Interest Rate Swaps Turning Harvard Crimson

Bloomberg has another article about Harvard’s financial bets gone awry, with a focus on interest rate swaps. As I’m closely tied to Harvard and traded billions of these financial instruments, I thought I might as well weigh in.

First, some background. Interest rate swaps themselves are a derivative. In fact, by overall notional (face value) outstanding, they’re the most common derivative. According to the Bank of International Settlements (the organization that has the enviable task of keeping track of this stuff), at the end of December 2008, there was approximately $600 trillion of notional outstanding in the over-the-counter (OTC) derivatives market. Interest rate swaps accounted for about $330 trillion of this amount, more than 50%. And the net present value of these swaps, about $16.5 trillion, was around 50% of the total net present value of all outstanding derivatives contracts (I’ll go into the difference between notional outstanding and net present value a little later for those interested).

Though a swap is classified as a derivative, at a basic level, it’s nothing more than a bond. One party agrees to pay a fixed interest rate tied to a notional, and the other party agrees to pay a floating rate in return for the fixed payment. Usually, the floating rate is linked to LIBOR, the London Interbank Offered Rate (a daily rate at which large banks make unsecured loans to one another).

Since this can get abstract pretty quickly, let me give some examples that should shed some light on how Harvard lost money (Note: I myself wasn’t involved in any of the Harvard transactions but am well aware of what they did). Suppose Party A and Party B have different views on interest rates at the long end of the yield curve: Party A thinks rates will rise, and Party B thinks they will fall. So, the two enter an interest rate swap, whereby for 30 years, Party A will pay 4% of $1 billion notional to Party B every six months (i.e., 2% per half year) and in return Party A will receive LIBOR on $1 billion every three months. For now, assume that 4% is the fair 30-year swap rate, meaning that the current net present value of this transaction is $0 for both parties. This is to say, the net present value of the expected quarterly LIBOR settings in the future will exactly offset the semi-annual payments of 4% on $1 billion.

Let me just summarize the terms of this swap for your reference.

Notional: $1 billion
Maturity: 30 years
Fixed Rate: 4% (Party A pays, Party B receives)
Floating Rate: LIBOR (Party A receives, Party B pays)

Because we’re assuming 4% is the fair rate today (and parties generally only enter into swaps at the fair rate), neither Party A nor Party B has to pay anything to enter into the contract. But, because interest rates move, the swap will not be worth $0 for more than a few seconds. To be more precise, when interest rates rise, Party A will make money and Party B will lose an equal and opposite amount. Conversely, when interest rates fall, Party A will lose money, and Party B will gain.

To see why this is true intuitively, it helps to view the matter from one party’s perspective (and then apply the opposite logic to the other party). I’ll choose Party B. As Party B is currently receiving a fixed amount, when rates fall, the net present value of what Party B will receive rises because the payments are being discounted at a lower rate. In this way, you can think of Party B as being long a bond, and perhaps you remember the basic principle that when interest rates fall, bond prices rise and vice versa. Actually, despite all the complicated terminology, there’s no need to think of anything other than a bond: Party B is long a bond, Party A is short; Party B wants rates to fall, Party A wants them to rise.

So, where does fair Harvard fit in?

Like Party A, Harvard was paying fixed on a long-dated swap. It therefore stood to gain if rates rose and stood to lose if they fell. Once again, the basic logic here isn’t complicated. While commentators often speak of innocent counterparties being “fooled” by swaps, even the most unsophisticated parties realize the fundamentals – you pay fixed, you pray rates rise; you receive fixed, you hope they fall.

So, as far as the swap was concerned (and ignoring the possibility of whether the school had anything to offset this swap), Harvard was no doubt hoping rates would rise. And they didn’t. No, they fell hard and fast. To give you a sense of how this precipitous collapse led to hemorrhaging at Harvard, I’ll briefly walk through a trade comparable to the ones put on in Cambridge.

Thinking that rates will rise, on January 2, 2007, I entered into a swap where I agree to pay fixed and receive floating on $3 billion notional for 30 years. The fair 30-year swap rate on that date was around 5.3%, so I had to make payments of roughly $100 million per year (split in two installments) in return for quarterly payments tied to LIBOR. Since I entered into the swap at the then fair rate, the value of the swap to me and my counterparty, Bank A, was $0.

Fast forward to January 2, 2009. A lot happened in the intervening two years – many of Bank A’s competitors went bust, the world’s very financial framework teetered on the brink of collapse, and, worst of all for me, rates fell across the board. The fair 30-year swap on that date was about 2.95%, 2.35% (235 basis points) lower than it was when I entered the agreement. Now, 235 basis points may not seem like a lot, but on a $3 billion 30-year swap, it’s gargantuan.

Roughly speaking (I am taking some liberties to hammer home the basic point but the calculation is essentially correct), in the intervening two years I stood to make about $4.5 million for every basis point the 30-year swap rate rose and stood to lose about $4.5 million for every basis point the rate fell. 235 basis points multiplied by $4.5 million per basis point is more than $1 billion. That’s a fair amount of coin for any place, including a school with an endowment exceeding the GDP of more than 90 countries (even after the fund’s 30% fall!).

If it's not that difficult for people within the university or beyond to see how Harvard could have lost so much money from an interest rate swap, I am somewhat perplexed as to why the institution amassed such a large amount of outright rate exposure. While the endowment is in the business of taking some (and perhaps too much) risk, I can't imagine why anybody at the university would just dump several billion dollars into illiquid long-dated swaps if not to hedge another obligation. Perhaps the university did put on the swaps to insulate against other rate risk. But if the swaps were meant to be a hedge, the institution's net losses suggest the instruments didn't do a very good job.

Indeed, the swaps did a far better job of making breakfast offerings colder, free cookies sparser, and endowment managers' faces redder.