Measurement of Intangible Investment by Industry and ... · Measurement of Intangible Investment by...

Policy Research Institute, Ministry of Finance, Japan, Public Policy Review, Vol.9, No2, Ｍａｒｃｈ 2013 405

Measurement of Intangible Investment by Industry and Economic Growth in Japan*

Tsutomu Miyagawa Gakushuin University and RIETI

Shoichi Hisa Yokohama City University

Abstract

We measure intangible investment by industry in Japan based upon the JIP (Japan Industrial Productivity) database and other primary statistics. Expenditures on intangibles in Japan amounted to around 43 trillion yen, accounting for around 9% of GDP in the 2000s. The ratio of intangible investment to gross value added in the IT sector was higher than that in the non-IT sector. The total intangible capital stock in 2008 was worth 136 trillion yen, but the growth rate of intangible capital turned negative in some industries in the 2000s, as they underwent a drastic restructuring process. Growth accounting including intangible assets shows that intangible assets in the IT sector, which has led Japanese economic growth, are contributing to the growth of the economy. Our above results suggest that we should not unnecessarily cling to economic policy based on existing industry classifications, but draw up growth-promoting policies based upon new industry classifications.

JEL classification numbers: E01, E22, O31, O32 Keywords: intangible investment, TFP, IT sector, Solow residuals

* This paper is based on a paper presented at an invitation lecture at the 2012 Spring Meeting of the Japanese Economic Association, and the RIETI Discussion Paper Series 12-E-037: “Measurement of Intangible Investment by Industry and Its Role in Productivity Improvement Utilizing Comparative Studies between Japan and Korea.” We thank Professors Hyunbae Chun (Sogang University) and Kyoji Fukao (Hitotsubashi University) for collaborating with us in the previous work. We also thank an anonymous referee, Professor Masahisa Fujita (President of RIETI), Mr. Masayuki Morikawa (Vice President of RIETI), Dr. Carol Corrado (Conference Board), Professor Ahmed Bounfour (University of Paris-South), Professor Kaoru Hosono (Ministry of Finance and Gakushuin University), and participants of seminars held by the Research Institute of Economy, Trade, and Industry, the National Institute of Science and Technology Policy, the Samsung Economic Research Institute, the OECD, and the 32nd IARIW General Conference. We also thank the attendees of the 2012 Spring Meeting of the Japan Economic Association, and Gakushuin University for their helpful comments and suggestions. The excellent research assistance by Mr. Tatsuya Makino is also appreciated. All errors in this paper are our own. This study was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology (No.22223004).

406 T Miyagawa, S Hisa / Public Policy Review

1. Introduction

There has been a great deal of research on intangible investment following the IT revolution and high productivity growth in the US economy in the 1990s. For instance, Hall (2000), (2001), Bresnahan et al. (2002) and Basu et al. (2003) focused on intangible assets and showed that intangible assets, which are complementary to IT equipment, have contributed to the acceleration of productivity growth. However, these papers have estimated the effects of intangible assets on productivity growth indirectly, due to the difficulty in measuring intangible assets.1

Corrado, Hulten and Sichel (2009) overcame this difficulty and measured intangible investment at the aggregate level in the US for the first time. Based on their estimation, they found that the ratio of intangible investment to GDP exceeded the ratio of tangible investment to GDP in the early 2000s, and that one third of the productivity growth in the late 1990s and the early 2000s was attributable to the growth in intangible assets. After their success in measuring intangible assets, many economists followed their method and estimated intangible investment in their own countries.2

However, the aggregate data does not provide enough information to conduct a productivity analysis. As Jorgenson, Ho, and Stiroh (2005), Inklaar, O’Mahony and Timmer (2005), Fukao, et al. (2012) have suggested, there is a significant productivity gap between IT and non-IT sectors. In addition, even in IT-intensive service industries, there is a productivity gap between the US and Japan. Understanding these gaps would require intangible investment data at the industry level. Moreover, the aggregate series also constrains our analysis. The measured time series of intangible investment data goes back at most 30 years. The sample size is not large enough to conduct several econometric analyses.

In this paper, we measure intangible investment at the industry-level in Japan to understand the relationship between intangible investment and the productivity gap across industries. There are only a few studies that focus on intangible investment at the sectoral level. For example, Fukao, et al. (2009) measure intangible investment in the manufacturing and service sectors. Barnes (2010) summarized the measurement of intangible investment at the sectoral level. However, the industry classification in the previous studies is close to the aggregate level. In our paper, we measure intangible investment at the two-digit industry level in accordance with the Japan Industrial

1 Miyagawa and Kim (2008) also considered the role of intangible assets on productivity improvement through indirect measurement of intangible assets using firm-level data. However, the estimation of intangible assets at the firm level is affected by changes in legal restrictions on firms’ financial account reports and is not suitable for analyzing the impact of intangible assets on long-run productivity growth. 2 Marrano, Haskel, and Wallis (2009) for the UK, Fukao et al. (2009) for Japan, Delbecque and Bounfour (2011) for France and Germany, Hao, Manole, and van Ark (2008) and Piekkola (2011) for major EU countries, Burnes and McClure (2009) for Australia, and Pyo, Chun and Rhee (2010) for Korea.


