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

Monetary Policy Rules, Macroeconomic Stability and Inflation: A View from the Trenches Athanasios Orphanides(cid:3) Board of Governors of the Federal Reserve System December 2001 Abstract I estimate a forward-looking monetary policy reaction function for the Federal Reserve for the periods before and after Paul Volcker’s appointment as Chairman in 1979, using information that was available to the FOMC in real time from 1966 to 1995. Theresults suggest broad similarities in policy and point to a forward looking approach to policy consistent with a strong reaction to inflation forecasts during both periods. This contradicts the hypothesis, based on analysis with ex post constructed data, that the instability of the Great Inflation was the result of weak FOMC policy responses to expected inflation. A di(cid:11)erence is that prior to Volcker’s appointment, policy was too activist in reacting to perceived outputgaps thatretrospectively proved overambitious. Drawing oncontemporaneous accounts of FOMC policy, I discuss the implications of the (cid:12)ndings for alternative explanations of the Great Inflation and the improvement in macroeconomic stability since then. Keywords: Monetary policy rules, real-time data, Greenbook forecasts, stagflation. JEL Classi(cid:12)cation System: E3, E52, E58. Correspondence: Division of Monetary A(cid:11)airs, Board of Governors of the Federal Reserve System, Washington, D.C. 20551, USA. Tel.: (202) 452-2654,e-mail: aorphanides@frb.gov. (cid:3)I am thankful to many colleagues and participants at presentations at the Bank of England, the Riksbank,the EuropeanCentralBank,the IMF and the NBERfor comments and discussions. The opinions expressed are those of the author and do not necessarily reflect views of the Board of Governors of the Federal Reserve System.

1 Introduction The performance of the U.S. economy during the past two decades has been impressive. From the early 1980s to the end of the 1990s, the economy steadily expanded (with but a brief interruption in 1990), while inflation remained fairly stable and subdued. The 1980s marked what was the longest peacetime expansion on record, only to be followed by the longest expansion ever. The \Long Boom" aptly describes this exceptionally long period of stability and growth (Taylor, 1998). By contrast, the essence of the (cid:12)fteen or so years before the Long Boom, in one word, is \stagflation." This single word describes both the perception of stagnation throughout the 1970s and also the Great Inflation, which started in the mid-1960s and became increasingly more virulent during the 1970s. What accounts for this dramatic change in economic outcomes, from the instability of the Great Inflation, to the steady expansion of the Long Boom? Broadly, explanations fall intotwonotmutuallyexclusivestrands,thoseemphasizingpossiblechanges inthestructure 1 of theeconomy andthose emphasizingchanges in policy. From apolicy perspective, explanations that emphasize the role of policy are of particular interest. To the extent a change in policy has contributed to such a drastic improvement in economic well-being, proper identi(cid:12)cation of the policy mistakes that were presumably corrected, or, more generally, of the characteristics of policy during the period of superior performance, would be of great economic signi(cid:12)cance. After all, the single most signi(cid:12)cant contribution of historical policy analysis is perhaps to identify and help avoid the repetition of past mistakes. Anumberofalternative hypothesesforhowapolicychangemayhavecontributedtothe improvement in macroeconomic performance during the Long Boom have been advanced. One widely known view is the result of recent influential studies on monetary policy rules, 2 notably Clarida, Gali and Gertler (2000) (henceforth CGG) and Taylor (1999a). This 1Ahmed,LevinandWilson(2001),BlanchardandSimon(2001)andKahn,McConnellandPerez-Quiros (2001) among others, emphasize helpful changes in the structure of the economy or a reduction in the frequency of disruptive disturbances as the primary sources of the improvement. Blanchard and Simon identify a decline in volatility starting in the 1950s|interrupted in the 1970s and early 1980s. Kahn et al. stressimprovementsininformation technologiessinceabout1984. Ahmedetal. identifyareducedvariance of exogenous shocks since about that time as themost important but not theonly source of improvement. 2SeeBlinder(1979),DeLong(1997),Mayer(1999)andreferencesthereinforearlierinvestigationsofthe role of policy for theunfavorable outcomes associated with theGreat inflation. 1

view emphasizes the important insight that successful monetary policy requires a strong response to expected inflation, such that an increase in expected inflation prompts a more thanproportionalincreaseofshort-termnominalinterestrates. CGGandTaylorarguethat the di(cid:11)erence in performance from the Great Inflation to the Long Boom can be squarely tracedtoashiftinthisresponseassociated withPaulVolcker’s appointmentasChairmanof the Federal Reserve in 1979. In essence, these authors arguethat duringthe Great Inflation the Federal Reserve pursued a policy that accommodated inflation and induced instability in the economy by lowering real interest rates when expected inflation increased and vice versa. Thisperversepractice, theysuggest, endedwithVolcker’s appointmentasChairman, 3 thus restoring monetary stability in the economy. AnalternativeviewonhowpolicymayhaveimprovedsincetheGreatInflationidenti(cid:12)es changes in the response of policy to economic activity, as opposed to expected inflation. In this view, policy was excessively activist during the Great Inflation, a result of policymaker overcon(cid:12)dence intheir ability tostabilize deviations of outputfrom theeconomy’s potential 4 supply|the output gap. As shown by Orphanides (1998), if policymakers mistakenly adopt policies that are optimal under the presumption that their understanding of the state of the economy is accurate when, in fact, such accuracy is lacking, they inadvertently 5 induce instability in both inflation and economic activity. According to this view, the instability associated with the Great Inflation was the unintended outcome of excessively activist policies chasing output targets that proved overambitious, retrospectively. By the 3Other studies, some building directly on the CGG empirical results, have advanced related arguments. Forexample,ChristianoandGust(2000)emphasizethatahighinflationexpectationstrapcanariseifpolicy accommodatesinflationassuggestedbyCGGforthe1970s. Becauseoftheattentionthathasbeenreceived by theCGG results, in particular, I focus my discussion here on that analysis. 4Thisconcernisbasedonthewellknownmonetaristcriticismagainstactivistcontroloftheeconomy|the \monetarists" versus\activists"debate. Seetheessayscollected inFriedman(1953)forearlyexpositionsof theissueand Meltzer (1987) fora morerecent exposition. Itspotentialfor understandingtheimprovement in macroeconomic performance since the Great Inflation has been recently investigated in the context of interest rate policy rules by Orphanides (1998, 2000). The problems associated with designing monetary policy without adequate treatment of uncertainty regarding real-time assessments of the output gap (and the closely related \unemployment gap") have been recently emphasized in a number studies, including, Estrella and Mishkin (1999), McCallum (2001), McCallum and Nelson (1999), Orphanides et al. (2000), Smets (1998) and Wieland (1998). 5The empirical evidence, briefly reviewed in section 3, indicates that assessments of the economy’s productive potential have historically been quite inaccurate. During the 1970s, in particular, misperceptions regarding adverse shifts in trend productivity resulted in outsized errors and overoptimistic assessments of theeconomy’s potential. 2

