3. Forward guidance: the expectation gets priced
Macro 2: Central banks
There is a scene that baffles anyone starting out: the central bank raises rates by half a point and the stock market closes up. Or it cuts and the market sinks. If you believe prices react to events, those days look irrational. They are not. What the market prices is not the decision: it is the difference between the decision and what was already discounted.
The event is in the price before it happens
A market with thousands of professional participants dedicated to anticipating meetings does not wait for the statement before acting. Weeks in advance it has already built a probability for every possible outcome and adjusted prices accordingly. When the announcement lands, the only thing left to absorb is the part nobody had.
Hence the three rules that govern any meeting day:
- A fully expected decision moves little, however large the number.
- What moves the price is the surprise, and the surprise usually lies less in today's decision than in the future path being hinted at.
- The press conference moves more than the statement. The statement confirms what was expected; the press conference is where new information about what comes next leaks out.
This turned communication into a monetary policy tool in its own right. If the long rate is the expected average of future short rates, then changing that expectation is moving the long rate, without executing a single trade. That is forward guidance: deliberately managing expectations about the path of rates.
The varieties of commitment
Not all forward guidance carries the same force, and it is worth telling them apart because the market treats them differently.
- Calendar-based guidance. A horizon is committed to: rates will not rise before such and such a date. It is clear, but it ages badly, because the economy does not consult the calendar.
- Threshold-based guidance. The decision is tied to observable variables: no rises while unemployment is above a given level and projected inflation does not exceed a given ceiling. The Fed used this format in 2012 with an unemployment threshold of 6.5%.
- State-contingent or data-dependent guidance. The reaction function is spelled out without committing to dates or figures: if the economy does this, we will do that. Less powerful and far more flexible.
There is also a useful distinction between guidance that is pure forecast ("we think the economy will do this, and therefore rates will do that") and guidance that is commitment ("we will tie our hands and keep rates low even when our own rule says they should go up"). The second is far more powerful and far more dangerous: it works precisely because it is costly to break.
Where to read the expectation already in the price
You do not need to guess what the market is discounting. It is published:
- Policy rate futures imply a probability for each outcome at each meeting. This is the usual reference when the press says the market assigns a given percentage to a rise.
- Overnight index swaps let you read the expected path for any horizon.
- The dot plot, published quarterly by the Fed, shows where each committee member sees the policy rate in the coming years. It is neither a promise nor a vote, and mistaking it for a commitment is a classic error.
The operational consequence is uncomfortable for an investor: you do not make money by getting right what the central bank will do. You make money by getting right where you depart from the consensus, and being correct.
The price of being wrong once you have committed
Forward guidance works on credibility, and credibility is a deposit that gets spent. In December 2021 the ECB's president described a rate rise during 2022 as very unlikely; the ECB raised in July 2022 and went on to deliver the fastest tightening cycle in its history. Across the Atlantic, the Fed maintained for months that inflation was transitory and ended up retiring the word late in 2021.
The cost is not public embarrassment. It is that, next time round, the market discounts the words with a skepticism factor, and moving the curve then requires shouting louder or actually acting.
Real case: the taper tantrum of May 2013
By May 2013 the Fed had been buying bonds for years and its policy rate was pinned at zero. On May 22, Ben Bernanke remarked to Congress that in upcoming meetings the committee might begin to slow the pace of purchases if the economy improved.
It did not raise rates. It did not sell a single bond. It did not reduce purchases: it merely said it might, later on, reduce them somewhat.
The reaction was violent. The yield on the American ten-year bond went from close to 1.6% to brushing 3% over the summer, almost double in a matter of months. The blow was far harder outside the United States: capital that had gone looking for yield in emerging markets turned around and punished a group of currencies the financial press christened the fragile five, with the Indian rupee marking record lows against the dollar. The actual slowing of purchases did not begin until December 2013, seven months later.
It is the cleanest experiment there is on this module's thesis: a sentence about the future moved markets more than most executed decisions, and the damage concentrated in long duration assets and in economies that depended on money borrowed in dollars. The next module deals with that dependence.