Seven Faces of “The Peril”

FEDERAL RESERVE BANK OF ST. LOUIS REVIEW SEPTEMBER/OCTOBER 2010 339

Seven Faces of “The Peril”

James Bullard

In this paper the author discusses the possibility that the U.S. economy may become enmeshedin a Japanese-style deflationary outcome within the next several years. To frame the discussion,the author relies on an analysis that emphasizes two possible long-run steady states for the econ-omy: one that is consistent with monetary policy as it has typically been implemented in theUnited States in recent years and one that is consistent with the low nominal interest rate, defla-tionary regime observed in Japan during the same period. The data considered seem to be quiteconsistent with the two steady-state possibilities. The author describes and critiques seven storiesthat are told in monetary policy circles regarding this analysis and emphasizes two main conclu-sions: (i) The Federal Open Market Committee’s “extended period” language may be increasingthe probability of a Japanese-style outcome for the United States and (ii), on balance, the U.S.quantitative easing program offers the best tool to avoid such an outcome. (JEL E4, E5)

Federal Reserve Bank of St. Louis Review, September/October 2010, 92(5), pp. 339-52.

The authors of the 2001 paper—Jess Benhabibat New York University and Stephanie Schmitt-Grohé and Martín Uribe both now at ColumbiaUniversity—studied abstract economies in whichthe monetary policymaker follows an activeTaylor-type monetary policy rule—that is, thepolicymaker changes nominal interest rates morethan one for one when inflation deviates from agiven target. Active Taylor-type rules are so com-monplace in present-day monetary policy discus-sions that they have ceased to be controversial.Benhabib, Schmitt-Grohé, and Uribe also empha-sized the zero bound on nominal interest rates.They suggested that the combination of an activeTaylor-type rule and a zero bound on nominalinterest rates necessarily creates a new long-runoutcome for the economy. This new long-runoutcome can involve deflation and a very low

THE PERIL

In 2001, three academic economists publisheda paper entitled “The Perils of Taylor Rules.”1

The paper has vexed policymakers and aca-demics alike, as it identified an important andvery practical problem—a peril—facing monetarypolicymakers, but provided little in the way ofsimple resolution. The analysis appears to applyequally well to a variety of macroeconomicframeworks, not just to those in one particularcamp or another, so that the peril result hasgreat generality. And, most worrisome, currentmonetary policies in the United States (andpossibly Europe as well) appear to be poised tohead straight toward the problematic outcomedescribed in the paper.

1 See Benhabib, Schmitt-Grohé, and Uribe (2001).

James Bullard is president and CEO of the Federal Reserve Bank of St. Louis. Any views expressed are his own and do not necessarily reflectthe views of other Federal Open Market Committee members. The author benefited from review and comments by Richard Anderson, DavidAndolfatto, Costas Azariadis, Jess Benhabib, Cletus Coughlin, George Evans, William Gavin, Seppo Honkapohja, Narayana Kocherlakota,Michael McCracken, Christopher Neely, Michael Owyang, Adrian Peralta-Alva, Robert Rasche, Daniel Thornton, and David Wheelock.Marcela M. Williams provided outstanding research assistance.

© 2010, The Federal Reserve Bank of St. Louis. The views expressed in this article are those of the author(s) and do not necessarily reflect theviews of the FOMC, the Federal Reserve System, the Board of Governors, or the regional Federal Reserve Banks. Articles may be reprinted,reproduced, published, distributed, displayed, and transmitted in their entirety if copyright notice, author name(s), and full citation are included.Abstracts, synopses, and other derivative works may be made only with prior written permission of the Federal Reserve Bank of St. Louis.

level of nominal interest rates. Worse, there ispresently an important economy that appears tobe stuck in exactly this situation: Japan.

To see what these authors were up to, considerFigure 1. This is a plot of nominal interest ratesand inflation for both the United States and Japanduring the period from January 2002 through May2010. The frequency is monthly. The Japanesedata are the circles in the figure, and the U.S. dataare the squares. The short-term nominal interestrate is on the vertical axis, and the inflation rate ison the horizontal axis. To maintain as much inter-national comparability as possible, all data arefrom the Organisation for Economic Co-operationand Development (OECD) main economic indica-tors (MEI). The short-term nominal interest rateis taken to be the policy rate in both countries—the overnight call rate in Japan and the federalfunds rate in the United States. Inflation in thefigure is the core consumer price index inflationrate measured from one year earlier in both coun-

tries. The data in the figure never mix duringthis time period: The U.S. data always lie to thenortheast, and the Japanese data always lie tothe southwest. This will be an essential mysteryof the story.

Benhabib, Schmitt-Grohé, and Uribe (2001)wrote about the two lines in the figure. The dashedline represents the famous Fisher relation for safeassets—the proposition that a nominal interestrate has a real component plus an expected infla-tion component. I have taken the real component(also the rate of time preference in the originalanalysis by Benhabib, Schmitt-Grohé, and Uribe)to be fixed and equal to 50 basis points in the fig-ure.2 Practically speaking, any macroeconomicmodel of monetary phenomena will have a Fisher

Bullard

340 SEPTEMBER/OCTOBER 2010 FEDERAL RESERVE BANK OF ST. LOUIS REVIEW

2 This is just for purposes of discussion—much of the formal analysisto which I refer later in the paper has stochastic features that wouldallow the real rate to fluctuate over time. Generally speaking,short-term, real rates of return on safe assets in the United Stateshave been very low during the postwar era.

