The U.S. economy’s current state reflects a long and tedious path to recovery. Growth has restarted, and unemployment has dropped significantly. Corporate profits have recovered, although new investments have been modest. The extended upswing came after one of the most daunting and disastrous recessions of the twenty-first century. While measures to reverse the decline cut unemployment and resulted in modest growth, investment has been slow to recover, and consumer spending is only beginning to show signs of life.
The first part of this review suggests that the recovery is likely to continue, and that some sectors, such as housing, could provide a short-term boost for the rebound. The second part of this paper argues that an important, but unexpected, source of growth is likely to come from innovation related to Internet technologies, and in the management of new infrastructure based on innovations that have emerged during the past few years. Many analysts are convinced that the consequences of artificial intelligence (AI), and automation discussed in the The Second Machine Age will impact job growth.
This part of the paper suggests that progress in technologies linked to the Internet and infrastructure produce very different outcomes than advances in automation and AI. Although advances in computing and software are likely to displace jobs that played a role in traditional information technology during the past two decades, there is a second, parallel path in technology and innovation that contains a story The Second Machine Age does not tell.
This second path is based on advances in the Internet that have reshaped computing, storage, and networking. These innovations are built on cloud computing, and the virtualization of machines and networks, i.e., the use of software to manage hardware in new and more efficient ways. The evidence in this part of the paper suggests that virtualization of devices (computers and switches) already has begun to transform the way businesses use computing and networking infrastructure. Netflix, Twitter, Walmart, Bank of America, Facebook, Gap, and other firms already are speeding and refining the creation of new services for customers.
While these new developments in the economy are not very visible in economic data, a number of observers, including Enrico Moretti from Berkeley and Cesar Hidalgo from Massachusetts Institute of Technology, have argued as I do here that we are on the way to creating a knowledge economy. This part of the essay argues that Internet and infrastructure innovations will create a large part of the new knowledge economy. Enterprises are rapidly implementing these innovations. I expect they could appear in productivity and growth figures quite soon, much to the surprise of many economists and forecasters.
A View of the U.S. Economy and Dynamism
There are divergent views about where the U.S. economy is going. The view I hold is that the economy is recovering slowly, but gradually, from a rather disastrous financial crisis spurred by a major collapse in the housing market. That implies that the economy is still very fragile and vulnerable to unforeseen crises, such as a Chinese economic slowdown, a Greek default, or a credit crisis due to something like a failure of the United States to pay its debt due to Congressional inaction. The crisis of 2007–2008 had many similarities to the 1929 crash, with the collapse of “unsinkable” institutions such as Lehman Brothers and the U.S. housing agencies. Its severity—at 18 months, longer than any since World War II—made the recovery far more difficult.
A more conservative view of the U.S. economy is that the country’s debt must be reduced, and the budget must be balanced to promote growth. This view has dominated Europe’s response to the recession. U.S. efforts, which stand in stark contrast to Europe’s, represent a more traditional Keynesian approach—namely, a sizable stimulus during the early stages of the downturn, as well as Federal Reserve policies to promote lending and consumption
Another view,1 espoused by Professor Robert Gordon, is that the economy has headed into such strong “headwinds” that the potential for future growth is very limited. In this view, a lack of innovation will curtail both productivity growth and consumption. In this scenario, U.S. gross domestic product (GDP) will fall to about 1 percent during the coming decades, and productivity growth will be almost nonexistent for perhaps the next 100 years—similar to what it was during the Victorian Age.
Let’s look at some of the data. First, what has been happening with GDP? Instead of accelerating to a 2.5 percent to 3 percent growth rate as in other, shorter recoveries, growth has been sluggish. One way to characterize the recovery is that it was “jobless” until the past year, when new jobs grew at nearly 300,000 a month. But, in December 2009, three years after the recession started, employment was still 4.5 percent below what it was at its peak; no other post-1946 downturn was so severe (see Table 1). This occurred in spite of a substantial economic stimulus at the beginning of the Obama Administration (see the chart below for government spending—in green—as well as GDP growth during the recovery period.)
