The Taylor rule is the celebrity of monetary policy rules, frequently touted as the go-to framework for central banks, especially the Federal Reserve. Its simplicity is its charm: set interest rates based on how far inflation deviates from the target and how much the economy is straying from its potential output.
However, let’s not kid ourselves into thinking the Taylor rule is the Holy Grail of monetary policy. It’s more of an empirical snapshot of what the Fed has done in the past rather than a universal panacea.
Enter the Mankiw rule—a somewhat underappreciated cousin in the monetary policy family. Greg Mankiw introduced this gem in his paper “Monetary policy during the 1990s”.
The appeal of the Mankiw rule is its simplicity. Mankiw argues that the Federal Reserve during the 1990s adjusted the federal funds rate based on the difference between core inflation (measured by the Personal Consumption Expenditures, or PCE, index) and the unemployment rate.
Yes, you heard that right: inflation and unemployment, which of course aligns with the Fed’s dual mandate.
Mankiw’s original formula is:
Federal funds rate = 8.5 + 1.4 (Core inflation – Unemployment)
In Mankiw’s paper, he states that “The parameters in this formula were chosen to offer the best fit for data from the 1990s”, but as the graph below shows, some slightly better parameters could probably have been chosen.
Obviously, Mankiw never argued that this was a perfect fit or even that it had been statistically estimated, but that’s not really important. We see from the graph that the rule fairly well captures the ups and downs in the Fed’s policy rate from the late 1980s to the early 1990s.
And the model is very simple – and that to me is certainly part of the charm of the model. The Fed’s two mandates expressed in a simple fashion pretty much dictated during the 1990s what the Fed did with interest rates.
A Slightly More Complicated Story
While the simplicity of the Mankiw rule is its main attraction, it is also the main problem with the rule, and it is in no way given that the structures in the US economy are stable over time.
I would particularly stress three factors that have not been stable over time and which are likely not to be stable over time.
Firstly, the Federal Reserve has changed its monetary policy targets numerous times over time before eventually implementing a 2% inflation target over the past decade.
Secondly, for the Fed it should be important to realise that looking at unemployment as a measure of inflationary pressures would be wrong if it fails to take into account that structural unemployment has changed over time.
Thirdly, the real natural interest rate, which is closely related to potential real GDP, has also changed over time.
Therefore, to fully understand why the Mankiw rule actually makes sense, we need to incorporate these factors in what I here call a Theoretical Mankiw Rule.
Now the Mankiw rule is somewhat less simple, but we can make it simple again.
Hence, if we look at the period from 1985 until 2000, then we can make assumptions about the variables in the theoretical Mankiw rule.
Now the Mankiw rule is somewhat less simple, but we can make it simple again.
Hence, if we look at the period from 1985 until 2000, then we can make assumptions about the variables in the theoretical Mankiw rule.
The Congressional Budget Office (CBO) makes estimates for potential real GDP growth and NAIRU, and I have used these estimates, around 3.5% and 5.5% respectively, and it is probably fair to assume that the Fed’s de facto inflation target at the time was 3%, and if I at the same time assume α equal to 1.5 (close to Mankiw’s assumption of 1.4), then I can insert these values into our Theoretical Mankiw Rule, and then I get the following rule:
Hence, we are now back at a simple version of the Mankiw rule, but based on a clearer definition of the theoretical foundation – and as we see from the graph below, this Theoretical Mankiw Rule (the red line) with these parameters gives a much better fit to the actual Fed funds rate than the original Mankiw rule does.
The Mankiw Rule: Has It Stood the Test of Time?
But how have the Mankiw Rule – and the Theoretical version – done over time? Does the Mankiw rule still provide useful guidance on what the Fed should do with interest rates, and can we use the Mankiw rule to predict Fed policy changes?
The graph below shows both versions of the Mankiw rule and the actual Fed funds rate since 1960.
The first thing we notice is that the two versions of the Mankiw rule – the original one and the theoretical one – are remarkably similar over time. Secondly, we see that both models have done a fairly good job explaining the ups and downs in the Fed funds rate over time – particularly until the major shocks in 2008-9 and 2020-22.
