You can think of a margin loan as being like an option on the underlying security: if I lend you $50 (nonrecourse) against a $100 share of stock, and tomorrow the stock is worth $45, then you've lost $50 and I've lost $5, same as if I wrote you a $50 strike put option on the stock.1 This isn't quite right - margin calls, etc. - but what it lacks in precision it gains in tax efficiency:
James H. Simons, who became a billionaire when he turned his extraordinary mathematical ability from defense work to investing, has deployed an unusual strategy at Renaissance Technologies LLC to skirt hundreds of millions of dollars in taxes for himself and other investors, said people with knowledge of the matter.
The Internal Revenue Service is challenging the technique, which it called “particularly aggressive,” without identifying the hedge fund in the dispute. ... Renaissance’s strategy involved buying an instrument called a “basket option contract,” from banks including Barclays, the people said.
That's from today's wonderful Bloomberg article about the IRS's investigation. Here's the IRS memo about the trade. Here's the trade.2 Actually wait: here's the trade, twice. You can just read down the left side if you enjoy getting mad at evil tax-dodging hedge funds, or just read down the right side if you don't want to believe that Jim Simons could ever get up to no good:
You might ask: who cares if you call this thing a "margin loan" or an "option"? Well, there's a key feature, which is that Renaissance actually gets to manage the basket: it can buy or sell whatever stocks it likes, as often as it likes, subject to some concentration/etc. limits, and its option becomes an option on whatever ends up in the basket.3 That is ... not how options work? If you make a bet that your favorite team will win the World Series, the bookie doesn't let you change your mind about who your favorite team is until the final out.4
The trick is that this transforms Renaissance's high-frequency trading in and out of stocks - in, call it, a margin account at its prime broker - into Renaissance buying and holding a two-year option on ... a basket of whatever stuff it wants to buy and sell each day. Trading every day gets you short-term capital gains, taxed at a 35-39.6% rate; holding for longer than a year gets you long-term capital gains, at a 15-20% rate. Renaissance, it appears, has been claiming that it owned only the option, not the stocks, and since it held for longer than a year it only owes long-term capital gains taxes. That is obviously absurd, but it is also cleverly absurd, and the tax code does reward cleverness.
Nonetheless, you might not be surprised to learn that the IRS is pissed. The IRS is pissed, because it knows that this shouldn't work, though it can't quite get its head around why.5 This isn't a long-term option: this is frequent trading in a margin account, subject to a margin loan, and wrapped up in some words that make it a long-term option.
We've talked before about synthetic prime brokerage: the practice that replaces "hedge fund buys securities and borrows against them from a prime broker" with a near-equivalent "prime broker buys securities and writes delta-one-ish derivatives on them to hedge funds." The idea, pretty much always, is to create some sort of regulatory efficiency, or - preferably - several sorts all at once. So synthetic prime brokerage can be less capital-intensive for the PB banks, and it can avoid transaction taxes, or dividend tax withholding, or margin rules, or various other unpleasantnesses that might afflict hedge funds forced by humorless bureaucrats to actually own the stocks that they own.
This story kicks it up a notch, providing an example of how synthetic prime brokerage can transform the very nature of time itself: Renaissance and its prime brokers invented a derivative that transformed hours into years. It's a pretty silly derivative, but it has a certain loveliness. And, perhaps, applications beyond tax avoidance.
One of my favorite things here, incidentally, is this defense of it (from Bloomberg):
Some of the people with knowledge of the option trades said they had at least one purpose unrelated to tax savings: they allowed Medallion to borrow more money from its banks than it otherwise could.
Get it? Margin loans are typically not allowed to be 90% LTV so if you do want to borrow 90% of the value of your portfolio, you need some sort of structure other than "we took out a margin loan."6 It doesn't need to be this structure but it needs to be something, and this structure comes with the tax savings. (Maybe!) Similarly, tax shenanigans always sound better if they come with some non-tax purpose. Here, the way Renaissance got around margin rules was with a tax trade, and the way Renaissance defends the tax trade is by saying "well it's not just a tax trade since helped us get around margin rules." How could you object to that?
