Today, the Senate's Permanent Subcommittee on Investigations is holding hearings about how some hedge funds, led by Renaissance Technologies, and with the assistance of some banks like Barclays and Deutsche Bank, may have evaded some $6 billion in taxes by using basket options. Fine. But the first thing to do about Renaissance Technologies' use of basket options to reduce taxes is to admire it aesthetically. It is so simple, yet so lovely. Here:
- Renaissance puts some money, say $1 billion, into a pot.
- Renaissance's prime broker, Barclays or Deutsche Bank, also puts some money, say $9 billion, into the pot.
- Renaissance manages the money in the pot, buying $10 billion worth of stuff and trading in and out of it rapidly: "The overall composition of the securities basket changed on a second-to-second basis."
- In a little over a year, the prime broker takes back its $9 billion, plus some extra money for its troubles, and Renaissance takes back whatever was left, the $1 billion it had put in, minus the money that the prime broker takes, plus any profits (or minus any losses).
That's a neutral description; that's just a thing that happened. Renaissance put some money into a pot, its prime broker put a lot more money into the pot, Renaissance managed the pot, and later the prime broker got its money back and Renaissance had the upside and the risk.
But the magic is that, in life -- in finance, anyway -- dreary reality imposes few limits on the power of the human imagination. There's a thing that happens, and there are words you use to describe it, and that thing and those words need to have some connection, but only a dullard would think that each thing has one single never-changing set of words appropriate to it. The joy is in trying out different sets of words on the same thing and seeing if they fit.
If you are a normal person with some financial understanding, you will read that list above and instinctively supply some magic words:
- Renaissance is putting equity into its account at the prime broker.
- The prime broker is providing financing, or margin, or leverage, or a loan, against Renaissance's account.
- Renaissance trades its account, subject to portfolio margining.
- At the end of the year, the prime broker's loan is repaid with interest and fees, and Renaissance keeps the levered upside or downside in the account.
Those are the normal magic words of prime brokerage, and their operation is straightforward and well understood. Everyone knows the effects that those magic words have. That's mostly good: Basic mechanical things such as how much this should cost, how it should be documented, what happens if the pot loses money instead of making money, etc., are all straightforward and predictable.
But those magic words also have some bad side effects, for Renaissance. For instance:
- Federal regulations limit how much leverage a prime broker can provide in a margin account; the actual limits are subject to their own subsidiary flavors of magic, but let's just say they top out at well below the 10 to 1 in my example above.
- Tax law says that if you trade in and out of securities rapidly, they are subject to the short-term capital gains rate of 39 percent, instead of the long-term capital gains rate of 20 percent (previously 15 percent) applicable to positions held for longer than a year.
- Tax law also requires withholding of 30 percent of dividends paid to people outside the U.S., including non-U.S. investors in hedge funds.
And if you are a deeply insightful magician at Renaissance, or at Deutsche Bank, you think: Is there a way to use different magic words to get better effects? Just words, mind you. That list of things that happened would still happen. The actual substance -- funding a pot, levering the pot, managing the pot, taking the profits -- is actually really important to Renaissance. You can't change the substance. But what if you muttered different incantations over the substance?
Here's the incantation they came up with:
- Renaissance pays $1 billion of option premium to the pot.
- The prime broker owns the pot, which is a subsidiary of the prime broker, and puts in $9 billion of its own money. But it writes a 3-year American call option on the subsidiary's performance to Renaissance.
- The subsidiary hires Renaissance as an investment manager, for a small fee. Renaissance advises the pot on how to trade.
- In a little over a year, Renaissance exercises its option. Its payout on the option is the upside above $9.X billion, where $9.X billion is basically equivalent to the prime broker's initial leverage plus interest and fees.
Why does this incantation work? Well, there's no margin rule that limits the leverage you can get on an option, so you can get 10 to 1 -- actually it got as high as 16.8 to 1 -- rather than 6 to 1 leverage. And if you have a three-year option and wait a year to exercise it, you have long-term capital gains on the option, rather than short-term capital gains on the underlying trading, so you save a lot -- $6 billion? maybe? -- in taxes. And if the payout on an option adjusts for dividends, there's no tax withholding on that adjustment (or those dividends), so you keep 100 percent rather than 70 percent of your dividends.
Alas this frolic had to end. In 2010, the Internal Revenue Service looked at it and said, come on, the right words are the margin-account words, not the option words, so you have to treat this as a short-term capital gain. The IRS and RenTec are still fighting over that. Obviously the Senate subcommittee takes the IRS's side.
