Back in 2005, New York Times columnist Thomas Friedman’s tome, “The World Is Flat,” laid out ten “flatteners” he claims are increasing global competitiveness. He did not see ahead to the then-looming financial crisis and the subsequent introduction of another key flattener.
In 2010 Barclays Bank plc issued a pair of complementary exchange-traded notes keyed to the shape of the U.S. Treasury yield curve. One was the iPath US Treasury Flattener ETN (FLAT). It was well positioned to attract investor interest when the Federal Reserve launched “Operation Twist,” a swap of the central bank’s inventory of short-term paper for longer-dated maturities.
That intervention was designed to nudge down market rates on notes at the long end of the yield curve while holding the short end fairly steady. Put simply, central bankers got into the business of flattening the yield curve in a big way and FLAT holders were there to reap the profit.
And profit they did. How so? FLAT, and its sibling, the iPath US Treasury Steepener ETN (STPP), track the Barclays Capital US Treasury 2Y/10Y Yield Curve Index from opposite directions. FLAT’s value increases as the interest rate spread between two-year and ten-year paper narrows. The Fed’s well-telegraphed Twist program, together with anticipatory trading, chopped the spread nearly in half between February and September 2011 and boosted FLAT’s price 31 percent (see Figure 1).
Meantime, STPP, geared to rise as the yield spread widens, lost 26 percent.
Why wouldn’t the results be the exact inverse? Because STPP and FLAT aren’t benchmarked to the cash yield spread. Instead, they’re tied to the rate differential in T-note futures. Investors familiar with the futures market know that prices for the same quantity and quality of a commodity can vary depending upon contracted delivery dates. In a normally well-supplied market for a storable commodity, prices for more immediate deliveries tend to be lower than those of deferred contracts. July crude oil, for example, might sell for $96 a barrel when October contracts trade for $97. The $1 “contango” represents the storage and insurance costs incurred between July and October, together with interest charges. The October premium, then reflects, the cost of “carrying” oil until an autumn delivery.
The same scheme normally holds true for T-note futures. Sort of. A normal market for notes and bonds exists when deferred delivery prices are lower than the cash or spot cost. That’s because there’s a benefit, not a cost, associated with holding a note over time, i.e., accrued interest.
So what’s this got to do with the securities’ gain/loss differential? Just this: the ETNs’ benchmark maintains constant yield curve exposure by notionally rolling its investment in T-note futures forward. A contango benefits one of the ETNs but costs the other, even before changes in the shape of the yield curve are factored in. To boot, the spread is levered to compensate for duration risk. The two-year exposure constitutes 75 percent of the index weight; ten-year paper accounts for 25 percent of the benchmark.
Now let’s roll forward to the present market. Since the top of the year, in fact, the two-year/ten-year spread is backed up 33 basis points (or nearly 22 percent) as the Fed jawboned about a possible curtailment of Operation Twist. The effect on FLAT and STPP is obvious.
In 2013, STPP’s become the outperformer while FLAT lags. More dramatic than the flip-flop in returns (see Table 1) is perhaps the fall-off in the once-popular FLAT’s volume. Daily turnover for the notes is down 85 percent this year as their total market capitalization has been slashed 25 percent.
STPP’s flip to profit hasn’t produced an uptick in daily volume, though asset flows have increased by eight percent this year. The ho-hum trading performance of the ETN may be due, in part, to its lack of visibility (how many pundits do you know who’ve recently pitched STPP?) More likely, investment alternatives have sucked up some of that flow.
This year’s best short Treasury play in the 10-year maturity bucket has been another Barclays exchange-traded note. The iPath US Treasury 10-Year Bear ETN (DTYS) booked a year-to-date gain of seven percent through the end of May, far and away ahead of competing products. That’s largely due to the note’s geared pricing. Essentially, DTYS is designed to increase in value by ten cents for every one basis point decline in the ten-year T-note yield (conversely, the notes’ price should fall as yields increase). Other products’ duration risk—usually lower than that of DTYS—can vary considerably.
Like other iPath products, DTYS charges an annual investor fee of 75 basis points. It’s not the cheapest product in the ten-year space, but neither is it the most expensive (see Table 2). The least costly, at 65 basis points, is the Direxion Daily 7-10 Year Treasury Bear 1X (TYNS). Its leveraged stable mate, the Direxion Daily 7-10 Year Treasury Bear 3X (TYO), charges 95 basis points as do the ProShares funds, bringing the asset-weighted average for the category to 82 basis points.
Another market-weighted stat is noteworthy: Asset flows into these short Treasury exchange-traded products has averaged 17 percent this year, more than double the inflow rate for the STPP note.
In contrast, bullish Treasury ETPs at the short end of the yield curve have leaked assets at an average five percent clip in 2013. There are a half dozen products tracking two-year T-notes, including the Schwab Short Term U.S. Treasury ETF (SCHO), the PIMCO 1-3 Year US Treasury Index ETF (TUZ), the SPDR Barclays Capital Short Term Treasury Bond (SST), the iShares Barclays 1-3 Year Treasury Bond (SHY), the Vanguard Short-Term Government Bond Index ETF (VGSH) and the iPath US Treasury 2-Year Bull ETN (DTUL).
Carrying exposure to short-term Treasuries is a lot cheaper than inverse commitments to longer term paper. The market-weighted annual expense for the two-year product set is just 15 basis points (see Table 3).
That last point appeals to the bargain hunter in me. If I want to play the steepening yield curve, I want it cheap which makes me wonder if the combination of a long two-year and a short ten-year ETP could produce results similar to STPP.
Turns out it can but not without concessions. Let’s test drive a couple of models to see how.
We could use the Schwab SCHO fund for inexpensive two-year exposure while relying upon the iPath DTYS note’s rate sensitivity in the ten-year space. However, if we use these products in a ratio replicating STPP’s underlying index, we’re likely to get sub-par results.
From the top of 2013, a 3:1 composite of SCHO-to-DTYS (a 25 percent-to-75 percent mix) would have been decidedly less volatile than an STPP position. That makes for comparatively lower highs and higher lows and a return about half of STPP’s. Still, we’d get a better Sharpe ratio and a blended cost just a third of STPP’s (see Table 4).
Divvying up the yield curve exposure into equal dollops improves the composite return, but nearly doubles its standard deviation. The aggregate expense goes up as well, but at 42 basis points is still much cheaper than STPP.
Turning STPP’s exposure ratio on its head, i.e., 25 percent two-year and 75 percent ten-year, produces a return that outstrips that of the iPath product. A relatively modest uptick in volatility keeps the combo’s Sharpe ratio well ahead of STPP’s at a price that’s 23 percent cheaper than the spread ETN.
Well, this exercise satisfies my bargain hunter instincts, but we have to consider that we’re looking retrospectively. Can similar results be expected going forward? Perhaps, but there’s no guarantee. We also must consider that the Fed’s avowed intention is to keep short rates low until the national unemployment rate is whittled down to 6.5 percent. That’ll keep a lid on two-year paper prices. So is it better to just sell the long-end? It’s certainly simpler, but when we consider an outright DTYS position costs 75 basis points and yields the same Sharpe ratio as that produced by the 25-percent/75-percent portfolio at only 58 basis points, the case for a cheaper, flatter world is certainly compelling.