Risk and Uncertainty in Monetary Policy

Document Sample
Risk and Uncertainty in Monetary Policy Powered By Docstoc
                          THE LAST FIFTEEN YEARS

                      Risk and Uncertainty in Monetary Policy
                                             By ALAN GREENSPAN*

   I plan to sketch the key developments of the                   has been operating in an environment particu-
past decade and a half of monetary policy in the                  larly conducive to the pursuit of price stability.
United States from the perspective of someone                     The principal features of this environment in-
who has been in the policy trenches. I will offer                 cluded (i) increased political support for stable
some conclusions about what I believe has been                    prices, which was the consequence of, and re-
learned thus far, though I suspect, as is so often                action to, the unprecedented peacetime inflation
the case, the passing of time, further study, and                 in the 1970’s, (ii) globalization, which un-
reflection will deepen our understanding of                        leashed powerful new forces of competition,
these developments. This is a personal state-                     and (iii) an acceleration of productivity, which
ment; I am not speaking for my current col-                       at least for a time held down cost pressures.
leagues on the Federal Open Market Committee                         I believe we at the Fed, to our credit, did
(FOMC) or the many others with whom I have                        gradually come to recognize the structural eco-
served over these many years.1                                    nomic changes that we were living through and
   The tightening of monetary policy by the                       accordingly altered our understanding of the
Federal Reserve in 1979, then led by my pre-                      key parameters of the economic system and our
decessor Paul Volcker, ultimately broke the                       policy stance. The central banks of other indus-
back of price acceleration in the United States,                  trialized countries have grappled with many of
ushering in a two-decade long decline in infla-                    the same issues.
tion that eventually brought us to the current                       But as we lived through it, there was much
state of price stability.                                         uncertainty about the evolving structure of the
   The fall in inflation over this period has been                 economy and about the influence of monetary
global in scope, and arguably beyond the ex-                      policy. Despite those uncertainties, the trauma
pectations of even the most optimistic inflation-                  of the 1970’s was still so vivid throughout the
fighters. I have little doubt that an unrelenting                  1980’s that preventing a return to accelerating
focus of monetary policy on achieving price                       prices was the unvarying focus of our efforts
stability has been the principal contributor to                   during those years.
disinflation. Indeed, the notion, advanced by                         In recognition of the lag in monetary policy’s
Milton Friedman more than 30 years ago, that                      impact on economic activity, a preemptive re-
inflation is everywhere and always a monetary                      sponse to the potential for building inflationary
phenomenon, is no longer a controversial prop-                    pressures was made an important feature of
osition in the profession. But the size and geo-                  policy. As a consequence, this approach ele-
graphic extent of the decline in inflation raises                  vated forecasting to an even more prominent
the question of whether other forces have been                    place in policy deliberations.
at work as well.                                                     After an almost uninterrupted stint of easing
   I am increasingly of the view that, at a min-                  from the summer of 1984 through the spring of
imum, monetary policy in the last two decades                     1987, the Fed again began to lean against in-
                                                                  creasing inflationary pressures, which were in
                                                                  part the indirect result of rapidly rising stock
   * Board of Governors of the Federal Reserve System,            prices. We had recognized the risk of an adverse
20th Street and Constitution Avenue, N.W., Washington,            reaction in a stock market that had recently
DC 20551.
     I, nonetheless, wish to thank my colleagues David            experienced a steep run-up—indeed, we ac-
Stockton, David Wilcox, Don Kohn, Ben Bernanke, and               tively engaged in contingency planning against
John Taylor, for their many suggestions and reminiscences.        that possibility.
