Deutsche On QE3, It's $800bn Or Bust!
Buried deep in the 137 pages of Fixed Income 2012 Outlook, Deutsche's bond group looks at the implications of an extremely flat US Treasury Curve and implicitly low bond risk premium. Based on 5Y5Y rates relative to long-term growth and inflation expectations, tail inflation risks, and estimates of supply/demand shocks, the current bond risk premium are at levels that were witnessed ahead of the bond market sell-off of 1994, at the peak of the bond market conundrum of 2004-2006 and around QE announcements. This 100bps or so of 2s10s 'flatness' relative to real short rates and expected deficits also corroborates this risk premium. So what does this tell us? The extremely low risk premium fully captures QE expectations. Empirically, they find USD19bn of new QE tends to reduce real rates by 1bps and based on this and a model of fundamentals and risk aversion parameters, they find that Twist was fully priced in last September and since then the current dislocation suggests another full QE2-style package of about $800bn is already priced into the market (ex MBS reinvestment). We just hope the market is not disappointed.
Via Deutsche Bank's 2012 Fixed Income Outlook
A low risk premium
We estimate the risk-premium using two orthogonal techniques. The first technique consists of estimating a 5Y5Y bond risk premium by modeling the 5Y5Y rate relative to long term growth and inflation expectations (as per Consensus Economics), tail inflation risk (as captured by the skew in the SPF’s long-term inflation forecasts), and the correlation between growth and inflation (which captures the supply vs. demand shocks to the economy).
The graph below shows that the risk premium is currently particularly low and at levels that were witnessed ahead of the bond market sell-off of 1994, at the peak of the bond market conundrum of 2004-2006 and around QE announcements (see chart below).
The second technique consists of estimating the risk premium in the slope of the curve by modeling the UST 2s10s slope as a function of real short rates and expected deficits. This model corroborates the conclusion of the 5Y5Y risk premium model: the curve is too flat by about 100bp.
The low risk premium fully captures QE expectations
The low risk premium could be interpreted as the pricing of potential additional QE. We estimate the QE3 amount implied by the current level of 10Y real rates by modeling them as a function of the fundamentals (US economic surprise index) and risk aversion parameters (VIX and SovX WE).
Analyzing historically the reaction function of real rates to QE announcements, we find that USD19bn of new QE tend to reduce real rates by 1bp. Based on this estimate and on the model dislocation, we find that the 10Y real yield was fully pricing in Operation Twist in September and that since then the dislocation has increased to price in another full QE package, similar in size to QE2, of about USD800bn (excluding reinvestments of maturing agency and MBS holdings).