an attempt to explain or predict actual interde-

Input–output analysis pendence between industries or sectors (Christ

1955).
Dazhong Cheng The theoretical foundation of input–output
Fudan University, China analysis, or the input–output model, is viewed
by Leontief as a general equilibrium theory.
Peter W. Daniels In its most basic form, an input–output model
University of Birmingham, UK consists of a system of n linear equations with
n unknowns which form a matrix that can
Input–output analysis is one of the most widely be readily used to represent the mathematical
applied analytical frameworks in economics. Its structure of the model. An essential condi-
fundamental purpose is to analyze the interde- tion for empirical input–output analysis is the
pendence of various industries or sectors in an input–output table, which was first used by
economy, such as agriculture, manufacturing, the French economist François Quesnay in
or services. The original idea for input–output Tableau Économique (Kuczynski and Meek 1972).
analysis can be traced as far back as the early Input–output analysis became a widely used tool
work of a British physician and Oxford pro- of economic analysis after a standardized system
fessor, William Petty, in the mid-seventeenth of economic accounts built around input–output
century. The first historic breakthrough was concepts was developed, initially by Leontief
made by a Russian-born US economist, Wassily (1941) and then by Richard Stone (1961), who
W. Leontief (1906–1999), in the late 1930s, for was awarded the Nobel Prize in Economic
which he received the Nobel Prize in Economic Science in 1984 (Miller and Blair 2009).
Science in 1973 (Miller and Blair 2009). The basic structure of an input–output table
records the flows of goods or services from
each sector (as a producer or seller) to each
Two pillars of input–output analysis
of the other sectors or itself (as consumers or
purchasers) (Table 1). The distinction commonly
Over the past eight decades, the evolution of made in input–output analysis is between the
input–output analysis has been based upon production of goods and services and their
two major pillars: theoretical foundation and utilization, as reflected in the four quadrants of
data construction. Therefore, some studies may the input–output table. Quadrant I shows the
be purely theoretical, exploring the formal intermediate transactions; that is, the flows of
relationships that can be derived under vari- goods and services that are both produced and
ous assumptions from sets of equations; other consumed in the process of current production.
studies may be purely descriptive and depen- Quadrant II shows the sales by the producing
dent upon the construction of input–output industries to final uses such as private consump-
tables; others will incorporate a mixture of both tion, government consumption, fixed capital
empirical data and theoretical relationships in formation, and net exports. Quadrants I and

The International Encyclopedia of Geography.

Edited by Douglas Richardson, Noel Castree, Michael F. Goodchild, Audrey Kobayashi, Weidong Liu, and Richard A. Marston.
© 2017 John Wiley & Sons, Ltd. Published 2017 by John Wiley & Sons, Ltd.
DOI: 10.1002/9781118786352.wbieg0071
Table 1 Basic structure of a national input–output table.
Intermediate use Final use
(producers as consumers) (final demand)
Output
Sector 1 Sector 2 ... Sector n Final Final Gross fixed Net
consumption consumption capital export
Input
expenditure expenditure by formation
by households government
Producers Sector 1
(intermediate
inputs) Sector 2
Quadrant I Quadrant II
...

Sector n
Primary inputs Employees
(value added)
Business owners Quadrant III Quadrant IV: GDP
and capital
Government

Source: Based on Miller and Blair (2009, 3). Reproduced from Cambridge University Press.
 I N P U T–O U T P U T A NA LYS I S