Structural Changes in the Economy
So what is happening to the economy? Among economists, there is general agreement that there was a period in the late 1990s and early 2000s when productivity rose to nearly twice the 1.5 percent year-to-year rate of the 1970s and 1980s. There was hope that the economy might begin to grow more than modestly.
I plan to address this growth question later in looking at innovation. But first, I’d like to explore the economy’s health and recovery from one of the worst downturns in recent history.
The State of the U.S. Economy during 2015
Let’s look at some of the main indicators for the economy to see how they are doing. As we saw from the chart above, the recovery has been quite slow compared to other downturns, possibly because the recession was so closely linked to the financial services sector and was preceded by a large decline in housing values. The former squeezed lending to businesses, while the latter dramatically cut consumer spending.2
To facilitate the recovery, the Federal Reserve adopted a policy of “quantitative easing,”3 in which it actively purchased government bonds. While this had the aim of promoting lending, it actually created far more stability in the bond markets, while interest rates also dropped to their lowest points in years. The creation of this stability in the bond markets did much to speed the recovery. Consumer spending accounts for about 70 percent of GDP growth, so its recovery is entwined with overall economic health. As is clear from the figure below, consumer spending has rebounded nicely. This is one factor helping GDP growth recover.
The identity for GDP is that GDP = C + I + G + Net Exports, where C is consumption spending, I is investment, and G is government spending. As can be seen from the first graph above, government spending rose during the recession as part of the stimulus program, a Keynesian effort to promote growth. According to Keynes, government spending is a key way to reverse the economic effects of a downturn. In contrast to Keynes, many conservatives argue that tightening government spending would promote growth by reducing taxes and lowering debt. Governments in the UK, Germany, and Europe adopted this policy soon after the crisis, but it now has been reversed.
Although we identified overall patterns of investment in the first figure above, both investment and consumer spending had a difficult time during first quarter 2015. The decline—as well as the decline in overall GDP of 0.7 percent during the first quarter—was due to the sharp downturn in oil prices and a subsequent drop in investment in structures that include oil drilling and wells. This latter component fell 21 percent during first quarter 2015.
On the trade side, the United States has been doing well after the recession. But the recent increase in the value of the dollar has made imports cheaper and U.S. exports more expensive overseas. One consequence of this was that GDP growth was 1.9 percent less during first quarter 2015 due to the increase in imports. The euro was one key currency that fell in value relative to the dollar, dropping from $1.32 at this time last year to $1.09 May–July 2015. This made U.S. imports more expensive in Europe and U.S. imports from Europe less expensive. This is one reason why the U.S. trade deficit has increased and growth has slowed.
Somewhat compensating for these negative impacts on the economy, the housing market has been on fire in recent months. Household formation (people purchasing their first houses) has only recently risen above the trend line of the past fifteen years. In part, this is due to the long period of low household formation following the 2007 crisis, but also is due to continuing low mortgage interest rates.
Growth, and the Productivity and Employment Debate in the United States
Several writers have argued that U.S. growth will remain rather modest—at 2 or 2.5 percent per year due to a lack of improvement in productivity. One common view4 is that the U.S. economy has entered a stage in which investment has slowed and, without a surge in new investment—capital and “software” now defined as capital—productivity growth will remain low, and overall economic growth will slow.
The following chart shows that, with a few exceptions during the early 2000s, productivity growth for the United States has been roughly 1.5 percent for the past fifty years. One factor that has slowed growth has been the fall of capital investment, which is shown in the next chart.
Investment slowed after the recent crisis. It fell below the long-term, post-World War II trend line of 22 percent of GDP. The slowdown is more apparent in the second chart, which shows net additions to the U.S. capital stock as a percentage of GDP. After the recent financial crisis, this share has dropped below the trend line of 3.3 percent.
