The Anatomy of Single-Digit Inflation in the 1960s

Finance and Economics Discussion Series Federal Reserve Board, Washington, D.C. ISSN 1936-2854 (Print) ISSN 2767-3898 (Online) The Anatomy of Single-Digit Inflation in the 1960s Jeremy B. Rudd 2022-029 Please cite this paper as: Rudd, Jeremy B. (2022). “The Anatomy of Single-Digit Inflation in the 1960s,” Finance and Economics Discussion Series 2022-029. Washington: Board of Governors of the Federal Reserve System, https://doi.org/10.17016/FEDS.2022.029. NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminary materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors. References in publications to the Finance and Economics Discussion Series (other than acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers.

The Anatomy of Single-Digit Inflation in the 1960s Jeremy B. Rudd Federal Reserve Board* April 9, 2022 Abstract Recently, the experience of the 1960s—when the U.S. inflation rate rose rapidly and persistently over a comparatively short period—has been invoked as a cautionary tale for the present. An analysis of this period indicates that the inflation regime that prevailed in the 1960swasdifferentinseveralkeyregardsfromtheonethatprevailedontheeveofthepandemic. Hence, there are few useable lessons to be drawn from this experience, save that monetarypolicymakingremainsadifficultundertaking. *WithapologiestoBlinder(1982). E-mail: jeremy.b.rudd@frb.gov. Theanalysisandconclusionssetfortharemy ownanddonotnecessarilyreflecttheviewsoftheBoardofGovernorsorthestaffoftheFederalReserveSystem.

There are things of which the mind understands one part, but remains ignorant of the other; and when manisabletocomprehendcertainthings,itdoesnotfollowthatheisabletocomprehendeverything. MosesMaimonides,TheGuideforthePerplexed,I:31 I Introduction Economists—particularly economic forecasters—often try to understand current economic developmentsbydrawinganalogiestohistoricalepisodes. Anepisodethathasreceivedparticular attentionoflate,especiallyforthinkingabouttheoutlookforinflation,istheaccelerationinprices that occurred in the United States over the second half of the 1960s. Under the conventional telling, the pursuit of a “guns and butter” fiscal policy, combined with inappropriately accommodative monetary policy, caused the economy to overheat. The result was a relatively rapid andpersistentincreaseininflationoverthesecondhalfofthe1960sthatinturnlaidthegroundwork for the Great Inflation of the 1970s.1 This account of the 1960s has recently been invoked by a number of commentators to oppose expansions in government spending—in particular, the $1.9 trillion American Rescue Plan that wasenactedonMarch11,2021.2 Somehavealsousedtheexperienceofthe1960stopaintan especially bleak picture of the current economic outlook, arguing that in the same way that the overlyexpansionaryfiscalandmonetarypoliciesofthe1960slefttheU.S.economysusceptible to the energy and food price shocks of the 1970s, similar recent “policy errors” have once again “brought the U.S. to the brink of stagflation.”3 In this essay, I attempt to provide an analytical description of the sources and nature of the 1960s inflation increase. I argue that there is very little that we can learn from this period that canhelpustounderstandcurrentinflationdevelopmentsortheinflationoutlook,inasmuchasthe inflationregimethatprevailedinthe1960swasdifferentinseveralkeyregardsfromtheonethat has prevailed in recent decades. Moreover, we cannot really say why those features of 1960s inflationdynamics—namely,asteeptradeoffbetweeninflationandactivity,anunanchoredlongrun inflation trend, and strong and persistent feedback between wage and price growth—were presentatthattime,norcanweexplainwhytheywentawayorwhatmightleadthemtoreappear. Hence,whileitisconceivablethatasignificantbreak(akintoaregimechange)mightoccurthat would return the inflation process to its earlier form, the experience of the 1960s provides no guidance as to the probability of such an event or what policies would prevent it from occurring. 1Meltzer(2009)evengoessofarastolabelthesecondhalfofthe1960s“PhaseI”oftheGreatInflation,datingits startto1965“...whenbasevelocityrosenearly4percent”(p.368). However,Meltzerfailstoalsonotethatvelocity growthslowedthereafter(andeventurnednegativein1967), withthelevelofbasevelocityfinishingthedecadein linewithits1959–1964trend. 2See,forexample,Blanchard(2021),Mankiw(2021),andSummers(2021). (Theamountcitedinthetextrefers totheestimatedcumulative10-yeareffectoftheplanonthefederaldeficit.) 3Summers(2022). 1

II What happened (to inflation) in the 1960s? Ifyourememberthesixties,youreallyweren’tthere. Attributedtovarious4 AnalyticalstudiesofpostwarU.S.inflationtypicallystartin1959,asmeasuresof“core”inflation— which exclude changes in food and energy prices—do not exist prior to this date.5 However, it is possible to construct a price series for the 1950s that removes the prices of food and energy goods; figure 1 plots the four-quarter change in this series (defined for personal consumption expenditures, or PCE) together with the corresponding change in the total PCE price index over the period 1953–1971. The figure indicates that inflation had declined noticeably coming into the1960s;from1957to1967,itspathlooksmorelikea“U”thana“J.”(Acomparisonofthetwo seriesalsorevealsthatpricechangesforfoodandenergygoodsmaderelativelymodestcontributions to overall PCE inflation over this period.) Seen over this longer span, it is the first half of the 1960s—when inflation was relatively low and stable—that appears somewhat anomalous. Whenthepickupininflationdid occur,however,itwasbroad-based. Figure2plotsthreeinflation seriesthatarecomputedforthebottomandtop25percentofthefulldistributionofpricechanges andfortheremainingmiddleportionofthedistribution.6 Eachseriesmeasuresinflationasa12month change (using a 12-month change helps to deal with the possibility that prices might “bouncearound”betweendifferentpartsofthedistributionathigherfrequencies),withtheupper panel covering all PCE prices and the lower panel covering only the core. The rise in inflation over the second half of the decade clearly appears in each portion of the distribution, including the lower tail. The increase in inflation over this period was nearly coincident with an increase in trend unit labor cost growth—figure 3—which is measured here as the change in hourly compensation in the nonfarm sector minus an estimate of trend productivity growth.7 This co-movement certainly hintsat the presenceof atwo-wayfeedbackbetween wageand price growth,which is the underlying mechanism of a “wage–price spiral.” We might speculate that one source of this feedback was institutional. Formal cost of living adjustments (COLAs) tied to the CPI were a feature of some labor contracts in the 1960s. If 4IncludingPaulKantnerandCharlieFleischer. 5Thefocusoncoreinflationreflectstheimportanceoffoodandenergypriceshocksduringthe1970sandearly 1980s; such shocks are generally (though not universally) viewed as having resulted from factors unrelated to the stanceofmonetarypolicyorlevelofresourceutilization. 6IamindebtedtoErickSagerforsuggestingthisapproachtodescribingthepricechangedistribution. 7Detailsontheconstructionofthesevariablesisprovidedintheappendix. Therateoftrendproductivitygrowth actuallystartstoslowaroundthemiddleofthedecade—aphenomenonthatwasnotedalmostcontemporaneouslyby Perry(1971)andNordhaus(1972)—butthedeclineinmymeasureisrelativelysmall,andvirtuallyallofthevariability inlaborcostsshowninfigure3isattributabletonominalcompensationgrowth. 2

