Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
23
Part II. Input-Output Analysis of IT Producing Sectors
Yusuke YAMAMOTO
Akihiko SHINOZAKI
Abstract
In Part II we present the results of our input-output analysis of the IT industry in the broad
sense (content, communication, broadcasting and equipment) from 1990 to 2005. This analysis
reveals that both induced production and induced value-added of the IT industry have increased
overall since 1990. Although induced value-added of the IT industry decreased from 2000 to
2005, it was still greater than that of the automobile industry. Looking back further, the disparity
between the IT industry and the automobile industry widened from 1990 to 2005. Analysis using
disaggregated data reveals that the equipment sector was a factor not only in decreasing
economic ripple effects but also economic ripple effects per unit in the IT industry. An increase
in imports in the equipment sector caused a decrease in economic ripple effects. In contrast,
software-centric information services were a factor in the increase of the content sector,
particularly in the IT industry. We can conclude that the IT industry was a key sector in the
expansion of the economy; and software, not hardware (equipment) was a key sector in the IT
industry from 1990 to 2005.
1. Introduction
The demand for IT producing sectors creates a number of services and products. In other
words, demand for IT producing sectors causes a number of economic ripple effects. For
example, non-IT production and services such as plastic products, electric power, transport
24
services, etc. are required to produce cellular phones. Similarly, IT production and services such
as software, communication, etc. are also needed. IT producing sectors use not only IT
production and services; they also use non-IT production and services. Demand for IT producing
sectors results in increased demand for IT production and services. So it is important to analyze
the scale and trends of the economic ripple effects of IT producing sectors (the IT industry)
within Japanese industries.
To verify the economic ripple effects, in Part II we show the results of an input-output
analysis of the IT industry9 from 1990 to 2005 that analyzed the economic ripple effects caused
by final demand10 for the IT industry. From this analysis we can see how important the IT
industry is to the Japanese economy. We can further pinpoint changes in the IT industry by using
a more precise classification of data.
2. Analytical framework
Using the Input-Output analysis method to determine the economic ripple effects we
calculated induced production, value-added, and employment.
Equation (1) represents induced production.
(1) F)A)M(I(IX 1ˆ −−−= ,
where X is induced production, I is the unit matrix, M̂ is the import coefficient matrix
(diagonal matrix whose main diagonal is import/domestic demand), A represents the input
9 Regarding the input-output analysis of Japan’s information economy, researchers have analyzed the problems with information sectors within an organization and we are not concerned with this issue. For detailed arguments, see Hiromatsu, et al. (1990). 10 We analyzed the sum of induced production or induced value-added. It is beyond the scope of this analysis to verify which sectors these induced production or induced value-added affect. This problem concerns the external multiplier and the internal multiplier. For detailed arguments, see Kuriyama, et al. (2001).
25
coefficient matrix, F stands for final demand that generates economic ripple effects and F
equals E)YM(I +− ˆ , where Y is domestic final demand, and E stands for Exports.
Induced value-added and employment are calculated using induced production. Equation
(2) represents induced value-added.
(2) XVVA ˆ=
where V̂ is the value-added coefficient matrix (diagonal matrix whose main diagonal is
value-added/domestic production), value-added in our study does not include "Consumption
expenditure outside of households," hence the sum of value-added for each industry equals the
GDP. We likewise calculated induced employment; however, since the latest data available
(2005 quick estimation) does not contain employment data, induced employment is excluded.
In calculating ripple effects per unit, we used final demand including imports. The induced
production coefficient (induced production per unit) of final demand is the sum of induced
production divided by the sum of final demand, which includes imports ( EY + ). The induced
value-added coefficient for the IT industry (induced value-added per unit) is calculated in the
same way.
3. Dataset
All datasets described in this section are taken from the Ministry of Internal Affairs and
Communications Input-Output Tables (I-O Tables). Data for the 1990 to 1995 period emanate
from the 1990-1995-2000 Linked I-O tables; The 2000 I-O tables provide the 2000 year data;
and 2005 data are from 2005 I-O Tables quick estimation.
We aggregated I-O Tables data and designed a square matrix to calculate Equations (1) (2).
26
Table 2-1 shows the number of sectors in each I-O table and sector classification within the IT
industry.
(Table 2-1)
The IT industry is divided into 4 categories: content, communication, broadcasting and
equipment. Though Hiromatsu, et al. (2007) analyze a similar problem, their definition11 of the
IT industry is narrower than our definition. The reason for the inclusion of broadcasting and
content is the convergence of media and communications. We chose IT sectors from the 2005
I-O Tables Quick Estimation first12, then we selected the corresponding sectors from other I-O
tables. As Table 2-1 shows, since classification in 2005 I-O Tables Quick Estimation is limited
compared to other years, sectors such as toys and games are not included even though they are
classified as IT related sectors. Toys and games are included in the miscellaneous manufacturing
products category. Miscellaneous manufacturing products include: sporting and athletic goods;
musical instruments; audio and video records; other information recording media; stationery;
jewelry and adornments; "Tatami" (straw matting) and straw products; ordnance; and
miscellaneous manufacturing products, which are not part of the IT industry. Since the sector
classification of 2005 data is rough, the 2005 estimation is less accurate than other years.
In addition, automobile manufacturing is generally regarded as a core industry, so we use
the automobile industry13 for a comparative analysis.
11 Hiromatsu, et al. (2007) defined as communications, equipment, and a portion of content. 12 Internet based services are newly included in the 2005 I-O Tables and we have included them in the IT industry. These services account for less than 2% of domestic production in the IT industry and any influence is minor. 13 The automobile industry is an aggregation of passenger motor cars, other cars and motor vehicle parts and accessories in the classification of 2005 I-O Tables quick estimation and is an aggregation of passenger motor cars, trucks, buses and other cars, two-wheeled motor vehicles, motor vehicle bodies, internal combustion engines for motor vehicles and parts and motor vehicle parts and accessories in the classification of other I-O Tables.
27
4. Economic ripple effects of the IT industry
4-1. Induced production and induced value-added of the IT industry
Based on the formula and dataset described above, we calculated the economic ripple
effects14 of the IT industry and the automobile industry. Table 2 shows the economic ripple
effects from 1990 to 2005.
(Table 2-2)
According to Table 2-2, final demand for the IT industry increased from 1990 to 2000, but
decreased from 2000 to 2005. However, final demand for the IT industry is greater than that for
the automobile industry. Though final demand for the IT industry decreased, it is still greater
than for the automobile industry in 2005.
Induced production in the IT industry was 50 trillion yen in 1990 and increased to 55 trillion
yen in 2005. In the year 2000, at the peak of the IT boom, induced production in the IT industry
was greater than in the automobile industry. But induced production of the IT industry decreased
to less than the automobile industry from 2000 to 2005. Induced value-added of the IT industry
was 21 trillion yen in 1990, which is 2 trillion yen greater than that of the automobile industry.
This disparity between the IT and automobile industries grew to 6 trillion yen in 2005, and
suggests that the IT industry has become the main factor in the expansion of the economy.
Although the overall economic ripple effects of the IT industry increased, growth varied
from sector to sector. Figure 2-1 illustrates induced production in the IT industry by category.
(Figure 2-1)
The reason for the induced production decrease in the IT industry in 2005 is the 12 trillion
14 In what follows, the phrase "economic ripple effects" means induced production and induced value-added.
28
yen decrease in induced production of equipment. Though equipment accounted for a large share
(75%) of total induced production within the IT industry in 1990, the share of equipment
decreased to 41% in 2005. On the other hand, induced production of content and communication
rose. The increase of content is particularly noticeable. Content, not equipment, is going to play
a central role in the IT industry. An equipment decrease and content increase are notable in
induced value-added. Figure 2-2 shows induced value-added of the IT industry according to
category. In 2005, content induced value-added increased and became greater than equipment
and the difference between equipment and the communications industry decreased.
(Figure 2-2)
Since equipment and content data are classified in detail, it is possible for a more in-depth
analysis of economic ripple effects. Figure 2-3 shows induced production in the IT industry
according to sector, as found in tables 2005 I-O.