endofthe1970s, theinstabilityandinflationaryimpetusoftheseactivistpolicieswas(cid:12)nally recognized and policy subsequently improved by becoming less activist. The behavior of inflation since the 1960s o(cid:11)ers indisputable evidence that monetary policy was highly accommodative during the Great Inflation but much less so afterwards. Figure 1 compares the behavior of inflation and the federal funds rate from 1966 to 1995. As is evident, the federal funds rate was consistently much higher than inflation since the late 1970s than it was earlier. This change is suggestive of a dramatic reversal in policy at that time. It also con(cid:12)rms an important element of both hypotheses mentioned above. To identify moreprecisely whetherand how monetary policydi(cid:11)ered beforeand after Volcker’s appointment,CGGestimateandcompareforward-lookingmonetarypolicyrulesresponding totheoutlookofinflationandeconomicactivity foreach era. Theirestimation alsosuggests that, even after controlling for policy responses to economic activity, the Federal Reserve adjusted real interest rates in a perverse manner prior to Volcker’s appointment but not after. In their estimation, however, CGG do not employ information that was available to the Federal Open Market Committee (FOMC) when monetary policy decisions were made butinstead rely on ex postconstructed data as proxies. As theycarefully acknowledge, this 6 raises some questions regarding the interpretation of the results. Indeed, CGG conclude that the fundamental problem they raise for the Great Inflation is that the Federal Reserve maintained persistently low short-term real interest rates in the face of high inflation; they also point to other possibilities for the cause of this mistake, including the alternative view mentioned earlier. Given the signi(cid:12)cance of an accurate interpretation of possible changes in policy after the Great Inflation, in this paper I revisit the issue and examine the evolution of monetary policy from the 1960s to the 1990s using exclusively information that was available to 6In particular, this practice can lead to misleading descriptions of historical policy and obscure the behaviorsuggested byinformation available topolicymakers in real time. Foradetailed discussion of these pitfalls in the context of policy rules such as those examined by Taylor and CGG see Orphanides (2001). Briefly, the main di(cid:14)culty arises from the fact that monetary policy decisions are based on and reflect policymaker perceptions of the state of the economy at the time policy is made. As a result, to correctly identify behavior, it is imperative to account for the evolution of these perceptions in real time and not simply rely on the actual evolution of the state of the economy as recognized ex post. Obviously, when perceptions and reality match closely, the distinction may be inconsequential. On the other hand, when perceptions proveincorrect for a period of time, thedistinction becomes crucial. 3

the FOMC when policy decisions were made. Speci(cid:12)cally, I estimate a forward-looking monetary policy reaction function such as proposed by CGG for the periods before and after Paul Volcker’s appointment as Chairman in 1979 using this real-time information. Estimationresultssuggestbroadsimilaritiesinpolicyoverthetwoperiods. Inparticular, andincontradictionto(cid:12)ndingsbasedontheexpostconstructeddata,theevidencepointsto a forward looking approach to policy consistent with a strong reaction to inflation forecasts both before and after Volcker’s appointment as Chairman. This suggests that policymakers during the Great Inflation did not commit an error as egregious as the perverse response to inflation would suggest. The evidence, however, does not absolve monetary policy from the macroeconomic instability experienced during the Great Inflation. As I discuss, the policy rule describing policy during the Great Inflation was excessively activist in its response to the output gap, especially in light of the outsized misperceptions regarding potential output that were only understood much later. By contrast, the evidence suggests that policy after 1979 did not exhibit the same degree of activism, resulting in a reduction of emphasistotheoutputgaprelativetoinflationinsettingpolicy. Contemporaneousaccounts provideadditionalsupportfortheviewthatanintentionalreductioninpolicyactivismalong these lines followed Paul Volcker’s appointment as Federal Reserve Chairman. The policy record suggests that rapidly changing economic developments during 1979 forced a critical reconsideration of policy that year. This subtle policy improvement in the aftermath of the Great Inflation contributed to the improved macroeconomic performance of the Long Boom. 2 Forward-looking Policy Rules 2.1 Speci(cid:12)cation Iconsiderafamilyofsimplelinearruleswiththefederalfundsrateasthepolicyinstrument. Briefly, these rules specify that monetary policy decisions are mainly driven by two factors, the outlook for inflation, as measured by the rate of change of the output deflator, and the outlook for real economic activity, as measured by the deviation of output from the economy’s potential supply|the output gap. This family of rules was (cid:12)rst examined in 4

detailinthepolicyregimeevaluation projectreportedinBryant, Hooper,andMann(1993). As they explained, this speci(cid:12)cation was motivated by the \stated dual objective of many central banks to achieve a sustainable growth in real activity while avoiding inflation," (p. 225), which also broadly describes the stated policy objectives of the Federal Reserve over the past several decades. Following an influential study by Taylor (1993), these rules are commonly referred to as \Taylor rules." Over the past several years, a vast literature has spawned examining various variants of these policy rules from theoretical and empirical 7 perspectives and their usefulness remains an area of active research. For the purposes of this study, I limit my attention to simple forward-looking variants along the lines examined 8 by CGG, which have served as the focus of recent historical policy comparisons. Let f t (cid:3) denote thenotional target for thefederal fundsrate for quarter t, y tjt the outlook for the output gap for quarter t, as perceived during the quarter, and (cid:25) t;ijt the outlook for inflation, speci(cid:12)cally for the average rate of inflation from quarter t to quarter t+i, also as perceived during quarter t. 9 The rules I examine specify that the notional target for the federal funds rate evolves according to: f t (cid:3) = (cid:11)+(cid:12)(cid:25) t;ijt+γy tjt Here, (cid:12) reflects the responsiveness of policy to expected inflation and γ the responsiveness of policy to real economic activity. As can be easily seen, (cid:12) > 1 reflects a policy that raises real rates with inflation, a response that is generally stabilizing, while (cid:12) < 1 indicates the 7SeeBall (1999), CGG (1999), Hetzel(2000), McCallum (1999), Taylor(1999b), Williams (1999), Woodford (2000), and references therein. Particularly relevant for forward-looking variants of these policy rules, such as examined here, is the work of Amato and Laubach (1999), Batini and Haldane (1999), Batini and Nelson (2000), Levin,Wieland and Williams (1999, 2000), Nessen (1999), Rudebusch and Svensson (1999), and Smets (2000). These forward-looking rules also provide a useful analytical framework for the inflation targeting approach to policy, as discussed in Bernanke and Mishkin (1997), Bernanke, Laubach, Mishkin and Posen (1998), and Svensson (1997, 1999). 8This sidesteps a number of possibly important issues relating to the speci(cid:12)cation of the rule. For example,itrulesoutthepresenceofnonlinearities,suchassuggestedfromtimetotimebyFOMCmembers themselvesandexamined,amongothersby,Blinder(1997),CGG(1999),OrphanidesandWilcox(1996)and Orphanides and Wieland (2000). Also, it does not address di(cid:11)erences in speci(cid:12)cation within linear rules which may influence interpretations of historical policy changes. For example, Sims (1999) and Fair (2001) suggest that the evidence for a policy change associated with Volcker’s appointment as Chairman is weak, based on the policy rule speci(cid:12)cations theyexamine. 9ForanyvariableX,IusethenotationX tj(cid:28) todenoteperceptionsofthevalueofthevariableforquarter t held at quarter (cid:28). For inflation and output data, this involves a forecast when (cid:28) (cid:20) t and actual data (though always subject to revision) for (cid:28) >t. 5

perverse response of reducing real rates when expected inflation rises, which is generally destabilizing. 10 The role of the remaining parameter, (cid:11), is most clearly seen by noting that in steady state, inflation is equal to the policy target, (cid:25)(cid:3) , and the output gap is equal to zero. Letting r(cid:3) denote the equilibrium real interest rate, the policy rule above implies: (cid:11) = r(cid:3)−((cid:12) −1)(cid:25)(cid:3) . Thus, (cid:11) reflects a linear combination of the equilibrium real rate and the inflation target and is equal to the equilibrium real rate in the special case of a zero inflation target. The actual federal funds rate for the quarter, f t, reflects movements of the notional target, f(cid:3) , possibly with a degree of partial adjustment, (cid:26)2 [0;1), 11 t f t = (cid:26)f t−1+(1−(cid:26))f t (cid:3) +(cid:17) t : The error, (cid:17) t, is assumed to reflect other factors that might influence the federal funds rate during the quarter, independent of the inflation and economic activity outlook. Combining the notional target and partial adjustment equations yields the following policy reaction function: f t = (cid:26)f t−1+(1−(cid:26))((cid:11)+(cid:12)(cid:25) t;ijt+γy tjt)+(cid:17) t : (1) 2.2 Real-Time Information In estimating a policy reaction function such as (1), the objective is to describe how policy responded over time to the outlook of inflation and economic activity as understood when policy decisions were made. Ideally, to capture the intent of policy as closely as possible, estimation of (1) should be based on consistent forecasts of inflation and the output gap, as formed by policymakers themselves, and reflecting concepts of these variables with uniform meanings over time. In practice, several complications need to be addressed. Monetary policy in the United States is decided by the Federal Open Market Committee. Although individual members of the Committee have sometimes o(cid:11)ered their views of the outlook, 10Stabilityconditionsdi(cid:11)erdependingonmodelspeci(cid:12)cdetails. Insomemodels,stabilityispossiblewith values of (cid:12) slightly smaller than one and γ > 0. See Christiano and Gust (2000), CGG (1999), Kerr and King(1996),RotembergandWoodford(1999),andWoodford(2000),forexaminationsinalternativemodels with optimizing behavior. 11Here, (cid:26) can be interpreted as an indicator of interest rate smoothing. See Sack and Wieland (2000) for a discussion of theoretical justi(cid:12)cations for such smoothing. 6