There is another debate that is influencing thinking about productivity growth. Using evidence based on a study of the likely economic impact of automation and “thinking machines,” Erik Brynjolfson and Andrew McAfee have argued in The Second Machine Age that “computers and other digital advances are doing for mental power – what the steam engine and its descendants did for physical power.” Reinforcing this analysis, quantitative modeling described in The Future of Employment, by Carl Frey and Michael Osborne at Oxford University, uses a ranking of occupational characteristics to determine how vulnerable specific jobs are to automation during the next twenty years. Frey and Osborne’s main conclusion is that only the most creative and labor-intensive jobs at the extremes of the job scale are likely to survive the trend to automation. The U-shaped chart below was developed from the Frey and Osborne data.
In their analysis of U.S. employment, Frey and Osborne find that “47 percent of total U.S. employment is in the high-risk category, meaning that associated occupations are potentially ripe for automation during some unspecified number of years, perhaps a decade or two.” They postulate two “waves” of automation, with the first affecting “most workers in transportation and logistics occupations, together with the bulk of office and administrative support workers, and labour in production occupations. Those jobs likely are to be substituted by computer capital” [p. 38]. In the second wave, after a technological plateau, jobs in “services, sales and construction occupations” [p. 38] are automated.
Furthermore, Frey and Osborne argue that, during the coming decades, even non-routine jobs will be automated. The main constraint on automation will be if the positions require “fine arts,” “originality,” “negotiation,” “persuasion,” “social perceptiveness,” and “assisting and caring for others.” Jobs with these variables all exhibit relatively high values in the low-risk category [p. 40], which means they are likely to survive.
“By contrast, we note that the ‘manual dexterity,’ ‘finger dexterity,’ and ‘cramped work space’ variables take relatively low values [in the low-risk category]. Hence, in short, generalist occupations requiring knowledge of human heuristics, and specialist occupations involving the development of novel ideas and artifacts, are the least susceptible to computerization” [p.42].
A similar concern about productivity growth has led Robert Gordon5 of Northwestern to argue that “innovation does not have the same potential to create growth in the future as in the past.” [p. 11]. Gordon’s key thesis about innovation and the Internet is:
“The computer and Internet revolution began around 1960, and reached its climax in the dot-com era of the late 1990s. But its main impact on productivity has withered away in the past eight years. Many of the inventions that replaced tedious and repetitive clerical labour with computers happened a long time ago, in the 1970s and 1980s. Invention since 2000 has centered on entertainment and communication devices that are smaller, smarter, and more capable, but do not fundamentally change labor productivity or the standard of living in the way that electric light, motor cars, or indoor plumbing changed it” [emphasis added].
Gordon expects that, when successful innovation occurs, it will be discrete and very similar to successful innovation during the nineteenth-century industrial revolution. This was true of electricity and machine power. He argues that “it is useful to think of the innovative process as a series of discrete inventions followed by incremental improvements, which ultimately tap the full potential of the initial invention. For the first two industrial revolutions, the incremental follow-up process lasted at least 100 years” [p. 2]. Lacking evidence for any innovation of this type during the twentieth century,6 he expects that the economy will be unable to regain much dynamism during the coming two to three decades, and probably longer. In his view, growth could fall to just 0.2 percent in 2100, as shown in Figure 2 below. This implies almost no productivity growth whatsoever by that time.
Is there another way to think about the outlook for productivity, growth, and jobs? There may be.
Since we are entering a services-oriented world, it’s possible that innovation will not be discrete, as Gordon argues it should be. If the Internet and networking benefits proposed here are not discrete, but occur in a “software and services chain,” important benefits may occur as new innovations improve the chain, and are used to create new services. In the next section, we explore data about the growth of “unicorns,” Internet-connected firms that rapidly reach a $1 billion valuation. This data begins to suggest that today’s situation may require new thinking about how innovation occurs and what needs to be measured.
Innovation in the U.S. economy
I would like to argue that innovation in software—particularly software that runs computing and networking in large enterprises, and that constitutes the infrastructure for firms such as Facebook and Twitter—is establishing a base for substantial new growth. This provides a dramatic contrast to the analysis of artificial intelligence, which is one part of the story of emerging technology and innovation.
