View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Monetary Policy Rules?
The Jeffrey and Kathryn Cole Honors College Lecture
Michigan State University
East Lansing, Michigan
March 16, 1999
Published in the Federal Reserve Bank of St. Louis Review, March/April 1999, 81(2), pp. 3-12


ll aspects of our behavior, economic
and otherwise, are governed and
organized by various rules. The most
routine aspects of our lives become
unpredictable and even chaotic when not governed by well-defined and generally accepted
rules. For example, in the 20th century, every
country has adopted a legal rule that governs
automobile traffic. In the United States, it is
understood that automobiles are driven to the
right; in the United Kingdom, the rule is to drive
to the left. If, as a U.S. driver, you travel to
England and are unaware of the rule for driving
or choose to ignore it, it is quite likely that the
outcome will be tragic.
In a similar vein, the convention in the United
States is to walk to the right. Generally this works
well, and pedestrian traffic flows smoothly even
along crowded walkways. In some cultures, the
rules for pedestrian traffic do not appear to be as
clearly understood. There it is common to observe
considerable zigging and zagging in dense traffic
as pedestrians seek to avoid collisions in the
absence of a systematic decision process on how
to proceed. The absence of a well-understood
convention for pedestrian traffic increases transit
times and can result in considerable irritation.
Resources, which could be used productively,
are wasted.
What is a rule? A rule can be defined as
“nothing more than a systematic decision process
that uses information in a consistent and predict1

able way.”1 The concept of a monetary policy rule
is the application of this principle in the implementation of monetary policy by a central bank.
Why, then, the question mark in my title? There
is a large body of economics literature on the
rules-versus-discretion debate over monetary
policy. I do not intend to analyze, or psychoanalyze, this debate here. I especially do not
intend to address the political side of this debate—
whether it would be desirable for a national legislature to enact a monetary policy rule to be
executed by the country’s central bank. Rather,
my purpose is to examine what we mean by a
monetary policy rule followed by a central bank,
and to examine what we know about the construction, or design, of the rule.
I will first discuss some general issues in the
design of rules. Next, I’ll discuss what monetary
policy can achieve—we need to be clear about
what a policy rule is supposed to accomplish.
I’ll also review how the Federal Reserve conducts
monetary policy today to provide the background
necessary to understand the practical design issues
for a monetary rule. My final topic will be the
critical requirements that must underlie any satisfactory rule. That will bring me to our current
understanding of a practical rule.
Before digging into this topic further, however,
I want to emphasize that the views I express here
reflect my thinking and do not necessarily reflect
official Federal Reserve views. Bob Rasche, until
recently a member of the department of economics

Allan H. Meltzer, “Commentary: The Role of Judgment and Discretion in the Conduct of Monetary Policy,” Changing Capital Markets:
Implications for Monetary Policy, Federal Reserve Bank of Kansas City, 1993, p.223.



at Michigan State University, is really a co-author
of this lecture; he deserves credit for its strengths
and I’ll accept responsibility for its errors.

How are rules constructed? In many cases,
rules govern our interaction with the environment,
and the optimal rule can be established as the
solution to a well-defined engineering problem.
Given advances in technology, many rules, formerly executed manually, are now performed by
automatic devices. For example, thermostats
that control the heat in our homes and autopilots
that control the progress of aircraft embody such
rules. A second type of rule governs our interactions with others. When the absence of welldefined rules of behavior or the failure to conform
to a convention seriously harms the welfare of
others, the government implements a rule by law
and imposes penalties for disobedience. You will
be apprehended quickly if you insist on driving
on the right side of the road in England.
A third type of rule involves the formulation
of policy decisions. Here, a systematic decisionmaking process is complicated because individuals and market participants observe or infer the
actions of the policymakers and adjust their
behavior in ways that work to their benefit, given
their understanding of the policy regime. This
is the type of problem faced by monetary policy
This point is so important that it deserves
special emphasis. Compare two card games:
solitaire and poker. Solitaire is a game against
nature—the characteristics of a deck of cards.
Those characteristics do not change from one day
to the next. Poker is a game against intelligent
agents. The players may change from one day to
the next. Players learn about each other’s playing styles over time and change how they play.
Designing an optimal rule, or strategy, for poker
is a much more difficult problem than designing
an optimal rule for playing solitaire.

