Determinants of Total Factor Productivity and Japan's Potential Growth: An International Perspective

Four presentations were made during the session, the underlying themes being the issues of total factor productivity (TFP) and economic growth. The presentations were in the form of research reports relating to developments in TFP and economic growth, the first focusing on Japan, the second on the European Union and the United States, the third on countries worldwide, and the fourth on Korea, Taiwan, and China.

MIYAGAWA Tsutomu and FUKAO Kyoji gave the first presentation, a report on "Determinants of TFP Growth and Potential Growth in Japan." The following points were made:

Why are studies about the potential growth rate and TFP so important for the Japanese economy?

As the level of economic growth nears cruising speed, considerable interest is currently being devoted to the GDP growth rate (potential growth rate) which is sustainable over the medium to long term without causing inflation. Using the neoclassical economic growth model, and making certain assumptions (e.g., balanced growth, Harrod-neutral technological progress), the potential growth rate is determined by the labor input growth and TFP (production per unit of labor and capital inputs combined) growth rate.

As the Japanese population, which forms the basis of labor input, is declining, the rise in TFP is a key factor to the potential growth rate in Japan.

Outline of the projects on measuring productivity at industry/firm level

We have constructed the Japan Industrial Productivity Database 2006 (JIP 2006), covering the entire Japanese economy, and the JIP Microdatabase, which covers firms and establishments from the entire manufacturing sector and most of the non-manufacturing sector.

The JIP 2006 database includes data for calculating TFP, including total output, intermediate inputs, capital stock, and labor inputs, for 108 industries for the period from 1970-2002. For ancillary tables, data are also being prepared for trade, FDI, deregulation indicators, capacity utilization ratios, etc. These data have already been posted on the RIETI website (currently only in Japanese).

We also participate in and collaborate with EU KLEMS, a research project for the analysis of productivity in the EU.

Uptrend in TFP at the macro/industry level

We used the JIP 2006 database to calculate the TFP growth rate and confirmed that TFP growth during the 1990s was low on a macro basis. Similar results have been obtained in other research.

Examining the manufacturing industry and the non-manufacturing industry separately, we found that there was a decline in the TFP growth rate, particularly in the manufacturing industry. In addition, by looking in detail at which industries have been driving the rise in macro TFP, we confirmed that the growth could be accounted for principally by the commercial sector, specifically the wholesale and retail industries, and the financial and IT-related industries. With regard to the 1990s, the wholesale industry alone accounted for the macro-level TFP growth rate, with the increases and decreases in the other industries offsetting each other.

Sources of the TFP growth

By using microeconomic data to analyze the factors behind the decline in the TFP growth rate in the manufacturing industry, we identified that it has arisen because of reductions in internal effects and also negative effects arising from the withdrawal of firms with high productivity. The overseas relocation of plants by high-productivity firms can be identified as a reason for the negative character of the withdrawal. Also, a possible reason for the fall in the internal effects may be the reduction in the productivity increase effect due to the drop in the number of new plants, which normally raise productivity significantly.

Taking IT as the criterion, we aggregated industry-specific data in the JIP 2006 database into six industrial categories, and measured the rate of increase in TFP for each. This revealed that "IT manufacturing," the sector engaging in the production of IT goods had the highest rate of increase in TFP. We also found that, in recent periods, there was a high rate of increase in TFP in "IT-intensive service industries," i.e., service industries that make intensive use of IT goods as capital goods.

We also examined the role of intangible assets which is considered to be complementary to IT capital. There was a strong correlation between the average number years of school attendance and the ratio of IT investment, suggesting complementarities between the degree of IT usage and the usage of highly educated human resources. Also, the contribution of organizational capital to TFP growth was estimated to be approximately 10%.

The Japanese economy: Future issues and growth potential

TFP growth is a key to increasing growth potential, but this is not limited to technological progress alone: fostering greater IT usage and addressing globalization are also important. In addition, since declining fertility and aging of the population are negative factors for improving productivity, sustained policy measures to counter declining fertility are desirable. With respect to promoting the spread of IT, steps should be taken to increase productivity, principally by the IT manufacturing industry and by service industries that use IT. As for tackling globalization, an effective method would be to engage in mergers and acquisitions in order to secure markets rapidly for that purpose.