34                                  AEA PAPERS AND PROCEEDINGS                                       MAY 2004

   In the event, the crash in October 1987 was       left those pressures unchecked, we would have
far more traumatic than any of the possible          put at risk some of the hard-won gains that had
scenarios we had identified. Previous planning        been achieved over the preceding decade and a
was only marginally useful in that episode. We       half. So, starting from a real federal funds rate
operated essentially in a crisis mode, respond-      that was close to zero, a preemptive tightening
ing with an immediate and massive injection of       was initiated. The resulting rise in the funds rate
liquidity to help stabilize highly volatile finan-    of 300 basis points over 12 months apparently
cial markets. However, most of our stabilization     defused those nascent inflationary pressures.
efforts were directed at keeping the payments           Though economic activity hesitated in early
system functioning and markets open. The con-        1995, it soon steadied, confirming the achieve-
cern over the possible fallout on economic ac-       ment of a historically elusive soft landing. The
tivity from so sharp a stock price decline kept us   success of that period set up two powerful ex-
easing into early 1988. But the economy weath-       pectations that were to influence developments
ered that shock reasonably well, and our easing      over the subsequent decade. One was the ex-
extended perhaps longer than hindsight has in-       pectation that inflation could be controlled over
dicated was necessary.                               the business cycle and that price stability was an
   That period was followed by a preemptive          achievable objective. The second expectation,
tightening that brought the federal funds rate       in part a consequence of more stable inflation,
close to 10 percent by early 1989. In the sum-       was that overall economic volatility had been
mer of that year, we sensed enough softening of      reduced and would likely remain lower than it
activity to warrant beginning a series of rate       had previously.
reductions. However, the weakening of demand            Of course, these new developments brought
already under way, some pullback of credit by        new challenges. In particular, the prospect that a
lenders, and the spike in oil prices associated      necessary cyclical adjustment was now behind
with Iraq’s invasion of Kuwait were sufficient        us fostered increasing levels of optimism, which
to produce a marked contraction of activity in       were manifested in a fall in bond risk spreads
the fall of 1990. But perhaps aided by our           and a rise in stock prices. The associated decline
preemptive action, the recession was, to then,       in the cost of equity capital further spurred
the mildest in postwar history.                      already developing increases in capital invest-
   However, the recovery also was more modest        ment and productivity growth, both of which
than usual, in large measure because of the          broadened impressively in the latter part of the
notable financial “headwinds” that confronted         1990’s.
businesses. Those headwinds were primarily              The rise in structural productivity growth was
generated by the constriction of credit in re-       not obvious in the official data on gross product
sponse to major losses at banks, associated with     per hour worked until later in the decade, but
real-estate and foreign lending, coupled with a      precursors had emerged earlier. The pickup in
crisis in the savings and loan industry that had     new bookings and order backlogs for high-tech
its origins in a serious maturity mismatch as        capital goods in 1993 seemed incongruous
interest rates rose. With their access to managed    given the sluggish economic environment at the
funds threatened and the quality of their loan       time. Plant managers apparently were reacting
portfolio (and hence their capital) uncertain,       to what they perceived to be elevated prospec-
these depositories were most reluctant to lend.      tive rates of return on the newer technologies, a
   Policy eased gradually but persistently to        judgment that was confirmed as orders and prof-
counter the effects of these developments, with      its continued to increase through 1994 and
the funds rate falling to 3 percent by September     1995. Moreover, even though hourly labor com-
1992, its lowest level since the early 1960’s.       pensation and profit margins were rising, prices
The uptilt to the term structure of interest rates   were being contained, implying increasing
in a generally low-interest-rate environment re-     growth in output per hour.2
stored bank profitability and, eventually, bank
capital. The credit crunch slowly lifted.
   By early 1994, as the headwinds of financial         2
                                                         That growth was showing through in gross income per
restraint abated, it became clear that underlying    hour. An increasingly negative statistical discrepancy was
price pressures were again building. If we had       masking the rise in productivity as measured by the official
VOL. 94 NO. 2                     INNOVATIONS AND ISSUES IN MONETARY POLICY                                                  35

   As a consequence of the improving trend in                   price bubble of the latter part of the 1990’s.4
structural productivity growth that was apparent                Looking back on those years, it is evident that
from 1995 forward, we at the Fed were able to                   technology-driven increases in productivity
be much more accommodative to the rise in                       growth imparted significant upward momentum
economic growth than our past experiences                       to expectations of earnings growth and, accord-
would have deemed prudent. We were moti-                        ingly, to stock prices.5 At the same time, an
vated, in part, by the view that the evident                    environment of increasing macroeconomic sta-
structural economic changes rendered suspect,                   bility reduced perceptions of risk. In any event,
at best, the prevailing notion in the early 1990’s              Fed policymakers were confronted with forces
of an elevated and reasonably stable NAIRU                      that none of us had previously encountered.
