First, Bill Kaye-Blake suggests they can be a way of capturing otherwise untaxable non-market activity. The idea here is that DIY work does not involve a market exchange and so is taken out of the tax net. The whole optimal tax literature from Ramsey on is predicated on how to minimise the inefficiency arising from this inability to tax some goods. As an example, I had an uncle who was a keen DIYer. He liked buying houses in need of repair and then doing them up. Once finished, he would sell and move on to a new fixer-upper. For him it was just a hobby, but an unintended consequence was that he was able to move into better and better houses through the untaxed return on his DIY labour that manifested itself as a capital gain. The argument here is perfectly valid, but as a reason or having a CGT, it would be like giving yourself acupuncture with a fork (Ron Jones’ expression). For the very small amount of non-market income-generating activity that would be captured indirectly by a CGT (not my uncle, if first houses are exempt), one would have to bear the other costs of the tax. Even if the net efficiency and equity benefits of the tax for all other reasons were zero and not negative, it is pretty hard to see that this benefit would outweigh the enforcement and compliance costs.
Second, Michael Reddell cites the 1990 consultative committee on the taxation of capital income as stating that capital-gains taxes can be efficiency enhancing when the changes in gains or losses in asset values are not windfall changes. I have not been able to track this document down, but I am guessing that the idea here is related to a 1964 paper by Samuleson, which Andrew Coleman has pointed me to. Samuelson’s paper is very elegant, but it isn’t an easy read, so let me try to illustrate his result with a simple model. Those who don't like models expressed without equations just jump two paragraphs at this point.
Consider a world in which there is a perfectly elastic demand for domestic saving in international markets at an interest rate of 5%. This means that, in the absence of taxes, there is an opportunity cost to both a saver and the country of investing in the creation of a local physical asset in the form of an ongoing return of 5%. It will therefore only be efficient for the country to invest in the asset if the present value of the future payments from the asset exceed the cost of the creating the asset, when the future is discounted at 5%. If we now put on a 40% tax on interest income, then local savers will use the after-tax interest rate of 3% to discount payments, which will lead to a higher present value of a given flow of payments from an asset. If, however, the 40% tax is also levied on the payments from the asset, there will be an offsetting reduction in the present value. In the special case where there is a constant flow of payments, these two effects will exactly offset. In this case, private incentives will lead to the efficient level of investment in local assets—namely invest only if the social rate of return is in excess of 5%.
Now consider an asset which will produce a constant payment in perpetuity but starting in one year’s time, with zero payment before then. The price of the asset in one year will be unaffected by the tax, but the private saver will discount this value back one year at a rate of 3% rather than the socially optimal 5%. As a result, the present value will be overstated and there may well be overinvestment. The reverse situation applied for assets whose payments fall over time. The problem arises because of the compounding of interest, which means that taxes on interest imply a larger proportionate reduction in the rate at which values are discounted at longer time horizons than in the near term.
The bottom line of this is that the tax system creates a distortion in which there is an incentive to invest too much for long-term gain, and not enough for shorter term gain.
Now, if payments from an asset are going to predictably increase over time, then so will the market value of the asset—that is, there will be capital gains. Similarly, predictable decreasing payments will lead to predictable capital losses. Samuelson’s very neat mathematics shows that if you tax capital gains and subsidise capital losses at the same rate of tax as is applied to interest and payments from the asset, then the distortion in favour of long-term returns over short-term returns is removed.
So why is this not a strong argument in favour of capital gains taxes: I can see three reasons:
- Samuelson’s result requires symmetric treatment of capital losses and capital gains, and no distinction between realised and unrealised gains. If this is not going to be the system implemented (which real-world capital-gains-taxation regimes ever are), the result does not apply.
- The result is partial equilibrium, dealing only with the distortion across time horizons. But the excess discounting that arises from the magic of compounding is exactly the reason that capital taxation is found to have greater efficiency costs than labour taxation. A regime without capital-gains taxation, may lead to some inefficiency in the time horizon of investments, while enhancing efficiency in the margin between consumption and saving.
- With a 5% before-tax interest rate and a 40% tax rate, it takes a horizon of 25 years before the future is excessively discounted. Do we really need a complicated tax regime to dampen the amount of saving going into investments with payoffs beyond 25 years? Can any proponent of capital gains taxes state with a straight face that this is the problem they are seeking to solve?