Monetary policy, needless to say, is more like
poker than solitaire. The goal of monetary policy,
however, is to make the economy better off, not
to go home at the end of the evening with your
friends’ money. The fact is, though, that monetary
policy affects interest rates, people make or lose
money from interest-rate fluctuations, and therefore, the markets are constantly trying to forecast
the next monetary policy adjustment. This interaction—policymakers trying to understand and
interpret markets, and markets trying to predict
what policymakers will do—makes the task of
designing an optimal monetary policy rule a very,
very difficult problem.

At the beginning of the 1960s, economists
generally believed that central banks could and
should be significant players in the effort to
achieve multiple social objectives: low inflation,
high growth, low unemployment and low nominal
interest rates. In addition, the Federal Reserve
was expected to contribute to specific efforts such
as encouraging balanced payments with the rest
of the world and a strong housing sector.
The notion that central banks can provide a
low-cost, over-the-counter “aspirin” that will
alleviate almost any ill that a society can face is
no longer credible. There is now a consensus
among economists and central bankers that the
only long-run effect a monetary authority can have
on an economy is to determine the sustained, or
trend, rate of inflation. That rate will result from
the rate at which the monetary authority injects
money into the economy.
The view that price stability depends upon
monetary conditions has a long history in monetary economics. Indeed, the basic proposition that
the amount of money determines the price level
originated long before economics was recognized
as a discipline. Simply put, I would like to note
that in the late 19th and early 20th centuries,
economists were precise about the nature of the
connection between money and the general price

Monetary Policy Rules?

level. Irving Fisher, among others, made important
contributions to monetary theory long before the
Great Depression. This idea—that the general
price level and its rate of increase depends primarily on the level of the money stock and its
rate of increase—fell out of favor with the rise of
Keynesian analysis in the 1930s and 1940s. The
idea was revived in the 1950s by Milton Friedman,
who has lived to win the intellectual battle that
sustained inflation is everywhere and always will
be a monetary phenomenon.
A consensus also exists that erratic monetary
policy has sometimes produced instabilities in
the economy. Most analysts now agree that Federal
Reserve actions contributed significantly to the
severity of the Great Depression in the United
States. Monetary policies can make the economy
either more or less stable. It is generally acknowledged that at least some—I think a lot—of the
credit for the stability of the U.S. economy during the past 15 years is due to Federal Reserve
policy under Chairmen Paul Volcker and Alan
Greenspan. Finally, it is generally accepted that
central banks are responsible for acting as a lender
of last resort in the event of a generalized liquidity crisis to maintain the soundness and function
of the payments mechanism. Most economists
accept the view that prompt Federal Reserve
actions in October 1987, after the stock market
crash and again last September after the Russian
default, were appropriate policy interventions.
Of course, we continue to debate the appropriate extent of Fed actions to alleviate a liquidity
crisis—indeed, even what truly qualifies as a
liquidity crisis—but that debate does not detract
from acceptance of the general principle that a
central bank response is desirable when the crisis
is severe.
The problem of designing monetary policy to
achieve sustained low inflation and more, rather
than less, stability is far from trivial. We know
much less about this task than we should. At this
point, no consensus exists on the size or reliability
of the short-run impact of monetary policy on an
economy. A considerable amount of professional
opinion, the general popular feeling, and financialmarket commentary hold that monetary policy

