View metadata, citation and similar papers at core.ac.

uk brought to you by CORE

provided by Research Papers in Economics

How Useful Are Leading Indicators

of Inflation?

By C. Alan Garner

any economists expect inflation to rise in How useful are these leading indicators for

M 1995. These expectations are based on

various approaches to forecasting infla-
tion. One approach is based on the standard eco-
forecasting inflation? This article examines five
widely watched leading indicators. The first section
evaluates the strengths and weaknesses of these
nomic theory that inflation rises when slack is indicators based on economic theory. The second
eliminated from the economy and production ex- section evaluates the leading indicators empirically,
ceeds capacity constraints. According to this view, looking at how the indicators have performed by
measures of economic slack such as unemployment themselves and whether the indicators add useful
and capacity utilization provide useful information information to a standard model relating inflation
about the inflation outlook. But the relationship to economic slack. It is concluded that, of the five
between slack and inflation is complicated and leading indicators, the composite indicators have
subject to variable lags. given the most useful early warning signals of
Uncomfortable with this complex relationship, inflation turning points, but none of the indicators
some analysts rely on alternative approaches to has recently been successful in predicting inflation
forecasting inflation. One approach is based on magnitudes.
“leading indicators” of inflation. The leading indi-
cators typically incorporate information on selected
prices to augment or replace information on eco- FIVE LEADING INDICATORS OF
nomic slack. The prices selected are usually key INFLATION
commodity prices that fluctuate more or less con-
tinuously in response to changing economic condi- Five leading indicators of inflation are de-
tions. Prominent leading indicators of inflation scribed in this section. The first is the price of gold,
include the price of gold, broader indexes of com- a commodity that once played an important role in
modity prices, and composite indicators that com- the world monetary system and is still held as a store
bine several economic series believed to predict the of value by investors in many countries. The next
inflation rate. two indicators are the Commodity Research Bureau
(CRB) index of commodity futures prices and the
C. Alan Garner is a senior economist at the Federal Reserve Journal of Commerce (JOC) index of industrial
Bank of Kansas City. David Hensley, a research associate at materials prices. These leading indicators are dif-
the bank, helped prepare the article. fering broad-based baskets of commodities that
6 FEDERAL RESERVE BANK OF KANSAS CITY

play a more important role than gold in current reasons. The price of gold might give a false signal
economic activity. The last two indicators are the of inflation by increasing in response to expecta-
Center for International Business Cycle Research tions of higher inflation based on erroneous infor-
(CIBCR) leading inflation index and the PaineWeb- mation, such as an inaccurate preliminary estimate
ber (PW) leading index. These indexes are composite of the economic growth rate. When investors even-
leading indicators of inflation that combine broad- tually discover the error, inflation expectations and
based commodity indexes with other economic the price of gold would adjust down to their correct
variables believed to be useful in inflation forecasting. levels.1 Moreover, the price of gold might rise rela-
tive to the general price level for reasons that are
specific only to the gold market, such as an increase
The price of gold in the industrial demand for gold.
The price of gold could also be a misleading
The price of gold is viewed by some analysts indicator because its price fluctuates in response to
as a leading indicator of inflation because gold is foreign economic and political factors. Gold essen-
widely held as a store of value. Gold is a store of tially trades in a world market. As a result, an
value partly because of its physical characteristics, increase in the demand for gold might be due to
such as durability and attractiveness, and partly rising inflation expectations abroad rather than in
because of its historical role as the centerpiece of the United States. Foreign political uncertainties
the world monetary system (Laurent). Many coun- also might increase the demand for gold as a store
tries issued gold coins and held stocks of gold of value, causing its market price to rise. For exam-
bullion to fully or partially back their paper curren- ple, some analysts attributed the higher price of gold
cies. Thus, although gold has industrial uses, much in 1993 and 1994 to rising inflation expectations in
of the demand for gold has always been as a store Southeast Asia and political uncertainties in Russia
of value. Moreover, the supply of gold is relatively rather than to increasing U.S. inflation expectations.
fixed because new gold production is small com-
pared with the existing stock of the metal.
Even though gold no longer plays a key role in CRB commodity futures index
the world monetary system, the price of gold might
be a good leading indicator of inflation. The ratio- A diversified commodity index, like the futures
nale is that if enough people regard gold as a good price index prepared by the Commodity Research
store of value, the expectation of rising inflation Bureau, may be a better leading inflation indicator
could cause some investors to shift their funds out than the price of gold. The CRB index reflects the
of financial assets with fixed nominal interest rates prices of futures contracts for 21 commodities
into gold coins or jewelry. Because the gold supply (Table 1). The CRB index is monitored widely by
is relatively fixed, the price of gold might rise financial market participants in part because the
sharply with even a small increase in demand. The index is updated continuously throughout the busi-
general inflation rate, in contrast, tends to rise more ness day as commodity futures prices fluctuate. A
slowly because the prices of many goods and ser- broad index of commodity prices might act as a
vices adjust sluggishly. As a result, an increase in leading indicator of inflation because, like gold,
the price of gold might precede an increase in the these commodities can be held as stores of value
general inflation rate—provided the expectation of when the general inflation rate is expected to rise.
rising inflation was correct in the first place. The prices of these commodities can also adjust
The price of gold is unlikely to be a highly quickly to a change in general inflation expectations
reliable leading indicator of inflation for several because commodity futures contracts trade in highly
ECONOMIC REVIEW • SECOND QUARTER 1995 7