(Figure 2-3)
As for equipment, induced production of communication equipment and related products is
greater than electronic computing and accessory equipment. The most important point about
content is that induced production of information services increased notably from 1990 to 2005.
The information services sector accounts for a large share of induced production of content, and
induced production of information services is greater than other sectors. But induced production
of information services almost equals that of communication equipment and related products.
The scale of information services is notable in induced value-added. Figure 2-4 shows induced
value-added of the IT industry by sector as found in Table 2005 I-O.
(Figure 2-4)
29
Induced value-added of information services is greater than other sectors and the difference
between information services and the second-largest sector (communication) was 2 trillion yen
in 2005. The fact that induced value-added of communication is greater than communication
equipment and related products, as well as electronic computing equipment and accessory
equipment, verifies that the equipment induced value-added decrease was significant in 2005.
Since software is a significant segment of information services, software, rather than hardware
(equipment), was a key sector in the IT industry from 1990 to 2005.
It is important to note that an increase in imports caused a decrease in economic ripple
effects. Figure 2-5 illustrates domestic production by category and the import penetration rate
(import coefficient) of communication equipment and related products.
(Figure 2-5)
The import penetration rate of communication equipment and related products increased
from 1990 to 2005. Imports, which do not cause economic ripple effects, accounted for 20% of
domestic demand in 2005. An increase in imports caused a decrease in economic ripple effects.
This decrease is more serious in the case of electronic computing and accessory equipment.
Figure 2-6 shows domestic production by category and the import penetration rate of electronic
computing equipment and accessory equipment.
(Figure 2-6)
Between 1990 to 2005, the increase in the import penetration rate of electronic computing
and accessory equipment was greater than communication equipment and related products.
Imports accounted for 70% of domestic demand in 2005 and the economic ripple effects loss
was substantial. It is clear that an increase in equipment imports caused a decrease in economic
30
ripple effects in the IT industry.
4-2. Induced production coefficient and induced value-added coefficient for the IT
industry
Table 2-3 shows economic ripple effects per unit (induced production coefficient and
induced value-added coefficient) from 1990 to 2005. Economic ripple effects per unit in the
automobile industry are greater than in the IT industry. It is important to note that the induced
production coefficient for the IT industry decreased from 1990 to 2005, although the trend of
induced production increased. On the other hand, the induced production coefficient for the
automobile industry increased from 1995 to 2005. As for induced value-added from 1990 to
2005, both IT and automobile industries showed decreasing trends. Trends of economic ripple
effects per unit vary from sector to sector. Figure 2-7 shows the induced production coefficient
for the IT industry by category.
(Figure 2-7)
It should be stressed that the induced production coefficient of equipment decreased from
2.2 (in 1990) to 1.6 (in 2005). Though the induced production coefficient for content decreased
from 1990 to 2005, the induced production coefficient for the equipment sector decreased more.
The induced production coefficient for communication and broadcasting did not decrease from
1990 to 2005. It is clear that equipment is the main factor in the falling off of the induced
production coefficient of the IT industry.
Figure 2-8 shows the induced value-added coefficient for the IT industry by category.
(Figure 2-8)
31
The induced value-added coefficient for equipment decline from 1990 to 2005 is significant.
Similar to the trend seen in the production coefficient, equipment is the main factor in the
decreasing induced value-added coefficient for the IT industry
It makes sense to use data with detailed classification for an in-depth analysis. Figure 2-9
shows the induced production coefficient for the IT industry by sector as seen in Table 2005 I-O.
(Figure 2-9)
As for equipment, the decrease in the induced production coefficient for communication
equipment and related products from 1990 to 2005 was greater than electronic computing and
accessory equipment. It is reasonable to assume that the increase in the import penetration rate
(Figure 2-6) caused the induced production coefficient decrease. As for content, the information
services induced production coefficient increased from 1995 to 2005. It is important to note that
the information services sector was not only the main factor in the increase of the economic
ripple effects of the IT industry; it was also the main factor in the increase of the economic ripple
effects per unit. Figure 2-10 illustrates that the main increase factor for the induced value-added
coefficient was also information services. In addition, the decrease in the induced value-added
coefficient for communication equipment and related products and electronic computing and
accessory equipment is quite notable.
We can conclude that software accounted for a substantial share of the information services
increase and hardware decrease from 1990 to 2005, not only in economic ripple effects but also
in economic ripple effects per unit.
32
5. Conclusion
With this research we have examined the economic ripple effects of the IT industry,
including content, communication, broadcasting and equipment. Our analyses revealed that both
induced production and induced value-added of the IT industry have increased since 1990.
Although induced value-added of the IT industry decreased from 2000 to 2005, it was greater
than that of the automobile industry. Looking back further, the disparity between the IT industry
and the automobile industry increased from 1990 to 2005. From 2000 to 2005, the decreasing
factor for not only the economic ripple effects but economic ripple effects per unit of the IT
industry was the equipment sector. An increase in imports in the equipment sector caused a
decrease in the economic ripple effects. In contrast, the content sector was an increasing factor in
the IT industry. The increase of software-centric information services is especially notable. We
can conclude that the IT industry has become the main sector in the expansion of the economy,
and software, not hardware (equipment), was a key sector in the IT industry from 1990 to 2005.
33
References
Kuriyama, Tadashi, Atsuko Ishikawa, Yan Cheng (2001) “Joho Keizai no Sangyo Renkan
Bunseki (Input-output Analysis of the Information Economy)”, Innovation and IO
Technique – Business Journal of PAPAIOS, Vol.10, No.2, December 2001, pp.4-17, in
Japanese.
Shinozaki, Akihiko (2003) “Tsuushin Sangyo ni Okeru Setsubitoushi no Keizai Koka Bunseki
(Economic Impact of Investment in Telecom Industry)” InfoCom Research, Inc., InfoCom
REVIEW, No.31, August 2003, pp.36-45, in Japanese.
Hiromastu, Takeshi, Akihiko Shinozaki, Yusuke Yamamoto (2007) “Joho Netowaku Sangyo no
Keizai Hakyu Koka (Economic Ripple Effects of The Information Network Industry)”
InfoCom Research, Inc., InfoCom REVIEW, No.43, December 2007, pp.30-35, in
Japanese.
Hiromastu, Takeshi, Gosei Ohira (1990) Joho Keizai no Makuro Bunseki (Macro Analysis of The
Information Economy), Toyo Keizai Shinposha., June 1990, in Japanese.
34
Tables and Figures
Table 2-1. Sector classification in the IT industry in Tables I-O
Category
Sector Classification in 1990-1995-2000Linked I-O Tables
(393sectors)
Sector Classification in 2000 I-O Tables
(399sectors)
Sector Classificationin 2005 I-O TablesQuick Estimation
(108sectors)
Newspapers Newspapers Video picture, character informationproduction and distribution
Publishing Publishing Video picture, character informationproduction and distribution
Advertising services Advertising services Advertising services
Information services Information services Information servicesNews syndicates and privatedetective agencies
News syndicates and privatedetective agencies
Video picture, character informationproduction and distribution
Motion picture and videoproduction, and distribution
Motion picture and videoproduction, and distribution
Video picture, character informationproduction and distribution
- - Internet based servicesPostal service Postal service CommunicationTelecommunication Fixed telecommunication CommunicationTelecommunication Mobile telecommunication CommunicationTelecommunication Other telecommunication CommunicationOther services relating tocommunication
Other services relating tocommunication Communication
Public broadcasting Public broadcasting BroadcastingPrivate broadcasting Private broadcasting BroadcastingCable broadcasting Cable broadcasting Broadcasting
Electric audio equipment Electric audio equipment Communication equipment andrelated products
Radio and television sets Radio and television sets Communication equipment andrelated products
Video recording and playbackequipment
Video recording and playbackequipment
Communication equipment andrelated products
Personal Computers Personal Computers Electronic computing equipmentand accessory equipment
Electronic computing equipment(except personal computers)
Electronic computing equipment(except personal computers)
Electronic computing equipmentand accessory equipment
Electronic computing equipment(accessory equipment)
Electronic computing equipment(accessory equipment)
Electronic computing equipmentand accessory equipment
Wired communication equipment Wired communication equipment Communication equipment andrelated products
Cellular phones Cellular phones Communication equipment andrelated products
Radio communication equipment(except cellular phones)
Radio communication equipment(except cellular phones)
Communication equipment andrelated products
Other communication equipment Other communication equipment Communication equipment andrelated products
Content
Communication
Broadcasting
Equipment
35
Table 2-2. Economic ripple effects
(Billion yen)1990 1995 2000 2005
IT industry 24,862 27,466 37,030 34,162Automobile industry 21,619 19,261 19,479 24,472IT industry 49,678 50,404 63,965 55,671Automobile industry 60,539 53,458 54,591 70,458IT industry 20,892 22,017 27,854 24,990Automobile industry 18,188 16,253 16,166 19,088
InducedValue Added
Final Demand
Induced Production
Source: Author’s calculation.
