Thread by @SHamiltonian

Feldstein made so many contributions to public finance, but the one that's had the greatest influence on me is his pivotal insight (REStat 1999) that we can measure the entire welfare impact of taxes via the response of taxable income. A thread. 1/12

mitpressjournals.org/doi/abs/10.116…

Historically, empirical work on the impact of taxes on behavior focused mainly on labor supply. But the income tax has the potential to affect many decisions. How many hours to work. But also work effort. Human capital accumulation. Tax evasion. Whether to give to charities. 2/12

Many of these margins are difficult or impossible to measure individually. This makes it hard to measure the total deadweight loss of the income tax, and in turn what the optimal income tax rate should be. Further, many of these margins are not observed in admin tax data. 3/12

In 1999, Feldstein published this remarkably lucid paper that blew all of this away. The idea is very simple. The income tax rate applies to taxable income. All of the aforementioned margins of behavior are, over time, aggregated in taxable income. 4/12

If you work more or harder or smarter, you get more taxable income. If you evade taxes or give to charity, you have less taxable income. Because the tax rate applies to taxable income, all of your behavioral responses must, by construction, be bundled up in taxable income. 5/12

This is incredibly freeing empirically. Taxable income is a single variable. We no longer needed to worry about figuring out the impact of taxes on wages or on the multitude other behavioral margins. And taxable income is reported in high-quality administrative tax data. 6/12

Every paper written since on behavioral responses to taxes has been influenced by this insight. It gave impetus to improved empirical methods for estimating the ETI (see his JPE 1995 piece below) which would inform better tax policy across the world. 7/12

jstor.org/stable/2138698

The result always seemed too good to be true. I tried I don't know how many times to break it. Invariably to no avail. Because you really can't escape the elegance of it. It truly is brilliant. It seems so obvious in hindsight. But the paper was published just 20 years ago. 8/12

The closest anyone's come is Chetty (2009) who showed that because not all the costs of tax evasion are real resource costs (the penalties are a transfer), the evasion response needs to be measured separately. But that's not what I'd call kryptonite. 9/12

aeaweb.org/articles?id=10…

In my JMP, I made the point that Feldstein's result applies only to the income tax rate. But tax systems are more complicated than that, and if we want to set all tax instruments optimally, then we need to observe changes in all the bases to which those instruments apply. 10/12

But that's really just a reapplication of the Feldstein idea to other tax bases, so it's not even really a novel contribution. Feldstein strikes again! It truly is a legendary result. It is the result that will not die. 11/12

And it aptly highlights what was great about him. He had remarkable economic insight. He could see through the noise and frippery and cut right to the core economic intuition. He is the exemplar of a dying breed of economist's economists. We are impoverished by his loss. 12/12