The juggler

Monetary theory as a circus act

In a perfect world, I would hardly ever mention the term “interest rates”. It’s not a particularly important concept in macroeconomics, where variables such as nominal GDP, average wages and employment are the key factors. But I don’t live in a perfect world; I live in a world where pundits are obsessed with interest rates. To get my ideas across, I’m forced to “speak Keynesian”, just as a traveler to a remote country needs to learn the local language.

Unfortunately, the language of interest rates is vastly more complicated than the language of NGDP growth. To understand monetary economics through the interest rate lens, we need to grasp four distinct concepts. Each concept is non-obvious, and we must hold all four in our minds at the same time when evaluating any given monetary policy claim. It’s sort of like a circus juggler keeping four balls in the air at the same time:

1. What matters is not “the interest rate”, rather it is the relationship between three very different interest rates: the Fed target rate, longer-term market interest rates, and the unobservable neutral interest rate. The neutral rate is often called the “natural” or “equilibrium” interest rate. There is a neutral real interest rate and a neutral nominal interest rate. I’ll focus on nominal rates.

2. The Fed does not move its target interest rate around at random. Rather they generally try to follow the neutral rate of interest. For this reason, movements in the Fed target rate tell us little or nothing about the movements in the stance of monetary policy. But the Fed also make mistakes, in some cases quite serious mistakes. They don’t always follow the neutral rate.

3. The Fed can directly change its target interest rate, and the Fed can also take actions that effectively move the neutral interest rate. Contrary to popular opinion, Fed-induced changes in the neutral interest rate are far more important than changes in the target interest rate.

4. The Fed is usually behind the curve. This means that periods of “tight money” are usually characterized by falling interest rates, and periods easy money are usually characterized by rising interest rates. This is the opposite of what most people would infer from reading the financial news. If Fed policy were efficient, then the information content of Fed interest rate changes would be zero, at least with respect to its effect on aggregate demand. This is the well-known “thermostat problem”. Nick Rowe provides the best explanation that I’ve seen.

At the end, I’ll apply this juggling framework to the question of whether the Fed should cut its target interest rate.

Photo by alexey turenkov on Unsplash

1. A simple model of the neutral interest rate

Here’s a November 2024 article on Jay Powell’s view of monetary policy:

"The economy is not sending any signals that we need to be in a hurry to lower rates. The strength we are currently seeing in the economy gives us the ability to approach our decisions carefully," Powell explained. "Ultimately, the path of the policy rate will depend on how the incoming data and the economic outlook evolve."

In a question-and-answer session with Washington Post columnist Catherine Rampell, Powell was asked about how the Federal Reserve knows when it has reached the neutral rate and his past comments that "we know it by its works."

Powell explained that when the Fed moves the benchmark federal funds rate up or down, it has to have an "estimate of something that's kind of neutral, a level of interest rates that's neither pushing the economy up and supporting it or dragging it down, which would be tighter, restrictive policy."

Because this is such a complex and counterintuitive subject, I’ll illustrate the concept with a model that is simpler than the actual economy. I’ll assume that the neutral (nominal) interest rate is equal to the expected rate of nominal GDP growth. In the real world, the two are correlated, but the correlation is far from perfect. In practice, other things also matter, including (among other things) the level of NGDP relative to trend, and the budget deficit.

The higher the rate of expected NGDP growth, the stronger the correlation with the neutral interest rate. A good example occurred between the 1950s and the 1970s, when NGDP growth accelerated to double digit levels and nominal interest rates rose to double digit levels. In that case, it was clear that causation ran from higher NGDP growth expectations to higher nominal interest rates.

So what did Powell mean by “we know it by its works”? He is implicitly suggesting that accelerating NGDP growth is a sign that the interest rate target is currently set below the neutral rate—indicating an expansionary or accommodative policy stance, whereas slowing NGDP growth is an indication that the interest rate target is set above the neutral rate, indicating a contractionary or restrictive policy stance. In that case, why not just skip interest rates entirely and focus on NGDP growth? Good question.

2. Target interest rates are largely endogenous

So now we have one ball in the air. What matters is not so much the Fed’s interest rate target, rather what matters is whether the target interest rate is set above or below the neutral interest rate. This is not particularly controversial, at least among economists with expertise in monetary policy. If the neutral interest rate were stable, we could infer that a cut in the Fed’s target interest rate would represent an easing of monetary policy and an increase in interest rates would represent a tightening of monetary policy.

In fact, the neutral rate of interest is not stable. Indeed, in my view it is considerably more unstable than many economists assume. Even so, if Fed policy changes were completely random, then on average a reduction in the Fed’s interest rate target would represent easier money, and vice versa. But Fed policy is not random. The Fed tries to produce an outcome that avoids large swings in aggregate demand. To do this the Fed must reduce its target interest rate any time there is a decline in the neutral interest rate, and vice versa. This is what Powell meant when he said:

"Ultimately, the path of the policy rate will depend on how the incoming data and the economic outlook evolve."