theseclauseswerewidespread,itwouldhaveresultedinanear-automaticdependenceofwage growthoninflation. However,thenumberofworkerscoveredbyCOLAprovisionspeakedaround 1958–1960 and then fell back sharply over the next several years. Although the number of covered workers turned up again in the second half of the decade—partly in response to the higherconsumerpriceinflationthathadbeguntoemerge—itremainedwellbelowitslate-1950s peak, and by 1969 represented only about 41⁄ percent of the private workforce.8 2 Moreover, many escalation clauses did not provide for full indexation of wages to the CPI: What evidencewehaveindicatesthateveninthetworelativelyhigh-inflationyearsof1968and1969, a third or less of the increase in the CPI was reflected in wage increases for workers covered by these types of contracts.9 These and other findings led one well-known study of contractual wageescalationandinflationtoconcludethatCOLAprovisions“playednopart”inoriginatingthe 1960s inflation, and that their role in sustaining the inflation was likely not very great.10 Hence, it appears that much of the dependence of wage growth on inflation in this period must reflect a mechanism that is less formal than direct (contractual) indexation. Discussions of the 1960s often identify excessive rates of resource utilization as the proximate cause of the acceleration in wages and prices. However, conventional measures of utilization— specifically, the unemployment gap or output gap—tend to be informed in some manner by the behavior of inflation. (For example, one common way to measure the “natural” rate of unemployment involves backing it out from a Phillips curve.) As a result, it is difficult to gauge when utilization rates started to put significant upward pressure on inflation without simply coming full circle and stating that it was around the time that inflation started to rise. Withthatcaveatinmindconsiderfigure4,whichplotstheactualunemploymentrateagainstthe CBO’sestimateofthenaturalrate.11 TheCBOmeasurestartsthedecadeatabout51⁄ percent, 2 which implies that the economy began moving into a sustained period of high utilization around 1963.12 However, attempts to use the resulting unemployment gap in a price equation typically also find a role for a constant term that—if we attribute it to mismeasurement of the natural rate—implies an even higher average level of utilization over this period and a correspondingly earlierdateofentryintothehigh-utilizationperiod. Onthelowerendoftherange,theestimated NAIRU in Staiger, et al. (2001) declines by more than a percentage point over the course of the 8Seetable1ofDouty(1975). 9Seetable8ofDouty(1975). 10Douty(1975),pp.53–54. 11From1948 to 2004, the CBO’s naturalrate estimate is essentially a NAIRU measurewith demographicadjustments(seeShackleton,2018,appendixB).Hence,themodestincreaseintheCBO’snaturalrateestimateoverthe 1960s(0.4percentagepoint)reflectstherelativeunimportanceofchangesindemographicsduringthisperiod. 12Bycontrast,manufacturingcapacityutilizationmovesupsharplyattheendof1964,reachesanear-record-high levelin1966,andthendropsbackin1967toalevelthatisnottoodifferentfromtheaveragerateoverthesecond half of the 1950s, while the worker quit rate in manufacturing looks more like the mirror image of the aggregate unemploymentrate. 3

1960s, ending the decade at 41⁄ percent. 2 III Inflation dynamics in the 1960s (and now) Therewassomethingspecialaboutthesixties...Butifyouaskedmetobemorespecific,topinpointwhat itwasaboutthesixtiesthatwassospecial,Idon’tthinkIcoulddomorethanstammeroutsometritereply. HarukiMurakami(2006) How might we try to make sense of these historical events in a coherent way? The approach I take here involves using a VAR model that allows for time-varying parameters and stochastic volatility, and that includes enough relevant drivers of inflation to have at least some chance of capturing and describing key features of the inflation process.13 An approach like this one is notperfect,asitrequiresusingamore-parsimoniousdynamicspecificationthanwouldbefound in other types of empirical wage and price equations, such as a Phillips curve. (It also inherits the same questionable claim to being a structural model that any recursively identified VAR model does.) But by using a framework that explicitly models parameter drift and incorporates information from the full sample, the approach does have some appeal; in addition, allowing for changes in the volatility of shocks over time reduces the chances of identifing a change in the parameters of the inflation process when none is actually present. TheVARsystemthatIuseincludesarelativeimportpricemeasure,trendunitlaborcostgrowth, core market-based PCE price inflation, and the difference between actual unemployment and the CBO’s current estimate of its natural rate, with that causal ordering.14 Because most of the variation in trend unit labor cost growth reflects changes in compensation growth (rather than changes in trend productivity growth), including this variable allows us to gain some insight into how wage determination might have varied over time; more importantly, it allows the model to capture the dynamics associated with wage–price spirals in periods when they are present. The first mildly interesting result that obtains from the VAR estimates is shown in figure 5. The dashed line in the figure plots the VAR’s baseline forecast for core inflation, which moves up steadily over the decade. This rise in the model’s baseline projection in turn reflects a value for 13SuchanapproachhasbeenusedbyanumberofauthorstoanalyzehistoricalchangesinU.S.inflationdynamics (CogleyandSargent,2005, isonewell-knownexample). Theparticularmodelusedhereissimilartooneusedby PenevaandRudd(2017)intheirstudyofwage–pricepassthrough;thatstudyisinturncloselybasedonearlierwork byClarkandTerry(2010)thatusedaVARmodelofthistypetoanalyzewhetherandhowthepassthroughofenergy priceshockstoinflationhasvariedovertime. 14I use market-based PCE prices in the VAR because several nonmarket components of PCE are priced using input cost indexes that are in turn based on wage or compensation measures. (On average, core market-based prices account for roughly 90 percent of the overall core index.) In addition, I use PCE prices rather than the CPI because the published CPI contains methodological breaks and uses mortgage interest rates to measure housing costsinthe1960sand1970s. 4

