Fed Frontrunning Update: The 5-7 Year Space Gives Best Returns As The Fed Prepares To Run Out Of Treasurys To Buy
It is time to once again consider the options for the only trade that make sense: frontrunning the Fed. Last week we did an analysis on how much, in Goldman's opinion, the had market priced in in terms of QE. The result was not surprising, as it appears that double the anticipated $1 trillion in QE is already priced into bonds, and half of it in stocks. Yet at the end of the day, all of this is irrelevant: as long as there is even one basis point in 30 Years to be picked, the Fed will pursue it. And when the curve is as flat as a pancake, and all rates are at zero, that is when the Fed's last ditch desperation move will be to do what the BOJ did and buy REITSs, ETFs, stocks, hops, malt, grains, sugar, coffee and pretty much anything not nailed down. But we probably have at least 12 months before we get there. So what to do in the meantime? Morgan Stanley's Igor Cashyn, whose track record of predicting what the Fed and the FRBNY do is second only to Bill Gross' (wink wink), has posted an update on where investors will get the most bang for their buck once $1-1.5 trillion in QE2 is announced. As we speculated first several weeks ago, Cashyn takes into account the prepayments of MBS put to the Fed, and realizes that the lower rates drop, the greater the negative convexity to prepay even more, forcing the Fed to purchase even more bonds. Which is why unlike others, like Barclays for example which has a $100-120 billion a month monetization bogey, Cashyn has a more modest expectation of "only" $70 billion in USTs bought back monthly. However, that $70 billion also adds another $30 billion in MBS prepays, adding up to pretty much the same number. Of course, when all is said and done, the Fed could easily end up announcing $1.5 trillion in UST monetizations, which would effectively mean a total of $3 trillion in Treasurys to be bought as we speculated much earlier. The problem, as we also concluded, is that there are simply not enough Treasurys across the entire curve, in existing or projected issuance, to satisfy the Fed's possible total monetization needs! And this is precisely the same conclusion that MS reaches, however courtesy of their less dramatic total Fed demand number (for now), Cashyn is mostly concerned about bonds in the 5-7 year sector, which he thus finds most attractive for Fed frontrunning purposes.
But before we get into specifics, we would like to take a moment to gloat over a trade which we ridiculed when MS put it on initially: the 10s30s flattener. At the time we pointed this trade out as nothing more than a reactionary response to client anger aimed at MS for expecting steepening across the curve. When that trade was finished, MS immediately flipflopped and pithced long-end flatteners. Oh well, that did not work out too well. As of Friday: "We stopped out of our 10s30s flattener as this curve resumed its steepening move (currently at 132bp). We originally recommend entering a tactical 10s30s flattener to take advantage of the historical flattening of this curve that typically follows large steepening moves going into FOMC. We note that the steepest that the JGB 10s30s curve ever got was 123bp in 2000 and considered that the US might follow Japan when we put this trade on in September." Oh well, it didn't. But nobody's perfect.
Anyway, here are the most recent observations on a topic MS has been very good at: how to move a few steps ahead of the Fed:
Given the high probability attributed to a QE2 announcement in November, in our view, one question that comes up is regarding the amount of room the Fed has in purchasing Treasuries before hitting its 35% SOMA ceiling for any one given Treasury security. In our previous monthly publication (see Trading Opportunities Amidst Low Rates & Slower Growth, September 2, 2010), we addressed the issue of whether the belly of the Treasury curve can absorb the needs of the Fed’s agency debt/MBS reinvestment needs, concluding that the Fed will have access to ~$500 billion in the >5-10y sector alone over the next year in which to reinvest the $350-400 billion that we expect in agency debt/MBS paydown over that period.
However, with the prospect of QE2 now around the corner, we must update this analysis to see how much room the Fed has if, in addition to the $30 billion/month in reinvestment needs, the Fed was to ramp up its monthly purchases by another $70 billion/month and absorb all of the net issuance in coupon Treasuries (currently running just over $100 billion/month). This would amount to $1.2 trillion of purchases over the next one year, an upper-end estimate of how much we think the Fed may announce in November (especially given that the Fed appears to favor an incremental approach).
If we just look at what’s available now for the Fed to purchase, we see that the Fed has access to $1.37 trillion of eligible coupon Treasury securities that it may purchase, with another $660 billion becoming eligible due to new issuance over the next twelve months (see Exhibit 3 - left):
However, to the extent that the Fed continues to apply the same distribution that we’ve seen by combining its purchases in 2009 and 2010, we find that the available capacity for its purchases in the >5-7y sector is a bit unsatisfactory (Exhibit 3 - right). If we apply the Fed’s purchase distribution just from 2010 (only its latest purchases), we find that it is again the capacity of the >5-7y sector that falls even more short of the Fed’s potential buying needs (although again, it’s worth stating that we’re assuming a full-blown QE2).
A possible solution to this is for the Fed to a) spread out its purchases into the front end and >7-10y part of the curve, in a pursuit of a barbell strategy, or b) raise its 35% SOMA limit. We think the former is more likely, and again this is likely to be less of a problem initially if the Fed does pursue a more incremental approach to its Treasury purchases, which seems to be in fact favored at this time.
