RIETI Special Seminar

Special Lecture by Professor Jorgenson, Harvard University: World KLEMS Initiative (Summary)

Speaker Dale W. JORGENSON
Samuel W. Morris University Professor, Harvard University
Commentator & Moderator FUKAO Kyoji
Program Director and Faculty Fellow, RIETI / Professor, Institute of Economic Research, Hitotsubashi University


Professor Dale JORGENSON

The subject of my talk today is going to be a new World KLEMS Consortium that was established at Harvard University in August 2010.

KLEMS refers to the missing half of the national accounts. GDP is used as a measure of output for the economy, but when we start to analyze the sources of the growth of output, we find that national accounts are no longer sufficient. To find the sources of economic growth there was nowhere to go. But thanks to efforts around the world, people are beginning to fill this gap. There are now seven countries that include the input side of the economy as part of the national accounts.

The input side of the economy is comprised of capital (K), labor (L), energy (E), materials (M), and services (S), hence KLEMS. For seven countries it is possible to obtain official figures on the input side of the economy, not only in current prices, but also in constant prices. We now have the ability to collect such figures for all countries, and our goal with the World KLEMS Initiative is to incorporate KLEMS-type data into the official statistical systems in every country in the world. Beginning 10 years ago I advocated that this should be adopted in Japan, as it has the best statistics for the input side of the economy. This is gradually moving toward realization because of the effort to reform the highly decentralized Japanese statistical system and create a unified system that would incorporate the input side as well as the output side.

The book Productivity: Information Technology and the American Growth Resurgence, is a version of the U.S. national accounts which has the characteristics of a KLEMS database covering the period from 1960 to 2004. The focus, as you can see from the title, is on the role of Information Technology. Following the publication of this book, the European Union (EU) formed a project beginning in 2005 to extend this type of analysis to all of the EU members. In the end, the project involved 25 of the 27 EU members. At the same time, the Japan Industrial Productivity (JIP) database was developed in Japan and became an important component of the world KLEMS community.

Altogether a total of about 30 countries have begun to develop KLEMS-type databases. Of course, making such a database and including it into the statistical system are two different things. The reason is that it is necessary to have standards. In the international community of statisticians, the standards are the System of National Accounts (SNA), which is a United Nations organization that involves all UN members. The SNA has recently issued a revised set of rules, the 2008 System of Natgional Accounts. For the first time an explicit allowance for incorporating data on the input side of the economy including KLEMS and productivity as part of the official UN system of national accounts.

The key decision was made by the UN statistical commission in March 2007, shortly after our book was published. Then, when the new SNA was published in 2009, it contained a chapter explaining how this could be done in a way that is consistent with the System of National Accounts. I developed a system for implementing this within the U.S. called the New Architecture Project, and this can be found on the website of the Bureau of Economic Analysis. Today I want to describe a new version of this project, which will provide inputs at the industry level as is appropriate for a KLEMS type database. It will be integrated with the U.S. national accounts within the next year or so.

I made a more explicit proposal at the World KLEMS Consortium formed in August 2010 to a total of 40 countries. In the U.S. system we are going to be covering a longer period through 2007, before the crisis, looking at detailed industry classification emphasizing IT, including the boom that took place from 1995 to 2000, the dot-com crash, and the jobless recovery, before the financial and economic crisis. Another project will be to integrate this with macroeconomics, which will involve replacing the aggregate production function by a more sophisticated approach to aggregation over indusries.

The first KLEMS initiative was the EU KLEMS project, which focused on 25 of the 27 members of the EU. The initial release of the data was in 2007, before the UN decision. This began a shift from academia toward an audience of government and business analysts who are using data for understanding policy issues. Later the U.S., Canada, Japan and Korea joined.

Japan is the country, in addition to Canada, with the longest established input-output tables. Part of the work that Professor Fukao and his colleagues have done is to turn it into a continuous time series and make it available. We did something similar in the U.S., as part of our book, but it was only this year that the U.S. national accounting agency created an official time series going back to 1947. That is going to be the basis for the new official KLEMS database for the U.S. In the meantime, the UN endorsed the idea, and the OECD began to develop standards for producing these accounts around the world, which are now part of the System of National Accounts and can now form the basis for official statistics around the world.