there does not exist a consistent record of the Committee’s quantitative assessment of the economic outlook at the time most decisions are made. However, a detailed record of policy discussions and information presented to the Committee by Federal Reserve Board sta(cid:11) at regularly scheduled meetings is available. Since the end of 1965, when the sta(cid:11) started the systematic preparation of quarterly forecasts for the FOMC, discussion of the outlook of the economy has been organized around these forecasts. Thus, to reflect information regarding the economic outlook as available to the FOMC as closely as possible, I rely on these forecasts and information associated with them. Speci(cid:12)cally, for each quarter from 1966Q1 to 1995Q4, I collected information corresponding to the Greenbook prepared 12 during (or, when not available by) the middle month of the quarter. For each quarter, I collected information regarding the concepts of \nominal output", \real output" and \potential output" or \output gap," which I used to construct time series for (cid:25) t;ijt and y tjt. This requires some additional speci(cid:12)city because the exact de(cid:12)nitions of these concepts has changed over time and, at times, multiple concepts have been put forth. The guiding principle I employed was to use, in each quarter, concepts corresponding to the headline concept for \real output" as de(cid:12)ned by the Commerce Department during that quarter. Thus the data reflect shifts in the concept of \nominal output" from GNP to GDP during the sample and various rede(cid:12)nitions of \real output" to correspond to alternative deflators over time. For \potential output," I use the o(cid:14)cial government estimates corresponding to the relevant concept of \real output" as available to Federal Reserve sta(cid:11) until 1980 and internal Federal Reserve sta(cid:11) estimates since then. (These data are from Orphanides, 2000). As already mentioned, the quarterly dataset constructed in this way is not ideal. However, it o(cid:11)ers a characterization of perceptions regarding the outlook for inflation and the output gap relevant for setting policy that is arguably as close as is possible, based on the available historical record. Further, a reading of the record of FOMC deliberations suggests that policy discussions since the 1960s have revolved around the outlook of economic ac- 12I start in 1966Q1 because systematic one-quarter-ahead forecasts were not presented in the Greenbook before December 1965. I end in 1995Q4 because more recent forecasts were not available to the public at thetime the dataset for this study was constructed (in February 2001). 7

tivity and inflation in a way that could be informed by these data with rather surprising continuity. To illustrate this point, it is instructive to compare the following two examples from policy deliberations, separated by nearly thirty years in time but selected to capture monetary policy turning points under roughly similar economic conditions. The (cid:12)rst reflects comments by Vice Chairman McDonough from the February 1994 FOMC meeting and also illustrates the role of the Greenbook forecasts as a focal point for the discussion regarding the economic outlook. With regard to the national forecast, we are rather similar to the Greenbook with some exceptions. ... In general, we think the gap between actual and potential GDP is now quite small, and certainly that which remains will be used up in the course of 1994 with our forecast, the Greenbook’s, or any of those we’ve heard around the table. Consequently, with the unemployment rate coming down to what we think is a reasonable estimate of the NAIRU|in the low 6 percent area|we do have to be considerably concerned about inflation. ... I believe very strongly that we should (cid:12)rm policy and that we should do so today... We are very near potential GDP and all of our forecasts, whether they are (cid:12)ne-tunings of the Greenbook or right on it, say that we will reach full potential this year. The second example reflects comments by Vice Chairman Hayes during the November 1965 meeting, about the time discussion of sta(cid:11) forecasts became an important element of Committee meetings. With the likelihood that GNP will be growing at a rate of around $11-12 billion per quarter in 1966, the gap between actual and potential levels of activity will probably narrow further ... ... [T]he time has come for an overt move to signal a (cid:12)rmer monetary policy ... [W]e are probably very close to the point where continued sustainable domestic expansion dependson greater e(cid:11)ort tokeep inflationarypressuresundercontrol. Despite some di(cid:11)erences, the considerations and rationale for taking policy action in these two instances would appear to be remarkably similar. 8

These examples also point to the forward-looking nature of policy, con(cid:12)rming that forward-looking speci(cid:12)cations for a policy rule are likely most appropriate for describing policy throughout this period. The appropriate horizon is less clear, especially for the early period,soIestimate equation(1)forfourhorizons,i = f1;:::;4g. BecauseearlyGreenbooks onlyreportedveryshort-runforecasts, however, thecoverage of dataforthe1960s andearly 1970s is increasingly less complete as the horizon lengthens. Data are missing for 2, 10, 18 and 26 observations respectively for the one-, two-, three- and four-quarter-ahead horizons. Figures 2 and 3 provide a graphical illustration of the data for the one-quarter ahead forecast horizon. Figure 2 plots the inflation forecast together with the ex ante real federal funds rate corresponding to that forecast. Figure 3 plots the output gap together with the same ex ante real interest rate series. Broadly, fluctuations in the real interest rate point to comovements with both the expected inflation and output gap series, suggesting that estimation of a policy rule such as equation (1) could o(cid:11)er an informative summary description of policy decisions. I return to these two (cid:12)gures later on. 2.3 Estimation Table1presentsestimation resultsforequation(1). Foreach forecasthorizon, i= f1;:::;4g, twosetsofestimatesarepresented,onesetwithdataendingwith1979Q2(priortoVolcker’s appointment as Chairman) and the second starting with 1979Q3. Three observations are in order. First, for both samples, the estimated policy rules (cid:12)t the data about as well and suggest rather similar policy responses to the output gap and expected inflation for the alternative forecast horizons. Second, concentrating on the estimated response to inflation, (cid:12), the estimates exceed one in both samples and are only slightly higher in the sample starting with Volcker’s appointment. In this sense, the policy response to expected inflation, appears broadly similar in both periods. Third, concentrating on the estimated response to the output gap, γ, estimates for the 1960s and 1970s are more than twice as large as the corresponding estimates for the sample starting with Volcker’s appointment. In this sense, policy appears to have been more activist during the Great Inflation than later. 9