theestimatedlong-runmeanforinflationthatisaboveactualinflationformuchofthedecade,and that moves up by 13⁄ percentage points from 1961 to 1967 (see the dashed line in figure 6).15 4 In addition, the rise in the stochastic trend for price inflation is almost exactly mirrored by an increase in the stochastic trend for labor cost growth (the dashed line in figure 7). According to the VAR, what kept the rise in trend inflation from showing through to actual inflation over the first part of the 1960s was unusually low wage growth: As indicated by the blue dotted line in figure5,theincreasingtrendwasobscuredbyasequenceofnegative(andultimatelytransitory) idiosyncratic shocks to labor cost growth.16 While this result provides additional evidence that theapparentsuddennessoftheriseininflationinthesecondhalfofthe1960swasexaggerated by unusually low inflation earlier in the decade, it is the instability in the stochastic trends for inflationandlaborcostgrowththatrepresentsthemostnoteworthyfeatureofinflationdynamics in this period—one that seems to be completely absent in more-recent decades.17 One (mechanical) contributor to the rise in inflation’s long-run mean over this period is worth highlighting, even though it is difficult to come up with a compelling explanation for what it is capturing. Thetime-varyinginterceptintheVAR’sinflationequationisclosetodriftlessformuch of the estimation period, fluctuating in a narrow range (albeit one with a relatively wide credible set)untilthelate1990s. However,overthe1960sthereisasteadyincreasein“intrinsic”inflation persistence, as captured by the sum of the coefficients on the own lags in the inflation equation (see figure 8).18 All else equal, higher own-persistence will increase inflation’s long-run mean eveniftheinterceptoftheinflationequationremainsconstant. Butwhateconomicinterpretation might we give to that increase? One possible interpretation is suggested by Cogley, et al. (2010), who associate the stochastic trendininflationwiththeFed’sinflationtarget. Underthattelling,theown-persistenceofinflation at a point in time contributes to the persistence of the “inflation gap,” which they define as the deviation between actual inflation and the Fed’s long-run inflation goal; a rise in the persistence of the inflation gap might then be attributed to the Fed’s becoming less willing to bring actual inflation back to its target quickly.19 Here the interpretation runs into trouble, though. As noted, 15Theinflationtrendinfigure6doesnotmatchtheVARbaselineforecastshowninfigure5becausethelattergives thebesti-periodaheadforecastimpliedbytheVARstartingfromsomeinitialperiod(andconditioningontheinitial period’sdata),whilethelong-runtrendisthevaluetowhichinflationwouldeventuallyconvergeabsentanyshocks (aBeveridge–Nelsontrend). 16The model attributes virtually all of the remaining weakness in inflation to own-shocks. The high correlation betweenwageandpricegrowththatexistedoverthisperiodmakesithardtobetooconfidentaboutthisbreakdown; thatsaid,reversingtheorderingoflaborcostsandinflationintheVARyieldsacontributionoflaborcostinnovations toinflationthatisonly0.4percentagepointsmallerattheendof1965. 17As discussed in Peneva and Rudd (2017) and Rudd (2022), inflation’s stochastic trend has been essentially invarianttochangesineconomicconditionssincethemid-1990s,inmarkedcontrasttotheexperienceofthe1970s and1980s(andapparentlytheexperienceofthe1960saswell). 18Inaddition,anincreaseinthesensitivityoflaborcostgrowthtoinflationhastheeffectofmodestlyincreasingthe “effective”weightonlaggedinflationinthepriceequation. 19Thisdiscussionisloose—thepersistenceofthe“inflationgap”inthesenseofCogley,etal.(2010)andthevalue 5

eveniftheinterceptintheinflationequationisfixed,ariseinown-persistencewillraiseinflation’s long-runmean. Butitisunclearwhyweshouldthenassociatetheresultingincreaseininflation’s stochastictrendwithanincreaseintheFed’slong-runinflationtarget,whichseemsasthoughit wouldbemorenaturallycapturedbyanincreaseintheinflationequation’sintercept. (Ofcourse, we might be skeptical that a model such as this one would actually be capable of making this sort of distinction, which further complicates the task of giving economic—let alone structural— interpretations to the VAR’s trends.) The estimated stochastic trend for the unemployment gap over this period (not shown) is persistently negative.20 Translating this long-run trend into unemployment-rate terms (the solid red lineinfigure9)providesanestimateofthe“natural”rateofunemploymentinthesenseofFriedman(1968)—thatis,theunemploymentratethattheeconomytendstoreturntoonceanyshocks have died out and all other variables in the system are at their long-run values. The model’s estimated natural rate is relatively flat for much of the decade at around 41⁄ percent, and ends the 2 decadebelow5percent.21 Takenatfacevalue,themodel’sestimateimpliesthattheunemployment rate did not fall appreciably below the natural rate until mid-1968. However, this definition of the gap—and, by extension, the model’s estimate of the natural rate—is not necessarily relevant as a gauge of wage and price pressures: The stochastic trend for the unemployment rate istryingtomeasurethelong-runmeanoftheseriesinawaythatbestfitsitsowndynamics,not in a way that best explains wage or price inflation.22 Finally,priceinflationappearedtobemoresensitivetorealactivityinthe1960s. Figure10plots the response of inflation following a shock to the unemployment gap at various dates, together with integral multipliers (the values in parentheses) for the eight quarters following the shock.23 Theseresultsconfirmthat theprice Phillips curvewas steeper in the1960s thanin more-recent decades,buttheyalsosuggestthatthisfeatureofthePhillipscurvewasalreadyinplacebefore the mid-1960s inflation surge occurred. A similar conclusion obtains for the responses of labor costgrowth(figure11),exceptthatherethequantitativeeffectofashocktorealactivityismuch more similar over the entire sample period. ofinflation’sstochastictrendbothdependonthefull-systempropertiesoftheVAR—butitisnottoomisleadinginthe currentcontext. 20Mostoftheoverallmovementintheunemploymentgapoverthisperiodis“explained”bytheVARwithacombinationofvariationintheinterceptofthegapequationitselfandidiosyncraticown-shocks(eventhoughtheunemploymentgapisorderedlastintheVAR). 21Bywayofcomparison,thenaturalrateestimateobtainedbyCogleyandSargent(2005,fig.5)usingasomewhat differentspecificationisalittleabove5percentin1961anddeclinesto41⁄ 2 percentby1965. 22A similar issue attends state-space models that try to use both labor-market and inflation data to inform their estimateofthecyclicalpositionoftheeconomy: Theresultinggapserieshastobalancetheneedtofitthedynamics ofthelabormarketvariableswiththeneedtofitthedynamicsofinflation,anditmightnotcareallthatmuchabout howwellitaccomplishesthelattergoal(especiallyifthePhillipscurveisrelativelyflat). 23Theseintegralmultipliersaretheratioofthecumulatedeight-quarterresponseofinflationtothecumulatedeightquarterresponseoftheunemploymentgap;theythereforecontrolforanytimevariationinthegap’sownresponse. 6

