This article appeared in Forbes ASAP (March 1993).
A few years back, MIT economist Robert Solow quipped, "We see computers everywhere but in the productivity statistics." His remark set off a controversy, now called the "productivity paradox," that refuses to go away. Swarms of investment analysts, economists and journalists have tried to dismiss this indictment of the computer industry by arguing that the government's worker-productivity data are so flawed they can't prove anything. But even after making all the reasonable adjustments, the same bottom-line message keeps coming back. Although U.S. businesses plunked down well over $1 trillion for computer systems in the last decade, they have almost no measurable productivity increase to show for it.
When I first heard about the computer productivity paradox, I couldn't believe there had been "no productivity growth." I can't even bear to think about how much time-wasting drudgery I'd have to go through without my Macintosh. And yet, plenty of bright people have investigated the productivity paradox for long enough to convince me that there's more to it than a statistical delusion.
Searching for a plausible explanation, I stumbled upon a brilliant but obscure research paper by Paul David, professor of economic history at Stanford University. Though economic historians often share the same floor in faculty office buildings with economic theorists, they remain a breed apart. Instead of wasting time massaging highly aggregated national income accounts data with impenetrable mathematical models, economic historians roll up their sleeves and dig out the hard facts of daily economic life. Much like paleontologists, they re-create bygone eras by patiently piecing back together the most minute details of forgotten technologies and extinct industries.
Intrigued by the flap over the productivity paradox, David decided to take a closer look at a remarkably similar historical puzzle. Ninety years before Intel invented the dynamic random-access memory chip, or DRAM, (1970) and the microprocessor (1971), Thomas Edison invented the carbon filament incandescent lamp (1879) and the central generating station (1881). Just as it took 10 years before a refined microprocessor -- embodied in the original IBM PC -- began changing our daily lives, it took nearly 20 years before electricity began having any significant impact on society. Indeed, it wasn't until the Paris Exposition of 1900 that electricity was used to light an entire city.
As the 20th century began, business investment in electrical equipment skyrocketed. Between 1900 and 1920 the percentage of U.S. factories equipped with electric motors jumped from five percent to 55 percent. Yet, despite the obvious advantages of electric motors over steam engines, worker productivity showed almost no measurable increase.
Solving the riddle of today's computer productivity paradox is simple, once its century-old predecessor is understood. Back then, a state-of-the-art facility was a three- or four-story brick building. A coal-fired steam engine sat in the factory's basement. Its power was transmitted to the equipment on the floors above through an elaborate system of vertical and horizontal shafts and drive belts.
Factory owners began replacing steam engines with large electric motors because the new technology slashed coal bills by 20 to 60 percent. But the power generated by these first electric motors was still conveyed to lathes, drills, grinders and punch presses by the conventional system of shafts and belts. As time passed and somewhat smaller electric motors became affordable, separate motors were installed on each floor. This eliminated the need for the vertical shaft. However, because the motors were still relatively expensive, each one powered a group of machines via horizontal shafts and belts.
Electric motor technology was indeed revolutionary, but to pay for itself it had to be grafted onto the existing industrial infrastructure. Multistory factories made sense in the steam age because they reduced the costly friction losses incurred as horizontal shafts carried power to the equipment. Single floor factories would have eliminated all the labor wasted moving unfinished goods from floor to floor, but overall it was cheaper to staff the elevators than to pay for the coal turned to waste heat by long horizontal shafts.
For similar reasons, machines requiring the most horsepower were placed near the base of each floor's horizontal shaft. Smaller, low-power-consuming equipment sat at the far end of the work floor. Again, though this arrangement made economic sense from a power-conservation standpoint, it made the floor of materials ridiculously inefficient. In short, since the entire industrial infrastructure had evolved around the assumption of expensive steam power, factories were designed to be grossly inefficient in other dimensions. Once made, these fundamental economic trade-offs endured for decades in bricks and motor.
It wasn't until cheap, small electric motors became available in the 1920s that factories began to abandon "group drive" power for the "unit drive" approach, the familiar present-day system in which each machine is powered by its own internal motor. As this technology was installed, companies ripped out their drive shafts and belts, rearranged their machines and smoothed their flow of materials. The most innovative "high-tech" firms-companies in fast-growing new industries such as cigarettes, organic chemicals and electrical equipment-were the first to go all the way and take the radical step of building single-story factories. They were literally the first "flat-tened" corporations.
Finally, 40 years after Edison's breakthroughs, productivity growth took off. Where the annual labor productivity growth had hovered around one percent for the first two decades of the century, it jumped to more than five percent a year during the Roaring Twenties.
Real economic life, it seems, is a tad more complicated than the input-output models of economic theorists. Pumping investment into a powerful new technology does not instantly and automatically yield productivity gains. At the start of the steam engine-electric motor transition, steam-based investments still worked and were still carried on company balance sheets. To reap the full promise of electricity, an almost endless number of practical details had to be worked out. That took time and enormous creativity. As with today's dream of the paperless office, no one knew what a fully electrified factory would be like or how best to manage it.
Looking ahead, no one can say when all the critical elements of the Information Age infrastructure will come together. But it appears that we may be on the cusp of that long-awaited productivity surge. The explosive growth of PCs, workstations, LANs, WANs fiber optics, wireless communications and object-oriented software may be competing the new office paradigm just as the unit drive motor completed the new factory paradigm of the 1920s. As these communications technologies link the previously isolated power of microprocessors, the cost of delivering the right information where its needed will collapse, allowing completely new work flows and organizational infrastructures to emerge. We already see evidence of these revolutionary changes on the horizon. Without the arrival of networked computers, there is little to explain the massive and permanent disappearance of middle-management jobs. Like the power-transmitting drive shafts and belts made unnecessary by the arrival of unit drive motors, information-transmitting white-collar workers are being ripped out of the emerging economic infrastructure.
Hierarchical "multistory" bureaucracies are being flattened into "single-story" organizations, not because of any sudden change of heart by the top management but because the hard economics of information handling are being utterly transformed. Even though in a paper-based world it made economic sense to minimize communication costs by co-locating workers in urban office towers, in the world of E-mail, faxes and teleconferencing the costs imposed by that infrastructure have become an incredible waste of money. More important, this shift has revealed that our culture or poor inter-personal communication is today's greatest barrier to productivity growth.
By simply extending already well-understood production techniques, makers of microchips and fiber optics are certain to drive the cost of computer cycles and data transmission down another thousandfold before the end of the century. As a consequence, all our sacred assumptions about the nature of organizations will be overturned. Like our predecessors in the early 1920s, we are being forced to rethink all the old trade-offs and invent new designs for our working lives.
Like most puzzles, the productivity paradox is an artifact of superficial thinking. By getting beneath its surface, by examining the computer productivity paradox in the light of economic history, we can understand our computer frustration. We can also foresee a productivity explosion that will soon dwarf the productivity surge that made the Roaring Twenties roar.
Copyright 1993 The Bionomics Institute
| Home | Resources | Bionomics: The Book |