Productivity (JIP) database.3 In the next section, we explain how we measure intangible investment by industry.

Then, in the third section, we show some features of intangible investment at the industry level in Japan. In the fourth section, using industry level data, we examine empirically the effect of intangible assets on productivity improvement. In the final section, we summarize our results.

2. Measurement of Intangible Investment by Industry in Japan

Although there are several classifications of intangible assets, we follow the asset classification by Corrado, Hulten, and Sichel (2005, 2009). According to their definition, intangible assets are made up of (1) computerized information, (2) innovative property, and (3) economic competencies. Regarding the industry classifications, we follow those used by the Japan Industrial Productivity (JIP) database. The JIP database provides data for 108 industries based on two-digit industry classification. In the JIP database, the market economy consists of 92 industries, the manufacturing sector consists of 52 industries, and the service sector consists of 33 industries. By combining the JIP database with other statistics, we estimate intangible investment by industry in Japan. We explain the measurement in the following sub-sections. The detailed explanation of measurement is summarized in Appendix 1.

2-1. Measurement of Computerized Information

Computerized information consists of custom and packaged software, and own account software. Custom and packaged software is estimated in the SNA. From the JIP database, we obtain SNA data and distribute the total custom and software investment in each industry by using the Fixed Capital Formation Matrix (FCFM). Our estimation method follows that of the JIP database.

We estimate the cost of workers who are involved in the development of software for their own firms to measure the value of own account software. We estimate the ratio of system engineers (SEs) and programmers to total workers by industry using the Population Census. By multiplying this ratio by the total number of workers in the JIP database, we can obtain an estimate of the number of SEs and programmers by industry. We obtain wage data for SEs and programmers from the Basic Survey on Wage Structure. Multiplying this wage by the number of SEs and programmers, we estimate a value for own account software investment. The Cabinet Office (CAO) in Japan recently published its own account software investment at the aggregate level in the new estimation of the SNA. Our

3 The JIP database consists of 108 industries. The website of the database is http://www.rieti.go.jp/ en/database/JIP2011/index.html. Fukao and Miyagawa (2008) explain how this database was constructed.


estimation method is similar to that by the CAO. However, due to the difference in coverage, our estimate for 2008 is smaller, approximately 450 billion yen less than the value published by the CAO.4 However, the estimates by the CAO are only available at the aggregate level since 1989. We therefore used our own estimates for our analysis.

2-2. Measurement of Innovative Property

Innovative property consists of science and engineering R&D, mineral exploitation, copyright and license costs, and other product development, design, and research expenses. First, we estimate science and engineering R&D costs by using the Survey of Research and Development published by the Statistical Bureau of the Ministry of Internal Affairs and Communications. However, this survey does not cover R&D data in most service sectors prior to 2000. Using the service sectors' expenditures for R&D outsourcing in the JIP database, we calculate backwards to find the service sectors' R&D costs.

Because expenditures on mineral exploitation are allocated only to the mining industry, we follow the estimation by Fukao, et al. (2009) that measured the aggregate intangible investment in Japan. The Mining Industry Handbook and the Establishment and Enterprise Survey provide data on expenses for mineral exploitation (the total expenses for geological investigation). By combining the above two surveys with FCFM, we estimate the total expenditures on mineral exploitation.

Copyright and license costs are assumed to consist of input from the publishing industry (JIP industry no. 92) and the video picture, sound information, character information production and distribution industry (JIP industry no. 93) to JIP industries nos. 1-71 and 73-107. By using the I-O table in the JIP database, we allocate these inputs to 108 industries. The allocation mechanism is as follows. For example, when we estimate copyright and license costs in industry i in year t, we obtain the input data from JIP industry no. 92 and no. 93 to industry i by using the JIP I-O table in year t. As the JIP database has I-O tables from 1973 to 2008, we are able to obtain the data of copyright and license costs by industry in the above period.

We estimate the outsourcing costs of design, display, machine design and architectural design by using the sales data of these industries from the Survey of Selected Service Industries and the input from the other services for the business industry (JIP industry no.88). We calculate the ratio of sales of the design and display industries to the nominal output of the other services for business industry (JIP industry no.88) of the JIP database. Like the estimation for copyright and license costs, we allocate the estimated costs to 108 industries by using the I-O table in the JIP database.