actions initially affect output, unemployment,
and real interest rates, even though the long-run
impact on these real variables is nil. Research
efforts to quantify these initial effects, however,
have failed to provide precise measures of the
impact, and at least one school of thought maintains that such short-run effects are negligible.
To judge monetary policy and central bankers,
we must concentrate on what they can reasonably
be expected to achieve, given our current state of
knowledge. There is a compelling case, I believe,
that the success or failure of monetary policy must
be judged first and foremost by whether a central
bank is able to achieve a low-inflation environment on a sustained basis. That environment is,
in turn, conducive to maximum growth and efficient utilization of the resources available to a
society. High growth and efficient utilization of
resources depend on government policies beyond
the central bank’s control. That fact, however, does
not change the proposition that a central bank’s
contribution should be judged primarily by the
average rate of inflation, and secondarily by the
stability, or lack thereof, of the overall economy.
By this standard, the history of the second half
of the 20th century, in the United States and in
other countries, is not kind to central bankers. For
a short period during the late 1950s and early
1960s, the U.S. economy (for all practical purposes) experienced price stability. The return to
low inflation following the Korean War was consistent with prior U.S. experience that inflation
was a wartime phenomenon. During peacetime,
citizens generally were unconcerned about
Periodically, the Gallup poll has asked, “What
do you think is the most important problem facing
the country today?” One of the response choices
was: “Inflation or the high cost of living.” In 1956,
several years after the end of the Korean conflict,
only 13 percent of respondents indicated this
concern. By 1964, the fraction of respondents
selecting this response had dropped to six percent.
Sixteen years later, in 1980, the economy
was at the peak of the inflation that had started
during the Vietnam War. Inflation as measured
by the consumer price index (CPI) had reached


double-digits. Short-term interest rates exceeded
20 percent. People came to understand that this
inflation could not be attributed to war; the
Vietnam War had ended by the mid1970s, but
inflation persisted and indeed rose for the rest of
the decade. The fraction of Gallup poll respondents who ranked inflation as the No. 1 problem
facing the economy rose almost continuously—
from 27 percent in 1972, to 47 percent in 1976,
to 61 percent in 1980. All segments of society
shared this concern.
Today, the price tranquility of the late 1950s
and early 1960s has been reestablished, both in
the United States and in most other industrial
countries. In the United States, annual inflation
during each year from 1983 through 1989 was
close to 4 percent. With the exception of a brief
increase just before and during the Persian Gulf
conflict in 1991, the inflation rate has declined
steadily to the point where the economy last year
was close to practical price stability. By 1993, only
1 percent of Gallup poll respondents ranked
inflation as the most important problem facing
the country. An article in the Wall Street Journal
last month reflected on a survey conducted by
Yale University economist Robert Shiller:2
It is now widely accepted that high rates of
inflation can damage an economy by distorting
markets, undermining public faith in government and forcing all sorts of wasted effort.
Mr. Shiller’s 1996 survey found that 84% of
the public—though only 46% of economists—
felt that preventing high inflation was as important as preventing drug abuse or deterioration
of schools. A return of inflation would be
jarring: The last time the Labor Department
checked, only one in five major union contracts
included an automatic cost-of-living adjustment, down from 60% in the early 1980s. A
spread of the Japanese-style deflationary spiral,
in which falling prices exacerbate a recession,
would also be painful.
But right now, the U.S. has neither. Instead,
it is experiencing an absence of inflation. Without inflation, the old-fashioned notion that
young couples can buy a house and grow into


their mortgage payments with ever-bigger paychecks may soon be as quaint as a telephone
with a dial. And retirees living off interest on
certificates of deposit and government bonds
will be shocked when their securities mature
to see how far interest rates have fallen as inflation ebbs. Already, rates on six-month certificates of deposit are down to an average of 4%,
according to BanxQuote Inc., and some big
banks are paying little more than 1% on regular
savings accounts.

Clearly, our economy—indeed many aspects
of our broader society—is deeply affected by
inflation and by the absence of inflation. The
public has no doubt, and I have no doubt, that the
absence of inflation is better. A critical question
facing all of us at the present time is whether the
inflation experience of the past 16 years will be
sustained. Or, will this period ultimately be
viewed in history as a wonderful stroke of good
luck—an anomaly in an age of otherwise nearly
permanent inflation?

