Keywords

5.1 Introduction

In the 1990s, new types of firms such as Amazon and Google were founded and grew rapidly under the IT revolution. There are several characteristics of these firms. As Brynjolfsson (2004) pointed out, they developed new software, invested in human capital, and formed organizational structures that enabled faster decision-making. Due to the success of these firms, economists have paid attention to the role of intangible assets on firm performance and firm value. Corrado et al. (2009) measured comprehensive intangible investment including software investment, investment in human capital, and reform in organizational structure, and showed the significant contribution of intangible assets to US economic growth. Following Corrado et al. (2009), the positive effects of intangible assets on economic growth were found in the advanced countries.Footnote 1

At the firm level, there have been several studies on the effects of R&D investment, which is a part of intangible investment on firm performances and firm value.Footnote 2 However, Hall (2000, 2001) pointed out that after the IT revolution, the stock market may be evaluating not only R&D stocks but also other types of intangible assets positively. To examine the determinants of firm value after the IT revolution, we need to measure a broader concept of intangible assets beyond R&D assets like Corrado et al. (2009).

Thus, in our paper, we measure comprehensive intangible assets following Corrado et al. (2009) by using data of Japanese listed firms. Based on our measurement, we examine the relationship between firm value and intangible assets, and estimate Tobin’s Q using not only intangible but also tangible assets. From the above studies, we find that the mean value of Tobin’s average Q becomes close to 1 and its variance becomes small when we consider intangible assets, as Hall (2000, 2001) expected. We also find that intangible assets are positively correlated with firm value. The estimation results show that the accumulation of intangible assets significantly increases firm value. The effect is particularly pronounced and significant in the IT related industries.

Our study consists of six sections. In the next section, we review the existing literature on the measurement of intangible assets and how intangible assets are evaluated in the stock market. In the third section, we explain how we measure intangible assets. In the fourth section, we examine several features of Tobin’s Q that take intangible assets into account. In the fifth section, we examine the effects of intangible assets on firm value by estimating a standard average Tobin’s Q. In the last section, we summarize our findings.

5.2 Intangible Assets and Firm Value: A Literature Review

Hall (2000, 2001) pointed out that the Tobin’s Q in the US consistently exceeded 1. He subsequently argued that as adjustment costs of tangible investment are accumulated as intangible assets within a firm, the gap between Tobin’s Q and 1 is accounted for intangible assets.Footnote 3 To examine Hall’s proposition, Brynjolfsson et al. (2002) estimated firm value using non-IT capital and IT capital, and found that the coefficients of IT capital were much greater than those of non-IT capital. Then, they argued that these large coefficients were affected by intangible assets, complementary to IT capital. Cummins (2005) and Miyagawa and Kim (2008) estimated firm value using not only non-IT capital and IT capital but also with R&D capital and advertisement capital. Although Cummins (2005) did not find a higher than normal rate of return for intangible assets, Miyagawa and Kim (2008) obtained the opposite results to Cummins (2005).

Although Cummins (2005) and Miyagawa and Kim (2008) focused on R&D capital and advertisement capital, Lev and Radhakrishnan (2005) recognized a portion of sales, general and administrative expenditures as organizational capital. By estimating the difference between market value and book value using organizational capital, they found that organizational capital significantly contributed to market value. Hulten and Hao (2008) estimated firm value of pharmaceutical companies by R&D capital, and organizational capital measured from sales, general and administrative expenditures, and showed that both of these types of intangible assets contributed to increasing firm value.

Abowd et al. (2005) constructed their own measure with respect to quality of human capital from employer-employee datasets. They estimated firm value by obtaining Compustat data using the measure of quality of human capital, and found that their measure was positively correlated with the value of the firm. Bloom and Van Reenen (2007) also constructed their own management score taking organizational management and human resource management into account, using their interview surveys. They showed that this management score was positively correlated with Tobin’s Q. Görzig and Görnig (2012) measured intangible assets by estimating the share of labor costs of IT, R&D, and management and marketing employees. Once they considered intangible assets, they showed that the dispersion of rate of return on capital was reduced dramatically.

5.3 Measurement of Intangible Assets in Japanese Listed Firms

Although previous studies have shown the contribution of intangible assets to firm value, they did not capture comprehensive intangible assets like Corrado et al. (2009). Therefore, among intangible assets classified by Corrado et al. (2009), we measure five types of intangibles; software, R&D, brand equity, firm specific human capital, and organizational change. This concept of intangibles is broader than that of previous studies.Footnote 4

Corrado et al. (2009) classified intangible assets into three categories: computerized information, innovative property, and economic competencies. Software investment is a part of investment in computerized information consists of three types of software; custom software investment, packaged software investment, and own account software investment. R&D investment is included in investment in innovative property.Footnote 5 Investment in economic competencies consists of brand equity, firm specific human capital, and organizational change. We measure these three components depending on the data in DBJ Corporate Financial Databank. The detailed methods we use to measure the five items mentioned above for each firm are as follows:

  1. 1.