For an example, I am going to focus, as I did in my book, on information technology (IT). I am going to focus on three broad groups: IT producing industries, IT consuming industries, and the others. I would like to draw your attention to the changing character of IT. Discussions of IT previously focused on manufacturing, but now—as a result of more detailed industry classification—it is possible to bring in an explicit role for services. An example would be cloud computing. This is motivated by the fact that most desktop or notebook computers are used about 20% of the time. The economic rationale is compelling: use that other 80%, and thereby reduce the cost of computing by a factor of five. The other industry that has suddenly appeared is computer system design and related services.

An IT-using industry is defined as having 15% of capital in the form of IT software or hardware. The IT-using industries are a fairly heterogeneous group, but 80% of the use of IT services is in services and trade. The role of manufacturing is fairly modest. The non-IT sectors include manufacturing, and a few services. These are the industries that IT is going to conquer over the next few decades.

IT is important because of the rapid pace of technical change as indicated by the rapid decline in prices of IT. Beginning around 1970, the price of computers declined from approximately 400 to 1, at a rate of about 15% per year until 1995, then by about 30% per year until 2005, and then by about 15% per year again after 2005. It is this relentless decline in prices that drives the utilization of IT.

It should be noted that the share of IT in GDP is very modest. However, when we look at the relative importance of various industries in economic growth, IT is much more important. If we combine the growth rates of IT with the share of services (weighted by the value-added share in GDP), we see that the role of IT rises to about 10% of the growth of the U.S. economy, which was growing at about 3% per year.

The contribution of the IT-producing sectors, essentially 3% of the economy at the end of the period, was about 10% of economic growth. How can 3% of GDP account for 10% of the growth? This can only happen by having very rapid growth rates, which are the counterpart of those very rapid declines in prices. During the IT investment boom of 1995-2000, the role of IT rose to around 0.8% per year out of total growth of about 4% per year, which equates to about 20% of economic growth, and then went back to the more normal growth pattern of around 0.3%, which has prevailed since the dot-com crash of 2000. However, as we emerged from the crash we discovered that services are relatively more important as compared to hardware within IT.

The sources of economic growth are total factor productivity (TFP), or output per unit of input, together with capital input and labor input. TFP is an important source of growth because of the fact that it is TFP that indicates the progress of technology, which is an important source of economic growth. Looking at the contributions of individual industries to productivity, the IT producing sectors, which account for about 3% of the economy and 10% of economic growth, accounted for more than half of productivity growth from 1960 to 2007. This means that of the changes in technology in the whole economy, more than half are associated with 3% of GDP. This is why it is essential to look at economic growth at the industry level. You cannot understand the process of modern economic growth without looking at the role of individual industries.

During the boom from 1995 to 2000, the role of IT-producing sectors more than doubled, but that was a bubble, and the role of IT-using industries slightly contracted. However, as we came out of the boom, productivity continued to advance at the same rate, but the role of the IT-producing industries declined, and the role of IT-using industries came to the fore. So what emerged from the dot-com crash was a remarkable change in the character of changes in technology in the U.S. economy, from the IT producers to the IT users.

The IT-using sectors are trade and services and the character of technical change shifted from changing the hardware and software to changing the business organizations and systems within which IT was used. That accompanied the change in the IT- producing industry from hardware to services. Designing new businesses and new organizations became more important than designing hardware. The role of the non-IT sectors has slightly improved, but very modestly, and during the earlier period from 1960 to 1995, it was slightly negative. The conclusion is that changes in technology now take place in a totally different way than before the year 2000. This would never have become apparent without the kind of detailed industry data that is part of the World KLEMS Project.

In conclusion, the character of economic growth has relied very heavily upon IT, and that becomes even more apparent when you look at the sources of growth in output per unit of input which is the indicator of technical change around the world. IT production is very volatile, so we have to be very careful to distinguish between permanent and transitory changes. But the big change in trends since 2000 is the trend away from IT production to IT utilization as the locus for changes in technology. It is trade in services which after all make up 60% to 70% of the average industrial economy that have come to the fore as far as productivity growth is concerned.

In order to understand the role of these IT-using sectors, trade and services, it is going to be necessary to change our model of productivity change. It is now less a matter of scientists and more like engineers who are trying to recast the nature of business organizations in order to make more effective use of the rapidly developing IT that is taking place. The character of IT itself is changing. It is moving from hardware to services. And those services in many cases create intangible assets, like new business organizations.