0

1

2

3

4

5

–2.00 –1.50 –1.00 –0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00

Inflation (percent)

Nominal Interest Rate (percent)

Japan, Jan. 2002 to May 2010

U.S., Jan. 2002 to May 2010

Fisher Relation

Nonlinear Taylor-Type Rule

2003-2004

May 2010

(2.3, 2.8)

(–0.5, 0.001)

May 2010

Figure 1

Interest Rates and Inflation in Japan and the U.S.

NOTE: Short-term nominal interest rates and core inflation rates in Japan and the United States, 2002-10.

SOURCE: Data from the Organisation for Economic Co-operation and Development.

relation as a part of the analysis, and so this lineis hardly controversial. The solid line in the figurerepresents a Taylor-type policy rule: It describeshow the short-term nominal interest rate isadjusted by policymakers in response to currentinflation. In the right half of the figure, when infla-tion is above target, the policy rate is increased,but more than one for one with the deviation ofinflation from target. And when inflation is belowtarget, the policy rate is lowered, again more thanone for one. When the line describing the Taylor-type policy rule crosses the Fisher relation, we saythere is a steady state at which the policymakerno longer wishes to raise or lower the policy rate,and, simultaneously, the private sector expectsthe current rate of inflation to prevail in the future.It is an equilibrium in the sense that, if there areno further shocks to the economy, nothing willchange with respect to inflation or the nominalinterest rate. In the figure, this occurs at an infla-tion rate of 2.3 percent and a nominal interestrate of 2.8 percent (denoted by an arrow on theright side of the figure). This is sometimes calledthe “targeted” steady state.3

The “active” policy rule—the fact that nomi-nal interest rates move more than one for one withinflation deviations in the right half of the figure—is supposed to keep inflation near the target. Italso means that the line describing the Taylor-typepolicy rule is steeper than the line describing theFisher relation in the neighborhood of the targetedinflation rate. It cuts the Fisher relation frombelow. Taken at face value, the Taylor-type policyrule has been fairly successful for the UnitedStates: Inflation (by this measure) has been neitherabove 3 percent nor, until very recently, below 1percent during the January 2002–May 2010 period.

None of this so far is really the story told byBenhabib, Schmitt-Grohé, and Uribe. On theright-hand side of the figure, short-term nominalinterest rates are adjusted up and down to keepinflation low and stable. It’s all very conventional.The point of the analysis by Benhabib, Schmitt-Grohé, and Uribe is to think more carefully aboutwhat these seemingly innocuous assumptions—

the Fisher relation, the active Taylor-type rule,the zero bound on nominal interest rates—reallyimply as we move to the left in the figure, faraway from the targeted steady-state equilibrium.And, what these building blocks imply is onlyone thing: The two lines cross again, creating asecond steady state. In the figure, this secondsteady state occurs at an inflation rate of –50 basispoints and an extremely low short-term nominalinterest rate of about one-tenth of a basis point(see the arrow on the left side of the figure).4 TheJapanese inflation data are all within about 100basis points of this steady state, between –150basis points and 50 basis points. That’s about thesame distance from low to high as the U.S. infla-tion data. But for the nominal interest rate, mostof the Japanese observations are clustered between0 and 50 basis points. The policy rate cannot belowered below zero, and there is no reason toincrease the policy rate since—well, inflation isalready “too low.” This logic seems to have keptJapan locked into the low nominal interest ratesteady state. Benhabib, Schmitt-Grohé, and Uribesometimes call this the “unintended” steadystate.5

At the unintended steady state, policy is nolonger active: It has instead switched to beingpassive. The policy line crosses the Fisher relationfrom above. When inflation decreases, the policyrate is not lowered more than one for one becauseof the zero lower bound. And when inflationincreases, the policy rate is not increased morethan one for one because, in this region of thediagram, inflation is well below target. Fluctua -tions in inflation are in fact not met with muchof a policy response at all in the neighborhood ofthe unintended steady state. At this steady state,the private sector has come to expect the rate ofdeflation consistent with the Fisher relationaccompanied by very little policy response; thus,

Bullard


3 Steady states are considered focal points for the economy inmacroeconomic theories—the economy “orbits” about the steadystate in response to shocks.

4 This example is meant as an illustration only. The formula I usedto plot the nonlinear Taylor-type policy rule is R = AeBπ, where Ris the nominal interest rate, π is the inflation rate, and A and B areparameters. I set A = 0.005015 and B = 2.75. Taylor-type policy rulesalso have an output gap component, and in the literature that issueis discussed extensively. For the possibility of a second steady state,it is the inflation component that is of paramount importance.

5 I discuss the social desirability of each of the two steady statesbriefly in the section titled “Traditional Policy.”

nothing changes with respect to nominal interestrates or inflation. Where does policy transitionfrom being active to being passive? This occurswhen the slope of the nonlinear Taylor-type ruleis exactly 1, which is at an inflation rate of about1.56 percent in Figure 1.

