FRBSF Weekly Letter
96-03; January 19, 1996
The rate of wage growth has moderated recently, leading some observers
to predict that inflation will continue to remain subdued in the near
future. In this Letter we discuss what economic theory tells us about
the relationship between changes in wages and changes in prices and also
look at some of the empirical evidence on this issue, in order to determine
whether the recent slowdown in wage growth does, in fact, tell us something
about future inflation that we might otherwise miss.
In a traditional Keynesian model, changes in wages typically are assumed
to precede changes in prices. Wages are determined in the labor market.
Specifically, they depend upon the unemployment rate; an increase in unemployment,
for example, reduces the rate of growth of wages. Once wages are determined,
firms set prices by adding a markup which determines their profit margin.
In such a model, when wages increase slowly relative to productivity,
employers face lower costs to produce the same number of goods. They pass
on the lower costs in the form of lower prices.
If this model is a reasonable description of the economy, it is quite
possible that the recent slowdown in wage growth will be followed by a
slowing of prices. To see how recent monetary policy may have played a
role, consider how a tightening of policy would work in such a model.
Suppose the monetary authority decided to reduce inflation, and raised
interest rates to do so. Consumers and firms would react to higher interest
rates by reducing the amount of money they spend, and sales would drop.
Faced with an unwanted increase in inventories, firms would cut back production
and employment. This would push the unemployment rate up, and lead to
a reduction in the rate of wage increases. In response, firms would reduce
the rate at which they were raising prices. Once the desired reduction
in inflation had been achieved, the monetary authority could reduce interest
rates, allowing economy-wide spending to go up again. Thus, the recent
slowing of wages could have been caused by a tightening of monetary policy,
and according to this model, would be followed by a reduction in the rate
of inflation.
This sequence of events is in stark contrast to Friedman's (1968) classic
description of the effects of monetary policy on the economy. We follow
his example, which assumes that the monetary authority wishes to reduce
unemployment. To do so, it increases the rate of money growth, causing
interest rates to fall temporarily. Spending goes up in response. If prices
have been stable in the past, people will have set wages and prices on
the assumption that they will continue to be stable in the future. Consequently,
output and employment will expand in response to the higher spending.
However, the positive response of output and employment occurs ...because
selling prices of products typically respond faster than prices of factors
of production.... The fact that firms are quick to adjust their prices
in response to the increase in spending while workers are slower in demanding
higher wages makes it profitable for firms to raise output. Soon, however,
workers catch on to the fact that prices are rising, and they demand increases
in wages to compensate for the higher prices. Ultimately, wages adjust
fully to the rise in prices. This causes unemployment to rise to its previous
level, as labor no longer comes at a discount.
This view of the economy suggests that to the extent that the recent
slowdown in wages is related to monetary policy, it reflects an earlier
slowdown in inflation, and provides no information about future inflation.
Thus, it is possible to find models in which changes in wages lead to
changes in prices as well as models in which changes in prices lead to
changes in wages.
Nor are we restricted to choosing one over the other. Gordon (1988) discusses
a model in which both variables affect each other. In his model, both
firms and workers take expected inflation into account when trying to
determine the appropriate level of wages. To the extent that workers and
firms look at recent inflation rates when trying to form estimates of
future inflation, the wage rate at any point in time will be affected
by past inflation. Gordon also assumes that prices are determined as a
markup over labor costs (in addition to some other factors that are incidental
to the issues at hand). Thus, in his model changes in wages lead to changes
in prices and changes in prices lead to changes in wages.
Economic theory, then, is consistent with either type of causation between
wages and prices and does not rule out any prediction about the behavior
of prices in response to the recent deceleration in wage growth.
While there is a long history of empirical studies on the relationship
between wages and prices, we will focus on recent studies that have used
post-war U.S. data. Generally speaking, these studies differ in the variables
they use to measure wages and prices and the techniques they employ.
Shannon and Wallace (1985) use unit labor costs as a measure of wages.
Unit labor costs literally measure the labor cost of producing a single
unit of output. Changes in unit labor costs can be the result of different
events; a decrease in unit labor costs, for example, could reflect either
a decrease in wages or an increase in output per hour of labor (that is,
of labor productivity). Shannon and Wallace justify the use of unit labor
costs by pointing out that wage increases that reflect increases in productivity
will not lead to higher inflation. They find that when the GNP deflator
is used to measure inflation, changes in unit labor costs lead to changes
in inflation, while the evidence for the reverse causation is somewhat
weaker. However, when the Consumer Price Index (CPI) is used to measure
inflation, there is weak evidence that price changes lead to wage changes,
but no evidence that wage changes lead to price changes.
Gordon (1988) uses the change in the GNP deflator to measure inflation
and an index of average hourly earnings to measure wages. Wages are adjusted
for changes in the trend growth of productivity (instead of actual productivity
as in the unit labor cost variable). Gordon estimates separate equations
for inflation and wage growth, explaining each variable in terms of past
inflation and wage growth, as well as variables that control for the stage
of the business cycle and shocks, such as wage and price controls. He
finds that while previous values of inflation and an indicator of the
stage of the business cycle help predict inflation, wage growth does not.