Table 1
Components of the Leading Inflation Indexes

Commodity Research Bureau index Journal of Commerce index

corn, oats, soybeans, soybean meal, wheat, cotton, polyester, burlap, print cloth, steel
soybean oil, cocoa, coffee, sugar, cotton, scrap, zinc, copper scrap, aluminum, tin, hides,
orange juice, lumber, pork bellies, hogs, live rubber, tallow, plywood, red oak, benzene,
cattle, copper, gold, silver, platinum, crude oil, crude petroleum
heating oil

Center for International

Business Cycle Research index PaineWebber index

employment-population ratio NAPM price diffusion

JOC index NAPM vendor performance
growth of domestic nonfinancial debt CRB spot index of industrial prices
import prices unemployment rate
percent of businesses expecting higher selling trade-weighted dollar
prices domestic oil prices
NAPM price diffusion agricultural prices
NAPM vendor performance

efficient auction markets. In contrast, measures of however, leaving movements in the commodity
consumer price inflation adjust more sluggishly. index that more closely reflect changes in general
The diversified CRB index may be a better inflation expectations.
leading indicator than the price of gold for two Although the CRB futures price index may be
reasons. First, the commodities in the CRB index a better leading indicator of inflation than the price
play a more important role than gold in current of gold, many analysts believe the components of
productive activity, meaning a rise in the CRB index the index are not diversified enough to be a highly
is more likely to represent an increase in production reliable leading indicator (Feder). While the index
costs that must ultimately be passed on to consum- is diversified across several commodity groups, 62
ers. Second, a diversified commodity price index percent of the index represents agricultural com-
may be less likely to give false signals of general modities and livestock. The CRB index, therefore,
inflationary pressures because of factors affecting a is not representative of the broad mix of goods and
particular commodity market. Like the price of services purchased by U.S. consumers because the
gold, other commodity prices also fluctuate because index gives too much weight to agricultural prod-
of market-specific disturbances to supply or de- ucts. Moreover, agricultural products sometimes
mand having nothing to do with the overall inflation experience major supply shocks, such as a bad
rate. Such market-specific disturbances may aver- harvest caused by drought or crop disease. As a
age out across a broad basket of commodities, result, the CRB index might give misleading signals
8 FEDERAL RESERVE BANK OF KANSAS CITY