The implications for economics are momentous. The character of IT is transforming the world economy that is conducive to softer forms of the development of technical change. As analysts and economists trying to use the results to analyze policy issues, we need the kind of detailed information of patterns of changes in technology by industrial sector that are at the heart of the KLEMS international consortium.

We also need to generate KLEMS-type data for the emerging countries, which are going to become the central focus of economic growth during the coming decades. I would like you to think about how this is going to affect your particular activities. In particular, if you are responsible for the Japanese national accounts, you should be asking yourself how this can be turned into a regular data production process that can be used in daily analytical activities. There is no country better equipped to take advantage of this new development than Japan.


Professor FUKAO Kyoji

I would like to briefly introduce what we do at RIETI in response to Professor Jorgenson's research. Tomorrow, 27 July, we are going to have the First Asia KLEMS Conference at the Asian Development Bank Institute (ADBI) in Tokyo to help launch Asia KLEMS. The chair is Professor Hak K. PYO of Seoul National University, Korea. Researchers from India, China, and Singapore participate in Asia KLEMS, allowing Asia KLEMS to be analyzed comparatively as a valuable part of World KLEMS.

The conference tomorrow is for a limited number of participants, but we have distributed here a presentation from Japan for tomorrow's conference, which describes how KLEMS type data and analysis can be used in a quantitative way to analyze the extent of the economic impact of the damage occurring in Japan following the Great East Japan Earthquake.

At RIETI, as Professor Jorgenson mentioned, we have the Japan Industrial Productivity (JIP) database, which analyzes the sources of growth in Japan, and also we have a China Industrial Productivity (CIP) database which is a similar database on China, produced in Japan. RIETI intends to make the JIP Database 2011 and the CIP Database Round 1.0 covering at least labor productivity available by autumn 2011.

The JIP database results show that IT investment in Japan is stagnant by international comparison. There are only a handful of industry sectors contributing to the rise of TFP.

I would now like to ask Dr. Jorgenson three questions:

First, Dr. Jorgenson's book was published by MIT Press in 2005, and Professor Jorgenson's design of a KLEMS framework was spread worldwide, first starting from the EU KLEMS Project. We all know the EU KLEMS Project and it is widely used. Now it has become the World KLEMS Initiative. In comparison to the EU KLEMS, the scope in terms of countries included in the World KLEMS is broader, but has anything additional been added compared to the EU KLEMS Project?

Second, when we think about the IT revolution, Professor Jorgenson mentioned that the characteristics shifted from hardware to services, the IT-using sector has become more important, and organizational change in the IT-using sector is becoming more important. After the IT revolution, productivity increased in the U.S., but will there be a similar pace of TFP rise in the future, and if there is an obstacle, what would the obstacle be, and what is the impact of the global economic crisis?

Third, since the nature of IT is changing, do we need to have a new kind of government statistics system? We have all worked under the KLEMS framework, but if the nature of IT has changed, would not the statistics that we collect also change? What should the United Nations do, and what is the future direction?

Professor Jorgenson

Professor Fukao's comments illustrate very well the problems that lie ahead for the use of this very important concept in the Japanese statistical system.

Let me focus on the last question first. What is the implication of this on the way we construct economic statistics, and what is going to change? The answer is that what you saw before was the story about the output of the economy, and specifically the growth of GDP. What you will find in the new system is exactly the same information as before, but there is going to be a whole new section for the input side of the economy, the missing half of the accounts. There is no barrier to doing this except to require that the system be augmented to provide this additional information on a regular basis along with the standard measures of output.

I laid out a detailed plan of how to provide information on inputs in January 2009, and what I said was that a complete system needs to include a production account that includes both the input side, that includes KLEMS and productivity, and on the other side the output. And that has to be in current and constant prices. In addition to that, you need to have the income and expenditure account that gives the income data in current and constant prices and wealth data in current and constant prices. What time is better than the present for doing this? We have an example here for the benefits of this new way of looking at economic growth in the most important economic event of the last decade in Japan, the Great East Japan Earthquake and its aftermath.

With regard to the role of the IT revolution, IT shifted from hardware to services in the production of IT, but the central message that I would like you to take away is that after the dot-com crash in 2000, productivity continued at the same high rates despite a 40% drop in the relative proportion of IT in the economy. The only way that could happen is through a shift in the character of technical change which I emphasized earlier, from hardware design and implementation to the creation of a new mode for changing technology through the reorganization of business systems. This is going to be the central issue in Japanese economic growth going forward.

