Arnold Kling  

Is the Taylor Rule Really a Rule?

Lawrence Lindsey Vs. Christy R... Unchecked and Unbalanced

John Taylor writes,

The Taylor rule says that the federal funds rate should equal 1.5 times the inflation rate plus .5 times the GDP gap plus 1. Currently the inflation rate is about 1.5 percent and the GDP gap is about -5 percent (using the average of the seven estimates of the gap provided in the recent update by Justin Weidner and John Williams).

So he gets that the Fed funds rate should be 0.75 percent.

If you had asked me, I would have used different numbers. I think that core inflation (the CPI excluding food and energy) has been running at about 0.5 percent. I look at an unemployment rate of 9.5 percent and think that is 5 percent above full employment. Using Okun's Law (the 2-for-1 version), that says that GDP is 10 percent below its full-employment level. So I get that the Fed funds rate should be -3.25.

(I asked my high school econ class to read Bernanke's August 27th speech and ask questions. One student asked what Bernanke meant be conventional vs. unconventional monetary policy, and I did not know quite how to answer, since class has just started and I have not explained anything about monetary policy. I think this -3.25 percent calculation would be a good way to introduce the conversation about conventional vs. unconventional.)

The point is not to claim that the assumptions I would have used are more reasonable than Taylor's. The point instead is to suggest that there is a lot more play in his "rule" than you might otherwise presume. The GDP gap, in particular, is a very elusive fellow to estimate, particularly when you are far from full employment.

Just intuitively, if the inflation rate is at a 50-year low and the unemployment rate is near a 50-year high, it's hard to believe that the monetary dials are set right.

All of the foregoing assumes, of course, that you believe that AD is the problem. I am willing to take a Pascal's wager approach to that assumption, but I personally think that what we have is a huge destruction of labor capital. In Taylor-rule terms, that means that the GDP gap is much, much smaller than the Okun's Law calculation suggests.

Comments and Sharing

COMMENTS (7 to date)
Steve writes:

A Pascal's wager approach works if there is no downside to trying it. However, is the degree of wealth transfer to banks based on monetary policy? Would a looser monetary policy cause a greater wealth transfer to banks?

John Hall writes:

He uses GDP deflator as his inflation measure.
You make a good point about the output gap. The housing crisis could have caused a "one-time" drop in potential GDP that has made the output gap smaller than simply a trend line would suggest.

The thing about the Taylor rule is that most people who talk about it never read the paper where he showed how he got it. He basically uses a model and simulates outcomes using different monetary rules. The Taylor rule is roughly the best one, after rounding. It only applied to the U.S. and most likely didn't include a situation where the zero lower bound would have been reached. Hence, it makes less sense to follow it for other countries or situations that are substantially different than when it was tested.

liberty writes:

Wasn't it you, Arnold, who said you put the "Taylor rule" in the same category as the "don't step on a crack" rule?

That is among my favorite quotes about economics. Sums it up perfectly. (Arbitrary correlations + desire to avoid badness => silly superstitions).

Ted writes:

Is John Taylor serious about using headline inflation? Who in the world would use headline inflation? Headline CPI is far to volatile for use in the conduct of monetary policy. If we conducted monetary policy based on these movements, it would be all over the place and we wouldn't have any sort of stability - it would actually be disastrous. Let's do a hypothetical. You are in the midst of a housing bubble popping, unemployment is ticking up a bit, financial firms are in really, really bad shape. Yet there is a massive spike in oil prices pushing up headline CPI. What would happen if you tightened monetary policy at that point? Well, you would cause a stock market crash, financial houses would be going bankrupt left and right, nominal GDP would plummet, and unemployment would skyrocket. Oh, and when I said "hypothetical" I actually meant 2007 to early 2008 when despite an increasingly weak economy oil prices would rising to extremely high levels. Any so-called "rule" that would tell you to tighten monetary policy in the midst of a recession where credit markets are freezing up and a housing bubble is deflating surely can't be trusted. Everybody knows we use core, although Mankiw and Reis have made a convincing case that any price index should give substantial weight to the nominal wage.

Anyway, I hate Taylor Rules. They are backward-looking; they are limited by the inaccuracy of real-time data; the coefficients are based on statistical regressions and thus do not guarantee optimality; they are not robust as their optimality is based on specific specifications of preferences and technology that may not be correct; they can easily converge to a liquidity trap and become nonsensical once they are in one; they can generate chaotic properties; and they allow no use of judgment which is boneheaded due to the uncertainty about the 'correct' model of the macroeconomy. The only thing Taylor Rules supposedly have going for them is that it reduces the discretion problem, but they actually don't. No central bank mechanically follows a Taylor Rule (mostly because that would be a bad idea). But without another rule to explain why you are deviating from the Taylor Rule you are again back at discretionary policy, just one stepped removed.

These reasons are actually why I support forecast targeting as advocated by Lars Svensson. Preferably nominal income, but I'd even settle for inflation forecast targeting.

Joey Donuts writes:


Take a look at the Fed's open market operations during the oil price hikes. The were net sellers of US securities during May and the summer of 2008. Could it be they were following Taylor's rule.
Were they idiots better yet are they idiots?

Noah Yetter writes:

Say it with me, class: Goodhart's Law

John Hall writes:

The Taylor rule was not based on statistical regressions. It's just that many people who use policy rules, create them with statistical regressions.

Comments for this entry have been closed
Return to top