As for in-house expenditures, we estimate only in-house designing. We estimate the ratio of the designers to total workers by industry using the Population Census.

4 The CAO seems to use its own survey to calculate the ratio of SEs and programmers to the total number of workers and estimate the own account software.


Multiplying this ratio by the number of total workers in the JIP database, we obtain the number of designers by industry. Census data is available every five years. For the other intermediate years when Census data is not available, we estimate the ratio by linear interpolation. We obtain wage data from the Basic Survey on Wage Structure, and multiply it by the number of estimated workers. Like the estimation in the own account software investment, we do not take other expenditures into account, except labor costs.

As for the estimation in product development in financial services, the estimation method by Corrado, Hulten, and Sichel (2005) was very controversial, especially after the downturn after the bankruptcy of the Lehman Brothers, because they assumed that 20 percent of intermediate inputs produced by financial services could be regarded as expenditures in intangible assets. Recently, Corrado et al. (2012) suggested that the cost of new product development in financial services makes up only about 8% of the compensation of highly skilled workers in the financial industry, in order to harmonize their estimate to estimates in EU countries by COINVEST and INNODRIVE projects.5 Thus, following Corrado’s suggestions, we have assumed that 8 percent of the compensation of college graduate workers in the financial industry (JIP industry no. 69) and the insurance industry (JIP industry no. 70) can be regarded as expenditures in intangible assets. These expenditures are treated as those for the financial sector and insurance industry respectively.

2-3. Measurement of Economic Competencies

Economic competencies consist of three components: brand equity, firm specific human capital, and organizational structure. Regarding the measurement of brand equity, we obtain the input data of the advertising industry (JIP industry no. 85) and allocate it amongst 108 industries by using the I-O table in the JIP database. The allocation device is similar to the case of copyright and license costs or the outsourcing costs of design, display, machine design and architectural design.

In estimating firm specific human capital, we focus on off-the-job-training costs. We estimate the ratio of off-the-job training costs to total labor costs from the General Survey on Working Conditions by industry as published by the Ministry of Health, Labor and Welfare. For the opportunity cost of off-the-job training in terms of working hours lost, we use the results obtained by Ohki (2003). Using micro-data from the “Survey on Personnel Restructuring and Vocational Education/Training Investment in the Age of Performance-based Wage Systems (Gyoseki-shugi Jidai no Jinji Seiri to Kyoiku/Kunren Toshi ni Kansuru Chosa),” conducted by the Japan Institute for Labor Policy and Training, Ohki calculated the average ratio of the opportunity cost of off-the-job training to direct firm expenses for training in 1998 for the entire business sector. The value was 1.51. We use this same ratio to estimate the opportunity cost.

5 Intangible investment in the Euro-zone is available at www.INTAN-invest.net.


To estimate expenditures for organizational restructuring, Corrado, Hulten, and Sichel (2005) assumed that 20% of the remuneration of executives could be considered as intangible assets for organizational structure. However, we replaced 20% with 9%, because only 9% of the total working time of executives is spent on organizational reform and the restructuring of the organization, according to Robinson and Shimizu (2001). We calculated the ratio of the remuneration of executives to value added using the Financial Statements Statistics of Corporations by Industry published by the Ministry of Finance. Then, we found the expenditure for the organizational restructuring by industry by multiplying this ratio to the value added in the JIP database.

2-4. Measurement of Capital Stock in Intangible Assets

Based on the measurement of expenditures in intangible assets, we estimate capital stock in intangible assets. Corrado, Hulten, and Sichel (2005) pointed out that some of the expenditures in intangible assets should not be accounted for as capital formation, because their service lives are too short. Therefore, based on the argument in Corrado, Hulten, and Sichel (2005), we revised our estimates of expenditures on intangible assets as follows to find a capital formation series:

(1) New product development costs in the entertainment industry are assumed to be short-lived.

(2) 60% of advertising costs are accounted for as capital formation. (3) 80% of remuneration of executives that is spent on organizational reform is

accounted for as capital formation. The capital formation series is measured in nominal terms. Using the deflator by assets

shown in Table 1, we construct a real capital formation series in intangible assets. Then, we accumulate the capital formation series by using the perpetual inventory method and find the real capital stock in intangible assets. The depreciation rate by asset that is used for the perpetual inventory method is shown in Table 2. Since the perpetual inventory method adds up the flow of intangible investment, the cumulative value of the intangible assets in the initial period does not account for the stock from the previous period. Thus, we only begin reporting the estimate of the stock of intangible assets from 1985, 5 years after the beginning of our estimation of the intangible investment.


