From now on, these posts will be reviews of the established literature. I will assume, as is standard, that people are completely selfish. The point is to show that, even if people are selfish, there are desirable incentive effects of income taxation.
As is often the case, the results of 50 year old theory turn out to depend on the assumption of symmetric information. If some agents know things that other agents don’t know, then free market outcomes can easily be Pareto inefficient.
One very old example is due to A Michael Spence who considered a model in which people don’t learn anything useful in college, but get degrees just to prove they can. Spence was dean of Harvard’s faculty of Arts and Sciences, so he should know. Basically the idea is that what matters is having the diploma not knowing how to read it. In fact, I personally possess a diploma signed by a Michael Spence which is written in Latin — a languge which I can’t read (although I can guess that Cantabrigia is the very old word for Cambridge).
In this model getting a degree requires unpleasant effort and is not accompanied by any useful or delightful knowledge. The point is that more able people can get the degree with less unpleasant effort. Also employers want able employees and pay people they know are able more.
The model has 2 Nash equilibria. In one, no one gets a degree. In the other, only the able get a degree — the effort demanded (potential including the major and maybe even a GPA above some level) is just enough to scare away the less able.
This equilibrium is inefficient. Students study and suffer for no socially valuable reason. The (partial) solution is an income tax. By reducing the (post tax) reward given to those who demonstrate ability, the temptation of the less able to try and fake it is reduced. This means the required amount of signaling is reduced. This helps the able.
If there are lots of able people and few less able people, the tax can cause a Pareto improvement. It can be cheaper for the able to pay the less able to admit that they are less able, than to scare them off with a very demanding signal..
Even if the tax and transfer is not Pareto improving, it increases money metric welfare (the silly measure which is not increased by lump sum taxes and transfers which take from the rich and give to the poor).
Signaling is probably a very general phenomenon. This is suggested by the phrase “dress for the job you want, not the job you have”. My mom tried to convince me that it was worth the effort to learn how to spell as it worked as a signal (she didn’t convince me but — hey spell checkers eliminated that signaling mechanism — also taught me how to spell most words with instant feedback).
There is another still more obvious way in which asymmetric information might create good incentive effects of income taxation. The more informed party is often tempted to lie to the less informed party. There is a huge literature on mechanisms which can prevent this. In the real world, one of them is prosecution for fraud. Here the gain from lying is often some number of dollars, and the potential loss is not. The income tax can keep people out of grey areas of the law.
It is true that the income tax doesn’t tell people which way to head out of the grey area. The income tax is not relevant to blatant criminals who must hide what they do and who, with notably rare exceptions, do not suffer if they add income tax evasion to their other crimes.
On the third hand, another desirable effect of the income tax was the conviction and incarceration of Al Capone.
I’m going to add absurd CEO compensation to the list. I consider extremely generous CEO compensation to be legal embezzlement. It isn’t prosecuted, but, if the CEO goes to far, it can be profitable for a predator to get 5% of the rent (boodle) by taking over the firm. Or shareholders might revolt and elect a board which fires the too greedy CEO. It is a fact that tax cuts were followed by the explosion of absurd CEO compensation. I don’t think that was an accident (although allowing people to hide income as capital gains was as important as the top marginal income tax rate).
Here again, I have to not an undesirable effect. The point is that the income tax makes greedy CEOs more risk averse. That is good. It also makes greedy potential entrepreneurs more risk averse, which is less ideal. The point is that, short of prosecution, there are lots of relatively rare punishments which scare people and which are not purely economic. I think that they are generally socially useful. Their relative strength compared to monetary incentives is increased by income taxation. I think that, on balance, this is a good thing.