An Access Advertising EconBrief:
Is GDP NDG in the Digital Age?
For years, we have heard the story of stagnant American wages, of the supposed stasis in which the real incomes of the middle and lower class are locked while the rich get richer. Various sophisticated refutations of this hypothesis have appeared. Households have been getting smaller, so the fact that “household income” is falling reflects mainly the fact that fewer people are earning the income that comprise it. “Wages” do not include the (largely untaxed) benefits that have made up a steadily larger share of workers’ real incomes ever since World War II.
But there is something else going on, something more visceral than statistics that leads us
to reject this declinism. It is the evidence of our own senses, our eyes and ears. As we go about our daily lives, each of us and the people around us do not exhibit the symptoms of a people getting materially worse off as we go.
For over thirty years, we have been forsaking the old broadcast trinity of network television stations, at first in favor of cable television and recently for a broadening array of alternative media. For over twenty years, our work and home lives have been dominated by desktop computers that have revolutionized our working and personal lives. For over ten years, an amazing profusion of digital products have taken over the way we live. Cell phones, smart phones, tablets, pads and other space-age electronic wonders have shot us out of a consumer cannon into a new world.
Can it really, truly be that we are worse off than we were before all this happened? As the late John Wayne would say if he were here to witness this phenomenon: “Not hardly.”
The pace of this technological revolution has not only been too fast for most of us to stay abreast of it. It has left many of our 20th century institutions blinking in the dust and gasping for breath. Mainstream economic theory and national income accounting, in particular, are trying to gauge the impact of a 21st-century revolution using the logic and measurement tools they developed in the first half of the 20th century.
The Case Study of Music
Music was one of the great consumer success stories of the 20th century. Thomas Edison’s invention of the phonograph paved the way for the recording of everything from live artistic performances to studio recordings of musicians and singers to the use of recorded sound tracks for motion pictures. The recordings themselves were contained on physical media that ranged from metal discs to vinyl to plastic. At first, these “records” were sold to consumers and played on the phonographs. Sales were in the hundreds of millions. Artists included some of the century’s most visible and talented individuals. The monetary value of these sales grew into billions of dollars.
Since recordings were consumer goods rather than capital goods, sales of records were recorded in the national income and product accounts. Or rather, the value added in the final, or retail, transaction was included. The value-added style of accounting was developed with the inauguration of the accounts in the late 1930s and early 40s in order to do three things: (1) show activity at various stages of production, but (2) highlight the new production of consumption goods each year to reflect the fact that the end-in-view behind all economic activity is consumption (3) by including only the additional value created at each stage to avoid double-counting.
As the 20th century came to a close, however, record albums were replaced by small audio discs that could be played on more compact devices. And these were soon supplanted by computers – that is, the playing medium became a computer and the music itself was housed within a computer file rather than a substantial physical object. As technology advanced, in other words, the media grew smaller and less substantial. But the message itself was unaffected; indeed, it was even improved.
How do we know that the value people derive from music has not been adversely affected by this transition to digitization? In The Second Machine Age, authors Erik Brynjolfsson and Andrew McAfee consider the question at length. In terms of physical units, sales of music have fallen off the table. Just in the years 2004-2008, they fell from roughly 800 million units to less than 400 million units – a decline of over 50% in four years! And the total revenue from sales of music fell 40% from $12.3 billion to $7.4 billion over the same period. By the standards we usually apply to business, this sounds like an industry in freefall.
In this case, though, those standards are misleading. During that same time span, the total unit-volume of music purchased still grew when purchases of digitized music where factored in. And acquisitions of music free of charge by various means swelled the total much, much larger. One of the things economists are best at is analyzing non-traditional markets, which is why Joel Waldfogel of the University of Minnesota was able to infer that the quality of music available to consumers has actually increased in the digital era. Today, anybody with a smartphone can access some 20 million songs via services like Spotify and Rhapsody. For those of us who recall the days of LPs and phonograph needles, the transition to today has been dizzying.
But the economics of the digital age have driven prices through the floor. As Brynjolfsson and McAfee observe, it is the same process that has driven the newspaper business to the wall and its readers online; the same one that has driven classified-advertising from newspapers to Craigslist; the same one that impels us to share photos on Facebook rather than buying prints for friends and family. “Analog dollars,” they conclude, “are becoming digital pennies.”
This creates an unprecedented marketplace anomaly. Measured by the value it creates for human beings, which is how economists want to measure it, the music industry is booming. But measured in dollars’ worth of marketplace transactions, which is how economists are currently able to measure it, the music industry is declining rapidly.
