I know what's on everyone's mind these days: Incoming Treasury Secretary Henry Paulson. Specifically, why would a CEO of Goldman Sachs want to be (1) Treasury Secretary, (2) in the last two years of an administration (3) that doesn't have the coziest relationship with the Treasury Department?
My colleague Marty Ginsburg has two possible explanations: (1) perhaps he's public-spirited?, and (2) section 1043.
Here's what §1043 of the Internal Revenue Code (26 U.S.C. §1043) says (my translation into English):
Suppose you take a position in the Executive Branch. To comply with federal conflict of interest rules, you have to get rid of some of your property — say, stock in companies that do business with the Treasury. Normally, you would have to sell the offending property — and pay tax on your capital gain. This could come out to a hefty chunk of change for people who have a lot of their wealth tied up in unrealized capital gains (i.e., assets that have gained a lot in value since they acquired them). But thanks to §1043, you get to instead use your capital gain to buy new, non-conflicting property, for instance a diversified fund approved by the Office of Government Ethics, tax-free!
Section 1043 was passed in 1989, in the days of Bush, Sr., apparently to make wealthy people more willing to go into government. Now I know what you're thinking: Can I, too, save by becoming Treasury Secretary? To answer your question, let's work through an example. Suppose you bought (or got as compensation) stock for $100 million. We law types call $100 million your "basis." Now it's worth $500 million. If you sell the stock, you realize a capital gain of $400 million. Normally, when you sell the stock, the gain is also "recognized," i.e., taxed. Section 1043 allows you to avoid recognition if you buy $400 million of other stuff.
The tax code is clever, though. Your $400 million unrecognized gain sticks around, and goes to reduce your basis on your new property. Even though you bought your $400 million of stock at a price of $400 million, your basis is now considered to be $0. One day, if you sell your new stock, you'll still have to pay tax on that $400 million (plus whatever extra gains you've had since then). So the recognition of your capital gain has only been delayed, not avoided.
Bottom line: Suppose you're perfectly happy with your portfolio, and (rather than holding it until you die, when you'll get some beneficial tax treatment) plan to sell some day in the future. If you become Treasury Secretary, you can rejuggle your portfolio, but when you sell in the end, you haven't saved any money. So what's the point? I suspect that, if you're the CEO of Goldman Sachs, you're not perfectly happy with your portfolio. In particular, you're heavily invested in Goldman Sachs stock, and would prefer to diversify. Even if you think Goldman Sachs will do fine, you're risk averse and would rather not have too many eggs in one basket. But to diversify is to sell, and to sell is to pay tax. So if you're in that sort of position, becoming Treasury Secretary lets you do what you'd like to do — diversify — without getting penalized.
More seriously, talking about §1043 allows us to talk about the "lock-in" effect of capital gains taxation generally — i.e., why do we have a system that taxes transactions, and thus discourages efficient portfolio reshuffling and diversification, rather than taxing capital gains as they happen? See, e.g., George R. Zodrow, Economic Analyses of Capital Gains Taxation: Realizations, Revenues, Efficiency and Equity, 48 Tax L. Rev. 419, 467 (1993). Why can't we all be Henry Paulson? But that more serious discussion is a topic for another post.
Thanks, Marty! (Cross-posted on Georgetown Law Faculty Blog. See also this New York Times article, Paul Caron's discussion on TaxProf Blog, and this Bruce Bartlett op-ed.)
UPDATE: Preliminary notes on the serious question at the end of the main body of this post -- why not tax capital gains as they happen? (This is called "marking to market.") The main reasons we have a "realization rule" are administrability -- once you sell, you know how much you've made -- and liquidity -- once you sell, you definitely have the cash to pay the tax. Check out the excellent comment by commenter nc3274, going into more detail on potential problems with "mark to market."
You'd have to sell your business in order to pay the capital gains tax on your business.
Somebody thinks that long term investment is very good and short term investments are bad. They have decided to reward holding and punish selling (maybe in the Keynsian hope of using government to soften recessions). They also tax long-term capital gains with a lower rate than short term capital gains.
(I'll add that Marty's class was by far the toughest class I took in law school.)
Paulson, a billionaire, thinks that by being Treasury Secretary for two years his influence on economic policy will save him more than he could earn working at Goldman.
I'm not claiming he's going to be unethical, just that he's nervous about the economy and doesn't want some idiot in the job.
In other words, it's the diminishing returns of money at work. To most people, making one million dollars in obscurity seems preferable to making $100,000 in a position with a lot of prestige but that view is probably reversed for many people who already have $100 million.
Because then the government would have to stump up refunds any day the stock market went down.
2. Ron Hardin: True, taxing gains as they happen would have a liquidity problem. At least, when you sell stocks, you have cash to pay the tax. This might be enough of a reason to keep the current system. But, there's an easy enough workaround: If your stock appreciates and you have no cash, sell some of the stock to pay the tax. This is similar to the death tax: You might inherit a home and have to sell the home to pay the tax; but with stocks you probably don't have a sentimental attachment to them. Plus, it's a minus to distort things so people prefer to have their wealth in illiquid (non-taxable) form rather than (immediately taxable) cash.
3. Cornellian: True, he's rich enough that he could take a low-paying prestige job. That's where considerations (2) and (3) from the top of the post come in: This is the end of an Administration that doesn't really listen to Treasury. Moreover, CEO of Goldman Sachs is really prestigious. Not that I deny that there's a certain cachet to being a cabinet secretary -- but it's not an open-and-shut case as to whether he's going up or down in prestige.
On the broad question, I'd put the troubles with a "mark to market" system into three basic buckets (setting aside whether Eisner v. Macomber creates a constitutional realization requirement, which seems spurious at this point):
(1) Administrability. The system would require asset valuation at the end of each year. With liquid securities markets, that's pretty easy, but with privately/closely held assets, there's a lot of potential for manipulation. And limiting mark to market to liquid securities means that they would be tax-disfavored relative to illiquid assets--the "minus" identified above. Same goes for taxpayers facing liquidity problems--it's easy to sell off 1000 Vanguard shares to pay the tax, but much more difficult to sell 20% of your wholly-owned business. Not to mention that you'd recompute basis every year.
(2) Volatility. Presumably we'd keep the end of the year as the "date of record" for determining gain or loss. That could then result in exogenous market shocks increasing prices at the end of the year, triggering gains that only last a short time, but are taxable in year N and offset by losses in year N+1 (and that assumes the capital loss limits are jettisoned). This is particularly true for non-USD denominated shares, the dollar value of which rise and fall with currency values and share price movements. Again, limiting mark to market to USD shares would make shares tax-disfavored, which seems like strange US policy.
(3) Manipulation. I haven't given any thought to how one would manipulate a mark to market rule, but some seem easy--go ex dividend on December 28, 2007 and pay the dividend Jan 2, 2008. Shareholders compute gain/loss on the ex dividend stock price as of the end of 2007, but don't have taxable dividend income until 2008. This could get largely priced out by market participants, but again who wants the distortion?
Note that there are instances of mark-to-market, but they relate to people who can easily deal with it (i.e., dealers).