Taylor Rules on the Taylor Rule

The rule for nominal interest rate setting that John Taylor proposed in his 1993 paper “Discretion versus Policy Rules in Practice“, Carnegie-Rochester Conference Series on Public Policy 39, 195-214, has had an enormous influence in the macroeconomics profession.  It is safe to say that numerous economists, practitioners and academics alike, since that paper have evaluated monetary policymaking using the Taylor rule as some kind of reference point. Empirically, a plethora of papers have estimated coefficients of Taylor-type rules for different countries during different periods. Theoretically, paper after paper on monetary policymaking adopt some form of the Taylor rule as a default specification of monetary policymaking (even undergraduate text books routinely incorporate a Taylor rule as part of the determination of the economy’s AD curve). In practical monetary policymaking, some authorities assess policies relative to what a Taylor rule would have ordered.

Because of its heavy usage in so diverse settings, it has to some gradually become slightly unclear what John Taylor actually meant the rule should represent when he proposed it (this confusion, I am not afraid to admit, has also cursed me). Was it meant as a descriptive or normative concept? In the original paper, he shows, in a now famous Figure 1, how the “example policy rule” tracks the actual US Federal Funds Rate 1987-1992 “remarkable well”. However, he did not choose his rule with the aim of fitting the data. In any case, this may also have contributed to confusion about the proper, and original, interpretation of the rule.

Luckily, there is no need to speculate anymore, as John Taylor himself has now ruled out any potential confusion. On his blog, “Economics One,” he writes in the post “Misunderstanding Prescriptive Versus Descriptive Monetary Policy Rules”:

“. . . the Taylor rule was not meant to be descriptive as I made clear in my original paper. Rather it was very explicitly meant to be prescriptive. I derived it by experimenting with different types of rules in stochastic simulations of different monetary models, including my multi-country model at Stanford, and by studying the results of other people’s simulations. This pinned down the left-hand side variable and the right-hand side variables, and led to simple functional forms and coefficients”

This should clear out any misunderstandings. The following is a prescription for monetary policymaking:

r =  p + 0.5y + 0.5(p-2) + 2,

where r is the nominal (policy-) interest rate, p is the rate of inflation over the past four quarters, y is output’s percentage deviation from trend, and “2” indicates both the (indirect) percentage inflation target and steady-state real interest rate.

In the original paper, Taylor notes that “there is not consensus about the size of the coefficients of policy rules” but labels his rule a “representative policy rule”. That is fine, as there is obviously nothing magic about the number “0.5” (or the now more famous “1.5”. which is the sum of coefficients on the inflation term). Its “derivation” also seems guided by experimentation on some existing models – which exact criteria were used is not spelled out clearly in the paper, which ultimately makes it appear quite judgmental. And a good prescription for policy should be anything but judgmental. Note, by the way, that the precise numbers in the rule do seem important to Taylor today, as he writes:

“You can’t justify QE2 by saying that the interest rate is negative with the prescriptive policy rule I proposed, because the implied rate is not negative, it’s close to 1 percent”

In any case, I am still confused. How can a “representative” interest-rate setting rule that resembles a couple of years of actual policymaking in the US in six years, which is derived by some experimentation, have generality for other time-periods and other countries? Any exercise in optimal interest rate determination in modern DSGE models gives rise to much more complicated rules than the Taylor rule. Nevertheless, the literature often presents arguments like “the Taylor rule performs well”, and sometimes even backs it up by proper welfare measures. However, in my view this merely reflects the fact that such DSGE models feature price rigidities as the overwhelming (monetary policy-relevant) distortion. Thereby, pure inflation stabilization often does a good job, and a Taylor rule will replicate this well with a high coefficient on inflation. So, the rule’s success may reflect simplicity in the models’ policy tradeoffs. This point is clear from some of the work by Stephanie Schmitt-Grohé and Martín Uribe, where they show that the principal role of monetary policy in a common DSGE model is to secure a determinate low-inflation outcome (see, e.g., their 2007 “Optimal Simple And Implementable Monetary and Fiscal Rules,” Journal of Monetary Economics 54, 1702-25).

Now, adherence to rule-based behavior is of course advantageous in face of time-inconsistency problems of monetary policy. However, for a rule to work, it should be credible, i.e., policymakers should stick to it; also when it hurts. On this point the Taylor rule still confuses me. Given that it is a policy prescription, how come that most of the original paper is devoted to discussions of how to deviate from the rule? In the original paper’s concluding remarks it says:

“This paper has endeavored to study the role of policy rules in a world where simple, algebraic formulations of such rules cannot and should not be mechanically followed by policymakers” (from Taylor, 1993, p. 213).

I just don’t get it. A prescriptive rule should not be mechanically followed? But then it is not a complete policy prescription in my view. The terms “rule” and “not mechanically” should simply not appear in the same sentence.

Therefore, despite Taylor’s clarification, I prefer to think about the Taylor rule as a useful descriptive short-hand for policy experiences around the world, as well as a convenient modeling tool to simplify the complications of real-life monetary policymaking.