Figure 2-1 Induced production of the IT industry by category
0
10,000
20,000
30,000
40,000
50,000
60,000
70,000
80,000
1990 1995 2000 2005
(Billion yen)
Contents Communication BroadcastingEquipment IT Industry Automobile industry
Source: Author’s calculation.
36
Figure 2-2. Induced value-added of the IT industry by category
0
5,000
10,000
15,000
20,000
25,000
30,000
1990 1995 2000 2005
(Billion yen)
Contents Communication BroadcastingEquipment IT Industry Automobile industry
Source: Author’s calculation.
Figure 2-3. Induced production of the IT industry by sector in 2005 I-O tables
0
3,000
6,000
9,00012,000
15,000
18,000
21,000
1990 1995 2000 2005
(Billion yen)
Information services Video picture, character information production and distribution
Advertising services Internet based services
Communication Broadcasting
Communication equipment and related products
Electronic computing equipment and accessory equipment
Source: Author’s calculation.
37
Figure 2-4. Induced value-added of the IT Industry by sector in 2005 I-O tables
01,0002,0003,0004,0005,0006,0007,0008,0009,000
1990 1995 2000 2005
(Billion yen)
Information services Video picture, character information production and distribution
Advertising services Internet based services
Communication Broadcasting
Communication equipment and related products
Electronic computing equipment and accessory equipment
Source: Author’s calculation.
Figure 2-5. Domestic production by category and import penetration rate of communication
equipment and related products
-2,000
0
2,000
4,000
6,000
8,000
10,000
12,000
1990 1995 2000 2005
(Billion yen)
-5
0
5
10
15
20
25
30(%)
Exports Imports
Domestic Demand Domestic Production
Import Penetration Rate (right axis)
Source: Author’s calculation.
Note: Since domestic production equals domestic demand + exports-imports, the value of imports is negative.
38
Figure 2-6. Domestic production by category and import penetration rate of electronic computing
equipment and accessory equipment
-4,000
-2,000
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
1990 1995 2000 2005
(Billion yen)
-20
-10
0
10
20
30
40
50
60
70(%)
Exports Imports
Domestic Demand Domestic Production
Import Penetration Rate (right axis)
Source: Author’s calculation.
Note: Since domestic production equals domestic demand + exports-imports, the value of imports is negative.
Table 2-3. Economic ripple effects per unit
1990 1995 2000 2005IT industry 1.998 1.835 1.727 1.630Automobile industry 2.800 2.775 2.803 2.879IT industry 0.840 0.802 0.752 0.732Automobile industry 0.841 0.844 0.830 0.780
Induced Production
InducedValue Added
Source: Author’s calculation.
39
Figure 2-7. Induced production coefficient for the IT industry by category
1.25
1.50
1.75
2.00
2.25
2.50
2.75
3.00
1990 1995 2000 2005
Contents Communication BroadcastingEquipment IT Industry Automobile industry
Source: Author’s calculation.
Figure 2-8. Induced value-added coefficient for the IT Industry by category
0.50
0.60
0.70
0.80
0.90
1.00
1990 1995 2000 2005
Contents Communication BroadcastingEquipment IT Industry Automobile industry
Source: Author’s calculation.
40
Figure 2-9. Induced production coefficient for the IT industry by sector in 2005 I-O tables
1.2
1.4
1.6
1.8
2.0
2.2
2.4
1990 1995 2000 2005Information services Video picture, character information
production and distributionAdvertising services Internet based services
Communication Broadcasting
Communication equipment and related products
Electronic computing equipment and accessory equipment
Source: Author’s calculation.
Figure 2-10. Induced value-added coefficient for the IT industry by sector in 2005 I-O tables
0.300.40
0.500.600.700.80
0.901.00
1990 1995 2000 2005
Information services Video picture, character information production and distribution
Advertising services Internet based services
Communication Broadcasting
Communication equipment and related products
Electronic computing equipment and accessory equipment
Source: Author’s calculation.
41
Part III. Growth Accounting Analysis
Akihiko SHINOZAKI
Abstract
In Part III we show the results of a growth accounting analysis of the past 30 years,
reviewing the contribution of information technology to economic growth. This analysis yielded
two observations. First, Japan experienced a massive IT investment boom in the late 1980s and a
resultant productivity surge in both aggregate labor productivity and total factor productivity
(TFP). Second, the investment boom, however, ended abruptly in the early 1990s when new
types of open-network technology surged throughout the world. Since then information
technology has not contributed to changes in productivity growth. We can conclude, therefore,
that there is neither a “Solow Paradox” or a “New Economy” in Japan.
1. Introduction
A controversial discussion has arisen related to the potential growth rate of the Japanese
economy. The majority view claims that the economy can grow at merely one and half percent
annually at most15, although some analysts argue that it is feasible to raise the growth rate to
around three percent annually16. In this argument, the major difference between pessimism and
optimism derives, apparently, from whether the Japanese economy can reap the benefits of
globalization and innovation in information technology.
As described herein, we specifically examined the magnitude of the effects of information
technology on Japanese economic growth because recent empirical studies revealed that
15 See, for example, Council on Economic and Fiscal Policy (2005).
42
information technology contributed without a doubt to the surge in productivity in the United
States and its consequent economic growth since the mid-1990s17. In the U.S., a driving force of
this drastic change has been massive investment in information technology since the early 1990s.
Eventually, the consensus formed that a “new economy” emerged, as the “Solow paradox,”
(from Solow’s famous quip, “You can see the computer age everywhere but in the productivity
statistics18,”) evaporated in the United States.
Japan, in contrast, experienced a “lost decade” in the 1990s, when business investment was
sluggish and the economy grew at only 1.3 percent annually. The matter in question in the
contrast between Japan and the U.S. is whether Japan’s investment in information technology
contributed to its economic growth over the last few decades and what will happen over the next
few decades. To address this question, we conducted a growth accounting analysis of the
previous 30 years, reviewing Japan’s economic growth and the contribution of information
technology to that growth. We then studied the periodic changes in Japan’s productivity and IT
investment to assess whether the “new economy” as well as the “Solow paradox” holds true for
Japan.
2. Analytical framework
In this part, we used a growth accounting method pioneered by Solow (1957). This
method is based on the framework of a neoclassical production function that estimates the
contributions to output per hour derived from increases in capital assets per hours worked and
16 See, for example, Adams, et al. (2007). 17 For detailed arguments, see Jorgenson, et al. (2008), Oliner, et al. (2007). 18 See Solow (1987). Until the early 1990s, most empirical studies of the U.S. economy found no evidence of a positive correlation, and some found a negative correlation, between IT and productivity (U.S. Department of Labor
43
total factor productivity (TFP), where TFP is estimated as a residual of technological or
organizational improvements that increase output for a given amount of input.
Equation (3-1) presents the fundamental concept of the growth accounting method, capital
assets divided into IT and non-IT assets, where IT assets include not only computer hardware but
also software and network infrastructure. One reason for this is that intangible assets have
become more important. Another is that recent extraordinary innovations have resulted from the
convergence of computers and telecommunications equipment, as in:
(3-1) Q=TKoαKi
βLγ ,
where α, β, and γ respectively represent income shares of inputs such that α+β+γ=1.