    Software: First, the ratio of workers engaged in information processing to the total number of employee is multiplied by the total cash earnings in order to measure the value of software investment. Then, we add the cost of information processing to this number to find total software investment. All the information is obtained from Basic Survey of Business Activities of Enterprises (BSBAE). We deflate this number by the deflator for software investment in the Japan Industrial Productivity (JIP) database.Footnote 6 , Footnote 7

  2. 2.

    Research and Development (R&D): We subtract the cost of acquiring fixed assets for research from the cost of R&D (i.e., in-house R&D and contract R&D) to estimate the value of investment into R&D. All the information is obtained from BSBAE. The output deflator for research (private) in the JIP database is used to deflate this R&D investment.

  3. 3.

    Brand equity: Brand equity is measured based on expenditures on advertising. The data of advertising expenses are obtained from the DBJ Corporate Financial Databank. We use the output deflator for advertising in the JIP database as the deflator for advertising investments.

  4. 4.

    Firm specific human capital: First, we estimate each firm’s investment on firm-specific skills by multiplying (1) the total labor cost in the DBJ Corporate Financial Databank with (2) the industry-average ratio of total employee training cost to the total labor cost for each firm from the General Survey of Working Conditions and (3) the ratio of the on-the-job and off-the-job training costs for firm-specific skills to the total education cost (0.37).Footnote 8 In order to further consider the opportunity cost of the off-the-job training cost for skill improvement, we multiply the number computed in the abovementioned procedure to 2.51.Footnote 9

  5. 5.

    Organizational change: Following Robinson and Shimizu (2006), who conducted a survey of the time-use of Japanese CEOs, we assume that 9 % of board members’ compensation—which we can obtain from the DBJ Corporate Financial Databank—accounts for investment in organizational change. This is deflated by the output deflator for education (private and non-profit) in the JIP database.

For all five investment category data detailed above, we employ the Perpetual Inventory (PI) method, in which we use FY1995 as the base year, to construct a data series of intangible assets from FY2000. All depreciation rates used for this computation follow that of Corrado et al. (2012).Footnote 10

5.4 Tobin’s Q with Intangibles

The conventional Tobin’s Q (Q C it ) at the firm level is measured as the ratio of firm value (V it ) to the replacement value of tangible assets ((1 − δ K )K it − 1) at the initial period of t.Footnote 11

$$ {Q}_{it}^C=\frac{V_{it}}{\left(1-{\delta}_K\right){K}_{it}} $$
(5.1)

where δ k is the depreciation rate of tangible assets. We measure the conventional Tobin’s Q as follows:

The conventional Tobin’s Q = (Stock value + Book values of commercial paper, corporate bond, and long-term debt)/(1 − δ K ) * (Replacement values of tangible assets + Inventory-Short-term debt).

As shown by Lindenberg and Ross (1981), and Hall (2000, 2001) for the US and Tanaka and Miyagawa (2011) for Japan, the standard Q expressed by (1) has persistently exceeded 1. The mean value of the conventional Tobin’s Q shown in Table 5.1 is also 1.40.

Table 5.1 Conventional Tobin’s Q (all sectors)
Full size table

Lindenberg and Ross (1981) explained the gap between the measured conventional Q and 1 as being due to monopoly rents, although they knew that unmeasured intangibles affected this gap. When we measure the Tobin’s Q considering intangible assets (N it − 1) as measured in Sect. 5.3, the revised Tobin’s Q (Q R it ) is expressed as follows:

$$ {Q}_{it}^R=\frac{V_{it}}{\left(1-{\delta}_K\right){K}_{it}+\left(1-{\delta}_N\right){N}_{it}} $$
(5.2)

where δ N is the depreciation rate in intangible assets.

We show a revised Tobin’s Q including intangible assets in Table 5.2. The mean value of the revised Tobin’s Q is 0.99 which is almost equal to 1. The difference between the two mean values is significant. The standard deviation of the revised Q is smaller than that of the conventional Q, which is consistent with the results of Görzig and Görnig (2012), who showed that the dispersion of profit rates when including intangible assets is smaller than that without intangibles. The distributions of two types of Tobin’s Q are shown in Fig. 5.1. We find that the revised Tobin’s Q is distributed around 1 compared to the conventional one. The Kolmogorov-Smirnov test rejected the hypothesis that the two distributions are the same.