Again, the data in this figure do not mix atall—it’s boxes on the right and circles on the left.But the most recent observation for the UnitedStates, the solid box labeled “May 2010,” is aboutas close as the United States has been in recenttimes to the low nominal interest rate steady state.It is below the rate at which policy turns passivein the diagram. In addition, the Federal OpenMarket Committee (FOMC) has pledged to keepthe policy rate low for an “extended period.”This pledge is meant to push inflation back towardtarget—certainly higher than where it is today—thus moving to the right in the figure. Still, as thefigure makes clear, pledging to keep the policyrate near zero for such a long time would also beconsistent with the low nominal interest ratesteady state, in which inflation does not returnto target but instead both actual and expectedinflation turn negative and remain there. Further -more, we have an example of an important econ-omy that appears to be in just this situation.

A key problem in the figure is that the mone-tary policymaker uses only nominal interest rateadjustment to implement policy. This is themeaning of the nonlinear Taylor-type policy rulecontinuing far to the left in the diagram. The pol-icymaker is completely committed to interest rateadjustment as the main tool of monetary policy,even long after it ceases to make sense (long afterpolicy becomes passive), creating a second steadystate for the economy. Many of the responsesdescribed below attempt to remedy the situationby recommending a switch to some other policywhen inflation is far below target. The regimeswitch required must be sharp and credible—policymakers have to commit to the new policyand the private sector has to believe the policy-makers. Unfortunately, in actual policy discus-sions nothing of this sort seems to be happening.Both policymakers and private sector players con-tinue to communicate in terms of interest rateadjustment as the main tool for the implementa-

tion of monetary policy. This is increasing therisk of a Japanese-style outcome for the UnitedStates.

My view is that the 2001 analysis by Benhabib,Schmitt-Grohé, and Uribe is an important one forcurrent policy, that it has garnered insufficientattention in the policy debate, and that it is indeedclosely related to the current “extended period”pledge of the FOMC. Below I relate and critiqueseven stories—both formal ones and informalones—that I have encountered concerning thisanalysis. The fact that there are seven “faces”shows just how fragmented the economics pro-fession is on this critical issue. These stories rangefrom reasons not to worry about the implicationsof Figure 1, through ways to adjust nominalinterest rates to avoid the implications of Figure 1,and on to the uses of unconventional policies asa tool to avoid “the peril.”

I conclude that promises to keep the policyrate near zero may be increasing the risk of fallinginto the unintended steady state of Figure 1 andthat an appropriate quantitative easing policyoffers the best hope for avoiding such an outcome.

SEVEN FACESDenial

I think it is fair to say that, for many whohave been involved in central banking over thepast two or three decades, it is difficult to thinkof Japan and the United States in the same game,as Figure 1 suggests. For many, the situation inJapan since the 1990s has been a curiosum, an oddoutcome that might be chalked up to particularlyByzantine Japanese politics, the lack of an infla-tion target for the Bank of Japan (BOJ), a certainlack of political independence for the BOJ, or someother factor specific to the Land of the Rising Sun.The idea that U.S. policymakers should worryabout the nonlinearity of the Taylor-type ruleand its implications is sometimes viewed as anamusing bit of theory without real ramifications.Linear models tell you everything you need toknow. And so, from this denial point of view, wecan stick with our linear models and ignore thedata from Japan (Figure 2).

Bullard


In Figure 2, the targeted steady state remainsat an inflation rate of 2.3 percent, but the Taylor-type rule is now linear. The policy rate still reactsto the current level of inflation, and more thanone for one; that is, the Taylor-type rule is stillactive. In fact, in the neighborhood of the targetedsteady state, there would be very little differencein choosing the linear or the nonlinear versionsof the Taylor-type rule. For lower values of infla-tion, however, the linear Taylor-type rule nowextends into negative territory, violating the zerobound on nominal interest rates. Some contem-porary discussion of monetary policy pines for anegative policy rate exactly as pictured here. Oneoften hears that, given the state of today’s econ-omy, the desired policy rate would be, say, –6.0percent, as suggested by the chart. This is non-sensical, since under current operating proceduressuch a policy rate is infeasible and therefore wecannot know how the economy would behavewith such a policy rate.6

The most disturbing part of Figure 2, how-ever, is that the Japanese data are not part of the

picture. This tempts one to argue that, becausecore inflation is currently below target, there islittle harm in keeping the policy rate near zeroand, indeed, in promising to keep the policy ratenear zero in the future. There is no danger asso-ciated with such a policy according to Figure 2.There is a sort of faith that the economy will nat-urally return to the targeted steady state, sincethat is the only long-run equilibrium outcomefor the economy that is part of the analysis.

Stability

There is another version of the denial viewthat is somewhat less extreme but neverthelessstill a form of denial in the end. It is a view thatI have been associated with in my own research.

Bullard


6 Over the years, some discussion in monetary theory has contem-plated currency taxes as a means of obtaining negative nominalrates, but that is a radical proposal not often part of the negativerates discussion. See Mankiw (2009). Interestingly, even negativerates would not avoid the multiple-equilibria problem—see, forinstance, Schmitt-Grohé and Uribe (2009).