Q: There are a number of methods for calculating TFP, but which one is best to minimize bias? What calculation method did you use in this research?

A: (Prof. Fukao) We used the method known as growth accounting, which calculates TFP as a residual. As for the production factor inputs, we believe that we estimated labor and capital rigorously, taking their "quality" into consideration.

Q: It is not possible to convert Harrod-neutral technological progress to Hicks-neutral technological progress with a general production function, so you will need to use a special production function such as a Cobb-Douglas type.

A: (Prof. Fukao) Our calculation was made based on such a premise. For example, if the nature of technological progress is of the capital-saving type, the balanced growth rate will change.

Q: When calculating TFP for individual industries, it seems a bit unreasonable to apply the postulation of linear homogeneous production functions to every industry. What are your thoughts as to the bias that arises from doing that?

A: (Prof. Fukao) We applied growth accounting with the assumption of constant returns to scale, and it is possible that the scale factor may have been incorporated in the TFP figure which was calculated as the residual. Since we believe that the scale factor is especially important when conducting analyses at the plant level, we hope to improve this point in the future.

(Prof. Miyagawa) Earlier research has shown that constant returns did not deviate in many industries regarding the economies of scale. Because of this, we believe it is possible to use the assumption of linear homogeneity.

Q: The graph of the ratios of IT capital in service industries and the average number years of school attendance shows a correlation that IT inputs are also high in places with high-level of education on average. How should we view this connection and the question of why an increase in IT capital inputs in Japan has not led to growth in TFP unlike the case of the U.S.?

A: (Prof. Miyagawa) As you have pointed out, this does not show any particular causal relationship. All it shows is that if we consider people with high average numbers of years of education to be skilled labor, there is the possibility of complementarities between skilled labor and IT capital. What kind of impact this actually has on productivity is something that we will have to study in the future.

Q: There has already been extensive research on the causes of Japan's so-called lost decade, but have you made any new findings in your research?

A: (Prof. Fukao) Our research examined such matters as the quality of labor, and clarified the problem of the constant decline in the rate of growth of labor quality. In addition, our research shows a smaller rate of decline in TFP between the 1980s and 1990s compared to that found in earlier research, but we believe this to be the result of our having controlled the quality of labor and capital in our research.

Also, no other research examined industries in such detail. This focus in our research allowed us to shed light on the fact that it was the increase in certain industries that accounted for the macro-level TFP growth. For example, the wholesale industry showed a relatively high rate of productivity growth.

In analyses using microdata, as was seen in the analyses on the manufacturing industry, we confirmed that Japan's industrial metabolism has been low for a long time, although it had been believed that it had not declined until the 1990s. The findings of research hitherto, particularly those of foreign studies, have been that the withdrawal of low-productivity companies during recession causes growth in productivity through reallocation. However, the new finding from our research is that this effect was not so significant from the 1980s.

Q: IT capital and non-IT capital are not regarded as substitutes, so if the Japanese economy recovers and inputs of non-IT capital increase, would the use of IT capital then also rise?

A: (Prof. Miyagawa) Certainly there is a cyclical element in IT investment. In Japan's case, the economy was in recession in the beginning of the 1990s, and IT investment did not increase. This is a major difference with the U.S. Also, IT investment fell in 2002, indicating that the cyclical nature of IT investment was significant in Japan.

Nevertheless, looking at the long-term rate of increase, something that has been evident in Japan recently is that whereas IT investment increases structurally, non-IT investment tends not to increase substantially, even if the economy improves somewhat.

Q: In your research have you given consideration to energy problems and food risk?

A: (Prof. Fukao) Energy and food issues are important, but we have not considered them here.

The second presentation, by Professor Marcel P. TIMMER, was titled, "Can the EU Catch Up with the U.S. in Terms of the Level of Total Factor Productivity?" The following were among the points made.

A comparison of the EU (data covering 15 countries) and the U.S., based on GDP, GDP per capita (economic welfare), and GDP per working hour (productivity) data, shows that since World War II the EU has been catching up with the U.S. in terms of welfare and productivity. However, since 1975, although the EU's productivity has continued to catch up with the U.S., its welfare has not, resulting in a persistent gap. This can be attributed to the following phenomenon: although productivity has risen since the 1970s because of the progressive reduction of working hours, GDP itself has not risen, and thus the increase in welfare has stagnated.

