In a speech last week Fed Chair Janet Yellen made use of policy rules, and in particular the Taylor rule, to explain her views on normalizing policy. This comes on the heels of Fed Vice-Chair Stanley Fischer’s reference to the Taylor rule in a speech earlier in the week, two influential Bloomberg View columns by Clive Crook (here and here) making the case for the Fed to use such rules, the Shadow Open Market Committee’s unanimous recommendation to use rules that way, and continued discussion of a bill in Congress which would require the Fed to state its rule and compare it with a reference rule.
In fact, Janet Yellen’s discussion of how current and upcoming policy might differ from a reference rule (the Taylor Rule) is not unlike what you might see if the policy rule legislation under consideration in Congress became law. So one can think of her discussion as sort of test drive or trial run. If so, it raises a number of questions.
Let me first quote from the relevant section of Janet Yellen’s speech starting on page 7 and embedding an important explanatory footnote at the end:
“Even with core inflation running below the Committee’s 2 percent objective, Taylor’s rule now calls for the federal funds rate to be well above zero if the unemployment rate is currently judged to be close to its normal longer-run level and the “normal” level of the real federal funds rate is currently close to its historical average. But the prescription offered by the Taylor rule changes significantly if one instead assumes, as I do, that appreciable slack still remains in the labor market, and that the economy’s equilibrium real federal funds rate–that is, the real rate consistent with the economy achieving maximum employment and price stability over the medium term–is currently quite low by historical standards. Under assumptions that I consider more realistic under present circumstances, the same rules call for the federal funds rate to be close to zero…
“For example, the Taylor rule is Rt = RR* + πt + 0.5(πt -2) + 0.5Yt, where R denotes the federal funds rate, RR* is the estimated value of the equilibrium real rate, π is the current inflation rate (usually measured using a core consumer price index), and Y is the output gap. The latter can be approximated using Okun’s law, Yt = -2 (Ut – U*), where U is the unemployment rate and U* is the natural rate of unemployment. If RR* is assumed to equal 2 percent (roughly the average historical value of the real federal funds rate) and U* is assumed to equal 5-1/2 percent, then the Taylor rule would call for the nominal funds rate to be set a bit below 3 percent currently, given that core PCE inflation is now running close to 1-1/4 percent and the unemployment rate is 5.5 percent. But if RR* is instead assumed to equal 0 percent currently (as some statistical models suggest) and U* is assumed to equal 5 percent (an estimate in line with many FOMC participants’ SEP projections), then the rule’s current prescription is less than 1/2 percent.”
So the main argument is that if one replaces the equilibrium federal funds rate of 2% in the Taylor rule with 0%, then the recommended setting for the funds rate declines by two percentage points. The additional slack due to a lower natural rate of unemployment is much less important. But little or no rationale is given for slashing the equilibrium interest rate from 2% percent to 0%. She simply says “some statistical models suggest” it. In my view, there is little evidence supporting it, but this is a huge controversial issue, deserving a lot of explanation and research which I hope the Fed is doing or planning to do.
If one can adjust the intercept term (that is, RR*) in a policy rule in a purely discretionary way, then it is not a rule at all any more. It’s purely discretion. Sharp changes in the equilibrium interest rate need to be treated very carefully, as Andrew Levin pointed out recently.
Another important issue is that Janet Yellen does not make the argument here that the coefficient on the output gap in the Taylor rule should be 1.0 rather than 0.5 as she has in previous speeches advocating a lower rate. The argument is different here, and no reason for dropping the old argument is given. Perhaps the reason is that the gap is small now, so the coefficient on the gap does not make much difference. This gives the appearance that one is changing the rule or the inputs to the rule to get an answer. I do not think that reason would go over well in Congressional testimony.