Unlike Morgan Stanley whose calls on the shape of the yield curve have been pretty much wrong all year, and which have changed consistently, Goldman has been rather quiet on what it expects the curve to do. Today Francesco Garzarelli has come up with expectations that, unlike MS, the bulk of the buying will be concentrated at the 5Y locus, even shorter than the MS-preferred 7-10 Y. While this does make sense as there are far more bonds of shorter duration available for purchase, it also means the average holdings of the Fed will soon be cut in duration even more, which will eventually become a sufficiently large political factor that we expect Congress to soon get involved in discussions over the viability of the Fed's balance sheet (think massive asset-liability duration mismatch). Goldman also notes that it expects the 30 Year to be purchased and that the 10-30s will flatten, even as the 5s10s steepens. In other words a 5s10s30s butterfly may be the right way to play the Goldman trade... or to, inversely, fade it as so often is the right way to trade Goldman recos.
From Francesco Garzarelli:
A ‘Catenary Effect’ on the Yield Curve
We expect the Fed to announce a program of public debt purchases next week, starting with US$ 500bn and potentially growing up to US$ 2 trillion. This would allow the desirable ‘horizontal easing’ to become effective.
Consistent with these expectations and our baseline macro projections, over the next 3-to-6 month we see US 5-yr USTs trading around 100bp, or possibly even temporarily lower– a forecast we have held since the start of August and to which we are sticking.
Further out on the term structure, we expect yields to be the resultant of two opposing forces:
- By pinning down maturities in the 3-5-yr area through direct bond purchases, the Fed will create what amounts to a ‘catenary effect’, keeping longer-dated bonds better anchored. Readers familiar with nautical terms can think about the 5-yr maturity as a ‘kellet’, the lead weight that sailors usually attach half way along the anchor cable to increase the horizontal tension, thus keeping a boat more stable in choppy waters.
- Admittedly, the yield curve has already sagged considerably as expectations of Fed purchases have increased. Our GS-Curve model –which relates the term structure of rates to consensus macro expectations over different horizons - indicates that 5-yr Treasuries now trade around 5 standard deviations too ‘rich’ relative to a linear combination of 2-yr and 10-yr maturities. This same ‘butterfly’ was 2 standard deviations out of kilter in early June. Pushing the nautical metaphor to its limit, our expectation is that this ‘anomaly’ will persist until the tide (i.e., the economic recovery) comes in.
- Expectations on the macroeconomic environment, both domestic and overseas, will likely apply an opposing force. Expectations on US growth have come down considerably, and the deceleration in aggregate demand growth we were expecting to see from H1:10-H2:10 has largely been seen. The final reading on our Global Leading Indicator, due this coming Monday, will provide more information on these dynamics. Furthermore, economic conditions in the rest of the world, particularly in Europe, continue to surprise to the upside.
Combining these considerations with the fact that US 10-yr yields are trading below their ‘fair value’ according to our valuation models, we forecast that 10-yr Treasuries will not stray far from 2.5% until year-end, before gradually increasing to 3.25% by end-2011.
A final word on the implementation of the purchases. We think these will likely be spread across the entire term structure, including the 30-yr maturity, although the target average duration will probably be around 5-yrs. To the extent that this is not entirely discounted, it could result in a flattening in the 10-30s portion of the yield curve after the FOMC announcement.