If the incoming data suggests strongly rising NGDP growth, the Fed will generally assume that the neutral interest rate is increasing. It will respond by raising its target rate. This means that the Fed often tends to follow the bond market. But it’s also clear that the Fed frequently makes mistakes, and NGDP growth often ends up higher than the Fed would prefer (the 1970s and 2021-23) or lower than the Fed would prefer (1929-33 and 2008-14). One should never assume that declining interest rates are a sign that the Fed is easing.

3. The Fed also affects the neutral interest rate

So now we have two balls in the air. What matters is the relationship between the policy target and the neutral rate, and the policy rate often (but not always) responds to changes in the neutral rate. Thus far I’ve treated changes in the neutral rate as an exogenous shock, a random event that strikes the economy like a thunderstorm. But perceptive readers will have already noticed a problem. If the neutral rate is highly correlated with the expected rate of NGDP growth, and if Fed policy impacts NGDP, then Fed policy should also have a powerful effect on the neutral rate of interest. I believe that monetary policy is the dominant factor affecting the neutral rate.

Previously, I discussed how the rapid NGDP growth of the 1970s led to high nominal interest rates. But that NGDP growth didn’t just happen in a vacuum; it was caused by Fed policy, which was highly expansionary from the mid-1960s to 1981.

Suppose we start from a position where the economy is in equilibrium. NGDP is growing at a rate consistent with the Fed’s policy goals (which might be something like 2% inflation and high employment), and the Fed’s current interest rate target is equal to the neutral interest rate. Now assume that the Fed just cannot stand success and decides to cut the target interest rate to a level below the neutral rate. That represents an expansionary monetary policy, and this would likely lead to faster NGDP growth.

In a world where investors have rational expectation and financial markets are efficient, this action will also increase the expected future rate of NGDP growth, which will push the neutral interest rate higher. But that would make policy even more expansionary, which would tend to cause NGDP growth to accelerate even further. Rinse and repeat. Eventually you’d end up with hyperinflation.

In fact, you almost never end up with hyperinflation, as central banks are not that dumb. Recall from section 2 that the Fed tends to follow the market. They would almost certainly recognize their mistake when they saw the economy overheating (as in late 2021) and begin to tighten monetary policy by sharply raising the target interest rate. Nonetheless, the policy mistake may lead to somewhat higher NGDP growth for a period of time, and this will tend to push market interest rates higher.

While we almost never end up with hyperinflation (or extreme deflation), we do see smaller policy mistakes on occasion. And when these policy mistakes occur, they have the effect of pushing the neutral interest rate in the opposite direction from the Fed’s target interest rate. Because this sort of policy mistake would lead to somewhat higher NGDP growth, this will also tend to push longer-term market interest rates higher, as investors expect the Fed to eventually raise rates to correct its mistake.

If interest rates were the only Fed policy tool, then we’d now be done with section 3. But they are not. Central banks have other tools, including money printing (QE), exchange rate pegs and forward guidance. The Fed doesn’t target the foreign exchange rate but does use QE and forward guidance.

Previously, I said that Fed policy mistakes generally do not lead to hyperinflation, because the Fed corrects the mistake long before the economy spirals into extremely high inflation. This suggests that the effect of Fed policy will partly depend on expectations of future Fed policy, which is why forward guidance is so important. A given quarter point interest rate cut has a very different effect in a world with Fed credibility than in a world where investors have no confidence in the central bank—perhaps fearing “fiscal dominance”.

The Fed can use tools such as QE and forward guidance to shape expectations of future NGDP growth, Those tools can then impact the stance of monetary policy even if the Fed doesn’t move the target interest rate. A quarter point increase in the neutral interest rate is just as expansionary as a quarter point cut in the target interest rate, ceteris paribus.

An extreme example of this phenomenon occurred in Switzerland in January 2015. The Swiss National Bank cut interest rates sharply, but policy nonetheless ended up moving in a more contractionary direction, as the neutral interest rate fell even more sharply on the same day. That occurred because the SNB abandoned a currency peg with the euro, allowing the Swiss franc to sharply appreciate—a strongly contractionary shock to the economy.

A good argument can be made that something similar happened in the US in early October 2008, when the expansionary effect of a reduction in the Fed’s interest rate target was more than offset by the contractionary effect of the new policy of paying interest on bank reserves. Similarly, a December 2007 announcement of a 1/4% decline in the Fed’s target interest rate ended up being contractionary, as the smaller than expected rate cut led to a sharp reduction in the neutral interest rate because investors became more bearish, making policy more restrictive.