about inflation if agricultural prices were to rise composite index combining several economic vari-
sharply because of a supply shock at a time when ables that tend to move up or down before the
other consumer prices were stable or decreasing. general inflation rate. In addition to commodity
prices, such variables might include measures of
economic slack or variables providing an early
JOC industrial materials price index warning of major supply shocks. Such a composite
index may be less likely to give a false signal of
The Journal of Commerce index of industrial general inflationary pressures because of demand
materials prices might be a better leading inflation or supply disturbances that are specific to the com-
indicator than the CRB index because it is not domi- modity markets.
nated by agricultural prices. The JOC index, which The pioneering work on composite leading in-
is compiled by the Center for International Business dicators of inflation was done by Geoffrey Moore
Cycle Research at Columbia University, tracks the and his associates at Columbia University’s Center
prices of 17 industrial commodities (Table 1). These for International Business Cycle Research. The
commodities, while fewer in number than in the CIBCR leading inflation indicator has recently in-
CRB index, may be less subject to supply shocks, cluded seven components (Table 1).3 The compo-
such as bad harvests, which create volatility in the nents were selected based on their theoretical
CRB index. Industrial materials prices also may be relevance and their historical record in predicting
a better leading indicator if the demand for such cyclical peaks or troughs in consumer price infla-
materials is more closely related to the level of tion. The first three components are the ratio of
current business activity than is the demand for employment to the population, the JOC industrial
agricultural products. materials price index, and the growth rate of domes-
Industrial materials prices, however, could tic nonfinancial debt. These components reflect
sometimes give misleading signals about future “the intensity of demand pressures in the labor
inflationary pressures. For example, a large shift in market, in the commodities markets, and in the
household demand away from consumer services capital markets” (Moore and Kaish). Another com-
toward consumer goods might make the industrial ponent is the change in nonfuel import prices, which
sector substantially stronger than the rest of the reflects prices that are especially sensitive to fluc-
economy, causing industrial materials prices to rise tuations in the foreign exchange value of the dollar.
at a time when other prices are relatively stable or The remaining components of the CIBCR in-
even declining.2 Temporary shortages of one or two dex are derived from surveys of business execu-
commodities also might be severe enough to cause tives. The percent of businesses expecting higher
a large increase in a broad commodity price index selling prices comes from a Dun and Bradstreet
even though most commodities are in ample supply. survey of nonfinancial firms. The last two compo-
Thus, even an index of industrial materials prices nents are from a monthly survey by the National
will probably not be a perfectly reliable indicator of Association of Purchasing Management (NAPM).
inflation. The price diffusion component measures the per-
cent of manufacturers experiencing higher prices
for materials they purchase. The vendor perform-
CIBCR leading inflation index ance component reflects how quickly suppliers are
able to make deliveries. Slower deliveries may
Because commodity prices are not perfectly imply that economic slack is diminishing, which
reliable inflation indicators, financial market par- may put upward pressure on consumer price infla-
ticipants and policymakers may wish to monitor a tion in the future.
ECONOMIC REVIEW • SECOND QUARTER 1995 9

Even a composite leading indicator of inflation and raised oil prices dramatically. Supply-induced
is unlikely to be totally accurate. The choice of shortages of agricultural products have also wors-
components for a composite index is based on past ened consumer price inflation at times. Yet like the
inflationary experience, yet the primary causes of CIBCR index, the PaineWebber leading indicator
inflation may differ over time. As a result, move- may not predict future inflation accurately if its
ments of a leading indicator might be misleading if components do not adequately reflect the current
the index does not give enough weight to components primary causes of inflation.
reflecting the current causes of inflation. In the
extreme case, a composite index might miss an
inflation upturn altogether if the inflationary pres- EMPIRICAL EVIDENCE
sures are caused by a factor that was unimportant in
the past. The previous discussion explained why the
price of gold, broad-based commodity price in-
dexes, and composite indexes might be useful lead-
PaineWebber leading inflation index ing indicators of consumer price inflation.
Theoretical arguments also suggest these indicators
Another example of a composite leading indi- might sometimes be misleading because of market-
cator is the PaineWebber leading inflation index, specific shocks to commodity markets, an improper
which is designed to anticipate turning points in weighting of components, or a false increase in
consumer price inflation by about 12 months. There inflation expectations. Empirical evidence is
are seven components of the PaineWebber indicator needed to more fully assess the usefulness of such
(Table 1). Although some components of the leading indicators to financial market participants
PaineWebber and CIBCR indexes overlap, the two and policymakers. Two kinds of empirical evidence
leading indicators differ in important ways. Both will be considered: a turning point analysis focusing
composite indexes include the NAPM price dif- on the timing of sustained upward or downward
fusion and vendor performance series. Industrial movements in the inflation rate, and a regression
materials prices also play an important role in both analysis designed to predict the magnitude of future
composite indexes, but the PaineWebber indicator inflation.
uses a CRB spot index of industrial materials
prices rather than the JOC index. The PaineWebber
indicator also includes demand pressures in the Behavior at turning points
labor market but measures these pressures by the
unemployment rate rather than the employment- The leading indicator approach to forecasting
population ratio. The PW indicator includes the has emphasized the prediction of turning points
trade-weighted foreign exchange value of the dollar rather than the prediction of inflation magnitudes.
as a measure of international influences rather than Previous research has found that leading inflation
the import price index in the CIBCR indicator. indicators typically signal major inflation peaks and
The PaineWebber indicator also may give troughs, but sometimes also give false signals of a
greater emphasis to inflationary disturbances origi- turning point. For example, Roth evaluated the
nating on the supply side of the economy. The PW turning point signals of five leading inflation indi-
index includes oil prices and agricultural prices as cators from 1948 to 1986. Roth found composite
separate components. Oil prices were a major con- indexes provide more reliable signals of inflation
tributor to rising inflation in 1973-74, and again in peaks and troughs than a commodity price index.
1979 when foreign oil producers curbed their output The composite indexes signaled every turning point
10 FEDERAL RESERVE BANK OF KANSAS CITY