Table 1: Deflators for Intangible Investment

Table 2: Depreciation Rates for Intangible Assets

Source: Corrado, Hulten and Sichel (2005).

3. Accumulation of Intangible Assets

3-1. Expenditures on Intangible Assets

Our estimates of expenditures on intangible assets at the aggregate and the sector levels are summarized in Table 3. The total annual spending on intangible assets in Japan for the period 2001-2008 was about 42 trillion yen on average. Annual capital spending on intangibles was about 39 trillion yen in the same period. In the market economy, the annual expenditures for the same period were about 40 trillion yen and the annual capital spending was about 35 trillion yen.


Table 3: Estimated Spending on Intangible Assets in Japan (Billions of JPY)

*Capital spending on intangibles is shown in parenthesis.

When we focus on spending on intangible assets at the sector level, spending in the manufacturing sector for the period 2001-08 was about 18 trillion yen, which was almost the same as that for the period 1991-2000. Spending on intangible assets in the manufacturing sector for the period 1980-90 was about 10 trillion yen annually on average, and the amount almost doubled for the period 1991-2000 and stayed flat since then. On the other hand, spending on intangible assets in the service sector was about 20 trillion yen for the period 2001-08, and increased about 24% from the previous period. In the service sector, outlays on intangible assets for the period 1980-90 was around 8 trillion yen, almost doubling during the period 1991-2000, and still continues to grow.

However, unlike these traditional sectors, more and more attention is being paid to the IT sector and its related sectors. We define the IT sector as follows. The IT sector consists of industries that provide IT products and IT services, as classified by OECD, as well as industry in which the ratio of investment in IT to output is above the median value. Other industries are defined as the non-IT sector. The detailed industry classification in the IT and non-IT sectors is described in Appendix 26. As shown in Table 3, the spending on intangible assets in the IT sector was approximately 25 trillion yen annually in the 2000s, and investment spending was approximately 22 trillion yen, far exceeding that in the non-IT sector. In particular, the spending on computerized information in the IT sector was almost 1.7 times greater than that of the non-IT sector.

In Table 4, we measure the ratios of intangible investment to GVA. In the 1980s, the total intangible investment/GVA ratio in Japan was 6.1% on average, which was higher

6 In this paper, we have included in the non-IT sector the estimates for intangible assets in industries that were categorized as neither IT sector nor non-IT sector, such as Agriculture, Forestry and Fisheries Industry.


than that in Korea (3.4%).7 This gap between Japan and Korea was reduced in the 2000s. In 2008, while the Japanese intangible/GVA ratio was 9.4%, the ratio in Korea was 7.4%.8 At the sectoral level, the intangible investment/GVA ratio in the manufacturing sector was larger than that in the service sector due to large investment in innovative property in the manufacturing sector. The values were similar between the manufacturing sector and the service sector. Note that both the IT sector and the non-IT sector consist of industries in the manufacturing sector and in the service sector, and the intangible/GVA ratio in the IT sector was higher than that in the non-IT sector.

Table 4: Intangible Investment/GVA Ratio in Japan

*CI: computerized information; IP: innovative property; EC: economic competencies.

Table 5 shows the intangible investment/GVA ratio by industry. Intangible investment ratios in the IT sector such as machinery industries and the information and communication service industry were much greater than those in other industries. On the other hand, in some service industries such as education, health and social work, and culture and entertainment industries, these ratios declined in the 2000s.

7 Intangible investment in Korea is estimated in Chun, et al. (2012). 8 In Korea, software investment may not include own account software which is included in the figure for Japan. If own account software investment in Korea is correctly estimated, the gap between Japan and Korea will be lower.


Table 5: Intangible Investment/GVA Ratio by Industry in Japan

3-2 Capital Stock in Intangible Assets

The amount of capital stock and its growth rate in Japan are summarized in Table 6. The total capital stock in 2008 was about 136 trillion yen. The amount of capital stock in the manufacturing sector was almost the same as that in the service sector. The annual growth rate of intangible capital in the market economy from 1985 to 2008 was 4.2%. Although the growth rate in the late 1980s exceeded 10%, it fell after 1990. In particular, the growth rate in the 2000s was a mere 1.3% in the market economy, due to the negative growth in capital in economic competencies. Although assets in computerized information increased in the late 1990s due to the IT revolution, their growth rate also declined in the 2000s. As a result, the increase in the overall intangible capital stock in the 2000s can mostly be attributed to that in the IT sector.