In general, people tend to greatly overestimate the impact of changes in the Fed’s target interest rate, and greatly underestimate the effect of forward guidance and/or money printing (QE or QT) that impacts the neutral interest rate. This leads to people often being surprised when the economy is doing well “despite” rising interest rates or doing poorly despite lower interest rates. When it’s doing well despite rising interest rates, the rising rates generally do not constitute a contractionary monetary policy. And that’s because . . .

4. . . . the Fed is usually behind the curve

So now we have three balls in the air. What matters is the relationship between the policy target and the neutral rate, the policy rate often (but not always) responds to changes in the neutral rate, and the Fed can affect the neutral rate by making monetary policy more expansionary or contractionary. Thus far, there’s no reason to assume that policy errors would be anything more than a random walk, changing in an unpredictable fashion at each of the Fed’s eight annual policy meetings. In fact, Fed errors generally are serially correlated.

Unfortunately, we do not have NGDP targeting. Nonetheless, with rare exceptions like the 2020 Covid lockdowns, the Fed would generally like to see fairly stable NGDP growth, consistent with roughly 2% PCE inflation. Sharp declines or large increases in NGDP are almost always viewed as policy mistakes, even by the Fed itself. Powell has acknowledged that Fed policy was too expansionary in late 2021, Bernanke acknowledged that policy was too contractionary in late 2008, and Fed researchers have almost universally characterized Fed policy as being too contractionary in the early 1930s and too expansionary throughout most of 1966-81.

If policy were efficient, then roughly 50% of the time the Fed might be expected to reduce its interest rate by more than the decline in the neutral rate and 50% of the time they would reduce their target rate by less than the decline in the neutral rate. In that case, a decline in the Fed’s target interest rate would provide almost zero evidence as to whether policy was becoming more expansionary or more contractionary. The thermostat problem.

In practice, the Fed tends to be somewhat inertial, moving the target interest rate more slowly than changes in the neutral rate. As a result, policy actually tends to be relatively contractionary during periods of falling interest rates (such as 2008) and expansionary during periods of rising interest rates (such as the 1970s and 2022). Over the business cycle, rates tend to fall during recessions (when policy is usually too tight), and they tend to rise during inflationary booms (when policy is usually too expansionary.)

5. Wrong question

I hope it is clear by now that “what should the Fed do with interest rates” is the wrong question. In January 2015, the SNB adopted a tighter monetary policy with sharply falling interest rates. Is that what people want when they advocate a rate cut? Probably not. Although the Swiss case was a particularly striking example of a NeoFisherian outcome, it is often true that falling interest rates are associated with a contractionary monetary policy.

But even if, “Where should interest rates be set?” is the wrong question, it’s hard to avoid the topic if interest rates are the universal language of monetary policy discourse. Unfortunately, to evaluate policy we need a policy objective that is a bit more definitive than, “the Fed likes fairly stable NGDP growth.” And the Fed’s recent framework review did not provide any sort of specific guidance as to the Fed’s policy goals.

If we assume a 2% PCE inflation target, then there’s no obvious reason to currently advocate a rate cut. But markets are pricing in a rate cut this fall, and TIPS markets don’t seem to forecast much more than 2% PCE inflation plus a one-time 1% price level bump due to tariffs. So it’s also not obvious that a 1/4% cut would be a bad idea.

Long time readers know that my objection to Fed policy since 2020 is not about interest rate settings, it’s the failure of the Fed to do the average inflation targeting that it promised. It’s hard to give policy advice if you don’t even know what “average inflation targeting” actually means. In my view, the upcoming rate decision itself is not very important, what matters is whether the Fed is committed to maintaining 2% inflation, on average, in the long run.

This morning, I saw a financial analyst on Bloomberg TV advocating that the Fed assist the Treasury with an easy money policy. Unlike interest rate decisions, I do have strong views on using monetary policy to ease debt burdens. It’s a bad idea, at least when nominal interest rates are positive.

PS. As is now the case with all policy issues, and many non-policy issues, the discussion inevitably turns to what the president wants. I view that as a distraction. What the president wants most of all is power and conflict. He’ll support any monetary policy that leads to those objectives. Thus, he’ll support low interest rates when he’s in power and oppose low rates when his opponents are in power. People who analyze presidential statements from a traditional analytical perspective don’t understand the game he’s playing.

As an analogy, when TikTok was popular back in 2020, Trump tried to ban it. After Congress banned it, he refused to enforce the law. “Is allowing TikTok to operate good public policy?” is a question that has no meaning to Trump. “Should a public official that lied on a loan application resign?” has no meaning to Trump. “Should Covid vaccines be encouraged?” has no meaning to Trump. He doesn’t think in terms of optimal public policies. Everything is conditional—what’s in it for him at this moment in time?