Table 5.2 Revised Tobin’s Q (all sectors)
Full size table
Fig. 5.1
figure 1

Density of Tobin’s Q

Full size image

We divide all samples into two sectors: IT sectors and non-IT sectors.Footnote 12 The mean value of Tobin’s Q in IT sectors is higher than that in non-IT sectors in both cases. However, the mean value of the revised Q in the IT sectors is 1.13, which is much closer to 1 than the mean value of the conventional Q in the IT sectors. Also, the standard deviation of the revised Q in the IT sectors is reduced compared to that of conventional Q in the IT sectors (Tables 5.3, 5.4, 5.5, and 5.6).

Table 5.3 Conventional Tobin’s Q (IT sectors)
Full size table
Table 5.4 Revised Tobin’s Q (IT sectors)
Full size table
Table 5.5 Conventional Tobin’s Q (non-IT sectors)
Full size table
Table 5.6 Revised Tobin’s Q (non-IT sectors)
Full size table

Arato and Yamada (2012) measured aggregate intangible assets based on DBJ data. Their estimated ratio of intangible assets to tangible assets is 0.47 in the 1980s. As shown in Table 5.7, the corresponding rate of our estimates is 0.45, which is similar to that of Arato and Yamada (2012). The result shows that the ratio of intangible assets to tangible assets has not changed in Japan.

Table 5.7 Statistics of the ratio of intangible assets to tangible assets (N/K)
Full size table

5.5 Do Intangible Assets Explain the Overvaluation of Tobin’s Q?

5.5.1 The Relationship of the Conventional Tobin’s Q with Intangibles

Although the revised Q is almost equal to 1 on average, the Tobin’s Q in each firm deviates from 1. Thus, we econometrically check the effects of intangible assets on the variation of Tobin’s Q. As we introduced in Sect. 5.2, Brynjolfsson et al. (2002), Cummins (2005) and Miyagawa and Kim (2008) estimated the effects of intangible assets on firm value. However, these studies focused on fewer components of intangibles than those classified by Corrado et al. (2009). Therefore, we examine the effect of intangibles following the classification by Corrado et al. (2009) on firm value.

Following Bond and Cummins (2000), the profit function (π) depends on tangible and intangible capital. Dividends at firm i (Di) are expressed as follows:

$$ {D}_{it}=\pi \left({K}_{it},{N}_{it}\right)-{I}_{it}-{O}_{it}-G\left({I}_{it},{K}_{it}\right)-H\left({O}_{it},{N}_{it}\right) $$
(5.3)

where I is investment in tangible assets, O is investment in intangible assets, and G and H are adjustment cost functions in tangible investment intangible investment, respectively.Footnote 13

$$ \begin{array}{cc}\hfill G\left({I}_{it},{K}_{it}\right)\hfill & \hfill =\frac{a}{2}{\left(\frac{I_{it}}{K_{it}}\right)}^2{K}_{it}\hfill \\ {}\hfill H\left({O}_{it,}{N}_{it}\right)\hfill & \hfill =\frac{b}{2}{\left(\frac{O_{it}}{N_{it}}\right)}^2{N}_{it}\hfill \end{array} $$

Capital accumulation in tangible assets and intangible assets is expressed as follows:

$$ \begin{array}{cc}\hfill {K}_{it}\hfill & \hfill ={I}_{it}+\left(1-{\delta}_K\right){K}_{it-1}\hfill \\ {}\hfill {N}_{it}\hfill & \hfill ={O}_{it}+\left(1-{\delta}_N\right){N}_{it-1}\hfill \end{array} $$

We solve the optimization problems of firm i with respect to I, and O.

$$ {q}_{Kt}=1+a\left(\frac{I_{it}}{K_{it}}\right) $$
(5.4a)
$$ {q}_{Nt}=1+b\left(\frac{O_{it}}{N_{it}}\right) $$
(5.4b)

where qK and qN are Lagrange multipliers.

