The skirt smoothing! How could you forget the skirt smoothing!

Megan McArdle has a pretty good post up questioning the memo's legitimacy here.

"All tax jurisdictions that I know tax the whole amount, not the unrealized amount. "

I see where you're coming from now. If that's the case then, yes, you're right a wealth tax would increase the take significantly.

Can you imagine teaming this up with a VAT? It would be a tax Armageddon!

I think we may be imagining two different scenarios.

First, I was operating with the understanding that in any given tax period you are only liable for the unrealized gain since the last tax period.

Secondly, I was talking about buying and then selling a particular security/asset and I was assuming away (for simplicity's sake) dividends and re-investments thereof and any tax advantaged accounts like a defined contribution plan.

If that's the case, then the inefficiencies of administering multiple tax collection events are going to reduce the actual revenue take, not increase it.

The limits on capital losses would have to be re-evaluated as well. If I recall correctly, capital losses are currently capped at $3K per year (in the US) although you can carry those forward. Capital gains have no such limit on taxation. At $3K a year, a sufficiently large loss can't be realistically written off in anyone's expected life-span. I'm not sure if capital gains on a primary residence are impacted by this rule though or if it only applies to stocks, etc.

The real problem with the tax proposed in the article is that it is highly inefficient and imposes a significant dead-weight loss without any additional benefit. The poster is proposing a choice between two ways to collect a tax on a realized gain:

1) The current way: tax it when the asset is sold.
2) The proposed way: tax it periodically on the unrealized gain and presumably true-up when the asset is sold.

Both taxes should result in the same amount of taxes being collected (assuming away differences in short and long term rates) but the second method has the disadvantage that the collection of the tax has to be administered at least once and probably many times for a long held asset. This administration has a cost but no actual benefit. At best it is as inefficient as the current method. At worst it's many times less efficient depending on the frequency at which you assess the tax. For small capital gains it's easy to imagine that the cost of collecting the tax in any given time period could eclipse the revenue collected.

The scheme also has the problem that you're also probably going to have to _refund_ unrealized losses. In addition to the cost overhead of implementing refunds you've now impacted actual revenues collected in the assessment time period. Tax revenue gets harder to project/budget and in a period of economic contraction, this will magnify the reduction of tax income.

In the end, the current system is more efficient and a lot more predictable.

Also see: http://en.wikipedia.org/wiki/Wickard_v._Filburn in which the SCOTUS ruled that a crop grown on a private farm for private consumption could be regulated by the federal government.