The Federal Reserve has practiced a consistent approach to the implementation of monetary
policy at least since the mid1980s. Monetary
policy decisions are the responsibility of the
Federal Open Market Committee (FOMC). The
FOMC consists of the seven governors of the
Federal Reserve System, the president of the
Federal Reserve Bank of New York, and four of
the presidents of the remaining 11 regional Federal
Reserve Banks, on a rotating basis. This committee
meets eight times a year, to discuss the current
state of the economy and the prospects for nearterm developments. The committee then votes
on instructions—the Directive to the System Open
Market Account Manager—that specify a target
value for the federal funds interest rate. The federal
funds rate is the rate at which depository institutions borrow and lend to each other their reserve

David Wessel, “With Inflation Tamed, America Confronts an Unsettling Stability.” Wall Street Journal, February 22, 1999, p. A.1.

Monetary Policy Rules?

balances on the books of the Federal Reserve
Banks. The Fed usually refers to the target federal
funds rate as the intended rate.
Once these instructions have been approved,
it is the responsibility of the staff of the Open
Market Desk at the Federal Reserve Bank of New
York, in consultation with the Chairman and members of the Open Market Committee, to keep the
actual funds rate close to the intended rate. The
Desk proceeds by buying and selling U.S. government securities for the Federal Reserve’s account,
or by engaging in transactions that are the practical equivalent of buying and selling government
securities. When the Account Manager desires to
offset market forces that are driving the funds rate
above the intended rate set by the FOMC, the
Desk purchases securities in the open market for
the Fed’s account. When the Account Manager
desires to offset market forces that are driving the
funds rate below the intended rate, the Desk sells
securities into the market from the Fed’s account.
The direct result of such purchases and sales is
that the amount of currency and/or balances of
depositories at the Federal Reserve Banks is
increased or decreased.
This approach to implementing monetary
policy is not new. Exactly the same procedures
were employed during the late 1960s and throughout the 1970s, the period of rising inflation. Therefore, there is no guarantee that the tactics of
monetary policy, as currently practiced by the
FOMC, will be successful in maintaining a lowinflation environment; the exact same procedures
delivered the Vietnam-era inflation.

We must address two critical issues in the
process of designing a rule for monetary policy.
First, the rule must take into account the fact that
the individuals’ regarding the Fed’s future actions
are an important determinant of economic outcomes. Second, the rule must be very explicit
about the information the FOMC uses to determine
how to change the intended federal funds rate.

An important development—if not the important intellectual development throughout the
past 25 years in our understanding of how the
macroeconomy works—is the recognition that
expectations play a central role in affecting economic behavior. Previously, to the extent that
expectations were considered at all, they were
treated in a rather mechanical fashion. Contemporary analyses now postulate that individuals
do not simply look to past economic outcomes
to project the future path of important conditions
like the inflation rate. Instead, individuals understand that it is in their self-interest to contemplate
seriously what path the Federal Reserve likely
will pursue for monetary policy and to align their
expectations about future inflation with their
perceptions of Fed actions.
Such a role for expectations is not just an element of elegant and stylized economic theories.
Expectations influence market activities day in
and day out. Traders in the federal funds futures
contracts on the Chicago Board of Trade, for example, pore over testimony and speeches of the
Chairman and Federal Reserve officials, searching
for hints about whether the FOMC will change
the intended federal funds rate at its next meeting,
or some meeting after that. Financial markets can
gyrate widely in response to a remark whose interpretation is contrary to the prevailing impression.
A monetary policy rule must take into account
these market expectations and speculations. The
goal should be that interest rates and other market
prices will respond to objective information about
the economy—the same information that monetary
policy itself depends on. The fact that markets so
often respond to comments and speeches by Fed
officials indicates that the markets today are not
evaluating monetary policy in the context of a
well-articulated and well-understood monetary
rule. The problem is a deep and difficult one. The
Fed does not know how to specify its monetary
policy decisions so that the market can look at
the same data the Fed looks at and arrive at the
same conclusion. I make this statement not by
way of any criticism of my Fed colleagues or staff,
but simply as an honest statement of how things
are today. We apply our best judgment to the task


