Company training session
VoxEU Column Productivity and Innovation

The decline in capital formation in Japan

The problem of weak physical investment has hindered firms in advanced economies since the global financial crisis. But evaluating corporate earnings based on physical assets alone ignores intangible assets such as R&D, which spur innovation and productivity. This column studies investment in Japan and finds that firms have not shifted their investment from physical to intangible assets. Instead, investment in education and training remains weaker than in other advanced economies. Supporting investments in both tangible and intangible assets – rather than focusing on either individually – will bolster Japan’s declining capital formation.

Japanese firms’ rate of return on capital in the late 2010s was higher than in the early 2010s or the 1990s. Generally, a high return on capital encourages aggressive physical investment. After the bubble economy collapsed, the Japanese economy faced a long period of secular stagnation, and in recent decades especially, the Japanese economy needed more investment to recover. Nevertheless, firms in Japan have not sufficiently invested in physical assets. Figure 1 shows the rate of return on capital and the rate of physical investment in Japan.

Figure 1 Rate of return and physical investment ratio

Figure 1 Rate of return and physical investment ratio

Source: Data from SNA (Cabinet Office in Japan).

The rate of return on capital was higher during 2002–2007 and 2013–2017 than before 1997. In such circumstances, firms would be expected to increase their physical investment, but the actual investment ratio has not increased.

Before 1997, the rate of physical investment was high. From 1997 to 2009, the rate dropped drastically. In particular, during the 2000s, the average rate of physical investment fell below the depreciation rate of physical asset capital, about 8%. After 2011, the rate of physical investment recovered, but it has not reached the pre-1997 level. Why haven’t Japanese firms sufficiently increased physical asset investment?

The problem of weak investment in physical assets like this has been a common problem among advanced countries since the global financial crisis. Gutiérrez and Philippon (2017) and Crouzet and Eberly (2018) argue that shifting to intangible asset investments causes lower physical investment. Corporate earnings are the result of both physical assets and intangible assets, which are not observable. Evaluating corporate earnings based only on physical assets overestimates the contribution of physical assets by ignoring unobservable, intangible assets. Therefore, there may be a significant relationship between the shortage of physical investment and investment in intangible assets. Gutiérrez and Philippon (2017) and Crouzet and Eberly (2018) show that weak physical investment in the US is a problem of mismeasurement, as intangible assets are unmeasured. Various studies demonstrate the recent trend of stagnation in physical investment. Zhang (2020) reached a similar conclusion in the UK. Miyagawa and Ishikawa (2021) confirmed this relationship using industrial databases (the JIP database and EU KLEMS database).

Have Japanese firms shifted their investment from physical assets to intangible assets, as Crouzet and Eberly (2018) and others have argued? Using firm-level data, we verify whether their argument applies to current Japanese investments.

My study (Ishikawa 2023) is based on the standard investment theory, Tobin’s Q. Tobin’s average Q is the market value of a corporation divided by the replacement cost of capital. The corporate value equals the market value of the stock plus total liabilities. And the replacement cost of capital is measured using the current value of physical assets. Here, the market value includes the contribution of intangible assets. But the replacement cost measures only physical assets, so it does not take into account intangible assets. Therefore, the average Q is an overestimation because the replacement cost of capital is underestimated.

Mismeasurement of Tobin’s Q shows the difference between the theoretical and the actual value of physical investment. This difference is what I have named ‘the investment gap’. If Tobin’s Q is high because of the contribution of intangible assets, the theoretical Q would indicate a high physical investment ratio. However, the actual measured physical investment ratio would be low because of the ignored intangible assets. In this situation, the investment gap has a negative value.

I study the degree to which this gap has occurred in Japan and the degree to which research and development (R&D) investment explains this gap, using Japanese listed firm panel data from the Nikkei Economic Electronic Databank System (NEEDS).

The results show that an investment gap exists in the service sector in Japan. Figure 2 shows the investment gap and revised investment gap in the service sector. The revised investment gap is a residual that R&D expenditure cannot explain.

Figure 2 Investment gap and revised investment gap in the service sector

Figure 2 Investment gap and revised investment gap in the service sector

Before the first half of the 1990s, the revised investment gap showed positive value. This means Japanese companies spent a large amount on R&D. During the late 1990s and the early 2010s, R&D expenditure covered the shortage of physical investment. Especially in 2010, R&D covered almost 80%. In this period, the Japanese service sector’s physical investment was consistent with the analysis of Crouzet and Eberly (2018). However, after 2013, the revised investment gap significantly expanded with the drastic expansion of the investment gap. In this framework, R&D expenditure is not a sufficient explanation for weak physical investment.

This period, after 2013, was when Prime Minister Shinzo Abe launched his extremely prominent economic policy. ‘Abenomics’ set activation of private sector investment as one of its goals. However, R&D does not account for the shortage of physical investment (see Figure 1). This suggests the growing importance of non-R&D intangible assets. For example, human capital and brand assets may have gained more market value. If a firm has gained from monopoly profit, the market value may be increased.

It will be necessary to continue actively promoting intangible asset investment activities in the future, as intangible assets are becoming increasingly important. For example, investment in education and training is essential for accumulating human capital. In fact, according to the JIP database, investment in training human capital is weaker than in other advanced countries. And ICT investments, including software, are also crucial for more efficient production activities.

Intangible investments, including R&D expenditures, are a source of innovation and are key to improving productivity. Weak intangible investment affects the rate of economic growth in the long term. Economic growth also requires physical investment, as traditional investments in physical assets such as construction, machinery, and equipment are important for capturing new technologies and for innovation. Therefore, it is necessary to support investments in both tangible and intangible assets rather than focusing on either individually.


Crouzet, N and J C Eberly (2018), “Understanding Weak Capital Investment: The Role of Market Concentration and Intangibles”, presented at Federal Reserve Bank of Kansas City Conference.

Gutiérrez, G and T Philippon (2017), “Investment-less growth: An empirical investigation,” Brooking Papers on Economic Activity (2): 89–190.

Ishikawa, T (2023), “The Decline in Capital Formation in Japan: Empirical research on Japanese listed firms’ data”, RIETI Discussion Paper Series, 23-E-008.

Miyagawa, T and T Ishikawa (2021), “Declining Capital Formation in Japan and the Role of Intangibles -Empirical Studies using Industry-level Data”, RIETI Policy discussion paper Series, 21-J-020 (in Japanese).

Zhang, D (2020), “Intangibles and the UK Under-investment Puzzle: Evidence from Firm-level Panel Data”, Economics Letters 194(109389).