View original document

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

Is Inflation Too Low?
16th Annual Monetary Conference
Cato Institute
Washington, D.C.
October 22, 1998
Published in the Federal Reserve Bank of St. Louis Review, July/August 1999, 81(4), pp. 3-10


hat is today’s big monetary policy
issue? It is, surely, the extraordinary volatility of the financial
markets and the wide quality
spreads that opened up between riskier bonds
and Treasury bonds following the Russian default
in mid-August 1998. No one forecast these problems; the financial-market upset certainly was
not a real, live policy issue back in the spring
and early summer.
We should not underestimate the magnitude
of the current disturbance in the U.S. financial
system. Monetary policy today is, I believe, appropriately focused on dealing with the possible
effects of the financial-market disturbance on the
U.S. economy. The size of that disturbance and
the circumstances surrounding it are so unusual
in the context of U.S. history that policymakers
must concentrate on dealing with this situation
for the time being.
The financial upset, however, will disappear
from the radar screen of pressing policy issues as
the markets settle down in due time. All of us
will then return—or should return—to analyzing
longer-run issues. With regard to the current outlook, I will say only that I am optimistic that we
will work through current problems, painful as
they have been for many, with no significant damage to the U.S. economy. My optimism stems from
the economy’s strong initial conditions of low
inflation, low and stable inflation expectations,
and a well-capitalized banking system. These are

about as favorable a set of initial conditions as
one can imagine for getting through financial
turmoil with minimal effect on the real side of
the economy.
The issue I wish to explore is this: Is zero
inflation, abstracting from measurement error in
the broad price indexes, too low? I think zero is
a very nice number, especially when it comes to
inflation. But there is a serious argument that the
economy is likely to work better with a moderate
inflation of, say, 2 or 3 percent per year. I disagree
with that argument.
I will concentrate on two arguments for moderate inflation. The first argument holds that inflation facilitates the smooth operation of labor
markets and thereby promotes maximum employment in the face of nominal wage rigidity. The
second argument contends that inflation, via the
Fisher relationship, keeps nominal interest rates
from falling too close to the zero bound, and
thereby gives the Fed sufficient room to ease—
that is, to cut rates—should a recession appear
In my view, both arguments are wrong. I will
begin by outlining some reasons why I believe
that zero inflation should be the paramount objective of monetary policy.

As Chairman Alan Greenspan has pointed out
on numerous occasions, our economy’s fine per-

This speech is reprinted with the permission of the Cato Institute. It was previously published in the Cato Journal, Winter 1999, 18(3).
© 1999, The Cato Institute.



formance since the early 1990s was accompanied
initially by declining inflation and, more recently,
by low and stable inflation. Clearly, the U.S. experience of the last five years casts doubt on the old
claim that falling inflation will inevitably bring
slower real growth or a higher rate of unemployment. This experience also suggests that reducing
the variability of inflation need not increase the
variability of output, as some people argue.
Although the performance of other countries
with low inflation is somewhat mixed, my point
is simply that there is little evidence to suggest
that zero inflation necessarily implies slow real
growth. Indeed, Robert Barro (1996) and others
have reported systematic evidence to the contrary.
Certainly, there are good reasons to expect that a
zero-inflation monetary regime, sustained over
the long run, would enhance an economy’s
If the monetary authority is committed credibly to zero inflation, then one source of interference with the efficient working of markets—
uncertainty about expected inflation—would be
reduced. Inflation uncertainty makes it difficult
for individuals and firms to distinguish changes
in relative prices among goods and services from
movement in the aggregate price level. Mistakes
in the allocation of resources are more likely to
occur because of this uncertainty, with real growth
consequently less than it could be.
By confusing the meaning of individual price
changes, inflation uncertainty also raises uncertainty about the prospects of investment returns.
A rising rate of inflation can lead both borrowers
and lenders to be overly optimistic about likely
returns, resulting in inefficient resource commitment. If the price expectations that are assumed
when funds are committed are not realized, borrowers may encounter difficulty repaying their
debts, which in turn puts stress on lenders. Thus,
it is reasonable to expect that eliminating uncertainty about the rate of inflation will enhance,
although obviously not guarantee, financial
Presumably, to eliminate uncertainty, the rate
of inflation need not be zero, but simply predictable. For at least two reasons, however, I believe