Table 6: Capital Stock in Intangible Assets in Japan

*CI: computerized information, IP: innovative property, EC: economic competencies

Figure 1 shows the growth in intangible assets by industry. As shown in Table 6, we can see negative growth in intangible assets in the 2000s in some industries, such as textile and leather, construction, and the wholesale and retail trade. Figure 2 shows the growth rate in intangible assets by industry and by component in the 2000s. In most industries, assets in economic competencies fell. In only a few industries, such as the petroleum, coal and chemicals, transport equipment, and information and communication industries, did these assets grow. In the cultural and entertainment, education and health, and social work industries, assets in computerized information declined greatly after 2000. Assets in economic competencies declined through intensive restructuring in response to the long-term economic slump. This decline reflects the reduction in spending on firm specific human capital (measured by cost of training) and in spending for organizational restructuring (measured by the fraction for remuneration of executives). On the other hand, assets in computerized information in the non-market sector declined, because the network system among institutions has not improved due to regulations and a shortage of management skills.9

9 For instance, in the health industry, nursing facilities are generally managed independently, and workers at these facilities are not appropriately allocated.


Figure 1: Growth Rate in Intangible Assets by Industry

Figure 2: Growth Rate in Intangible Assets by Industry and Component (2000-08)


4. Growth Accounting Including Intangible Assets

Based on the estimates from the previous section, we use growth accounting including intangible assets to examine their impact on productivity growth. However, when we do growth accounting using the JIP database, we have to create the Divisia index for capital and labor input for the production function, in order to aggregate up various types of capital and labor.10 This applies to the case when we incorporate intangible assets in growth accounting. As we use JIP data for labor and tangible assets, we first create the Divisia index for the intangible assets and use it as an input to produce the estimation.

The production function in industry i for value added, Vit , can be expressed as:

(1) Yt,i Ht,i Vt,i At,iF(Lt,i,Kt,i,Zt,i)

where A is TFP, Y is value added before considering intangible assets, L is labor, K is tangible assets, and Z is intangible assets. We assume that production function, F, is constant returns to scale, which in turn implies that it is homogenous of degree 1 in each input, L, K, and Z. Then, equation (1) becomes,

(2) vt ,i F

Lt,i

Lt ,i

Vt ,i

lt ,i F

Kt,i

Kt ,i

Vt ,i

kt ,i F

Zt,i

Zt ,i

Vt ,i

zt ,i at ,i

where X

Xx

Figure 7 shows the results of growth accounting based on equation (2). As for the estimation strategy for growth accounting, we can estimate the production function in equation (1) using the growth rates of the value added and inputs, while interpreting coefficients for each input as a share. Alternatively, we can calculate the share of the inputs using the data on labor income and capital spending and use the growth rates of value added and inputs to conduct the above growth accounting exercise. Using the former estimation method, we can estimate the production function either for the whole of industry or at the sector level, and thus can obtain input shares common over time and across industries. The Japanese economy has experienced drastic structural change during the period for which we conducted our analysis. Therefore, although the above equation expresses input shares as the ratio of returns to each input to value added, we calculate input shares as the ratio of returns to each input to total costs. This is due to the fact that capital share may become negative when calculated as 1 – labor share using the actual data. In this case, we calculate capital share by adding capital stock to the cost of capital (CC). The cost of capital j is calculated as

10 Please refer to Chapter 1 of Fukao and Miyagawa (2008) for growth accounting using the Divisia index.


CCt,i pt ,i(i t ,i ) /(1 u) where pj is investment goods price for capital j , i is the nominal interest rate

(calculated as the weighted sum of the bond rate and long-term prime rate).11 j is the depreciation rate of capital j, and j is expected growth rate of investment goods price. We assume perfect foresight for the expected growth rate of investment goods price and use the growth rate of the price between this period and the next period. We can calculate the cost of capital (CC) independent of tangible and intangible assets.

The growth accounting exercise based on the above calculations indicates that the market economy saw a slight recovery in the growth rates of value added after 1995, following the end of the bubble economy in Japan. This is due to both sustained contributions by the intangible assets and the recovery of the TFP growth rates. Although the contributions from intangible assets declined in the 2000s, the growth rates of value added during this period were sustained at approximately the same level as in the 1990s, due to regulations on the reduction in labor input.

Table 7: Growth Accounting Incorporating Intangible Assets

LI: labor input; TAI: tangible asset input; IAI: intangible asset input.

11 The weights used here are the debt to total assets ratios by industry.


When we compare the manufacturing sector and the non-manufacturing sector, the growth rate of value added in the service sector has declined, while that in the manufacturing sector recovered rapidly in the 2000s. In both sectors, contributions from intangible assets have declined. There is a difference in the growth rates of value added across the two sectors, because the manufacturing sector saw an increase in the residual term (i.e. TFP growth rate), while the service sector saw a decline in investment in tangible capital. Therefore, it is uncertain whether the accumulation of intangible assets has contributed to the growth in productivity.

