The Reaction Function Fundamentalists Are Lost in a Maze of Mirrors
Yesterday, we argued that Federal Reserve Chairman Kevin Warsh should resist demands that he publicly articulate the Fed’s monetary-policy so-called “reaction function.” But since many of Warsh’s critics persist in their errant views, it’s worth expanding on that argument.
Although Warsh’s critics say they accept the idea of dropping forward guidance, calling what they want a “reaction function” does not change what is really being demanded. What they want is actually state-contingent forward guidance, a description of how future economic developments will translate into future interest-rate decisions.
The strongest attack on Warsh’s approach comes from Bill Dudley, the former president of the Federal Reserve Bank of New York. Dudley says he agrees that ordinary forward guidance is generally unnecessary when interest rates are not pinned at zero. But he says Warsh goes dangerously too far by refusing to tell markets how the Fed will respond as circumstances change.
Without that information, Dudley argues, markets will do a poor job anticipating monetary policy. Financial conditions will adjust more slowly. And the gap between what markets expect and what the Fed ultimately does will become larger.
Many others have made versions of the same argument. The apparent consensus among Warsh’s critics is that modern monetary policy requires a well-communicated reaction function.
There is one problem with this consensus: monetary policy requires no such thing.
The Fed’s Hall of Mirrors
For more than two decades, economists Stephen Morris of Princeton and Hyun Song Shin, now at the Bank for International Settlements, have developed a serious theoretical challenge to the assumption that more central bank communication necessarily produces better informed markets.
The problem is that the Fed has two potentially conflicting relationships with financial markets. It uses markets to influence the economy, and it uses market prices to learn about the economy.
The more successfully the Fed imprints its own expectations on market prices, the less independent information those prices may contain. Investors want to predict the central bank, but the central bank wants to extract information from investors. At some point, the Fed looks to markets for guidance and discovers its own prior judgment staring back at it.
This is known as the “reflection problem.” Dudley comes remarkably close to describing it himself. He warns that Warsh’s approach will leave “the Fed looking at the markets, the markets looking at the Fed.”
Dudley thinks an explicit reaction function would solve this problem. Morris and Shin argue that it can create the problem.
Their most directly relevant work is a 2018 paper titled “Central Bank Forward Guidance and the Signal Value of Market Prices,” published in the American Economic Association’s Papers and Proceedings. The paper models precisely what Warsh’s critics are demanding: a central bank that announces a state-contingent reaction function.
In their model, the central bank must decide how much weight to place on market prices and how much to place on independent information, such as surveys. Market participants know the reaction function and want to predict the central bank’s next move.
As the central bank places more weight on the market, investors place less weight on their private information and more weight on predicting one another. They know the Fed is watching the market, so they try to anticipate what the market will tell the Fed and how the Fed will respond.
The Reflexive Beauty Contest
In other words, the reaction function creates the classic Keynesian reflexive beauty contest. In Keynes’s famous analogy, contestants win by choosing the face they think everyone else will choose, rather than the face they personally find most beautiful. Likewise, once investors know the Fed is watching market prices, they have an incentive to anticipate how other investors will move those prices—and therefore how the Fed will respond—instead of relying on their own assessment of the economy.
Morris and Shin find that excessive reliance on market signals becomes self-defeating. The more the central bank responds to the market, the less information the market supplies. The optimal policy still makes use of market prices, but the central bank should commit to placing less weight on them than it would be tempted to place after seeing them.
That is very close to Warsh’s stated position. He has said markets can provide useful information while insisting that the Fed will not be constrained by them or take their message verbatim. He wants to avoid fogging up market signals with Fed forecasts and rolling commentary.
Predicting the Fed Is Not Understanding the Economy
The reaction function fundamentalists repeatedly confuse two kinds of accuracy.
A market can become better at predicting the Fed without becoming better at understanding inflation, employment, growth, or financial risk. It may even become worse.
The research by economists Michael Bauer, Carolin Pflueger, and Adi Sunderam that we discussed yesterday shows that beliefs about the Fed’s reaction function can make interest rates much more sensitive to economic news. That proves reaction function beliefs move markets. Morris and Shin show why those larger movements do not necessarily mean markets have become better informed about the economy.
Reaction function talk may be extremely effective at getting markets to follow the Fed. And that effectiveness is part of the danger.
The transitory inflation debacle provides an obvious illustration. In 2021, Fed officials insisted that the inflation surge would soon pass. Bond yields barely responded to accelerating inflation, and market measures continued to show inflation returning toward the Fed’s target. The Wall Street Journal‘s Greg Ip cites this as evidence of the danger of trusting markets. It is better understood as a warning against treating market prices shaped by Fed communication as independent confirmation of the Fed’s forecast.
The market absorbed the Fed’s transitory narrative and its expected reaction. The Fed then looked at anchored market expectations and saw reassurance. It was the monkey looking into the mirror and mistaking its reflection for another monkey’s judgment.
The Missing Half of the Case
Dudley argues that a clear reaction function would allow markets to adjust financial conditions earlier and more accurately. That is plausible. It is also only half of the analysis.
Greater predictability can improve the transmission of monetary policy while degrading the information contained in market prices. It can reduce surprises while encouraging investors to organize their decisions around the Fed. It can align market expectations with official expectations without bringing either closer to economic reality. But if the Fed gets things wrong and investors are following the Fed’s signals, you can wind up with a disastrous result like the explosion of inflation in 2021 and 2022.
The Morris-Shin work establishes the point relevant to the current dispute: monetary theory does not require the chairman to supply markets with an explicit reaction function. There is a genuine trade-off between managing expectations and learning from markets. Warsh’s critics discuss the first side as though the second did not exist.
The reaction function fundamentalists think Warsh is abandoning monetary theory. In reality, he appears to understand a part of it they have forgotten.
Read the full article here