zero should be the target. First, maintaining a
steady but positive inflation rate probably would
be harder politically than maintaining a steady
zero inflation. The reason is that we live in a world
where both politicians and economists often argue
that just a little more inflation would generate
positive real economic gains. If we accept the
argument that 2 percent inflation is okay, why
not 2.5 percent? Let me emphasize that when I
advocate zero inflation, I am ignoring measurement questions, such as whether or not bias exists
in the relevant price index. As a practical matter,
policy is probably best specified in terms of a
measured inflation range that accounts for our
best estimate of measurement errors.
A second reason I advocate zero inflation
concerns the distortions caused by the interaction
of inflation with the tax code. Inflation indexing
is incomplete, especially for investment income,
because nominal interest income and nominal
capital gains are subject to tax. Martin Feldstein
(1997) has estimated that reducing inflation from
its current level of about 2 percent to zero would
yield substantial, permanent real income gains.
Theoretical analysis by James Bullard and Steven
Russell (1998), and others, also suggests that tax
distortions cost the economy substantial real
performance at higher rates of inflation.
In short, I think the case is strong that monetary policy should aim for zero inflation as its paramount objective. Moreover, I reject the approach
that zero inflation must be shown to be superior
to a poorly specified alternative of some positive
inflation. The burden of proof really should fall
on those who contend that positive inflation is
better. So, let me now consider the arguments
advanced for a positive rate of inflation.

One perennial argument in favor of positive
inflation is that certain wages must fall relative to
other prices or other wages, and inflation allows
this adjustment of real wages to occur in the face
of nominal wage rigidity. The centerpiece of this
argument is the claim that downward nominal

Is Inflation Too Low?

wage adjustments occur too infrequently to be
consistent with flexible real wages in a world
where microeconomic shocks continuously alter
the relative positions of particular firms, industries, or occupations. With zero inflation, the argument goes, rigid nominal wages prevent optimal
adjustment to relative price disturbances with
the result that employment varies inefficiently.
Therefore, a little inflation is a good thing because
it allows wages to fall relative to other prices;
inflation “greases the wheels” of labor-market

Zero Inflation in a Different Regime
There are, in my opinion, serious flaws at
three levels of this argument. First, the argument
claims that nominal rigidity creates a large inefficiency that inflation ameliorates. But, if the claim
of a large inefficiency is true (and I will question
it later), a simple theoretical argument creates
the presumption that nominal wages would not
continue to be sticky in a zero-inflation regime.
There is some dispute about the extent to
which nominal wages are downwardly rigid. But,
no doubt some employers have found it difficult
to reduce nominal wages during the periods covered by the most popular data sources. One data
source, the Panel Study of Income Dynamics,
started during the late 1960s. I mention the sample
period because making an empirical regularity
the foundation, rather than an implication, of
economic theory always is dangerous. Robert
Lucas (1976) elegantly demonstrated this point
more than 20 years ago. To the extent that downward nominal wage rigidity exists, it presumably
serves some economic function. After all, putting
minimum wage laws aside, fixed nominal wages
are not required by law. We cannot assume that
the present degree of wage rigidity—whatever it
is—would continue into a different inflation
regime. Indeed, a compelling case can be made
that the extent of wage rigidity we observe would
not survive in a zero-inflation regime.
Consider an environment where, broadly
speaking, the annual changes in broad price
indexes usually are close to zero and have been

