Information Technology and the US Economy

Dale Jorgenson

The role of information technology and productivity growth has become the subject of hot debate in the US for a number of reasons. The US is now officially in recession, as of March 2001. And there are added uncertainties associated with the events of September 11. My view is that we should expect to see a continued impact of information technology on productivity growth and an acceleration in economic growth relative to periods when IT was less important.

This is not the only view out there. Last year, at a meeting at Jackson Hole, Wyoming, Larry Summers of Harvard University was optimistic and said productivity would grow at three percent a year. Others, such as Martin Bailey, the former chairman of President Clinton's Council of Economic Advisors, were less optimistic and predicted two percent growth. The projection of productivity growth is important for economic policymaking, such as setting budgets and taxation.

Last year I dated the US productivity revival at 1995. This was based on my view of the role of IT in the economy. If we look at the more recent evidence, business sector productivity has grown at about 1.5% per year from the third quarter of 2000 to the third quarter of 2001. During this period, output grew at only 0.1%. What has happened is that there has been a sharp jump in the unemployment rate. For the last quarter of 2001, there was also strong productivity growth.

Nevertheless, there are reasons for pessimism. In July 2001, the official national accounts were revised downward. In particular, estimates of investments in software were sizably reduced. And then there was a cyclical slowdown (the recession). But productivity growth has been more rapid during this recession than the past nine recessions. So clearly something different is happening here.

Productivity growth has been high since 1995, peaked in 2000, and has been coming down since. Nevertheless, during this recession, productivity growth has continued. The resurgence in productivity has continued into the current recession. These are the facts about labor productivity.

But let me remind you that economists use the word "productivity" in at least two different senses: labor productivity (output per hour worked) and overall productivity or total factor productivity (output per unit of all inputs including labor). So what happens next?

Long-term projections for analyzing fiscal policy in the US look at ten-year periods, so I will look at a period of ten years too. For my presentation today, I would like to look at the role of IT in productivity growth and to make a projection of growth eliminating the cyclical component.

First I would like to review the historical record. I would like to compare output and labor productivity for the periods of 1995-2000 and 1973-1995. Output will be broadly defined to include housing services, non-profits, and service flow from consumer durables. My data comes from the Department of Commerce. From 1973 to 1995, hours growth was 1.55%, while labor productivity growth was 1.44%. Meanwhile, from 1995 to 2000, those numbers were 2.24% and 2.36%. Hours growth and labor productivity both grew. What are the sources of labor productivity growth?

One source is capital deepening, which is the result of investment; investment provides more and better capital to labor. A second source, which is extremely important, is increases in the quality of labor. The idea is that underlying trends for increased educational attainment and age and gender composition of the work force all impact labor quality. Labor quality is defined as labor input per hour worked. Finally, there is output per unit of input (technology and everything else) or total factor productivity (TFP).

What happened after 1995? First of all, capital deepening increased substantially. One reason is because of the increased role of IT. Between 1995 and 2000, there was a huge increase (0.76%) in capital deepening associated with IT. What does that represent? It represents the higher levels of investment in IT, specifically computers, telecommunications, and software.

Meanwhile, during the late 1990s, labor quality actually slowed. Why? There was a huge increase in the growth of hours worked. With this increase, there was both an increase in people employed with high skills and people employed with low skills. We had welfare reform that resulted in an increase of people with very low skills in the labor force. The consequence is that labor quality was reduced.

But TFP growth accelerated. IT TFP contribution almost doubled. There was a change in the product cycle of semiconductors. Overall, the change in labor productivity growth after 1995 was about 0.9%, a sizable number. Of that, slightly less than half was due to IT capital deepening. Eight percent was due to other types of capital deepening. Labor quality fell. TFP in IT production contributed to about one quarter, as did TFP in other sectors. That is the historical picture. Despite the slowdown, the productivity revival remains strong.

How about looking forward? How much is transitory? The two key assumptions we will make to remove transitory effects are the following: first, in the long run, output and reproducible capital (producible capital, which excludes land) have to grow at the same rate; second, hours growth matches labor force growth.

There is a lot of uncertainty. So I will give you a pessimistic, optimistic, and base-case. As is customary, I will assume that hours and labor quality are the same, from demographic projections. I will also assume that capital, labor, and IT output shares are equal to historical averages.

