|
November 1989
Federal Reserve Bank of Dallas
The Texas Industrial Production Index
 |
Abstract
The Texas Industrial
Production Index (TIPI) measures the output
of the manufacturing, mining, and utility
sectors of the Texas economy. These sectors
are of special interest because of their
sensitivity to business cycles and because
of the size (albeit declining) of the Texas
mining sector. The Federal Reserve Bank
of Dallas has published TIPI since 1958.
Revisions are implemented when new data
sources are available, when existing data
are revised, or when methodological improvements
are devised. The most recent major TIPI
revision came in the fall of 1988.
Berger and Long examine
TIPI’s performance during the volatile
1980s and relate this performance to the
broader economic environment in which it
took place. They find that the Texas industrial
sector has grown more slowly than that of
the nation since 1982 and that TIPI clearly
depicts the oil-price induced 1986–87
Texas recession. They also find that Texas
industrial production was buoyed by the
manufacturing sector and hindered by mining,
although the effects of two periods of drastic
oil price declines spilled over to the manufacturing
sector as well. |
|
|
The Federal Reserve Bank of Dallas
has produced the Texas Industrial Production Index (TIPI)
continuously since 1958. The index measures the output
of Texas’ mining, manufacturing, and utilities
sectors and provides a regional counterpart to the national
industrial production index compiled by the Board of
Governors of the Federal Reserve System. Regional production
indexes published by several other Federal Reserve Banks
are limited to the manufacturing sector. The importance
of oil and gas extraction in the Teas economy, through
both direct and secondary effects on the manufacturing
sector, necessitates its inclusion.
In this article, we consider what
an industrial production index attempts to measure,
why it is useful in conducting economic analysis, and
what TIPI tells us about the performance of the Texas
economy during the 1980s. In the appendix, we explain
TIPI’s construction in detail, including recent
methodological improvements such as the incorporation
of gross state product data.
The Texas Industrial Production
Index clearly shows that the manufacturing sector buoyed
Texas industry during the 1980s. Texas manufacturing
output rose faster than overall industrial output, especially
after the national recovery began in 1982. Nevertheless,
until recently, drastic declines in oil process and
the resulting entrenchment in the energy industry held
the growth of manufacturing output in Texas below that
of the nation. TIPI clearly depicts the 1986 Texas recession
caused by the oil price collapse and the weak recovery
that began in early 1987.
Why a regional production index?
The Federal Reserve Bank
of Dallas is interested in monitoring the economic activity
of the Eleventh Federal Reserve District, which includes
Texas and parts of Louisiana and New Mexico. This interest
is motivated, in part, by the special role the Bank
plays in contributing to the formation of the nation’s
monetary policy and in gaining awareness of the varying
impacts of monetary policy at the regional level. Another
motivating factor is the research Bank economists conduct
on such issues as the causes and consequences of regional
economic growth and future trends in regional economic
activity. Further, the Bank is committed to providing
information of interest to the public as well as the
business and academic communities.
TIPI is intended to supplement
the wealth of other data that are used for regional
analysis but are limited in timeliness or scope. For
example, employment data are timely but not comprehensive.
Employment data are available with a fairly short time
lag for most states as well as the nation. But employment
may not accurately reflect the income that a particular
industry or region generates if technological change
permits output to increase without increasing employment.
Moreover, employment and output will not have a constant
relationship because of the substitution of capital
for labor in production processes.
 Texas
manufacturing illustrates how output and employment
can diverge (Chart 1). Through the 1970s, manufacturing
employment and output tracked one another fairly closely.
In the early 1980s however, output growth exceeded employment
growth, as shown by the divergence of the two indexed
series. One explanation for this divergence is that
technological change during the 1970s reduced the importance
of labor in many manufacturing processes. As capital
assumed more importance, manufacturers were able to
increase output without corresponding increases in employment.
TIPI can also provide more timely
information than other measures of regional economic
activity. Direct measures of income or output, such
as personal income and gross state product, are available
for all states, including industry detail. Yet, the
time lag in reporting these data ranges from several
quarters to several years. Such long lags make analyzing
current conditions and forecasting future activity difficult.
Economic activity in many industries can reverse course
quickly. If reporting lags are long, then business and
government decisionmakers will be unable to react appropriately
to such changes. Furthermore, in forecasting it is necessary
to predict the values of variables not only for the
future but also for current or past periods for which
data are not available. To the extent that historical
data are not available on a timely basis, forecasts
will be weakened.[1]
What is TIPI?
