My presentation is entitled “Monetary Policy Inertia,” and I will be referring to a handout that was distributed. Based on monetary policy rules estimated from quarterly data, many economists hold the view that the Fed adjusts monetary policy at a very sluggish pace, specifically, that it distributes desired changes in the funds rate over several quarters—a behavior that is often termed “monetary policy inertia.” I will argue, instead, that there is essentially no policy inertia at a quarterly frequency and that, in fact, the funds rate typically is adjusted fairly promptly to economic developments—within a single quarter anyway. In large part, my argument is based on evidence of a very limited ability of financial markets (for example, Eurodollar futures or fed funds futures) to forecast the next few quarterly changes in the funds rate. Such evidence is informative about policy inertia because any partial policy adjustment obviously means that there is some remaining portion of the policy action that is postponed to the future and is thus predictable. Therefore, the absence of funds rate predictability implies the absence of significant partial adjustment by the Fed.
Before I flesh out this argument, however, let me start on page 1 of the handout and delineate two types of monetary policy inertia. These two types of inertia are often confused in the literature, with discussion of one type often mistakenly applied to the other; however, the two types of policy inertia operate at very different horizons. First, there is very short-term policy inertia or interest rate smoothing that I think does exist, but I won’t be discussing it in detail today. This short-term or week- to-week partial adjustment of the funds rate involves, for example, cutting the funds rate by two 25 basis point moves in fairly quick succession, rather than reducing the funds rate just once by 50 basis points. This type of gradualism or interest rate smoothing was more prevalent in the past, when intermeeting moves were more frequent, though it still goes on to some extent—perhaps induced by concerns of financial market fragility. In any case, this short-term policy inertia is essentially unrelated to the quarterly policy inertia that is relevant for most macroeconomic discussions—including my own.
Quarterly policy inertia is defined as the slow partial adjustment of the federal funds rate on a quarter-to-quarter basis. For example, if the Fed knew it wanted to increase the funds rate by 1 percentage point, it actually would raise the rate only about 20 or 25 basis points per quarter for the next few quarters. It is this type of inertia that I find suspect. The apparent evidence supporting quarterly policy inertia is summarized on the second page of my handout. This evidence is based on estimates of monetary policy rules or reaction functions—that is, estimated equations that attempt to model Fed behavior. These estimated equations usually take a standard partial-adjustment form, where the current funds rate is set as a weighted average of last quarter’s actual funds rate and the current quarter’s desired rate. This partial adjustment form is displayed as the first equation on page 2. The parameter ρ is the weight on last quarter’s funds rate, and 1-ρ is the weight on the current desired level. A high ρ means that the funds rate will be adjusted very slowly to its desired level. Based on quarterly data, estimates of ρ are typically around .75, which puts a ¾ weight on the lagged rate and a ¼ weight on the desired rate. Thus, these empirical rules imply a very slow speed of adjustment of the policy rate—specifically, the Fed would change the funds rate only 25 percent each quarter toward its desired level. This sluggish adjustment of the funds rate over several quarters is widely interpreted as evidence of “interest rate smoothing” or “monetary policy inertia.”
For example, before each FOMC meeting, the financial indicators packet that is distributed contains two estimated monetary policy rules: one with and one without quarterly policy inertia. Both rules set the desired funds rate according to the Taylor rule, which is displayed as the second equation. In the Taylor rule, the desired level of the funds rate is based on current readings for the output gap and inflation rate. The α and β parameters calibrate the policy response to these determinants. The funds rate recommendations from these two rules are shown in the chart at the bottom of page 2. The solid line shows the actual path of the funds rate as a quarterly average. The dashed line shows the estimated rule with no inertia—so ρ = 0. The dotted line shows the rule with inertia—where ρ is estimated by the Board staff to be .75. The Taylor rule without inertia (the dashed line) fits the actual funds rate fairly well, but there are some large persistent deviations. Notable deviations include 1992 and 1993, when the actual rate was held below the rule; 1996, when the rate was pushed above the rule; and 1999 and 2002, when again the funds rate fell below the rule. The nature of these deviations is a key element in understanding the evidence for policy inertia. The Taylor rule with inertia (the dotted line) largely eliminates these deviations and matches the historical path of the funds rate much more closely. That is, the lagged funds rate in this estimated equation is statistically significant. Although this type of econometric evidence appears convincing, it is valid only if the equation is specified correctly. If the desired funds rate also depends on persistent factors other than the current output and inflation in the Taylor rule, then such a misspecification could result in a spurious finding of partial adjustment. Accordingly, based only on these types of policy rule estimates, it is very difficult to distinguish between whether the Fed’s adjustment is sluggish or whether the Fed generally follows the Taylor rule with no policy inertia but sometimes deviates from the rule for several quarters at a time in response to other factors.
I will discuss the nature of these persistent deviations in a minute, but first—on page 3—I want to provide some evidence against policy inertia from a different source. This evidence is based on a key implication of policy inertia—namely, that the presence of inertia should imply predictive information in financial markets about the future path of the funds rate. Intuitively, if the funds rate typically is adjusted 25 percent toward its desired target in a given quarter, there’s a remaining 75 percent of the adjustment that should be expected to occur in future quarters. In a wide variety of settings, such delayed policy adjustment ensures forecastable future variation in the funds rate. Assuming that financial markets understand the inertial nature of policy, they should anticipate the future partial adjustment of the funds rate. In this case, a regression of actual changes in the funds rate on predicted changes should yield a good explanatory fit and a fairly high R². In fact, researchers have found the opposite. They have estimated interest rate forecasting regressions and, using financial market expectations, have found little predictive information beyond a few months. For example, Eurodollar futures have essentially no ability to predict the quarterly change in the funds rate three quarters ahead. The R² of such a regression is zero.