This fight is in some deep sense uninteresting. It is boring and unimaginative to say that this thing that happened "is" margin financing, or "is" a derivative: It's a thing that happened, and there's no a priori reason to think that you can only characterize it one way. On the other hand, I mean, this thing really doesn't look like an option. Two things that most people would probably say about options are that:
- they cost money -- "premium" -- beyond just a time-value-of-money cost, and
- their value changes over time in a way that does not strictly mirror the value of the underlying asset.
Here it seems like Renaissance paid minimal option premium, beyond financing costs, though frankly the Senate report is not very clear on this. And it seems certain that Renaissance's option value had a 1-to-1 relationship with the value of the underlying portfolio, or, at least, that everyone acted as though that were the case. So it doesn't look especially option-y to me. The Senate report calls it a "fictional derivative," and that's fair enough. But then, isn't there a little bit of fiction in every derivative? Every derivative is a work of imagination.
Anyway, after the IRS cracked down, RenTec's prime brokers did two wonderfully illustrative things. Deutsche Bank "worked to develop a new product that would not raise the same tax problems." This product was code-named Project Dawn, and, well, let's let them describe it. From the Senate report:
In a discussion between Satish Ramakrishna, then Deutsche Bank’s risk management head, and Anthony Tuths, a former member of the Tax Structuring and Planning Division at Deutsche Bank, Mr. Ramakrishna raised the concern that under the new proposal, RenTec would not receive back any portion of its premium. Mr. Tuths responded: “That’s the result of having a real option,” in contrast to the previous transactions between the parties. In an interview with the Subcommittee, Mr. Ramakrishna said that RenTec had indicated that it was not interested in Deutsche Bank’s proposed transactions because they did not closely track the performance of the trading recommended by RenTec’s algorithm. Ultimately, RenTec rejected both proposed transactions.
Like I said: You want to do the thing you were going to do anyway and apply the most useful magic words to it. You don't want to do a different thing. That's right out. Deutsche ultimately ended up doing the same transaction it was doing anyway, but making the option's term less than a year, so that it could no longer be used to create long-term capital gains. (It can still be used to get leverage, and I guess to avoid dividend withholding.)
Essentially, the memorandum contended that since Barclays was already at risk, due to previous basket option transactions, it would not “meaningfully increase” its reputational risk by participating in additional transactions, so there was no reason to stop offering the COLT options at that point in time.
Where does this stand? Well, after the IRS memo pushed Deutsche -- and, yes, eventually Barclays -- to stop offering the long-term-capital-gains version of this product, it's no longer really a tax trade. But the IRS is still fighting over back taxes, and this is still a leverage trade, though: It's one of (numerous!) ways for hedge funds to get extra financing on their positions.
Here's the Senate subcommittee's cri de coeur:
More generally, the financial sector and the corporate community are using derivatives to try to achieve a variety of favorable outcomes in accounting, tax, financial, and other regulatory contexts, even when the derivative instruments mimic economic activities that by themselves yield different results. Two examples have been highlighted in this report. Congress and the appropriate agencies should closely examine the growing use of derivatives to circumvent accounting, tax, or regulatory rules, and what steps should be taken to prevent disparate outcomes, particularly when they may pose a threat to the transparency, safety, soundness of our financial system or the economy as a whole.
I guess? This seems so humorless. You can, as I guess the Senate does, think that there are things called "economic activities," and other things called "derivatives," and still other things called "fictional derivatives." But in the world, there are only the things. Someone needs to decide how to name them. If the Senate wants to decide, I guess it's free to. It makes laws and stuff, and laws are good for naming things. But right now, mostly, the banks are the ones handing out the names. And when they name things, they take a certain amount of poetic license.
Also if the pot loses $1 billion before the year is out, the prime broker seizes the pot, sells whatever's left and takes the money; Renaissance gets nothing. This gets a little more complicated in some of the versions of the trade, but nobody seems that fussed about it for this amazing reason: Renaissance never lost money. From the Senate report:
Moreover, in more than ten years of operation, in which the COLT options not only experienced millions of trades as detailed below but also operated throughout the worst financial crisis to affect the market in generations, no knockout event ever took place. In fact, the facts indicate that no losses took place in any year with respect to any COLT option. ...
The accounts of the two largest users of this version of the MAPS options, RenTec and George Weiss, which purchased and held the options from 2000 to 2007, never experienced any losses or knockout events. Instead, every option, every year, provided a cash payment to the hedge fund that exercised the option.
The Senate report gets into this on pages 79-82. The basic rule is 2:1 margin for U.S. equity brokerage positions under Regulation T, but there are ways around that including using "joint back office arrangements" which the Senate report variously says allow for 6.7:1 or 7.6:1 leverage.
