Instead of speaking about “the interest rate” or even “the yield curve”, I wish people would speak more frequently about the yield surface: put duration on the x-axis, per-period default risk on the y-axis and the yield on the z-axis. Banks do not just borrow short and lend long; they also borrow safe and lend risky.
Liquidity is not uniform over the duration-instantaneous-default-risk space. Liquidity is not even monotonic over the duration-instantaneous-default-risk space.
There is still a trade-off for the Fed in wanting lower interest rates for long-duration, medium-to-high-risk borrowers to spur the economy and wanting a steep yield surface to help banks with weak balance sheets improve their standing.
By keeping IOR above the overnight rate, the Fed is sterilising their own QE (the newly-injected cash will stay parked in reserve accounts) and the sole remaining effect, as pointed out by Brad DeLong, is through a “correction” for any premiums demanded for duration risk.
Nevertheless, packaging the new QE as a collection of monthly purchases grants the Fed future policy flexibility, as they can always declare that it will be cut off after only X months or will be extended to Y months.
It seems fairly clear to me that the announcement was by-and-large expected and so “priced in” (e.g. James Hamilton), but there was still something of a surprise (it was somewhat greater easing than was expected) (e.g. Scott Sumner).
Menzie Chinn thinks there is a bit of a puzzle in that while bond markets had almost entirely priced it in, fx-rate markets (particularly USD-EUR) seemed to move a lot. I’m not entirely sure that I buy his argument, as I’m not entirely sure why we should expect the size of the response to a monetary surprise to be the same in each market.