Chart 1
Consumer Price Inflation
Percent change from 12 months ago
0.7 6.4 3.0 12.2 5.0 14.6 1.2 6.4
16

12

0
1960 ’64 ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’95

Source: U.S. Department of Labor.

in CPI inflation and gave fewer false signals of an of the chart. The exact months for the inflation
inflation peak or trough than did the JOC index of peaks and troughs are shown in Table 2.5
industrial materials prices.4 An informal evaluation of turning point predic-
Further turning point analysis is desirable, tions is possible with Chart 2. Each panel of the
however, because additional data have become chart shows one of the leading inflation indicators
available since Roth’s study, and because Roth did described earlier with shaded regions marking the
not evaluate the price of gold. The first step in the episodes of rising CPI inflation. Chart 2 shows
process is to identify peaks and troughs in consumer 12-month growth rates for the three commodity
price inflation. Because no generally accepted dat- price indexes and levels of the CIBCR and PW
ing of inflation turning points is available, peak and indexes.
trough dates were selected for this article based on Gold prices have tended to signal past inflation
an inspection of past CPI inflation rates. Chart 1 upturns, but have also fluctuated erratically in ways
shows consumer price inflation, measured by the that might be misleading to financial market partici-
percent change in the consumer price index (CPI) pants and policymakers. The price of gold is plotted
from 12 months earlier. The shaded areas represent in the first panel of Chart 2 only back to 1970
periods of rising consumer price inflation, with because the price of gold was previously fixed by
peak and trough inflation rates shown across the top the international monetary system.6 The 12-month
ECONOMIC REVIEW • SECOND QUARTER 1995 11

Table 2
Turning Points in the CPI Inflation Rate

Inflation trough Inflation peak

Percentage
Month Inflation rate Month Inflation rate point increase

Nov. 1961 .7 Feb. 1970 6.4 5.7

Aug. 1972 3.0 Nov. 1974 12.2 9.2
Dec. 1976 5.0 Apr. 1980 14.6 9.6
Dec. 1986 1.2 Oct. 1990 6.4 5.2

Note: Inflation is measured by the percent change in the CPI from 12 months earlier.

growth rate of gold prices always rose before the leading indicators turned up before peaks, and down
shaded inflationary episodes. However, the growth before troughs, in the CPI inflation rate. As is true
rate was negative—that is, gold prices were fall- for the commodity price indexes, however, the
ing—before the 1976 inflation trough, giving an amount of time by which movements in the CIBCR
inflation signal that many forecasters would have index or the PW index preceded movements in
found unconvincing at the time. Gold prices also consumer price inflation was variable from one
fluctuated substantially during the 1972-74 infla- inflation cycle to the next. For example, the PW
tion upturn. During this period, the sharp declines index peaked near the middle of the 1972-74 infla-
in the 12-month growth rate of gold prices could tion upturn, but the index peaked at the very end of
easily have misled forecasters into thinking an in- the 1986-90 inflation upturn. Such variable lead
flation peak was close at hand. times hamper the usefulness of leading inflation
The two broader commodity price indexes also indicators.
tended to lead the inflation cycle, but with erratic Although the leading indicators have not been
fluctuations that were potentially misleading to fi- perfectly reliable, these indexes have often given
nancial market participants and policymakers. For early warning signals of inflation turning points.
example, the 12-month growth rate of the CRB Moreover, even when the signals came slightly after
index historically started rising before inflation the turning points, such signals may have helped
troughs, but the growth rate was negative at the analysts confirm that a turning point occurred. Such
December 1986 inflation trough. The level of the confirming signals are useful since recognizing a
CRB index, therefore, was still declining at the time turning point may be difficult in practice because of
of the inflation upturn. Forecasters who were look- data lags and inflation measurement problems. Lead-
ing for an actual rise in the index thus received a ing inflation indicators, therefore, can be useful in
confirming signal only after the inflation upturn had forecasting or confirming turning points despite
already occurred. occasional false signals and variable lead times.
In contrast, the levels of the two composite While the analysis conducted here does not
12 FEDERAL RESERVE BANK OF KANSAS CITY