When the profit function is linear homogeneous, the firm value of firm i is expressed as a linear combination of each asset (Wildasin (1984) and Hayashi and Inoue (1991)).

$$ {V}_{it}={q}_{Kt}\left(1-{\delta}_K\right){K}_{it}+{q}_{Nt}\left(1-{\delta}_N\right){N}_{it} $$
(5.5)

From Eq. (5.5),

$$ {q}_{Kt}=\frac{V_{it}}{\left(1-{\delta}_K\right){K}_{it}}-{q}_{Nt}\frac{\left(1-{\delta}_N\right){N}_{it}}{\left(1-{\delta}_K\right){K}_{it}} $$
(5.6)

Substituting Eqs. (5.4a) and (5.4b) into Eq. (5.6), we obtain:

$$ \begin{array}{cc}\hfill {Q}_{it}^C-1\hfill & \hfill = a\left(\frac{I_{it}}{K_{it}}\right)+\{1+ b(\frac{O_{it}}{N_{it}})\}\frac{\left(1-{\delta}_N\right)}{\left(1-{\delta}_K\right)}(\frac{N_{it}}{K_{it}})\hfill \\ {}\hfill \hfill & \hfill = a\left(\frac{I_{it}}{K_{it}}\right)+\frac{\left(1-{\delta}_N\right)}{\left(1-{\delta}_K\right)}(\frac{N_{it}}{K_{it}})+ b\frac{\left(1-{\delta}_N\right)}{\left(1-{\delta}_K\right)}(\frac{O_{it}}{K_{it}})\hfill \end{array} $$
(5.7)

where \( {Q}_{it}^C=\frac{V_{it}}{\left(1-{\delta}_K\right){K}_{it}} \) is the standard average Q at firm i.

Equation (5.7) implies that the gap between the conventional Q ratio and 1 is explained by the ratio of intangible assets to tangible assets, the gross tangible investment/tangible assets ratio, and the gross intangible assets ratio.

5.5.2 Estimation Results

Based on Eq. (5.7), we estimate the following equation:

$$ {Q}_{it}^C-1= const.+{\alpha}_1\left(\frac{N_{it}}{K_{it}}\right)+{\alpha}_2\left(\frac{I_{it}}{K_{it}}\right)+{\alpha}_3\left(\frac{O_{it}}{K_{it}}\right)+{\displaystyle \sum_{j=1}^n{\beta}_j{X}_{ijt}+{\varepsilon}_{it}} $$
(5.8)

In Eq. (5.8), Xij is a control variable. Lindenberg and Ross (1981) pointed out that monopoly rents explained the overvaluation of firm value. In addition, financial constraints may affect the gap between a standard Q and 1. Then, we also estimate Eq. (5.8) with a price cost margin or external finance dependence as defined by Rajan and Zingales (1998). We expect that the coefficient of external finance dependence will be negative because a greater dependence on external finance reduces firm value. The basic statistics of the variables used in our estimation are summarized in Table 5.8.

Table 5.8 Statistics of the sample
Full size table

First, we estimate Eq. (5.8) by OLS. To avoid endogeneity, we take a 1-year lag for all explanatory variables except firm age. The estimation results are shown in Table 5.9. In Column (1), we focus on the effect of intangible assets on the overvaluation of the conventional Q. In this estimation, the ratio of intangible to tangible assets significantly explains the overvaluation of the Q ratio. In Column (2), we regress firm value on three variables included in Eq. (5.7). The estimation results show that all variables are positive and the ratio of intangible to tangible assets, and the tangible investment/tangible assets ratio are significant. Due to the strong correlation between intangible assets/tangible assets and intangible investment/tangible assets ratio, the coefficient of intangible investment/tangible assets ratio may be not significant.

Table 5.9 OLS estimates of determinants of conventional Tobin’s Q-1
Full size table

In Columns (3) and (4), we estimate Eq. (5.8) including control variables. In Column (3), all three variables in Eq. (5.7) are positive and significant. In addition, the coefficient of external finance dependence is negative and insignificant, as we expected. In Column (4), the ratio of intangible assets to tangible assets and the price cost margin are positive and significant, while intangible and tangible investments are not significant.

Next, we estimate Eq. (5.8) utilizing the instrumental variable method. Instruments are the ratio of white-collar to total workers, and external finance dependence. The results in Table 5.10 indicate that the ratio of intangible assets to tangible assets is positive and significant in all estimations. However, the intangible investment/tangible assets ratio is negative in Columns (2) and (3). It is possible that negative coefficients of intangible investment/tangible assets are caused by the multicollinearity between intangible assets and intangible investment.

Table 5.10 Instrumental variable (IV) estimates of determinants of conventional Tobin’s Q-1
Full size table

We also conduct panel estimations. As the Hausman test suggests that the random effect estimation is better than fixed effect estimation, we show the results of random effect estimations in Table 5.11. Table 5.11 shows that the ratio of intangible assets to tangible assets is positive and significant in all estimations. As the coefficient of price cost margin is also positive and significant, monopoly rents also contribute to the valuation of firm, as Lindenberg and Ross (1981) suggested.