(non-accelerating inflation rate of unemploy-                    Aside from the then-recent experience of Japan,
ment). Those views were reinforced as inflation                  only remote historical episodes gave us clues to
continued to fall in the context of a declining                 the appropriate stance for policy under such
unemployment rate that by 2000 had dipped                       conditions. The sharp rise in stock prices and
below 4 percent in the United States for the first               their subsequent fall were, thus, an especial
time in three decades.                                          challenge to the Federal Reserve.
   Notions that prevailed for a time in the                        It is far from obvious that bubbles, even if
1970’s and early 1980’s that even high single-                  identified early, can be preempted at lower cost
digit inflation did not measurably impede eco-                   than a substantial economic contraction and
nomic growth were gradually abandoned as the                    possible financial destabilization—the very out-
evidence of significant benefits of low inflation                  comes we would be seeking to avoid. In fact,
became increasingly persuasive. Moreover, the                   our experience over the past two decades sug-
variance of GDP growth markedly lessened as                     gests that a moderate monetary tightening that
inflation tumbled from its double-digit high in                  deflates stock prices without substantial effect
the early 1980’s. To preserve these benefits, we                 on economic activity has often been associated
engaged in our most recent preemptive tighten-                  with subsequent increases in the level of stock
ing in early 1999 that brought the funds rate to                prices.6 Arguably, markets that pass that type of
61⁄2 percent by May 2000.
   Our goal of price stability was achieved by
most analysts’ definition by mid-2003. Unstint-                     4
                                                                      It is notable, that in the United States, surges in price–
ing and largely preemptive efforts over two                     earnings ratios, a presumed essential characteristic of an
decades had finally paid off. Throughout the                     equity price bubble, are not observed with elevated inflation
period, a key objective has been to ensure that                     5
                                                                      But, as the Federal Reserve indicated in congressional
our response to incipient changes in inflation                   testimony in July 1999, “... productivity acceleration does
was forceful enough. As John Taylor has em-                     not ensure that equity prices are not overextended. There
phasized, in the face of an incipient increase in               can be little doubt that if the nation’s productivity growth
                                                                has stepped up, the level of profits and their future potential
inflation, nominal interest rates must move up                   would be elevated. That prospect has supported higher stock
more than one-for-one.3                                         prices. The danger is that, in these circumstances, an un-
   Perhaps the greatest irony of the past decade                warranted, perhaps euphoric, extension of recent develop-
is that the gradually unfolding success against                 ments can drive equity prices to levels that are
inflation may well have contributed to the stock                 unsupportable even if risks in the future become relatively
                                                                small. Such straying above fundamentals could create prob-
                                                                lems for our economy when the inevitable adjustment oc-
                                                                curs” (Greenspan, testimony before the Committee on
                                                                Banking and Financial Services, U.S. House of Represen-
data that relied on gross product per hour. As I indicated in   tatives, 22 July 1999).
the fall of 1994, “we are observing ... [an] opening up of            For example, stock prices rose following the comple-
margins ... . But unit labor costs apparently have been so      tion of the more than 300-basis-point rise in the federal
well contained by productivity gains at this stage that cost    funds rate in the 12 months ending in February 1989. And
pressures have not flowed into final goods prices” (FOMC          during the year beginning in February 1994, when the
transcripts, 27 September 1994, p. 37).                         Federal Reserve again raised the federal funds target 300
     See, for example, “A Half-Century of Changes in Mon-       basis points, stock prices initially flattened. But as soon as
etary Policy,” John B. Taylor’s remarks delivered at the        that round of tightening was completed, prices resumed
conference in honor of Milton Friedman, 8 November 2002         their marked upward advance. From mid-1999 through May
(unpublished manuscript, U.S. Department of the Treasury,       2000, the federal funds rate was raised 150 basis points.