Also, although productivity in the U.S. has risen since the mid-1990s, productivity in the EU has not risen as much as in the U.S., resulting in the tendency for the gap to widen somewhat.

When considering the reasons for the acceleration of productivity growth in the U.S., as well as the reason why it has not accelerated in the EU, IT may be cited as a factor in the acceleration. As for the EU, on the other hand, the underlying reason for the delay in the acceleration can be seen as being that productivity is lagging behind until the impact of IT appears.

Another aspect to be considered is that as a result of changes in the labor market, the number of people in employment has risen and productivity has declined, or the catch-up process has ceased, and in consequence the rise in productivity has slowed.

When we examined the contributions made by individual industries to macro labor productivity in the U.S. and the EU, we found that the contribution of the service industry in the U.S. was significant, whereas it was very small in the EU. Comparing the labor productivity of individual industries in the service sector in detail, we found that most industries in the U.S. had higher labor productivity. This trend was particularly significant in the commercial sector.

When we broke down TFP growth in the entire market economy into TFP growth in individual industries, we found that, in most EU countries, the contribution of service sectors was small, and was negative in the Netherlands and Germany. In contrast, in the U.S., the contribution of TFP growth in the service sector was shown to be substantial.

With regard to the TFP level in 1997, when we compared the U.S. and the EU we found several European countries in which TFP in their manufacturing industry was higher than in the U.S., however, almost all EU countries showed lower TFP in the service sector than that of the U.S.

Comparing the U.S. and the EU with respect to inputs of IT capital and non-IT capital, we found that the use of IT capital was more intensive in the U.S. in both the manufacturing and the service sectors. In contrast, non-IT capital was used more intensively in the EU.

As of the mid-1990s, productivity in the EU was on par with that of the U.S., but productivity subsequently rose in the U.S. as a result of the IT revolution, while in the EU there was no sign of any acceleration of productivity, owing to labor-market and other problems.

There are a number of views regarding the growth of productivity in the EU. The optimistic view is that the problem is the IT lag, so in a few years' time the EU will achieve a productivity increase similar to that in the U.S. The pessimistic view is that the productivity growth hitherto has been artificial, and it is possible that further deregulation in the labor market may possibly lead to a decline in labor productivity. The European Commission holds a qualified optimistic view that productivity growth can be expected with the advancement of deregulation in the manufacturing and labor markets, as well as the increase in intangible asset investment.

Initiatives to raise productivity in the EU include the Lisbon Agenda to promote R&D investment and the EU Services Directive aimed at opening up the market for services. It is essential to have an accurate database to monitor the progress of these processes, and we intend to take steps to that end and to make international comparisons with Japan's cooperation.

Q: The pessimistic viewpoint you mentioned was that Europe's high productivity has been induced artificially, and could be said to be the result of the regulation of labor. Although that holds true for arguments about labor productivity, it is not sufficient to explain high TFP.

A: As you have pointed out, the high level of labor productivity has been caused in part by the regulation of labor supply, but it cannot explain the high level of TFP.

Q: In the comparison of TFP in the EU and the U.S., the contributions made by TFP in the manufacturing and service sectors have been examined. The contribution of the manufacturing sector to TFP in the EU was high, whereas the contribution of the service sector to TFP in the U.S. was high. In view of this, could it not be said that the difference is largely eliminated overall? Also, perhaps this difference was affected by data measurement problems.

A: As of the mid-1990s, there was very little difference in the levels of TFP, but because the rate of TFP growth subsequently rose in the U.S. and Canada, a difference in the levels now exists. Accordingly, I believe that this difference cannot be ignored.

The third presentation, by Professor Dale W. JORGENSON, was on the subject of "Japan's Potential Growth in a World Perspective." The following were among the points made.

The Japanese economy experienced a lost decade, but the world economy was in a period of renewed growth. It is essential to identify the kind of role played by IT during this period, but when doing so the prices of IT goods should also be studied.