1.Or: same as if I owned the stock and wrote you a $50 strike call on it. Anyway, cf. repo to maturity vs. CDS.
2.The IRS's description is pretty clear:
HF [Hedge Fund, viz. Renaissance] entered into a contract ("Basket Contract") with Foreign Bank (“FB”), which is a U.K. public limited company. The Basket Contract was styled as a call option on a basket of securities (“Reference Basket”) held in a specified prime brokerage account administered by FB. The value of the securities in the Reference Basket was $10x when the parties entered into the Basket Contract, and the Reference Basket was funded with $1x in “premium” paid by HF and $9x paid by FB. FB determined HF’s $1x premium through its finance department, rather than through option valuation formulas typically used when pricing standard options. HF had the right to terminate the Basket Contract at any time during a two-year term and receive a specified “Cash Settlement Amount,” which was based on the performance of the Reference Basket.
The Basket Contract provided for a strike price equal to the initial value of the Reference Basket ($10x). The Cash Settlement Amount that HF was entitled to receive upon termination of the contract equaled the greater of (1) zero or (2) the reimbursement of the $1x premium, plus “Basket Gain” or less “Basket Loss.” ... Basket Gain or Loss fully reflected all of the net economic return or loss on the performance of the Reference Basket, including the financing charges on $9x. ...
The Basket Contract contained a “Knock-Out” provision that automatically terminated the contract at any time that Basket Losses reached 10%, which was the same amount as HF’s initial premium investment. Thus, if Basket Losses breached the Knock-Out barrier, the contract would terminate and HF would receive a Cash Settlement Amount of zero. FB also had the right to require HF to enter into risk reducing trades even before losses in the Reference Basket reached the 10% barrier. Consequently, HF bore the risk of loss, dollar-for-dollar, for Basket Losses up to the amount of its investment.
3.ACTUALLY Renaissance gets to specify what's in the "basket" (for option settlement purposes) but not what the prime broker actually buys to "hedge" its "option," so you could imagine ALL SORTS of front-running/fading/whatever shenanigans (LIKE: "our delta on the option is 110," etc., if you think Jim Simons is good at trading, which: you should). But, naah:
Although not contractually obligated to follow GP’s specific trading instructions as long as the net value of the Reference Basket nevertheless reflected GP’s instructions, FB in fact executed all of GP’s trading instructions, which could entail numerous trades per day.
Actually I guess that net value thing would be hard to game.
4.Though that's kiiiind of how CDOs work. Also I'm sure there are some legit basket options that work closer to this than I'm implying.
5.The bulk of the IRS memo is spent explaining why this shouldn't be treated as an option for tax purposes and I am aesthetically unimpressed by it. It's a lot of blather:
HF’s control over the Reference Basket caused the Basket Contract to operate unlike an option. As explained by the court in Saviano, an option provides one party with the choice of accepting an offer, while the other party is obligated to keep the offer open for a specified period of time. Options on property allow the holder to accept an offer buy or sell specified property at defined price. In this case, the Basket Contract purports to identify the Reference Basket as specific property subject to an option, yet the IMA contradicts that characterization by allowing HF (through GP) to alter the Reference Basket while the Basket Contract remained open. HF’s ability to trade component securities within the Reference Basket calls into question whether the Reference Basket constitutes specific property apart from its components; thus, rather than having the right to buy the Reference Basket, HF could be viewed as having a series of separate contractual rights for each security within the Reference Basket such that each trade HF executes while the Basket Contract is open would generate a taxable event attributable to that trade under sec. 1001. FB permitted HF to have this control because the terms of the Basket Contract ensured that FB was protected from HF’s investment decisions; as noted above, the Basket Contract imposed potential costs upon HF that were more consistent with a party that had an obligation to buy than upon a party with a mere option to buy.
Is there not a more elegant way to put that? I dunno. Like, "a call option on whatever you're thinking about in two years" is not a call option. Anyway, though, it's blather, but the cumulative effect is persuasive. This is some bullshit!
6.I know, I know, hedge fund portfolio margining is more complicated / more permissive than that, whatever.