is not that same as irrationality). Then look to see if the localities are consistently different. If there are consistent differences, next step: does it correlate with certain chains? Or with demographics of the area (education, income, income distribution)? Or some cultural intangible (i.e. San Diego is too relaxed for its own good)?
The square mile should be controlled so that there aren't geographic barriers (busy highways or streets, water, etc.) that would actually make consistent price differences more rational. The differences I've seen are ones between gas stations across the street or opposite ends of the block from each other. Since the index would consist purely of posted price as reported online, rather than sales, it's assuming that people are actually buying gas at the more expensive stations. But unless there's some bizarre detail at work here, this is a fair assumption - service stations aren't going to set a price at which nobody buys gas. (The bizarre detail could be something like - the local area is dominated by a single chain that profits from products other than gasoline, so its prices are less sensitive to actual consumer behavior; or, a frequent-flyer style gas club not available in both cities. This is why I should actually ask people what they're thinking filling up at the more expensive station, because maybe they are actually thinking something.)
In fairness, I'm very sad to have left San Francisco, and very eager to jump on anything that points the Bay Area or its residents as better than San Diego. But what I've observed anecdotally suggests strongly that San Diegans just can't be bothered to drive an extra 2 minutes to save three dollars. That is to say, it seems that San Diegans are truly less rationally self-interested than San Franciscans, and this provides a way for me to make sure it's not just my own confirmation bias operating against poor San Diego.
The next step is to actually do the calculation, which will take a while. If you run across this post and know of a similar index that's already established, please comment.
C’est la vis, teaching kids how to visualize data
12 hours ago