We’re often told that for every published paper there are many other projects that never saw the light of day, and I’ve always thought that it was a shame that we don’t know what those projects are or how many of them there are. So I thought I’d sit down and make a brief list of projects that I worked on for at least some period of time, what I found, and why they didn’t work out. At first I thought I could get all of them in one sitting, but quickly realized that I have too many of these “dead” projects. So this is part 1 of X. These are mostly in no particular order.

    1. Traffic fatalities decrease significantly during a hurricane watch/warning. I thought there might be a tradeoff between warning people and avoiding traffic deaths, but it seems like when you warn people, they stay home and traffic fatalities go down. The overall welfare implications seemed like they would be impossible to pin down, so I abandoned this project.
    2. People are more willing to sacrifice economic growth to protect the environment when the local unemployment rate is low/local incomes are high. This one was even a working paper at one point, but the finding was based on a survey answer which economists don’t like andthere was a similar paper published while we were working on ours, so after a few rejections my coauthor and I abandoned this one.
    3. Corporations have higher sales and higher stock returns in years when average temperatures in the state of their physical headquarters are higher. This one was also a working paper, but welfare implications were unclear and heterogeneity patterns were strange, so we abandoned it.
    4. By allowing individuals to deduct disaster losses from their income but not allowing them to deduct insurance premiums, the tax code creates disincentives to insure against risk. This one was mostly a theoretical project. I thought that to make it really convincing we would also have to empirically show that people’s insurance choices react to their tax rate, but we couldn’t come up with a credible (i.e., causal) way to show this empirically (tried looking at homeowners insurance in the AHS combined with simulated marginal tax rates, but ran into endogeneity issues).
    5. Women under-report their weight and men over-report their height. As a result, BMIs of both genders are under-reported. We found this by comparing the self-reported distributions in BRFSS to the measure quantities in NHANES. Wasn’t clear what the point of such a paper would be, but it was fun.
    6. People who are in their 2nd+ round of the CPS are less likely to be unemployed than people who are in their 1stround. The idea is that this is because there are many detailed questions in the 1stround about howpeople might be looking for jobs and this can plausibly have a causal effect on job search. But the alternative explanation is differential attrition and I didn’t see how this could be ruled out. Plus, there was an experimental paper showing a similar effect (that being surveyed can change behavior).
    7. The benefits of buying crop insurance indexed to county yields in the US can be predicted using a factor that’s not included in pricing this insurance – last year’s yields.But farmers don’t seem to buy more of this insurance when the returns are high; instead, they buy less of it, potentially because of countervailing incentives in other, non-indexed, plans. This was actually the third chapter in my dissertation. I submitted to four journals, no one seemed excited, I didn’t see a credible way to get at the mechanism for advantageous selection and had other projects in the pipeline, so out went this one.