A previous post discussed the issue of the deferral of capital gains tax and why, if there is going to be a capital gains tax, this deferral distorts economic decision-making. The issue of this post is whether mandatory mark-to-market taxation can fix the problem.
Mark-to-market reporting and taxation seems to be a relatively straight-forward fix. That way our investor A would pay capital gains taxes every year the stock appreciates, even if he does not sell, and would be left with the same amount of post-tax cash as our investor B. This would eliminate the distortion.
One objection to this would be that A, not having sold his shares, may very well not have the cash on hand to pay the tax. But that is a minor issue. A could just sell some shares to get the cash to pay the tax, just like B does, or borrow the money perhaps with a loan secured by the shares themselves.
The more profound problem is valuation. To tax a gain, you must first determine what the gain in any year is and for that you need to value the investment. Moreover, this value must be determinable by some objective measure or there is going to be endless fudging of the reported numbers and endless litigation about that fudging.
One case in which determining the value of an investment objectively is easy is when there just was an arm's length purchase or sale of it, like when A acquired it or when A eventually sells it. Moreover, for investments with a liquid, public market (let's call them marketable), it is also easy to objectively determine the value at any time.
But many perfectly valid investments are not marketable. Think of your house—you may be able to guestimate some value at the end of the year, but that is not nearly precise or objective enough for you to file your taxes on that basis—or investments in closely-held companies which trade only rarely and never publicly.
Ok, so then you say, let's at least require market-to-mark taxation on marketable investments and leave deferral in place for non-marketable securities. Most investments are marketable anyway, so that would solve most of the problem, wouldn't it?
It would, but for the ingenuity of the tax bar and financial engineers. It turns out that it is remarkably easy to turn a bundle of marketable securities into a bundle of non-marketable securities with essentially identical performance characteristics. Given this ease, and the great advantages of deferral, such a tax law change would just result in most marketable securities being replaced by non-marketable ones quickly.
How would one do this? Consider a common security like SPY, an ETF linked to the S&P 500 stock index. It is a marketable security par excellence; one can look up its value to a high degree of certainty any second of any day. So if this was an investment you wanted to be in, but you do not want it deemed marketable in order to gain the deferral benefit?
Easy. Your friendly neighborhood financial engineer sets up a trust in a hospitable jurisdiction. Into this trust he places, let's say one million SPYs, the trust then issues a million different securities. Each of these securities guarantees a return which calculated by a complicated formula—different for each security—based on the return of the SPYs and something random, unpredictable but subsequently objectively determinable (Greek minor league soccer scores?). These formulas are specified so as to guarantee that the average return of the million new securities exactly equals the return of the S&P 500 (minus perhaps a small administrative fee to keep the financial engineer out of the poor house), so the trust is guaranteed always to have enough funds to pay out the returns, regardless of how the teams in Greek minor league soccer do. But the return of any one of these securities may vary widely.
Now if you want to hold 10,000 SPYs, but don't want them to be marketable, you go and swap them for a random set of 10,000 different securities issued by the trust. Each one of your new securities will not be marketable (who'd bother to make a market in something so minuscule and ridiculous?), so you can defer gain on them. At the same time, the law of large numbers practically guarantees that your bundle of 10,000 different securities is going to give you the same return at the original 10,000 SPYs.
So requiring mark-to-market on investments whenever there is a market is going to accomplish nothing except reducing transparency of the financial markets and diverting the brains of financial engineers (and some of them are quite clever) away from more productive tasks.