The Texas Industrial Production
Index provides timely monthly estimates of changes in
the level of output of the manufacturing, mining, and
utilities sectors of the Texas economy. TIPI includes
indexes for aggregates such as durable and nondurable
goods, manufacturing, mining, utilities, and total industrial
production, plus all two-digit Standard industrial Classifications
(SIC)[2] that have significant representation in Texas
industry. The indexes begin with January 1967.[3]
Actual monthly physical output
is available for several industries, and is incorporated
into the calculation of TIPI.[4] For the rest, two alternative,
but related, measures of output are available at the
state and two-digit SIC level of detail. Both are based
on the concept of value added. Value added
is the market value of produced goods less the cost
of the materials and services purchased from others
to produce those goods. Equivalently, value added is
the income earned by the factors of production such
as labor, capital, land, or entrepreneurship. We compare
the two sources of value added data later in this article.
To ensure that changes in estimated
output correspond as closely as possible to changes
in real physical output and are not due to changes in
price levels, we deflate nominal value added by using
two-digit SIC price deflators from the U.S. Bureau of
Economic Analysis (BEA).[5]
Unfortunately, data on output
(value added) for most industries at the regional level
are available only annually. The purpose of an industrial
production index, however, is to facilitate analysis
of recent developments. Therefore, a regional
industrial production index methodology must transform
data that are available on a timely and monthly basis
into a measure of monthly production. Texas employment
and average weekly hours worked are available
monthly from the U.S. Bureau of Labor Statistics.[6]
Data on electric power sold to each two-digit SIC Texas
industry are available from the Federal Reserve Bank
of Dallas Statistics Department. These are the principal
data transformed into output for industries that do
not have actual monthly output data available. The appendix
to this article explains in detail how the Texas Industrial
Production Index is constructed.
What does TIPI show about Texas
industries?
This section covers what
TIPI reveals about the performance of the Texas industrial
sector during the 1980s. We begin with a discussion
of the major factors affecting Texas industry in the
1980s. It is in the context of these events that movements
in Texas industrial output occurred. Then for TIPI as
a whole and for selected subindexes, we examine index
movements from two perspectives. First, we compare the
performance of the index with its national counterpart.
We also examine how the performance of the major components
of TIPI affected the overall index. Where possible,
we offer explanations for the behavior of these industries,
but we do not attempt to use a formal model to quantify
what factors contributed to fluctuations in individual
industry.[7] Previous research examined how growth in
various industries affects the volatility of regional
economies and how exogenous shocks affect different
industries.[8]
Events affecting Texas industry.
Three related events dominated
the performance of Texas industry during the 1980s.
The first and most obvious was the decline in oil prices,
first beginning at the end of 1982 and again at the
end of 1985. Even worse than the large decline in prices
that actually occurred was the plunge in the
level that people expected oil prices to reach.
Forecasts that oil prices would exceed $100 per barrel
by the year 2000 were not uncommon before 1982. Clearly,
such high prices failed to materialize, but many oil
producers and consumers based plans on drastically higher
prices. This is significant because much economic activity
is based on expectations about the future state of the
economy. Thus, the reduction in economic activity that
followed tumbling prices was much worse than it would
have been had price expectations been more realistic.[9]
A second major influence that
buffeted both the Texas and national economies was the
rise, then prolonged fall, in the foreign exchange value
of the U.S. dollar. The rising dollar made foreign goods
less expensive relative to domestic goods, which hurt
U.S. manufacturers, including those in Texas. The decline
in the value of the dollar beginning in March 1985 reversed
this effect, helping domestic firms sell more goods
to other countries. Cox and Hill (1988) conclude that
Texas was a significant beneficiary of the decline in
the dollar, though the impact for the state was slightly
less than for the nation.
The third factor affecting Texas
industry during the 1980s was the record-length recovery
of the nation’s economy since 1982. As with the
decline in the dollar, the recovery has benefited Texas
industry, notwithstanding the problems generated in
the energy and financial sectors by the oil price decline.
The response to these factors clearly has not been uniform
across industries.
 The
performance of Texas industry in the 1980s. Since
1982, Texas industrial output overall has grown more
slowly than national output (Chart 2). The
decline in oil prices prevented the Texas economy from
rebounding from the 1981–82 recession as strongly
as did the nation. Later, while national output was
boosted by lower oil prices and a lower dollar, Texas
output remained weak as the state economy became less
energy-dependent. Nevertheless, several industries in
the state have outperformed their national counterparts
since 1982. These include electric power generation,
oil and gas extraction, instruments, transportation
equipment, and electric and electronic equipment.
Three of the five faster-growing
Texas industries—instruments, transportation equipment,
and electric and electronic equipment—are durable
goods manufacturing industries. These three industries
are relatively insensitive to fluctuations in energy
prices. They are also among the industries that stood
to benefit most from declines in the exchange value
of the dollar.[10] While these factors explain their
overalls strong performance, they do not explain why
these industries grew faster in Texas than in the nation
as a whole. The factor that best explains the stronger-than-national
performance of these industries is their relationship
to the defense industry in Texas. The defense buildup
during the first half of the 1980s benefited Texas defense
contractors. These contractors tend to be heavily concentrated
in aircraft and electronic industries. It is likely
that strength in the instruments industry is also related
to military spending. Although Texas no longer greatly
exceeds the nation in per capita defense spending, defense
outlays have likely boosted these industries more than
their national counterparts.