Chart 2
Five Leading Indicators of Inflation

Percent change from 12 months ago

200 Price of gold

150

100

50

-50
1960 ’64 ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’95
Source: Handy and Harmon.

80 Commodity
Research
60
Bureau
futures
40 index

20

-20
1960 ’64 ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’95
Source: Commodity Research Bureau.

60 Journal of
Commerce
40 index

20

-20
1960 ’64 ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’95
Source: Center for International Business Cycle Research.
ECONOMIC REVIEW • SECOND QUARTER 1995 13

Chart 2 (continued)
Five Leading Indicators of Inflation

1967=100
140 CIBCR
130 leading
inflation index
120

110

100

90
80
1960 ’64 ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’95
Source: Center for International Business Cycle Research.
1982=100
130 PaineWebber
leading index
120

110

100

90
1960 ’64 ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’95
Source: Economics Group, PaineWebber Incorporated.

provide a ranking of the five leading indicators, Regression analysis

theoretical reasoning and previous research suggest
the composite indexes are generally more reliable Financial market participants and policymak-
than commodity prices alone. Theoretical reason- ers are interested not only in inflation turning points
ing showed that composite indexes are less likely but also in the magnitude of future inflation. Al-
to give false signals of turning points because the though the leading indicator approach has focused
composite indexes are more diversified than the on predicting peaks and troughs, it is natural to ask
other indexes. Roth’s empirical research also found whether such indicators can also predict inflation
that composite indexes gave fewer false turning magnitudes. As a result, regression equations are
point signals. A good historical performance, how- estimated to determine how well each indicator—
ever, does not guarantee future reliability because taken by itself—explains subsequent CPI inflation.
the structure of the economy may change over time. The analysis then considers whether these indicators
14 FEDERAL RESERVE BANK OF KANSAS CITY