This lack of predictive ability is well illustrated by the most recent episode of easing. The chart at the bottom of the page gives the actual funds rate target and various expected funds rate paths as of the middle of each quarter based on fed funds futures. Under quarterly policy inertia, the long sequence of target changes in the same direction in 2001 would be viewed as a set of gradual partial adjustments to a low desired rate. However, although the funds rate gradually fell in 2001, market participants actually anticipated few of these declines at a six-to-nine-month horizon, as they would have if policy inertia were in place. Instead, markets assumed at each point in time that the Fed had adjusted the funds rate down to just about where it wanted the funds rate to remain based on current information available. Under this interpretation, the long sequence of declines is the result of a series of fairly prompt responses to new information that turned progressively more pessimistic. That is, the presence of quarterly partial adjustment or policy inertia is contradicted by the lack of the forecastability of changes in the funds rate.
Turning to page 4, I will reconcile the evidence for and against quarterly policy inertia. As I said, the persistent deviations of the actual rate from the Taylor rule without inertia are key to understanding what is going on. Under policy inertia, these persistent deviations are explained as sluggish responses to output and inflation, but that interpretation is inconsistent with the lack of funds rate predictability. An alternative explanation is that the Taylor rule is an incomplete description of Fed policymaking and that the Fed responds to other persistent variables besides current output and inflation. Under this interpretation, the Fed does not exhibit quarterly policy inertia, but it responds promptly to a variety of developments that unfold over time.
What would cause such persistent deviations from the Taylor rule? Well, in John Taylor’s original analysis, he noted that occasional deviations from his rule were appropriate responses to special circumstances. Two such special circumstances are noted at the bottom of page 4. The deviations in 1992 and 1993 can be interpreted as the Fed’s response to a disruption in the flow of credit, in which the funds rate was kept lower than might be expected given the macroeconomic context because of severe financial headwinds. The 1992-93 episode is better described as a persistent “credit crunch” deviation from the Taylor rule than as a sluggish partial adjustment to a known desired rate. In terms of the Taylor rule, the disruption of credit supply can be treated as a temporary fall in the equilibrium real rate, to which the Fed responds by lowering the funds rate (relative to readings on output and inflation). Similarly, in 1998 and 1999, a worldwide financial crisis following the Russian default and devaluation appeared to play a large role in lowering rates—again relative to what the Taylor rule would have recommended. Expectations also can play an important role in tempering the policy response to current readings on output and inflation. Indeed, some have suggested that expectations of future inflation—and, in particular, inflation scares in the bond market—are an important consideration for policy.
Finally, on page 5, I highlight two questions that are in some sense two sides of the same coin. The first question is, How should we think about analyzing or modeling recent Fed behavior? The second question is, To what extent should actual Fed behavior conform to our models of optimal monetary policy? Let me start with the first question. In modeling the Fed’s decisionmaking process, I argue that the Taylor rule is an incomplete description of Fed behavior. However, adding partial adjustment to the policy rule is not a solution; instead, in my opinion, partial adjustment adds another layer of misspecification that substitutes for a clearer understanding of the policy process. Of course, more research is required to characterize the full set of influences and determinants of policy beyond those contained in the Taylor rule.
A closely related question focuses on modeling the underlying motives of policy—more specifically, what kind of loss function should represent Fed behavior? Currently, the policymaker “perfect foresight” path in the Bluebook uses a loss function that assumes the Fed would be equally displeased with any of the following: (1) an unemployment rate 1 percentage point above the natural rate; (2) an inflation rate 1 percentage point above target; or (3) a 100 basis point decrease in the quarterly average funds rate. These equal weights place an implausibly high penalty on funds rate variability. However, without a substantial funds rate volatility penalty, the constructed optimal policy path does not match the recent historical path of the funds rate, and this is true across a variety of models. In my opinion, the high funds rate volatility penalty may be another misspecification that is compensating for some unknown flaw in our calculations of optimal policy. In essence, if policy over the past two decades has been close to optimal, then an important element is missing from the current specifications used by economists to construct optimal monetary policy.
An alternative possibility is that our economic models—without interest rate smoothing in the loss function—are basically correct in finding that under an optimal policy the Fed should be more aggressive in reacting to economic news than it has been. This suggests a second question: Should the Fed deviate from its historical behavior and become more aggressive in changing the funds rate? The analysis above suggests that the Fed has not been sluggish in reacting to economic developments: It has likely set the funds rate equal to its desired rate in each quarter. However, questions remain about whether the desired rate should react more forcefully to economic news. Indeed, researchers typically find that the parameters of an optimal Taylor rule—that is, the α and β shown on page 2—are much larger than the estimated parameters of a historical Taylor rule. Thus our models, even after trying to take into account various types of uncertainty, recommend much more vigorous policy responses. This raises the question, Has monetary policy been too timid in its responses to economic developments? I personally remain less than convinced that our models capture all the important factors influencing policy.