The forces of globalization and the demographic changes taking place in Japan are going to force changes in every business organization in the country. This is going to require a fundamental system of reforms of the Japanese economy focusing on trade and services that will facilitate the completion of the IT revolution in Japan. Although IT has always been one of the great success stories on the production side of the Japanese economy, Japan's effectiveness in utilizing IT has lagged behind the rest of the world. Therefore this is a revolution that is going to have to take place if Japan is going to make any progress.

On what's new in the World KLEMS Project, the most important thing is the change in scope, from only the advanced industrial countries to include the emerging economies that will dominate the next decades of world economic growth, such as China, India, Brazil, and Russia. The World KLEMS Project will include the Asia KLEMS project, the Latin America KLEMS Project, altogether including 40 countries around the world. The second thing that will be new is the possibility of integrating these data with a world input-output database, involving trade relations among all the economies in the world, because that is the key to understanding the role of globalization. I think you will find it to be one of the most stimulating new developments in empirical economics in our discipline's history.

Question and Answer

Q: Mr. SAKAMOTO Masahiro , Japan Forum on International Relations
First, do you think that the trend of shifting from the IT-producing sector to the IT-using sector will continue in the future? And second, what are the implications of this shift for the advanced economies? I read that there is a huge shift of employment and income from advanced economies to emerging economies, leaving a huge unemployment problem in the United States, Europe and Japan.

Professor Jorgenson
That is a very challenging question. My view is that this represents a net addition to world economic growth, and the advanced economies will continue to grow at rates comparable to long-term historical averages of about 0.5% per year for Japan, or about 2.2% per year in the United States. Japan in the 1950s invented a new mode of economic growth. This created double-digit growth in Japan for an extended period until roughly the early 1970s. This pattern was learned by people in Taiwan, Korea, Singapore and Hong Kong, the four Asian tigers, and was successfully applied outside of Japan. The people who learned this most thoroughly was the Chinese, who learned from Taiwan, Singapore and Hong Kong, what these countries had learned from Japan.

The period of very rapid growth will continue for about a decade in China and the next two decades for India, and this will create opportunities for the advanced economies that will eventually make the unemployment adjustment period after the financial and economic crisis a short-term transitory phenomenon. However, the relative importance of the so-called advanced economies is going to shrink, so there is going to be a total transformation of the world economy within the next two decades, built on the Japanese pattern as it spreads around the world, specifically within Asia, with the most important being China and India, and we are very grateful to Professor Fukao for taking leadership in the development of work for China KLEMS here in Japan.

Q: Dr. Martin SCHULZ, Fujitsu Research Institute
Linking Asia growth with your research, one important element for growth is extremely high investment rates in Asia. When you have very high levels of investment, you also have very high levels of depreciation of capital. How important do you think it is to take care of the depreciation of capital when linking input and output in extending the GDP statistics?

Professor Jorgenson
The North Atlantic banking crisis was triggered by the subprime and banks that had globalized their funding sources and mismanaged them. It was a financial crisis for the U.S. and Europe, but it had nothing to do with finance for the rest of the world. Rather, the uncertainty that led to the postponement especially of durable goods consumption was what brought about the world's largest trade collapse ever. However, the trade collapse is now over, and we are seeing a two-speed recovery. I do not see this as having a primary effect on the world trading system.

That is a bit of a technical question, but depreciation is a part of the measurement of capital services, so it is in the production account, but to take full account of the importance of depreciation it is necessary to shift the focus to income and expenditure. Expenditure is the sum of consumption plus saving. Saving is defined net of depreciation. When you look at the uses of economic growth in terms of consumption, current and future, depreciation is central. And the importance of depreciation has been much higher with higher levels of investment. But it is not so much the level of investment as a proportion of GDP, it is the fact that it increases.

In Japan the initial proportion of savings to GDP was less than 20% around 1950, but by the end of the boom decade of the 1970s that proportion had risen to historic highs of 40%. That is not sustainable. The saving rate level then fell off to a steady state level of about 25%, which brought the growth rate of the Japanese economy down to a rate consistent with its status as an advanced economy.