Semiconductor technology has advanced hitherto in accordance with Moore's Law, which holds that the numbers of transistors on microchips doubles every one to two years. However, it will likely be difficult for this law to continue to hold good in the future.

By examining trends in relative prices we can see that the relative price of personal computers has declined to one-ten thousandth of its former level over the past 40 years or so. The average rate of decline was around 15%, and was particularly rapid in the 1990s, when the annual rate was 30%.

If we examine the share of IT goods inputs, computers accounted for approximately 6% during the current decade. The figure, 6%, should not be viewed as small.

In addition, with respect to the share of IT and non-IT capital in capital inputs, there was no great difference between Japan and the U.S. during the 1980s. In the second half of the 1990s, the IT boom in both Japan and the U.S. caused the contribution of IT capital to increase, but only in Japan did the contribution of non-IT capital shrink.

In Japan, the contributions of IT capital and non-IT capital have declined even further during the current decade. If we look at this point in Asia, we can see that the contribution of IT capital is gradually increasing, although the share of non-IT capital is still large.

By breaking down the factors of GDP growth using growth accounting, we found that TFP in Japan showed stable growth of approximately 1% p.a. from the 1980s to the present decade. In the U.S., in addition to the contribution of IT capital, the size of the non-IT capital contribution was also approximately 1%; however, the contribution of non-IT capital was very small in Japan. With respect to labor, whereas labor inputs in Japan have been negative, their contribution in the U.S. during the current decade has remained positive, although the contribution has been declining.

Applying the same method to the world as a whole, the contribution of TFP has been approximately 2% during the current decade and has been increasing gradually.

Finally, we will forecast Japan's future growth. Here, we assumed that that output and capital would grow at the same rate, and the increase in labor inputs was projected to be in line with the population growth rate. The value of labor inputs was assumed to be 3% in the U.S. and -0.5% in Japan. Although there is a negative increase in Japan, we believe that this may be offset by improvement in quality.

Based upon premises such as these, the rate of Japan's economic growth is forecast to be 1.5%. TFP is forecast at approximately 2%, which does not differ substantially from the corresponding figure of 2.2% in the U.S.

Q: In the U.S. both incomes and living standards are rising, but there does not appear to be a particularly large disparity in TFP between the EU and the U.S. Given this, do you think that differences in working hours are a major contributing factor to the difference in incomes?

A: (Prof. Jorgenson) Working hours in the U.S. were certainly longer than in the EU, but there were background factors such as labor regulations. However, I do not believe that that was the only cause. For example, the quantity of IT capital per capita is larger in the U.S., and TFP is also higher in the U.S. I believe that the disparity in economic growth arose as a result of multiple factors such as these.

(Comment by Prof. Timmer) Working hours in the EU are no doubt shorter, a fact that can be attributed both to regulation and worker choice. Also, in view of the problem of the aging of society, it is essential to increase the ratio of participation in the workforce by elderly people.

Q: Is it possible to offset the reduction in labor by the use of IT?

A: As regards the reduction in labor, it will be important to increase the ratio of women's participation in the labor force. In addition, I believe that IT can offset the negative impact of the reduction in labor force to a certain degree. However, it is not necessary to dwell solely on IT; for example, an improvement in labor quality may lead to a rise in non-IT investment. Various factors are important, and a task for the future will be to study which is the most important.

Q: The contribution of TFP to Japan's economic growth appears to be relatively large, but do cyclical economic factors also play a part in this? Also, since growth in Japan stagnated during the lost decade, is there sufficient potential growth capacity?

A: In Japan there are still some industries that have not been exposed to foreign competition. I believe that in view of this there is still room to raise the TFP of industries with low productivity. As to whether it is advisable to achieve that by means of liberalization, the issue will have to be discussed with policymakers.

Q: In Korea, IT investment has been conducted vigorously since 1995 and the IT sector has been expanding, but it accounts for only around 10% of GDP, which is not as large as one would expect. How should we interpret this? Also, it is being said that the next generation of IT is fusion technology (FT). What are your thoughts on that?

A: So far there has been a great deal of discussion concerning IT users, but let me now add an examination of the IT manufacturing sector.

First, as regards the proportion of GDP accounted for by the IT manufacturing sector, this appears to be high in Korea, as it is in the U.S. and Japan, and productivity is also very high. These countries are all major producers of IT goods.