The fundamental case I put forward is that enterprises are only now shifting from the initial efforts they have made to put much of their computing activity into the cloud. They are undertaking far more substantial transformations for several reasons:
- The Internet of Things is being implemented much more rapidly in the auto industry due to the introduction of driverless car technology. Without human intervention, you already can call a BMW from a spot in a parking lot using “valet assistant” and have it drive to the entry to pick you up. Mercedes has a “stop and go” autopilot, and Cadillacs soon will have an autopilot that manages lane changes. This effort will help similar initiatives, which are changing health care delivery and retailing.
- Pressed by the need to include mobile devices in the way they operate, and to connect with consumers, corporations are ripping out their old infrastructure. They are introducing a type of agile “chained software” to create new services and applications faster than was previously possible. With 70 percent of all purchases coming from mobile devices, Facebook, Twitter and others, firms such as Walmart must be agile, or they will lose out to more innovative and nimble competitors. The old infrastructure will not meet such new demands. So twenty-first century firms are ripping out their traditional infrastructure and replacing it with software-defined and “chained infrastructure”—virtualizing7 not only computing, but also networking—that creates new types of pipelines for continuous service delivery.
- New networking technologies are solving the problem of how to function in a 24-hour-a-day world. This is a shift to a global view that many scholars have not appreciated. It also has been facilitated by technologies derived from the Internet. This change is essential for financial, logistical, and transportation firms. Once they adopt it, other firms are likely to follow.
In sum, these changes are likely to revitalize a number of firms and alter their business models.
They also will create a shareable, often “open source,” software that is published on easily accessible websites, such as GitHub. In essence, innovations that larger firms initiate can provide the foundation for smaller, innovative firms. Subsequently, this can open the way for entirely new industries to develop, including the mobile financial technology firms I mention below.
To offer more information about the types of innovations that are the basis for these changes, I have created the chart below to identify the main innovations in different aspects of the Internet and enterprise computing, storage, and networking.
I also have drawn upon studies to illustrate the early economic benefits associated with such changes. This information is provided in the following chart:
I think these innovations will serve as fundamental building blocks for the knowledge economy, creating entirely new areas for economic activity. From recent investments by venture firms, we can identify a number of “emerging” growth sectors. Writers, including Cesar Hidalgo in Why Information Grows? and Enrico Moretti in The New Geography of Jobs, have espoused views very similar to my own.
If we are on the verge of a big shift in innovation, it is possible that traditional measurements have been unable to decipher what is actually happening. Several analysts have argued that young firms are declining in importance to the U.S. economy. One study that compiles data showing this trend is by Decker and others.
Katherine Kobe8 finds a similar decline in the share of GDP produced by small business 2002–2010, but finds that “small businesses continue to be incubators for innovation and employment growth during the current recovery. Small businesses produced 46 percent of the private non-farm gross domestic product (GDP) during 2008 (the most recent year for which source data are available to make these estimates). That is down from the 48 percent share of GDP produced by small businesses during 2002. Preliminary information indicates that the weak business conditions through early 2009 affected small businesses about as much as large businesses, resulting in only a minor change to the shares.” While acknowledging a decline in the small business contribution during 2010, she finds that small businesses held their own for much of the recovery, although large firms recovered more successfully, as shown in the figure below.
What might suggest that innovation is making more positive contributions to the economy? I would argue that venture capitalists’ recent funding for new businesses identifies a vibrant core of new firms in some key sectors. Quite a number of these firms are connected to the growth of infrastructure and the Internet.
Venture capitalists have named firms that rapidly attain a valuation of $1 billion or more “unicorns.”9 The explosive growth of unicorns may say more about innovation and growth than do the aggregate statistics presented above. Research from CB Insights indicates that the pace of unicorn creation is ramping up rapidly during 2015, likely to double the 2014 pace.
CB Insights also provides data about investments in emerging industries. A good example is mobile financial technology. In this sector, funding has supported more than twenty firms each quarter during the past three years, with investments almost twice the size during second quarter 2015 compared with three years earlier.