–3

–2

–1

0

1

2

3

4

5

6

0.50 1.00 1.50 2.00 2.50 3.00

2003-2004

May 2010

(2.3, 2.8)

Inflation (percent)

U.S., Jan. 2002 to May 2010

Fisher Relation

Linear Taylor-Type Rule


Figure 2

Denial

In this view, one accepts the zero bound on nom-inal interest rates and the other details of theanalysis by Benhabib, Schmitt-Grohé, and Uribe.One accepts that there are two steady states. How -ever, the steady states have stability propertiesassociated with them in a fully dynamic analysis,and the argument is that the targeted steady stateis the stable one, while the unintended, low nomi-nal interest rate steady state is unstable. There -fore, according to this argument, one shouldexpect to observe the economy in the neighbor-hood of the targeted steady state and need notworry about the unintended, low nominal interestrate steady state.

The original analysis by Benhabib, Schmitt-Grohé, and Uribe entailed much more than whatI have described in Figure 1. The figure outlinesjust the big picture. In fact, the authors wrotedown complete DSGE economies7 and analyzedthe dynamics of those systems in a series ofpapers. In the original analysis, the 2001 paper,they endowed both the central bank and the pri-vate sector in the model with rational expecta-tions. They then showed that it was possible forthe economy to begin in the neighborhood of thetargeted steady state and follow an equilibriumpath to the unintended, low nominal interest ratesteady state. These dynamics in fact spiraled outfrom the targeted steady state.

I did not find this story very compelling, fortwo reasons: because the dynamics describedseem unrealistic—they imply a volatile sequenceof interest rates and inflation rates followed bysudden arrival at the low nominal interest ratesteady state—and because they rely heavily onthe foresight of the players in the economy con-cerning this volatile sequence.

A 2007 paper by Stefano Eusepi, now at theFederal Reserve Bank of New York, addressedsome of these concerns. Eusepi accepted the non-linear nature of Figure 1 with its two steadystates. He also backed off the rational expecta-tions assumption that characterized the originalanalysis by Benhabib, Schmitt-Grohé, and Uribe.Instead he assumed that the actors in the model

might learn over time in a specific way by consid-ering the data produced by the economy itself.8

One key result in the Eusepi paper (2007)was the following: If the monetary authority, withits nonlinear Taylor-type policy rule, reacts toinflation one period in the past (as perhaps onemight expect of many central banks), then theonly possible long-run outcome for the economyis the targeted steady state. I found this comfort-ing. It suggests that one need not worry about theunintended steady state and that exclusive focuson the targeted steady state is warranted. To besure, a careful reading of the Eusepi paper revealsthat many other dynamic paths are also possible,including some that converge to the unintendedsteady state. Still, one might hope that the targetedsteady state is somehow the stable one—and thatfor this reason one can sleep better at night.

I’ve said this is a form of denial. First, as fasci-nating as they are, the results are not that clean,as many dynamics are possible depending onthe details of the model. It is hard to know howthese details truly map into actual economies.But more importantly, Figure 1 suggests that atleast one large economy has in fact converged tothe unintended steady state. The stability argu-ment cannot cope with this datum, unless one iswilling to say that conditions are subtly differentin Japan compared with the United States, pro-ducing convergence to the unintended steadystate in Japan but convergence to the targetedsteady state in the United States. I have not seena compelling version of this argument. I concludethat the stability argument is actually a form ofdenial in the end.9

The FOMC in 2003

In Figure 1, a set of data points is circled.These data are labeled “2003-2004” and are asso-ciated with a policy rate at 1.0 percent and theinflation rate between 1.0 and 1.5 percent. This

Bullard


7 That is, dynamic, stochastic, general equilibrium economies.

8 See Evans and Honkapohja (2001).

9 For an argument that, under learning, the targeted steady state islocally but not globally stable, see Evans, Guse, and Honkapohja(2008). In their paper, the downside risk is much more severe, asunder learning the economy can fall into a deflationary spiral inwhich output contracts sharply.

episode was the last time the FOMC worried abouta possible bout of deflation. While core inflationdid move to a low level during this period—notquite as low as the current level—inflation movedhigher later and interest rates were increased.This episode surely provides comfort for thosewho think the Japanese-style outcome is unlikely.It suggests that the economy will ultimately returnto the neighborhood of the targeted steady state,perhaps even indicating that the stability story isthe right one after all. The 2003 experience didnot involve a near-zero policy rate, however.

One description of this period is due toDaniel Thornton, an economist at the FederalReserve Bank of St. Louis.10 The Thornton analy-sis emphasizes (i) how the FOMC communicatedduring this period and (ii) how the market expec-tations of the longer-term inflation rate respondedto the communications. At the time, some meas-

ures of inflation were hovering close to 1 percent,similar to the most recent readings for coreinflation in 2010. At its May 2003 meeting, theCommittee included the following press releaselanguage: “[T]he probability of an unwelcomesubstantial fall in inflation, though minor, exceedsthat of a pickup in inflation from its already lowlevel.” At several subsequent 2003 meetings, theFOMC stated that “the risk of inflation becomingundesirably low is likely to be the predominantconcern for the foreseeable future.” By the begin-ning of 2004, inflation had picked up and FOMCreferences to undesirably low inflation ceased.Thornton shows that before any of these state-ments were made the longer-run expected infla-tion rate, as measured from the 10-year Treasuryinflation-indexed security spread, was 1.74 per-cent during the period from January 2001 throughApril 2003. After the statements, from January2004 to May 2006, the longer-run inflation expec-tation averaged 2.5 percent. Thornton interprets