could add useful information to a standard eco- in the last two columns of Table 3. In this experi-
nomic model that relates consumer price inflation ment, one-month ahead CPI inflation is explained
to past inflation and economic slack. by month-to-month movements of the inflation in-
Much of the variation in future CPI inflation dicators over the previous 12 months. The compos-
cannot be explained by a single leading indicator. ite indexes again perform better than the other
Table 3 summarizes two forecasting experiments in leading indicators for 1973-94, although neither
which CPI inflation is explained by a single leading composite index predicts better than past CPI infla-
indicator. The first two columns of the table con- tion. For 1983-94, the explanatory power of all the
sider an experiment in which inflation over a future indicator variables again drops substantially. The
12-month period is explained by the change in com- PW composite index performs slightly better than
modity prices over the last 12 months or by the level the other leading indicators but not as well as past
of the CIBCR or PW index. The statistics show the CPI inflation.
percent of variation in future CPI inflation ex- The results in Table 3 provide a cautionary
plained by the single indicator variable. For compari- note about using leading indicators, by themselves,
son, the last row of the table shows how much of the to predict the magnitude of future CPI inflation. The
future inflation is explained by consumer price infla- leading indicators leave much of the variation in
tion over the previous 12 months, making no use of consumer price inflation unexplained for the 12-
information from any of the five leading indicators. month and one-month forecast horizons. The
Although all of the leading indicators leave a CIBCR index and the PW index generally per-
large part of the variation in future inflation unex- formed better than the commodity price indexes
plained, the composite indexes perform best over but still explained poorly the magnitude of inflation
1973-94. The CIBCR index explains 60 percent of in recent years.
the variation in future CPI inflation, and the PW Many financial market participants and policy-
index explains 54 percent. The other leading indi- makers may, however, be more interested in
cators explain less than half of the variation in CPI whether a leading indicator adds any additional
inflation. Only the CIBCR and PaineWebber in- information to a standard economic model of infla-
dexes predict future inflation better than past CPI tion. Most economists believe that economic slack
inflation over this period. is a key fundamental determinant of the inflation
The leading inflation indicators have even less rate. Two alternative measures of slack are considered
explanatory power over the shorter 1983-94 period. here—the civilian unemployment rate and the
This period was marked by a lower and more stable manufacturing capacity utilization rate.7 Lagged CPI
inflation rate than in the 1970s and early 1980s. inflation was also included to account for sluggish
Supply shocks to the agricultural and crude oil price adjustment and the influence of prevailing
markets also played less of a role in the inflationary inflation expectations on price and wage bargaining.
process during 1983-94. The second column shows Table 4 presents some tests of whether the five
that all of the leading indicators explain less of the leading inflation indicators contain additional pre-
variation in future CPI inflation for 1983-94 than dictive information.8 The tests consider whether
for 1973-94 as a whole. The price of gold, the adding a leading indicator to the standard economic
CIBCR index, and past CPI inflation experience model improves CPI inflation forecasts. Some tests
notable declines in predictive power. Although the also examine whether a leading indicator adds use-
PW and JOC indexes perform best, these indicators ful information to a forecast based solely on lagged
leave about 80 percent of the variation in future CPI CPI inflation for the last 12 months. Two periods
inflation unexplained. are considered, 1973-94 and 1983-94. A leading
A different forecasting experiment is summarized inflation indicator contains additional predictive
ECONOMIC REVIEW • SECOND QUARTER 1995 15

Table 3
Explanatory Power of Leading Inflation Indicators
(Percent of inflation variation explained)

12-month change Distributed lag

1973-94 1983-94 1973-94 1983-94

Price of gold .43 .02 .25 .09

CRB index .41 .13 .24 07
JOC index .37 .21 .20 .09
CIBCR index .60 .15 .45 .09
PW index .54 .19 .39 .17
Past CPI inflation .48 .01 .54 .23

Note: This table gives R 2 statistics from various regression models. In the first two columns, the dependent variable is a
simple percent change in the CPI from month t to month t+12. The explanatory variables are either the simple percent
change in commodity prices or the CPI from month t-12 to month t or the level of a composite index in month t. In the
last two columns, the dependent variable is the monthly change in the CPI. The explanatory variables are a 12-month
distributed lag on past monthly changes in commodity prices or the CPI, or past monthly levels of a composite index. All
regressions included a constant term.

information if the marginal significance level in the model in recent years.

table is less than 0.05, meaning there is less than a Taken as a whole, the empirical evidence
5 percent chance of concluding the indicator adds suggests that leading indicators of inflation may
information when it really does not.9 have some value to financial market participants
The leading inflation indicators add little useful and policymakers. The leading indicators often pro-
information to the standard model of inflation. vide signals that help predict or confirm a turning
When past CPI inflation is the only other explana- point in consumer price inflation. However, the
tory variable, the leading inflation indicators add indicators have signaled inflation turning points
predictive information for 1973-94. However, only with variable lead times and sometimes have given
the CRB futures price index adds information if false signals of an inflation peak or trough. Such
either the unemployment rate or manufacturing ca- indicators, by themselves, have not been successful
pacity utilization is also included. For 1983-94, in predicting the magnitude of CPI inflation in
none of the leading indicators adds predictive infor- recent years. Moreover, the indicators have typi-
mation except for one test involving the price of cally not added useful predictive information to a
gold. Thus, even the CRB futures index does not standard model of consumer price inflation that
add useful information to the standard economic includes economic slack and past inflation.
16 FEDERAL RESERVE BANK OF KANSAS CITY

Table 4
Tests for Additional Predictive Information
(Marginal significance levels)

1973-94 sample period

Leading inflation indicator
Other variables Price of gold CRB index JOC index CIBCR index PW index

Lagged inflation .020 .000 .002 .000 .021

Lagged inflation
and unemployment .065 .040 .188 .126 .358
Lagged inflation
and capacity use .126 .041 .189 .131 .761