Intotal,theseresultsprovideashredofevidencethatthesharpaccelerationinpricesandwages that occurred after 1965 was probably not the result of a discrete break or “regime shift” in the inflationprocess.24 TheslopeofthepricePhillipscurveappearstohavebeenroughlythesame oneithersideofthe1960sinflationbreakout;moreimportantly,thelong-runtrendsforlaborcost growth and inflation had already moved up by a percentage point by mid-decade.25 Likewise, wage–price spirals were certainly a feature of the inflation landscape in the 1950s, making it unsurprising that they could have emerged again in the 1960s. If we buy into this conclusion (note the conditional), we are left with the job of coming up with a plausible explanation for why large negative shocks to wage growth (and inflation) emerged in the first half of the decade and then rapidly vanished. Since we can give unexplained residuals whateverinterpretationwelike,wemightventureastoryinwhichthedisappearanceofthenegativeown-shockstolabor-costgrowththatweseeinfigure5reflectedanewfoundattentiveness to inflation on the part of households and firms, or perhaps even a shift in worker bargaining power as the labor market tightened. But it is difficult to see from the PCE price data what the precipitatingfactormighthavebeenintheformercase(givenhowlowandstableactualinflation was until the mid-1960s), or why a change in worker bargaining power wouldn’t have also been associated with a steepening of the wage Phillips curve (which we don’t find). On the other hand, Douty (1975) does specifically identify 1965–1966 as a watershed, arguing thatabroad-basedaccelerationintheCPIatthattimecoupledwithlargeincreasesinfoodprices meant that “...rising prices became an explicit factor in wage determination” by the middle of 1966.26 ThebreadthofthepriceincreasesDoutypointstoisinfactevidentinthePCEpricedata (recallfigure2). However,thecontributiontooverallinflationmadebythemid-1960sacceleration in food prices is only really striking if we look at the CPI—see figure 12—mainly because food had a much larger weight in the CPI in this period. (Of course, during this time only the CPI 24TheseresultsalsocallintoquestionReis’s(2021)characterizationofthe1960sinflationexperience.Reisargues that inflation was “anchored” prior to 1965 thanks to the Fed’s ability to control inflation after 1951 (something not immediately evident from figure 1), and constructs a series for an “expected inflation anchor” that rises more than 2percentagepointsbetween1967and1970(theseriesthendropsabout11⁄ 2 percentagepointsin1971). However, theresultspresentedheresuggestthattrendinflationwasincreasingsteadilyoverthefirstpartofthe1960s,withno decline thereafter (the trend estimates continue to rise after 1971—not shown). It should also be noted that Reis’s paperdoesnotdemonstrateacausallinkageinwhichhisexpectationsmeasureinfluencesactualinflation(asAlan Blinder’sdiscussionofthepapernotes,Reis’s“anchor”appearstosimplymoveinlinewithtotalCPIinflation).Absent such a demonstration, I would argue that Reis’s measure does not provide any useful insights into the state of the inflationprocessduringthisperiod. 25Wemightbeconcernedthatthesteadyincreaseintheselong-runtrendsreflectsthemodel’stryingtosmooth throughasuddenshiftthathappenedlaterinthedecade. Butstochastictrendsfromthistypeofmodelcanactually change relatively rapidly when they want to—for instance, in the early 1980s the estimated trends for inflation and laborcostgrowthdeclinebyabout3percentagepointsoveraneight-quarterperiod. 26Douty(1975),p.9.Cecchetti(1987)questionsDouty’sconclusion,arguingthatthefrequencyofwageadjustment appears unrelated to the level of inflation over this period. But Cecchetti’s finding bears more directly on the slope ofthewagePhillipscurve(notonthedependenceofwagegrowthonpastinflation)—andasfigure11suggests,the responseoflaborcoststoachangeinrealactivitywasindeedquitestableinthe1960s. 7

would have been available as an indicator of consumer prices.) An alternative explanation for the negative wage (and price) shocks and their subsequent reversal that seems at least equally (im)plausible is that these shocks reflect the initial success and later collapse of the Kennedy–Johnson wage–price guideposts. Regarding their timing and effect, one study (Congressional Budget Office, 1977) argued that the guideposts “...worked fairlywellinmoderatingpriceandwageincreasesthrough1965”(p.17),breakingdownin1966 as “major labor unions’ antagonism toward the policy intensified” (p. 19). But while there are certainly examples from this period where high-level intervention by Administration officials was able to enforce the guideposts in particular sectors, it is less clear whether adherence to the guideposts was widespread enough to explain aggregate wage and price developments. IV Fiscal policy in the 1960s (and now) Abillionforthis,abillionforthat,abillionforsomethingelse... SenatorEverettDirksen(circa1961) The VAR model does not contain any direct measure of fiscal policy, instead capturing any change in aggregate demand from this source through its effect on the unemployment gap. It is undeniable, though, that sustained expansionary fiscal policy played a major role in pushing up overall spending in the 1960s. What quantitative statements can we make about fiscal policy actions in the 1960s, and how do those actions compare in size to the ones taken in response to the pandemic? Figure 13 plots a measure of fiscal stance that—in contrast to measures like the actual or highemploymentbudgetdeficit—attemptstocontrolforthedifferentialeffectsthatgovernmentspending and taxation have on aggregate demand. This measure, which is labelled “fiscal impetus” in the chart, gives the first-round contribution to Q4-over-Q4 real GDP growth from discretionary federalandstateandlocalfiscalpolicies. Inordertoprovidearoughideaofthecombinedeffect of current and past fiscal actions in this period, the plot also shows the contribution to output growthifanexpendituremultiplierisappliedtotheautonomousfiscalimpulses.27 Theestimates in figure 13 suggest that fiscal policy did in fact turn strongly expansionary after 1964; to put thesemagnitudesintoperspective,asimpleOkun’sLawrelationimpliesthattheunemployment rate would have been 2 percentage points higher at the end of 1969—so around 51⁄ percent— 2 27ThefiscalimpetusestimatesarebasedonFolletteandLutz(2011). Forthemultipliercalculation,Iassumean aggregatemarginalpropensitytospendof0.4(implyinganeventualexpendituremultiplierof1.7),whichisobtained byweightingconventionalmeasuresofexcesssensitivityinconsumptionandoftheoutputelasticitiesofinvestment andimportdemandby1964nominalGDPshares. (Acomparablecalculationusing2019sharesyieldsamultiplier of1.5,mostlybecauseimportleakagesarelargernow.) 8