Table 5.11 Panel estimate (random effect) of determinants of conventional Tobin’s Q-1
Full size table

Brynjolfsson et al. (2002), Basu et al. (2003), and Cummins (2005) emphasized that intangible assets are complementary to IT assets. Miyagawa and Hisa (2013) found that intangible investment in the IT sectors improve TFP growth. In Sect. 5.4, we found that the Tobin’s Q in IT sectors is higher than that in non-IT sectors. Then, we divide all samples into those in the IT sectors and non-IT sectors and estimate Eq. (5.8) by the instrumental variable method in each sector. Table 5.12 shows that estimation results in IT sectors are similar to those in Table 5.10. The ratio of intangible to tangible assets is positive and significant in all estimations when the coefficients of intangible and tangible investments are not significant. However, in the non-IT sectors, the coefficients of the ratio of intangible to tangible assets are not necessarily significant, while the signs of the coefficients are positive in all estimations. The estimation results in Table 5.12 imply that only intangible assets in the IT industries contribute significantly to the evaluation of firm value.Footnote 14 In addition, the price cost margin is positive and significant in the IT and non-IT sectors, as can be seen in Tables 5.10 and 5.11.

Table 5.12 Instrumental variable (IV) estimates of determinants of conventional Tobin’s Q-1 (IT or non-IT sectors)
Full size table

As explained in Sect. 5.3, we measure five types of intangible assets; software, R&D, brand equity, firm specific human capital, and organizational change. We examine what kind of assets the stock market assesses favorably. Estimation results in Table 5.13 show that the stock market assesses assets in software and firm specific human capital favorably, while the assessments of R&D, brand equity, and organizational change are inconclusive. These results imply that the stock market does not necessarily consider all components of intangibles as positive.

Table 5.13 Instrumental variable (IV) estimates of determinants of Conventional Tobin’s Q-1 (software, R&D, brand equity, human capital, and organizational change)
Full size table

Figure 5.1 shows that the sample deviation from the mean value is not symmetric. In this case, quantile regression—that estimates parameters based on the error measured as a deviation from the median value in each quantile—is useful to check the robustness of our results. We separate the distribution of a conventional Tobin’s Q into four quantiles and conduct quantile regression. Table 5.14 shows the estimation results of quantile regression that correspond to the OLS estimations in Table 5.9. As in Table 5.9, the firm value reflects intangible values in all estimations. In addition, intangible investment also contributes positively and significantly to the increase in firm value (Column (2)), while the coefficient of this variable is not significant in Table 5.9. As a result, the above two alternative estimations confirm the positive and significant contributions of intangible assets to firm value.

Table 5.14 Quantile regression of determinants of conventional Tobin’s Q-1
Full size table

5.6 Concluding Remarks

The IT revolution has changed the growth strategy of firms. Software investment has become as important as tangible investment. Firms have focused on accumulation in human capital and restructured their organizations to be compatible with the new technology. Many economists such as E. Brynjolfsson, C. Corrado, R. Hall, C. Hulten, B. Lev, and L. Nakamura summarized these new types of expenditures as intangible investment and examined its effects on firm value. However, many studies have focused on the effects of specific components of intangible assets on firm value, because it is difficult to measure intangibles at the firm level.

Based on the classification of intangibles by Corrado et al. (2009), we measure a broader concept of intangibles than those in the previous studies using the listed firm-level data in Japan. The mean value of Tobin’s Q including intangible assets is almost equal to 1, while the mean value of conventional Tobin’s Q exceeds 1, as Hall (2000, 2001) suggested. The standard deviation of the revised Q is smaller than that of the conventional Q, which is consistent with the results of Görzig and Görnig (2012). These results imply that stock prices reflect the value of intangibles.

Although the results also imply that the market concludes that there are no growth opportunities of Japanese listed firms on average in the 2000s, there are still differences in Tobin’s Q. The Tobin’s Q in the IT industries is consistently higher than that in the non-IT industries. This difference in market value suggests that firms in the IT industries should expand their businesses, and firms in the non-IT industries should restructure their businesses. The result is consistent with Miyagawa and Hisa (2013), who argued that intangible investment improves productivity in the IT industries. The Japanese government should take growth strategies such as to promote investment including intangibles in the IT industries and to assist firms in the non-IT industries transform themselves to a business in a growth industry.

Using our measures, we examined the effects of intangibles on firm value. Estimation results following Bond and Cummins (2000) showed that greater intangible assets increase firm value. As these results are robust in the IT industries in particular, they support our policy implications. However, not all intangible assets are valued in the stock market. The values of innovative property and economic competencies are inconclusive. One possible reason for the long-term slump of the Japanese stock market is that investors are not valuing high level R&D investment and human resources in Japanese firms. The upcoming reform in accounting standards that will evaluate intangible assets will contribute to the revitalization of the Japanese stock market.