pp. 9 –10).                                                     However, equity price increases were largely undeterred
36                                            AEA PAPERS AND PROCEEDINGS                                            MAY 2004

stress test are presumed particularly resilient.                  June 2003 to its current 1 percent, the lowest
The notion that a well-timed incremental tight-                   level in 45 years. We were able to be unusually
ening could have been calibrated to prevent the                   aggressive in the initial stages of the recession
late-1990’s bubble while preserving economic                      of 2001 because both inflation and inflation
stability is almost surely an illusion.7                          expectations were low and stable. We thought
   Instead of trying to contain a putative bubble                 we needed to be, and could be, forceful in 2002
by drastic actions with largely unpredictable                     and 2003 as well because, with demand weak,
consequences, we chose, as we noted in our                        inflation risks had become two-sided for the first
mid-1999 congressional testimony, to focus on                     time in 40 years.
policies “to mitigate the fallout when it occurs                     There appears to be enough evidence, at least
and, hopefully, ease the transition to the next                   tentatively, to conclude that our strategy of ad-
expansion.”8                                                      dressing the bubble’s consequences rather than
   During 2001, in the aftermath of the bursting                  the bubble itself has been successful. Despite
of the bubble and the acts of terrorism in Sep-                   the stock-market plunge, terrorist attacks, cor-
tember 2001, the federal funds rate was lowered                   porate scandals, and wars in Afghanistan and
43⁄4 percentage points. Subsequently, another 75                  Iraq, we experienced an exceptionally mild
basis points were pared, bringing the rate by                     recession— even milder than that of a decade
                                                                  earlier. As I discuss later, much of the ability of
                                                                  the U.S. economy to absorb these sequences of
during that period despite what now, in retrospect, was the       shocks resulted from notably improved struc-
exhausted tail of a bull market. Stock prices peaked in           tural flexibility. But highly aggressive monetary
March 2000, but the market basically moved sideways until         ease was doubtless also a significant contributor
September of that year.                                           to stability.9
    Such data suggest that nothing short of a sharp increase
in short-term rates that engenders a significant economic             The Federal Reserve’s experiences over the
retrenchment with all its attendant risks is sufficient to         past two decades make it clear that uncertainty
check a nascent bubble. Certainly, 300 basis points proved        is not just a pervasive feature of the monetary
inadequate to even dent stock prices in 1994.                     policy landscape; it is the defining characteristic
      Some have asserted that the Federal Reserve can de-         of that landscape. The term “uncertainty” is
flate a stock price bubble, rather painlessly, by boosting
margin requirements. The evidence suggests otherwise.             meant here to encompass both “Knightian un-
First, the amount of margin debt is small, having never           certainty,” in which the probability distribution
amounted to more than about 13⁄4 percent of the market            of outcomes is unknown, and “risk,” in which
value of equities; moreover, even this figure overstates the       uncertainty of outcomes is delimited by a
amount of margin debt used to purchase stock, as such debt
also finances short sales of equity and transactions in non-       known probability distribution. In practice, one
equity securities. Second, investors need not rely on margin      is never quite sure what type of uncertainty one
debt to take a leveraged position in equities. They can           is dealing with in real time, and it may be best
borrow from other sources to buy stock. Or, they can
purchase options, which will affect stock prices given the
linkages across markets.
    Thus, not surprisingly, the preponderance of research
suggests that changes in margins are not an effective tool for         Some have argued that, as a consequence of the 1995–
reducing stock-market volatility. It is possible that margin      2000 speculative episode, long-term imbalances remain,
requirements inhibit very small investors whose access to         having been only partly addressed since early 2001, the
other forms of credit is limited. If so, the only effect of       peak of the post-bubble business cycle. For example, large
increasing margin requirements is to price out of the market      residues of household and external debt are perceived as
the very small investor without addressing the broader issue      barriers to future growth. But in the past, imbalances that
of stock price bubbles.                                           led to business contractions were rarely fully reversed be-
    If a change in margin requirements were taken by inves-       fore the subsequent economic upturn began. Presumably
tors as a signal that the central bank would soon tighten         they were fully reversed in later periods, or they continued
monetary policy enough to burst a bubble, then there might        to fester, but not by enough to halt economic growth.