As shown in the various Tables, we conducted a growth accounting exercise, after classifying industries into the IT sector and the non-IT sector. There is a large difference in the growth rate of value added between the IT sector and the non-IT sector. While the IT sector has contributed to the GDP growth for the nearly three decades, the non-IT sector has continuously experienced negative growth rates since the end of the bubble economy in Japan. The non-IT sector barely contributed to the hiring of labor, investment in tangible capital, or investment in intangible capital. It was the IT sector alone that contributed to the increase in hiring and investment. However, in the 2000s, both tangible and intangible investment in the IT sector declined while hiring of labor slightly increased. The TFP growth rate is also declining, and the economy as a whole continues to be stagnant.

Traditional growth accounting does not include intangible investment and the TFP growth rate (called Solow residual) based on this traditional approach is considered to be including the contribution to TFP growth from intangible assets. The question remains whether or not a TFP growth rate (Solow residual) based on growth accounting that incorporates intangible assets is smaller than one based on the traditional approach. Figure 3 shows a comparison of the TFP growth rate based on the traditional growth accounting and that based on the growth accounting that incorporates intangible assets for the market economy and the IT sector.


Figure 3-1: Standard TFP Growth Rate and Modified TFP Growth Rate (Market Economy)

Figure 3-2: Standard TFP Growth Rate and Modified TFP Growth Rate (IT sector)


Figure 3 shows that in most cases, the Solow residual is higher than the TFP growth rate based on the alternative approach. However, the modified TFP growth rate exceeds the Solow residual for the IT sector in the late 1990s. Why does this happen?

We can rearrange equation (2) as

(3) yt,i

Yt,i

Vt,i

F

Lt ,i

Lt,i

Yt,i

Yt ,i

Vt ,i

lt ,i F

Kt,i

Kt,i

Yt,i

Yt,i

Vt,i

kt,i F

Zt ,i

Zt,i

Yt,i

Yt ,i

Vt ,i

zt,i Ý H t,iYt,i

Yt,i

Vt,i

Yt,i

Vt,i

at,i

We can further rearrange this into

(4) yt ,i F

Lt,i

Lt ,i

Yt,i

lt,i F

Kt,i

Kt ,i

Yt ,i

kt,i F

Zt,i

Zt ,i

Yt ,i

zt,i Ht,i

Yt,i

ht,i at,i

If we define the growth rate of Solow residual as SR, equation (4) implies

(5) SRt,i yt,i F

Lt,i

Lt ,i

Yt ,i

lt ,i F

Kt,i

Kt ,i

Yt ,i

kt ,i F

Zt,i

Zt ,i

Yt,i

zt ,i Ht,i

Yt,i

ht,i at,i

Equation (5) indicates that the deviation between the growth rate of the Solow residual and the modified TFP growth rate based on growth accounting that incorporates intangible asset investment depends on the growth rate of intangible capital stock and that of intangible investment. In other words, intangible assets include what were traditionally considered to be intermediate inputs, and counting these as part of intangible investment increases the value of value added, making it possible for the residual from this alternative approach to exceed the residual based on standard growth accounting. On the other hand, intangible investment, as a new production input, contributes to the production of value added and reduces the Solow residual. If the accumulation rate of intangible capital stock is high, the modified TFP growth rate subtracts intangible investment from the Solow residual. However, there are cases in which intangible investment rapidly increases and exceeds the accumulation rate of investment assets, and the modified value added growth rate increases. In such cases, the modified TFP growth rate will exceed the Solow residual. In the late 1990s, intangible investment rapidly increased in the IT sector due to an increase in new IT equipment. As a result, the modified TFP growth rate exceeded the Solow residual.12

Lastly, we compare our growth accounting results with those for the US and Europe published by Corrado et al (2012). Their accounting also incorporates intangible assets. Corrado et al (2012) decompose the labor productivity growth rate, which was measured using hours worked, into tangible assets, the capital deepening rate of intangible assets,

12 The argument developed in this paper is based on Basu et al. (2003). On a balanced growth path, the modified TFP growth rate would surely be lower than the growth rate of the Solow residual.


change in labor, and the TFP growth rate. We follow their method in decomposing the labor productivity growth rate, measured using hours worked, and use data from 1995 to 2007 as the comparison period.