and do not rely on a formal rule, because we do
not have a formal rule we trust.
My point here is that a critical part of designing a rule is dealing with the interaction of the
Fed and the markets. Given this interaction, an
important first feature of any rule must be that it
is formulated in a systematic fashion and can be
communicated easily to the public. A rule the
public does not understand will not work satisfactorily because policy changes resulting from
application of the rule will constantly take the
markets by surprise. The public will not and
should not accept a procedure that creates policy
changes that seem totally unpredictable and,
therefore, arbitrary and capricious.
Second, a policy rule must have the correct
long-run properties. A rule that, if followed religiously, would permit inflation to rise or fall to
unacceptable levels would obviously be deficient.
This aspect of designing a rule is, fortunately,
relatively straightforward.
Third, I think it is desirable, though I confess
substantial uncertainty on this point, that a rule
rely heavily on the market itself. On the whole,
markets do a good job in allocating resources
efficiently and making judgments about things
that are difficult to predict. I think a rule will work
best if it can establish a solid and predictable base
for monetary policy, leaving maximum room for
markets to set interest rates and other prices. For
example, if we knew of a direct way to set the rate
of inflation to zero directly, then market interest
rates could be free to rise and fall as credit
demands rise and fall.

Now I’m down to the bottom line of this lecture—what might an actual monetary policy rule
look like?
The place to begin is with the policy rule
advocated by Milton Friedman, among others,
starting during the 1950s. The long-standing
controversy over monetary rules derives in large
part from this particular rule. Friedman’s proposed rule was that the Federal Reserve should

establish a constant rate of growth for the stock
of money and maintain that growth rate no matter
what emerged from the state of the economy.
Friedman’s opponents argued that, should the
Fed adopt such a rule, it would default on its
responsibilities to stabilize cyclical fluctuations
of the economy. They felt that such stabilization
required that the Fed exercise discretion in the
conduct of monetary policy. Friedman countered
that historically the Fed was the principal cause
of cyclical fluctuations in the economy and that
much of the desired stabilization of the economy
would be achieved if money growth were constant.
Nothing in the modern concept of a monetary
policy rule requires that the Fed pursue a policy
invariant to the state of the economy. The restrictions imposed on Fed decision-making by the
monetary-rule process, as defined here, only
require that decisions to change or not change
the intended federal funds rate repeatedly incorporate the same information and respond to that
information in the same way.
Critics of the rule approach argue that the Fed
must consider all available information about the
economy. There is nothing in the concept of a
monetary rule, however, that precludes the FOMC
from reaching its decisions based on a wide variety
of information. All that is required is that the same
information be considered and incorporated into
the decision-making process in the same fashion
each time the intended federal funds rate level is
reassessed. Neither the data consulted, nor the
weight placed on particular pieces of information,
should be altered in repeated decisions. Furthermore, operating under a monetary rule requires
that the basis for deciding whether to change the
intended federal funds rate be clearly communicated to the public. Everyone should be able to
make informed predictions about the future course
of policy, given knowledge of the same facts about
the state of the economy. The rule calling for a
constant growth rate of the money stock has many
desirable features:
• It is easy for the public to understand.
• The rate of inflation cannot take off toward
plus infinity or minus infinity if money
growth is held constant.

Monetary Policy Rules?