If the music industry were a singularity, we might treat it as a mere curiosity. It is not, of course; the gap between price/quantity product and value created yawns wide across the spectrum of industry. “By now, the number of pages and digital text and images on the Web is estimated to exceed one trillion…children with smartphones today have access to more information in real time via the mobile web than the President of the United States had twenty years ago. [!] Wikipedia alone claims to have over fifty times as much information as Encyclopedia Britannica, the premier compilation of knowledge for most of the twentieth century.”
“…Bits are created at virtually zero cost and transmitted almost instantaneously worldwide. What’s more, a copy of a digital good is exactly identical to the original… Because they have zero price, these services are virtually invisible in the official statistics. They add value to the economy, but not dollars to GDP… When a business traveler calls home to talk to her children via Skype, that may add zero to GDP, but it’s hardly worthless. Even the wealthiest robber baron would have been unable to buy this service [in the 19th century]. How do we measure the benefits of free goods or services that were unavailable at any price in previous eras?”
This understates the case. As Brynjolfsson and McAfee acknowledge, most of the new digital services substitute for existing services whose sales contribute to GDP. Thus, the digital bonanza actually lowers measured GDP at the same time that our well-being rises. In economic jargon, the effect on GDP’s function as index of national welfare is perverse.
This leads many people, including these authors, to the conclusion that GDP is no longer an adequate measure of national output. If this is true, it makes our monthly, quarterly and annual preoccupations with the growth rate of GDP seem pretty silly. The government agency whose task is the compilation of economic statistics is the U.S. Bureau of Economic Analysis. Its definition of the economy’s “information sector” aggregates sales of software, publishing, movies, audio recordings, broadcasting, telecommunications, and data processing and information services. These sales account for about 4% of measured GDP today. Yet we are commonly understood to be chest-deep in a new “economy of information” that is replacing the economy of tangible goods and services. Either this perception or that 4% metric is wrong; the latter seems vastly more probable.
What’s more, the irrelevance of GDP increases by the nanosecond.
Of course, not all digital products and services are substitutes for existing counterparts. Some of them are genuinely new. If these are similarly hard to incorporate in GDP, the distortion may be only half as great as that described above. But the digital revolution has displayed a propensity for creating things that were unknown heretofore but that soon became necessary accoutrements of daily life.
Longtime macroeconomist and textbook author Robert Gordon estimated the value of new goods and services added but missed by GDP at about 0.4% of GDP. That may not sound like much, but since the long-term average annual rate of productivity growth is around 2%, it would mean that we are overlooking 20% of annual productivity.
GDP and Investment: The Bad News Gets Worse
GDP is failing because it neglects to measure the tremendous increases in consumption and well-being conferred by the digital age. But GDP also measures investment, or purports to. Are its failings on the consumption side mitigated by its performance with investment?
No, they are magnified. The production of digital goods and services is heavily dependent on intangible assets rather than the familiar plant and equipment that are the focus of traditional investment. Brynjolfsson and McAfee identify four categories of these intangibles: intellectual property, organizational capital, user-generated content and human capital. It comes as no surprise to find that the measurement of these assets largely eludes GDP as well.
Intellectual property encompasses any creation of the human mind to which legal ownership can be attached. Patents and copyrights form the backbone of this category. A great deal of spending on research and development (R&D) constitutes investment in intellectual property.
Yet R&D has long been recognized as almost impossible to accurately measure because only its cost is transparent, while the value (e.g., capital) it creates often escapes measurement.
Organizational capital is an even broader concept intended to capture the value inhering in brands, processes, techniques and conceptual structures owned by particular businesses. This category long predates the digital age but is idealized by companies like Apple, whose brand and unique corporate style complement its portfolio of intellectual property to create perhaps the world’s most productive company. Accountants have long sought to put a price tab on things like “good will” and “brand name.” We have observed that the transition to a computer-savvy work force has necessitated investment in procedures and processes far greater than the initial spending on the computer hardware and software – spending that doesn’t show up in the national income accounts as investment.
User-generated content is a true digital innovation. Facebook, Twitter, YouTube, Pinterest, Instagram, Yelp and countless other websites are largely created by their users. The value of this approach is both undeniable and subjective, as anybody who has every previewed a restaurant on Yelp or planned a vacation with TripAdvisor can testify. The feedback generated these sites provides an object lesson in the generation of information – the kind of information that economists had to assume that people already knew because we didn’t know how markets could make it available to them. Now we do.
Human capital was a concept invented and popularized by economists Theodore Schultz and Gary Becker decades before the Internet existed. The talents, skills and training that we receive make us better productive “machines,” which inspired the analogy with physical capital.
