BackgroundQuantitative Tightening/Easing can have an effect on bond yields as a result of two plausible channels. (The obviously incorrect channel is the Quantity Theory of Money, which is dead as a doornail as a result of the empirical QE test.)
- Signalling effects (what do policymakers think about the stance of monetary policy currently, which should translate into expectations about future moves).
- Portfolio balance effects. Central bank selling/buying of bonds disturbs private sector balance sheets; prices need to shift to allow the market to clear.
I am going to ignore the second channel (portfolio balance effects). Although there are some obvious disclaimers (an unexpected dump of $200 billion of bonds on a Friday afternoon at 3 pm should move bond prices), the effect is probably not measurable. (Furthermore, the portfolio balance effect is presumably correlated with the signalling effect, and so it would be difficult to untangle them.) Therefore, I will just discuss signalling effects.
QE and Signalling
The signalling effect of QE is easier to discuss, since we just passed through that dismal period of monetary policy voodoo.
The problem that Fed policymakers faced in the post-crisis period is that they were at, or very close to the effective lower bound for the policy rate. (Negative rates can be achieved, but institutional factors preclude short rates of -5%.) In order to create greater interest rate stimulus, the Fed needed to lower longer-term interest rates. (Assuming the standard view that lower interest rates stimulate activity.)
They could try jawboning markets, through speeches and devices like the "dot plots." The problem is that talk is cheap; there is no reason to believe that future policymakers will follow the hints of earlier policymakers. Buying bonds to create excess reserves was a way of signalling policy makers' views about the future, and this signal is backed by actual purchases in the market. (The fact the Fed can raise rates regardless of the size of its balance sheet once it had the ability to pay interest on reserves was not fully appreciated by some American monetary policy wonks, as said wonks never bothered looking at central bank practice in any other country than the United States.)
The problem with QE was that bond purchases are in the wrong units; there is no obvious way to map the size of purchases to what policymakers think about interest rates. (The dot plot does not have this defect.) That said, the Fed could leak a way of translating the size of purchases to interest rates, and then the market could create an informal calibration. To what extent this was effective is left as an exercise to the reader.
I used to be a secular bond bull, so anyone familiar with me would not be surprised when I write that a policy move is bond-bullish. Therefore, I will start with a scenario in which Quantitative Tightening is in fact bond-bearish.
Imagine that the Fed has entered a "steady state tightening" mode: they hike by 25 basis points every second meeting. This would not be unusual; the tightening cycles after 1994 featured a steady drumbeat of 25 basis point hikes every meeting.
In such a scenario, there is little debate about the near-term path of interest rates. Implied volatility is not zero; the possibility of a crisis always exists. However, realised short-term rate volatility is low, as the Fed will just ratify the rate hikes prices in on a 1-2 year horizon. The volatility in the market revolves around the pricing of the terminal rate: at what point does the steady drum beat of hikes stop? Forward rates will trade in a range around the plausible terminal rates, which has a net effect that bond yields only rise slowly. (This causes economists who refuse to understand their own models to write op-eds about the "bond yield conundrum.")
In such an environment, the Fed has little ability to jawbone markets. Adjusting the near-term path of rate hikes was largely precluded; faster hikes would create a 1994-style uncontrolled panic -- an experience Fed policymakers generally want to avoid. In order to convince markets that forward rates should be higher, the Fed would need to convince market participants that they had a very good long-range economic forecast (which would move the terminal rate). Although the Fed has a good short-term forecasting track record (excluding recessions), nobody sensible takes long-term forecasts too seriously.
Quantitative Tightening adds a new signalling mechanism. If the Fed ramps up its bond sales, while at the same time hiking rates at 25 basis points at every second meeting, that implies that Fed would be open to hiking 25 basis points every meeting. The difference between 100 and 200 basis points a year would be enough to crush bond bulls, but causes less of a panic. (This is because the faster rate hikes still remains a threat, rather than reality.) This allows for more control over the near-term path of short-rates, without triggering a 1994-style panic.
However, this only works if the Fed is hiking rates at the same time. If the sales are a substitute for rate hikes, we actually are pushing expected rate hikes further into the future. These lower forward rates -- which can be measured -- will almost certainly have a larger effect on bond yields than the hypothetical portfolio balance effect (which cannot be measured).
Furthermore, the current cycle is already long in the tooth. Credit numbers, and auto data are already signalling possible problems. Delaying rate hikes means the terminal rate will be much lower if a recession hits within 1-2 years.
It should be noted that many economists follow a hidden assumption that a recession are only caused by "too high" interest rates. (Monetarists -- to what extent they still exist -- will say something about the money supply.) If you believe this, delaying rate hikes will not affect the terminal rate, since the "terminal rate" is the mechanism that terminates the business cycle. Conversely, if you downplay the importance of interest rates on the economy, a recession can occur at any level of interest rates for reasons endogenous to the economy, and so delaying rates does lower the terminal rate. I am not going to resolve that debate here.
Unless the cycle is terminated by a recession soon, expect to be inundated by research "proving" any number of things about the Fed's balance sheet and bond yields. However, what matters is how Quantitative Tightening to relates to rate hikes, and not the dollar size of flows.
(c) Brian Romanchuk 2017