• Interest rates are free to fluctuate in response
to changing market conditions.
The Friedman rule, however, has not gained
general acceptance. One reason why is that the
term “money” must be defined in an acceptable
way if the rule is to work and be easily understood by the public. Many, and perhaps most,
economists today believe that changes in the
amount of money demanded by the public are
of sufficient size and duration that keeping the
money stock on a steady path will likely lead to
much larger fluctuations in the inflation rate and
level of economic activity than we’d like. The
better way of stating this point is to say that a
central bank using its best judgment can beat the
performance of the Friedman rule, and that this
claim is well demonstrated by the Fed’s performance since it began to attack inflation in 1979.
There are other criticisms of the Friedman rule.
My purpose here, however, is not to review this
whole debate but to discuss the issue of rules more
Others—especially Allan Meltzer and
Bennett McCallum—have worked on variants of
the Friedman rule. These are quantity-based rules
that yield a changing growth rate of the money
stock or the monetary base. The research is promising and deserves more attention than it gets.
Another approach is an interest-rate rule, in
contrast to the quantity rules just discussed. The
policy rule for interest rates that has been discussed most often for several years now was proposed by Stanford economist John Taylor in 1993.
His rule is an attractive one to consider because
it is so closely linked to traditional Fed practice
in setting an intended federal funds rate.
Taylor proposed that the federal funds rate
be determined by a rule with three basic terms in
it. First, the funds rate should equal an estimate
of the economy’s real rate of interest at a zero rate
of inflation plus the Fed’s target rate of inflation.
For example, with an estimate of an equilibrium
real rate of interest of 2 percent, and a long-run
target rate of inflation of 1 percent, the base rate
for the federal funds rate would be 3 percent.

The second term in the Taylor rule calls for
an adjustment to the intended federal funds rate
when the inflation rate deviates from the FOMC’s
target inflation rate. Continuing with the illustration that the target inflation rate is 1 percent, if
the actual inflation rate is 2 percent, then the inflation deviation is 1 percentage point. The Taylor
rule multiplies that deviation by a specified coefficient and adds the product to the intended federal
funds rate. For example, if the coefficient is 1.5,
then the inflation deviation of 1 percentage point
yields an intended federal funds rate that is higher
by 1.5 percentage points.
Taylor has emphasized the importance of
having a coefficient on the inflation deviation
term that is higher than 1.0. If the coefficient is
below 1.0, then an increase in inflation will call
forth an increase in the intended federal funds
rate that is smaller than the increase in inflation.
That means that the real rate of interest would
fall when inflation rises, which is a recipe for a
never-ending increase in inflation. Everyone agrees
that the coefficient on the inflation deviation needs
to be above 1.0, but how much above is unknown
at this time. There also is an issue of how to define
the inflation rate—what index to use and what
time period. Using the inflation rate over the last
month would introduce a great deal of random
noise into the federal funds rate set by the rule;
using the inflation rate averaged over the last five
years would yield a rule that responds too slowly
to changing conditions. The optimal length of
the averaging period is not known at this time.
Using an inflation forecast might be better, but
whose forecast?
The third term in the Taylor rule is the deviation of real gross domestic product (GDP) from the
path of potential GDP. We can call this quantity
the GDP deviation. Several ideas lie behind this
term. One is that if the Fed were to follow a money
growth rule in an economy in which the problems
with that rule did not exist, then interest rates
would rise and fall as credit demands rise and
fall with the strength of the economy. In this sense,
the Taylor rule mimics the behavior of interest
rates under a constant-money-growth rule. More
generally, it seems sensible that if the economy