Toexamine morepreciselywhetherandhow thepolicyrulesfor thetwo periodsdi(cid:11)erin a statistically signi(cid:12)cant sense, I estimated equation (1) for the whole sample and examined restrictions on the constancy of some or all of the policy rule parameters in the two periods. Results are reported in Table 2. Examiningall parameters ((cid:12)rst row in thetable) suggested rejections of the joint constancy hypothesis, at the 10% level for the one-quarter-ahead horizon, at the 5% level (but barely) for the two-quarter-ahead horizon, and tighter levels for the longer horizons. Examining the parameters one at time, while restricting remaining parameters to be constant across periods, suggested a statistically signi(cid:12)cant di(cid:11)erence in onlyoneparameter,γ. (Thisisreflectedinthefourthrowinthetable.) Forallhorizons,the response to the output gap was signi(cid:12)cantly smaller in the sample starting with Volcker’s appointmentasChairman. Noevidenceofasigni(cid:12)cantdi(cid:11)erenceintheresponsetoexpected inflation, (cid:12), was present (third row). Surprisingly, constancy of (cid:11) could not be rejected either, as would be expected if the inflation target or equilibrium real interest rate had changed signi(cid:12)cantly (at least in the linear combination r(cid:3)−((cid:12) −1)(cid:25)(cid:3) ). Figure 4 plots the notional targets implied by the parameter estimates for the onequarter ahead inflation. These permit a counterfactual comparison of the suggested setting for the federal funds rate, conditioning on the outlook for inflation and economic activity perceived at each quarter from 1966Q1 to 1995Q4. One interesting observation is that the two rules are not very di(cid:11)erent in the (cid:12)rst few and last several years in the sample. They do di(cid:11)er substantially from about 1974 to about 1985, with the rule estimated for the pre- Volcker period providing systematically easier policy prescriptions. The main di(cid:11)erence, again, is that the rule estimated for the period after Volcker’s appointment, would not have suggested as large a policy ease as was adopted in practice in response to the severe downturn and recovery associated with the 1974 recession. Similarly, it did not suggest as large a policy ease as the earlier rule would have suggested in responseto the downturn and recovery associated with either the 1980 or 1982 recessions. This tighter policy, of course, was the driving force behind the stabilization of inflation in the early 1980s. In summary, the estimated policy rules suggest broad similarities in policy before and after Volcker’s appointment as Chairman in 1979, with only a rather subtle (though not 10

unimportant) di(cid:11)erence, a reduced response to perceived output gaps. 3 Interpretation 3.1 Output Gap Misperceptions Given the importance of the policy response to perceived output gaps apparent in the estimation results above, it is usefulto examine the evolution of theseperceptions over time in order to gain a better understanding of the historical evolution of policy. As discussedin detail in Orphanides(2000), contemporaneous perceptions of theoutput gap during the period covered in this sample exhibited serious flaws. An important source of di(cid:14)culty was the failure to recognize su(cid:14)ciently quickly the persistent adverse shifts in trend productivity in the economy that were experienced during the late 1960s and early 13 1970s. As a result, estimates of potential output during this period appeared consistently more optimistic than what could be justi(cid:12)ed based on ex post data. Throughoutthe 1970s, output appeared to fall short of the economy’s potential supply, increasingly so in the early 14 and mid 1970s. Ex post, this gradual deterioration in the economy’s prospects can be captured by approximating potential output, for example, by using a quadratic time trend or a smooth trend such as based on the Hodrick-Prescott (cid:12)lter. To assess the pattern of output gap misperceptions in this sample, Figure 5 provides a comparison of the real-time perceptions of the output gap with two ex post constructs. These are based on current data for GDP detrended over the 1966 to 1995 sample using the HP (cid:12)lter and a quadratic time trend, respectively. I selected these two methods as representative of alternatives that are frequently employed, also noting that these two have beenemployed forestimatingpolicyruleswithexpostdata, includingbyCGG(1998, 2000) 13Inadetailedstudyofo(cid:14)cialreal-timeoutputgapestimatesintheUnitedKingdom,NelsonandNikolov (2001)reportaremarkablysimilarpatternoferrorsinthatcountry. Giventhatmanyindustrializedcountries experienced a slowdown in productivity during the 1960s and 1970s, it is likely that similar patterns may characterize errors in the measurement of the output gap in some of these countries as well. To the extent monetarypolicyinthesecountriesexhibitedactivismsimilartothatexhibitedbytheFederalReserve,these misperceptions could explain, at least in part, the common rise and fall in inflation observed in so many countries from the1960s to the1990s. 14Errors associated with the GNP data as originally reported by the Commerce Department also contributedimportantlytotheproblem. Forexample,Orphanides(2000b)showsthattheseerrorscanaccount for about 5 percentage points of themismeasurement of theoutput gap in 1975. 11

15 and Taylor (1999a). The HP and quadratic trends produce very similar results over this sample. As a result, I concentrate my comparisons of the real-time series with the ex post concept based on the quadratic trend which is also the one favored by CGG. Using this gap as a reference series suggests a number of interesting observations regarding the real-time output gap perceptions. First, the real-time series is very similar to the ex post construct at the beginning and end of the sample. The two series also exhibit similar comovements with the business cycle, registering cyclical peaks and troughs at about the same times. But the real-time series diverges from the ex post construct from the late 1960s to the mid 1970s before the two series slowly converge again over the late 1970s and 1980s. The divergence suggests a U-shaped pattern of misperceptions, with a low point around 1975. This, of course, is the pattern that would be expected with a process of gradual learning of the reductioninpotential outputgrowthassociated withthedeterioration oftrendproductivity 16 in the economy that was experienced during the late 1960s and early 1970s. One would also expect that such misperceptions would lead to systematic errors in inflation forecasts. Indeed,ascanbeseeninFigure6,inflationforecastssystematicallyunderpredictedinflation 17 during the late 1960s and early 1970s. Elements of this comparison prove useful as a device for reconciling di(cid:11)erences in alternative interpretations of the historical evolution of policy and macroeconomic outcomes. 3.2 Correlations and Biases The estimated policy parameters of a linear policy reaction function such as (1), reflect the correlation patterns of the underlying data. One way to understand di(cid:11)erences between the results in Table 1 and those based on ex post constructed data is to compare relevant correlations of the real-time and ex post constructs. Consider, for example, alternative 15However,itiswellknownthatneitherofthesemethodsisusefulforreal-timeanalysisduetothelackof reliability oftheresultingend-of-sampletrendestimates. SeeOrphanidesandvanNorden(1999) fordetails on themagnitude of this unreliability. 16Note that because the sample ends in 1995, these data do not reflect the reversal of this deterioration in trend productivity that was experienced in the late 1990s. Of course, the quadratic detrending concept described herewould be totally inappropriate for examining that reversal. 17This is evident from comparisons of the forecasts with either current data (as shown in the (cid:12)gure) or (cid:12)rst-publisheddata. Mayer(1999)o(cid:11)ersadetailedanalysisoftheinflationforecasterrorsduringthisperiod. 12

estimates of the parameter (cid:12) which is the critical parameter for the hypothesis that the policy response to inflation was perverse during the period before Volcker’s appointment. To that end, compare the estimates for the one-quarter ahead inflation forecast, the case i = 1 in Table 1, with the corresponding estimates reported by CGG using quadratic trend 18 concepts of the output gap. CGG (1998) and CGG (2000) report estimates of 0.80 and 0.75 for (cid:12), respectively. By contrast, the estimate in Table 1 above is about twice as large, 1.64. An important di(cid:11)erence, in this case, is associated with the correlation of the output gap with the inflation forecasts. The ex post gap based on quadratic detrending is not correlated with the inflation forecast series. The correlation coe(cid:14)cient in the 1966Q1 to 1979Q2 sample is 0.04. By contrast, the real-time output gap series exhibits a signi(cid:12)cant negative correlation with the inflation forecast series −0:54. 19 The implications of this di(cid:11)erence on estimated policy parameters are easy to see. Since (cid:12) and γ are positive, if policymakers in real time responded strongly to both expected inflation and the output gap, omitting the real-time gap from the estimation of the policy rule would lead to a downward bias in the estimate of (cid:12). And this downward bias in estimating (cid:12) would remain if the ex post construct were used in place of the real-time gap, since the ex post construct is uncorrelated with expected inflation. To illustrate the signi(cid:12)cance of this bias, I reestimated equation (1) imposing the restriction γ = 0. The resulting estimate of (cid:12) was below one, 0.94 to be exact, con(cid:12)rming a substantial downward bias. Returning to Figures 2 and 3, concentrating on the movements of expected inflation, the real interest rate and the output gap up to mid-1979, provides a visual rendition of this argument. The data suggest that two forces were pulling increasingly the real interest rate in opposite directions with roughly equal force. While rising inflation suggested that the real rate should be raised, perceptions that the economy was getting further away from its potential suggested a reduction of the real rate was in order. The policy rule estimated 18Note that CGG use the subscript t+1 to denote output produced during period t. Instead, I employ theusualtimingconvention. Thus,y tjt referstotheoutputgapforquartertwhichmatchestheoutputgap CGG denotewith thesubscript t+1 and employ in their baseline speci(cid:12)cation. 19Thiscollinearity alsoexplainstherelatively largestandarderrorsin Table1,despitethehighoverall(cid:12)t of the regressions. 13

