Yesterday, Goldman's Jan Hatzius, piggybacking on what has now become a prevalent belief among Wall Street economists following a "leak" from the WSJ's Jon Hilsenrath, predicted that the FOMC minutes would hint at more easing, in the form of "sterilized" interventions, or in other words, an extension of Operation Twist. There is, however, one problem with this analysis. It is total BS, for a simple reason that for every bond on the long end that the Fed buys (and it has bought a whopping 91% of the 20-30 year gross Treasury issuance), it has to sell one in the 3 Month - 3 Year maturity interval. And therein lies the rub. As Bank of America shows below, at the end of Twist in June there will be just 2 months worth of Treasurys available for sale. What could fix this? Well, instead of ZIRP until 2014, Bernanke could say the Fed would keep rates at zero until 2016 or even 2018, and proceed to sell all Fed holdings in the 3 month - 5 year or 3 month - 7 year intervals. This however, would make the entire bond curve an epic farce, shifting the belly to beyond the 10 year point, and in the process blowing up the MBS market due to total collapse of traditional convexity heding strategies. Which we don't think is likely unless the world is coming to an end. In other words, anyone hoping that Twist will be extended, is wrong, and in turn it means that any real option for the Fed's NEW QE will be the outright monetization (aka LSAP) of either USTs or MBS, ala QE1 and QE2.
BofA explains further:
Extending Twist is a limited option, as the Fed will have only about $175 bn of short-dated Treasuries (3 months to 3 years) in its SOMA portfolio on June 30. That would allow two to perhaps three months of further twisting at the current pace — i.e., into September. That does buy some time, but the Bernanke Fed has not been one to go for half-measures or small steps since the crisis began. If the outlook warrants more easing, we still see QE3 as the most likely tool chosen.
Math - it's fundamental.