for some time. Suppose, also, that the Fed’s commitment to maintaining this regime is clear. In
such an environment, nominal wage rigidity,
according to the grease-the-wheels argument,
would generate a large inefficiency that inflation—
now zero—would no longer ameliorate. This
inefficiency, however, is exactly what should
make us doubt that nominal wage rigidity would
continue to exist. The main function of the price
system is to allocate resources by setting relative
prices. Competitive forces likely would eliminate
anything that interferes with relative price adjustment, particularly if failure to adjust is very costly,
unless there is some compelling reason for it to
exist. Could we imagine that nominal wage rigidity would continue during a sustained 10 percent
deflation? Of course not. Why? The private costs
of interfering with relative price adjustment would
be too high. It may take longer for competitive
forces to erode nominal rigidity under zero inflation, but the principle is the same.
Keep in mind that the magnitude of ongoing
resource reallocation in U.S. labor markets dwarfs
the employment growth that makes headlines on
the first Friday of every month. Jobs appear, jobs
disappear, and people move into and out of them
at rates far higher than net employment growth.
This is prima facie evidence that U.S. labor markets do not suffer from any massive inefficiency.
If nominal wage rigidity creates significant
economic inefficiency, it seems entirely plausible
that it is perpetuated by inflation. I admit I do not
know for sure. Based on the current state of economic theory, however, I think we must favor the
presumption that inefficient wage rigidity would
disappear in a zero-inflation economy. This position makes sense if we take economic theory

Other Mechanisms for Relative Wage
A second flaw in the grease-the-wheels argument is that it imagines only two mechanisms for
achieving adjustments to a worker’s relative wage:
Either cut the nominal wage, or let all other prices
around it rise. In fact, the workings of labor mar3


kets suggest at least two other mechanisms, and
so the presence of nominal wage rigidity— were
it to exist—might not be a hindrance in a zeroinflation world.
First, average compensation tends to rise over
time, as overall productivity improves. Thus, in
a zero-inflation environment, nominal wages may
not need to fall, even in some declining occupations. Proponents of the grease-the-wheels view
sometimes ignore this mechanism.
Internal labor markets provide yet another
adjustment mechanism. Compensation tends to
increase with seniority, partly because of an
individual’s accumulation of human capital.
Edward Lazear (1981) has argued that an upwardsloping path for earnings also acts as a mechanism
to overcome agency problems within the firm.
James Malcomson (1984) and others have argued
that promotions may play a similar role; rather
than simply filling positions necessary for the
technological operation of the firm, promotions
provide necessary incentives for those at lower
levels of the hierarchy.
The common theme in these observations
about internal labor markets is that an individual
worker typically will expect an increasing real
wage. Therefore, the kind of base adjustment
achieved by inflation can also be accomplished
by delaying wage change relative to an individual’s
upward-sloping real wage path.
Of course, there is a segment of the labor
market where little human capital accumulation
exists and long-term implicit contracts are rare.
But, for obvious reasons, this is exactly the segment where turnover costs are low on both the
supply and demand sides of the market. Hence,
any nominal wage rigidity that is present is not
especially costly.

The third flaw in the labor-market case for
positive inflation is perhaps the most transparent.
Inflation tends to increase the sort of microeconomic shocks—because cross-sectional variation

in price changes tends to rise with higher aggregate
inflation—that underlay the case for pursuing a
positive rate of inflation. Thus, the claim that inflation helps the economy cope efficiently with relative price changes is suspect immediately, since
there is more relative price variation to cope with
if there is more inflation.

Labor Market Costs as Well as Benefits
Overall, I believe that the benefits of inflation
as labor-market grease are exaggerated. Furthermore, inflation itself seems to worsen the problem
it ostensibly alleviates. In addition to these theoretical arguments, we now have some direct
evidence, supplied by Erica Groshen and Mark
Schweitzer (1996, 1999). They recognize that
compensation typically is set for at least a year,
and that there are, in essence, two pieces to a
firm’s wage-setting process. First, management
decides on the overall change in the wage pool,
based in part on the rate of inflation expected to
prevail during the following year. This wage pool,
in effect, sets the firm-wide budget constraint.
Second, individual wages and salaries are adjusted
in a way that satisfies the budget constraint. This
two-step process is explicit in many organizations.
Mistakes occur during the first stage when
managers misforecast inflation. “Sand-in-thewheels” effects occur if higher average levels of
inflation result in more inflation variability, causing larger inflation forecasting errors. A consequence is that inflation will cause more interfirm
variation in wage adjustment. Grease effects operate, as I outlined earlier, and imply more dispersion of interoccupational wage adjustment. The
grease effects should taper off as inflation rises
because some level of inflation enables employers
to make all of the relative wage adjustments they
would make in a frictionless labor market. Because
they view wage setting as a two-stage process,
Groshen and Schweitzer estimate the grease and
sand effects separately. They find evidence of
both effects, with sand effects rising rapidly with
the inflation rate. Comparing the grease and sand
effects directly, Groshen and Schweitzer find that
even for low inflation rates the net benefit of infla-

