Monetary Policy Rules Based on Real-Time Data

This 2001 American Economic Review paper by Athanasios Orphanides asks how large the informational problem is when Taylor-type interest rate rules are implemented with the data actually available to policymakers in real time, rather than with the ex post revised data conventionally used in the academic literature, and whether reaction functions estimated on revised data give an accurate description of historical policy. Orphanides builds a real-time data set from Federal Reserve Greenbook staff forecasts prepared for each FOMC meeting from 1987:1 to 1992:4 (extended to 1993:4 for forward-looking specifications, for a total of 24 and 28 quarterly observations respectively), pairing within-quarter Greenbook estimates of inflation and the output gap against later-revised vintages from 1993:1 and 1994:4, evaluated under both Taylor’s original linear-trend output-gap concept and the Federal Reserve staff’s Q* concept of potential output. He first evaluates a standard Taylor rule (r*=2%, π*=2%, coefficients 1.5 on the inflation gap and 0.5 on the output gap) quarter by quarter under each data vintage, then estimates backward- and forward-looking partial-adjustment policy reaction functions by OLS/IV. The real-time output gap is on average far more negative than the final revised estimate (-1.25% versus -0.23%), and the discrepancy between the actual federal funds rate and the Taylor-rule prescription varies sharply with both the data vintage and the output-gap concept used (Taylor’s original linear-trend concept versus the Federal Reserve staff’s Q* concept): real-time (within-quarter) data implies the Fed was on average 96 basis points too tight relative to the rule under the linear-trend concept and 123 basis points too tight under the Q*-concept (both significant at 1%), the 1993:1 revised vintage implies a statistically insignificant -2 basis points under the linear-trend concept but 40 basis points too tight under the Q*-concept (significant at 1%), and the 1994:4 revised vintage implies about 20 basis points too easy under the linear-trend concept (significant only at the 10% level) but 27 basis points too tight under the Q*-concept (significant at 5%) - with a standard deviation of 0.58 percentage points and discrepancies approaching 200 basis points at times. Revisions to the output gap and inflation grow substantially over the year following a Greenbook forecast, and characterizing the revisions as either “news” (rational forecast updates) or “noise” (measurement error), Orphanides finds the correlation pattern is more consistent with noise. Estimating a contemporaneous backward-looking reaction function on real-time data with partial interest-rate smoothing yields an inflation-response coefficient of about 0.10 (statistically indistinguishable from zero), compared with roughly 1.0-1.5 on revised data - a result the paper argues reflects misspecification (the true rule is forward-looking) rather than genuinely unstable policy; re-estimating at forecast horizons of zero to four quarters ahead shows the real-time-data inflation coefficient only reaches a plausible value (1.64) at the four-quarter-ahead horizon, a horizon Orphanides links to Greenspan’s own stated policy horizon, and the implied equilibrium real rate from that specification (1.8-3.1%, depending on the assumed inflation target) is reasonable, whereas shorter, misspecified horizons yield “considerably higher and quite misleading” implied real rates. The results are specific to this small, Greenspan-era pre-ZLB sample and to Greenbook forecasts as the real-time proxy, but the paper’s broader conclusion is that policy reaction functions estimated from revised data can seriously misdescribe historical monetary policy and complicate the identification of monetary policy shocks.

Monetary Policy Rules, Macroeconomic Stability, and Inflation, A View from the Trenches

Re-estimating the Clarida-Galí-Gertler forward-looking Fed policy rule using only the real-time Greenbook forecasts available to the FOMC at the time, rather than revised after-the-fact data, this paper finds the inflation-response coefficient was above one both before and after Volcker’s 1979 appointment. Using quarter-by-quarter reconstructions of Federal Reserve Board staff forecasts for inflation and the output gap from 1966 to 1995, the estimated response to expected inflation is about 1.64 (one-quarter horizon) pre-Volcker and only modestly higher after – roughly twice the pre-Volcker estimate obtained by Clarida, Galí, and Gertler using ex post constructed data (0.75-0.80). The paper traces this discrepancy to the correlation structure of the data: real-time output-gap perceptions are strongly negatively correlated with inflation forecasts (-0.54) while ex post output-gap constructs are essentially uncorrelated with them (0.04), so omitting the real-time gap biases the estimated inflation response downward. The genuine, statistically significant difference across eras is in the response to the perceived output gap, which was more than twice as large before Volcker as after. Drawing on real-time output-gap data that show persistent, U-shaped overoptimism about potential output from the late 1960s through the mid-1970s (as trend productivity slowed unrecognized), the paper derives an index, γ/(β − 1), showing this activist pre-Volcker response to a badly mismeasured output gap implied roughly four times the steady-state inflationary bias of the post-1979 rule – about 4.4 percentage points versus 1.2. The paper also traces the 1979 policy shift through FOMC meeting records and Humphrey-Hawkins testimony, arguing the “stop-go” pattern of policy easing and abrupt tightening through the 1970s reflects activist policy chasing a mismeasured output target rather than a change in the weight placed on inflation itself, and that the “subtle” 1979 shift was toward de-emphasizing short-run output stabilization rather than toward a stronger anti-inflation stance per se.

Athanasios Orphanides | Macro Paper Warehouse

Monetary Policy Rules Based on Real-Time Data

Monetary Policy Rules, Macroeconomic Stability, and Inflation, A View from the Trenches