Goldman Explains Why The "Orphaned" 30 Year Will Soon See Buying Interest, Expects A Drop In The 10s30s Back To 110bps
FUG (as Goldman's Francesco U. Garzarelli signs his emails) has released another note with his outlook for the Treasury curve over the next several months. As readers will recall, it was Goldman's call that the long-end would be bought up by the Fed, leading to an implicit flattening of the curve. Nonetheless, the New York Fed disclosed that, as Morgan Stanley expected, the bulk of purchases would occur around the belly, resulting in, as we highlighted yesterday, what turned out to be a record steepening of the 5s-30s, which could merely be the last trump card the Fed has to generate some profitability for the banks, whose core business model, now that the hedge fund and sales and trading model is in shambles with plunging market participation, is the treasury curve trade (long near, short far). Despite Brian Sack's attempt at giving taxpayer capital to banks in this last attempt to goose the banking sector, Goldman continues to be skeptical about further steepening of the curve: "Our best guess (corroborated by our GS Curve estimates) the slope between 10-yr and 30-yr will retrace to around 110-125bp, from 160bp currently." The explanation for why the Fed would ignore purchases of the long bond even as it bids up everything else is as follows: "While arguably increasing transparency and predictability, the Fed has also lost degrees of freedom, and the costs and benefits are yet to be seen. In the eyes of many, the long bond now appears to have been orphaned by the Fed. One explanation for this may be that the FOMC wanted to have some quantification of the potential costs of the asset purchase policy under future interest rate scenarios before its launch." Well, it now knows. And now that the UST curve look literally like a hockeystock, and the Fed is about to be accused of massive telegraphing of intentions by Ron Paul, we expect that Goldman will be proven right as the yield chase game continues, and the Fed ultimately makes it clear that it will have no choice but to gobble up the long-end as well, especially since if as we expect, MBS repurchases will be far higher than expected, resulting in a far greater contribution to monetization due to QE Lite, and leaving far less available for purchase across the balance of the curve.
Goldman's summary notes:
- The Fed’s asset purchase program is broadly in line with our expectations.
- The target average duration of 5-6-yr underpins our 1% yield forecast for 5-yr T-Notes.
- We remain of the view that 10-yr USTs will not stray far from 2.5% to year-end.
- And that the cyclical lows for long-dated yields have already been seen.
- The Fed’s actions should continue to encourage a rotation into pro-cyclical assets
And here is the full note:
On Wednesday, the Fed announced that between now and next June it intends to purchase an additional US$600bn of US Treasury securities (Notes, Bonds and inflation-linkers) with maturities ranging from 1.5-yr to 30-yr at monthly clips of around US$75bn. Moreover, the Fed will continue replacing maturing Agency/MBS with Treasuries, as initially indicated in early August. These switches are expected to amount to around US$250-300bn, or roughly US$35bn per month.
The additional monthly demand for US Treasuries coming from the Fed to the end of Q2:11 will amount to US$110bn, compared with an estimated average monthly gross issuance of coupon bonds of around US$170bn over this period. By the middle of next year, the Fed will be sitting on a portfolio of US$1.7 trillion-worth of Treasury securities (roughly 20% of the marketable stock), with an average duration of 5-6-yr.
The New York Fed has provided a breakdown of the planned cumulative security purchases by maturity bucket: 63% of the purchases, or roughly US$560bn, will be concentrated in maturities ranging between 2.5-yr and 7-yr. This amounts to around three-quarters of the expected gross issuance of corresponding maturity Treasury securities from now until next June. Purchases of bonds with maturities between 10-yr and 30-yr will amount to 6% of the total, or around US$60bn. We estimate the long bond issuance to be in the region of US$115bn over the corresponding period.
The New York Fed’s website contains exhaustive operational details on the purchase program in Q&A format. Previously planned purchases for reinvestment will take place as scheduled on November 4 and 8, after which the two programs will be combined with the expected publication of a new purchase schedule on November 10 at 2pm NY time.
