The blog of the Tax Policy Center points out that the latest tax law revision has created a new tax bracket for singles making between $398,350 and $400,000—a $1,650 range.
It is an odd little thing. The blog estimates that under 500 people fall into the range. Of course, this isn’t how tax brackets work, and I’m certain that everybody at the TPC knows this and ignored it for the purposes of poking fun. Everybody making over $400k (a lot more than 500 people) will be paying 35% on their income in this range. But it’s still erratic: if this bracket were merged with the one below it to create a more natural under-$400k bracket, these people would be paying 33% on the income in this range. So having this bracket makes for a 2% of $1,650 increase, which comes out to…let me get out my calculator here…$33.
[I admire that the serious people at the TPC are willing to express some humor from time to time.]
The TPC blog thinks this is silly, but then I started to wonder: why not? Why do we value simplicity in tax rates at all?
Complex laws can distort incentives and cause unpredictability. If the law sets a broad sales tax, then provides an exception stating that furbys, as animate creatures, are not taxable, then the market will distort in predictable ways: furbys will become marginally more expensive because the maker knows its competition is taxed, other makers will start trying to fit the legal definition of furby, and so on. Doll makers would have to take guesses as to whether their products in planning will be taxed or not. A real-world mess caused by complexity in the legal code.
That’s not what’s going on here. The law is a definite formula that applies to everybody. If you make more money, you pay more in taxes. Say that you are planning some future change, like a new job offer paying more, and you actually want to calculate post-tax income before deciding whether to take the job. You do the same thing no matter how many little tax brackets there are: you look up the tax liability in a tax calculator, tax preparation software, the tax tables in the 1040 instructions, or some other system that knows the full tax formula. If you’re me, you write a program to take in incomes and spit out tax liabilities; see p 118 of Modeling with Data for sample code. Whatever mechanism you choose, the procedure is the same: you come to your oracle with a number—I’m going to be making $x dollars—and it spits out a number: you’ll be paying y% of that in taxes.
In fact, if the tax rate weren’t just a set of brackets, but a crazy function, like the natural log of income plus the square root of income over $100k, or a Loess-smoothed curve, we would all be doing the same lookups that we do now, even though nobody knows how to calculate a Loess regression.
There are still questions about what formula we should use. There are people who advocate for a flat tax, but that’s really a political argument about whether people who make more money should pay more in taxes, dressed up in an argument for simplification. But I question even that surface argument: is there a real, practical cost to having a more complex tax formula? Given that we already have computers doing our taxes for us, is there more than æsthetic benefit to having formulæ that are calculable by hand? Is simplicity in legal code a desirable end in itself?