Oil and gas extraction.
Despite sharp increases in
oil prices during 1973–74 and 1978–81, output
in oil and gas extraction in both Texas and the rest
of the United States remained essentially flat throughout
the 1970s and early 1980s. Oil and gas extraction comprises
crude oil production, natural gas production, and oil
and gas field services, which are largely exploration-related.
Texas crude oil production was about 24 percent lower
in 1981 than in 1970. For the rest of the United States,
it was only about 4 percent lower. Increased exploration
activity resulting from oil price increases prevented
crude oil production from falling even further. Exploration
activity contributes directly to oil and gas extraction
because the drilling activity increases value added
regardless of whether oil or gas is discovered. New
discoveries, principally in Alaska, and increases in
output from higher-cost wells account for the smaller
decline in the rest of the United States. Because Texas’
oil fields are relatively old and cost less to operate,
Texas crude oil production is not as responsive to price
increases.
 The
oil price declines of 1981–82 and 1985–86
tell a different story. These decreases caused precipitous
declines in Texas oil and gas extraction of about 11
percent and 16 percent, respectively (Chart 3).
These declines also had indirect
effects, such as lowering the demand for the goods and
services of other industries and reducing the incomes
of royalty owners, drillers, and producers. The lower
demand for other goods and services contributed to the
recession in Texas through multiplier effects.[11] During
the first oil price decline in early 1980s (see the
left shaded area of Chart 3), the 10-percent decline
in total Texas industrial output was similar to the
decline in oil and gas production. During the second
period (see the right shaded area of Chart 3), however,
the 6-percent decline in overall output was much smaller
than the 16-percent decline in oil and gas production.
This difference may reveal how much the Texas industrial
sector had already adjusted to lower oil prices by the
mid–1980s. Of course, the national recession that
occurred during the first period confounds our ability
to be more conclusive on this point. Effects of the
state’s weakened economy also were observable
in other industries.
Total manufacturing.
Manufacturing may be the
most interesting sector to analyze with a regional industrial
production index such as TIPI. The economic events mentioned
earlier—oil price movements, exchange -rate fluctuations,
and the national economic expansion—affected this
sector in various ways. Lower oil prices benefited some
industries and hurt others. The falling value of the
U.S. dollar benefited some industries more than others,
and the national economic expansion allowed some industries
to avoid a more severe downturn as a result of falling
oil prices. Before examining the durable and nondurable
components of the manufacturing sector, we will compare
the sector’s overall performance to that of the
nation.
 Output
in Texas manufacturing industries grew faster than that
of the nation until 1982 (Chart 4). Since the
national recession and the 1982 decline in oil prices,
Texas manufacturing output has grown more slowly, on
average, than that of the nation. Texas manufacturing
output fell as a result of the 1985–86 oil price
decline, whereas the nation’s output did not.
In early 1987, Texas manufacturing output, boosted by
the falling dollar and the continued national expansion,
began to rebound and actually grew faster than national
manufacturing output.
Manufacturing constitutes roughly
54 percent of industrial output in Texas, compared with
78 percent for the nation.[12] After the first oil price
decline, which occurred at the end of a national recession,
manufacturing output in Texas declined about 10 percent,
roughly the same as the decline for overall state industrial
output (Chart 5 and Table 1). After the retrenchment
and readjustment in the early 1980s, manufacturing output
declined only 2.7 percent after the severe 1985–86
oil price decline. The beneficial effect of the oil
price decline on the national economy probably offset
some of that event’s negative effect on Texas
manufacturing. Manufacturing output was further helped
by the lagged effect of the decline in the dollar’s
value beginning in March 1985.
| Table 1 |
| Declines in Industry Output in
Texas in Selected Periods |
| Industry |
First
downturn |
Decline |
Second
downturn |
Decline |
| Total
industrial production |
Aug.
1981–
March 1983 |
–9.53 |
Jan.
1986–
March 1987 |
–6.26 |
|
Mining |
Sept. 1981–
March 1983 |
–11.12
|
Aug.
1985–
September 1986 |
–15.64 |
| Manufacturing |
July
1981–
Feb. 1983 |
–10.07
|
Jan.
1986–
Dec. 1986 |
–2.70 |
| Durable
|
Sept.