II together allocate the total output of each Two crucial advances can be identified in the
industry in the economy. Quadrant III shows evolution of these national-level input–output
the primary (factor) inputs that are required by tables. One is the compilation of regional and
each producing industry; these constitute the multiregional input–output (or MRIO) tables,
gross value added (including employee com- which provide the basis for impact analysis
pensation, depreciation of capital, and indirect and examining the interconnections between
business taxes) in each industry. Quadrants I different regions within a country. The other is
and III combined show the total inputs used for the construction of noncompetitive import type
production by each industry in the economy. input–output (or noncompetitive IO) tables, In
Quadrant IV records two equivalent measures these tables, domestically produced intermediate
of gross domestic product (GDP), one derived inputs and imported intermediate inputs are
from expenditure components, and the other treated separately, as they are presumed to be
from value-added components. poor substitutes for each other. If the inputs
imported by a country are differentiated further
Recent major developments by country of origin, it is possible to trace its
in input–output analysis external economic and trade relationship with
other countries (of which more below).
Second, as more and more countries have
The widespread availability of high-speed mod-
become engaged with the process of globaliza-
ern computers has made input–output analysis
tion, industrial linkages now reach well beyond
more comprehensive and practical. Among
national borders and it has become more difficult
the most important advances are the improved
implementation of computable general equilib- for the conventional national-level input–output
rium (CGE) models and the ability to continually tables to capture the details of global industrial
update and expand input–output tables. and value chains. Therefore, various attempts
CGE models use the input–output model as to harmonize input–output tables for differ-
a benchmark for working with actual economic ent countries and to construct international
data to estimate how an economy might react input–output tables are ongoing. Examples
to changes in policy, advances in technology, include the Asian IDE-JETRO tables compiled
or other external factors. So far, there are two by the Institute of Developing Economies and
versions of mathematical programming and opti- Japan External Trade Organization (Inomata and
mization software suitable for CGE modeling and Okamoto 2015), the GTAP tables constructed
analysis: the general algebraic modeling system by Purdue University (see https://www.gtap.
(GAMS) (see www.gams.com/) and the general agecon.purdue.edu/), and the recently released
equilibrium modeling package (GEMPACK) World Input–Output Database (WIOD), which
(see www.copsmodels.com/gempack.htm). covers 27 countries of the European Union
There have been two important developments and 13 other major countries in the world
in the use of input–output tables in recent for the period from 1995 to 2011. The latter
decades. First, since 1936, when Leontief first attempts to analyze the effects of globalization
presented US interindustry transactions tables on trade patterns, environmental pressures, and
for 1919 and 1929, an increasing number of socioeconomic development (see www.wiod.
countries and regions around the world have org/new_site/data.htm) (Timmer 2012). In
been routinely constructing input–output tables. contrast to a national input–output table, a world

3
I N P U T–O U T P U T A NA LYS I S

input–output table traces flows of products or classification of producer services cannot accu-
services for both intermediate and final use that rately detail their status, their role, and their
are divided into those that are produced domes- contribution to a national economy. In order
tically and those that are imported (see Table 2). to circumvent such problems, some researchers
have turned to input–output analysis (Khayum
An application of input–output analysis 1995; Windrum and Tomlinson 1999; Cheng
to producer services and Daniels 2014).
By way of illustration, two examples are pre-
sented here. First, the output of any one specific
In contrast to consumer services, producer
service sector must be classified into two parts:
services enter the production process of other
manufacturing and services firms as an input output consumed by final users and output that
(Grubel and Walker 1989; Stibora and de Vaal will be used by firms as an input. The ratio of the
1995). Firms can obtain producer services via former part in the output is the consumer ser-
two channels: in-house provision and/or market vices ratio, while the latter part (SI/SO = services
transactions. The former reflects the specialized input/services output) is the producer services ratio.
division of labor inside the producers or firms, This identifies the extent to which a specific
and thus the internal resource allocation and service-sector category is a producer service.
industrial linkages directed by firms’ decisions. A large producer services ratio means that more
Unfortunately, these types of producer services of the output of a specific service sector is being
cannot be captured by input–output tables in used as input to other parts of the economy; this
the System of National Accounts (SNA) and suggests that the sector is more like a producer
will frequently be grouped according to firms’ service. If the ratio is low for a specific service
main business (e.g., manufacturing). The latter, sector, it is considered to be functioning more
market-driven group reflects the specialization like a consumer service. As a general rule, a
of, and division of labor between, different firms ratio of 50% or above is considered the threshold
in the market, and hence the resource allocation for judging whether a service activity can be
and industrial linkages based on market com- characterized as a producer service. Data for the
petition. These types of producer services are USA (Table 3) show that the sectors with a pro-
captured in the SNA input–output tables, with
ducer services ratio averaging 50% or more are
the number of such records increasing as the
research and development (R&D) (about 90%),
outsourcing and marketization of producer ser-
other business activities (about 80%, including
vices, a natural evolution of specialized division
of labor and resource allocation from inside the knowledge-intensive business services (KIBS)
firm to the market, continues. such as consulting, design, legal services, etc.),
However, from a statistical perspective it is transport and storage (70%), post and telecom-
difficult to distinguish between producer services munications (over 60%), and finance and insur-
and consumer services; activities such as banking ance (above 55%). The remaining sectors with
or transport not only fulfill intermediate demand values below 50% are more closely aligned with
but also meet the needs of final consumers, consumer services. For China in the mid-2000s
even though individual firms may emphasize (see Table 3) six activities have more than 60% of
the provision of services to one group or the their output used as producer services: wholesale
other. As a result, research that uses an arbitrary and retail trade and repairs, hotels and restaurants,