hadfiscalpolicy’scontributionfrom1965to1969insteadsimplybeeninlinewithCBO’sestimate of potential output growth for that period.28 Moreover, fiscal policy remained a contributor to aggregate demand growth well into the late phases of the 1961–1969 expansion—in sharp contrast to today. Figure 14 plots the raw fiscal impetusmeasure(thatis,withoutanymultipliercontribution)againsttheCBOoutputgapovera longer period; as is evident from the chart, the output gap had closed around the time that the post-1964fiscalexpansionstartedgettingunderway. Morerecently,thefiscalmeasuresputinto place during and after the Covid pandemic—while extremely large—are also largely temporary, implying that they will become a drag on GDP growth in later years.29 V Monetary policy in the 1960s To try effectively to wipe out hard-core inflation by squeezing the economy is possible but disproportionatelycostly. Itisburningdownthehousetoroastthepig. RobertM.Solow(quotedinMermelstein,1979) William McChesney Martin, the chairman of the Federal Reserve during this period, was a dedicated inflation fighter (he had pushed to tighten policy to counteract inflation in the 1950s, thereby contributing to the 1957–1958 recession).30 Martin also put an emphasis on inflation expectations—as both a driver of actual inflation and as a contributor to the persistence of inflation—that would not be out of place in academic or policy circles today.31 Finally, a numberofFOMCmembersbecameincreasinglyconcernedaboutrisinginflationasthe1960swore on.32 Sowhydidmonetarypolicyfallfurtherandfurther“behindthecurve”overthesecondhalf of the decade? • Overreliance on unobservables. As Orphanides (2003a, b, 2004) has argued, the actual federal funds rate over this period was in line with what a standard Taylor-type rule (using data available at the time) would have prescribed—in particular, the Fed’s response to price de- 28Noneofthisisnews,ofcourse—seeBlinderandGoldfeld(1976)forarelatedexercisefrom50yearsago. 29Thefiscalstanceestimatesfor2020to2023thatareshowninfigure14arebasedontheBrookingsInstitution’s HutchinsCenterfiscalimpactmeasure. Asthismeasureiscalibratedtobezerowhenspendingandtaxesarerising withpotential,IadjustitusingCBO’sestimatesofpotentialoutputgrowthinordertomakethemmorecomparableto theFollette–Lutzmeasureshownoverhistory. (Theseadjustmentsaddabout0.35percentagepointtotheBrookings measureineachyear.) 30Bremner(2004),pp.126–129. 31SeeBremner(2004),p.266;alsoMeltzer(2009),p.155,n.176. Ofnote,Martinsoundlyrejectedthelate-1950s argumentthat“creepinginflation”wastolerableoraninherentfeatureoftheU.S.economy(Bremner,2004,pp.127– 129), which contradicts Meltzer’s (2009) claim that this view kept the Federal Reserve from using monetary policy moreaggressively(pp.530–531). 32Bremner(2004),pp.253–255;Meltzer(2009),p.533. 9

velopments conformed to the “Taylor principle” that the nominal rate should move more than one-for-one with actual or expected inflation. Orphanides concludes that the real problem was that the Martin Fed was targeting an estimate of potential output that—with the benefit of hindsight—was too high to deliver stable inflation.33 • Financial stability considerations. The Federal Reserve started tightening in December 1965, resulting in a credit crunch in mid-1966. In contemplating another tightening cycle in 1968, Martin was apparently wary of repeating the 1966 episode by moving too abruptly.34 In addition, over the course of the decade financial markets made increasing use of hedging and othershort-runinvestmentstrategiesthatraisedtheriskoffinancialinstabilityweretheFedto disappoint markets.35 • Desire for a fiscal policy retrenchment. Fed Chairman Martin viewed fiscal policy as too expansionary over this period, and started advocating for fiscal restraint (in the form of a tax increase) in mid-1965.36 Martin pushed harder for a tax increase toward the end of 1966 whentheJohnsonAdministrationreleasedareviseddeficitprojectionforfiscalyear1967and its first estimate for 1968.37 Martin felt that additional monetary tightening would take the pressure off of Congress and the Administration to rein in fiscal policy; by the end of 1967, however, he realized he had waited too long.38 (A surtax was finally enacted in 1968.) • Overreliance on model forecasts. Once the surtax was in place, model-based forecasts implied a larger negative effect on aggregate demand than what actually occurred, leading the Fed to try to cushion the anticipated effect on real activity with monetary easing that was—in retrospect—toomuch. Martinadmittedthat“[t]heSystemhadpulledaboner”bybasingeasier policy on these projections.39 33Part of the reason involved maintaining an estimate of the “natural” rate of unemployment that was—again in retrospect—too low to deliver stable inflation (the CEA kept its estimate of the natural rate at 4 percent until 1970, when it was lowered to 3.8 percent). With the flattening of the empirical Phillips curve in recent decades, the uncertaintyaroundsuchnaturalrateestimates—whichwasalreadyhighenoughtomakethembasicallyworthlessfor policy purposes—has only become greater. (For different reasons, the dislocations caused by the pandemic have madethissituationexponentiallyworse.) 34Bremner(2004),pp.252,254–255. 35Bremner (2004), p. 256. One of the reasons Martin saw keeping inflation expectations in check as being so vital was that he thought higher expected inflation distorted economic and financial decisionmaking and made the economyvulnerabletoaboom–bustcycle(Bremner,2004,p.123). 36Bremner(2004),p.219. 37Bremner (2004), pp. 224–225. Martin had evidence that President Johnson was understating the fiscal costs associated with the intensification of the U.S. war in Vietnam: As early as 1965, FOMC staff had concluded that unreported military spending was causing their econometric model to underpredict real activity, while Martin had receivedconfidentialinformationalongtheselinesfromfriendsintheAdministration(Bremner,2004,pp.204–205). NorweretheFedtheonlyoneswithmisgivings:WhenArthurOkun,amemberofthePresident’sCouncilofEconomic Advisers,receivedamemoonwarcostsmarked“ForInternalUseOnly”headdedthenote“Butnottobeswallowed” (Halberstam,1993,p.607). 38Bremner(2004),pp.237–238;Meltzer(2009),pp.521,524. 39Meltzer (2009), p. 561. Staff forecasts had started to be incorporated into FOMC deliberations in 1966, and had become “an integral component in the FOMC policy-making process” by 1968 (Bremner, 2004, p. 253; also 10

• Concerns over Fed independence. The Fed faced numerous challenges to its independence over this period. In 1964, Martin had to ask President Johnson for help in dealing with Representative Wright Patman’s attempts to remake the Fed into an institution that would be less autonomous (Patman was chair of the House committee that oversaw the Fed).40 The Fed then faced pressure from the Administration to keep policy expansionary; when Martin advocated the 1965 policy tightening, he saw it as a way to reassert Fed independence, but acknowledged that it ran the risk of provoking Johnson.41 Later (in 1968), Johnson applied less-subtle pressure by appointing a task force to consider changes to the Federal Reserve System that would effectively increase the Administration’s influence on the FOMC.42 By 1968, however, it was clearly evident that inflation had ratcheted higher, and that the surtax had failed to cool the economy. So why didn’t the Federal Reserve push against inflation more aggressively in the last part of the decade? Were the considerations described above enough to preclude greater Fed tightening even at this late date, or was there another reason that might have loomed even larger in policymakers’ minds? Probably the simplest explanation for the Martin Fed’s inaction at this point is that policymakers did not want to incur the economic cost that they felt would be required in order to engineer a persistentreductionin inflation. Martin was clearly awarethat inflation had becomeentrenched, and attributed the cause to a rise in inflation expectations.43 At the September 1968 FOMC meeting, he stated that: “The existing momentum of inflationary pressures is because both fiscal and monetary restraint have come much too late. It is asking too much of the available tools of monetary policy to expect them to deal with the inflationary psychology that resulted from the delay.” (Bremner, 2004, p. 253) Martinwasalsoawarethatfullysqueezingoutthis“inflationarypsychology”wouldrequirealong and grueling campaign, one that he did not want to undertake without Administration support. He therefore gave a speech in May 1969 that suggested such a policy course in the hopes that President Nixon would take up the gauntlet and make a strong public commitment to fighting Meltzer,2009,p.498). 40Bremner(2004),pp.190–191. 41Meltzer(2009),p.456. JustpriortotheDecember1965discountrateincrease,agroupofthreeAdministration economistsandtheFederalReserveBoard’sdirectorofresearchproducedastudythatconcludedthatanaccelerationinpriceswasunlikely,andthatapolicytighteningshouldthereforebedeferred(Bremner,2004,p.206). Martin chosenottocirculatethestudy, fearingitwouldreducesupportforarateincrease. (Oneoftheeconomistsonthe Administration’s team was Paul Volcker, who was at that time Johnson’s deputy undersecretary of the Treasury for monetaryaffairs.) 42Meltzer(2009),p.549. 43A concern over rising inflation expectations had been one of Martin’s justifications for the 1965 discount rate increase—seeMeltzer(2009),p.480;alsoBremner(2004),p.206. 11