How important are these intangible assets in the modern economy? Nobody knows with certainty, but – as always – economists have made educated guesses. Brynjolfsson and McAfee estimate the value of organizational assets as some $2 trillion. The preeminent theorist of investment, Dale Jorgenson, estimated that human capital is worth 5-10 times as much as the stock of all physical capital in the U.S. Investment in R&D has been estimated at roughly 3% of GDP in recent decades.
The degree of distortion in GDP numbers – specifically in measures of productivity, which compares growth in inputs and output – is harder to gauge in this case than in the consumption example. Some intangible assets, like R&D and human capital, are longtime thorns in the sides of statisticians; their measurement has always been bad and may be no worse now than before. In some cases, the distortions in investment may offset those in consumption, so that the measure of productivity may be accurate even though the numerator and denominator of the ratio are inaccurate. But the elements most closely associated with the digital revolution, such as user-generated content, impart a huge downward bias to measured productivity in the national income accounts.
A New, Improved GDP?
Economists and other commentators have done a good job of diagnosing the havoc wreaked on GDP by the digital revolution. Alas, they have rested on those laurels. In the “solutions and policy proposals” section of their work, they have fallen back on the tried and trite. GDP was a sibling of macroeconomics; the economic logic underlying the two is the same, with the operative word being “lying.” Macroeconomists are loathe to repudiate their birthright, so their reflex is to cast about for ways to mend the measurement holes in GDP rather than abandon it as a bad job. Hence the rosy glow cast by Brynjolfsson and McAfee over nebulous concoctions like the “Social Progress Index” and the “Gallup-Healthways Well-Being Index.” As for the touted “Gross National Index” of Bhutan, the less said about this laughable fantasy (treated in a previous EconBrief), the better.
The authors cite the comments of Joseph Stiglitz, whom they call “Joe” to profit by the implied familiarity with a Nobel laureate: “…Changes in society and the economy may have heightened the problems at the same time that advances in economics and statistical techniques may have provided opportunities to improve our metrics.” The “improvements” don’t seem to have included the ability to stop the scandalous misuse of the concept of “statistical significance” that has plagued the profession for many decades.
In fact, GDP has been known to be a failure almost since inception. Introductory economics textbooks routinely inculcate students in the shortcomings of GDP as a “welfare index” by listing a roster of flaws that predate the digital age, the Internet and computers. It has ignored the value of household services (predominantly provided by women), ignored the value created by secondary transactions of all kinds of used goods, undervalued services and thrown up its figurative hands when confronted by non-market transactions of all kinds. Its continued use has been a grim tribute to Lord Kelvin’s dubious dictum that “science is measurement,” the implication being that measuring badly must be better than not measuring at all.
What’s more, the blame cannot be laid at the feet of economic theory. It is certainly true that the digital age has brought with it a veritable flood of “free” goods – seemingly in contradiction with Milton Friedman’s famous aphorism that “there is no such thing as a free lunch.” Hearken back to Brynjolfsson and McAfee’s words that “bits are created at virtually zero cost.” A fundamental principle – perhaps the fundamental principle – of neoclassical microeconomics is that price should equal marginal cost, so that the value placed on an additional unit of something by consumers should equal its (opportunity) cost of production. When marginal cost equals zero, there is nothing inherently perverse about a price approaching zero. No, the laws of economics have not been suspended on the Internet.
Careful comparison of the age-old flaws of GDP and its current failure to cope with the challenges posed by digital innovation reveal a common denominator. Both evince a neglect of real factors for lack of a monetary nexus. The source of this insistence upon monetary provenance is the Keynesian economic theory to which the national income accounts owe their origin. Keynesian theory dropped the classical theory of interest in favor of a superficial monetary theory of liquidity preference. That is now proving bogus, as witness the failure of Federal Reserve interest-rate policies since the 1960s. Keynesian theory gives spending the pride of place among economic activity and relegates saving and assets to a subordinate role. Indeed, the so-called “paradox of thrift” declares saving bad and spending good. No wonder, then, that the national income accounts fail to account for assets and capital formation in a satisfactory manner.
Instead of tinkering around the margins with new statistical techniques and gimmicks when they have not even mastered basic statistical inference, economists should instead rip out the rotting growth root and branch. Reform of macroeconomics and reform of the national income accounts go hand in hand.
End the Reign of GDP
The digital age has merely exposed the inherent flaws of GDP and widened its internal contradictions to the breaking point. It is time to dump it. The next measure of national output must avoid making the same mistakes as did the founders of the national income accounts nearly 80 years ago.
The next EconBrief will outline one new proposal for reform of the national income accounts and explain both its improvements and shortcomings.