In a similar manner, CB Insights notes that investment in on-demand firms such as Uber is quite strong. These firms are Internet-based, largely implementing new techniques to develop software, such as Microservices,10 and able to keep up with spikes in demand while creating innovative new services. Although this area is dominated by Uber, which raised 39 percent more funding during 2014 than all other on-demand startups combined, it is still interesting to see the rate at which funds have focused on startups in this area.
In addition, the accompanying figure on funding for on-demand startups shows that they are maturing and receiving funds in follow-on rounds beyond the initial round of angel/seed funding. The shift in later-round funding 2010–2014 illustrates this trend.
While the U.S. economy appears to be recovering well, it has taken a while to get to a more dynamic state. Unemployment remained high long into the recession, and business profits did not recover until recently.
There is an open question about what happens next. Some economists believe that automation will destroy a large number of jobs during the next two decades. Others are concerned that sluggish productivity gains will dampen growth and result in a return to an economy similar to that of the Victorian era, with few productivity gains and very modest growth.
I have suggested that there is a dynamic group of firms that already is exploiting new Internet and networking technologies to operate in far more dynamic ways than its competitors are. This trend is apparent in many of the more infrastructure-oriented firms that are “early adopters” of the technologies I mentioned above.
In addition, another group of startup firms is exploiting many of the same technologies to build new markets and create new services in industries such as finance. These firms, the most successful examples of which are called “unicorns”—firms that have attracted $1 billion in financing very rapidly—have potential to change investment patterns as well as productivity growth.
These two types of firms, rapidly adjusting older firms and highly innovative new ones, indicate that a brighter future for innovation and growth in the U.S. economy is a possibility. For it to come to pass, there would have to be more evidence that firms in emerging industries will attract large new investments. If this happens, it could raise investment trends above the levels in the data presented here. It also probably would contribute to a rise in productivity.
As the U.S. economy becomes more service-oriented, U.S. government statisticians may need to develop better tools to measure what is happening. One reason to do this is that innovations in services may not be as discrete as they were during the industrial revolution, when electricity and machine power were new. Rather, there may be incremental but significant innovations that we do not yet know how to measure very well.
If we can move productivity estimates higher based on the emerging industries of the Internet, I think we can revise Gordon’s forecast to be more optimistic.
- Gordon, Robert J. “Is U.S. Economic Growth Over? Faltering Innovation Confronts the Six Headwinds.” www.cepr.org.
- Using data from the Florida Department of Taxation, I tracked consumption of consumer products during the months following the collapse of housing prices within several counties that were the focus of the state’s housing crisis. This data shows a rather dramatic decline in consumption spending within a few months of significant declines in housing prices.
- Blinder, Alan S. “Quantitative Easing: Entrance and Exit Strategies.” Federal Reserve Bank of St. Louis Review, pp. 465–480, (November/December 2010).
- Smithers, Andrew. “U.S. Labor Productivity: Key to Lasting Growth.”
- Gordon, Robert J. “Is U.S. Economic Growth Over? Faltering Innovation Confronts the Six Headwinds,” Centre for Economic Policy Research, CEPR Policy Insight No. 63 (September 2012).
- I provide a counter view that cites a number of examples of innovation in the Internet-related economy in a paper, “A Software-Defined Economy: Innovation, the Internet and Future Growth,” Columbia University CITI Conference on Internet and Employment (June 5, 2015).
- See Intel IT Center, “Virtualization and Cloud Computing.” (August 2013).
- Kobe, Kathryn. “Small Business GDP: Update 2002–2010,” Economic Consulting Services, LLC, Washington, DC, prepared for the Office of Advocacy, United States Small Business Administration, p.3 (January 2012).
- CB Insights, “Unicorns Are Breeding Like Rabbits: Set to Double 2014′s Record Pace.” (May 21, 2015).
- Cockcroft, Adrian. “Creating Business Value with Cloud Infrastructure.” Open Networking Users Group conference, May 13–14.