1981–
March 1983 |
–16.30
|
July
1985–
Feb. 1987 |
–5.24 |
|
| SOURCE: Federal Reserve Bank
of Dallas. |
 Durable
good manufacturing. Texas
durable goods manufacturing output also grew faster
than that of the nation during the 1970s (Chart
6). In addition, with the oil economy as a buoy,
durable goods manufacturing in the state did not suffer
swings in output as severe as those of the nation. The
effect of the 1982 fall in oil prices was to delay the
recovery of Texas durable manufacturing industries after
the 1981–82 recession. When the 1985–86
oil price decline hit, durable goods production in Texas
went into another slump while that in the nation did
not.
 The
durable goods industries performing the poorest clearly
have been those with the strongest ties to the energy
industry. Oil field machinery, for example, is an important
component of nonelectrical machinery production in Texas
(Chart 7). Another example is primary metals
production, which provides drill pipe and structural
steel to the extraction industry, and which has not
recovered as much in Texas as it has nationally since
the 1981–82 recession (Chart 8).
 Texas
durable goods manufacturing constitutes about 27 percent
of total industrial output in the state and 49 percent
of manufacturing output. During both the 1981–82
and the 1985–86 oil price downturns, durable goods
manufacturing output declined more than overall manufacturing
output (Chart 9 and Table 1). For the earlier
decline, durables output declined 16.3 percent, compared
with roughly 10 percent for total manufacturing. During
the most recent downturn, durables output fell by slightly
more than 5 percent, roughly double the percentage decline
in overall manufacturing output. The larger decline
for durable goods production conforms to the typical
behavior of this industry during a business cycle—durable
goods are generally subject to larger output swings.
Nevertheless, during the second oil price downturn,
Texas durable production fell less than did overall
industrial production because, in percentage terms,
mining output fell by much more.
 Nondurable
goods manufacturing. Nondurable
goods manufacturing in Texas has not performed as well
as its national counterpart (Chart 10). In
fact, not a single nondurable component industry has
achieved faster output growth than its national counterpart.
The Texas index for paper and allied products rose above
that for the nation in early 1989, after lagging behind
for most of the previous decade.
Chemicals and related products
and petroleum and coal products, the latter group being
primarily the refining industry, are the two largest
Texas manufacturing industries (Table 2).
| Table 2 |
Texas Industry Weights and Factor
Shares, 1986
(Percent) |
| |
Gross
product |
Factor
shares |
| Industry
(SIC Code) |
Share
|
Labor |
Capital |
| Lumber
and wood products (24) |
1.5
|
36.9
|
63.1 |
| Furniture
and fixtures (25) |
0.5 |
50.5 |
49.6 |
Stone,
clay, and glass products (32)
|
2.5 |
34.6 |
65.4 |
| Primary
metal industries (33) |
1.8
|
40.9
|
59.1 |
| Fabricated
metal products (34) |
3.5
|
52.5
|
47.5 |
| Machinery,
except electrical (35) |
6.0
|
45.3
|
54.7 |
| Electric
and electronic equipment (36) |
5.8 |
54.4 |
45.6 |
| Transportation
equipment (37) |
4.2
|
64.7
|
35.3 |
| Instruments
and related products (38) |
1.0 |
60.1 |
39.9 |
| Total
durable goods |
26.8
|
— |
— |
| |
|
|
|
| Food
and kindred products (20) |
4.8
|
39.6
|
60.4 |
Apparel
and other textile products (23)
|
1.1 |
50.0 |
50.0 |
| Paper
and allied products (26) |
1.2 |
51.2
|
48.8 |
| Printing
and publishing (27) |
2.9 |
52.1 |
47.9 |
| Chemicals
and allied products (28) |
7.2 |
— |
— |
| Petroleum
and coal products (29) |
8.8 |
— |
— |
| Rubber
and miscellaneous plastics products (30) |
1.4 |
45.2 |
54.8 |
Total
nondurable goods |
27.4 |
— |
— |
| |
|
|
|
| Total
manufacturing |
54.2 |
— |
— |
| |
|
|
|
| Mining
except oil and gas (10, 12, 14) |
0.6 |
66.7 |
33.3 |
| Oil
and gas extraction (13) |
34.7 |
— |
— |
| Total
mining |
35.3
|
— |
— |
| |
|
|
|
| Electric
utilities (491) |
8.0 |
— |
— |
| Gas
utilities (492) |
2.4 |
— |
— |
| Total
utilities |
10.4 |
— |
— |
| |
|
|
|
| Total
industrial production |
100.0
|
— |
— |
|
| SOURCES OF PRIMARY DATA: American
Gas Association, Bureau of Economic Analysis, U.S.
Department of Commerce, Energy Information Administration,
U.S. Department of Energy. |
 Chemicals
output in 1988 and early 1989 grew faster in Texas than
in the nation. However, Texas chemicals output showed
slower growth rates, if not declines, when compared
to the nation for most of the previous six years (Chart
11).