is booming—running well above potential—then
interest rates should be somewhat higher to check
the excessive pressure on available labor and
capital resources. If the economy is slack—operating below potential—then interest rates should
be somewhat lower to encourage greater utilization of available resources.
The argument for a term in the policy rule
reflecting the GDP deviation is attractive. The
optimal size of the coefficient on this term is not
known, however. Whether the federal funds rate
should change by an amount equal to 0.5, or 2.0,
or some other number times the GDP deviation
is being investigated. Moreover, potential GDP is
not an observable variable.
Although different researchers have different
ideas about which is the best method of measuring
potential GDP, I do not regard this issue as critical,
because an error here will not send the economy
off permanently in one direction or the other. The
rule will be stabilizing, though not perfect, if
potential GDP is misestimated. The size of the
coefficient on the GDP deviation is an important
issue, however, as it determines how stabilizing
the rule is likely to be. This is a complicated
matter— too large a coefficient might induce
economic cycles around potential GDP and too
small a coefficient might permit sustained departures from a desirable path that would tend to
destabilize the rate of inflation.
Taylor argued that the behavior of the federal
funds rate incorporated in his rule is a reasonable
approximation to the actual process of adjustment
of the funds rate targets the FOMC used between
1987 and 1993—a period during which monetary
policy was quite successful. But his rule is much
more than a simple effort to fit the data. It incorporates important and sound theoretical principles
that need to be followed if monetary policy is to
be successful. As I have already emphasized,
however, there is much we do not understand
about the optimal construction of a monetary
policy rule. We should be wary about accepting
coefficients that seem to come out of a small slice


of history, no matter how successful policy was
during that period. Every empirical economist is
all too familiar with the phenomenon of a great
model fit during a sample period, followed by
utter disillusionment with the performance of
the model outside the sample period.
As you can imagine, there are numerous
potential problems with the Taylor rule; I’ve
mentioned just a few. But these problems also are
problems with the current conduct of monetary
policy. Gauging the current interest rate against
the equilibrium real rate of interest is an important part of our job today, but we don’t know what
the equilibrium real rate is in precise numerical
terms. We need to judge the current level of the
economy against its potential, but we don’t know
in precise numerical terms what that potential
is. Criticizing the Taylor rule, or any other rule,
for such reasons does not help solve the problems
policy-makers face, nor does it make a convincing
case that what we do now is better than following
an imperfect rule.
The Taylor rule has figured more prominently
than other proposed rules in recent discussions
of monetary policy rules. Undoubtedly, interest
in the Taylor rule reflects the fact that the current
implementation of monetary policy around the
globe focuses on manipulating a short-term interest rate. Thus, the adoption of a Taylor-type rule
would not require the Federal Reserve to alter its
operating procedures; the Taylor rule is an effort
to formalize the decision-making process that
generates the operating instructions in the Fed’s
current practice.
Another approach to monetary policy, known
as inflation targeting, has been instituted by the
central banks in several foreign countries.3 The
practice, which varies from country to country,
started with the Reserve Bank of New Zealand,
and has been adopted by the Bank of Canada, the
Bank of England, the Bank of Finland, the Swedish
Riksbank, and the Reserve Bank of Australia.
These central banks announce in advance their
policy objective for an inflation rate. This

Guy Debelle, Paul Masson, Miguel Savastano, and Sunil Sharma, “Inflation Targeting as a Framework for Monetary Policy,” Economic
Issues 15, International Monetary Fund, Washington, D.C., September 1998.

Monetary Policy Rules?

announcement reflects a public commitment of
the central bank to the policy objective. In none
of these cases has the central bank specified the
decision rule that it will use to achieve the stated
objective. In no market economy can the central
bank control the inflation rate or the price level
directly. It must intervene in some market to
manipulate a price or a quantity to steer the economy toward the desired objective—whether this
objective is announced or not. The variable manipulated by the central bank is defined as the policy
instrument. The monetary rule indicates the
process by which the central bank adjusts the
policy instrument when information on the performance of the economy relative to its policy
objective is received. Thus, merely announcing a
policy objective in terms of the inflation rate does
not assure that the central bank is operating under
a monetary policy rule. However, I believe that it
is desirable for central banks to be clear about
their objectives, and in this regard I believe that
inflation targeting is a desirable practice.