Furthermore, Q is the private output, T is TFP, Ko represents non-IT capital assets, Ki denotes IT
capital assets, and L is labor input representing work hours of total employees. Consequently,
eqn. (3-1) can be transformed to
(3-2) Q.
-L.=T.+α(K
.o-L.)+) β(K
.i-L.),
where a dot over a variable denotes the rate of change expressed as a log difference. In eqn. (3-2),
Q.
-L. represents changes in output per hour, or average labor productivity, T
. stands for changes in
TFP, and K.
-L. represents changes in capital assets per hours worked, labeled capital deepening.
The capital-deepening component is further divided into the contribution from IT assets and
other non-IT assets in eqn. (3-2).
The basic equation presented above must be adjusted for the following two factors. The first
is the business cycle effect. Productivity is well known to be so pro-cyclical that the structural
trend of productivity must be distinguished from business-cycle-related changes to productivity.
[1994]). Therefore, it is likely that the “Solow paradox” pertained there.
44
For this discussion, the utilization rate of capital assets was used as a proxy of business cycle
effects to remove the influence of the business cycle from labor productivity. The second
adjustment we made was to consider labor quality. An important trend that drives economic
development is knowledge. In a knowledge-based economy, economic prosperity depends
deeply on labor quality as well as capital stock and technology. We employed education records
as a proxy of labor quality for these analyses. Therefore, eqn. (3-1) can be modified so
(3-3) Q= T(pKo)α(pKi)β(eduL)γ,
that p is the utilization rate of capital assets, assuming that the utilization rate is homogeneous in
each asset, and edu signifies the education records of employees as a proxy of labor quality.
Consequently, eqn. (3-3) can be transformed into the expression shown below.
(3-4) Q.
-L.=T.+α(K
.o-L.)+ β(K
.i-L.)+(α+β)p
.+ γ ed
.u
Here, we can measure the contributions to changes in labor productivity, or output per
hour, through decomposition into four factors: changes in TFP (T.), non-IT capital assets per
hours worked (capital deepening of non-IT: K.
o-L.), IT capital assets per hours worked (capital
deepening of IT: K.
i-L.), the utilization rate of capital assets (p
.) as a proxy of the business cycle
effect, and the education records of employees (ed.u) as a proxy of labor quality.
3. Dataset and overview of IT investment in Japan
All datasets described in this paper are culled from officially published data compiled by
government ministries or research institutes: output data and overall capital input data come
from the Cabinet Office, labor input data and education record of employees are from the
Ministry of Health, Labour and Welfare, utilization rates emanate from the Ministry of Economy,
45
Trade and Industry, and information technology assets and the ubiquitous index come from
InfoCom Research, Inc.
(Figure 3-1)
Before examining the growth accounting analysis and production function analysis, it is
useful to review Japan’s IT investment history. As Fig. 1 depicts, total investment in information
technology amounted to 14 trillion yen (120 billion US dollars) in 2007, which accounted for 2.7
percent of the nominal Gross Domestic Product (GDP), and 16.9 percent of total nonresidential
fixed investment. The amount of investment in software technology, approximately 6.8 trillion
yen (57 billion dollars), was as much as investment in hardware, which amounted to 7.2 trillion
yen (61 billion US dollars). However, the amount of investment in hardware including
computers, communications, and office equipment was twice what was invested in software up
until the late 1990s. Regarding computer investment, it was for a time the largest component of
IT investment, but it is now merely 2.7 trillion yen (23 billion US dollars), less than the current
figure of 3.3 trillion yen (28 billion US dollars) investment in communications equipment.
Several characteristics are readily apparent from Fig. 3-1. The first is the long-run
investment boom in the late 1980s. Second is the increased technology investment in the early
1990s and a cyclical fluctuation from the mid-1990s to the late 1990s. The third is the end of the
downward trend in hardware investment that became apparent in the early 2000s. Finally, there
has been a steady expansion of software investment since the late 1990s. It must be emphasized
that Japanese private businesses invested aggressively in “legacy” types of technology based on
mainframe computers and closed switched network systems in the 1980s, but they were much
less apt to invest in the new open-network technology of the 1990s.
46
In Japan in 1985, deregulation was just beginning in the telecommunications market but
banking industry leaders were enthusiastic about enhancing online transaction systems based on
“legacy” technology with little attention given to the “Solow paradox.” Consequently, they
successfully adopted “legacy” information systems even as U.S. firms were confronting the
productivity paradox.
The Japanese IT investment boom, however, halted abruptly in the early 1990s when new
types of open-network technology surged throughout the world, resulting in a downsizing from
mainframe computers and an increase in personal computer use with its wide Internet reach. By
that time, Japan’s investment in information technology had shown repeated cyclical
fluctuations that marked the decade.
(Figure 3-2)
That major shift in investment trends––the boom in the 1980s and the slump in the
1990s––affected the accumulation of information technology assets. Figure 3-2 illustrates that
the annual growth rate of Japan’s IT capital assets increased in the 1980s up to 19 percent.
Nevertheless, the rate of increase fell drastically in the early 1990s and has never since achieved
the high rate of the 1980s. Indeed, it is much more instructive to examine the United States. The
rate of accumulation of Japan’s IT assets jumped to more than double the U.S. rate in the latter
1980s, sliding to a lower level than that of the U.S. by the end of the 1990s. Therefore, we can
conclude that Japan missed a window of opportunity to ride the dynamic wave of information
technology innovation in the 1990s. In sharp contrast, the United States has had a long and
smooth journey and has reaped the benefits of the Internet revolution.
47
4. Results of growth accounting analysis
4-1. Japan’s past economic performance
Based on the formula and dataset described above, we analyzed the long-run economic
performance of Japan and the contribution of information technology. Table 1 shows the results
of the measurements of economic growth, with labor productivity shown as hourly output, since
the second half of the 1970s. The first line in the table traces the growth rate of the entire
economy; the third line shows the productivity growth rate as a formula of the first line (growth
rate of output) minus the second line (growth rate of labor input). The fourth and fifth lines show
this productivity growth rate with the business cycle effect and fundamental trend.
(Table 3-1)
Japanese macroeconomic performance has changed drastically over the last three decades
and the figures in the first line accurately portray the transformation. The economy apparently
enjoyed a powerful boom in the late 1980s and plunged into a deep slump in the 1990s. The
economy grew at a healthy 3.3 percent annually in the early 1980s and a vigorous 5.0 percent
annually in the late 1980s. That growth was accompanied by a rapid advance in labor
productivity. Output per hour rose at an annual rate of 2.4 percent in the early 1980s and a robust
3.7 percent in the late 1980s. This improvement was not driven by a cyclical effect in those days,
but rather by a fundamental trend in productivity improvement. More precisely, it was driven by
the surge in TFP and the capital deepening of IT assets.
In the 1990s, however, the economy plummeted into a deep slump, especially in the second
half of the decade. The economy grew at a mere 1.3 (1.6 in the first half, 0.9 in the second half)
percent annually with sluggish productivity improvement during the 1990s. The growth rate of
48
the economy was less than one-third the rate of the late 1970s or late 1980s, and less than half the
rate of the early 1980s. This sluggishness is also apparent in productivity figures. The
fundamental trend of output per hour rose 2.7 percent annually in the early 1990s and at the even
worse pace of 1.4 percent in the late 1990s. The productivity growth in the latter 1990s fell
sharply by two percentage points from the late 1980s. In fact, TFP also fell by more than one
percentage point. These figures well represent the stagnant economic condition that is often
referred to as the “lost decade” of the Japanese economy.
(Figure 3-3)
Nevertheless, the economy finally seemed to show slight signs of recovery in the early
2000s when, led by the Koizumi Administration, Japan underwent several important reforms.
The aggregate growth rate of the economy was one and one-half percent in the first half of the
2000s but that was mainly because of the decreasing trend in labor input, a reflection of the
private business sector’s efforts at downsizing and restructuring. Regarding the fundamental
productivity trend, productivity apparently bailed the country out of its deepest slump in the late
1990s. The productivity trend has recovered 0.7 percentage points from 1.4 percent to 2.1
percent since 2001, mainly because of the resurgence of TFP. The annual growth rate of TFP,
which fell to 0.0 percent in the late 1990s, has improved 1.1 percentage points to 1.1 percent now
and compensates somewhat for the weak contribution of capital deepening. The resurgence of
TFP reflects the recovery of aggregate efficiency in the Japanese economy.