Likewise, Texas’ output
of petroleum and coal products has generally increased
less than the nation’s during the 1980s (Chart
12). Slow growth in these two components would
have been sufficient to cause Texas nondurable goods
manufacturing output to lag that of the nation, even
if growth in other components had been near the national
rate.
Apparel manufacturing provides
a contrast with other nondurable goods manufacturing
industries in Texas. After experiencing a far more severe
slump in the state than in the nation in the early 1980s,
apparel manufacturing has grown far faster in Texas
than is has nationally (Chart 13).
We know that, nationally, nondurable
goods manufacturing exhibits smaller output swings during
business cycles than durable goods manufacturing. TIPI
confirms this for Texas (Chart 14) for the
1974–75, 1981–83, and 1986–87 recessions.
 Utilities.
Electric power and natural
gas utilities constitute about 10 percent of total industrial
output in Texas. Natural gas utilities include only
those firms involved in the transmission and distribution
of natural gas, not in its extraction. Both sectors,
of course, respond to changes in other energy markets.
But because both are heavily regulated, their output
behavior is also influenced by changing regulatory environments.
No attempt is made here to describe in detail the effects
of energy prices and regulation or their interaction.
One interesting effect of the oil price decline in 1982,
however, is the sharp acceleration of electric power
production (Chart 15). After a flat performance
from 1970 to 1982, electric power generation rose sharply
from 1982 to 1986, before flattening again after 1986.
During the 1970s and early 1980s,
natural gas utilities suffered declining output. Two
factors account for this decline. One is the restriction
placed on the industrial uses of natural gas during
this period. Because of its importance for residential
heating, some industries, notably the electric power
industry, were prohibited from using natural gas in
new facilities. A second factor is the substitution
of other fuels as energy sources. Both of these effects
resulted form rising prices.
Conclusion
TIPI provides a useful tool
for analyzing the Texas economy. By providing monthly
estimates of the output of the Texas industrial sector,
the index can be used to analyze intertemporal, interregional
and interindustrial changes in economic conditions.
TIPI shows the degree to which
manufacturing buoyed the Texas economy in the 1980s
and the degree to which the mining sector has hindered
it. TIPI also shows that despite declining oil prices,
the importance of defense spending in the state, the
national economic expansion, and the declining dollar
enable some Texas industries to grow faster than their
national counterparts in recent years.
—Franklin D. Berger and
William T. Long III
|
| About
the Authors
Berger is manager
of research support and Long is an economist
at the Federal Reserve Bank of Dallas.
Notes
- We do not forecast Texas industrial
production in this article. For an example
of such a forecast, see Gruben and Long
(1988).
- See Standard Industrial Classification
Manual (1972).
- Although we have produced TIPI since
1958, methodological and data changes
over time prevent calculation of the indexes
before 1967.
- These industries are oil and gas extraction,
petroleum and coal products, electric
utilities, and gas utilities. Details
on how the production indexes for these
industries are created are provided in
the technical appendix.
- Because regional price deflators do
not exist, we must use national deflators.
To the extent that the distribution of
industries constituting any two-digit
SIC code varies regionally, inaccuracy
is introduced into the process of constructing
real value added.
- Average weekly hours worked are available
for production workers only. We assume
that using these figures for nonproduction
workers does not introduce serious error
into the estimates.
- Because TIPI serves as a timely monthly
indicator of economic activity in Texas,
it has been used in econometric forecasting
models for the state. See Gruben and Long
(1988).
- See Gruben and Phillips (1989) for a
discussion of which industries contribute
to stability in the Texas economy. Sherwood-Call
(1988) examines which states have had
the most stable economies in recent years.
- For a discussion of the adjustment of
the Texas economy to lower oil prices,
see Fomby and Hirschberg (1989).
- See Cox and Hill (1988, 7).
- For models that describe how these multiplier
effects influence the Texas economy, see
Hill (1986) and Brown and Hill (1988).
- Manufacturing is about 16 percent of
total gross state product in Texas and
about 19.7 percent of gross domestic product
for the nation.
Appendix
How is TIPI Constructed?
Researchers at
the Federal Reserve Bank of Atlanta pioneered
the methodology used to construct TIPI.[1]In
fact, each of the regional production indexes
currently in use in the Federal Reserve
System relies on the Atlanta method.[2]
Numerous articles have been written on the
methodological aspects of constructing industrial
production indexes. Previous research conducted
at the Federal Reserve Bank of Dallas supports
the use of the Atlanta method on grounds
of both accuracy and minimization of the
resources necessary to produce the index
on a continuing basis.[3] For several industries,
however, we employ techniques that differ
considerably from the basic Atlanta method.
We will describe these deviations subsequently.