4
Table 2 Basic structure of world input–output table with three economies.
Intermediate use Final use/consumption
Output Country A Country B Rest of World Country A Country B RoW
(RoW)
Input Sectors: Sectors: Sectors:
1 ... n 1 ... n 1 ... n
Intermediate Country A Sectors: Intermediate Intermediate Intermediate Final use of Final use by Final use by
inputs 1 ... n use of use by B of use by RoW domestic B of RoW of
domestic exports of exports output exports exports
output from A from A from A from A
Country B Sectors: Intermediate Intermediate Intermediate Final use by Final use of Final use by
1 ... n use by A of use of use by RoW A of domestic RoW of
exports domestic of exports exports output exports
from B output from B from B from B
RoW Sectors: Intermediate Intermediate Intermediate Final use by Final use by Final use of
1 ... n use by A of use by B of use of A of B of domestic
exports exports domestic exports exports output
from RoW from RoW out put from RoW from RoW
Primary inputs Value-added in Value-added in Value-added in
(value-added) A B RoW

Source: Based on Timmer (2012).

I N P U T–O U T P U T A NA LYS I S

transport and storage, post and telecommunica- (downstream) sectors will depend on sector i (as
tions, finance and insurance, and other business a seller) (e.g. transport services)? The greater this
activities. The comparison suggests that although coefficient, the higher the pressure of demand
R&D behaves more as a producer service in the experienced by sector i.
United States, in China it is more akin to a con- Indeed, there has been considerable literature
sumer service, along with real estate activities, arguing about the linkage measures and offering
education, and health and social work. On the numerous suggestions for differing definitions
other hand, wholesale and retail trade and repairs and refinements of them (Beyers 1976; Miller
and hotels and restaurants in China act more as and Blair 2009). Our aim here is just to intro-
producer than consumer services. duce the simplest forms of these measures and
Input–output analysis is also used to compute in particular to show how they are derived from
two linkage coefficients: the backward linkage information in the IO table.
coefficient (BLj ) It is common for the same kind of services
( n )/[ ( n n )] (e.g. transport services) to be used as both
∑ ∑∑ upstream and downstream services. Referring
BLj = bij 1∕n bij (1) again to Table 3, in the mid-2000s, R&D,
i=1 i=1 j=1
other business activities, education, and health
and the forward linkage coefficient (FLi ) and social work have larger backward linkage
( n )/[ ( n n )] coefficients in China than in the USA. How-
∑ ∑∑ ever, the backward linkage coefficients of real
FLi = bij 1∕n bij (2) estate activities and finance and insurance are
j=1 i=1 j=1
obviously much smaller (only 0.57 and 0.74,
(from the Leontief inversion matrix B = (bij )n×n = respectively) in China. The values of the forward
(I − A)−1 , the Leontief complete consumption linkage coefficients for wholesale and retail trade
coefficients bij can be obtained, where A is an and repairs, transport and storage, and other
input coefficient matrix based on Quadrant I in business activities in China are greater than 1,
Table 1 and I is an identity matrix). The first while the values for the other sectors are much
coefficient measures the backward economic lower; five scored less than 0.5 in the mid-2000s.
linkage of one specific sector j to the rest of the In the USA, the forward linkage coefficient
economy; that is, when the output of sector j for R&D was well above 1.7 from the early
(e.g. transport services) increases by 1 unit, how 2000s, approximately four times the equivalent
much of the increased demand from sector j value for China. Indeed, almost all the service
(as a purchaser) will rely on (upstream) sectors sectors in China, except hotels and restaurants
(e.g. transport vehicles) whose outputs are used and transport and storage, have much smaller
as inputs to production in sector j? The greater forward linkage coefficients.
the value of this coefficient, the stronger the The fact that every sector of the economy in
pulling power of sector j on the rest of the China depends to a significant degree on hotels
economy. The second coefficient measures the and restaurants and on transport and storage again
forward economic linkage of one specific sector confirms that these services are utilized more as
i to other sectors; that is, when the output in producer services and have therefore performed a
every sector of the economy increases by 1 unit, key role in industrialization. Once again, the for-
how much of the increased demand from these ward linkage coefficient for real estate activities