4-2. Neither a “Solow paradox” nor a “new economy”
In the discussion presented in this subsection, we specifically address the contribution of
49
information technology to productivity improvement and the resultant economic growth. As
Table 1 shows, capital deepening, which reflects business investment, largely accounts for the
labor productivity improvement in each period. For example, the growth rate of productivity
trends during 1976–1980, 1981–1985, 1986–1990, 1991–1995, 1996–2000, and 2001–2005
were, respectively, 2.3, 2.4, 3.4, 2.7, 1.4, and 2.1 percent (see the fifth line of the table), of which
capital deepening contributed 1.7, 1.5, 1.8, 1.6, 1.0, and 0.7 percentage points, respectively (see
the sixth line of the table).
Although the overall contribution of capital deepening seems to have changed little, its
composition has shifted substantially. The capital deepening of IT assets gained in influence,
from 0.1 in the late 1970s to 0.4 in the late 1980s. It has remained almost unchanged until now
(see the eighth line of the table), although non-IT assets have become less important, from 1.6 to
0.4 percent (see the seventh line of the table). The surge of IT capital deepened in the late 1980s,
reflecting the increased importance of information technology (see increase of income share in
the addendum of Table 3-1) and the faster growth of information technology assets (see growth
rate of input in addendum of Table 3-1).
In the first half of the 1990s, however, the capital deepening of IT assets lessened somewhat
and has remained almost unchanged since then, accounting for one-seventh of the 2.1 percent of
the productivity trend growth in the 2000s. During the same period, the capital deepening of
non-IT assets became remarkably less productive, from 1.3 percent in the late 1980s to 0.4 in the
early 2000s. Consequently, the impact of IT assets on the economy has recently become as great
as that of non-IT assets.
The matter at issue, however, is not a comparison of IT assets to non-IT assets, but rather the
50
periodic changes in IT assets in terms of their contribution to productivity improvement and
resultant economic growth. The last five columns of Table 3-1 present important data.
Acceleration of TFP (see the tenth line) and the contribution of IT assets (see the eighth line) are
described as periodic changes in each of the five-year periods. The remarkable fact is that the
changes in TFP and contribution of IT capital assets ran in the same direction instead of in
opposite directions until the mid-1990s. This characteristic differs greatly from the growth rate
of TFP and the contribution of IT assets, which ran in opposite directions in the U.S. until the
mid-1990s (Table 3-2). In the United States, therefore, “economists were puzzled as to why
productivity growth was so slow despite the widespread use of information technology.”19 It was,
demonstrably, a “Solow paradox.”
(Table 3-2)
The Japanese economy is a case in contrast. For example, during 1981–1985, TFP increased
0.2 percentage points from the preceding five year period with IT capital assets contributing 0.1
percentage points. There was a 0.7 percentage point TFP growth with IT capital assets
contributing 0.3 percent points during 1986–1990, in addition to a -0.4 percentage point TFP
growth with a -0.1 percent point IT capital assets input during 1991–1995. Accordingly, TFP
was positive when capital deepening of IT capital assets contributed positively, and so TFP was
negative when IT capital assets contributed negatively. In other words, there was no “Solow
paradox” in Japan before the mid-1990s.
Conversely, no manner of clear correlation has been shown between TFP and the
contribution of IT assets since the second half of the 1990s. For example, during 1996–2000,
19 Baily (2002), p. 4.
51
TFP decreased 0.8 percentage points from the preceding five-year period with an unchanged IT
capital assets contribution; during 2001–2005, there was 1.0 percentage point TFP growth with a
slight negative (-0.1 percent point) contribution of IT capital assets. Therefore, it seems that
more significant changes in TFP, from 0.8 to 0.0 to 1.1, were never affected by a capital
deepening of IT assets, which remained almost unchanged during those periods. It follows that
there was no “Solow paradox” before the mid-1990s nor was there a “new economy” after the
mid-1990s in Japan. These observations are in clear contrast to those of the U.S., where the
“paradox” was noticeable before the mid-1990s, as was the “new economy” after the mid-1990s.
In light of the above descriptions, it seems reasonable to conclude that the former
observation (lack of a “Solow paradox”) represents successful investment in “legacy”
information technology in the 1980s, and the latter observation (lack of a “new economy”)
represents unsuccessful investment in open-network technologies of the Internet in the 1990s.
5. Implications of the new economy in the U.S.
As our analysis shows, Japan seems to have missed the chance to reap the rewards of the
“new economy.” It could be argued, however, that huge potential beckons in Japan’s current
economy. In other words, the Japanese economy could even now accelerate productivity and
resultant economic growth if it were to embrace the “new economy” and take full advantage of
the dynamism of IT innovation as the U.S. economy certainly did over the last decade. Therefore,
in this subsection, we put forth some simple measurements of Japan’s potential growth rate
assuming that the Japanese economy reaps the benefits of IT investment, as the U.S. economy
has done since the mid-1990s.
52
Before we make such an estimate, it is useful to review the long-run outline of Japan’s
fundamental productivity trends so we can measure a baseline of the potential growth rate. As
seen in section 4-1, the growth rate of the economy has fluctuated greatly over the last three
decades. Regarding the fundamental productivity trend, however, the changes were not so
drastic. They were moderate because fluctuation effects on labor input and the business cycle
were removed, as Fig. 3-3 clearly illustrates.
Furthermore, disregarding the exceptional periods of the late 1980s, a period of an
overheating economic boom or bubble, and the late 1990s, a period of financial crisis and
deflation, fundamental productivity trends were stable: 2.3 percent in the late 1970s, 2.4 percent
in the early 1980s, 2.7 percent in the early 1990s, and 2.1 percent in the early 2000s. As the data
in Table 3-3 underscore, the average growth rate of fundamental productivity trends in these
stable periods is around two and half percent. It therefore seems appropriate to conclude that an
annual productivity growth rate of two and half percent is the baseline for the minimum potential
of the Japanese economy.
(Table 3-3)
For measurement of economic growth, demographic trends must be considered as well as
the productivity baseline. The Japanese national population is predicted to decrease for some
time. According to the National Institute of Population and Social Security Research, the
working-age population will be decreasing just less than one percent annually over the next few
decades. Under this diminishing demographic trend, the potential economic growth rate would
be 1.6 percent annually even if the recent level of IT contribution were sustained.
What should not be disregarded is that the U.S. economy accelerated its productivity by
53
more than one percentage point over the last decade. Major contributions to this rising tide
derived from IT assets and TFP. If we simply presume that the Japanese economy will catch up
with the level of the U.S. or achieve a rate of acceleration similar to that of the U.S., the economy
will grow at a healthy clip, two and half percent or more annually, rather than just below the two
percent that has been generally accepted in Japan. Although adapting figures of the U.S. to the
Japanese economy might seem simplistic and naïve, it does suggest that the Japanese economy
has the potential to realize faster economic growth than the prevailing consensus.
6. Conclusion
As described in this part, we examined the impact of information technology on Japanese
economic growth, conducting an empirical analysis of the growth accounting model. This
analysis revealed that the Japanese economy successfully introduced a “legacy” type of IT
before the mid-1990s, but that Japan failed to keep pace with the drastic changes in technology
that occurred in the 1990s. Namely, there has been neither a “Solow paradox” nor a “new
economy” in Japan.
This is in sharp contrast to the economic performance in the United States where the
paradox apparently existed before the mid 1990s but then evaporated and a new economy
emerged in the late 1990s. Nevertheless, implications of a new economy in the U.S. might
suggest that the Japanese economy, which has fumbled innovation to date, still has the potential
to realize faster economic growth than the prevailing consensus, if the economy catches up and
achieves an acceleration of growth similar to that of the U.S.
54
References
Adams, Gerard F., Lawrence R. Klein, Yuzo Kumasaka, and Akihiko Shinozaki (2007)
Accelerating Japan’s Economic Growth, Routledge Studies in the Growth Economies of
Asia, Taylor & Francis, UK., October 2007.