1983-94 sample period

Leading inflation indicator
Other variables Price of gold CRB index JOC index CIBCR index PW index

Lagged inflation .198 .839 .755 .662 .272

Lagged inflation
and unemployment .273 .940 .869 .848 .512
Lagged inflation
and capacity use .000 .628 .850 .732 .358

Note: These tests are based on regressions relating the monthly change in the CPI to 12-month distributed lags on past
CPI inflation and one of the indicator variables. The commodity price variables are entered as percent changes, and the
composite indexes are entered as levels. In addition, some of the regressions included 12-month distributed lags of either
the unemployment rate or manufacturing capacity utilization. All regressions included a constant term.

CONCLUSION of turning points. But other economic models and

statistics can provide more accurate forecasts of
Because the empirical evidence provides only future inflation magnitudes.
qualified support for the leading indicators of infla- Given that the leading indicators provide some
tion, financial market participants and policymakers useful information about future inflation, what are
should probably use a variety of economic models such indicators saying about the current situation?
and indicators to predict CPI inflation. The leading The leading indicators generally gave a warning
inflation indicators are useful primarily as a signal signal of rising inflationary pressures in 1994 but
or confirmation of an inflation turning point. have turned down slightly in 1995. The price of gold
Among the set of leading indicators, composite in- rose in 1993 and early 1994, but stabilized in late
dexes are likely to provide the most reliable signals 1994 and early this year. As noted earlier, some of
ECONOMIC REVIEW • SECOND QUARTER 1995 17

the increase in gold prices may have reflected rising Thus, it is too early to conclude with any certainty
inflation expectations in Southeast Asia and political that upward pressures on the leading indicators
uncertainties in Russia. The other leading indicators have abated. Taken with the evidence of declining
of inflation rose sharply last year, although these economic slack, the leading indicators of inflation
indicators have either stabilized or declined slightly justify recent concerns about potential inflationary
so far in 1995. Recent declines in the indicators have pressures.
been small, however, relative to last year’s increases.

ENDNOTES
1 This discussion assumes that the monetary authority does because this indicator is not widely followed by financial
not accommodate the upward shift in the short-run aggregate market participants or policymakers. The money supply is not
supply curve caused by the erroneous expectations. Monetary considered because M1 has not been a reliable indicator of
accommodation could produce a self-fulfilling prophecy in real activity or inflationary pressures in the 1980s or 1990s.
which higher inflation expectations, even though based on Roth measured consumer price inflation by the six-month
erroneous information, ultimately lead to a higher inflation smoothed change in the CPI, which gives a somewhat
rate. different dating of inflation turning points than the 12-month
change used in this article.
2 Technological progress also may change the equilibrium
price of industrial commodities relative to other goods and 5 An inflation upturn also occurred in 1959 and the first few
services. Technological progress in the postwar era has often months of 1960. However, the early months of 1960 are not
reduced the demand for industrial commodities by making shaded in Chart 1 or Chart 2 because the charts do not show
products smaller and lighter or by allowing the producer to the full inflationary episode. The higher inflation in late 1983
use cheaper synthetic materials. Many new products, such as and 1984 is not classified as an inflation upturn because this
computers and software, require little commodity input rise in the inflation rate “most likely is a statistical artifact”
relative to the value of the product. And, technological associated with a change in the way the CPI measured
advances have helped commodity producers open new homeownership costs (Roth).
sources of supply that previously were not feasible. Using another price index or another method of computing
Reflecting such influences, the relative price of commodities the percent change might result in different turning points for
has drifted downward over the long term (Reinhart and the inflation cycle. The peak and trough dates in Table 2 are
Wickham). Such factors may have weakened the link similar to those in PaineWebber, although the exact dates of
between commodity price movements and the general the inflation turning points sometimes differ by a few months.
inflation rate because commodities have become less Unlike PaineWebber, this article does not classify the brief
important in the production process. increase of CPI inflation in 1981 as an inflation upturn.

