How does the Fed decide the appropriate setting
for the policy instrument?
The Fed's job of stabilizing output in the short run and promoting
price stability in the long run involves several steps. First, the Fed
tries to estimate how the economy is doing now and how it's likely to
do in the near term—say, over the next couple of years or so. Then it
compares these estimates to its goals for the economy and inflation.
If there's a gap between the estimates and the goals, the Fed then has
to decide how forcefully and how swiftly to act to close that gap. Of
course, the lags in policy complicate this process. But so do a host
of other things.
What things complicate the process of determining how
the economy is doing?
Even the most up-to-date data on key variables
like employment, growth, productivity, and so on, reflect conditions
in the past, not conditions
today; that's why the process of monetary policymaking has been compared
to driving while looking only in the rearview mirror. So, to get
a reasonable estimate of current and near-term economic conditions, the
tries to figure out what the most relevant economic developments
these might be things like the government's taxing and spending policies,
economic developments abroad, financial conditions at home and abroad,
and the use of new technologies that boost productivity. These developments
can then be incorporated into an economic model to see how the economy
is likely to evolve over time.
Sounds easy—plug the numbers into the
model and get an answer. So what's the problem?
There are lots of problems. One problem is that models are
only approximations—they can't capture the full complexity of the
economy. Another problem is
that, so far, no single model adequately explains the entire economy—at
least, you can't get economists to agree on a single model; and no
single model outperforms others in predicting future developments in
Another problem is that the forecast can be off base because of unexpected,
even unprecedented, developments—the September 11 attacks are a case
in point. So in practice, the Fed tries to deal with this uncertainty
by using a variety of models and indicators, as well as informal methods,
to construct a picture of the economy. These informal methods can include
anecdotes and other information collected from all kinds of sources,
such as the Directors of the Federal Reserve Banks, the Fed's various
advisory bodies, and the press.
So now are we in a position to compare the Fed's estimates with its
Not so fast. Coming up with operational measures of the goals
than you might think, especially the goal for the rate of maximum sustainable
output growth. Unfortunately, this is not something you can go out and
measure. So, once again, the Fed has to turn to some sort of model or
indicator to estimate it. And it's hard to be certain about any estimate,
in part because it's hard to be certain that the model or indicator the
estimate is based on is the right one. There's one more important complication
in estimating the rate of maximum sustainable growth—it can shift over
What problems does a shift in the rate of maximum sustainable growth
The experience of the late 1990s provides a good example of the
policy problems caused by such a shift. During this period, output
surged at the same time that rapid innovation was transforming the
information technology industry. In the early stages, there was no way
for the Fed—or
anybody else—to tell why output was growing so fast. In other
words, the Fed had to determine how much of the surge in output was
unusually rapid technical progress and whether this implied an increase
economy's trend growth rate.
This was a crucial issue because policy would respond differently depending
on exactly why the economy was growing faster. If it was largely due
to the spread of new technologies that enhanced worker and capital productivity,
implying that the trend growth rate was higher, then the economy could
expand faster without creating inflationary pressures. In that case,
monetary policy could stand pat. But if it was just the economy experiencing
a more normal business cycle expansion, then inflation could heat up.
In that case, monetary policy would need to tighten up.
The Fed's job was complicated by the fact that statistical models did
not find sufficient evidence to suggest a change in the trend growth
rate. But the Fed looked at a variety of indicators, such as the profit
data from firms, as well as at informal evidence, such as anecdotes,
to conclude that the majority of the evidence was consistent with an
increase in the trend growth rate. On that basis, the Fed refrained from
tightening policy as much as it would have otherwise.
Does the trend growth rate ever fall?
Yes, it does. A good example, with a pretty bad outcome, was what happened
in the early 1970s, a period marked by a significant slowdown in the
trend growth rate. A number of economists have argued that the difficulty
in determining that such a slowdown had actually taken place caused the
Fed to adopt an easier monetary policy than it might otherwise have,
which in turn contributed to the substantial acceleration in inflation
observed later in the decade.
What happens when the estimates for growth and inflation are different
from the Fed's goals?
Let's take the case where the forecast is that growth will be
below the goal. That would suggest a need to ease policy. But that's not
Fed also must decide two other things: (1) how strongly to respond to
this deviation from the goal and (2) how quickly to try to eliminate
the gap. Once again, it can use its models to try to determine the effects
of various policy actions. And, once again, the Fed must deal with the
problems associated with uncertainty as well as with the measurement
problems we have already discussed.
Uncertainty seems to be a problem at every
stage. How does the Fed deal with it?
Uncertainty does, indeed, pervade every part of the monetary policymaking
process. There is as yet no set of policies and procedures that policymakers
can use to deal with all the situations that may arise. Instead,
policymakers must decide how to proceed by going case by case.
For instance, when policymakers are more uncertain about their reading
of the current state of the economy, they may react more gradually to
economic developments than they would otherwise. And because it's hard
to come up with unambiguous benchmarks for the economy's performance,
the Fed may look at more than one kind of benchmark. For instance, because
it's hard to get a precise estimate of the trend growth rate of output,
the Fed may look at the labor market to try to figure out where the unemployment
rate is relative to some kind of benchmark or "natural rate," that
is, the rate that would be consistent with price stability. Alternatively,
it might try to determine whether the stance of policy is appropriate
by comparing the real funds rate to an estimate of the "equilibrium
interest rate," which can be defined as the real rate that would
be consistent with maximum sustainable output in the long run.
These issues are far from settled. Indeed the Fed spends a great deal
of time and effort in researching various ways to deal with different
kinds of uncertainty and in trying to figure out what kind of model
or indicator is likely to perform best in a given situation. Since
issues aren't likely to be resolved anytime soon, the Fed is likely
to continue to look at everything.