Somewhat disappointingly for somebody who wants to have a theoretical explanation as well as an empirical explanation, the original Mankiw rule does a slightly (and it is really slightly) better job in terms of correlation and fit (measured by R2 and MSE) in explaining the development in the Fed funds rate.
That said, I feel a lot more comfortable using a policy rule for analysis and prediction when I understand it and it is not just a black box.
Both Versions of the Mankiw Rule Are in Agreement – It Is Time for the Fed to Cut Rates
One thing is how the Mankiw Rule has done historically – another thing is what it is saying now and what it predicts for the future.
Let’s first zoom in on developments in recent years.
As the graph above shows, both versions of the Mankiw rule have had a hard time explaining why interest rates were as low as they were until 2020, but this is not really what we are interested in here as there is general consensus that numerous factors such as negative demographics, financial regulation and likely also a de facto acceptance from central banks around the world that inflation was going to undershoot inflation targets without central banks taking any actions to do anything about this.
But the important thing is what happened from 2020 onwards.
In early 2020, the lockdown shock hit the US (and global) economy and caused unemployment to skyrocket. As a consequence, the Mankiw rule(s) stipulated that the Fed should cut rates strongly. The Fed did this and once it hit the Zero Lower Bound, it started – in my view correctly – to undertake massive quantitative easing of monetary policy.
However, as we also see, the US economy responded swiftly to the reopening of the economy as well as the massive monetary and fiscal stimulus. As a consequence, unemployment dropped very fast in the second half of 2020 and the Mankiw rules are quite clear – the Fed should have started the interest rate hiking cycle already in 2021. The Fed, as we know, didn’t do that and that in my view is the main reason inflation shot up sharply starting already in 2021.
However, in 2022 the Fed started to undo its mistakes and the Fed funds rate was hiked significantly, and during 2022 and 2023 the actual rate and the Mankiw rules have approached each other – and presently mid-2024 the theoretical Mankiw rule is basically exactly at the same level as the actual Fed funds rate.
But it is also notable that the theoretical Mankiw rule is slightly closer to the actual policy rate than the original Mankiw rule is.
An Updated Mankiw Rule
But if we want a simple rule of thumb, we really would like an updated version of the simple Mankiw rule.
We can calculate that by using the theoretical version of the Mankiw rule and assuming that the inflation target is 2%, that NAIRU is 4% and potential real GDP growth is 2.5%.
Then we get the following reduced form for the Mankiw rule:
We then assume that (core) inflation will have returned to 2% by the end of the year (we are already quite close) and that unemployment will have increased slightly to 4.5% also by the end of 2024.
Then we can calculate a “target” for the Fed funds rate over the next 6-12 months:
The current federal funds rate is set in the range of 5.25% to 5.50%, which means that the (updated) Mankiw rule indicates that the Fed should cut rates by as much as 1.50 percentage points over the coming year.
That could sound like a lot, but this is in fact extremely close to actual market pricing as illustrated by the table below from CME Group.
As we see, the most likely scenario (32.4%) is that the Fed will cut its key policy rate to 3.75-4.00 over the coming 12 months. Exactly as indicated by the updated Mankiw rule.
In conclusion, even though the Mankiw rule is extremely simple, it remains a rather useful—and easy to use—tool for assessing and predicting US monetary policy. This is especially true, in my view, if you use the updated parameterization of the rule.
The crucial question going forward is, of course, whether we will return to the “predictable 1990s” or if we will once again be hit by major economic and financial shocks as we have seen since 2008. I am optimistic that these are very rare events and that it is likely that we are, in fact, returning to something similar to the Great Moderation that lasted from 1985 until 2008—and returned again in 2010-2012 until 2020.
If that is the case, we might find it useful to have an updated Mankiw rule in our toolbox.
PS I have build a simple Mankiw Rule Calculator using artifacts in Claude 3.5 Sonnet. You can test the Calculator here. With the calculator you can make your own predictions for the Fed funds rate based on your expectations for macroeconomic variables in the US.