3 The empirical work in this paper was conducted before a 6 The price of gold is a spot price from Handy and Harmon
recent revision in the CIBCR leading index. Thus, the results derived from Tuesday quotes in The Wall Street Journal.
may not be fully representative of how the current version of
the index would have performed historically. The empirical 7 Garner (1994) and Weiner provided evidence relating these
tests do, however, assess the version of the CIBCR index that measures of economic slack to the U.S. inflation rate. Supply
received substantial attention in the last two years as concern shock variables also might be included in a mainstream
about future inflationary pressures mounted. The CIBCR economic model because most economists agree supply-side
index has been revised several other times in the past. variables, such as the price of crude oil, have caused major
fluctuations in the inflation rate. Such supply shock variables
4 Roth evaluated turning point predictions by Moore’s were not included here to focus on the key role of economic
leading inflation index, the Niemira composite index, the slack. In addition, omitting supply shock variables probably
Morosani indicator, the JOC index, and the growth rate of the gives the leading indicator variables a better opportunity to
M1 measure of the money supply. The Niemira index was a add useful information to the mainstream model. Including
predecessor of the PaineWebber index, but the current PW supply shock variables separately in the regression equations
index is substantially revised and has more components. The would probably strengthen the finding that the leading
Moore index and the JOC index have also been revised since indicators do not add useful information to the mainstream
Roth’s study. The Morosani indicator is not considered here model.
18 FEDERAL RESERVE BANK OF KANSAS CITY

8 Previous research has reached differing conclusions about 9 Table 4 reports marginal significance levels to test the
whether commodity price indexes help predict future hypothesis that the coefficients of the leading indicator are
inflation. Garner (1989) found commodity price indexes zero. The number 0.05 is widely used as a criterion for
contained information that was useful in predicting CPI statistical significance, but other analysts might prefer 0.01
inflation. However, these tests did not include the measures or 0.10. The tests for predictive usefulness in Table 3 are
of economic slack that are used here and covered substantially described in Granger. Monthly CPI inflation is regressed on
different time periods. Branson and Boughton also presented 12 lagged values of CPI inflation and 12 lagged values of the
some evidence that commodity prices help predict the general leading inflation indicator. Sometimes, the regressions also
inflation rate. But Fuhrer found a measure of commodity include 12 lagged values of either the civilian unemployment
price inflation, by itself, did not significantly improve CPI rate or the manufacturing capacity utilization rate.
inflation forecasts.

REFERENCES
Feder, Barnaby J. 1994. “Inflation Concerns Rise with Com- Granger, C.W.J. 1969. “Investigating Causal Relations by
modity Prices,” The New York Times, June 22, pp. C1 and Econometric Models and Cross-Spectral Methods,”
C15. Econometrica, July, pp. 424-38.
Boughton, James M., and William H. Branson. 1991. “Com- Laurent, Robert D. 1994. “Is There a Role for Gold in
modity Prices as a Leading Indicator of Inflation,” in Kajal Monetary Policy?” Federal Reserve Bank of Chicago Eco-
Lahiri and Geoffrey H. Moore, eds., Leading Economic nomic Perspectives, March/April, pp. 2-14.
Indicators: New Approaches and Forecasting Records. Moore, Geoffrey H., and Stanley Kaish. 1983. “A New Infla-
Cambridge: Cambridge University Press. tion Barometer,” Morgan Guaranty Survey, July.
Fuhrer, Jeffrey C. 1993. “Commodity Prices, the Term PaineWebber. 1995. Monthly Economic Review, January 16,
Structure of Interest Rates, and Exchange Rates: Useful p. 5.
Indicators for Monetary Policy?” Federal Reserve Bank Reinhart, Carmen M., and Peter Wickham. 1994. “Commod-
of Boston New England Economic Review, November/ ity Prices: Cyclical Weakness or Secular Decline?” Inter-
December, pp. 18-32. national Monetary Fund Staff Papers, June, pp. 175-213.
Garner, C. Alan. 1989. “Commodity Prices: Policy Target or Roth, Howard. 1986. “Leading Indicators of Inflation,” Fed-
Information Variable?” Journal of Money, Credit, and eral Reserve Bank of Kansas City Economic Review, No-
Banking, November, pp. 508-14. vember, pp. 3-20.
. 1994. “Capacity Utilization and U.S. Infla- Weiner, Stuart E. 1993. “New Estimates of the Natural Rate
tion,” Federal Reserve Bank of Kansas City Economic of Unemployment,” Federal Reserve Bank of Kansas City
Review, Fourth Quarter, pp. 5-21. Economic Review, Fourth Quarter, pp. 53-69.