Baily, Martin Neil (2002) “The New Economy: Post Mortem or Second Wind?” Journal of
Economic Perspectives, Spring 2002, 16:2, pp. 3–22.
Brynjolfsson, Erik and Lorin Hitt (1996) “Paradox Lost?: Firm-level Evidence on the Returns to
Information Systems Spending,” Management Science, April 1996, 42:4, pp. 541–558.
Council on Economic and Fiscal Policy (2006) The Report of the Special Board of Inquiry for
Examining “Japan’s 21st Century Vision,” Cabinet Office, April 2005.
InfoCom Research, Inc. (2008) InfoCom ICT Keizai Houkoku (InfoCom Economic Report on
ICT), November 2008, in Japanese.
Jorgenson, Dale W. (2001) “Information Technology and the U.S. Economy,” American
Economic Review, March 2001, 91:1, pp. 1–32.
Jorgenson, Dale W., Mun S. Ho, and Kevin Stiroh (2008) “A Retrospective Look at the U.S.
Productivity Growth Resurgence,” Journal of Economic Perspectives, 22:1, pp. 3–24.
Oliner, Stephen D. and Daniel E. Sichel (2000) “The Resurgence of Growth in the Late 1990s: Is
Information Technology the Story?” Journal of Economic Perspectives, 2000, 14:4, pp.
3–22.
Oliner, Stephen D., Daniel E. Sichel, and Kevin J. Stiroh (2007) “Explaining a Productive
Decade,” Brookings Papers on Economic Activity, 1:2007, pp. 81–153.
Shinozaki, Akihiko (2006) “Does the sun rise again in the ubiquitous information age?:
Feasibility of a vigorous economic growth for Japan under the diminishing demographic
trend ,” Kyushu University, Journal of Political Economy (Keizaigaku–Kenkyu), March
2006, 72:5–6, pp. 99–124.
Shinozaki, Akihiko (2008) “Japan’s IT puzzle: Neither a Solow paradox nor a new economy,”
InfoCom Research, Inc., InfoCom Review, March 2008, 44, pp. 22–31.
Stiroh, Kevin J. (2002) “Information Technology and the U.S. Productivity Revival: What Do the
Industry Data Say?” American Economic Review, 92:5, pp. 1559–1576.
Solow, Robert M. (1957) “Technical Change and the Aggregate Production Function,” Review of
Economics and Statistics, 39:3, pp. 312–320.
Solow, Robert M. (1987) “We’d Better Watch Out,” New York Times Book Review, July 12, 1987.
U.S. Department of Labor (1994) Integrating Technology with Workers in the New American
Workplace, Washington D.C., Government Printing Office.
55
Tables and Figures
Figure 3-1. Japan’s nominal investment in IT
0
5
10
15
20
25
30
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07
(trillions of yen)
software
communications and office equipment
computers
Source: InfoCom Research, Inc. (2008).
Figure 3-2. Growth of IT assets and non-IT assets
0
5
10
15
20
25
75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07
(%)
Growth of Japan's IT assets
Growth of US IT assets
Source: InfoCom Research, Inc. (2008) and U.S. Department of Commerce NIPA tables.
56
Table 3-1. Economic growth, labor productivity, TFP, and the contribution of IT
76-80 81-85 86-90 91-95 96-00 01-05 changes from previous five yearsa b c d e f b-a c-b d-c e-d f-e
Private output 4.8 3.3 5.0 1.6 0.9 1.5 -1.5 1.6 -3.3 -0.7 0.6Hours worked 1.4 0.9 1.3 -0.3 -0.5 -0.8 -0.4 0.3 -1.5 -0.3 -0.3 Output per hour 3.4 2.4 3.7 1.9 1.5 2.3 -1.1 1.3 -1.8 -0.4 0.9 Business cycle effect 1.2 -0.0 0.3 -0.8 0.1 0.2 -1.2 0.3 -1.1 0.9 0.2 Fundamental trend 2.3 2.4 3.4 2.7 1.4 2.1 0.1 1.0 -0.7 -1.3 0.7 Capital deepening 1.7 1.5 1.8 1.6 1.0 0.7 -0.2 0.3 -0.2 -0.6 -0.3 of non IT-assets 1.6 1.3 1.3 1.2 0.6 0.4 -0.3 0.0 -0.1 -0.6 -0.2 of IT assets 0.1 0.2 0.4 0.3 0.4 0.3 0.1 0.3 -0.1 0.0 -0.1 Labor quality 0.3 0.4 0.3 0.3 0.4 0.3 0.1 -0.1 -0.0 0.1 -0.0 Total factor productivity 0.3 0.6 1.3 0.8 0.0 1.1 0.2 0.7 -0.4 -0.8 1.0 [Income shares (percentage)]
share Ko (α) 31.1 29.5 29.7 25.5 22.4 22.1 -1.6 0.2 -4.3 -3.1 -0.3 share Ki (β) 1.9 1.9 3.0 3.6 4.5 5.6 -0.0 1.1 0.6 0.9 1.2 share L (γ) 66.9 68.5 67.3 71.0 73.2 72.3 1.6 -1.3 3.7 2.2 -0.9
[Annual growth rate of inputs] dKo 6.5 5.3 5.7 4.5 2.3 1.2 -1.2 0.4 -1.2 -2.2 -1.1 dKi 5.3 9.0 15.9 9.1 7.9 4.8 3.6 6.9 -6.8 -1.2 -3.1 dedu 3.5 -0.1 0.9 -2.6 0.2 0.7 -3.5 0.9 -3.5 2.8 0.5
Source: Author’s calculation.
Note: Figures might not add precisely because of rounding.
Table 3-2. Acceleration of the U.S. economy and the contribution of IT assets difference 2006
1959–73
(a)
1973–95
(b)
1995–
(c) 95-2000(d)
(b)-(a)
(c)-(b) (d)-(b)
Output per hour 2.8 1.5 2.6 2.7 -1.3 1.1 1.2 Capital deepening
of IT assets Labor quality Total factor productivity
1.4 0.2 0.3 1.1
0.9 0.4 0.3 0.4
1.4 0.8 0.3 1.0
1.5 1.0 0.2 1.0
-0.5 0.2 0.0 -0.7
0.5 0.4 0.0 0.6
0.6 0.6 -0.1 0.6
Source: Jorgenson et al. (2008).
Note: Figures might not add precisely because of rounding.
57
Figure 3-3. Economic growth and sources of productivity growth
2.3 2.4
3.4
2.7
1.4
2.1
-2.0
-1.0
0.0
1.0
2.0
3.0
4.0
5.0
6.0
76-80 81-85 86-90 91-95 96-00 01-05
annual rate: %
labor inputbusiness cycle effectlabor qualitycapital deepening of non-ITcapital deepening of ITtotal factor productivityfundamental productivity trendaggregate economic growth
Overheating Economic Boom↓
Financial Crisis & Deflation
↓
Source: Table 3-1 in this paper.
Table 3-3. Japan’s potential growth rate and acceleration estimation Japan’s economic growth rate Estimation of acceleration
Average Potential rate Case I Case II Growth rate Labor input Labor Productivity
Business Cycle Trend Capital deepening
of non-IT of IT assets Labor quality Total Factor Productivity
2.83 0.31 2.52 0.14 2.38 1.36 1.14 0.22 0.32 0.70
1.61 -0.86 2.47 ―
2.47 1.45 1.14 0.31 0.32 0.70
2.34 -0.86 3.20 ―
3.20 (+0.73) 1.92 1.14 0.78 0.32 0.96
2.55 -0.86 3.41 ―
3.41 (+0.94) 1.82 1.14 0.69 0.32 1.27
Note: Average excludes extraordinary periods of the late 1980s and the late 1990s. The potential rate incorporates
demographic trends of the next few decades and recent contributions of IT assets. Case I illustrates that
Japan’s TFP and capital deepening of IT assets can catch up with those of the U.S. Case II shows that Japan can
accelerate its productivity by 0.94 percentage points, as the U.S. has done since the late 1990s.
