In a new working paper, the St. Louis Fed’s Christopher Neely argues The Large Scale Asset Purchases Had Large International Effects.
The Federal Reserve’s large scale asset purchases (LSAP) of agency debt,
MBSs and long-term U.S. Treasuries not only reduced long-term U.S. bond yields also
significantly reduced long-term foreign bond yields and the spot value of the dollar. …
changes were much too large to have been generated by chance and they closely followed LSAP
announcement times. These changes in U.S. and foreign bond yields are roughly consistent with
a simple portfolio choice model. Likewise, the exchange rate responses to LSAP announcements
are roughly consistent with a UIP-PPP based model. The success of the LSAP in reducing longterm
interest rates and the value of the dollar shows that central banks are not toothless when
short rates hit the zero bound.
This is an important finding because it re-affirms the scope for Fed policy even when the Fed Funds rate has hit zero. It also buttresses the results of Gagnon et al., showing that Fed purchases of long term bonds did have an impact on long term interest rates.
I’ve discussed the portfolio balance model in the past, noting the limited success in estimating risk premia.   . The Fed LSAPs have constituted an experiment (albeit not completely controlled) of unprecedented size, and that large variation is what was needed in order to obtain evidence in favor of the portfolio balance model.
How do the predictions of [the] simple portfolio model compare to the actual
changes in observed prices? Recall that the simple model predicted that a 22 percent reduction
in the quantity of U.S. debt outstanding would increase expected real U.S. bond returns by 178
basis points and foreign expected real returns (in U.S. goods) by 100 to 160 basis points. These
estimates are subject to uncertainty because the mean and covariance of returns must be
estimated and because the number of assets is limited. 11 If the LSAP did not change inflation
expectations, then these expected real returns translate directly into expected nominal returns.
Table 2 showed that 10-year U.S. Treasury yields fell by 107 basis points over the 5
“buy” announcement windows. The troubled state of global credit markets in the autumn of
2008 meant that long-term high-quality asset prices were already very high, by historical standards, when the LSAP was announced. Given the existing high prices caused by low risk
tolerance, the actual falls in U.S. Treasury prices seem reasonably consistent with the very
simple model’s predictions.
The model incorporates “risk-adjusted” uncovered interest parity (including at the 10 year horizon, which is consistent with Chinn-Meredith (2004) ) and long run purchasing power parity — so that the Dornbusch overshooting effect holds. Figure 7 from the paper depicts the impact on the dollar’s value.
Figure 7 from Neely (2010).
These results clearly have policy implications, above and beyond expanding our understanding of how international financial markets work. In particular, they suggest that the Fed can affect the value of the dollar even if the policy rate has hit zero — something I had wondered about previously   . It also reminds us that, as the economy’s rebound weakens, monetary policy can still offset (some of) the effects of ill-advised attempts by some to withdraw fiscal stimulus.    (That last point is my view, and not in the paper).