Ben Bernanke will face difficulties in disentangling a slowdown in purchases with a tightening in monetary policy conditions. I develop this further below, as the Fed meets on Tuesday and Wednesday.
One of the consequences of the Fed QEs was to separate the UST yields from the economic cycle. As can be seen in the chart below, the 10-year Treasury yield (UST 10Y) disconnected from the news flow in the end of 2011. We are currently seeing a recoupling of the series, suggesting a re-pricing of the UST 10Y.
One of the most difficult challenges for the Fed over the next few quarters will be to communicate and properly explain the separation principle:
1. Management of the liquidity (LSAP or Large Scale Asset Purchase) and the associated tapering off; and
2. Guiding expectations of future rate hikes (forward guidance).
A successful exit would start with a steepening of the yield curve as the short end is still pegged by the numerical targets (2.0% for inflation and 6.5% for the unemployment rate), while the tapering is expecting to reduce liquidity and add some upward pressure on long yields.
The recent move in the U.S. yield curve is clearly a reflection of this duality.
The 2/10 year yield curve has even recoupled with the S&P 500 implied level.
The model on the chart below is an estimate of the 10-year UST yield based on Fed Funds, the euro dollar slope (long dated euro dollar minus first nearby), the quarterly changes of the Fed's balance sheet, long-term inflation expectations (University of Michigan) and short-term stock returns. Estimated between 2000 and today, it suggests that long bond yields are close to fair value.
Getting into the details of my model I find that:
1. the slope of the euro dollar curve (forward guidance proxy) is positively correlated to 10-year UST yields (100 bp rise translates into an increase of 60 bps of UST 10-year yields); and
2. the same goes for the quarterly changes in the Fed's balance sheet. I highlight this in the chart below: any time the Fed increased its balance sheet, yields went upward, not downward (except during Operation Twist). This is counter to the conventional wisdom view that Fed purchases put a cap on long-term yields.
This can easily be explained by the flows towards riskier assets that the Fed purchases triggered. But it suggests that if you want to use a statistical model to gauge the impact on yields of the Fed's tapering (lower purchases, stop of purchases, sale of bonds), then you have to make another assumption on the stability (or lack thereof) of the relationship between US Treasuries and other asset classes.
As, if the model holds, the relationship would be:
Expectations of a reduced flow of liquidity => fewer inflows in risky assets +> lower risky asset prices => lower UST yields
The periods when the Fed was buying were also times when the correlation between U.S. Treasury prices and risky asset prices was weak. Conversely, a status quo brought the correlation to a more significant negative level.
As a result, if the Fed stops buying in late 2013, then risky asset returns might once again be more negatively correlated to UST prices and positively correlated to UST yields. This would be consistent with what my model implies: lower stocks and lower yields!
The case for the Fed tapering bringing UST 10-year yields higher is not as simple as it seems. The most difficult part for the Fed will be to communicate efficiently on the forward guidance side of its action since, as we have seen recently, higher long-terms rates are mostly a reflection of a move of the Money Market curve.