58
Part IV. Estimation and Simulation of Production Function Models
Akihiko SHINOZAKI
Abstract
In Part IV we estimated and simulated production function models that clearly incorporate
IT capital stock and network effects, exploring whether it is realistic to assume that information
technology will contribute to and accelerate Japan’s future economic growth. These analyses
yielded two observations. First, estimation of the production function model proves that IT
capital stock and network effects markedly influenced the economy, which suggests that
sluggishness of IT investment has plunged the economy into a lower growth path since the 1990s.
Second, simulations of the production function model demonstrate that the economy has the
potential to grow at a higher rate than the consensus of less than two percent. Consequently, it
can be argued that the Japanese economy still has room to accelerate economic growth if it is
able to somehow maximize the benefits of innovation, which the economy has fumbled during
the last decade.
1. Introduction
As demonstrated in Part III, growth accounting analysis revealed that there is neither a
“Solow paradox” before the mid-1990s nor a “new economy” after the mid-1990s in Japan.
Those observations are in clear contrast to what is happening in the U.S., although they do
suggest that the Japanese economy still has room to accelerate economic growth. In other words,
the economy can grow faster than the pessimistic consensus, if the economy successfully adopts
technology and achieves growth acceleration similar to the U.S. To verify this potential, in Part
59
IV we first conducted estimations of a production function model in which IT capital assets and
network effects are clearly incorporated. Based on the estimation results, we then simulated the
economy’s growth paths for the next few decades. Through these analyses, we explored whether
it is realistic to assume that information technology can contribute to and accelerate Japan’s
economic growth under the country’s diminishing demographic trends.
2. Analytical framework and dataset
In Part IV, we used a production function model describing the mapping of quantities of
inputs to quantities of an output as generated by a production process. We employed and
modified the traditional Cobb–Douglas model to estimate and simulate the Japanese economic
growth path using several scenarios. We then estimated three types of production function
models: a base model of a traditional Cobb–Douglas function with two simple input factors of
total capital assets and labor input; an IT assets model, in which the impact of IT capital assets
can be measured respectively by dividing total capital assets into IT and non-IT capital assets;
and a network effect model, which accommodates increasing returns to scale where network
externality is incorporated as the spread of IT infrastructure and their sufficient use.
[Base model]
Equation (4-1) presents the fundamental Cobb–Douglas type production function model
with capital assets and labor input in which labor quality is incorporated.
(4-1) Q=A(eduL)α(pKall)β
In this equation, α and β respectively signify output elasticity with respect to labor input and
60
capital stock, assuming constant returns to scale (i.e. α+β=1). In addition, Q is the private output,
A is the level of technology, edu represents the education records of employees, L is the labor
input for work hours of total employees, p is the capital asset utilization rate, and Kall stands for
total capital assets without distinguishing IT and non-IT capital assets (Kall =Ki+ Ko).
[IT assets model]
Based on the model shown above as eqn. (4-1), it is impossible to clearly estimate and
simulate the impact of IT investment on economic growth because IT capital assets are
contained in total capital assets in the base model. Therefore, the base model must be modified to
a model in which IT capital assets are represented clearly by dividing capital assets into IT and
non-IT capital assets, as in
(4-2) Q=A(eduL)αKoβKi
γ,
where α, β, and γ respectively signify output elasticity with respect to labor input, non-IT capital
stock, and IT capital stock, assuming constant returns to scale (i.e. α+β+γ=1).
[Network effect model]
Equations (4-1) and (4-2) portray the hypothesis of constant returns to scale (i.e., α+β=1 or
α+β+γ=1). The theory of the information economy, however, demonstrates a “network effect”
or a “network externality” that eases the hypothesis of constant returns to scale: it supports
increasing returns to scale (i.e. α+β+γ>1). The following model, in which network effects are
considered, was estimated in JCER (2000).
(3) Q=A(eduL)αKallβKi
γ
61
Therein, α+β=1, γ>0, and Kall =Ki+ Ko, implying that IT capital assets (Ki) contribute to the
output in two such paths: where they ordinarily serve as capital input for their own production
processes and where they additionally serve as a kind of public good, or infrastructure, for
others’ production processes. In the former path, their contribution is presented as a part of Kall
input, although it is exhibited in the form of the explicit contribution of Ki in the latter path.
When JCER (2000) conducted an estimation of the model in eqn. (4-3), it proved that IT
capital assets (Ki) show a positive network externality, or 9% increasing returns to scale (i.e.
α+β+γ=1.091>1, γ=0.091) as does Shinozaki (2003), with an estimate of 16% increasing
returns to scale (i.e. α+β+γ=1.162>1, γ=0.162).
The model shown above as eqn. (4-3), however, does not incorporate an important aspect of
the network effect. Given the same amount of IT capital assets value, the model does not
distinguish a small number of mainframe computers from a large number of personal computers.
For example, the network effects of a single mainframe computer valued at 1 million US dollars
and one thousand personal computers valued at 1,000 US dollars each differ greatly, but the
model of eqn. (4-3) treats them equally.
Furthermore, the model does not take into account whether IT assets are sufficiently used or
not, i.e., aggressive use and lackluster use of technology are treated as identical given the same
amount of IT assets, even though their network effects must be quite different. To address these
limitations and improve the network effects model, we modified the model in the following
manner:
(4-3’) Q=A(eduL)αKallβ(ubqKi)γ
Therein, α+β=1, and Kall =Ki+ Ko, ubq is the ubiquitous index that comprises the number of PC users,
62
cellular phone users, circulation volume of information, and several other related figures.
Consequently, ubq is considered an appropriate proxy to denote the pervasion and effective use of
information technology20. In this model, the network effect is identified if we attain the statistically
significant parameter γ>0, i.e. α+β+γ>1.
All datasets employed in this part are the same as we use in Part III except for the ubiquitous
index that is constructed and published by InfoCom Research, Inc.
4. Estimation of the production function model
As demonstrated in Part III, Japan seems to have fumbled the “new economy.” It could be
argued, however, that huge potential beckons in the current economy. In other words, the
Japanese economy could even now accelerate productivity and resultant economic growth if it
were to embrace the “new economy” and take full advantage of the dynamism of IT innovation
as the U.S. economy certainly has done. Indeed, the U.S. economy accelerated its productivity
by more than one percentage point over the last decade (see table 3-2 in the part III). Major
contributions to this rising tide derived from IT assets and consequent productivity
improvement.
Although adapting U.S. figures to the Japanese economy might seem simplistic and naïve,
it suggests that the Japanese economy has the potential to realize faster economic growth than
the current pessimistic consensus. To verify this potential, further empirical studies are needed,
such as estimations and simulations of production function models that explain explicit
contributions of IT innovation.
20 For details, see Noguchi, et al. (2008).
63
As clarified above, we estimated three types of production function models: base, IT
assets, and network effect. Constant returns to scale are assumed in the base model and the IT
assets model, whereas the network effects model allows increasing returns of scale.
To estimate each model, eqns. (4-1), (4-2), and (4-3’) shown in section 2 are transformed
to eqns. (4-4), (4-5), and (4-6) respectively, dividing both sides by eduL and taking logarithms of
both sides.
[Base model]
(4-4) ln(Q/eduL) = lnA + β ln (pKall /eduL) + e
In that equation, α and β represent output elasticity with respect to labor input and capital stock
respectively, assuming constant returns to scale (i.e., α+β=1), Q is the private output, A is the
level of technology, edu represents the education records of employees, L is the labor input for
work hours of total employees, p is the utilization rate of capital assets, Kall represents total
capital assets without distinguishing IT and non-IT capital assets, and e is an error term.
[IT assets model]
(4-5) ln(Q/eduL) = lnA + β ln (pKo /eduL) +γ ln (pKi /eduL) + e
where α, β, and γ represent output elasticity with respect to labor input, non-IT capital (Ko) stock,
and IT capital stock (Ki) respectively (Kall =Ki+ Ko), assuming constant returns to scale (i.e.
α+β+γ=1).
[Network effect model]
64
(4-6) ln(Q/eduL) = lnA + β ln (pKall /eduL) +γ ubqKi + e
Therein, α+β=1, Kall =Ki + Ko, and ubq clarify the ubiquitous index, assuming increasing returns
to scale (i.e. α+β+γ>1).
