Interest rate swaps

The Bank for International Settlements reports that interest rate swaps are the largest component of the global OTC derivative
market. The notional amount outstanding as of December 2006 in OTC
interest rate swaps was $229.8 trillion, up $60.7 trillion (35.9%) from
December 2005.

That’s from Wikipedia.
You’ll see other estimates as well, although they fall within a few
hundred trillion of this number.  If it makes you feel any better, swap
numbers usually measure the total liabilities in the market, not the
size of the swapped payments.  So you could argue that "the real
number" is maybe 1/20th of this or so, with error margins of only
trillions remaining.

Oddly economists don’t have a clear explanation for swaps.  In a
classic "plain vanilla" swap you trade a fixed rate interest payment
for a floating rate payment and of course the swaps occur across
currencies as well.  So here’s a typical story: Bank A takes out a
floating rate loan in terms of Swiss francs (from C) and Bank B takes
out a fixed rate loan in terms of Japanese yen (from D).  Bank A and
Bank B then decide they each would rather have each others’ liabilities
and so they swap interest payments.  That’s called the comparative
advantage theory.

But why didn’t Bank A borrow in yen from D to begin with?  And why
didn’t Bank B borrow in Swiss Francs from C to begin with?  OK, they
"changed their minds."  Is that how you get to all those trillions?

Or maybe lender D didn’t trust Bank borrower A in the first place
and would have charged an excess risk premium.  But then why does Bank
B trust Bank A so much? 

Is there a regulatory arbitrage argument here?  Under Basel I, a
bank might prefer to get a non-risky loan off its books to avoid the
associated capital requirements.  Clearly that drives some of the
market but regulators have been working on
remedying that problem and no one was expecting the swaps market to
disappear as a result.  Furthermore the interest rate swaps predated
the Basel agreements.  Another regulatory arbitrage argument cites the
difference between the U.S. and Eurodollar markets.

Here is one survey of explanations for interest rate swaps.  The explanations mostly seem lame and question-begging to me.  Here is another survey of potential explanations of interest rate swaps.  Good luck and I hope you have JSTOR access.  Here is a useful non-gated summary.

It is a shame that economists have devoted so little attention to
understanding interest rate swaps.  It’s hard to get the data for doing
first-rate quantitative finance work, so the topic tends to be
ignored.  Right now it would be nice to know how much of this market is
real gains from trade and how much is a zero- or even negative-sum game
of some kind.  I believe that practitioners have a better sense of this
than do the academics, myself included.

The bright side is that — as far as we’ve been told — this
massive, unregulated interest rate swaps market has not been a major
driver for troublesome counterparty risk.  The credit default swaps
have been the culprit there, in part because those latter markets are
based on large, discrete default events, which kick in quickly and
require very large surprise payments.

Comments

Why is a currency swap called a currency swap and not a repurchase agreement? That's confused me for a while.

Many corporate borrowers sign credit agreements at floating rates and simultaneously swap to fixed rates. The glib explanation is that banks prefer floating rates due to match their funding and borrowers prefer fixed for certainty (lower risk).

I don't think this is a very hard question. In the example given, that Bank B is likely headquartered in Japan. The only people who can understand the bank well enough to be willing to lend it money are Japanese, and the only money they have to lend is yen. But Bank B doesn't want all its liabilities to be in yen (because it has non-yen assets), so it does the swap.

More generally, financial institutions generally have preferred access to certain kinds of assets, and to certain kinds of liabilities, but those assets and liabilities don't always match, so they do various kinds of swaps. In real estate lending, for example, institutions have easy access to floating rate liabilities, priced at LIBOR (if you want to be more sophisticated, you would say that LIBOR approximates the bank's marginal cost of funds), and they have a large pool of customers/borrowers who want to borrow at fixed rates. So either the bank borrows at LIBOR, does a swap, and makes a fixed rate loan, or the bank makes a floating rate loan and tells the customer to buy a swap.

I agree with y81. Here's a rewording of the question in more concrete terms: Let's say Joe Consumer, A, wants to buy a box of cereal from grocery store B. But first A goes to the bank, C, deposits a check and withdraws some cash from his account. And first B goes to wholesaler D, and borrows boxes of cereal. If B just wanted money so badly, why didn't it just go straight to the bank A and trade its boxes of cereal for money? If A just wanted cereal so badly, why didn't A just didn't take his check to D?

I fully agree with the point that this is a form of arbitrage (and I do a very small amount of this for clients, intermediating for those who can't easily access the swap markets). Some lenders do not like to lend in fixed, and as a result some borrowers cannot access the right type of funding (typically fixed rate longer term). (the same applies to currency). Other banks have the best access to the right type of funding (i.e. from the market), but may not be comfortable with the credit risk of smaller clients - but could be comfortable with a large notional swap exposure (which may be considered to be a very small equivalent credit exposure because of netting/offsets - like 2-3% or even less.) Or they may provide it with the help of some intermediaries...

