Thread by @jasonfurman

I had a broader thread on @paulkrugman's column on Dem economists. Here is a short technical one on an error he made linking theory & empirics. Am belaboring the point but if he is going to accuse economists of being wrong then best not be wrong himself.

https://twitter.com/jasonfurman/status/1496232259371868165?s=20&t=TJApITOQIElPfcF1GozLvQ

This plot shows an almost perfect correlation of gas prices and gas taxes. We show a similar one in Ec10.

It says nothing about whether the incidence of the gas tax is on consumers or on producers. In fact, even if the tax was 100% on producers the graph would look the same.

Imagine global oil prices are $80/barrel. There are two countries: one levies a tax of $1/barrel and the other $2/barrel.

If the tax is fully paid by consumers then the prices would rise to $81 and $82 respectively and you would observe a perfect correlation.

BUT suppose the tax is paid 100% by producers. The price of oil would fall to $78.50. It is a global price so would be the same everywhere. Including the tax consumers in country 1 would pay $79.50 and consumers in country 2 would pay $80.50.

Again, a perfect correlation.

So observing a perfect correlation in the scatterplot does not test the hypothesis of whether the incidence is 100% on consumers, 100% on producers, or somewhere in between. All of these theories yield exactly the same prediction for that scatterplot.

Now it might be that incidence is 100% on consumers so they fully benefit from the tax holiday. That would be because of arguments about supply being perfectly elastic. My thread explains why that is almost certainly wrong, especially in the short run.

https://twitter.com/jasonfurman/status/1496232281887088643?s=20&t=TJApITOQIElPfcF1GozLvQ

But why show a scatterplot that is irrelevant to the point you're making because even if the opposite was true the scatterplot would still look the same?