Gordon Reikard

In March 2011, Gordon Reikard published a widely read Amstat News piece, “Stimulating Economic Growth Through Technological Advance,” exploring the connection between investment in technology and economic growth. Revisiting the topic four and half years later, Reikard discusses the prospects for economic growth through the end of the decade.

—Steve Pierson, ASA Director of Science Policy

Gordon Reikard is a statistician at U.S. Cellular. His research interests include the economics of growth and application of statistical methods in the physical sciences.

Over the last 10 years, the most commonly used measure of technological advance—total factor productivity (TFP)—has slowed considerably, especially when compared with the growth rates achieved during the previous two decades. This has led to a protracted debate about the future of economic growth. From one perspective, the United States is in an ongoing period of secular stagnation. The countervailing view is that this is a temporary hiatus and growth rates will pick up.

To get some perspective on this, it is useful to look at the long-term data. TFP, or multifactor productivity as defined by the Bureau of Labor Statistics, is defined as output less the contribution of labor and capital. In the short run, this measure can be volatile, but over longer periods—a full business cycle—the trend in TFP approximates the rate of technological advance. This insight owes to the seminal paper published by Nobel Laureate Robert Solow, “Technical Change and the Aggregate Production Function,” published in Review of Economics and Statistics in 1957. Using this framework, over the long run, technical advance accounts for roughly 38% of growth in gross domestic product (GDP).

Figure 1: Multifactor Productivity, 1949–2014. Left scale: Annual percent change. Black: Actual. Blue: 7-year moving average

Figure 1 shows the percent change in TFP in the private business sector from 1949 to 2014 and a seven-year moving average. Several trends are immediately visible. From the late 1940s through the late 1960s, TFP was consistently strong. There was an extended dip in the 1970s, but starting in the early 1980s, there was a second rise that lasted roughly until 2004. Since this time, however, TFP has clearly slowed down.

The early post-WWII period has often been described as a golden age for the American economy. Growth rates were consistently high, while recessions were brief. Unemployment declined, and inflation remained stable until the Vietnam War. One key factor was the high growth rate of technical advance. From 1949 to 1973, TFP averaged 2.16% per year.

By the late 1960s, there were indications that the boom was slowing down. The slowdown hit with a vengeance in the mid-1970s. Some of this was due to shocks. OPEC raised oil prices by 400% from 1973 to 1974, and an additional 150% from 1979 to 1980. The inflation rate increased and the economy underwent deep recessions in 1975, followed by a double recession in 1980 and 1981 to 1982. From 1970 to 1982, TFP slowed to only 0.49%. TFP normally declines during downturns, and the successive recessions were unusually deep.

But the energy shocks and business cycle volatility masked a more important underlying problem. The gains in standards of living from the major technological advances of the 20th century had already been achieved. To revive, the economy needed new technologies.

This came in the form of computers and software. By the early 1980s, personal computers were becoming widely available, and they were increasingly being equipped with powerful software. This led to a series of spillover effects into other industries, which generated both gains in efficiency and entirely new products and services. Electronic banking made capital more mobile. Computerized inventory control enabled manufacturers to reduce fluctuations in stocks. CAD-CAM made it possible for engineers to design new products online. Word processing revolutionized offices. Local area networks made it possible to equip all the workers in a company with advanced technology. The Internet led to a revolution in worldwide communications. For statisticians, regressions could be run instantaneously, and even very sophisticated models such as neural networks could be programmed with a few commands.

The economy also benefited from lower inflation and a milder business cycle: Expansions were long, while recessions were shallower. From 1983 to 2004, TFP grew by 1.4% per year. The heyday was the period from 1996 onward, when TFP grew by better than 1.7% per year.

But then, starting around 2005, gains slackened again. From 2005 to 2014, TFP grew by 0.57%, only marginally higher than in the 1970s. The 2008–2009 recession played a role, but the business cycle is only part of the story. By the middle part of the last decade, many of the gains from computers had already been achieved.

This has serious implications for long-term growth. As the baby-boom generation retires, the labor force will level off, or even decline. As the contribution of labor inputs falls to zero, all future growth will have to come from capital investment or new technology. Adding TFP to the effect of capital, the growth rate would average about 1.5% annually.

At the same time, there are embryonic trends toward a new wave of technological advances. Artificial intelligence has now advanced to the point where much greater automation will be possible using intelligent machines. Manufacturing was already heavily automated during the 1990s, but with AI, it will be possible to integrate production lines with new techniques such as 3D printing and nanotechnology.

Productivity has been consistently strong in manufacturing industries, in part because these technologies can be deployed more easily to produce goods. But even in the service sector, the capabilities of software are only beginning to be exploited. The Microsoft Office suite raised the productivity of office workers in the 1990s, but much more powerful statistical and analytical programs—ranging from SAS to the open software package R—have now been widely disseminated.

The emergence of large databases and the software to mine them (“Big Data”) has excited considerable interest. To date, most of the applications have been in marketing, with limited impact on growth. Instead, the real potential lies in the sciences, particularly fields like biochemistry.

A third area is biotechnology, which has applications in several areas. These include health care, agriculture, non-food use of crops, and other products such as biodegradable plastics and biofuels. Research in genetics has the potential to achieve major gains in medical care.

An even more intriguing possibility is that technological advances are already showing up in the national income accounts. In 2014, income ran ahead of product, and in the second quarter of 2015, GDP picked up. In the mid-1990s, income also grew faster than product, leading to a major revision that revealed growth had been higher than previously realized. One of the key elements of this revision was the explicit modeling of qualitative improvements in information technology using hedonic price increases.

While one should not read too much into a few quarters worth of data, the bulk of the evidence argues that a revival in technical advance is, in fact, quite possible. During each prior wave, it has taken some time—often well over a decade—to translate new designs into usable products. In this sense, the productivity slowdown of the last decade may be a transitory lull before the next set of technologies comes online.