The key part about why the numbers sound so astoundingly large is for two reasons: a) the notional exposure (say, a $100 million swap) has a small amount changing hands at each settlement date, and (in theory) they net, so the notional amounts don't mean much; and b) because swaps don't cancel each other out (although they may effectively match in terms), the amount outstanding increases every time one counterpart enters into another swap. This is why you get what looks like crazy exponential growth.

Imagine if, on the stock market, instead of each trade being settled in five days or whatever, each trade was an agreement to exchange the paper five years from now. So if I bought something today and sold tomorrow 100 shares of Microsoft, and my counterparty did the same, the notional number of underlying "share contracts outstanding" has doubled. (Each side of the trade has gone from 100 shares to 200 - but with two new counterparties). If this goes on every day for 30 days, you now have 100*2^30 underlying notional contracts outstanding. They don't get netted/settled until five years from now.(Please note, I'm not intending to get this exactly right math-wise, just illustrate). I see only my trade counterparty (citibank, say), so I'm not worried, and so on down the line. Citibank, however, probably never wanted to hold/sell the stock, so concluded (originally) a back-to-back contract with someone else (meaning more notional contracts outstanding).

This is why there is so much talk of getting these on an exchange, where the trades get 'netted' and the numbers are less crazy.

Of course, on that settlement day, you might have some complicated exchanges of shares, particularly if any of the counterparties disappears. And in practice, a relatively small number of banks concentrate the contracts, they're staggered, and there may be some cancelling out internally and a lot of other stuff happens.

Tyler, I am sorry that I am going to sound like some cracked record that keeps on repeating the same line, but you entirely fail to understand the logic behind derivatives - sadly, a feat not uncommon among people primarily trained as economists and which has long kept me wondering.

Anyway, the aim behind interest rate swaps is to get FUNGIBLE discount factors. Again, as I pointed out to you a couple of days ago, derivatives exist only because the underlying has not the necessary characteristics to be easily tradable. For often-used instruments, like an interest-rate hedge, easily tradable means fungible. Normal interest rates instruments are not fungible. That is why we have occasional topsy-turvy periods like for instance the 1998 LTCM blow-up or the current mess, when all "normal" (ie "usual") arbitrage relationships break up and instruments drift apart.

Oh, and plain-vanilla interest rate swaps are all-purpose hedging instruments. So there is no use to try and build "you have two cows"-type stories, as you do, to try to explain their purpose. Comparative advantage? Please, you must be joking. This is about RISK TRANSFER. You heard about risk transfer? No? Well, Google says you never used the expression in all your posts. As it is what capital markets are for, primarily, maybe you should get acquainted with the blasted notion...

Sorry about the briskiness/forcefulness of the tone. I just want the message to come across. Regards. Henri

Tyler,

I only worked with a very small subset of swaps in the affordable housing bond market, but from where I was standing swaps were used to try to capitalize on low variable rates and the bet that the variables rates stayed that way.

Banks don't like to issue or back long term variable debt because its impossible to baseline your loan to value ratios going into the future, and also therefore your estimated probability of refinancing at a given point in the future.

At the same time, the rate on variable rate debt was lower than the similar rate for fixed debt and so the total 'stack' of the variable rate plus the fee charged by the swap provider to turn the rates into fixed rates were still cheaper than the fixed interest rates.

As to why that differential existed in the first place, i'm less sure about but I'd imagine that a lot of swap and cap derivative instruments were accomplishing similar things.

Converting from fixed to floating is kinda of important, but the reason the market is so massive is that it's the most popular tool for hedging out duration. So, almost all trading in fixed income requires both sides to enter into a swap to hedge out the new duration in their books. Also, if the position has any convexity to it, swaps are often used to delta hedge the position.

Why are swaps more popular than treasuries for hedging duration?
1. They don't take up balance sheet. This is huge.
2. There are less technical problems and no need to roll swaps every couple of months (either to the on-the-run treasury or with treasury futures).
3. Typically, swaps are a better hedge than treasuries. One way to think of this is that banks fund at libor not treasuries, so they will discount cashflows with libor not treasuries.

Oops, typo! Above, instead of "If desk A inside a bank LENDS a large sum of money wholesale in one gulp on Monday" read "If desk A inside a bank BORROWS a large sum of money wholesale in one gulp on Monday". Sorry.

Henri: Your IBM and World Bank example is an absolutely cannonical example of comparative advantage. So I don't quite see how it's supposed to support your thesis that "the notion of comparative advantage has nothing to do with swaps".

Henri: Thanks for the clarification.

I don't really see how the existence of intermediary financial institutions that warehouse swaps changes the underlying dynamic. In the world of "normal goods," wholesalers also warehouse goods to ease counterparty-finding, bridge desired transfer times, and reap economies of scale. But it is still comparitive advantage between the ultimate counterparties that drives the trades. Are you claiming that for modern swaps are different?

You mention risk transfer, but I don't see how that's different from comparative advantage. Isn't differing risk tollerance, or tollerance for differering kinds of risks, a form of comparative advantage?

Comments for this post are closed