Me me me!!!

So the bottom line is that the original Holmstrom intuition works, but only in small markets when the stakes are relatively low. Otherwise, heterogeneity in costs across competitors matters more for their strategies.

so airlines don't want to enter when demand is high because this good state of the world gets filtered out, and they are less likely to exit when demand is low because they are not punished for this common shock.

RPE has no effect on entry-exit strategies in large markets. Demand is so high that all the airlines always want to enter. In medium-sized markets, removing RPE helps weak competitors, who are no longer compared to their more efficient rivals. In small markets, RPE filters out common shocks,

So we estimated a dynamic game in which airlines compete with one another under the assumption that their contracts contain RPE incentives. (I know. Not an optimal contract. But it is what we see in the data, so our estimation is just characterizing an observed equilibrium with observed contracts.

Not heavily researched. Why? No data. Measuring how much an executive's compensation depends on competitors' performance by looking at proxy statements is a mess. There's information there, but it does not provide clean measures. So it's a casualty of lampost economics.

Just way too many mistakes. It's kinda impressive. This is for coding in anything but python. Even other open-source languages.

I mostly use Claude, so this is why the change in GPT was a surprise.

I haven't used GPT in a while because I've mostly been using Claude.