Atlanta method
Assuming that
firms maximize profits in perfectly competitive
markets and employ a two-factor linear homogeneous
production function, then according to Euler’s
Theorem,[4] the net physical product of
an industry, Q can be written
(1)
Q = MPL • L + MPK
• K,
where MPL
is the marginal product of labor, L
is units of labor, MPK
is the marginal product of capital,
and K is units of capital. Multiplying
both sides of equation 1 by product price,
P, reveals that
(2)
P • Q = P • MPL
• L + P • MPK •
Kt
or
(3)
VA = VMPL • L + VMPK
• K
where VA
is nominal value added, VMPL
is the value of the marginal product of
labor, and VMPK is the
value of the marginal product of capital.
Under the assumptions of profit maximization
under competition, VMPL =
PL and VMPK
= PK, where PL
is the price of labor, or wage rate, and
PK is the price of capital. Therefore,
(4)
VA = PL • L + PK
• K,
which states that
nominal value added is the sum of the wage
bill and the capital bill. Multiplying the
first term of equation 4 by (VA/VA)
• (L/L) and the second term
by (VA/VA) • (K/K),
and then rearranging terms, we have
Noting VA/P =
Q, dividing through by product price,
and rearranging terms results in
To simplify the notation,
let labor’s share in value added,
(PL • L)/VA,
be denoted SL, let the
productivity of labor, (Q/L), be
denoted ?L let capital’s share in
value added, (PK • K)/VA,
be denoted SK, and let the productivity
of capital (Q/K), be denoted ?K.
Letting t denote a given time period,
equation 6 can be rewritten as
which states that,
in any time period, physical output (or
real valued added) consists of the weighted
contributions of labor and capital, where
each factor’s contribution is the
amount of that factor used, multiplied by
its productivity, and where the weights
are each factor’s share of total nominal
value added. Equation 7 provides the basis
for estimating monthly industrial production.
Substituting actual or estimated values
for factor usage, factor productivities,
and factor shares into equation 7 results
in the monthly estimate of production for
an industry.
Data considerations
Practitioners
of industrial production indexes using the
Atlanta method must make choices regarding
the data to use in equation 7. Data availability
and quality can vary regionally and over
time. The following sections describe decisions
we made at the Federal Reserve Bank of Dallas.
Benchmarking.
Benchmarking
an index is the process of ensuring that
long-run movements in the generated monthly
production index correspond to long-run
movements in the known annual output measure.
TIPI is the first regional index to be benchmarked
to new data on gross state product available
from the Bureau of Economic Analysis of
the U.S. Department of Commerce. Other regional
production indexes are benchmarked (as were
previous versions of TIPI) to value-added
data published by the Census Bureau of the
U.S. Department of Commerce in its Censuses
and Annual Surveys of Manufactures.
There are several
advantages to using the BEA data. First,
BEA has devoted considerable effort to providing
value added estimates that improve on those
available from the Census Bureau.[5] The
terms value added and gross
product will be used interchangeably.
The principal improvement is that BEA has
subtracted an estimate of the cost of purchased
services from the Census Bureau measure
of value added. Thus, some output attributed
by the Census Bureau to each industry is
properly attributed to the service sector
by BEA. Second, using gross production eliminates
a methodological inconsistency that results
from using GNP price deflators with value-added
data. Third, the BEA data are available
annually from 1963 to 1986, including data
from 1979 to 1981, when the Census Bureau
did not publish Annual Surveys of Manufactures
at the state level. Finally, BEA provides
annual estimates of gross product in mining
industries, whereas the Census Bureau’s
value-added estimates are available only
at five-year intervals in the Census of
Mining.
Labor. We
take the labor input to be the product of
employment and average weekly hours, as
reported monthly in the Bureau of Labor
Statistics’ Establishment Survey.
With one exception, data on Texas total
employment and average weekly hours are
available for all two-digit industries back
to at least 1967. Average weekly hours for
instruments and related products (SIC 38)
are unavailable before 1972. During the
1967–71 period, variation in the labor
input for that industry is solely due to
variation in employment.
Capital.
Because electricity
powers much modern capital equipment, the
TIPI methodology uses electricity consumption
to proxy the usage of capital. This is a
common technique validated by previous research.[6]
The Federal Reserve Bank of Dallas Statistical
Department collects electric power data
from a panel of Eleventh District electric
utilities that report electricity sales
by SIC code.
A phenomenon that
somewhat limits the value of the electric
power data is that of cogeneration. Cogeneration
is the simultaneous generation of electricity
and useful heat from a single fuel source.
The Public Utility Regulatory Policies Act
of 1978, which requires utilities to buy
power from private cogenerators, and the
Natural Gas Policy Act of 1978, which limits
the use of natural gas as fuel for utilities
but not for cogenerators, combined to spur
the growth of cogeneration.
