Where is the Fed’s interest rate heading?

Financial markets have capitulated in their expectations of US interest rate cuts this year. At the start of January, they expected at least six quarter-point cuts by the end of 2024. Now they are only sure about one. Traders are even building up bets that the next move might be up.

Borrowing costs at the end of 2024 are therefore highly uncertain. Do we know anything more about where interest rates in the US and elsewhere are ultimately heading?

Economists define this destination as R-star or the neutral rate of interest. This is the rate that balances desired savings and investment and would apply when output is at an economy’s potential and inflation was at target. It cannot be seen but can be estimated.

We know R-star occupies the Fed because chair Jay Powell said last year he was “navigating by the stars under cloudy skies”. The neutral rate was unknown, he said, but the Fed policy rate then (and now) of between 5.25 per cent and 5.5 per cent was restrictive and “well above mainstream estimates of the neutral policy rate”.

Even though the median Fed official estimates R-star to be 2.6 per cent, the important question with a resilient US economy and higher than expected inflation is whether the committee can know this. The dilemma also applies to all other central bankers.

The easy days

There was a time when all estimates of the neutral rate, whether they were based on market prices, structural models or something in between, gave similar answers. R-star had been high and declined until the pandemic.

The following well-known chart shows market rates for 10-year government bonds in the US, eurozone, Japan and UK between 1980 and the pandemic in 2019. This can be seen as a very rough proxy for R-star and other economic estimates gave similar results.

Economists and policymakers told many stories explaining the decline in nominal and real interest rates including the control of inflation, secular stagnation, a savings glut from Asian economies, rising inequality increasing savings and lower productivity and growth rates reducing the desire for investment.

Borrowing costs have risen since the pandemic, complicating this story and raising the question to what degree has R-star risen too? Do these stories still make sense or have we moved into a persistently higher interest rate world?

A world of disagreement

I am going to simplify here, but we have a big divergence. Market data, as above, suggests neutral interest rates have risen sharply. These are replicated in some much more sophisticated estimates of R-star. The alternative view, generated by more structural modelling, such as the IMF’s estimates or those from the New York Fed, say that nothing has changed. Soumaya Keynes wrote a fabulous piece about the differences in modelling last year.

The broad reasoning behind the difference of view is that methods of estimating R-star which show little change amplify the importance of long run slow-moving concepts such as the potential growth rate or demographics, while those showing the world has changed put more weight on financial market prices.

This would be difficult enough if it was not for horrible estimation problems and fundamental uncertainties over the economic meaning and drivers behind R-star.

Nightmares for economic researchers

Estimating things using econometrics is difficult. There are always data problems. These are extreme when it comes to R-star.

In March, for example, the Bank for International Settlements noted that the one standard deviation interval (in which the true value will sit only 68 per cent of the time) around the Richmond Fed R-star estimate for the US was 1.73 percentage points wide. It was 3.2 percentage points for the New York Fed model. For the eurozone, these blew out to totally unusable figures of 5.6 and 12 percentage points wide.

In other words, the New York Fed model was reasonably confident that nominal eurozone interest rates were heading to a value between -4.7 per cent and 7.3 per cent. Brilliant.

The uncertainty is in many ways worse than that. These models do not even agree with past versions of themselves, partly because the main determinant of R-star in the model is the potential growth rate — itself a made up number.

The New York Fed is extremely transparent about its results and publishes these every quarter. I have collated all of its data to produce the animated chart below.

When you play the animation, the same model for the same period gives different results depending on when it was estimated. This is partly down to revisions in economic data itself, a regular difficulty for economies. More recently, model specification changes also matter and show that the model is unstable post pandemic for all countries. The New York Fed has stopped publishing UK results.

If you are a central banker working with one number in mind for the destination of interest rates and then it changes significantly in new estimates, you have a problem. A big one.

Even more fundamental models such as those produced by the IMF are no more reliable, sadly. As BIS head of economics Claudio Borio and colleagues documented in 2017, the long-run forces, which do a good job in explaining the decline in R-star between 1980 and the 2010s, did not work in previous periods.

That suggests strongly that the model is wrong. This point was made forcefully in a recent lecture at the US Monetary Policy Forum by Harvard professor and former Fed governor Jeremy Stein.

Nightmares for central bankers and markets

That is not the end of the difficulty in searching for R-star because market data has a serious problem too. The big decline in all interest rates between 1980 and 2019 did not happen evenly.

A remarkable paper by Sebastian Hillenbrand shows that all of the decline since 1989 came in the three-day period around Fed meetings. At all other times there was no decline in borrowing costs at all.

If you do not believe this, the ECB has replicated Hillenbrand’s data and the chart below shows both sets of data being essentially identical.

What this means is that the Fed is clearly important in influencing the market proxies of R-star. This means if officials look to the market to get an estimate of R-star, they will be looking back at their own actions. They are simply looking at the mirror.

More fundamentally, the research challenges the whole concept of R-star because it is theoretically supposed to be an anchor set by fundamental forces balancing savings and investment.

If, instead, Fed policy influences R-star directly, the honest thing to do would be to think about binning the concept of an interest rate we are trending towards. The same logic applies if governments influence R-star with their stance on fiscal policy.

Well before Powell used the cloudy analogy, Borio said: “When the sky is cloudy, it is hard to see where the stars are and even how many there are.”

Others have rationalised this result with less catastrophic consequences for central bankers. ECB executive board member Isabel Schnabel said that although the results were “puzzling”, the results would not be difficult to explain if “financial markets . . . ultimately look to the central bank for information about the long-run evolution of the economy”.

This is forward guidance on steroids and suggests financial markets have few ideas of their own apart from what they learn from the august central bankers.

But there are other theories. Perhaps central banks are important in shaping R-star with their actions not words. If low interest rates create zombie companies that do not invest, it can lower R-star and be self-reinforcing. Or as Stein mused, perhaps low interest rates had become addictive and no longer were as stimulative as before.

The upshot

It is clear that numerical estimates of R-star are not worth much even if economists need to think about the concept. We do not know where the Fed or other central bank interest rates are heading. We have lost the anchor we thought we had, but probably never did.

That does not mean economic analysis of interest rates is hopeless. Central banks need to constantly assess whether the monetary policy they set is expansionary or contractionary in the short term. There are many measures they can look at to see how the transmission mechanism is working.

What they should not do is pluck some R-star number from a model and give it a credence it does not deserve.

What I’ve been reading and watching

A chart that matters

While almost everyone thinks the Fed now needs to pause a bit longer and assess the inflationary outlook, Tej Parikh, our economics leader writer, thinks everyone is freaking out unnecessarily. He is rather disappointed that Powell also now thinks more time is needed to see disinflation, which he thinks is still on track. In one important chart, he shows how US wage growth is still moderating nicely.