His results also suggest that price changes do not provide any information
about wage changes, though here the evidence is weaker in that an alternative
version of his tests shows that price changes do help predict wage changes.
A significant problem with the tests that we have described so far is
that they focus on the short-run relationship between prices and wages,
but do not take into account any long-run relationship between these variables.
Recent advances in econometric techniques allow one to do just that. Mehra
(1991) employs these techniques in a framework that is similar to Gordon
s. Using unit labor costs as a measure of wages, he finds that prices
and wages move together over the long run, and that this relationship
results from the fact that changes in prices lead to long-run changes
in wages. Changes in wages do not have any effect on prices. However,
in later work Mehra finds that this conclusion is sensitive to the price
index he uses; specifically, while this result is confirmed for the GDP
deflator, he finds that wages and prices affect each other when the CPI
is used to measure inflation.
Another potential problem with these studies has to do with the use of
unit labor costs to measure wages. It turns out that changes in unit labor
costs caused by changes in productivity could lead to changes in inflation
even if changes in nominal wages provided no information about changes
in future inflation. Assume that there is an increase in worker productivity,
so that output goes up. If the monetary authority does not increase the
rate of money growth in response, firms must lower prices to sell the
higher output (or slow down the rate of price increases, depending upon
the prevailing rate of inflation). Thus, higher productivity leads to
lower prices, with no change in wages. Since unit labor costs have fallen
as well, however, they will be seen as leading to lower inflation, and
a researcher using unit labor costs to measure wages will conclude that
wage changes lead to price changes, even if firms did not move wages at
all.
These considerations suggest that in examining the relationship between
wages and prices it may be useful to isolate the effects of changes in
productivity. Such an exercise is carried out by Huh and Trehan (1995).
They focus on the nonfarm business sector, using total compensation per
hour as a measure of wages, the implicit price deflator as a measure of
prices, and output per hour as a measure of productivity. Employing techniques
that allow them to take long-run relationships into account, they find
that changes in wages that are not related to changes in productivity
do not provide any information about future inflation. By contrast, changes
in inflation do help to predict future wage growth. Finally, they also
find some evidence which suggests that increases in productivity have
little effect on the wage rate but do lead to lower inflation, consistent
with the discussion above.
Theoretical models suggest that causality between wages and prices could
go in either direction. While empirical evidence on the issue is mixed,
tests that allow for a long-run relationship between wages and prices
tend to find that price changes predict future wages, while wage changes
do not have much effect on prices. In addition, tests that find that wages
cause prices generally use unit labor costs as a measure of wages, which
makes the results difficult to interpret because changes in wages can
get confounded with changes in productivity. Huh and Trehan (1995) explicitly
controls for productivity changes and does not find any effect of wage
changes on prices. While we are wary of claiming that this last result
is conclusive (for instance, it would be useful to verify this result
using different measures of inflation and wage growth), we see the evidence
as favoring the view that changes in wages do not provide much information
about future changes in inflation, especially once we take into account
certain commonly watched indicators.
The last caveat is important; in general, the results we have surveyed
do not rule out an increase in inflation following an increase in wage
growth. For instance, if wages react more quickly than prices do to cyclical
conditions in the economy, wages still could be seen to lead prices when
the two series were compared to each other. Thus, wage growth could still
act as an indicator of inflation. The evidence does suggest that it is
not a very useful indicator, however, since once we take into account
certain other commonly watched indicators such as the price of oil and
the stage of the business cycle, wage growth does not appear to provide
any additional information about future inflation. Consequently, we are
skeptical about claims that the recent slowdown in wage growth provides
significant, independent evidence about slower inflation in the near future.
Benjamin Bridgman
Research Associate
Bharat Trehan
Research Officer
Friedman, Milton. 1968. The Role of Monetary Policy. American Economic
Review, pp. 1-17.
Gordon, Robert J. 1988. The Role of Wages in the Inflation Process. American
Economic Review, pp. 276-283.
Huh, Chan, and Bharat Trehan. 1995. Modeling the Time-Series Behavior
of the Aggregate Wage Rate. FRBSF Economic Review, pp. 3-13.
Mehra, Yash.1991. Wage Growth and the Inflation Process: An Empirical
Note. American Economic Review, pp. 931-937.
Shannon, Russell, and Myles S. Wallace. 1985. Wages and Inflation: An
Investigation into Causality. Journal of Post Keynesian Economics, pp.
182-191.
Opinions expressed in this newsletter do not necessarily reflect
the views of the management of the Federal Reserve Bank of San Francisco,
or of the Board of Governors of the Federal Reserve System. Editorial
comments may be addressed to the editor or to the author. Mail comments
to
Research Department
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120
|