be the appearance of a causal effect. But it is the prospect of      Even if imbalances persist in our current environment,
monetary policy action, not the margin increase, that should      the business decline that began in March 2001 came to an
be viewed as the trigger. In a similar manner, history tells us   end in November of that year, according to the National
that “jawboning” asset markets will be ineffective unless         Bureau of Economic Research. We experienced tepid re-
backed by action.                                                 covery until the second half of last year, when GDP accel-
      Quoted from testimony before the Committee on Bank-         erated considerably. Hence, when the next recession arrives,
ing and Financial Services, U.S. House of Representatives,        as it inevitably will, it will be a stretch to attribute it to
22 July 1999.                                                     speculative imbalances of many years earlier.
VOL. 94 NO. 2               INNOVATIONS AND ISSUES IN MONETARY POLICY                                   37

to think of a continuum ranging from well-              A policy action that is calculated to be opti-
defined risks to the truly unknown.                   mal based on a simulation of one particular
   As a consequence, the conduct of monetary         model may not, in fact, be optimal once the full
policy in the United States has come to involve,     extent of the risks surrounding the most likely
at its core, crucial elements of risk management.    path is taken into account. In general, different
This conceptual framework emphasizes under-          policies will exhibit different degrees of robust-
standing as much as possible the many sources        ness with respect to the true underlying struc-
of risk and uncertainty that policymakers face,      ture of the economy.
quantifying those risks when possible, and as-          For example, policy A might be judged as
sessing the costs associated with each of the        best advancing the policymakers’ objectives,
risks. In essence, the risk-management ap-           conditional on a particular model of the econ-
proach to monetary policymaking is an applica-       omy, but might also be seen as having relatively
tion of Bayesian decision-making.                    severe adverse consequences if the true struc-
   This framework also entails devising, in light    ture of the economy turns out to be other than
of those risks, a strategy for policy directed at    the one assumed. On the other hand, policy B
maximizing the probabilities of achieving over       might be somewhat less effective in advancing
time our goals of price stability and the maxi-      the policy objectives under the assumed base-
mum sustainable economic growth that we as-          line model but might be relatively benign in the
sociate with it. In designing strategies to meet     event that the structure of the economy turns out
our policy objectives, we have drawn on the          to differ from the baseline. A year ago, these
work of analysts, both inside and outside the        considerations inclined Federal Reserve policy-
Fed, who over the past half century have             makers toward an easier stance of policy aimed
devoted much effort to improving our under-          at limiting the risk of deflation even though
standing of the economy and its monetary trans-      baseline forecasts from most conventional mod-
mission mechanism. A critical result has been        els at that time did not project deflation; that is,
the identification of a relatively small set of key   we chose a policy that, in a world of perfect
relationships that, taken together, provide a use-   certainty, would have been judged to be too
ful approximation of our economy’s dynamics.         loose.
Such an approximation underlies the statistical         As this episode illustrates, policy practition-
models that we at the Federal Reserve employ         ers operating under a risk-management para-
to assess the likely influence of our policy          digm may, at times, be led to undertake actions
decisions.                                           intended to provide insurance against especially
   However, despite extensive efforts to capture     adverse outcomes. Following the Russian debt
and quantify what we perceive as the key mac-        default in the autumn of 1998, for example, the
roeconomic relationships, our knowledge about        FOMC eased policy despite our perception that
many of the important linkages is far from com-      the economy was expanding at a satisfactory
plete and, in all likelihood, will always remain     pace and that, even without a policy initiative, it
so. Every model, no matter how detailed or how       was likely to continue doing so.10 We eased
well designed, conceptually and empirically, is      policy because we were concerned about the
a vastly simplified representation of the world       low-probability risk that the default might trig-
that we experience with all its intricacies on a     ger events that would severely disrupt domestic
day-to-day basis.                                    and international financial markets, with out-
   Given our inevitably incomplete knowledge         sized adverse feedback to the performance of
about key structural aspects of an ever-changing     the U.S. economy.