I believe that the Taylor rule is a promising
approach to better understanding monetary policy.
From what I have already said, however, it should
be clear that I do not believe that this rule is ready
for adoption. We need a lot more research.
Two areas deserve special attention. One
concerns a role for monetary aggregates and the
second a role for market interest rates, or some
other piece of market information.
As a general proposition, we can extract information from markets by studying both prices and
quantities. The total neglect of information about
the monetary aggregates in the Taylor rule opens
up a natural avenue to extend the rule. This is
not the place to present current research results,
especially since I don’t have any of this kind. I am
simply saying that, given all the evidence supporting Milton Friedman’s proposition that “inflation
is now and everywhere a monetary phenomenon,”

it seems to me that we ignore the behavior of the
monetary aggregates at our peril.
Obviously, interest rates reflect market expectations about the future. Might we incorporate
rates in the rule in some fashion? The idea is
intriguing, but there is a circularity problem
because it appears that the bond and money
markets respond significantly to changes in Fed
policy and to changes in expectations about Fed
policy. The more confidence the market has in
the Fed, the more the market will concentrate on
what the Fed is doing and the less the market will
concentrate on fundamentals other than the Fed.
Consider an analogy: If you are an investor but
know little about investing, it makes sense to
choose your investments by observing the decisions of an investor known for his superior performance. This strategy is likely to work better
for you than concentrating on investment fundamentals themselves, which by assumption, you
may not understand very well. The market
watches the Fed because the Fed is well informed
and because the Fed is the dominant player in
the money market.
The more confidence the market has in the
Fed’s willingness to do whatever is necessary to
maintain low inflation, the more sense it makes
for the market to concentrate on the Fed’s actions
rather than forming an independent judgment
about future inflation prospects. Therefore, the
Fed cannot use the behavior of interest rates in
the bond market to provide useful information
on how it should adjust the federal funds rate.
If, however, the Fed is able to adopt a precise
rule in the future, which is based on information
everyone can observe—such as that employed in
Taylor’s rule—then there may well be a place for
an interest rateterm in the rule. If the market could
be confident that the Fed would change the federal funds rate, only in response to the observed
inflation rate and the GDP deviation (and whatever other observable information the Fed
included in the rule), then market forecasts
would be incorporated in interest rates. Adding
an interest rate term to the rule would be a way
to add forecasts to the rule, which, in principle,
should make it work better than a rule based
solely on past data.


The concept of a monetary rule is attractive
for many reasons. To repeat the definition offered
at the beginning of this lecture, a rule is “nothing
more than a systematic decision-making process
that uses information in a consistent and predictable way.” Operating under a monetary rule
imposes accountability and transparency upon a
central bank. It requires that the policymakers be
specific about the rationale behind their policy
actions. The record of the decisions will then
contain information from which future decisionmakers can learn.
A policy rule ought not be considered irrevocable or unchangeable. At any time, our understanding of the short-run impact of monetary
policy on the economy is imperfect. Policymakers
necessarily operate within constraints imposed
by the current state of knowledge and should not
be blamed for outcomes that are impossible to
avoid given that knowledge. One of the benefits
of a policy rule is that a historical record will
exist that can be analyzed, and from this analysis
we can obtain an understanding of why past policy
actions did not produce the intended results. The
knowledge gained from such analysis can and


should be incorporated into the formulation of
future policy rules. Possible changes in the rule
should be studied and debated. When the analysis
indicates that the rule can be improved, the Fed
should announce the changes in advance and
explain the rationale for them.
A policy rule also provides the surest method
to pass the accumulated knowledge about the
effective operation of monetary policy to further
generations. This, after all, is how engineers, using
engineering theory and observation of skilled
human pilots, constructed autopilots. These
devices had to be tested and refined, and the
process took time. The devices are subject to
continuous improvement. Under normal flying
conditions, an autopilot now does a better job
than a human pilot at keeping an aircraft on its
desired course. As I have emphasized, designing
a monetary rule is a difficult task intellectually,
but it seems obvious to me that this is the path
we must travel.
In short, pursuing the path to developing
and then adhering to a rule provides the best
approach—perhaps the only approach—to
improving the practice of monetary policy over
the long run.