6
Table 3 Application of input–output analysis: China and the USA.
Producer services ratios (SI/SO) BLj ) FLj )
Early 2000s Mid-2000s Early 2000s Mid-2000s Early 2000s Mid-2000s
Service sector China USA China USA China USA China USA China USA China USA
Wholesale and retail 66.42 30.21 61.96 30.14 0.97 0.85 0.8 0.8 1.65 1.9 1.29 1.93
trade and repairs
Hotels and 62.17 22.11 67.82 21.16 1.01 1.04 1.01 0.97 0.77 0.73 0.9 0.68
restaurants
Transport and 80.46 70.17 77.96 70.81 0.94 1.02 0.97 0.95 1.17 1.54 1.72 1.47
storage
Post and telecom- 81.14 61.88 75.29 65.88 0.97 1.04 0.97 1.02 0.79 1.19 0.84 1.19
munications
Finance and 76.1 56.63 74.59 56.05 0.66 0.97 0.74 0.85 1.02 1.79 0.9 1.73
insurance
Real estate activities 31.14 29.87 20.13 32.05 0.8 0.79 0.57 0.77 0.54 1.25 0.49 1.3
Research and – 89.22 37.29 89.65 – 0.91 1.09 0.88 – 1.71 0.4 1.83
development
(R&D)
Other business 54.77 79.9 76.22 78.78 1.06 0.85 1.14 0.78 1.12 1.38 1.07 1.38
activities
Education 12.55 16.52 9.46 17.38 0.89 0.95 0.78 0.86 0.47 0.57 0.42 0.53
Health and social 4.49 2.25 13.18 2.09 1.18 0.91 1.16 0.84 0.42 0.55 0.45 0.51
work

Source: Calculation based on input–output tables of China and the USA. “–” denotes data vacancy. Both coefficients have in their denom-
inator the average value of a coefficient in the inverse matrix.
I N P U T–O U T P U T A NA LYS I S

in China is less than 0.5 and is just one-third of edited by Conference on Research in Income and
the equivalent USA value. It is clear that Chinese Wealth, 137–182. Princeton: Princeton University
real estate activities cannot exert strong pulling or Press.
pushing powers on the rest of the economy: this Grubel, Herbert, and Michael Walker. 1989. Ser-
sector functions independently from (or is not vice Industry Growth: Causes and Effects. Vancouver:
Fraser Institute.
strongly connected to) the rest of the economy
Inomata, Satoshi, and Nobhiro Okamoto. 2015.
since both linkage coefficients are well below 1. “Measuring the Globalization of Industrial Link-
To sum up, subject to the availability of suitable ages.” http://www.ide.go.jp/English/Research/
and reliable data, an input–output analysis can Topics/Eco/Io/(accessed October 7, 2015).
greatly improve our understanding of the devel- Khayum, Mohammed. 1995. “The Impact of Ser-
opment and contribution of producer services to vice Sector Growth on Intersectoral Linkages in the
economic change and restructuring, especially at United States.” The Service Industries Journal, 15(1):
the national scale but also at the global scale as 35–49.
the appropriate models for analysis begin to be Kuczynski, Marguerite, and Ronald Meek, eds. 1972.
introduced. Quesnay’s Tableau Économique. London: Macmillan.
Leontief, Wassily W. 1941. The Structure of American
Economy 1919–1929. New York: Oxford Univer-
SEE ALSO: Deindustrialization; Demand and sity Press.
supply for producer services; Externalization; Miller, Ronald, and Peter Blair. 2009. Input-Output
Analysis: Foundations and Extensions, 2nd edn. Cam-
Factors of production; Global commodity/value
bridge: Cambridge University Press.
chains; Global production networks; Stibora, Joachim, and Albert de Vaal. 1995. Services
Globalization; Industrial linkage; International and Services Trade: A Theoretical Inquiry. Amsterdam:
division of labor; Producer services: definition Thesis Publishers.
and classification; Value added Stone, Richard. 1961. Input-Output and National
Accounts. Paris: Organization for Economic Coop-
eration and Development.
References Timmer, Marcel, ed. 2012. “The World
Input-Output Database (WIOD): Contents,
Beyers, William B. 1976. “Empirical Identification of Sources and Methods.” WIOD Working Paper
Key Sectors: Some Further Evidence.” Environment Number 10. http://www.wiod.org/publications/
and Planning A, 8(2): 231–236. papers/wiod10.pdf (accessed October 7, 2015).
Cheng, Dazhong, and Peter Daniels. 2014. “What’s Windrum, Paul, and Mark Tomlinson. 1999.
So Special about China’s Producer Services?” “Knowledge-Intensive Services and International
China and World Economy, 22(1): 103–120. Competitiveness: A Four Country Comparison.”
Christ, Carl. 1955. “A Review of Input-Output Technology Analysis and Strategic Management, 11(3):
Analysis.” In Input-Output Analysis: An Appraisal, 391–408.