for the Great Inflation period indicates that policy responded strongly to both of these concerns, and balanced them nearly one for one. By contrast, the pattern of correlations in the two (cid:12)gures changes somewhat after mid-1979, indicative of the relatively greater emphasis on expected inflation reflected in the estimated policy rule. 3.3 Inflation and Disinflation The strong response to perceived output gaps coupled with the pattern of misperceptions suggestedinFigure4,provideastraightforwardexplanationfortheacceleration ofinflation, especially during the 1970s. To see this, it is useful to examine how far from the inflation target, (cid:25)(cid:3) , the economy would settle if policy responded to an output gap persistently measured with an error equal to −x (de(cid:12)ned so that x > 0 measures an overoptimistic assessment of the economy). Recall that in the absence of such a systematic error, the rule implies that in steady state, r(cid:3) + (cid:25)(cid:3) = (cid:11) + (cid:12)(cid:25)(cid:3) . With a persistent error, −x, the correspondingsteadystate rateofinflation wouldbe(cid:25)(cid:3)x suchthatr(cid:3) +(cid:25)(cid:3)x = (cid:11)+(cid:12)(cid:25)(cid:3)x−γx. Bringing these two together yields: (cid:25)(cid:3)x −(cid:25)(cid:3) = γ x. Thus, the ratio γ provides a ((cid:12)−1) ((cid:12)−1) useful index of the inflationary consequences of a persistent overoptimistic assessment of the economy, indicating by how much (in percent) inflation would be expected to deviate from its target if the output gap were persistently believed to be one percent below its true value. The inflationary potential associated with sustained overoptimistic assessments of the economy’s potential supply di(cid:11)ers importantly for the rules followed during and after the Great Inflation. To illustrate the extent of these di(cid:11)erences, Table 3 shows the values for this index corresponding to the policy reaction functions estimated in Table 1. As can be seen, the index has values around 1 for the policy rules describing the Great Inflation but only about one quarter as high for the post-1979 sample. (The data reject the hypothesis that the index is constant over the two samples, for all forecast horizons.) Noting that realtime misperceptions of the output gap averaged −4:9 percent in the sample to mid-1979 and −3:6 percent later on, we can use the index to obtain a rough estimate for the extent of the inflationary bias embedded in the policies followed duringthe two subsamples. Using 14

the index corresponding to the one-quarter ahead forecast horizon, suggests an inflationary bias of about 4.4 percent before mid-1979. That is, if policymakers implemented policy aiming towards a long-run inflation target of 2 percent, their actions were actually pushing the economy to an inflation rate above 6 percent. This bias would have been much smaller, 1.7 percent, if the policy rule describing the post-1979 period was in place. Likewise, the inflationary bias for the post-1979 period is only about 1.2 percent but would have been considerably larger, about 3.2 percent, had the policy rule describing the pre-1979 period been in place. Thus, there were signi(cid:12)cant di(cid:11)erences in the two policy rules which have important implications for understanding the Great Inflation and subsequent disinflation. 3.4 Stop-Go Policy Instability In addition to generating high average inflation, the excessive activism exhibited by policy during the Great Inflation, coupled with the increasingly optimistic assessments of potential economic activity in the early 1970s, increased instability in the economy. Mixing these two ingredients is essential for understanding the problem. Under ideal conditions, activist policies such as followed during the Great Inflation could be e(cid:14)cient and result in greater stability than policies placing less emphasis on perceived output gaps. However, this requires a solid understandingof the structure of the economy and reliable assessments of the economy’s potential. With that in place, deviations of actual output from potential output can, in principle at least, be a useful guide for setting policy. But what if this guide is error prone,as happenedsosystematically duringthe1970s? Then,following theactivist policies deemed e(cid:14)cient under the presumption of accuracy, can lead policymakers to a futile chase of the wrong target. The result, is a pattern of stop-go policy reversals that retrospectively appear to be out of sync with the economic fundamentals. Retrospectively, policy keeps falling \behind the curve." A formal accounting of the role of persistent output gap misperceptions in generating instability, when policy follows an activist interest rate policy rule such as (1), requires comparisons based on an estimated model and estimates of the persistence and magnitude of historical misperceptions. Orphanides (1998) presents such comparisons and his results 15

provide a useful perspective for the di(cid:11)erences one could expect from the rules estimated here before and after Volcker’s appointment. The model is su(cid:14)ciently simple so that policy rules such as (1) are optimal for a policymaker who values inflation and output stability. Using the 1980s and early 1990s as a benchmark period, Orphanides shows the degree of instability induced when policy follows an activist rule that is optimal with perfect information when, in fact, mismeasurement is present, and computes e(cid:14)cient rules that properly account for mismeasurement. To illustrate the di(cid:11)erences in the alternative rules, it is useful to compare how activist they are in terms of the index described earlier, γ=((cid:12)− 1). For a policymaker who places equal emphasis on inflation and output stability, the 20 optimal rule with perfect information has an index value of 0.84. This compares with 0.47 for the e(cid:14)cient rule in the presence of mismeasurement such as seen in the data in the 1980s and early 1990s. And while the more activist rule is by design optimal under perfect information, it yields asymptotic standard deviations for inflation and output that 21 are about 10 percent higher than those corresponding to its less activist counterpart. These comparisons suggest that the reduction in policy activism that followed Volcker’s appointment could explain, at least in part, the improved performance of the economy during the Long Boom. An intuitive understanding of the \stop-go" problem, as it applies to the 1970s, can be gained simply by returning to Figure 5 and relating the path of policy to the output gap as perceived in real-time, and as suggested by the ex post constructs. A useful starting point is the recession of 1970. At the turn of the year, as signs of a recession appeared, monetary policy started on a path of policy easings to restore economic growth. But how long could policy maintain an expansionary stance without facing a threat of worsening inflation from an overheated economy? Looking at the ex post constructs in Figure 5, 20This isbased onthevaluesreportedin Table3for thepreference weight!=0:5in Orphanides(1998). The index monotonically increases with therelative preference towards greater outputstability. 21Thee(cid:14)cientdegreeofpolicyactivismvariesgreatlywiththemagnitudeofmismeasurement. Thus,with greater mismeasurement, such as in the 1970s, the e(cid:14)cient policy is even less activist and the performance loss associated with the activist rule greater. On the other hand, with better measurement more activist rules would be e(cid:14)cient. Thus, if a reduction in the volatility of the economy (as suggested in the studies mentioned in footnote 1) reduces the variance of y tjt and its associated measurement error, more activist rules would be e(cid:14)cient relative to therules that would be e(cid:14)cient for the1970s or 1980s. 16

