Income tax time is an appropriate moment to go to the heart of President Obama’s complaint about the taxes Warren Buffett and other rich people pay, or don’t pay. What the president is really complaining about is that the tax rate on capital gains is too low.

But there is a more basic question to be asked: why tax capital gains at all?

Did you know that the term “capital gains” does not even appear in the official income accounts for the U.S. economy? That’s right. No matter how high stock prices climb, they do not affect the official reckoning of national income one iota.

“Capital losses” aren’t included either.

When stock prices soar, stock owners are wealthier—at least they feel wealthier. When stock prices plunge, owners of stocks feel less wealthy. But none of these ups and downs have any bearing whatsoever on the official calculation of the income for the economy as a whole.

So here is the policy question: If we are going to have an income tax, should we tax only income? Or should we tax activities, events and transactions that are not counted as part of our national income?

At the New York Times Economix Blog, Princeton University economist Uwe Reinhardt argues that capital gains should be taxed at the same rate as ordinary income (which is included as part of national income, by the way). I had a debate about all of this with Michael Kinsley at Slate some time back. Interested readers may want to refer to the text of that debate for more details than I plan to go into here. Also, don’t miss Steven Landsburg’s devastating critique of Uwe’s piece.

Imagine a poker game. At the end of the evening, some players walk away winners. Some are losers. No real income has been produced at this event. It’s strictly entertainment. The winnings of the winners are exactly equal to the losses of the losers. Should the IRS get involved? If your answer is “no” I like the way you think.

As it turns out, however, the IRS does get involved and it does so in a very unfair way. It taxes the winner’s gains but limits the ability of the losers to deduct their losses. (Gambling losses can only be deducted from gambling winnings, not from other income.) If the IRS treated everyone fairly (symmetrically) there would be no point to taxing gambling income. The deductions by the losers would offset the gains of the winners and there would be no net revenue for Uncle Sam.

Now let’s turn to stock prices. One way to view the stock market is to see it as a place where people also make bets. They are betting on the future income of corporations. Eventually the future will arrive, however. The companies will realize their actual income and they will pay taxes on it. If the firms return some of this income to investors (stockholders), the investors will pay a tax on their dividend income. If the firms pay interest to bondholders, they will be able to deduct the interest payments from their corporate taxable income, but the bondholders will pay taxes on their interest income.

Here is the bottom line: There is no need for the IRS to tax the bets that people make along the way—as stock prices gyrate up and down. Eventually all the income that is actually earned will be taxed when it is realized and those taxes will be paid by the people who actually earned the income.

But, as in the case of the poker game, let’s suppose the IRS decides to foolishly get involved anyway. What will be the outcome? Uncle Sam almost certainly won’t collect much money unless (as with gambling) it treats people asymmetrically. And that is exactly what it does. It taxes capital gains ferociously while limiting the ability of people to deduct their losses.

Moreover, unlike the poker game, I have complete discretion over when I choose to sell a share of stock. I can time my gains so they fit into this taxable period or the next. I can time my losses in the same way. It is because of this discretion that the federal government would get almost no net income from the taxation of capital gains if it treated losses the same way it treats gains. That’s why the government imposes so many arbitrary restrictions on how both losses and gains can be realized. But these restrictions interfere with the flexibility of the capital market. And they do so for no good reason—because eventually all corporate earnings will be realized and taxed anyway.

Uwe’s post was in response to a post by Greg Mankiw, in which Greg gives five examples of capital gains—all involving housing. The problem with all five examples is that a home is an asset that produces income that is counted as income in the national accounts. If the house is rented, the income is rental income. If the homeowner lives in his own house, he is enjoying “imputed” rental income. The IRS taxes the former, but does not tax the latter. If tax policy were consistent, all “income” from housing would be taxed the same way and (as in the case of a share of stock) there would be no reason for the IRS to worry about capital gains and losses from home sales.

Greg also addressed the question of “carried interest,” which I want to handle as part of a more general problem.

Suppose one person sitting at the poker table is not just enjoying an evening of pleasure and entertainment. Suppose he makes a living gambling, just like you and I make a living in some other occupation. How should the tax law treat that person? Answer: differently. Similarly, suppose someone makes a living trading stocks and bonds. If she works on commission, her income will be taxed as ordinary income. But suppose her income comes in the form of a share of some of the trades she makes. Should her income be treated differently than the gains and losses of the passive investors she is managing money for? Answer: yes.

I am not going to propose a technical solution here. I only want to say that these special cases are problems that can be dealt with in special ways without taxing capital gains generally.

Finally, I want to address an argument Paul Samuelson once made. If capital gains are taxed at a lower rate than ordinary income, Samuelson argued, he would find ways to convert ordinary income into capital gains. Actually, the IRS has created a lot of obstacles to keep him from doing that. But a more basic response is: why not avoid all these problems by reforming the entire tax system along the lines of a flat tax?

The idea behind a flat tax can be summarized in one sentence: In an ideal system, (a) all income is taxed, (b) only once, (c) when (and only when) it is realized, (d) at one low rate.