We then calibrate alternative assumptions, which vary across scenarios according to TFP growth in IT production, TFP growth elsewhere in the economy, and capital quality growth. The base-case scenario will be based on the "International Technology Roadmap for Semiconductors," which argues an eventual reversion to the three-year product cycle. The optimistic scenario is based on Intel's view that we will see a continuation of the two-year product cycle. The pessimistic scenario says we will revert to the 1973-1995 experience.

The TFP contribution from IT is significant for all three scenarios: for the pessimistic view it is 0.37%, for the base-case it is 0.44%, and for the optimistic view it is 0.52%. Capital quality growth was exceptional, at 2.45% average growth for 1995 to 2000, but it is unlikely to continue at this rate. The pessimists say capital quality growth will go to 0.84%; the base-case would be at 1.75%; and the optimists say it will go to 2.45%.

Reasonable demographic assumptions see hours growth declining to 1.1% per year or half of what it was during the late 1990s (2.24%). We are not going to have the kind of increase in labor participation increases we have had because that is simply unsustainable. This is the core of the pessimistic view, that IT was a bubble and cannot be repeated. Hours growth can not continue as it did during the 1990s. So we put it at 1.1% for each scenario. The other assumptions determine the range of productivity growth projections. Adding hours growth and productivity together, the pessimists project productivity will be at 2.43% and that the late 1990s were just a bubble (this is a view held only by the extreme pessimists); the base-case say 3.34%; the optimists (like Larry Summers) say 4.08%.

Keep two things in mind: First, there will be a slowdown in US economic growth. Second, there is an enormous range of uncertainty, which has to do with labor productivity in different scenarios. The key to the uncertainty is the future of technology. It is what drives IT-related TFP and capital quality growth.

The US productivity remains intact. IT use and production play important roles in TFP and capital deepening. Post-1995 productivity growth can continue. 2.25% per year seems reasonable; this is the consensus view (views of publications such as The Wall Street Journal, which is pessimistic, and BusinessWeek, which is optimistic, are converging). It is important to avoid giving the impression that these estimates are precise. There is tremendous uncertainty. The virtue of IT is that it grows rapidly. In ten years, there were three separate product cycles, so it is difficult to be precise. Planning for the future must reckon with this uncertainty related to IT.

Rizaburo Nezu

At the OECD, I spent a great deal of time studying this subject. You say that US productivity will continue to grow, despite the economic downturn. This has sparked some debate inside the OECD. I do have a couple of questions. If in the years to come, productivity will increase, shouldn't the US then change its macroeconomic policy? As for the semiconductor product cycle, how quickly will consumers want to adopt this new technology? I am worried about the demand side of the equation. The price of semiconductors is only a small chunk of the total price of IT.

Q: The hardware proportion of IT expenditure is shrinking. To what extent would you projection change from this trend?

Dale Jorgenson: At first there were two views at the OECD: one represented by Mr. Nezu's science, technology, and industry division and one by the economics division. By the time Mr. Nezu left the OECD, the economists had appropriated the views of Mr. Nezu's division. In fact the economists stole the ideas from STI. So it was a total triumph for Mr. Nezu.

The role of these projections on economic policy is that they are a basic input for deciding fiscal policy (taxation and expenditure levels). The Congressional Budget Office and the Office of Management and Budget are pretty much in line with this thinking.

The most prominent of the pessimists, Paul Krugman, wrote a piece called "Setting the Speed Limit," in the Harvard Business Review. He took a pessimistic view and said there was no significant role for IT in productivity. He said that productivity growth would stay at 2.5% per year. And this was the consensus view. Alan Blinder wrote a similar piece.

My, relatively moderate, view is now the consensus. It reflects the triumph of Mr. Nezu and his colleagues at the OECD and his counterparts in the US.

Most publicity has been about monetary policy. The Federal Reserve let things get out of hand in 2000. The impact of tightening monetary policy is unpredictable. There is a surge in productivity, so some feel we have to worry less about inflation. That is true; but it is also true that the variability in productivity growth makes the job of monetary policymakers much more difficult. That lesson has yet to sink in.

There are mysteries about IT. The IT productivity TFP is not totally explained by the product cycle. It is not only about IT production, but also it is about IT absorption and adoption (which is reflected in capital deepening and capital quality). Capital deepening is associated with the demand for different skills. The relative demand for skills changes. The labor force is slow to change. We get a rapid change capital allocation, but slower changes in the labor force.