Can a tiny fish and an old hot dog teach us anything about whether health insurance reduces mortality?
How a day of fishing raises questions about one of the canonical studies in the Medicaid expansion literature.
I just got back from a wonderful week of fishing, solving crossword puzzles, and reading decades-old newspapers at a beach house in the south of Norway.
After several attempts fishing for hours on end without as much as a bite (aside of course from that “monster-bite” my brother still claims but that the rest of us never saw much of) I decided we had established a solid baseline and that it was time for an experiment. In the absence of live bait I would try to bait my hooks with an old hot dog I found in the back of the fridge. The hotdog was small and shriveled, so to avoid making the pieces so small that they would disintegrate in the water, I cut it in four equal pieces that I spread across my six-hook rig, leaving two unbaited.
As you might expect, the idea was thoroughly ridiculed by my siblings who among other things pointed out I’m now officially an American, but my genius idea paid off almost immediately — or so I thought — when I got a bite and reeled in the first fish of the year..
The glory ended abruptly once I saw the fish, a beautiful but oh so slight cod, that I’m guessing weighed 8 oz. To my mind the letdown wasn’t just its size, but that it had bitten one of the unbaited hooks, which I thought served as what economists call a placebo test: essentially, if the hotdog works as bait, the fish should have bitten one of the baited hooks.
Disappointed, it seemed I had to release not only the cod but the “hot-dog as bait” hypothesis as well.
But perhaps not, so this is where it gets interesting. If we compare fishing yields on baited and unbaited hooks, we’d say we’ve found evidence against the hot-dog as bait hypothesis — this was my gut reaction in the spirit of a placebo-test. But if we instead compare yields on days fishing with and without hotdogs, we’d say we’ve found evidence in favor of the hypothesis, and we’d be right!
So, to frame it differently, could it be true that hotdogs work as bait even if the fish bites the unbaited hook? I think so. For instance, it's possible that the scent of the hotdog attracts fish, but that the fish I happened to attract was too small to bite the baited hook. But the story gets even more complicated.
Being more of an economist than a fisherman, the moment I saw the fish on the unbaited hook I thought of Miller, Johnson and Wherry’s 2021 paper on Medicaid and mortality. As I’ve written before, it’s an excellent paper that, akin to my comparison of days fishing with and without hotdogs, finds states that expanded Medicaid experienced larger reductions in mortality relative to states that didn’t.
Most people interpret this to mean that expanding health insurance coverage (lowering the uninsurance rate) reduces mortality, but for that to be the correct interpretation you would expect the mortality reductions (or at least the largest reductions) to occur among previously uninsured people, but instead the effects they found were actually very similar for previously insured and uninsured people, or to continue the fishing analogy, akin to comparable increases in bites on both baited and unbaited hooks.
I know this is maybe just a cute story but it did make me think about the subtle art of econometric science and policymaking more generally. Now back to work!