Next, as regards your second question, it may be more appropriate for a technical expert to answer it. However, what I will say is that we are shifting from a single-core generation to a multi-core generation that requires software to be rewritten completely, and we do not know whether, amid this transition, the rapid growth of the semiconductor and other fields will continue. This may well be one of the issues to be studied next.

The fourth presentation, by Professor MOTOHASHI Kazuyuki, was a report on "Japan's Potential Growth from an East Asian Perspective." The following were among the points made.

One of the major differences between the Japanese economy of the 1980s and that since the 1990s is the decline of the growth rate in the 1990s. Another is the Trade Specialization Index (Difference between exports and imports / Total trade). With respect to trade with other parts of Asia, comparing this index on a category-by-category basis in 1988 and 2001 shows that for computers and television receivers the index turns from positive to negative. That shows a shift from a trade surplus to a trade deficit, or in other words a change from exports to imports.

This research aims to shed light on such points as the international competitiveness of individual industries from the perspective of productivity, and whether the productivity of Japan's service industries is truly low, by comparing it with the productivity of the service industries in other countries.

This research has been conducted by the International Comparison of Productivity among Asian Countries (ICPA) Project, a RIETI project that makes international comparisons of the productivity of five countries and territories (Korea, China, Japan, Taiwan, and the U.S.). The data are comprised of output, intermediate inputs, labor inputs, and capital inputs in 33 industries.

Based on these data, the relative prices of individual inputs and outputs have been calculated, and productivity has then been calculated from there. To enable correct comparisons to be made when carrying this out, it is essential to control strictly for attributes and other factors.

Looking at the results on a macro basis, the product price level of Chinese manufactures in 1995, for example, were very low, at approximately 30% of those in Japan. With regard to labor, wages for matching categories were one-fiftieth of those in Japan. When comparing productivity, it was 64% of that in Japan, which showed that it was not particularly low. This could be attributed to the fact that the value of capital was not particularly low, considering the international nature of capital flow.

Looking at TFP in the other countries and territories, we found that, assigning the base value of 1 to Japan , the value for Korea was 0.77 and Taiwan 0.91, both somewhat higher than China, while the U.S. is higher than Japan, with a value of 1.07.

By dividing industry into manufacturing and non-manufacturing sectors and examining individual industries, we find that productivity in manufacturing industry is relatively high, while in the non-manufacturing sector Japan had numerous industries with low TFP compared with the other countries and territories.

From the perspective of the convergence of productivity, we found that on a macro basis the productivity gap between the U.S. and Japan is widening. As for the other three countries and territories, we also identified the fact that they are catching up with Japan and the U.S. With respect to the manufacturing industry, in 1995 the U.S. overtook Japan, and the gap has been widening thereafter. In regard to the other three countries and territories, as in the case of productivity on a macro basis we also identified that they are catching up with Japan and the U.S.

With regard to the difference in TFP between Japan and the U.S., the IT sector and non-IT sector were examined separately. From this it was found that in Japan the rate of growth in the TFP of the IT sector has been rising, but in the non-IT sector the rate of TFP growth has been declining, with the result that overall growth in TFP on a macro basis is falling.

In the U.S., meanwhile, the rate of TFP growth in both sectors is increasing, and this is contributing to the rise in productivity on a macro basis. When comparing Japan and the U.S. on a company level we found that the rate of increase in productivity when IT is used was 4% in the U.S. and 2% in Japan. Although in Japan there is extensive IT investment, it can be inferred that this may not be being utilized effectively.

When examining relations with other parts of Asia, we found that in the electronics industry, for example, they are catching up with Japan, similarly to what has been described above. Nevertheless, in China's case, for instance, in the electronics industry almost half of the output is accounted for by multinationals. These multinational firms are probably using state-of-the-art technologies, indicating that the advance of globalization will facilitate the convergence of productivity.

I have referred to the possibility that IT has not been utilized effectively in Japan, but viewed from the opposite perspective this shows that there is still opportunity for growth. In addition, when economic growth in the rest of Asia advances, that will lead to the expansion of markets, and Japan should be able to seize that as a growth opportunity.