I have a new working paper with Josh Hendrickson titled Nominal GDP Targeting and the Taylor Rule on an Even Playing Field. Using a standard New Keynesian model, Josh and I show one of the advantages of a nominal GDP target is that it is more robust to the knowledge problem facing central bankers than is the standard approach that implicitly invokes some kind of Taylor rule.
Here is an excerpt from the paper on the knowledge problem:
One of the key challenges facing monetary policy authorities is the knowledge problem. As first noted by Hayek (1945), this problem arises because the information needed for optimal economic planning is distributed among many individual firms and households and therefore outside the knowledge of a central planning authority. This observation, when specifically applied to central banking, means that the information required to make activist countercyclical policies work is not available. Consequently, monetarists like Friedman (1953, 1968), Brunner (1985), and Meltzer (1987) argued early on against central bank discretion and instead called for simple rules that committed monetary authorities to stable money and nominal income growth.
The knowledge problem was later shown by Orphanides (2000, 2002a, 2002b, 2004) to apply not only to central banks that conduct discretionary monetary policy but also to ones that follow a “constrained discretionary” approach to monetary policy. That is, even central banks that follow some kind of Taylor rule in a flexible inflation-targeting regime are susceptible to the knowledge problem...
The biggest information challenge comes from attempting to measure the output gap in real time. The output gap is the difference between the economy’s actual and potential level of output and is subject to two big measurement problems. First, real-time output data generally get revised and often on the same order of magnitude as the estimated output gap itself. Second, potential output estimates are based on trends that rely on ever-changing endpoints. Orphanides finds the latter problem to be the biggest contributor to real-time misperceptions of the output gap. This means that even if real-time data improved such that there were fewer revisions, there would still be a sizable problem measuring the real-time output gap.
To illustrate these problems, figure 1 replicates Orphanides’s (2002b) construction of real-time output gap measures using vintage real output data and compares them to final output gap measures using the Hodrick-Prescott and Baxter-King filters…
The top panel in figure 1 shows both the real-time and final output gap measures. To help discern how different these measures are, the second row plots the real-time output gap misperceptions, or the difference between the real-time and final output gaps. Both the HodrickPrescott and the Baxter-King filters reveal sizable measurement problems, particularly in the 1970s. The Hodrick-Prescott filter shows real-time output gap misperceptions reaching as much as 5 percentage points, while the Baxter-King filter shows up to 2 percentage points in the 1970s.
Orphanides (2004) sees these large measurement errors as a key contributor to the unmooring of inflation in the 1970s. He shows that, if the real-time estimates of the output gap and inflation from the 1970s are plugged into a Taylor rule like equation... the result is pretty close to the actual monetary policy that occurred during this time. The Great Inflation, in other words, was not the result of the Federal Reserve failing to properly respond to the economic developments of the time. It was the result of the Federal Reserve failing to properly measure the output gap.
Interestingly, figure 1 also indicates that the Great Moderation period of 1984–2007 was characterized by relatively smaller real-time output gap misperceptions. These findings raise questions about the claims of Taylor (1999), Clarida, Galí, and Gertler (2000), and others who see the Federal Reserve’s Federal Open Market Committee after Chairman Paul Volker’s term as more disciplined in its response to inflation. They suggest, instead, that Walsh (2009, 216) may be correct in his assessment that the success of targeting inflation has more to do with the “good luck” coming from a “benign economic environment” than from improved monetary policy.Below is Figure 1:
P.S. Here is an earlier post on the knowledge problem facing central bankers when they try to divine changes in the inflation rate.