Is Inflation Too Low?

tion is statistically indistinguishable from zero,
although point estimates of the gross benefit do
slightly exceed estimates of gross cost.
One might quibble with the specifics of their
empirical strategy, but Groshen and Schweitzer’s
emphasis on evaluating costs as well as benefits
is absolutely correct. From the standpoint of labor
markets, I think it is fair to say that the evidence
of net benefits from an inflationary monetary
policy is slim to none.

Now let us consider whether concerns about
conducting countercyclical monetary policy in a
low-inflation environment can justify a positive
rate of inflation. Specifically, does price level
stability cause special problems for monetary
policy because nominal interest rates cannot be
less than zero?
The zero-bound view is an old and much
debated one in macroeconomics. With rising inflation during the 1970s and early 1980s, the issue
largely became moot, as policymakers scrambled
to get inflation back under control and to regain
lost credibility. Recently, however, the topic has
resurfaced as inflation rates in the industrialized
countries have fallen and stayed low during the
1990s, and as central banks around the world
have adopted inflation targeting as a method of
achieving and committing to price stability.
The zero-bound view holds that moderate
inflation aids in the implementation of stabilization policy by keeping nominal interest rates from
falling too low. The bottom line, according to this
argument, is that an inflation target of zero interferes with the attempts of monetary policymakers
to stimulate an economy in recession because the
nominal interest rate obviously cannot fall below
zero. Put another way, with moderate ongoing
inflation the policymakers have room to push
the real rate of interest below zero, which they
cannot do when the steady inflation rate is zero.
The zero-bound story begins with the commonplace idea that monetary policy is concerned

with setting a short-term nominal interest rate—
in the United States, the nominal federal funds
rate. A higher nominal federal funds rate is often
described as a tighter policy, while a lower nominal federal funds rate is described as an easier
policy. When the economy is weak, the monetary
authorities lower the nominal federal funds target
in an effort to stimulate interest-rate-sensitive
sectors of the economy. So according to this view,
when a recession hits, the current level of the
federal funds rate determines the number of
basis points the Fed has available to combat the
recession: the lower the initial funds rate, the
less scope for subsequent easing. As you might
guess, I dislike this characterization of monetary
policy, but let me finish the story.
Of course, financial market participants are
interested mainly in the real interest rate, not the
nominal interest rate. A simple Fisherian decomposition divides any nominal interest rate (with
zero default risk) into two major components—a
real component determined by equilibrium conditions in the economy and a nominal component
determined by the expected inflation rate.
The zero-bound view holds that the expected
inflation component of nominal interest rates
moves little over periods as long as a year, so that
adjustments in the nominal federal funds rate
mainly change real returns at the very short end
of the term structure. Movements in short-term
rates then lead to adjustments in longer-term real
interest rates.
What hampers stabilization policy in a lowinflation environment, according to the zerobound view? If inflation is zero and expected to
remain that way, then the expected inflation component of nominal interest rates is zero, and the
nominal rate is lower on average than it would
be in a world of persistent inflation. Thus, in a
recession, the Fed would have less room to cut
interest rates because of the zero nominal bound.
The end result, according to this view, is a longer
and deeper recession than would otherwise be
the case. The message is clear: If the Fed is to
help the economy in times of distress, nominal
interest rates must be kept high enough in normal


times, which requires maintaining a modest rate
of inflation.
The zero-bound view has raised many counterarguments over the years. Perhaps most obviously,
this view places heavy emphasis on the idea that
monetary policy can be used to fine tune the
macroeconomy, downplaying well-known concerns that attempts to fine tune can contribute to
economic instability. Leaving that issue aside,
however, there are still several reasons to doubt
the validity of the zero-bound argument for pursuing a policy of positive inflation.