Bullard


10 See Thornton (2006, 2007).

0

1

2

3

4

5

–2.00 –1.50 –1.00 –0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00

2003-2004

May 2010

(1.75, 2.25)

(–0.5, 0.001)

(2.5, 3.0)

Inflation (percent)



U.S., Jan. 2002 to May 2010

Fisher Relation

Nonlinear Taylor-Type Rule (π* = 1.75)

Nonlinear Taylor-Type Rule (π* = 2.5)

May 2010

Figure 3

Inflation Expectations, Interrupted

NOTE: The 2003-04 episode. Thornton (2006, 2007) argues that FOMC communications increased the perceived inflation target ofthe Committee.

the FOMC language as putting a lower bound onthe Committee’s implicit inflation target range.This had the effect of increasing the longer-runexpected rate of inflation.

Figure 3 shows how such a change in longer-run inflation expectations might play out. Accept -ing the other premises of the analysis by Benhabib,Schmitt-Grohé, and Uribe, the private sector nowviews the central bank as taking action to containinflation only once inflation has attained a some-what higher level. Those in the private sectorthought they understood policy as the solid blackline, but after the FOMC communication, theyunderstood policy as the dashed blue line. Thisalters the targeted steady-state inflation rate ofthe economy from 1.75 percent (the second arrowfrom the right in Figure 3) to 2.5 percent (right-most arrow).

It is not immediately obvious from Figure 3why this should have a desirable impact onwhether the economy ultimately returns to theneighborhood of the targeted steady state or con-verges to the unintended, low nominal interestrate steady state. Credibly raising the inflationtarget is actually moving the target steady-stateequilibrium to the right in the diagram, fartheraway from the circled data from 2003 and 2004.One might think that creating more distance fromthe current position to the desired outcome wouldnot be helpful.

In the event, all worked out well, at leastwith respect to avoiding the unintended steadystate.11 Inflation did pick up, the policy rate wasincreased, and the threat of a Japanese-styledeflationary outcome was forgotten, at least tem-porarily. Was this a brilliant maneuver, or did theeconomic news simply support higher inflationexpectations during this period?

Discontinuity

If the problem is the existence of a second,unintended steady state—and this is partly causedby the choice of a policy rule that is controlled by

policymakers—why not just choose a differentpolicy rule? This can, in fact, be done and wasdiscussed by Benhabib, Schmitt-Grohé, and Uribein their original paper. Furthermore, some partsof the current policy discussion have exactly thisflavor.

The problem illustrated in Figure 1 is pre-cisely that the two lines, one describing policyand one describing private sector behavior, crossin two places. But the policy line can be alteredby policymakers. A simple version is illustratedin Figure 4. Here, the nonlinear Taylor-type policyrule is followed so long as inflation remains above50 basis points. For inflation lower than that level,the policy rate is simply set to 1.5 percent andleft there. This creates the black bar in Figure 4between an inflation rate of –1.0 percent (or lower)and 0.5 percent. The policy would be that, forvery low levels of inflation, the policy rate is setsomewhat higher than zero, but still at a veryaccommodative level. After all, short-term nomi-nal interest rates at 1.5 percent would still beconsidered aggressively easy policy in nearly allcircumstances.

Of course, this policy looks unusual and per-haps few would advocate it, but again we are try-ing to avoid all those circles down there in thesouthwest portion of the diagram. The discon-tinuous policy has the great advantage that it is avery simple way to ensure that the unintended,low nominal interest rate steady state no longerexists. The only point in the diagram where theFisher relation and the policy rule can be in har-mony is the targeted equilibrium. This wouldremove the unintended steady state as a focalpoint for the economy.12,13

Some of the current policy discussion hasincluded an approach of this type, although notexactly in this context. The FOMC’s near-zero

Bullard


11 Many have criticized the FOMC for allowing the target rate toremain too low for too long during this period. For a discussion,see remarks by Fed Chairman Bernanke (2010) delivered at theannual meeting of the American Economic Association.

12 The academic literature regarding the use of fiscal policy measures,as described below, has the same goal—unintended outcomes areeliminated as equilibria. But the fiscal policy route is far moreconvoluted.

13 Two astute reviewers—Costas Azariadis and Jess Benhabib—bothstressed that a discontinuous policy could create a pseudo steadystate (at the point of discontinuity) and that the economy mightthen oscillate about the pseudo steady state instead of convergingto the targeted outcome. This has not been a subject of research inthis context as far as I know.

interest rate policy and the associated “extendedperiod” language have caused many to worry thatthe Committee is fostering the creation of new,bubble-like phenomena in the economy that willeventually prove counterproductive. One antidoteto this worry may be to increase the policy ratesomewhat, while still keeping the rate at a histori-cally low level, and then to pause at that level.14

That policy would have a similar flavor to the onesuggested in Figure 4, although for a differentpurpose.