inflation. Nixon actually did broach the idea of taking “the bad medicine now” with his economic advisers, but was dissuaded by George Shultz, his Secretary of Labor.44 Chairman Martin’s view that a large amount of economic pain would be required to secure a persistent reduction in inflation was arguably vindicated by later events. Following the food and energy shocks of 1973–1974, core PCE inflation settled down to a 6 percent pace for the next three years.45 A second round of supply shocks got underway in 1978 (initially for food, later for energy),resultinginanotherlargespikeininflationthatledtheFed(underChairmanVolcker)to initiateanaggressiveanti-inflationpolicythatdeliveredtwoback-to-backrecessionsandapeak unemployment rate of just under 11 percent. Relative to 1975–1978, core inflation did end up falling persistently by a little more than 2 percentage points (core inflation averaged 3.8 percent from 1983:Q4 to 1990:Q4), the benefit of which should be weighed against the actual damage and potential risk to the economy that this policy entailed.46 Moreover, trend inflation moved downagain inthewakeofthe1990–1991recession,suggestingthatVolcker’sdrasticapproach to inflation control was not by itself sufficient to deliver an inflation regime in which inflation’s long-run trend was fully anchored. ButwhateverexplanationonechoosesfortheMartinFed’sfailuretoreinininflation,itisapparent from the documentary history that by 1969, monetary policymakers were basically out of ideas (see figure 15).47 VI What (if anything) have we learned? Veryfewpeopleareentitledtotheiropinion. JackDouglas Perhaps the most striking fact about the 1960s is that policymakers at the Federal Reserve held views about inflation—and faced concerns about how best to deal with it—that seem quite 44See Bremner (2004), pp. 266–268. Interestingly, Meltzer (2009) makes no reference to this event, in which Shultz—whowasanacolyteofMiltonFriedman—usedmonetaristargumentstopersuadeNixonthatfurthermonetaryrestraintwasunnecessary. 45TheaverageannualizedlogdifferenceinthecorePCEpriceindexwas6.15percentfrom1975:Q2to1978:Q2. 46Temin (1989), p. 40, even goes so far as to argue that a second Great Depression might have resulted “[h]ad thedeflationarypolicybeencontinuedforanothersixmonths.” Amoregenerousassessmentwouldattributethefull 31⁄ 4 percentagepointdeclineintrendinflationfrom1979to1983toVolcker’spolicyactions. However, thatvalueis likelyanoverstatementgiventhatpartofthereductionininflationreflectedtheunwindingoftheeffectsofthesupply shocks(therelativecontributionoffoodandenergypricestoPCEpriceinflationbecomesnegativeafter1981:Q1); relatedly, VAR specifications that control for the effects of the supply shocks on core inflation directly (by including relative food and energy price terms) as opposed to indirectly (through their effects on wage growth) estimate a declineintrendinflationoverthisperiodthatisroughlyhalfaslarge. 47However, one explanation that receives no support from the history of this period is Reis’s claim that policy remained loose because Martin “inferred that the [inflation] anchor was firmly in place” based on the low levels of long-termbondyieldsseenoverthe1965–1968period(Reis,2021,p.7). 12

recognizable from the vantage of 2022. Perhaps the most sobering fact, though, is how little practicalbenefitsixdecades’worthofadditionalexperiencehasprovidedus: Ourunderstanding of how the economy works—as well as our ability to predict the effects of shocks and policy actions—is in my view no better today than it was in the 1960s. Regardinginflationspecifically,wehavesomeempiricalbasisforclaimingthatduringthe25-year periodthatprecededtheCovidpandemicchangesinrealactivityhadrelativelymutedeffectson priceinflationandbothwageandpriceinflationtendedtoreverttolong-runmeansthatappeared largelyinvarianttomacroeconomicconditions. Moreover,thisstateofaffairsrepresentedaclear departure from the preceding three decades. But we have no deep understanding of why or how this change occurred, and so can say very little about what might cause it to change again or what sorts of actions might prevent such a change from occurring. In particular, we have no convincing evidence that this situation came to be because inflation expectations became well anchored or because the Fed won its credibility as an inflation fighter, which suggests that invoking a need to “keep expectations anchored” or to “restore inflation-fighting credibility” does not provide an especially compelling justification for a proposed policy action. Nor does the experience of the 1960s really help us to assess likely outcomes going forward. One conclusion of this paper is that the sharp rise in inflation that took place in the mid-1960s wasnottheresultofaregimechangethatwasinducedby“overheating”theeconomy: Asfaras areasonablysophisticatedstatisticalmodelcandiscern,thefeaturesoftheinflationprocessthat caused inflation to rise significantly and persistently in the face of sustained demand pressure werealreadyinplaceatthestartofthedecade. Bycontrast,thesefeatureswerenot presentat theonsetoftheCovidpandemic,andwehavenoevidencesofarthattheeconomicdislocations associated with the pandemic have caused them to reemerge. The present situation also differs importantly from the 1960s in that rather than resulting from successive large boosts to aggregate demand sustained over a number of years, much of the recentriseininflationinsteadreflectsasetofrelativepriceshocksthathaveoccurredasalarge and rapid realignment of demand across broad categories of consumption has run up against a temporary inability of producers and suppliers to fully meet that demand. If the inflation regime remains similar to what it was on the eve of the pandemic, this relative price shock—unlike the food and energy price shocks of the 1970s—should not be expected to yield a permanently higher inflation rate. But since we can make no credible assessment of the likelihood that a different outcome will result, or whether even a large reduction in aggregate demand would be sufficient to preclude it, the policy dilemma is especially acute at present. Hence, perhaps the most useful lesson from the 1960s inflation experience is how difficult it is to successfully conduct economic policy in the real world and in real time. Policymaking unfolds 13

on a “darkling plain,” and its practitioners—as well as those who seek to advocate an alternativecourse—willinvariablybeburdenedbyahighlyimperfectunderstandingofhowtheeconomy works;noisyandrevision-plagueddata;andoutcomesthatcannotevenbespecifiedinadvance, let alone be assigned a credible probability weight. Of course, policymakers face an additional burden that these others don’t: They are the ones responsible for making consequential decisions, and they are the ones held accountable for the results. 14