Before we get into the actual trade, let's discuss this first: adding the various total available bonds for purchase comes to just over $2 trillion over the next year. In other words, assuming even MS' lowball estimate of just $30 billion a month in MBS prepays, or $360 billion in one year, it means that the absolute maximum the Fed can issue in the form of QE, limited entirely by the supply side, is $1.7 trillion. Anything above this means that the 30 Year will effectively drop to zero as the Fed will be forced to purchase the most expensive bond at any price up to and including a 0% yield. Which also provides a handy shortcut to how much buffer there is inbetween QE2 and QE3 before the Fed has no choice but to start buying equities.
Taking this further, if instead of $360 billion, the Fed sees $750 billion in MBS put to it, and it has already committed to $1.5 trillion in UST QE, it means that the Fed will have no choice but to purchase non-UST asset purchases. And with MBS already pretty much irrelevant, and providing no bang for the buck, it means that Bernanke will be one step closer to admitting he will have to buy equities outright and in the open.
But, as we noted before, there is time before this hyperinflation moment sets in. here is Cashyn on how to trade the next 12 months:
From a trading perspective, this favors being long the >5-7y sector, and this is the sector that now offers the best rolldown + carry on the curve at roughly 14-14.5bp over a 3-month horizon (slightly above the 14bp of the on-the-run 5y!).
It also makes sense to be overweight the belly from the expected distribution of Treasury supply in F2011 versus the blended 2009/10 distribution rates of where the Fed tends to purchase Treasuries (Exhibit 4).
The asset swap curve has already started to reflect this, richening up the 5-15y sector in a more homogeneous fashion than was previously the case (Exhibit 5).
And for those who prefer pair-trades instead of outright long/short positions, Igor also suggests the following Matched-Maturity Switches:
With the probability of QE2 running high, the Fed’s demand for cheap issues will only increase, and we’ve already seen how this tends to benefit RV/convergence plays on the Treasury curve since its decision in August to restart Treasury purchases in order to keep the size of its SOMA portfolio constant.
To that end, this week we introduce a new analytic to our Liquid Rates Tracker daily data package that focuses on convergence plays between bond pairs of different original tenors (e.g., 3s and 5s) but of the same month/year in maturity (e.g., 0.750% Aug13s vs. 3.125% Aug13s). Essentially, we are comparing the valuations of more seasoned bonds that have rolled down the curve versus more recently issued bonds, and we utilize their ASW spreads to conduct our analysis.
Exhibit 6 illustrates the matched-maturity pairs of original-maturity 5s that rolled down into the 3y sector versus original-maturity 3s, illustrating the ASW spread of spreads between the two issues (we use OAS to take the coupon difference between the two bond pairs into account), the 3-month average of that spread of spreads, and the 3- and 6-month z-scores of that spread of spreads. Because the spread of spreads is calculated as the more seasoned bond minus the less seasoned bond, a positive z-score means that the less seasoned bond has cheapened to the more seasoned bond. The report has similar tables for 2s/3s, 2s/5s, 5s/10s, 7s/10s and 10s/30s as well. Finally, we aggregate the average yield pick-up/give up involved in owning one of the original series versus the other and provide a time series of these pick-ups as well, illustrated for the 3s/5s and 7s/10s matched-maturity pairs below (Exhibit 7):
Above we can see that while all of the cheapness of owning off-the-runs 7s has been corrected, owning off-the-run 3s is still preferred to owning matching off-the-run 5s. That is, if we were to switch from original-maturity 5s that rolled down the 2011/13 part of the curve and into original-maturity 3s, we would pick up an average of 3.6bp. In the graphs above, we can also see that the yield pick-up of owning original-maturity 3s (and 7s for that matter) was greatest after the start of the EU sovereign crisis in April/May, but has since come back in once the Fed started targeting cheap issues on the UST curve in August/September.
Net, the purpose of the aggregate graphs is to identify more widespread dislocations across these bond switches, and while the most obvious dislocations have already been corrected, the sector grids like Exhibit 6 can still identify individual pairs of interest. To that end, the following pairs look attractive, both, in terms of the large yield pick-up they offer and the favorable momentum indicators via their 3- and 6-month z-scores:
- In the 2012 sector, we recommend buying 1.375 of Sep12s at Libor - 15.8bp vs. Dec TU contracts at an invoice spread of Libor - 20.5bp (CTD: 4.5% of Sep12s)
- In the 2017 sector, we recommend buying 1.875% of Aug17s at Libor - 21bp vs. Dec TY contracts at an invoice spread of Libor - 24.5bp (CTD: 4.75% of Aug17s), for a yield pick-up of 3.5bp
Risks: The risks to the above trades include a flight-to-liquidity that favors the Treasury futures contracts over off-the-run bonds, albeit we think that this risk is manageable considering that if the Fed does decide to implement QE2 (and our economists think the probability of this is very high at this point, with the recent poor ADP print only further supportive of this view), the off-the-run sectors will outperform as the Fed proves supportive of the cheaper issues on the curve.
So there you have it: want to frontrun the Fed and feel good about yourself for stealing from America's labor class? This is how it's done.