Dollars and Jens
Friday, August 17, 2007
capital gains taxes
I want to re-float an idea I mentioned in April, but with more numbers. (I like numbers.)

The idea, again, is that we make a tax that is savings-neutral by having you index the cost basis of long-term capital gains, but have you then pay regular income tax on it. This seems particularly satisfying to me in the case of an individual earning money from trading on a frequent basis; economically, it seems to me that the money you make in excess of the risk-free interest rate is the actual gain attributed to your labor, rather than a simple decision to defer spending. The decision as to whether to defer spending is not affected by taxes up to the point at which you simply earn the risk-free interest rate.

The best argument against this is that it would be complicated — though I raised an economic point in the April post, and have given no more thought to it since then; I still suspect it's invalid — but I don't imagine this adding an appreciable amount of complexity to the tax code. I imagine the 1040 instructions would be supplemented by a table like this:


When you report capital gains, you're already asked for the date of acquisition. If you buy a stock in 1997 for $100, it's not that hard to look up in the table that we're treating that as $140 in current dollars; if you sell it in 2007 for $200, add $60 to your Adjusted Gross Income.

To keep things this simple, your cost basis only accrues interest at midnight on January 1; this would create some preference for buying late in the year and selling early in the year, and I'm only giving credit for entire calendar years in which a position is maintained in part to reduce the value of exploiting that distortion, which I think is, on a practical basis, pretty small if you have to hold the position for a full calendar year in order to realize it. The chart starts in 1990 largely because that's the most recent time we went from a tax regime in which capital gains (as measured as though dollars in one year were the same as dollars in another) were treated the same as new income to one in which they were not; my sense of fairness wouldn't be too aggrieved if we simply said that any position established before 1990 is priced as though dollars before that point were constant, so that a position established for $100 in 1970 is treated as $195 in today's dollars; it would be more fair, probably, to extend the table back, though, and not too complicating either. So whatever.

Powered by Blogger