Each estimation is calculated taking first order serial correlation (AR[1]) into account.
Estimation results of eqns. (4-4), (4-5), and (4-6) are shown respectively in the following eqns.
(4-4’), (4-5’), and (4-6’).
[Base model]
(4-4’) ln(Q/eduL)=-2.303+0.537 ln(Kall/eduL)+0.626 AR[1]
(-16.01) (23.51) (3.32)
adjR2=0.994, D.W.=1.728, t-statistics: shown in (), sample year: 1976-2007.
[IT assets model]
(4-5’) ln(Q/eduL)=-0.888+0.229 ln(Ko/eduL)+0.149 ln(Ki/eduL)+0.662 AR[1]
(-1.81) (2.25) (3.73) (3.01)
adjR2=0.996, D.W.=1.654, t-statistics: shown in ( ), sample year: 1976–2007.
[Network effect model]
(4-6’) ln(Q/eduL)=-1.542+0.359 ln(Kall/eduL)+0.018(ubqKi)+0.952 AR[1]
(-5.34) (6.06) (2.19) (17.32)
adjR2=0.993,D.W.=1.519, t-statistics: shown in ( ), sample year: 1976–2007.
65
(Table 4-1)
Table 4-1 presents a summary of the estimation results of three production function types,
demonstrating that IT capital assets significantly affect economic growth and identifying a
positive network effect even though Japan has not reached the “new economy” stage yet. These
results suggest that sluggish IT investment drove the economy to a lower growth path after the
1990s but that, going forward, the economy nevertheless has the potential to reap the benefits of
IT innovation through intensive investment in technology. Accordingly, the estimation results
led us to another empirical study: to simulate alternative perspectives of the Japanese economic
outlook, using the production function models described above.
5. Simulation of Japan’s next growth path
We then attempted to simulate Japan’s economic growth path to 2025 based on the
estimation results shown above. Based on eqn. (4-4’), we cannot explicitly measure the
opportunities of IT innovation, although we can simulate another growth path based on eqn.
(4-5’) or (4-6’), where we can incorporate IT innovation into economic projections.
We used the following assumptions for the simulation: For a short term business cycle we
assume that the timeframe of 2008 through 2010 is a recession period, with the economy
recovering after 2011. Regarding the labor input, we adopt the moderate decreasing trend of the
working age population projected by The National Institute of Population and Social Security
Research, while we assume that labor quality continues to improve at the same clip as the
average rate during 1991–2005. For capital input, we employ the average growth rate of types of
capital assets during 1991–2005, with the exception of IT capital assets. Regarding IT capital
66
assets, we assume that capital deepening of IT assets (i.e. IT assets per hour worked) during
2011–2020 will grow as fast as in the late 1980s. It is predicted that information network
industries will be revitalized and will compete more vigorously in several innovative markets
generated by the technological progress of digital convergence in telecommunications and
broadcasting as well as further deregulation. We also used the average growth rate of the
ubiquitous index during 2000–2005 for our robust projection period from 2011 to 2020,
assuming it will meet faster capital deepening of IT assets and will grow as fast as it has in the
early 2000s, when broadband networks and mobile internet services had just begun to be
adopted.
(Figure 4-1)
Figure 4-1 portrays the simulation results. The average growth rate of the economy in the
base model is measured as one and half percent annually, whereas the IT assets model shows a
growth rate of two percent or more. Moreover, the network effect model proves the economy can
grow at two and half percent or more annually, one percentage point faster than base model
suggests21. Therefore, as suggested by the experience of the U.S. economy, it can be concluded
that simulations of the production function model more strongly support the possibility of a
faster growth path than the pessimistic viewpoint would have us believe.
6. Conclusion
As described in this part, we examined the impact of information technology on Japanese
economic growth, estimating and simulating the production function models that elucidate
21 Several other empirical studies also suggest faster economic growth of Japan. See Adams, et al. (2007).
67
explicit contribution of IT assets and network effects. These analyses revealed that IT assets and
network effects significantly influenced the Japanese economy and that the economy has some
potential to grow at a higher rate than the consensus of one and a half percent annually even
though since the 1990s the economy has so far missed the chance to reap the benefit of
information technology innovation. It might be argued, therefore, that the Japanese economy,
which has fumbled innovation to date, has room to accelerate economic growth under the
country’s diminishing demographic trend if intensive investment and the efficient use of
technology take hold throughout the economy.
68
References
Adams, Gerard F., Lawrence R. Klein, Yuzo Kumasaka, and Akihiko Shinozaki (2007)
Accelerating Japan’s Economic Growth, Routledge Studies in the Growth Economies of
Asia, Taylor & Francis, UK., October 2007.
Council on Economic and Fiscal Policy (2006) The Report of the Special Board of Inquiry for
Examining “Japan’s 21st Century Vision,” Cabinet Office, April 2005.
InfoCom Research, Inc. (2008) InfoCom ICT Keizai Houkoku (InfoCom Economic Report on
ICT), November 2008, in Japanese.
JCER (2000) Nihon Keizai no Sai Shuppatsu II: IT Kakushin no Shogeki to Sono Hyoka (Born
Again of the Japanese Economy: Impact of IT innovation and its estimation), Japan Center
for Economic Research, May 2000, in Japanese.
Jorgenson, Dale W., Mun S. Ho, and Kevin Stiroh (2008) “A Retrospective Look at the U.S.
Productivity Growth Resurgence,” Journal of Economic Perspectives, 22:1, pp. 3–24.
Noguchi, Masato, Akihiko Shinozaki, Yusuke Yamamoto, and Shota Yamasaki (2008)
“Yubikitasu Shisu no Suitei ni Tsuite (How to Compose Ubiquitous Index)” InfoCom
Research, Inc., Technical Papers on ICT related Economic Indices, No. 08-2, February 2008,
pp. 1–15, in Japanese.
Oliner, Stephen D., Daniel E. Sichel, and Kevin J. Stiroh (2007) “Explaining a Productive
Decade,” Brookings Papers on Economic Activity, 1:2007, pp. 81–153.
Shinozaki, Akihiko (2003) “Tsuushin Sangyo ni Okeru Setsubitoushi no Keizai Koka Bunseki
(Economic Impact of Investment in Telecom Industry)”
Shinozaki, Akihiko (2006) “Does the sun rise again in the ubiquitous information age?:
Feasibility of a vigorous economic growth for Japan under the diminishing demographic
trend ,” Kyushu University, Journal of Political Economy (Keizaigaku–Kenkyu), March
2006, 72:5–6, pp. 99–124.
Shinozaki, Akihiko (2009) “Simulating Japan’s Alternative Growth Paths: Production Function
Model Analysis on the Impact of Information Technology,” InfoCom Research, Inc.,
InfoCom REVIEW, No.47, forthcoming.
69
Tables and Figures
Table 4-1. Results of estimation
coefficient t-statistics coefficient t-statistics coefficient t-statisticsC -2.303 ** -16.010 -0.888 -1.814 -1.542 ** -5.341
Kall/eduL 0.537 ** 23.510 0.359 ** 6.056Ko/eduL 0.229 * 2.250Ki/eduL 0.149 ** 3.725ubq*Ki 0.018 * 2.189AR(1) 0.570 ** 3.316 0.662 ** 3.008 0.952 ** 17.315
Labor shareCapital share(of non-IT)( of IT )
adjR2
D.W.growth rate % % %
(2010-20)(2010-25)
0.6220.3780.229
IT assets model
0.149
2.11.6
0.9961.654
1.6 2.4
1.728
Base model
0.4630.537
0.994
2.6
Network effect model
0.6410.359
2.7
0.9931.519
Source: Author’s calculation.
Figure 4-1. Simulation of Japan’s next growth path
0
100,000
200,000
300,000
400,000
500,000
600,000
700,000
800,000
900,000
74 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 06 08 10 12 14 16 18 20 22 24
(hundred millions of yen)
actual
simulation (network effect model)
simulation (IT assets model)
simulation (base model)
Source: Author’s calculation.