economy and the sometimes asymmetric costs              The product of a low-probability event and a
or benefits of particular outcomes, a central         potentially severe outcome was judged a more
bank needs to consider not only the most likely      serious threat to economic performance than the
future path for the economy, but also the distri-    higher inflation that might ensue in the more
bution of possible outcomes about that path.         probable scenario. That possibility of higher
The decision-makers then need to reach a judg-
ment about the probabilities, costs, and benefits
of the various possible outcomes under alterna-        10
                                                            See minutes of the FOMC meeting of 29 September
tive choices for policy.                             1998.
38                                  AEA PAPERS AND PROCEEDINGS                                MAY 2004

inflation caused us little concern at the time,       operate in a way that does not depend on a fixed
largely because increased productivity growth        economic structure based on historically aver-
was resulting in only limited increases in unit      age coefficients. We often fit simple models
labor costs and heightened competition, driven       only because we cannot estimate a continuously
by globalization, was thwarting employers’           changing set of parameters without vastly more
ability to pass through those limited cost in-       observations than are currently available to us.
creases into prices. Given the potential conse-      Moreover, we recognize that the simple linear
quences of the Russian default, the benefits of       functions underlying most of our econometric
the unusual policy action were judged to out-        structures may not hold outside the range in
weigh its costs.                                     which adequate economic observations exist.
   Such a cost– benefit analysis is an ongoing        For example, it is difficult to have much confi-
part of monetary-policy decision-making and          dence in the ability of models fit to the data of
causes us to tip more toward monetary ease           the moderate inflations of the postwar period to
when a contractionary event, such as the Rus-        accurately predict what the behavior of the
sian default, seems especially likely or the costs   economy would be in an environment of aggre-
associated with it seem especially high.             gate price deflation.
   The 1998 liquidity crisis and the crises asso-       In pursuing a risk-management approach to
ciated with the stock-market crash of 1987 and       policy, we must confront the fact that only a
the terrorism of September 2001 prompted the         limited number of risks can be quantified with
type of massive ease that has been the historic      any confidence. And even these risks are gen-
mandate of a central bank. Such crises are pre-      erally quantifiable only if we accept the assump-
cipitated by the efforts of market participants to   tion that the future will, at least in some
convert illiquid assets into cash. When con-         important respects, resemble the past. Policy-
fronted with uncertainty, especially Knightian       makers often have to act, or choose not to act,
uncertainty, human beings invariably attempt to      even though we may not fully understand the
disengage from medium- to long-term commit-          full range of possible outcomes, let alone each
ments in favor of safety and liquidity. Because      possible outcome’s likelihood. As a result, risk
economies, of necessity, are net long (that is,      management often involves significant judg-
have net real assets) attempts to flee these assets   ment as we evaluate the risks of different events
cause prices of equity assets to fall, in some       and the probability that our actions will alter
cases dramatically. In the crisis that emerged in    those risks.
the autumn of 1998, pressures extended beyond           For such judgment, policymakers have
equity markets. Credit-risk spreads widened          needed to reach beyond models to broader,
materially, and investors put a particularly high    though less mathematically precise, hypotheses
value on liquidity, as evidenced by the extraor-     about how the world works. For example, in-
dinarily wide yield gaps that emerged between        ferences about how market participants and,
on-the-run and off-the-run U.S. Treasuries.          hence, the economy might respond to a mone-
   The immediate response on the part of the         tary policy initiative may need to be drawn from
central bank to such financial implosions must        evidence about past behavior during a period
be to inject large quantities of liquidity— or, as   only roughly comparable to the current
Walter Bagehot (1873 p. 85) put it, describing       situation.