by early 1972 output had returned to its trend and the expansionary stance should have been long reversed. Based on the real-time perceptions of the output gap, however, the economy did not appear overexpanded even much later. The resulting policy activism ignited inflation|the go phase of the policy error. With inflation rising, policy tightened signi(cid:12)cantlybylate1973, raisingtherealratetoaboutfourpercent. Andwiththeeconomy already overextended, this action could only bring about a recession|the stop phase of the 22 policy error. Inretrospect,byinappropriatelychasingafteranoutputtargetthatwastoohighrelative to the economy’s potential, policy inadvertently pushed the economy beyond its potential, fueling inflation, prompting an abrupt tightening which precipitated a recession only to start the cycle once again. 4 What happened in 1979? Thecycle of\stop-go" policyerrorswas toberepeatedoncemoreneartheendofthe1970s. In1977,whenoutputhadreturnedtoitstrend,accordingtotheexpostconstructspresented in Figure 5, real interest rates were about zero. Perceptions in real-time, however, did not suggest that the economy was overheated. Once again, by responding to these perceived gaps, policy kept real interest rates too low for too long. In the second half of 1978, the FOMC recognized that the pace of economic expansion was too rapid while inflationary pressures were not abating. A weakening dollar elevated concernsthatinflationandinflationexpectationswouldremainhighevenifeconomicgrowth were to be brought down in line with the economy’s potential supply. Reflecting these concerns, the Committee raised interest rates in a series of policy moves, aiming to curb inflation in the following year. The situation at the turn of the year was described in the (cid:12)rst Humphrey-Hawkins 22Surely, the energy crisis and other shocks contributed importantly to the dismal outcomes of 1974 and 1975. The argument here simply points out that at least part of the inflationary problem and economic slowdown can be traced to the earlier policy mistakes. Barsky and Kilian (2001), Lansing (2001), and Orphanides (2000) provide counterfactual model simulations that attribute a large part of the problem duringthisperiod tosuch policy errors. Barsky and Kilian, in particular, argue that theenergy crisis itself was likely an endogenous response to the policy driven overheatingof theeconomy. 17

Report, submitted to the Congress on February 20, 1979: The narrowing of the gap between actual and potential output implies that a tighterholdonthenation’saggregatedemandforgoodsandservicesisnecessary if inflationary forces are to be contained. ... Real GNP increased 4.3 percent from the fourth quarter of 1977 to the fourth quarter of 1978|a bit slower than the average pace over the earlier part of the expansion, but still well above the trend growth of potential output in the 23 economy (p. 33-34). The Committee’s outlook in the Report exhibited cautious optimism, noting that \...it shouldbepossibletoslowthepaceofexpansion|andtherebyrelieveinflationarypressures| without prompting a recession." (p. 54). However, as the Record of Policy Actions for the February Meeting revealed soon after, some members harbored less sanguine views of the outlook. On one side, \a few members ... suggested that the onset of a recession before the end of the year ... was the most likely development" (p. 128). But others recognized a serious danger that inflation could intensify further. Both risks appeared well justi(cid:12)ed. As the year progressed, the Committee was once again facing the cruel dilemma of stagflation. Already by March, both inflation and economic weakness risks had deteriorated. The record of the March 20 meeting indicates that \many members" (as opposed to the \few members" who had expressed a similar concern in February) \believed that the chances of a recession beginning before the end of the year or in early 1980 were fairly high" (p. 138). Regarding the inflation outlook, signi(cid:12)cant disagreements became evident. One view was that the \slackening of economic activity later in the year could be expected to slow the rise of prices generally," but another view was that \inflation would remain rapid even during a recession," (p. 139). The meeting concluded with a decision not to change policy, but on a very close vote, with 6 votes in favor of the adopted directive, and 4 dissents in favor of a more restrictive policy. Incoming data prior to the April 17 and May 22 FOMC meetings continued to reinforce both concern of additional economic weakness and concerns regarding heightened inflation. In May, 23PagenumbersforreferencestotheHumphrey-Hawkinsreports,andRecordsofPolicyActionsforFOMC meetings during 1979 refer to theAnnual Report for 1979, Federal ReserveBoard (1980). 18

this resulted in an unusual split of the vote, with two dissents favoring an easing and one favoring a tightening. By the July 11 meeting, the situation appeared to have markedly deteriorated on both fronts. According to the record for the meeting, \no member of the Committee expressed disagreement with the sta(cid:11) appraisal that real gross national product had declined somewhat in the second quarter and that further declines were likely for the remaining two quarters of the year" (p. 171). The Humphrey-Hawkins Report, submitted to the Congress on July 17, noted that the consensus projection of Board members for real GNP growth for 1979 was −2 to −1=2 percent. Despite this dismal outlook, however, inflationary concerns were getting even worse, and started to shift the Committee’s view of the balance of risks squarely in that direction. Among a number of factors cited for intensi(cid:12)ed inflationary pressures, most important was continuing unexpected increases in oil prices, and a decline in the value of the dollar. In some ways, July 1979 marked a small but important turning point. Despite the view that the economy was likely already in recession, by the end of the month the Committee had raised the federal funds rate twice, (cid:12)rst on July 19 and then again on July 27. Soon after, starting with Paul Volcker’s (cid:12)rst meeting as chairman on August 14, the Committee moved even more decidedly in a tightening direction, despite the fact that the outlook for the economy appeared, if anything, even more uncertain. The Policy Record of the August 14 meeting o(cid:11)ers a glimpse of the unpleasant choices: In considering policy for the period immediately ahead, Committee members focused on the problems posed by emerging recession and its potential for substantial increases in unemployment, concurrent with strong monetary growth, high actual and expected rates of inflation, and an exposed position of the dollar in foreign exchange markets pending anticipated improvement in the U.S. foreign trade and current accounts. Any policy course in these circumstances necessarily involved unusual risks: prompt pursuit of a policy aimed at moderating the e(cid:11)ects of the curtailment in outputcould beperceived as exacerbating inflation and thus could have perverse e(cid:11)ects on economic activity and employment; a policy directed toward moderating inflation and lending support to the dollar in the foreign exchange markets could risk intensifying the recession (p. 183). 19

By moving decisively towards tightening, the Committee demonstrated that during the course of the summer policy had shifted in a subtle way, from the reluctance to raise interest rates in the face of concerns of economic weakness, to a focus on inflation. By October, the famous change in operating procedures further solidi(cid:12)ed this focus of policy towards reigning in inflation and set the economy towards a path of disinflation. Looking back, the delay in tightening policy during the (cid:12)rst half of the year proved a costly mistake. Despite all the fears and concerns, the widely anticipated recession that kept the Committee from tightening duringthe (cid:12)rst half of the year did not arrive. Despite the pessimism and gloomy forecasts for 1979, the economy grew in every quarter. By not tightening, the Committee compounded its earlier errors, allowing inflation to accelerate further only to postpone and raise the costs of restoring stability. But this lesson was not lost on the Committee. In his (cid:12)rst Humphrey-Hawkins testimony, on February 19, 1980, Chairman Volcker explained the subtle policy shift that had taken place: In the past, at critical junctures for economic stabilization policy, we have usually been more preoccupied with the possibility of near-term weakness in economic activity or other objectives than with the implications of our actions for futureinflation. To some degree, that has been true even duringthe long period ofexpansionsince1975. Asaconsequence, (cid:12)scalandmonetarypoliciesalike too often have been prematurely or excessively stimulative, or insu(cid:14)ciently restrictive. The result has been our now chronic inflationary problem, with a growing conviction on the part of many that this process is likely to continue. ... The broad objective of policy must be to break that ominous pattern. That is why dealing with inflation has properly been elevated to a position of high nationalpriority. Successwillrequirethatpolicybeconsistentlyandpersistently oriented to that end. Vacillation and procrastination, out of fears of recession or otherwise, would rungrave risks. Amid the present uncertainties, stimulative policies could well be misdirected in the short run; more importantly, far from assuring more growth over time, by aggravating the inflationary process and psychology they would threaten more instability and unemployment. (Federal Reserve Board, 1980b, p. 214) It is easy to understate the signi(cid:12)cance of the change Volcker articulated in this testimony. As the NBER later con(cid:12)rmed, the economy had already peaked in January and 20

was in recession during the (cid:12)rst half of 1980. But by then, the Chairman was not about to recommend repeating the policy errors of the recent past. The Committee had recognized that long term stability required setting short-run output stabilization concerns aside. 5 Conclusion In retrospect, there is little doubt that monetary policy during the Great Inflation was too activist, placing too much emphasis on short-run stabilization of economic activity at the expense of the Federal Reserve’s long-term price stability objective. However, policy was not flawed in an obvious manner; indeed it would appear entirely reasonable from the perspective of many modern policy-evaluation analyses. In theory, the activist approach to monetary policy that was followed during the Great Inflation would be workable, if only policymakers could have a solid understanding of the structure of the economy and reliable readings of the state of the economy upon which to base their actions. But what works in theory, often works in theory only. In reality, policymakers did not possess the knowledge necessary for an activist approach to monetary policy. Regrettably, they also lacked an appreciation of their ignorance. Despite the best of intentions, monetary policy itself became the engine of inflation and a source of instability during the Great Inflation. The subtle policy change in 1979 reflected a shift to more modest but attainable goals. Reducing the excessive emphasis on stabilizing the level of economic activity around its uncertain potential and concentrating instead on the inflation outlook for policy guidance providedthefoundationforstablesustainablegrowth. Thisallowedtheeconomytoprogress unimpeded|the Long Boom. 21