Table 8 shows an international comparison of growth accounting after taking into account intangible assets. We find the Japanese capital deepening rate of intangible assets to be 0.2%, which is low compared to the international standard. The share of capital deepening of intangible assets in the labor productivity growth rate is only 9.5% in Japan, compared to 33.7% in the US and 19.9% in the EU respectively. Since 1995, the accumulation of intangible assets has played a key role in productivity growth in developed countries. On the other hand, labor productivity growth in Japan since 1995 has been attributed to a compositional shift in the labor market: an increase in high quality labor due to the popularization of higher education. However, there is a limit to this trend and once the number of people pursuing higher education hits a ceiling, this compositional effect will be muted. In this sense, it is necessary to accumulate intangible assets up to a level comparable with other developed countries.

Table 8: Factor Decomposition of Labor Productivity Growth Rate (1995-2007)

*Source: Corrado, et, al (2012) and estimates by the current authors.

5. Conclusion

Based on the framework of Corrado, Hulten and Sichel (2005, 2009), we estimated intangible investment by industry. Using the panel data retrieved from the JIP database, we were able to estimate intangible investment in 108 industries for the period from 1980 to 2008. The total annual expenditure in intangible assets in Japan was about 43 trillion yen


on average for the period 2001-08. Annual capital spending on intangibles was about 38 trillion yen in the same period. Its ratio to GVA in 2008 was 9.8%.

When we look at the intangible investment/GVA ratio by industry, IT intensive industries, such as machinery industries, and the information and communication service industry, have much higher ratios than those of other industries. On the other hand, in some service industries such as education, health and social work, and culture and entertainment industries, these ratios -- which have declined in the 2000s -- are smaller than those in Korea, as estimated in Chun, et al. (2012).

Using intangible investment data, we estimate capital stock by industry using the perpetual inventory method. The total capital stock in 2008 was about 136 trillion yen and the average annual growth rate of intangible capital in the market economy from 1985 to 2008 was 4.2%. However, the annual growth rate in the 2000s was very low. The slow growth of intangible assets in the 2000s was due to the decline in capital accumulation in economic competencies in many industries. This decline is attributed to the severe restructuring that occurred due to the long-term economic slump. In the non-market economy, intangible capital stock in computerized information declined in several industries.

Using our estimated data on intangible assets and the JIP database, we examined the effect of intangible investment on TFP growth through growth accounting incorporating intangible assets. Results showed that intangible investment in the market economy has contributed to TFP growth positively during the past three decades, when categorizing industries into the IT and non-IT sectors. The contributions of the intangible assets were not clear when industries were categorized in a traditional way. The non-IT sector has been declining since the end of the bubble economy in Japan. Therefore, we can infer that the recent stagnant growth in the Japanese economy has largely been influenced by the decline in overall investment (including intangible investment) in the IT sector, which played the role of a driving force for Japanese economic growth previously. As shown by the international comparison, the contribution of the intangible capital accumulation to productivity growth is weaker in Japan than it is in other developed countries.

Our estimation results suggest two important policy implications for the long-term productivity growth of the Japanese economy. First, since the stock of intangible assets is still lower in Japan than in other developed countries, there is room for improving labor productivity through intangible investment. Second, our results shed light on which industries we should see higher returns on intangible capital accumulations in. In the 2000s, based on the classification of industries into two sectors, the manufacturing and the service sector, policies focused on promoting growth in the service sector, whose growth was slower than that of the manufacturing sector. Such policies included promoting intangible capital investment. However, our results, as well as the studies by Basu, et al. (2003) and Corrado, et al. (2012), suggest that intangible investment is complementary to IT investment. Therefore, while we can expect a positive impact of intangible asset accumulation in the service industries which belong to the IT sector, its effect is more


uncertain for the non-IT sector. However, further quantitative analysis is necessary before we can provide precise

policy suggestions. We would need to estimate the production function that incorporates intangible assets and examine whether the returns on intangible assets are much higher than the returns on alternative assets. As richer data on intangible investment becomes available in the US and in Europe, econometric analysis on intangible capital investment, including international comparison, will be on our next research agenda.


A

ppen

dix

1: M

easu

rem

ent o

f In

tang

ible

Inv

estm

ent


Appendix 2: Measurement of Intangible Investment


non-IT intensive manufacturing sector


References

Barnes, P. and A. McClure. (2009), “Investments in Intangible Assets and Australia’s Productivity Growth,” Productivity Commission Staff Working Paper.

Barnes, P. (2010), “Investments in Intangible Assets and Australia’s Productivity Growth: Sectoral Estimates,” Productivity Commission Staff Working Paper, Productivity Commission, Melbourne.

Basu, S. J. G. Fernald, N. Oulton and S. Srinivasan. (2003), “The Case of the Missing Productivity Growth: Or, Does Information Technology Explain Why Productivity Accelerated in the United States but not in the United Kingdom?” in M. Gertler and K. Rogoff (eds.), NBER Macroeconomics Annual, 2003, MIT Press, pp.9-63.