VII Appendix Additional details regarding the data and model used for this note are provided here. A Data documentation National Income and Product Accounts (NIPA) data are produced by the Bureau of Economic Analysis (BEA); data on unemployment, productivity, and compensation come from the Bureau of Labor Statistics; CBO’s natural rate and potential output series come from the July 2021 update of The Budget and Economic Outlook. (These data were downloaded from the Haver Analytics database on January 2, 2022.) Market-based PCE price index: Official data for the core market-based PCE price index are publishedfrom1987tothepresent. Priorto1987,themarket-basedseriesisconstructedbyusing detailed PCE data and a Fisher aggregation procedure routine that replicates the procedure followed by the BEA in constructing the NIPAs to strip out the prices of core nonmarket PCE components from the published overall core PCE price index, where the definition of “nonmarket” mimics the BEA’s. The core inflation series used in the statistical models also subtracts out Blinder and Rudd’s (2013) estimates of the effects of the Nixon-era price controls. Relative import price term: Import price inflation is defined as the annualized log difference of thepriceindexforimportsofnonpetroleumgoodsexcludingnaturalgas,computers,peripherals, and parts, which is computed using detailed NIPA series. (As the data required to construct this seriesonlyextendbackto1967:Q1,theannualizedlogdifferenceoftotalgoodsimportsisused prior to that date.) The relative import price inflation term used in the VAR model equals the differencebetweenthisseriesandcoremarket-basedpriceinflation(laggedoneperiod),weighted bythetwo-quartermovingaverageoftheshareofnominalimports(definedconsistentlywiththe import price measure) in nominal core PCE.48 Trend productivity growth: Trend productivity growth is defined as the low-frequency component of the annualized log difference of nonfarm business output per hour, which is obtained from a band-pass filter with the width and cutoffs set equal to the values used by Staiger, et al. (2001). An ARIMA(4,1,0) model is used to pad the actual productivity growth series prior to its 1947:Q2 starting point; after its 2020:Q4 endpoint, the series is padded with CBO’s forecast of average trendlaborproductivitygrowthfrom2021to2030(whichequals1.72inlogdifferences)andwith the 2031 value of the CBO forecast (which equals 1.64) thereafter. 48The relative import price term uses actual core market-based PCE prices (that is, unadjusted for the effect of pricecontrols),andthenominalimportshareisscaledbyitssamplemean. 15

Manufacturing quit rate: Series Ba4685, Historical Statistics of the United States, Millenial Edition. Cambridge, U.K.: Cambridge University Press (2006). Fiscal stance measures: Through 2019, the fiscal stance measure is based on an updated version of the fiscal impetus measure from Follette and Lutz (2011). Data and projections for 2020–2023 are derived from the Brookings Institution Hutchins Center fiscal impact measure (downloadedonFebruary6,2022). TomakethelattermeasuremorecomparabletotheFollette– Lutz measure, I add the CBO’s rate of Q4-over-Q4 potential output growth times the nominal share of government in GDP; these shares are set equal to 0.181 in 2020 and to 0.174 (the 2017–2019 average) in 2021–2023. B Additional estimation details The VAR system used in section III includes two lags of weighted relative core import prices, trend unit labor cost growth, core market-based PCE price inflation, and the unemployment gap (defined as the difference between the unemployment rate and the Congressional Budget Office’s estimate of the natural rate), with that ordering. Data are quarterly, and the sample period runs from 1959:Q1 to 2019:Q4. EstimationusesClarkandTerry’s(2010)implementationoftheMetropolis-within-Gibbsposterior sampler,whichinturnfollowsCogleyandSargent(2005).49 Thenumberofburn-indrawsequals 50,000; after that, 100,000 additional draws are run with every twentieth draw kept. The priors for the initial values are computed by estimating the VAR over a training sample that starts in 1949:Q2 (for this period, the core PCE price series is set equal to the price index for total PCE less food and energy goods). Following Clark and Terry (2010), the VAR uses an uninformative prior for the degree of time variation in the VAR coefficients (equal to 0.001 times the variancecovariance matrix of the VAR coefficients estimated over the training sample, with degrees of freedom set equal to the number of coefficients in the system plus one). 49IncontrasttoCogleyandSargent(2005),thesamplerdoesnottruncateexplosivedrawswithareflectingbarrier or“backstep”algorithm;inlinewiththerecommendationofKoopandPotter(2011),therefore,theimpulseresponse functions are median values. (The historical decomposition uses mean values in order to ensure that the sum of the baseline forecast and the contributions of all shocks will exactly equal the actual value of the variable being described.) 16

References Blinder,AlanS.(1982). “TheAnatomyofDouble-DigitInflationinthe1970s.” InInflation: Causes and Effects, edited by Robert E. Hall, pp. 261–282. Chicago: University of Chicago Press. Blinder,AlanS.,andStephenM.Goldfeld(1976). “NewMeasuresofFiscalandMonetaryPolicy, 1958–1973.” American Economic Review, 66, 780–796. Blinder, Alan S., and Jeremy B. Rudd (2013). “The Supply-Shock Explanation of the Great Stagflation Revisited.” In The Great Inflation: The Rebirth of Modern Central Banking, edited by Michael D. Bordo and Athanasios Orphanides, pp. 119–175. Chicago: University of Chicago Press. Blanchard, Olivier (2021). “In Defense of Concerns over the $1.9 Trillion Relief Plan.” PIIE Realtime Economic Issues Watch, February 18. Bremner,RobertP.(2004). ChairmanoftheFed: WilliamMcChesneyMartinJr. andtheCreation of the Modern American Financial System. New Haven, CT: Yale University Press. Cecchetti, Stephen G. (1987). “Indexation and Incomes Policy: A Study of Wage Adjustment in Unionized Manufacturing.” Journal of Labor Economics, 5, 391–412. Clark, Todd E., and Stephen J. Terry (2010). “Time Variation in the Inflation Passthrough of Energy Prices.” Journal of Money, Credit and Banking 42, 1419–1433. Cogley, Timothy, and Thomas J. Sargent (2005). “Drifts and Volatilities: Monetary Policies and Outcomes in the Post WWII US.” Review of Economic Dynamics, 8, 262–302. Cogley,Timothy,GiorgioE.Primiceri,andThomasJ.Sargent(2010). “Inflation-GapPersistence in the US.” American Economic Journal: Macroeconomics, 2, 43–69. Congressional Budget Office (1977). Incomes Policies in the United States: Historical Review and Some Issues. Washington, D.C.: U.S. Government Printing Office. Douty,H.M.(1975). Cost-of-LivingEscalatorClausesandInflation. ExecutiveOfficeofthePresident, Council on Wage and Price Stability Staff Report. Washington, D.C.: U.S. Government Printing Office. Follette,Glenn,andByronLutz(2011). “FiscalPolicyintheUnitedStates: AutomaticStabilizers, DiscretionaryFiscalPolicyActions,andtheEconomy.” InFiscalPolicy: LessonsfromtheCrisis, pp. 125–148. Perugia, Italy: Banca d’Italia. 17