Because the panel
of electric power producers that report
to the various Federal Reserve Banks was
defined prior to the rapid growth of cogeneration,
the impact of cogeneration is not captured
in the data available to us.[7] For example,
a decline in the amount of electricity purchased
by a manufacturing firm from an electric
utility may be due to bringing a new cogeneration
system online and not due to a decline in
its actual power consumption.
Cogeneration is so
important in the chemicals industry that
the historical data on electric power sales
do not adequately proxy electric power usage
by that industry. Therefore, we treat the
chemicals industry as if it employed a single-factor
production process. This is unfortunate
because chemicals production is one of the
most capital-intensive manufacturing industries,
and it has the second largest share of value
added in manufacturing. The fact that cogeneration
is important in the petroleum refining industry
as well does not present a problem since
we employ a modified technique to estimate
output in that industry, as seen below.
We suspect that the unusual volatility in
the electric power series for paper and
allied products is due to cogeneration,
although we have not investigated this matter
closely.[8]
Factor shares.
We calculate
the labor share as the ratio of payroll,
as reported in the 1986 Annual Survey of
Manufactures, to total 1986 nominal gross
Texas product, as reported by BEA. The capital
share must be, according to our assumptions,
1 minus the labor share.[9] We assume the
factor shares to be invariant over the period
covered by the index. As noted above, we
assume that the labor share for chemicals
is equal to 1. Table 1 in the main text
reports the factor shares for all industries
included in TIPI whose output we estimate
using the Atlanta method.
Productivity.
The Atlanta method
requires monthly estimates of the productivity
of each factor of production. Output, and
therefore productivity, however, is available
only annually. We derive the monthly factor
productivity estimates by assuming that
productivity grows exponentially between
the annual observations. The rate of factor
productivity growth during the period after
the last actual annual observation is an
important methodological concern. The methods
that practitioners of regional production
indexes most commonly employ include extrapolating
a long-run productivity growth
rate, extrapolating the most recently
observed growth rate, and fixing productivity
at its most recently observed level.
None of these choices,
in our view, are good. Examining the data
indicates that productivity tends to rise
and fall as output rises and falls—that,
is, productivity is pro-cyclical. None of
the above-mentioned methods account for
cyclical movements in productivity; in fact,
they are likely to result in estimates of
production that understate the magnitude
of both peaks and troughs in the business
cycle. We adopted a technique to extrapolate
factor productivity in a manner that allows
incorporation of both trend and cyclical
components. We regress first differences
in annual real gross product for each industry
on first differences in annualized man-hours
and/or electric power usage. Based on these
results, we forecast annual output as closely
to the present as possible. Finally, we
use the forecasted values to compute factor
productivity. At present, gross product
data are available only through 1986, so
we must forecast them through 1988, using
this procedure. We assume constant factor
productivity growth after 1988.
Exceptions to the
Atlanta method
The Board of
Governors of the Federal Reserve System,
in constructing the U.S. Industrial Production
Index, emphasizes the desirability of collecting
actual production data whenever possible
instead of estimating output from labor
and capital data.[10] At the regional level,
very little such data are available. Nevertheless,
for several important industries, it is
possible to collect timely monthly data
that we can use to estimate monthly changes
in production more closely than we could
be using changes in labor data or electric
power usage.
Oil and gas extraction.
Actual monthly
data on Texas crude oil production, natural
gas production and the level of exploration
activity are available. We use these measures
to drive monthly movements in the index
for oil and gas extraction (SIC 13), but
we benchmark the series historically to
BEA’s estimates of gross product in
that industry. The method we use is as follows:
we follow the normal Atlanta method, except
that in performing the calculations, we
use oil production, gas production, and
the Hughes rotary rig count as if they were
the factors of production in a three-factor
production process. The “factor shares”
in this case are estimated shares in SIC
13 gross product attributable to the three
“factors.” It is useful to think
of this as a method that combines measures
of oil production, natural gas production,
and exploration activity into an overall
index for SIC 13, while constraining long-term
movements to follow those of gross production
in the industry.
Petroleum and coal
products. Although
a measure of input, not output, the amount
of crude petroleum refined is often used
as a measure of refining output.[11] In
the current revision of TIPI we introduced
a modified procedure. Again, we follow the
Atlanta procedure computationally, except
that here we perform the computation as
if refiners use a single-factor production
process, where the “factor”
is runs of crude petroleum. Crude runs to
refineries, therefore, strongly influence
month-to-month movements in the index for
petroleum and coal products, although the
long-run pattern must follow that of gross
product in the industry.
Electric and gas utilities.
BEA reports gross
state product strictly at the two-digit SIC
level. SIC 49 covers electric and gas utilities
and sanitary services. We prefer not to include
sanitary services in TIPI, both for the sake
of comparability with previous versions of
TIPI and for the sake of comparability with
the U.S. Industrial Production Index. We therefore
estimate annual Texas value added ourselves—for
electric utilities, from the income and expense
statements of Texas electric utility companies[12]
and, for gas utilities, from data reported
in Gas Facts, published by the American
Gas Association.