Monetary Policy Is Fundamentally Not
About Nominal Interest Rates
First, we must remember that nominal interest
rates do not indicate the true stance of monetary
policy even though, as a practical matter, the Fed
implements short-term policy by targeting the
nominal federal funds rate. This method of implementation has been effective in recent years.
Controlling the funds rate is not, however, an
end in itself. Fundamentally, monetary policy is
reflected in the growth of the money stock and,
ultimately, the rate of inflation. So the idea that
central bankers are somehow trapped if the nominal short-term interest rate nears zero seems quite
a stretch to me. We are in the business of providing
liquidity to the macroeconomy, and if the situation calls for it, liquidity can always be injected,
regardless of the level of nominal interest rates.
The first years of the Great Depression offer
perhaps the clearest illustration that monetary
policy is fundamentally about providing liquidity
and not about controlling nominal interest rates.
During that time, nominal interest rates were
low, which seemed to indicate an “easy” monetary policy. But as Milton Friedman and Anna
Schwartz (1963) have noted, from 1930 to 1933
the money stock was falling rapidly, indicating a
far tighter policy than was intended. Of course,
that policy was an unmitigated disaster, as both
output and prices fell by a third and the unemployment rate hit 25 percent. That experience, as
well as other, less dramatic historical episodes,
should make it obvious that blind adherence to

nominal interest rates as indicators of the stance
of monetary policy can be tragically misleading.
We might also do well to remember that during the late 1950s and early 1960s, the nominal
annualized yield on three-month Treasury bills
fluctuated around 3 percent, while the yield on
10-year Treasury bonds was around 4 percent.
These yields are below, but in the general ballpark,
of those we observe today. Consumer price index
inflation during that period averaged about 2
percent on a year-overyear basis—not too different
from today’s inflation rate. So, while we have
not seen a sustained zero-inflation environment
in the United States during the postwar era, we
have seen an environment not too different from
today’s in terms of relatively low inflation. And
during the late 1950s and early 1960s, there was
no obvious impediment to the operation of monetary policy just because inflation was low.

Inflation and Output Variability
The relative stability of our economy in
recent years suggests that low inflation probably
contributes both to less inflation variability and
to less output variability. Throughout the 1970s
and early 1980s, by contrast, when inflation rose
sharply and then fell abruptly, the United States
suffered through three recessions, including two
of the most severe recessions of the postwar era.
To be sure, during that period, the U.S. economy
was hit with shocks from external sources, but at
the same time monetary policy was decidedly
uneven— resulting in a substantial inflation that
caused both unnecessary distortions and proved
difficult to tame. Thus, the postwar experience
strongly suggests that lower inflation is associated
with less volatile inflation, and lower inflation
volatility is reflected in lower volatility in real
output. Even in a zero-inflation environment the
lower bound on nominal interest rates probably
would not be a problem for stabilization policy
because economic volatility itself would likely
be lower.

Nonlinear Effects Near Zero
A final reason to doubt that monetary policy
would run aground in a zero-inflation world stems

Is Inflation Too Low?

from the nonlinearity of investment demand.
This nonlinearity implies that a given interest
rate change, measured in basis points, may well
have a larger impact when interest rates are low
than when they are high.
Much of the thinking behind the zero-bound
view is centered on the extrapolation of linear
effects to very low interest rate environments. In
much of the work on this issue, the average effects
of short-run monetary policy changes are estimated using postwar data, which include many
years of high inflation. There is little reason to
think that coefficients estimated from an environment of relatively high inflation would be
good proxies for the coefficients in the new
We might expect that a given basis points
change in the nominal federal funds rate target
would have a larger impact when interest rates
are lower. Certainly, there is no reason to expect
that the response of investment to changes in the
nominal interest rate is linear. At any point in time
countless investment projects are available, and
as the nominal cost of funds moves lower, the
net present value of many more of these projects
becomes positive. Accordingly, investment would
be unbounded at very low real interest rates,
implying that the Fed could conduct a countercyclical policy just as actively and effectively
when interest rates were low, even if the nominal
federal funds rate target was near zero.