Friedman, Milton (1968). “The Role of Monetary Policy.” American Economic Review, 58, 1–17. Halberstam, David (1993). The Best and the Brightest: Twentieth-Anniversary Edition. New York, NY: Ballantine Books. Koop, Gary, and Simon M. Potter (2011). “Time Varying VARs with Inequality Restrictions.” Journal of Economic Dynamics and Control, 35, 1126–1138. Mankiw, N. Gregory (2021). “The Biden Economy Risks a Speeding Ticket.” New York Times, February 26. Meltzer, Allan H. (2009). A History of the Federal Reserve: Volume 2, Book 1, 1951–1969. Chicago: University of Chicago Press. Mermelstein, David (1979). “The Threatening Economy.” New York Times, December 30. Murakami, Haruki (2006). “A Folklore for My Generation: A Pre-History of Late-Stage Capitalism.” In Blind Willow, Sleeping Woman, Philip Gabriel (translator), pp. 61–80. New York: Alfred A. Knopf. Nordhaus, William D. (1972). “The Recent Productivity Slowdown.” Brookings Papers on Economic Activity, 3, 493–536. Orphanides, Athanasios (2003a). “The Quest for Prosperity Without Inflation.” Journal of Monetary Economics, 50, 633–663. Orphanides, Athanasios (2003b). “Historical Monetary Policy Analysis and the Taylor Rule.” Journal of Monetary Economics, 50, 983–1022. Orphanides,Athanasios(2004). “MonetaryPolicyRules,MacroeconomicStability,andInflation: A View from the Trenches.” Journal of Money, Credit and Banking, 36, 151–175. Peneva, Ekaterina V., and Jeremy B. Rudd (2017). “The Passthrough of Labor Costs to Price Inflation.” Journal of Money, Credit and Banking, 49, 1777–1802. Perry, George L. (1971). “Labor Force Structure, Potential Output, and Productivity.” Brookings Papers on Economic Activity, 3, 533–565. Reis, Ricardo (2021). “Losing the Inflation Anchor.” Paper presented at the September 9, 2021 Brookings Papers on Economic Activity conference. 18

Rudd, Jeremy B. (2022). “Why Do We Think That Inflation Expectations Matter for Inflation? (And Should We?)” Review of Keynesian Economics, 10, 25–45. Shackleton, Robert (2018). “Estimating and Projecting Potential Output Using CBO’s Forecasting Growth Model.” Congressional Budget Office, Working Paper 2018-03 (February). Staiger, Douglas, James H. Stock, and Mark W. Watson (2001). “Prices, Wages, and the U.S. NAIRU in the 1990s.” In The Roaring Nineties: Can Full Employment Be Sustained?, edited by Alan Krueger and Robert Solow, pp. 3–60. New York: Russell Sage Foundation and Century Foundation Press. Summers, Lawrence H. (2021). “Opinion: My Column on the Stimulus Sparked a Lot of Questions. Here Are My Answers.” Washington Post, February 7. Summers, Lawrence H. (2022). “The Fed Must Do Much More to Fight Inflation—And Fast.” Time.com, March 17. Temin, Peter (1989). Lessons from the Great Depression. Cambridge, MA: MIT Press. 19

1. PCE price inflation, 1953−1971 Four-quarter log difference (x 100) 6 Total 5 Core Extended core 4 3 2 1 0 -1 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 20

2. Distribution of PCE price changes A. All components 12-month log difference (x 100) 10 Lower 25th 50th to 75th 8 Upper 25th 6 4 2 0 -2 -4 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 B. Core components only 12-month log difference (x 100) 10 Lower 25th 50th to 75th 8 Upper 25th 6 4 2 0 -2 -4 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 21

3. Trend unit labor cost growth, 1953−1971 Four-quarter log difference (x 100) 6 5 4 3 2 1 0 -1 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 4. One gauge of resource utilization Percent 8 Unemployment rate 7 CBO natural rate 6 5 4 3 2 1953 1955 1957 1959 1961 1963 1965 1967 1969 1971 22

5. VAR decomposition of core market-based inflation Four-quarter log difference (x 100) 6 Actual infla�on 5 Baseline forecast With effect of TULC shocks 4 3 2 1 0 1962 1964 1966 1968 1970 1972 23

6. Stochastic trend for price inflation Four-quarter log difference (x 100) 6 Core PCE infla�on 5 Long-run trend 4 3 2 1 0 -1 1961 1963 1965 1967 1969 1971 7. Stochastic trend for labor cost growth Four-quarter log difference (x 100) 6 Labor cost growth 5 Long-run trend 4 3 2 1 0 -1 1961 1963 1965 1967 1969 1971 24

8. Sum of inflation lag coefficients from inflation equation 1.0 0.8 0.6 0.4 0.2 0.0 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 2011 2016 Dashed lines denote 70 percent credible set. 9. Model-implied natural rate Percent 8 Unemployment rate 7 Model-implied natural rate 6 5 4 3 2 1961 1963 1965 1967 1969 1971 Dashed lines denote 70 percent credible set. 25

10. Response of core inflation to an unemployment gap shock Percentage point deviation from baseline 0.00 -0.05 -0.10 -0.15 -0.20 (−0.31) (−0.35) (−0.18) (−0.16) 1965 1969 1995 2015 -0.25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Quarter following shock Note: Eight-quarter integral multipliers in parentheses. 11. Response of TULC growth to an unemployment gap shock Percentage point deviation from baseline 0.00 -0.05 -0.10 -0.15 -0.20 (−0.43) (−0.44) (−0.47) (−0.40) 1965 1969 1995 2015 -0.25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Quarter following shock Note: Eight-quarter integral multipliers in parentheses. 26

12. Relative contribution of food and energy prices Percentage points 1.5 Consumer price index 1.0 PCE price index 0.5 0.0 -0.5 -1.0 1961 1963 1965 1967 1969 1971 27

13. A summary measure of fiscal policy in the 1960s Contribution to Q4/Q4 real GDP growth, pct. pts. 4 Fiscal impetus With mul�plier 3 2 1 0 -1 -2 1955 1957 1959 1961 1963 1965 1967 1969 1971 28

14. Fiscal policy and the output gap Percentage points (fiscal impetus) or percent (output gap) 6 4 2 0 -2 -4 -6 Fiscal impetus -8 Output gap (Q4 value, percent) -10 1955 1961 1967 1973 1979 1985 1991 1997 2003 2009 2015 2021 Notes: Shaded region denotes forecast. Output gap forecast uses July 2021 CBO potential output projection and December 2021 median FOMC SEP projection for real GDP. Fiscal impulse for 2020−2023 is derived from Brookings Hutchins Center fiscal impact measure (downloaded 2/6/2022), adjusted using CBO potential output growth. (See Appendix for additional details.) 29

15. Today this building is known as the Martin Building… 30