The same is happening today in China. China has surpassed the Japanese proportion of savings to GDP. Some estimate that it is as high as 50% of GDP. That too is unsustainable and will gradually decline in the coming decade and then decline more rapidly, and China will become a normal middle income country. In India, we can see the Japanese pattern beginning to spread with the rise of the savings ratio, and this is going to continue and ultimately peak and then decline. So depreciation and savings is a critical issue in understanding the processes. It is not enough to have a production account; you have to look beyond production to how the production is going to be used.

Q: Mr. SUZUKI Toshiyuki, UFJ Bank
After you got KLEMS data did you find any policy implication to promote economic growth?

Professor Jorgenson
Yes, the policy implication to promote economic growth is to try to understand the process by which changes in technology have shifted out of the laboratory and into the workplace and into the reorganization of business processes. That turns out to be a process that I believe has its roots in the nature of competition. For example, Europe has succeeded in unifying the market for goods, but has failed in unifying the market for services. The result is that productivity growth rates in Europe are falling at precisely the time they are rising in the United States and that has important lessons from Japan.

Q: Professor Robert F. OWEN, University of Nantes
IT, through organizational changes, which was already mentioned, and notably through its impact on labor and vintage effects and quality effects on human capital is producing a very complex identification problem with regard to the sources of growth. You mentioned in your non-IT sectors rail transportation, but the use of IT is saving the time of other agents and so there are important spillover effects in addition, and one needs to think more about the role of time. My conjecture is the differences in human capital across countries is probably one of the most important factors explaining productivity growth and is absolutely essential for policy.

Professor Jorgenson
I couldn't agree more. A very important part of this story is that of changes in labor quality associated with increasing educational attainment. However, more importantly the improvement in HR management post education, which is captured in a very abstract term of business reorganization, is a story about the management of human resources. Having people who can make chips and computers is only the beginning; without millions of people out there who could make effective use of the technology this would have been worthless. The rule of thumb would be that of the total amount of investment in IT, only 10% of it is in hardware and software, the rest of it is in reorganization and businesses and more effective use of human resources. How important is that as a source of economic growth? The part that is due to an improvement in the quality of tertiary education is relatively modest. The major source of economic growth is investment in IT and non-IT.

Q: Mr. YOSHIDA Yasuhiko , RIETI
The service sector and IT utilization are very important. My question is that when comparing IT-using and non-IT-using industries, as classification is done according to the proportion of IT utilization, heavily equipped industries would have a high relevance here. Also, those industries open to Chinese competition face downward price competition and therefore productivity gains are difficult to obtain. So factors other than the IT field may be at work.

Secondly, Japan has a fairly strong manufacturing industry compared with other economies, and some say we should make efforts in manufacturing. Is it more appropriate for Japan to emphasize the manufacturing sector? There are various viewpoints in considering possible policies, so can you elaborate more on policy implications from all of those viewpoints.

Professor Jorgenson
Let me say first that the answer to that is at the core of understanding the Japanese model. The most important thing that Japan discovered was globalization, and this was before the word was ever used in common discourse. At the time, globalization presented opportunities to the Japanese economy very much like those now being presented to the Chinese, and hopefully, before too long, to the Indians. And by globalization of the economy, it is possible to take advantage of participation in the world division of labor, and this is essentially the lesson of Japanese manufacturing.

Japan will continue to be among the world's leaders in manufacturing. Whatever can be done to enhance Japan's performance will strengthen its role in the international economy, so I am not anti-manufacturing. But, the real opportunities in terms of growth prospects for the Japanese economy are to bring up the lagging sectors, which are personal services and trade. So how is it possible to bring about such changes? Those changes will be forced by globalization and demography. What sort of policy should be adopted to facilitate the change? The answer is that it is necessary to strengthen the competition for these industries that are largely not integrated with the global economy.

It will be necessary for Japan to adopt a whole new range of policies that will focus on productivity changes and exploitation of opportunities to productivity growth on the neglected part of the Japanese economy. This part of the economy is disappearing as we speak, but the role of IT in these neglected sectors is going to produce a gain from the potential growth of the Japanese economy of less than half a percent per year to stretch to a target of 1% per year. It is necessary to focus on those neglected areas, in order to produce what a competitive market structure in the world economy produced in Japanese manufacturing. The JIP database can be used in analyzing these issues, but you have a lot of work ahead of you and it is not going to be easy.