Traditional Policy

According to the Bank of England,15 for 314years the policy rate was never allowed to fall

below 2.0 percent. During more than three cen-turies the economy was subject to large shocks,wars, financial crises, and the Great Depression—yet 2.0 percent was the policy rate floor until veryrecently. A version of this policy is displayed inFigure 5. This policy rule does not eliminate theunintended steady state; it simply moves it to beassociated with a higher level of inflation. In thefigure, this point occurs at an interest rate of 2.0percent and an inflation rate of 1.5 percent (thecenter arrow in the figure). This policy seems veryreasonable in some ways. To the extent that oneof the main purposes of the interest rate policy isto keep inflation low and stable, this policy cre-ates two steady states, but the policymaker maybe more or less indifferent between the two out-comes. Then one has to worry much less aboutthe possibility of becoming permanently trappedin an unintended, deflationary steady state. Thispolicy prevents the onset of interest rates thatare “too low.”

Bullard


14 My colleague Thomas Hoenig (2010), president of the FederalReserve Bank of Kansas City, has advocated such a policy. See hisspeech, “The High Cost of Exceptionally Low Rates.”

15 For historical data since 1694 on the official Bank of England policyrate, see www.bankofengland.co.uk/statistics/rates/baserate.xls.

(2.3, 2.8)

1

2

3

4

5

–2.00 –1.50 –1.00 –0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00

Inflation (percent)



U.S., Jan. 2002 to May 2010

Fisher Relation


0

2003-2004

May 2010May 2010

Figure 4

Discontinuity

NOTE: The discontinuous Taylor-type policy rule looks unusual but eliminates the unintended steady state.

The idea that policymakers might be more orless indifferent between the two steady statesbrings up an important question about the originalanalysis by Benhabib, Schmitt-Grohé, and Uribe.Why should one steady-state equilibrium be pre-ferred over the other? This question has someacademic standing—there is a long literature onthe optimal long-run rate of inflation, and loweris usually better. In the conventional policy dis-cussion, however, the targeted steady state isdefinitely preferred. Perhaps the most importantconsideration is that, in the unintended steadystate, the policymaker loses all ability to respondto incoming shocks by adjusting interest rates—ordinary stabilization policy is lost, possibly forquite a long time. In addition, the conventionalwisdom is that Japan has suffered through a “lostdecade” partially attributable to the fact that theeconomy has been stuck in the deflationary, lownominal interest rate steady state illustrated inFigure 1. To the extent that is true, the UnitedStates and Europe can hardly afford to join Japanin the quagmire. Most of the arguments I know of

concerning the low nominal interest rate steadystate center on the idea that deflation, even milddeflation, is undesirable. It is widely perceivedthat problems in the U.S. financial system are atthe core of the current crisis. Given that manyfinancial contracts (and, in particular, mortgages)are stated in nominal terms and given that thesecontracts were written in the past (under theexpectation of a stable inflation around 2.0 per-cent), it is conceivable to think that deflationcould hurt the financial system and hamper U.S.growth.16

If we suppose that deflation is the main prob-lem, then this could likely be avoided by simplynot adopting a rule that calls for very low—near-zero—interest rates. Instead, the rule could call

Bullard


16 Some have argued that ongoing deflation in Japan is not an impor-tant contributory factor for the nation’s relatively slow growth.See, for instance, Hayashi and Prescott (2002). In addition, theUnited States grew rapidly in the late nineteenth century despitean ongoing deflation. So, the relationship between deflation andlonger-run growth is not as obvious as some make it seem. Still,the conventional wisdom is that a turn toward deflation wouldhamper U.S. growth.

2003-2004

May 2010

(2.3, 2.8)

(1.5, 2.0)

1

2

3

4

5

–2.00 –1.50 –1.00 –0.50 0.00 0.50 1.00 1.50 2.00 2.50 3.00

Inflation (percent)



U.S., Jan. 2002 to May 2010

Fisher Relation


0

May 2010

Figure 5

Traditional Policy

for rates to bottom out at a level somewhat higherthan zero, as the traditional policy rule does. Ofcourse, a policy rule like the one depicted inFigure 5 does not allow as much policy accom-modation in the face of shocks to the economy atthe margin. But is it worth risking a “lost decade”to get the extra bit of accommodation?

Fiscal Intervention Given the Situationin Europe

In the academic literature following the 2001publication of the perils paper, some attempt wasmade to provide policy advice on how to avoidthe unintended steady state of Figure 1.17 Thisadvice was given in the context of trying to pre-serve the desirable qualities of the Taylor-typeinterest rate rule in the neighborhood of the tar-geted steady state. That is, even though interestrate rules are the problem here, the advice is givenin the context of those rules—as opposed to sim-ply abandoning them altogether.

The advice has a certain structure. It involvesnot changes in the way monetary policy is imple-mented, but changes in the fiscal stance of thegovernment. By itself, this makes the practicalityof the solution much more questionable. But itgets worse. The proposal is for the government toembark on an aggressive fiscal expansion shouldthe economy become enmeshed in a low nominalinterest rate equilibrium. The fiscal expansionhas the property that total government liabilities—money plus government debt—grow at a suffi-ciently fast rate. Inside the model, such a fiscalexpansion eliminates the unintended steady stateas an equilibrium outcome. By this roundaboutmethod, then, the only remaining longer-run out-come for the economy is to remain in the neigh-borhood of the targeted steady state.

