The curve trade | FT Alphaville

The curve trade

A “tempest in a teapot. That’s how JP Morgan CEO Jamie Dimon described the fuss caused by the bank’s Chief Investment Office apparently entering into large credit trades. It may well be teapot-sized, for him. The point for some hedge funds is that even if it was a swimming pool, you’d feel a bit cramped if Shamu joined you for your morning laps.

Which is to say that, for those actually trading credit indices, the thing that is such a big deal is whether the trading behaviour of JP Morgan’s CIO distorted the market. Less of a big deal is whether JP Morgan is going to land itself in trouble if the trades aren’t actually hedges but proprietary bets, hence go against the Volcker Rule.

To discuss the possibility of market distortions, we first need to review exactly what some of the trades might be. There has been a lot of speculation about this, so FT Alphaville would like to guide you step-by-step through the logic of what these trades (probably) are.

If you want a preview or an after-party of the discussion we are about to embark on, please go read the post Matt Levine put on Dealbreaker on Tuesday. In fact, we’re going to start by quoting the same WSJ article that Matt did:

On a conference call with analysts, [J.P. Morgan Chief Financial Officer Doug] Braunstein said the positions are meant to hedge investments the bank makes in “very high grade” securities with excess deposits. (J.P. Morgan has some $1.1 trillion in worldwide deposits.) Braunstein said the CIO positions are meant to offset the risk of a “stress-loss” in that credit portfolio. He added the CIO position is made in line with the bank’s overall risk strategy.

As Matt goes on to point out, having a hedge against a “stress-loss” that involves the indices could well mean having a curve trade. Specifically, a flattener. That’s the trade you’d put on to be bearish, and it would have paid off quite handsomely in the latter half of 2011.

There are different methodologies for putting a flattener on, but we’re going to discuss one that makes the holder indifferent to small moves in the index. The important bit to keep in mind that these are hedges/bets on how the shape of the curve will change.

To get warmed up, here’s what the main investment grade credit index for the US looked like at Wednesday’s close:

Sloping up like that is healthy, whereas an inverted (downward sloping) curve would necessitate tin hats, and maybe chocolate to make ourselves feel better in the face of impending doom.

Step #1 – Select the best index for the job

Let’s say JP Morgan’s CIO, and whatever other bearish investors are around, put the flattener trade on at the start of 2011. If they wanted to use credit indices, they’d have to find one that’s deeply liquid, that’s also likely to stay liquid, and that has more than one tenor (i.e. tradable maturity) meeting these characteristics, as we’ll need at least two tenors to take a view on the how the shape of the curve will change.

Also, we’re going to say it’s safe to assume, given all the reporting, that the trade was primarily in the Markit CDX.NA.IG (North American Investment Grade) index, which makes sense given that the CFO said the “positions are meant to hedge investments the bank makes in ‘very high grade’ securities”.

In the first post of this series we had the below graph. It’s of the net notional amounts outstanding on each series of CDX.NA.IG index. The net notional is a measure of the total economic interest.

As a reminder, the index “rolls” to a new series every six months, meaning that the 125 corporates that are in it can shuffle around, with some companies being removed and new ones going in.

(For info on the “data break” in the graph, see the note at the bottom.)

Looking at the graph, focus in on January 1st, 2011, and tell the person sitting next to you what the most liquid index is likely to be.

The Series 15 is one possible answer, BUT that one was on-the-run, i.e. the newest index, at that time. As we discussed in the first post, when indices stop being the newest one, they become less liquid as market participants roll their positions into the latest index. And indeed the open interest on the Series 15 plummeted once the Series 16 got up and running.

What about the Series 9 then? Being the last index before the financial crisis properly hit, it has high liquidity for structural reasons — it’s baked into a number of structured products.

So if you wanted to put on a curve trade, e.g. flattener, because you are feeling bearish and need a hedge for your high grade securities that you bought with excess deposits, the 9s would be a good series to use, wouldn’t they?

Step #2 – Learning to love curves

By now you should be wondering why the CIO didn’t just buy protection on the index outright to express bearishness.

Well, it may well be that JP Morgan and others did that … but such outright exposure to credit is risky and could be volatile, and even more to the point, there’s the not inconsiderable cost of having such a position.

What one could do instead with the CDX.NA.IG.9s is something along these lines:

  • Buy protection on the 5Y (matures December 2012)** on say $300m notional
  • Sell protection on 10Y (matures December 2017) on $100m notional

With such a trade in place at the start of 2011, credit spreads could improve or deteriorate (by small amounts), and as long as those movements were borne out equally across the curve, the mark-to-market of the overall trade would be zero (to be impervious to bigger movements, the trade would need to have the notionals rebalanced periodically). This type of trade is referred to as being “DV01-neutral”.

When, however, credit spreads start to express the view that in the near term — next year or so — things are going to get ugly before they get better again, the curve starts to flatten out, widening at the short end of the curve more than at the long end. It’s in these conditions that a flattener pays off.

FT Alphaville discussed back in November how anyone who placed a flattener on Italy, even just a few months beforehand, would have made some serious money when the curve did this:

Buying protection on any one point of that curve would have made money, but it would have been expensive to pay the coupons for it. If, however, one had employed a curve trade, it would have also been profitable, but less expensive and less risky. The cost of protection bought at the short-end of the curve could be funded by the payments received for selling protection on the long-end.

Step #3 Confirmation

Back to WSJ for a third time: “Braunstein said the CIO positions are meant to offset the risk of a “stress-loss” in that credit portfolio”.

Pair with what Bloomberg said back on April 9th:

The trade on the index, known as the Markit CDX North America Investment Grade Series 9 (IBOXUG09), probably isn’t a one-way bet, the people said. Iksil may be offsetting the trade by buying protection on the same index with contracts that expire about eight months from now, the people said. That strategy would pay JPMorgan the difference between the long-dated contracts and the short-dated ones, about 47 basis points as of April 6, and the trade would gain when the gap narrows.

If this, and the above logic, is to be believed then JP Morgan may have been hedging credit exposure, particularly of the investment grade variety, with a flattener.

Or, as Matt over at Dealbreaker put it when discussing a trade of this type:

This has the advantage of (1) actually hedging a stress loss in high-grade short-term corporate securities, (2) fitting in with the relative lack of noise in the CIO portfolio, (3) being what people have told Bloomberg Iksil was doing, and (4) being what JPMorgan has actually said it actually did in the CIO during the crisis.

In the next part of our tour will take us into the depths of credit markets to examine the distortions such flattener trades may have caused in the CDX.NA.IG.9. Having Shamu in the swimming pool does indeed cause waves. Ha! Ha! Erm, sorry…


* Data break in the charts. On April 29, 2011, DTCC started splitting each index series into “tranched” and “untranched” hence the series isn’t strictly continuous. After that date, the above charts only contain untranched data unless otherwise specified. It only creates a disconnect at that one point in time, hence not doing separate charts.

** It’s 5.25-years from the launch date of the index, which was September 2007, hence the 5-year will mature in December 2012. It’s still called the 5-year even though it matures in less than a year.

Related links:
JPMorgan’s Voldemort Probably Isn’t That Magical – Dealbreaker
Italian CDS curve inverts (and who made money?) – FT Alphaville