such policies of the Bank of England more than          Some critics have argued that such an
a century ago, in a panic the Bank should lend       approach to policy is too undisciplined—
at very high rates of interest “to all that bring    judgmental, seemingly discretionary, and diffi-
good securities quickly, freely, and readily.”       cult to explain. The Federal Reserve, they
This was perhaps an early articulation of a crisis   conclude, should attempt to be more formal in
risk-management policy for a central bank.           its operations by tying its actions solely, or in
   The economic world in which we function is        the weaker paradigm, largely, to the prescrip-
best described by a structure whose parameters       tions of a simple policy rule. Indeed, rules that
are continuously changing. The channels of           relate the setting of the federal funds rate to the
monetary policy, consequently, are changing in       deviations of output and inflation from their
tandem. An ongoing challenge for the Federal         respective targets, in some configurations, do
Reserve—indeed, for any central bank—is to           seem to capture the broad contours of what we
VOL. 94 NO. 2                INNOVATIONS AND ISSUES IN MONETARY POLICY                               39

did over the past decade and a half. And the          The success of monetary policy depends impor-
prescriptions of formal rules can, in fact, serve     tantly on the quality of forecasting. The ability
as helpful adjuncts to policy, as many of the         to gauge risks implies some judgment about
proponents of these rules have suggested. But at      how current economic imbalances will ulti-
crucial points, like those in our recent policy       mately play out.
history (the stock market crash of 1987, the             Thus, both econometric and qualitative mod-
crises of 1997–1998, and the events that fol-         els need to be continually tested. The first signs
lowed September 2001), simple rules will be           that a relationship may have changed is usually
inadequate as either descriptions or prescrip-        the emergence of events that seem inconsistent
tions for policy. Moreover, such rules suffer         with our hypotheses of the way the economic
from much of the same fixed-coefficient diffi-           world is supposed to behave. The anomalous
culties we have with our large-scale models.          rise in high-tech capital-goods orders in 1993,
   To be sure, sensible policy-making can be          to which I alluded earlier, is one such example.
accomplished only with the aid of a rigor-            The credit crunch of the early 1990’s is another.
ous analytic structure. A rule does provide a            The emergence of inflation-targeting in re-
benchmark against which to assess emerging            cent years is an interesting development in this
developments. However, any rule capable of en-        regard. As practiced, it emphasizes forecasts,
compassing every possible contingency would           but within a more rule-like structure that skews
lose a key aspect of its attractiveness: simplic-     monetary policy toward inflation containment
ity. On the other hand, no simple rule could          as the primary goal. Indeed, its early applica-
possibly describe the policy action to be taken       tions were in high-inflation countries where dis-
in every contingency and thus provide a satis-        cretionary monetary policy fell into disrepute.
factory substitute for an approach based on the          Inflation-targeting often originated as a
principles of risk management.                        fairly simple structure concentrating solely on
   As I indicated earlier, policy has worked off      inflation outcomes, but it has evolved into
a risk-management paradigm in which the risk          more-discretionary forms requiring complex
and cost– benefit analyses depend on forecasts         judgments for implementation. Indeed, this
of probabilities developed from large macro-          evolution has gone so far that the actual prac-
models, numerous submodels, and judgments             tice of monetary policy by inflation-targeting
based on less mathematically precise regimens.        central banks now closely resembles the prac-
Such judgments, by their nature, are based on         tice of those central banks, such as the Euro-
bits and pieces of history that cannot formally       pean Central Bank, the Bank of Japan, and the
be associated with an analysis of variance.           Federal Reserve, that have not chosen to
   Yet, there is information in those bits and        adopt that paradigm.
pieces. For example, while we have been unable           In practice, most central banks, at least those
to readily construct a variable that captures the     not bound by an exchange-rate peg, behave in
apparent increased degree of flexibility in the        roughly the same way. They seek price stability
United States or the global economy, there has        as their long-term goal and, accounting for the
been too much circumstantial evidence of this         lag in monetary policy, calibrate the setting of
critically important trend to ignore its existence.   the policy rate accordingly. Central banks gen-
Increased flexibility is a likely source of chang-     erally appear to have embraced a common
ing structural coefficients.                           model of the channels through which monetary
   Our problem is not, as is sometimes alleged,       policy functions, although the specifics and em-
the complexity of our policy-making process,          phasis given to those channels vary according to
but the far greater complexity of a world econ-       our particular circumstances. All banks ease
omy whose underlying linkages appear to be            when economic conditions ease and tighten
continuously evolving. Our response to that           when economic conditions tighten, even if in
continuous evolution has been disciplined by          differing degrees, regardless of whether they are
the Bayesian type of decision-making in which         guided by formal or informal inflation targets.