References Ahmed,Shaghil,AndrewLevinandBethAnneWilson(2001),\RecentU.S.Macroeconomic Stability: Good Luck, Good Policies, or Good Practices?" Federal Reserve Board. Amato, Je(cid:11)ery, and Thomas Laubach (1999), \Forecast-based Monetary Policy," mimeo. Ball, Laurence (1999), \Policy Rules for Open Economies," in Taylor (1999c). Batini, Nicoletta and Andrew Haldane (1999), \Forward-looking Rules for Monetary Policy," in Taylor (1999c). Batini, Nicoletta and Edward Nelson (2000), \Optimal Horizons for Inflation Targeting," Bank of England Working Paper 119, July. Barsky,RobertB.andLutzKilian(2001)\AMonetaryExplanationoftheGreatStagflation of the 1970s," mimeo. Bernanke, Ben S., and Frederic S. Mishkin (1997), \Inflation Targeting: A New Framework for Monetary Policy?" Journal of Economic Perspectives, 11(2), 97-116. Bernanke, Ben, Thomas Laubach, Frederic Mishkin and Adam Posen (1998), Inflation Targeting: Lessons from the International Experience, Princeton: Princeton University Press. Blanchard, Olivier, and John Simon (2001), \The Long and Large Decline in U.S. Output Volatility," mimeo, March. Blinder, Alan (1979), Economic Policy and the Great Stagflation, San Francisco: Academic Press. Blinder, Alan (1997), \What Central Bankers Could Learn from Academics|and Vice Versa," Journal of Economic Perspectives, 11(2), 3-20, Spring. Christiano, Lawrence, and Christopher Gust (2000). \The Expectations Trap Hypothesis," in Money, Monetary Policy and Transmission Mechanisms, Bank of Canada. Clarida,Richard,JordiGali,andMarkGertler(1998), \MonetaryPolicyRulesandMacroeconomic Stability: Evidence and Some Theory," NBER Working Paper, No, 6442, March. Clarida, Richard, Jordi Gali, and Mark Gertler (1999), \The Science of Monetary Policy," Journal of Economic Literature, 37(4), 1661-1707, December. Clarida,Richard,JordiGali,andMarkGertler(2000), \MonetaryPolicyRulesandMacroeconomic Stability: Evidence and Some Theory," Quarterly Journal of Economics, 147- 180, February. De Long, Bradford (1997), \America’s Only Peacetime Inflation: The 1970’s," in Romer and Romer, eds, (1997). Estrella Arturo amd Frederic Mishkin (1999), \Rethinking the Role of NAIRU in Monetary Policy: Implications of Model Formulation and Uncertainty," in Taylor (1999c). Fair, Ray C. (2001), \Actual Federal Reserve Policy Behavior and Interest Rate Rules," Federal Reserve Bank of New York Economic Policy Review, 61-72, March. 22

Federal Reserve Board (1965), FOMC Minutes, Washington D.C. Federal Reserve Board (1980), Annual Report for 1979, Washington D.C.. Federal Reserve Board (1980b), Federal Reserve Bulletin, Washington D.C.. Federal Reserve Board (1994), FOMC Transcripts, Washington D.C. Friedman, Milton (1953), Essays in Positive Economics, Chicago: University of Chicago. Hetzel,RobertL.(2000), \TheTaylorRule: IsitaUsefulGuidetoUnderstandingMonetary Policy?, Federal Reserve Bank of Richmond Economic Quarterly, 86(2), 1-33, Spring. Kahn, James, Margaret M. McConnell, and Gabriel Perez-Quiros (2001), \The Reduced Volatility of the U.S. Evonomy: Policy or Progress?" mimeo. Kerr,William, andRobertG.King(1996), \LimitsonInterestRateRulesintheISModel," Federal Reserve Bank of Richmond Economic Quarterly, 82, 47-76, Spring. Lansing, Kevin (2001), \Learning About a Shift in Trend Output: Implications for Monetary Policy and Inflation," Federal Reserve Bank of San Francisco working paper. Levin,Andrew,VolkerWielandandJohnWilliams(1999), \RobustnessofSimpleMonetary Policy Rules under Model Uncertainty," in Taylor (1999c). Levin, Andrew, Volker Wieland and John Williams (2000), \The Performance of Forecastbased Monetary Policy Rules under Model Uncertainty," Federal Reserve Board. Mayer, Thomas (1999), Monetary Policy and the Great Inflation in the United States: The Federal Reserve and the Failure of Macroeconomic Policy, 1965-1979, Edward Elgar, Cheltenham. McCallum, Bennett T. (1999), \Issues in the Design of Monetary Policy Rules," in Taylor and Woodford (1999). McCallum, Bennett T. (2001), \Should Monetary Policy Respond Strongly to Output Gaps?" mimeo, Carnegie-Mellon University. McCallum, Bennett T. and Edward Nelson (1999), \Performance of Operational Policy Rules in An Estimated Semi-Classical Structural Model," in Taylor (1999c). Meltzer, Allan H. (1987), \Limits of Short-Run Stabilization Policy" Economic Inquiry, 25, 1-14. Nelson, Ed, and Kalin Nikolov (2001), \A Real-Time Output Gap Series for the UK, 1965- 2000: Construction, Analysis, and Implications for Inflation," Bank of England, mimeo. Nessen, Marianne (1999), \Targeting Inflation over the Short, Medium and Long Term," Sveriges Riksbank. Orphanides, Athanasios (1998), \Monetary Policy Evaluation With Noisy Information," Finance and Economics Discussion Series, 1998-50, Federal Reserve Board, October. Orphanides, Athanasios (2000), \The Quest for Prosperity Without Inflation," European Central Bank Working Paper No. 15, March. Orphanides, Athanasios (2000b), \Activist Stabilization Policy and Inflation: The Taylor 23