The described solution has the following fla-vor: The government threatens to behave unrea-sonably if the private sector holds expectations(such as expectations of very low inflation) thatthe government does not desire. This threat, if itis credible, eliminates the undesirable equilib-rium. Some authors have criticized this type of

solution to problems with multiple equilibria as“unsophisticated implementation.”18

Today, especially considering the ongoingEuropean sovereign debt crisis, these proposedsolutions strike me as wildly at odds with therealities of the global economy. The proposalmight work in a model setting, but the practicali-ties of getting a government to essentially threateninsolvency—and be believed—seem to rely fartoo heavily on the rational expectations of theprivate sector.19 Furthermore, governments thatattempt such a policy in reality are surely playingwith fire. The history of economic performancefor nations actually teetering on the brink of insol-vency is terrible. This does not seem like a goodtool to use to combat the possibility of a low nomi-nal interest rate steady state.

Beyond these considerations, it is question-able at this point whether such a policy actuallyworks. Japan, our leading example in this story,has in fact embarked on an aggressive fiscalexpansion, and the debt-to-GDP ratio there isnow approaching 200 percent. Still, there doesnot appear to be any sign that their economy isabout to leave the low nominal interest rate steadystate, and now policymakers are worried enoughabout the international reaction to their situationthat fiscal retrenchment is being seriously debated.

Quantitative Easing

The quantitative easing policy undertakenby the FOMC in 2009 has generally been regardedas successful in the sense that longer-term interestrates fell following the announcement and imple-mentation of the program.20 Similar assessmentsapply to the Bank of England’s quantitative easingpolicy. For the United Kingdom in particular,both expected inflation and actual inflation haveremained higher to date, and for that reason theUnited Kingdom seems less threatened by adeflationary trap. The U.K. quantitative easingprogram has a more state-contingent character

Bullard


18 See Atkeson, Chari, and Kehoe (2010).

19 For a version that backs off the rational expectations assumption,but still eliminates the undesirable equilibrium, see Evans, Guse,and Honkapohja (2008).

20 See, for instance, Neely (2010).

17 See, in particular, Benhabib, Schmitt-Grohé, and Uribe (2002),Woodford (2001, 2003), and Eggertsson and Woodford (2003).

than the U.S. program. The U.S. approach was tosimply announce a large amount of purchasesbut not adjust the amounts or pace of purchasesaccording to changing assessments of macroeco-nomic prospects.

The quantitative easing program, to the extentit involves buying longer-dated government debt,has often been described as “monetizing the debt.”This is widely considered to be inflationary, andso inflation expectations are sensitive to suchpurchases. In the United Kingdom, all the pur-chases were of gilts (Treasury debt). In the UnitedStates, most of the purchases were of agency—Fannie Mae and Freddie Mac—mortgage-backedsecurities, newly issued in 2009. It has beenharder to judge the inflationary effects of thesepurchases, and so perhaps the effects on inflationexpectations and hence actual inflation have beensomewhat less reliable in the United States thanin the United Kingdom.

The experience in the United Kingdom seemsto suggest that appropriately state-contingentpurchases of Treasury securities are a good toolto use when inflation and inflation expectationsare “too low.” Not that one would want to overdoit, mind you, as such measures should only beundertaken in an effort to move inflation closerto target. One very important consideration is theextent to which such purchases are perceived bythe private sector as temporary or permanent.We can double the monetary base one day andreturn to the previous level the next day, andwe should not expect such movements to haveimportant implications for the price level in theeconomy. Base money can be removed from thebanking system as easily as it can be added, soprivate sector expectations may remain unmovedby even large additions of base money to thebanking system.21 In the Japanese quantitativeeasing program, beginning in 2001, the BOJ wasunable to gain credibility for the idea that theywere prepared to leave the balance sheet expan-sion in place until policy objectives were met.And in the end, the BOJ in fact did withdraw the

program without having successfully pushedinflation and inflation expectations higher, vali-dating the private sector expectation. The UnitedStates and the United Kingdom have enjoyedmore success, perhaps because private sectoractors are more enamored with the idea that theFOMC and the U.K.’s Monetary Policy Committeewill do “whatever it takes” to avoid particularlyunpleasant outcomes for the economy.

CONCLUSIONThe global economy continues to recover

from the very sharp recession of 2008 and 2009.During this recovery, the U.S. economy is suscep-tible to negative shocks that may dampen infla-tion expectations. This could push the economyinto an unintended, low nominal interest ratesteady state. Escape from such an outcome is prob-lematic. Of course, we can hope that we do notencounter such shocks and that further recoveryturns out to be robust—but hope is not a strategy.The United States is closer to a Japanese-styleoutcome today than at any time in recent history.

In part, this uncomfortably close circumstanceis due to the interest rate policy now being pur-sued by the FOMC. That policy is to keep thecurrent policy rate close to zero, but in additionto promise to maintain the near-zero interest ratepolicy for an “extended period.” But it is evenmore than that: The reaction to a negative shockin the current environment is to extend theextended period even further, delaying the dayof normalization of the policy rate farther into thefuture. This certainly seems to be the implicationfrom recent events. When the European sovereigndebt crisis rattled global financial markets duringthe spring of 2010, it was a negative shock to theglobal economy and the private sector perceptionwas certainly that this would delay the date ofU.S. policy rate normalization. One might thinkthat is a more inflationary policy, but TIPS-basedmeasures of inflation expectations over 5 and 10years fell about 50 basis points.