Timely
monthly data on total power generated by
Texas electric utility companies and natural
gas transmitted by gas utilities are available
from the Department of Energy and the Texas
Railroad commission, respectively. We use
these data to drive monthly movements in
TIPI’s electric and gas utility industries.
We employ the same
technique for these industries as we do
for petroleum and coal products. Movements
in electric power generation and natural
gas transmission strongly influence month-to-month
movements in the indexes for electric utilities
and gas utilities, respectively, but we
benchmark the series historically to estimates
of value added in these industries.
Miscellaneous considerations
Seasonal adjustment
and smoothing. We
seasonally adjust all primary monthly series
using a variant of the Census Bureau’s
X-11 procedure prior to incorporation into
the index calculation. In addition, to ensure
that monthly movements of all series contain,
on average, more information than statistical
noise, we convert the monthly series to
centered moving-average form wherever appropriate.[13]
Aggregation.
TIPI and other industrial production indexes
differ in how industry aggregates are calculated.
Ordinarily, production indexes are aggregated
using weights that are based on the distribution
of value added across industries. This is
necessary because one or more of the components
of the aggregate are not available in dollar
terms. For example, the amount of crude
oil refined is used as a proxy for refinery
production. It is interesting to note, however,
that because the industry weights are based
on a value-added distribution, a dollar
figure for value added is needed or at least
one year.
TIPI takes a somewhat
different approach. We construct monthly
real value-added series for all
industries. Therefore, it is a straightforward
procedure to sum the components into aggregates.
Only after we have formed all the real value-added
aggregates do we convert to index form.
The apparent avoidance of using industry
weights is illusory. In constructing real
value added, we are implicitly fixing relative
prices to be those existing in one single
year. In constructing TIPI, we denominate
value added in terms of 1986 dollars, that
is, we chose relative prices in 1986. Implicitly,
this is equivalent to using the
1986 value-added industry distribution to
weight the individual industry indexes
to form aggregates.[14] A corollary to this
is that it would be wrong to use 1986 price
deflators to construct real value added,
and then use 1982 value-added weights to
form the aggregates. Table 1 shows the relative
importance of all TIPI industries in total
industrial production according to BEA’s
1986 gross product estimates.[15]
Revision schedule
It is useful
to think of the Texas Industrial Production
Index as an ongoing experiment. As new techniques
or data become available, or when existing
data are revised by the issuing agency,
we will revise TIPI to incorporate the new
information.
At a minimum, we revise
TIPI annually, following the U.S. Bureau
of Labor Statistics’ annual two-year
revision of its Establishment Survey data.
Also at that time, we incorporate other
recently released or revised data, and we
update the seasonal adjustments. Finally,
whenever updated gross product data become
available from the Bureau of Economic Analysis,
we rebenchmark all series.
Sources of error
We will address
three main sources of inaccuracy. First,
TIPI can only be as accurate as the primary
data upon which it relies. As we mentioned
earlier, as these data are revised by the
issuing agencies, we will incorporate the
improved data into our TIPI estimates. Second,
in extrapolating factor productivities beyond
the most recently available gross product
data, statistical error is introduced. It
is possible for this error to be greater
than that generated by using alternative
procedures. For now, we prefer what we consider
to be a more accurate approach (that is,
to try to capture both trend and business-cycle-effects),
but we will reevaluate as we make future
revisions. Finally, the necessity of using
national price deflators is unfortunate.
Still regional price information may become
available.
- See Pyun (1970) and Strobel (1978).
- As stated earlier, the Federal Reserve
Bank of Dallas has produced the Texas
Industrial Production Index continuously
since 1958, using a variant of the Atlanta
method since 1983. The Federal Reserve
Bank of San Francisco introduced an index
in 1973, which is no longer in production,
also using the Atlanta method (Walsh and
Butler 1973). Since 1987, the Federal
Reserve Banks of Chicago, Cleveland, Richmond,
and Philadelphia have introduced manufacturing
indexes using the Atlanta method. See
Schnorbus and Israilevich (1987); Bryan
and Day (1987); Bechter, Chmura and Ko
(1988); and Hamer (1989).
- See Fomby (1986). Very recent research
conducted at the Federal Reserve Bank
of Chicago offers new insights into this
question. See Israilevich, Schnorbus,
and Schneider (1989).
- For a proof, see Chiang (1974, 407–10).
- For details, see Bureau of Economic
Analysis (1985).
- See Moody (1974).
- As this is written, the Board of Governors
of the Federal Reserve System is investigating
means to augmen
| | |