Incidentally, there is much sociological interest in who invented this, and how, and why. A rough version, from the Senate report:
- "The first basket option structure appears to have been designed between 1996-1997, by the Royal Bank of Canada (RBC)." This seems to have been a primitive version.
- "A few years later, Deutsche Bank developed and marketed a less costly structure it named Managed Account Product Structure (MAPS). The MAPS basket option product at Deutsche Bank evolved from a similar product called MAIDS, which was initially designed by National Westminster Bank (NatWest). In 1998-1999, Deutsche Bank acquired NatWest Securities, including the team that had worked on the MAIDS transaction and brought over almost the entire group to help Deutsche Bank establish its Global Prime Services business."
- "RenTec purchased its first MAPS option in 2000, followed by additional options in 2002. Later, several Deutsche Bank employees who worked with RenTec on the basket option structure moved over to RenTec, including operations personnel, management, and legal counsel."
- "In 2002, RenTec introduced the basket option concept to Barclays. According to RenTec, Barclays initially approached them to expand Barclays’ U.S based business. RenTec told the Subcommittee that, because RenTec wanted to spread its counterparty credit risk with multiple banks, it agreed to do business with Barclays, but only if the bank could develop a basket options type structure."
- "The first basket option structure appears to have been designed between 1996-1997, by the Royal Bank of Canada (RBC)." This seems to have been a primitive version.
The Senate gets very boringly exercised about how this fee is much smaller than the hedge-fund fees you'd actually pay Renaissance for managing your money. (You can't pay those fees anyway, though, since Renaissance is closed to outside investors.)
There is a somewhat sterile, but important for derivatives geeks, discussion on this point. Basically Renaissance describes this not as "Renaissance tells the pot what to purchase" but rather "Renaissance picks a reference portfolio and the pot can hedge however it wants." The Senate committee is very skeptical of this; see footnote 41 of the report. I am more sympathetic, because if you're just uttering magic words why not utter these words? But, sure, in practice, Deutsche "hedged" the "reference portfolio" by buying exactly the same positions in exactly the same quantities, and it is basically unimaginable that it could have done otherwise. From the report, it sounds like everyone at DB and Barclays, when asked about the idea of two separate positions -- the "synthetic position" on which RenTec had an option, and the "hedge position" that DB used to hedge its option risk to RenTec -- sort of scratched their heads and looked puzzled.
I mean, not literally for free. You had to pay lawyers. And your bank, which is pleased with itself for coming up with this idea, will probably charge you a premium fee for it. And demand a tax indemnity, as it happens.
Ooh incidentally there's a long boring story on pages 54-59 of the Senate report about how Barclays wanted to get its version of this thing off its balance sheet, which is kind of indicative. You're always optimizing on a bunch of axes -- tax, margin rules, economics -- and in recent years there's been much more of a focus on optimizing bank balance sheet. That shifts the calculation on trades like this; renting balance sheet to improve client tax outcomes used to be more attractive than it is now.
Technical appendix to that paragraph. First, on tax law, there's a vague but robust set of doctrines -- on "constructive ownership" and "substance over form" and so forth -- that inform the discussion. Look at pages 20-23 of the Senate report. In particular, Section 1260 attributes "constructive ownership" of the underlying positions if you have certain delta-one derivative positions. A rule of thumb that I have heard from tax lawyers, though I cannot cite it, is that an option with a delta above 90 percent runs constructive-ownership risk under these rules.
Second, on premium. There is some option premium here, because the banks take the gap risk that the basket's value will fall below 90 percent (in my example -- the actual knockouts were even higher in some cases; if the portfolio lost 6 or 7 percent of its value RenTec would eat through its equity), and they'll have to liquidate it at a loss. According to the Senate report, Deutsche had some liquidation rights before the basket entirely ate through margin, while Barclays apparently did not. A 90-strike put, even on a fast-changing basket, has some value, so Barclays seems to have been selling real options (whether or not it charged for them). It's less clear if Deutsche was, since the report is not clear on when the liquidation rights kick in. In any case, since RenTec never lost money, the realized value of these options was always zero, though that doesn't prove their initial value was always zero.
Third, on delta. The bare fact is, Renaissance had an "option" on a pile of stuff, call it X, and the prime brokers hedged that option always and only by buying X, in the precise amount that the "option" referenced. However you model that option in your mind, it was hedged by the prime brokers as a delta-one option. That makes it delta-one in my book, though my book is not the IRS's.
The Senate report has a bunch of other arguments, some of which are convincing, some of which are just mush ("Options and other derivatives usually require the participants to make careful calculations about how specified financial assets will perform ...").
To contact the author on this story:
Matt Levine at email@example.com
To contact the editor on this story:
Zara Kessler at firstname.lastname@example.org