Promising to remain at zero for a long time isa double-edged sword. The policy is consistentwith the idea that inflation and inflation expec-

21 For discussions of how forms of quantitative easing can helpachieve the intended steady state, in combination with appropriatefiscal policy, see Eggertsson and Woodford (2003, pp. 194-98) andEvans and Honkapohja (2005).

Bullard


tations should rise in response to the promiseand that this will eventually lead the economyback toward the targeted equilibrium of Figure 1.But the policy is also consistent with the idea thatinflation and inflation expectations will insteadfall and that the economy will settle in the neigh-borhood of the unintended steady state, as Japanhas in recent years.22

To avoid this outcome for the United States,policymakers can react differently to negativeshocks going forward. Under current policy inthe United States, the reaction to a negative shockis perceived to be a promise to stay low for longer,which may be counterproductive because it mayencourage a permanent, low nominal interest rateoutcome. A better policy response to a negativeshock is to expand the quantitative easing programthrough the purchase of Treasury securities.

Bullard


22 Evans and Honkapohja (2010) have made a version of this argumentmore formally.

REFERENCESAtkeson, Andrew; Chari, Varadarajan and Kehoe, Patrick. “Sophisticated Monetary Policies.” Quarterly Journalof Economics, February 2010, 125(1), pp. 47-89.

Benhabib, Jess; Schmitt-Grohé, Stephanie and Uribe, Martín. “The Perils of Taylor Rules.” Journal of EconomicTheory, January 2001, 96(1-2), pp. 40-69.

Benhabib, Jess; Schmitt-Grohé, Stephanie and Uribe, Martín. “Avoiding Liquidity Traps.” Journal of PoliticalEconomy, June 2002, 110(3), pp. 535-63.

Bernanke, Ben S. “Monetary Policy and the Housing Bubble.” Remarks before the Annual Meeting of theAmerican Economic Association, Atlanta, GA, January 3, 2010;www.federalreserve.gov/newsevents/speech/bernanke20100103a.htm.

Eggertsson, Gauti and Woodford, Michael. “The Zero Bound on Interest Rates and Optimal Monetary Policy.”Brookings Papers on Economic Activity, 2003, 1, pp. 139-211.

Eusepi, Stefano. “Learnability and Monetary Policy: A Global Perspective.” Journal of Monetary Economics,May 2007, 54(4), pp. 1115-131.

Evans, George; Guse, Eran and Honkapohja, Seppo. “Liquidity Traps, Learning, and Stagnation.” EuropeanEconomic Review, November 2008, 52(8), pp. 1438-463.

Evans, George and Honkapohja, Seppo. Learning and Expectations in Macroeconomics. Princeton, NJ: PrincetonUniversity Press, 2001.

Evans, George and Honkapohja, Seppo. “Policy Interaction, Expectations and the Liquidity Trap.” Review ofEconomic Dynamics, April 2005, 8(2), pp. 303-23.

Evans, George and Honkapohja, Seppo. “Expectations, Deflation Traps and Macroeconomic Policy,” in DavidCobham, Øyvind Eitrheim, Stefan Gerlach, and Jan F. Qvigstad, eds., Twenty Years of Inflation Targeting:Lessons Learned and Future Prospects. Chap. 11. Cambridge, UK: Cambridge University Press, 2010, pp. 232-60.

Hayashi, Fumio and Prescott, Edward C. “The 1990s in Japan: A Lost Decade.” Review of Economic Dynamics,January 2002, 5(1): 206-35.

Hoenig, Thomas M. “The High Cost of Exceptionally Low Rates.” Remarks before the Bartlesville Federal ReserveForum, Bartlesville, OK, June 3, 2010; www.kansascityfed.org/speechbio/Hoenigpdf/Bartlesville.06.03.10.pdf.

Mankiw, N. Gregory. “It May Be Time for the Fed to Go Negative.” New York Times, April 18, 2009;www.nytimes.com/2009/04/19/business/economy/19view.html.

Neely, Christopher J. “The Large Scale Asset Purchase Purchases Had Large International Effects.” WorkingPaper No. 2010-18A, Federal Reserve Bank of St. Louis, July 2010;http://research.stlouisfed.org/wp/2010/2010-018.pdf.

Schmitt-Grohé, Stephanie and Uribe, Martín. “Liquidity Traps with Global Taylor Rules.” International Journalof Economic Theory, 2009, 5(1), pp. 85-106.

Thornton, Daniel L. “The Fed’s Inflation Objective.” Federal Reserve Bank of St. Louis Monetary Trends, July2006; http://research.stlouisfed.org/publications/mt/20060701/cover.pdf.

Thornton, Daniel L. “The Lower and Upper Bounds of the Federal Open Market Committee’s Long-Run InflationObjective.” Federal Reserve Bank of St. Louis Review May/June 2007, 89(3), pp. 183-93; http://research.stlouisfed.org/publications/review/07/05/Thornton.pdf.

Woodford, Michael. “Fiscal Requirements for Price Stability.” Journal of Money, Credit, and Banking, August2001, 33(3), pp. 669-728.

Woodford, Michael. Interest and Prices: Foundations of a Theory of Monetary Policy. Princeton, NJ: PrincetonUniversity Press, 2003.

Bullard


Seven Faces of “The Peril”

Documents

Transcript of Seven Faces of “The Peril”