To summarize, it seems to me that neither
the arguments about wage stickiness nor those
concerning the zero bound for nominal interest
rates make a convincing case that monetary policy
should aim for a positive rate of inflation. Instead,
I believe the logic and the evidence both suggest
that the appropriate goal for monetary policy
should be price stability, that is, a long-run inflation rate of approximately zero.

Today we are enjoying the benefits of a low
and comparatively stable rate of inflation. In our
present state, we should not forget the high costs
of inflation. Inflation makes planning difficult
for individuals and firms, it interferes with the
operation of markets, and it interacts insidiously
with the tax code to discourage saving and investment. Moreover, inflation’s effects are felt most
acutely by members of society who are economically the most vulnerable. In arguing for a positive
rate of inflation, therefore, the burden of proof
should rest with those who contend that our
economy would perform better with inflation
than without it. Inflation proponents also should
explain how a moderate rate of inflation could
be maintained without inching ever higher. In
my view, the case for positive inflation has not
been proven.
A central bank’s single most important job is
preserving the value of the nation’s money. Monetary policy has succeeded if the public can reasonably trust that a dollar will buy tomorrow what it
will buy today. At this point, inflation will have
ceased to be a hindrance to the smooth functioning
of our market economy. I cannot promise that
price stability will mean an end to the business
cycle, to unemployment, or to occasional financial
distress. Indeed, I am willing to bet that a few
years from now we will look back on 1998 and
conclude that the stability of the inflation environment was important to containing the financialmarket upset that started in August.
We should not be seduced by arguments that
a little inflation is a good thing. Look at the record:
Over the past 50 years, our economy has performed better when inflation was low than when
it was high. There simply is no compelling evidence that we could foster sustained economic
growth by pursuing an inflationary monetary
policy. The evidence points in the other direction.
Thus, I am confident that our economy’s long-run
performance would be enhanced by a monetary
policy that aims at, achieves, and maintains a
zero rate of inflation.


I owe a great debt to the St. Louis Fed research
department, and especially to David Wheelock,
Jim Bullard and Joe Ritter, for assistance in preparing these remarks.

Barro, Robert J. “Inflation and Growth.” Federal
Reserve Bank of St. Louis Review, May/June 1996,
pp. 153-69.
Bullard, James B. and Russell, Steven H. “How Costly
Is Sustained Low Inflation for the U.S. Economy?”
Federal Reserve Bank of St. Louis Working Paper
97-012B, March 1998.
Feldstein, Martin. “Capital Income Taxes and the
Benefits of Price Stability,” NBER Working Paper
6200, September 1997.
Friedman, Milton and Schwartz, Anna Jacobson. A
Monetary History of the United States, 1867-1960.
Princeton University Press, 1963.


Groshen, Erica L. and Schweitzer, Mark E. “The
Effects of Inflation on Wage Adjustments in FirmLevel Data: Grease or Sand?” Federal Reserve Bank
of New York Staff Report No. 9, January 1996.
Groshen, Erica L. and Schweitzer, Mark E.
“Identifying Inflation’s Grease and Sand Effects in
the Labor Market,” in Martin Feldstein, ed., The
Costs and Benefits of Price Stability. University of
Chicago Press, 1999, pp. 273-308.
Lazear, Edward P. “Agency, Earnings Profiles,
Productivity, and Hours Restrictions.” American
Economic Review, September 1981, pp. 606-20.
Lucas, Robert E., Jr. “Econometric Policy Evaluation:
A Critique,” in Karl Brunner and Allan H. Meltzer,
eds., The Phillips Curve and Labor Markets.
Carnegie-Rochester Conference Series on Public
Policy, North Holland, 1976, pp. 19-46.
Malcomson, James M. “Work Incentives, Hierarchy,
and Internal Labor Markets,” Journal of Political
Economy, June 1984, pp. 486-507.