Rule in the 1970s," Finance and Economics Discussion Series, 2000-13, Federal Reserve Board, February. Orphanides,Athanasios (2001), \Monetary Policy Rules Based onReal-Time Data," American Economic Review, 91(4), 964-985, September. Orphanides,Athanasios, Richard Porter, David Reifschneider, RobertTetlow andFrederico Finan (2000), \Errors in the Measurement of the Output Gap and the Design of Monetary Policy," Journal of Economics and Business, 52(1/2), 117-141, January/April. Orphanides, Athanasios, and Simon van Norden (1999), \The Reliability of Output Gap Estimates in Real Time." Finance and Economics Discussion Series, 1999-38, Federal Reserve Board, October, August. Orphanides, Athanasios and Volker Wieland (2000), \Inflation Zone Targeting," European Economic Review 44(7), 1351-1387, June 2000. Orphanides, Athanasios and David Wilcox (1996), \The Opportunistic Approach to Disinflation," Finance and Economics Discussion Series, 1996-24, Federal Reserve Board, May. Romer, Christina and David Romer eds. (1997), Reducing Inflation: Motivation and Strategy, Chicago: University of Chicago. Rotemberg, Julio and Michael Woodford (1999), \Interest Rate Rules in an Estimated Sticky Price Model," in Taylor (1999c). Rudebusch, Glenn and Lars Svensson (1999), \Policy Rules for Inflation Targeting," in Taylor (1999c). Sack, Brian, and Volker Wieland (2000) \Interest-Rate Smoothing and Optimal Monetary Policy: A Review of Recent Empirical Evidence" Journal of Economics and Business, 52(1/2), 205-228, January/April. Sims, Christopher (1999), \Drift and Breaks in Monetary Policy," mimeo. Smets, Frank (1998), \Output Gap Uncertainty: Does it Matter for the Taylor Rule?" BIS Working Paper No. 60, November. Smets, Frank (2000), \What Horizon for Price Stability?" European Central Bank. Svensson, Lars E. O. (1997), \Inflation Forecast Targeting: Implementing and Monitoring Inflation Targets," European Economic Review, 41(6), June, 1111-1146. Svensson, Lars E. O. (1999), \Inflation Targeting as a Monetary Policy Rule," Journal of Monetary Economics, 43(3), June, 607-654. Taylor, John B. (1993), \Discretion versus Policy Rules in Practice," Carnegie-Rochester Conference Series on Public Policy, 39, December, 195-214. Taylor, John B., (1998), \The Long Boom," Federal Reserve Bank of St Louis Review, November/December, 3-11. Taylor, John B. (1999a), \An Historical Analysis of Monetary Policy Rules," in Taylor (1999c). 24

Taylor, John B. (1999b), \The Robustness and E(cid:14)ciency of Monetary Policy Rules as Guidelines for Interest Rate Setting by the European Central Bank," Journal of Monetary Economics, 43(3), 655-79, June. Taylor, John B., ed., (1999c), Monetary Policy Rules, Chicago: University of Chicago. Taylor, John B. and Michael Woodford, eds. (1999), Handbook of Macroeconomics, Amsterdam: Elsevier. Wieland, Volker (1998), \Monetary Policy and Uncertainty about the Natural Unemployment Rate," Finance and Economics Discussion Series, 98-22, Board of Governors of the Federal Reserve System, May. Williams, John C. (1999), \Simple Rules for Monetary Policy," Finance and Economics Discussion Series, 99-12, Board of Governors of the Federal Reserve System, February. Woodford, Michael (2000), Interest and Prices, manuscript. 25

Table 1 Estimated Policy Rules (cid:11) (cid:12) γ (cid:26) SEE R(cid:22)2 i = 1 1966:1{1979:2 1:53 1:64 0:57 0:70 0:81 0:86 (1:31) (0:38) (0:12) (0:07) 1979:3{1995:4 1:31 1:80 0:27 0:79 1:19 0:90 (1:84) (0:48) (0:30) (0:11) i = 2 1966:1{1979:2 2:12 1:61 0:60 0:67 0:80 0:87 (1:39) (0:36) (0:13) (0:08) 1979:3{1995:4 1:07 1:85 0:24 0:78 1:18 0:90 (1:83) (0:50) (0:23) (0:09) i = 3 1966:1{1979:2 2:13 1:65 0:62 0:69 0:88 0:85 (1:80) (0:42) (0:15) (0:08) 1979:3{1995:4 0:80 1:89 0:19 0:76 1:17 0:90 (1:56) (0:43) (0:19) (0:07) i = 4 1966:1{1979:2 3:53 1:44 0:61 0:72 0:95 0:84 (1:85) (0:41) (0:21) (0:10) 1979:3{1995:4 0:54 1:95 0:17 0:74 1:14 0:90 (1:41) (0:38) (0:15) (0:05) Notes: The table presents NLLS estimates of: f t = (cid:26)f t−1+(1−(cid:26))((cid:11)+(cid:12)(cid:25) t;ijt+γy tjt)+(cid:17) t for i 2 f1;2;3;4g: Robust standard errors in parentheses. f t is the federal funds rate (in percent per year), y tjt the output gap estimate for quarter t (in percent), and (cid:25) t;ijt the forecast of inflation from quarter t to quarter t+i (in percent per year). All regressions for the 1979:3{1995:4 sample have 66 observations. For the 1966:1{1979:2 sample, 52, 44, 36 and 28 observations are available for the 1-, 2-, 3- and 4-quarter ahead forecast horizons, respectively. 26

Table 2 P-values of Subsample Stability Tests Forecast Horizon 1 2 3 4 All parameters 0:068 0:046 0:019 0:022 (cid:11) 0:217 0:318 0:155 0:142 (cid:12) 0:316 0:334 0:163 0:111 γ 0:033 0:014 0:003 0:002 (cid:26) 0:127 0:081 0:156 0:476 Notes: Theentriesreflectp-valuesofparameterstabilitytestsacrossthesubsamples1966:1{ 1979:2 and 1979:3{1995:4. Columns correspond to the four alternative forecast horizons examined. For each horizon, the (cid:12)rst row examines the hypothesis of joint constancy of all parameters as shown in Table 1. Each of the remaining rows examines the hypothesis that the speci(cid:12)c parameter shown is constant, under the assumption that remaining parameters are constant. 27

Table 3 Policy Activism Index Forecast Horizon 1 2 3 4 1966:1{1979:2 0:90 0:99 0:95 1:37 (0:40) (0:48) (0:51) (1:04) 1979:3{1995:4 0:34 0:28 0:21 0:18 (0:45) (0:78) (0:22) (0:16) Notes: The index is computed as γ=((cid:12) −1). Entries are based on the parameter estimates shown in Table 1 for the corresponding forecast horizons and sample periods. Standard errors in parentheses. 28

Figure 1 Federal Funds Rate and Inflation Percent 18 17 Federal Funds Rate 16 Inflation 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 Notes: Inflation reflects the quarterly change in the chain-weighted GDP price index (January 2001 data, percent annual rate). The federal funds rate is the quarterly average of daily e(cid:11)ective rates. The solid and dashed vertical lines represent NBER business cycle peaks and troughs, respectively. 29

Figure 2 Real Federal Funds Rate and Inflation Forecast Percent 12 Real Federal Funds Rate 11 Inflation Forecast 10 9 8 7 6 5 4 3 2 1 0 -1 -2 -3 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 Notes: The inflation forecast is the one-quarter-ahead forecast of the change in the implicit output deflator (percent annual rate). The real rate is the federal funds rate minus the inflation forecast. See also notes to Figure 1. 30

Figure 3 Real Federal Funds Rate and Output Gap Percent 12 10 8 6 4 2 0 -2 -4 -6 -8 -10 -12 Real Rate -14 Output Gap -16 -18 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 Notes: The output gap shown is based on within-quarter forecasts for the quarter shown. See also notes to Figures 1 and 2. 31

Figure 4 Federal Funds Rate and Estimated Notional Targets Percent 20 19 Federal Funds Rate 18 1966:1-1979:2 17 1979:3-1995:4 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 0 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 Notes: Theestimatednotionaltargetscorrespondtotheestimatesfortheone-quarter-ahead horizon (i = 1) shown in Table 1. 32

Figure 5 Real-Time Perceptions and Ex Post Concepts of the Output Gap Percent 6 5 4 3 2 1 0 -1 -2 -3 -4 -5 -6 -7 -8 -9 -10 -11 -12 -13 Real-Time -14 HP Trend -15 Quadratic Trend -16 -17 -18 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 Notes: The real-time output gap shown is based on within-quarter forecasts for the quarter shown. The HP Trend and Quadratic Trend concepts reflect detrending of current data. See also notes to Figure 1. 33

Figure 6 Inflation Forecast and Ex Post Outcomes Percent 15 14 13 12 Forecast Actual 11 10 9 8 7 6 5 4 3 2 1 0 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 Notes: The one-quarter-ahead forecast of inflation is from Figure 2. The ex post actual rate of inflation is that shown in Figure 1, shifted one quarter to allow a direct comparison with the forecast. 34