Bresnahan, T. F., E. Brynjolfsson and L. Hitt. (2002), “Information Technology, Workplace Organization, and the Demand for Skilled Labor: Firm-Level Evidence”, Quarterly Journal of Economics 117, pp. 339-76.

Chun, H., K. Fukao, S. Hisa and T. Miyagawa. (2012), “Measurement of Intangible Investment by Industry and Its Role in Productivity Improvement Utilizing Comparative Studies between Japan and Korea,” RIETI Discussion Paper Series 12-0-037.

Corrado, C., C. Hulten and D. Sichel (2005), “Measuring Capital and Technology: An Extended Framework,” in Carol Corrado, John C. Haltiwanger, Daniel E. Sichel, eds., Measuring Capital in the New Economy, 11-46, The University of Chicago Press, Chicago.

Corrado, C., C. Hulten, and D Sichel (2009), “Intangible Capital and U.S. Economic Growth.” Review of Income and Wealth 55 (3), pp. 658-660.

Corrado, C., C. Haskel, C. Jona-Lasinio, and M. Iommi. (2012), “Intangible Capital and Growth in Advanced Economies: Measurement and Comparative Results” presented at the 2nd World KLEMS Conference at Harvard University.

Delbecque, V. and A. Bounfour (2011), “Intangible Investment: Contribution to Growth and Innovation Policy Issues” The European Chair on Intellectual Capital Management Working Paper Series, No. 2011-1A.

Fukao, K. T. Miyagawa, K. Mukai, Y. Shinoda and K. Tonogi. (2009), “Intangible Investment in Japan: Measurement and Contribution to Economic Growth”. Review of Income and Wealth 55(3), pp.717-736.

Fukao, K., T. Miyagawa. H. K. Pyo and K. Rhee (2012) “Estimates of Multi Factor Productivity, ICT Contributions and Resource Reallocation Effects in Japan and Korea” M. Mas and R. Stahler eds., Industrial Productivity in Europe: Growth and Crisis, Edward Elgar Publishing

Hall, R. (2000). “E-Capital: The Link between the Stock Market and the Labor Market in the 1990s,” Brookings Papers on Economic Activity, pp. 73-118.

Hall, R. (2001). “The Stock Market and Capital Accumulation,” American Economic Review 91, pp. 1185-1202.


Hao, J., V. Manole and B. van Ark. (2008). “Intangible Capital and Growth - An International Comparison, Economics Program” Working Paper Series, EPWP #08-14, Conference Board.

Inklaar, R., M. O’Mahony and M. Timmer. (2005). “ICT and Europe’s Productivity Performance: Industry-Level Growth Account Comparisons with the United States,” Review of Income and Wealth 51, pp. 505-536.

Jorgenson, D., M. Ho and K. Stiroh. (2005). Information Technology and the American Growth Resurgence, MIT Press.

Marrano, M, J. Haskel and G. Wallis. (2009). "What Happened to the Knowledge Economy? ICT, Intangible Investment, and Britain’s Productivity Revisited,” Review of Income and Wealth 55(3), pp. 661-716.

Miyagawa, T. and Y. Kim. (2008). “Measuring Organizational Capital in Japan: An Empirical Assessment Using Firm-Level Data”, Seoul Journal of Economics, 21, pp. 171-193.

Piekkola, H. (2011). “Intangible Capital in EU 27 – Drivers of Growth and Location in the EU” Proceedings of the University of Vaasa. Reports 166.

Pyo, H., H. Chun, and K. Rhee. (2010). “The Productivity Performance in Korean Industries (1990-2008): Estimates from KIP Database,” presented at RIETI/COE Hi-Stat International Workshop on Establishing Industrial Productivity Database for China (CIP), India (IIP), Japan (JIP), and Korea (KIP) held on October 22, 2010.

Robinson, P. and N. Shimizu. (2006). “Japanese Corporate Restructuring: CEO Priorities as a Window on Environmental and Organizational Change,” The Academy of Management Perspectives, 20(3) pp. 44-75.

Ohki E, (2003) “Gyousekishugi to Kyoiku Kunren Toushi”, K Imano eds, “Ko to Soshiki no Seikasyugi” Chuo Econonmic Publisher. (trans “Meritocracy and Investment in Educational Training”)

Fukao K. and Miyagawa T., (2008). “Seisan sei to Nihon no Keizai Seicho” University of Tokyo Publisher (trans “Productivity and the Japanese Economic Growth”).

Measurement of Intangible Investment by Industry and ... · Measurement of Intangible Investment by...

Documents

Transcript of Measurement of Intangible Investment by Industry and ... · Measurement of Intangible Investment by...