Relative to Tuesday’s close, 5-yr and 10-yr bonds have rallied about 6-9bp, while 30-yr yields have sold off 22bp, resulting in a sharp steepening of the curve. 30-yr zero coupons have sold off 25bp since Tuesday’s close. At 4.6%, they are back to the average since 2002.
Our thoughts on US rates are now as follows.
By restarting its asset purchase program, as our US economics team has predicted would occur since early August, the Fed is trying to provide further stimulus to the economy while circumventing the zero nominal bound on interest rates. A backward-looking ‘Taylor Rule’ tying the policy rate to contemporaneous values for the core inflation rate (CPI and PCE, and the unemployment gap (defined as the difference between the unemployment rate and the CBO’s estimate of NAIRU) would indeed put the Fed Funds target in deep negative territory, (see US Economics Analyst 10/42 ‘QE2: How Much Is Needed?’, 22 October). Having exhausted the ‘vertical’ range of easing, the Fed needs, figuratively, to ‘go horizontal’.
As the experience since last March has shown, anchoring the 2-to-5-yr maturity sector is hard to achieve with only a promise to keep policy rates low for an ‘extended period’. Indeed, since the distribution of expected returns is truncated, investors will require a bigger premium depending on the degree of uncertainty over possible future outcomes. Under these extreme circumstances, the Fed needs to participate directly in the bond market to make its commitment credible and effective.
Since the start of August, we have forecast that the Fed would concentrate purchases in the intermediate sector of the yield curve, pushing 5-yr Treasury yields to 1% at least over the current quarter and the next. This will continue to enable investors to extract ‘carry’ from rates between 3- and 5-yr (the roll-down on 1-yr swap rates 2-yr, 3-yr and 4-yr forward is currently around 80-90bp per annum). We are sticking to this forecast. As a cross-check, we run a regression from 1988 relating the same two macro factors entering the Taylor relationship introduced above to the 5-yr nominal Treasury yield, and find that the latter should be around 1%.
Moving further out on the curve, where international influences become more important in the determination of yields, we remain of the view that 10-yr USTs will not stray far from 2.5%, roughly where they are trading at the time of writing, until the end of this year. This has been our call for end Q3 and Q4 since early August and we have also argued that the cyclical lows for the long-end is now upon us. Our GS Bond Sudoku model, which feeds off our 1-yr-ahead macro forecasts, currently places the ‘fair value’ for 10-yr USTs at 2.8%, before rising gently over the coming quarters.
Where the 30-yr bond should trade in coming months based on macro factors is more contentious, and it is a segment of the curve where we admittedly have less conviction than on intermediates. The evolution of households’ long-dated inflation expectations will have a greater bearing, and these are hard to predict, particularly under current circumstances. Our best guess (corroborated by our GS Curve estimates) the slope between 10-yr and 30-yr will retrace to around 110-125bp, from 160bp currently.
Regarding the impact of the allocation of Fed purchases on the long bond, we have been of the view that these would span the entire term structure, and would not be concentrated at the ultra-long-end, as some commentators had argued. In our opinion, it is time-inconsistent for a central bank to attempt to lift inflation by keeping longer-dated nominal forwards capped for a sustained period, particularly given the potential for balance sheet losses down the line. Moreover, given that the long-end of the yield curve carries important information value on long-run expectations, the Fed may not want to distort this segment.
Nevertheless, previous intervention had referred more vaguely to the purchase of ‘longer-dated securities’, including the 30-yr, but the actual cumulative distribution was known to the market only ex post. While arguably increasing transparency and predictability, the Fed has also lost degrees of freedom, and the costs and benefits are yet to be seen. In the eyes of many, the long bond now appears to have been orphaned by the Fed. One explanation for this may be that the FOMC wanted to have some quantification of the potential costs of the asset purchase policy under future interest rate scenarios before its launch.