we have engaged.                                         As yet unresolved is whether the mere an-
   While all, no doubt, would prefer that it were     nouncement that a central bank intends to
otherwise, there is no way to dismiss what has        engage in inflation-targeting increases the
to be obvious to every monetary policymaker.          credibility of the central bank’s inclination to
40                                   AEA PAPERS AND PROCEEDINGS                                      MAY 2004

maintain price stability and, hence, assists in the      Yet, during the last quarter century, policy-
anchoring of inflation expectations. The Bank          makers managed to defuse dangerous inflation-
of England’s recent experiences may be en-            ary forces and dealt with the consequences of a
couraging in this regard. But, presumably, we         stock-market crash, a large asset price bubble,
will not know for sure the significance of for-        and a series of liquidity crises. These events did
mal inflation-targeting as a tool until the world      not distract us from the pursuit and eventual
economy is subjected to shocks of sufficient           achievement of price stability and the greater
magnitude to assess the differential perfor-          economic stability that goes with it.
mance of those who do not employ formally                As we confront the many unspecifiable dan-
announced inflation targets. To date, inflation has     gers that lie ahead, the marked improvement in
fallen for formal targeters, but it has fallen for    the degree of flexibility and resilience exhibited
others as well.                                       by our economy in recent years should afford us
   Under the rubric of risk management are a          considerable comfort.11 Assuming that it will
number of specific issues that we at the Fed had       persist, the trend toward increased flexibility
to address over the past decade and a half and        implies that an ever-greater part of the resolu-
that will likely resurface to confront future mon-    tion of economic imbalances will occur through
etary policymakers. Most prominent is the ap-         the actions of business firms and households.
propriate role of asset prices in policy. In          Less will be required from the risk-laden initi-
addition to the narrower issue of product price       atives of monetary policymakers.
stability, asset prices will remain high on the          Each generation of policymakers has had to
research agenda of central banks for years to         grapple with a changing portfolio of problems.
come. As the ratios of gross liabilities and gross    So while we eagerly draw on the experiences
assets to GDP continue to rise, owing to ex-          of our predecessors, we can be assured that we
panding domestic and international financial in-       will confront different problems in the future.
termediation, the visibility of asset prices          The innovative technologies that have helped
relative to product prices will itself rise. There    us reap enormous efficiencies will doubtless
is little dispute that the prices of stocks, bonds,   present us with challenges that we cannot cur-
homes, real estate, and exchange rates affect         rently anticipate.
GDP. But most central banks have chosen, at              We were fortunate, as I pointed out in my
least to date, to view asset prices not as targets    opening remarks, to have worked in a particu-
of policy, but as economic variables to be con-       larly favorable structural and political environ-
sidered through the prism of the policy’s ulti-       ment. But we trust that monetary policy has
mate objective.                                       meaningfully contributed to the impressive per-
   As the transcripts of FOMC meetings attest,        formance of our economy in recent decades.
making monetary policy is an especially hum-
bling activity. In hindsight, the paths of infla-                         REFERENCES
tion, real output, stock prices, and exchange
rates may have seemed preordained, but no such        Bagehot, Walter. Lombard Street: A description
insight existed as we experienced it at the time.        of the money market. New York: Scribner,
In fact, uncertainty characterized virtually every       Armstrong, 1873.
meeting, and as the transcripts show, our ability
to anticipate was limited. From time to time the
FOMC made decisions, some to move and some               11
                                                           See my testimony before the Committee on Banking,
not to move, that we came to regret.                  Housing, and Urban Affairs, U.S. Senate, 11 February 2003.

Shared By: