Thursday, March 30, 2006

March 2006 Early Indications I: What is the digital era?

It's commonplace to refer to our situation as a digital economy, an information age, or a post-industrial society. Because we have almost exactly 50 years of experience with commercial computers (GE purchased the first commercial Univac system for its Louisville appliance plant in 1954), it's worth trying to analyze how the U.S. has changed in that time. Isolating how information technology has shaped the economy and society turns out to be much less direct than looking for the impact of, say, the automobile.

To begin with, the nation has steadily produced more and more economic value. In 1900, U.S. GDP per capita, in current dollars, was $268. By 1950, that figure had multiplied seven times to about $2000. Between 1950 and 2000, the multiple was eighteen. To put that per capita figure in perspective, real GDP rose from $294 billion to $9817 billion: a 33-fold increase. (That ratio of 18 to 33 suggests that total population nearly doubled, which it did, from 151 million to 281 million. I don't know how much the apparently sudden growth could be methodological, but I was quite surprised to see a counter on the Census site indicating that the U.S. is just about to cross the 300 million mark. It probably has, depending on how illegal aliens are counted.)

The role of agriculture has changed in surprising ways. The number of farms in the U.S. in 1950 was almost the same as in 1900, a little over five and a half million after having peaked in the mid-1930s. By 2000, the number of farms had dropped to 2.2 million, but the average size, possibly reflecting the rise of organic farms, was actually dropping from its high in 1994. The amount of total acreage in farms reached its peak in 1953: for all the talk of urbanization in the late 19th century, it turns out that in 1950 the U.S. was still robustly rural, in land use terms anyway.

The surprising amount of farm acreage belied a strong population shift, however. In 1900, 41 percent of the U.S. workforce was employed in agriculture. The number fell to 16 percent in 1945, and in 2000 less than 2 percent did so, most part-time. Agriculture was less than 1 percent of GDP by that time, a ninth of what it had been in 1945.

Where did people go when they left the farms, which were mostly located in the Midwest? South and west, of course: the 13 states that constitute the Census Bureau's West region (every state west of Texas) combined to grow from 13 percent of U.S. population in 1950 to 22.5 percent fifty years later. I was surprised that the South only increased from 31 to 36 percent. Before talking about macroeconomic effects of computing, one has to appreciate the impact of air conditioning on where people live: such fast-growing states as Virginia, Georgia, Texas, and Arizona can be uncomfortable and unhealthy without climate control. Air conditioning in turn raises the importance of electric power, which was still a novelty in the rural South in 1950.

Along with internal migration, the second half of the twentieth century was marked by broad social change: political and economic involvement by women moved broader, in terms of numbers, and deeper, in terms of impact. Women doubled their participation in the workforce, from 30 to 60 per cent, and now often, but not routinely, hold seats as CEOs, senators, Supreme Court justices, and astronauts. Women also constitute well over half of the college population only a generation after having gained admittance to the leading private universities. The U.S. is also a much older nation than in 1950. Life expectancy at birth has risen from 68 to 77. Both of these trends relate closely to changes in medical technology and, for women, birth control. It will require further study to determine how much the increase in life expectancy relates to computing: trends in immunization, smoking, nutrition, and cardiac interventions are, I suspect, far more important.

In demographic terms, the automobile stands out among technologies with major impact. The pervasiveness of its influence tracks closely with the invention and habitation of the suburb - a term that did not exist at the time of the 1900 census. But from 1910 until 1950, when the percentage of population in suburbs more than tripled, to 23 percent of the population, the rise of the automobile literally reshaped the landscape. In the fifty years that constitute our focus here, the percentage of population in suburbs more than doubled: fully half the U.S. population now lives in suburbs, a striking testimony to the geographic transition caused by the automobile.

Shifting from residence to occupation, manufacturing surged at agriculture's expense as the air conditioning- and automobile-related figures would suggest. But while manufacturing employment peaked in 1979 at over 19 million jobs, it had been declining since 1953 as a percentage of total employment: the drop, from one job in three to one in ten, constitutes another defining characteristic of the past half-century.

The picture of how and why this shift occurred is complex, politically loaded, and still poorly understood, but one salient factor ties agriculture and manufacturing: massive productivity growth. Just as less than one percent of the workforce can go a long way toward feeding the U.S. population while exporting over $50 billion worth of food a year, so too has manufacturing been able to produce more and more with fewer inputs of labor. The role of computerization in productivity growth has yet to be fully analyzed, but it clearly contributes.

Given that the 1950 to 2000 half-century marked precipitous loss of employment, as a percentage of the workforce, in both agricultural and manufacturing arenas, we know that services have become the dominant economic sector. According to the CIA's World Factbook, "industry" constitutes 21% of the U.S. economy, while services add up to 78 percent. How could this sector grow so big so fast, and what are some of the implications?

For all the rhetoric about "becoming a nation of burger-flippers," government has become a much bigger economic entity that gets bundled into services: prison guards, for example, constitute one of the fastest-growing occupational categories. As of 2002, there were more government employees -- about 19 million at all levels, not counting military personnel -- than in any other standard industry group. Health care was second, and retail was third.

All three of these industries, writ large, have failed to harvest the productivity increases that manufacturing or, say, financial services have. How much and how well these three industries adopt information technology will go a long way toward dictating the fate of the tech sector. A further difficulty in assessing IT's impact comes in the measurement of services productivity: x units of input create y units of manufactured (or harvested) output, but if the input is time spent working, measuring the output of a teacher, software engineer, or bank teller is much less straightforward.

Because they frequently don't store or travel well, services are harder to export than manufactures or food. Once again using NAICS definitions from 2002, of the top five service categories by revenues, retail and health care are highly localized, finance and insurance face cross-border regulatory issues, and information and professional/scientific/technical services, while valuable and often portable, face price-based competition from off shore. If trends in offshoring are any indication, it may be easier for the U.S. to import services (such as remote radiographic analysis, call centers, and computer programming) than it is to export such activities as investment banking, neurosurgery, or engineering education.

The uncomfortable juxtaposition of globalization and locality is not a new phenomenon -- just look at England in the twilight of empire. If people only earn money from local sources but spend it on goods and, increasingly, services from "away," eventually money needs to come back into the locality: just as a multi-crop family farm is no longer a viable option, neither is a self-sustaining local economy. Somehow, money needs to come in as well as leave, and the current trade imbalance and federal debt levels both ratchet up that imperative.

Every new technology has at least three uses. You can use it to:

-do things you've been doing, but faster and easier (in shorthand, this is automation)
-make things to do things (capitalization)
-do new things that were previously impossible (innovation).

In looking over the government's service categories, it's hard to see the capitalization and innovation. Let me be clear: the automation effects are often substantial, and there are elements of technology-driven innovation in service industries. A word processor not only automates a typewriter but allows new freedom to revise and to reuse. Google more than automates a reference library. In terms of capitalization, eBay allows anyone to become a retailer with just a PC for infrastructure.

But the main service innovations appear to be in recreational and peripheral (no pun intended) activities rather than at the heart of the economy. Cellular telephony is a large industry, but a) it's cannibalizing an existing sector and b) it's no automobile in its impact. Digital gaming is an innovation, but hardly a core activity, and it has yet to disrupt adjacent industries like movies or education. The automation effects of ERP replacing a general ledger or CAD replacing drafting tables are significant, without a doubt, but business computing innovates far less often and powerfully in services. As a macroeconomic replacement for manufacturing, the information sector falls short - today.

Like many others, I persist in believing that the transformative power of computing lies ahead of us. Whether it's in genome-aware therapeutics, or rich media self-publishing, or low-cost avionics that make small jets feasible as air taxis, the majority of digital innovations that will remake the economy are as yet uncommercialized. And compared to such landmarks as the invention of the steam engine or the factory system, the fifty years of computing represents enormous change in short time. The daunting fact is that the change to come looms even bigger.

In the interim, the demands of a digital economy contribute to complex and difficult demographic issues: skill- and education-based bifurcation, along with a changing racial composition. While I need to research the matter further, my hunch is that factory work paid better than farm work and was widely accessible at the low end. People could leave farms, enter manufacturing with no or few skills and little education, and stay afloat. A further correlate here is decentralization: factory work collects resources in one place, while services industries (and powerful communications networks) disperse them. What are the consequences of the south and west's growth without a heavy reliance on industrial centers like Milwaukee, Pittsburgh, or Detroit?

Prior to and during World War II, the internal migration of black Americans from the rural south to the industrial midwest led to such varied changes as a rebirth of popular music, a power base for the Democratic party, and the rise of a black middle class. Only one or two cultural hops separate Henry Ford from Diana Ross, Lyndon Johnson, and the Cosby Show. Look a little closer and you see the Rolling Stones, Magic Johnson's NBA, and Oprah Winfrey, who was born in Mississippi but made her name in Chicago.

Now the opposite dynamic is at work as manufacturing automation and globalization release workers to take jobs in lower-paying categories such as hospital food service or big-box retail; in raw numbers, the biggest job-creators of the past few years have been Home Depot and Lowe's, and of course Wal-Mart's net role in employment is being hotly disputed. Retail and other services often teach their workers how to use automated systems but rarely prepare them to enter a better-paying sector. How the shift to services interrelates with America's racial picture, including of course the emerging Hispanic majority, will be critically important to track.

As the CIA's World Factbook puts the issue,

"The onrush of technology largely explains the gradual development of a 'two-tier labor market' in which those at the bottom lack the education and the professional/technical skills of those at the top and, more and more, fail to get comparable pay raises, health insurance coverage, and other benefits. Since 1975, practically all the gains in household income have gone to the top 20% of households."

The consequences of such a bifurcated populace touch sociology, politics, economics, and even ethics, so I won't even attempt a summary comment. Perhaps this trend is the result of moving farther and farther from a subsistence economy. One of the things we'll be tracking as this research progresses is the changing composition of the economy away from food, clothing, and shelter to transportation, entertainment, and other luxuries. The interplay of rapid population growth, rapid increase in the amount of liveable and available real estate, wider education, suburbanization, and the shift to a services economy all contribute to making the task of assessing information's role highly problematic.

But that won't stop us from trying.


Unless otherwise noted, all figures come from the U.S. Census Bureau.

CIA World Factbook

Carolyn Dimitri, Anne Effland, and Neilson Conklin, "The 20th Century Transformation of U.S. Agriculture and Farm Policy," USDA Economic Research Service Electronic Information Bulletin 3, June 2005.

Frank Hobbs and Nicole Stoops, "Demographic Trends in the 20th Century," U.S. Census Bureau, 2002.

Louis D. Johnston and Samuel H. Williamson, "The Annual Real and Nominal GDP for the United States, 1790-Present." Economic History Services, October 2005 (