22 July 2008

InformationGain('not fair') == 0.0

It appears there are new numbers available on the tax burdens of various income strata, which confirm my suspicion that Democrats' vociferous bellyaching about "the rich not paying their fair share" is nothing but high octane BS. Various other sources seem to agree.

First of all "fair" is a word that conveys almost no semantic information and does nothing to enhance my understanding. Hearing someone describe something as "fair" or "not fair" does not add a single bit of information to my knowledge of the situation. Calling something unfair is a lazy way of saying "I don't like it but I can't be bothered to explain why." "Fairness" as a philosophical concept is fine, but the labels "fair" and "unfair" are thrown about with such abandon, and have become so unmoored to any meaning or system of determining what is or is not fair, that they are almost pure noise. If you want to make some Rawlsian distribution argument, or invoke something like the max-min fairness algorithm, by all means go ahead and do so. But 99% of the time someone says "it's not fair" all I hear is "Waahhhhhh!! Me no like!"

As I mentioned once before, if you're going to make a qualitative claim like "the rich aren't paying their fair share," then it is incumbent upon you to define what a "fair share" would be. After all, how can you tell me that something is too high or too low if you don't have a way of recognizing what the right level is? You're not Goldilocks. You're not sampling bears' porridge. It's a little more complicated than "This tax burden is too low. This tax burden is too high. But this tax burden is juuuuust right!"

This was made exceptionally clear in a thought experiment presented by Steven Landsburg in The Armchair Economist. (Which is highly recommended, by the way. Far superior to the heavily lauded Freakonomics.) I don't have a copy with me, so I'll attempt to recreate an approximation of it:
Say you've got a town with two citizens, Anne and Bob, and one public service, a well. Anne makes $20K/yr and takes 10 gallons of water a day from the well. Bob makes $50K/yr and takes 7 gallons/day from the well. What taxes should Anne and Bob each pay?
It seems you have three options:
  1. Ignore income and just charge for water consumption. Simple, but runs the risk of being regressive, which most consider "unfair."
  2. Ignore water usage and just tax income. Okay, but what if Anne uses 10000 gallons/day and Bob uses 1 gallon/year because he dug his own well? Where's the "fairness" in making Bob pay for something he doesn't use?
  3. Specify some function which combines income and water usage. Then you have the unenviable task of justifying which function you chose (and the particular coefficients) with some kind of rigorous grounding, otherwise they're just random numbers which are no more inherently fair than any others.
If you can't pinpoint a "fair" tax regime for a nation with two people, one form of income, and one public service, what makes you think you know what will be fair for 300 million people, thousands of forms of income and wealth, millions of forms of consumption, and hundreds of thousands of often unquantifiable government services?

I believe that sooner or later — unless you take a very hard stance like "no taxes at all" or "no private property" — you're going to have to either derive your taxation parameters through empirical models attempting to maximize the economic growth, tax receipts, etc or you're going to have to pull magic numbers out of a hat, or both, since your models will probably need magic numbers of their own. At that point please have the decency to say something like "I pulled this number out of the aether. It is based on nothing more than my opinion that no one should have to pay more than 50% of their income to the taxman. That could have just as easily been 49% or 52% or 13%, but I chose 50% as a base constant of my system." If you can not build a system without introducing arbitrary postulates then you should have the honesty to declare those assumptions openly and recognize them as malleable and morally baseless.

Postscript: Previously wisdom from TJIC: "The word 'fair' means nothing more than 'what I think should happen.'"

1 comment:

  1. Great example from Landsburg I'd somehow forgotten. The one that sticks in my head (possibly from the second book) is the one with three guys on an island. One spends a lot of time to build a fancy house for himself and one builds a quick shelter and lounges on the beach all day. What sort of house-building assistance do each of them owe the third?