Enhancing Productivity Growth in the Information Age: Measuring and Sustaining the New Economy (2007)

This page intentionally left blank.

Faster, better, and cheaper semiconductors and computers as well as software and telecommunications equipment have led, especially over the past decade, to the widespread adoption and use of modern information and communications technologies. This, in turn, is rapidly ushering fundamental changes to the way in which (and the rapidity with which) goods and services are developed, manufactured, and distributed around the world and the way in which individuals and businesses everywhere consume, interact and transact. This “New Economy” poses new challenges, requiring new approaches to economic measurement and policy analysis.

To this end, the National Academies’ Board on Science, Technology, and Economic Policy (STEP) has since 2000 held a series of workshops to better understand the New Economy phenomenon and to develop policies needed to sustain the positive contribution of modern information and communications technologies to U.S. growth and competitiveness. This section of the report summarizes and provides background for some of the key issues raised over the course of the five conferences hosted by the STEP Board (listed in the Preface) on Measuring and Sustaining the New Economy.

The proceedings of each of these conferences have been published in separate volumes by The National Academies Press. Although the technologies of the industries considered at these conferences continue to evolve rapidly, the reports nonetheless capture conceptual issues of continued policy relevance to the industry leaders, academics, policy analysts, and others who participated in these workshops.

At the time of the STEP Board’s first conference in 2000, many economists were still reluctant to proclaim a technology-driven New Economy if only because there were few or no data reflecting economy-wide returns to the substantial investments made by U.S. business in new information and communications technologies.¹ Throughout the 1970s and 1980s, Americans and American businesses regularly invested in ever more powerful and cheaper computers and communications equipment. They assumed that advances in information technology—by making more information available faster and cheaper—would yield higher productivity and lead to better business decisions.

The expected benefits of these investments did not appear to materialize—at least in ways that were being measured. Even in the first half of the 1990s, productivity remained at historically low rates, as it had since 1973. This phenomenon was called “the computer paradox,” after Robert Solow’s casual but often repeated remark in 1987: “We see the computer age everywhere except in the productivity statistics.”²

At the National Academies first conference on the New Economy, however, Dale Jorgenson pointed to new data that showed that the U.S. economy was undergoing a fundamental change.³ While growth rates had not returned to those of the “golden age” of the U.S. economy in the 1960s, he noted, new data did reveal an acceleration of growth accompanying a transformation of economic activity. This shift in the rate of growth by the mid-1990s, he added, coincided with a sudden, substantial, and rapid decline in the quality-adjusted prices of semiconductors from an average of 15 percent annually before 1995 to 28 percent annually after 1995.⁴

In response to the rise in capability of computers and drop in price, investment in semiconductor-based technologies exploded, leading to a positive impact on economic growth. Jorgenson and Stiroh have calculated that computers’ con-

tribution to growth rose more than five-fold, to 0.46 percent per year in the late 1990s. Software and communications equipment contributed an additional 0.30 percent per year for 1995-1998. And their preliminary estimates through 1999 revealed further increases for all three categories.⁵ Jorgenson thus made the case for “raising the speed limit”—that is, for revising upward the intermediate-term projections of growth for the U.S. economy.⁶

Moore’s Law describes the speed at which semiconductor technology develops. Semiconductors are the core enablers for the wide array of information and communications technology. The pace of semiconductor development is, therefore, critical to the development of the broader range of computing and telecommunications technologies that are the basis for modern economic processes.

Moore’s Law is based on a prediction made by Gordon Moore in a 1965 paper titled “Cramming More Components onto Integrated Circuits,” where he noted:

Extrapolating this trend (see Figure 1), Gordon Moore predicted an exponential growth of chip capacity at 35 to 45 percent per year through 1975.⁸

Gordon Moore revised his original prediction in 1975 (the endpoint of his earlier projection) stating that increases in components per chip would continue, approximately doubling every 2 years, rather than every year.⁹ Believing that human ingenuity would further sustain the growth of chip capacity, he noted that manufacturers were using “finer scale microstructures” to engineer higher density of components per chip.

As Kenneth Flamm pointed out at the National Academies’ 2001 conference on semiconductors, the idea popularly known today as “Moore’s Law” (drawn from but not identical to Gordon Moore’s predictions) anticipates the doubling of

FIGURE 1 The original “Moore’s Law” plot from Electronics, April 1965.

the number of transistors on a chip every 18 months.¹⁰ While not deterministic, Moore’s Law accurately reflects the pace for growth in the capacity of memory chips and logic chips from 1970 to 2002, as shown in Figure 2.¹¹

FIGURE 2 Transistor density on microprocessors and memory chips.

As Kenneth Flamm further noted, Moore’s Law also captures an economic corollary that successive generations of semiconductors and related information technology products will not only be faster but also successively cheaper. Data from the Bureau of Economic Analysis (BEA), depicted in Figure 3 (and displayed by Dale Jorgenson at the conference on software), shows that quality-adjusted semiconductor prices have been declining by about 50 percent a year for logic chips and about 40 percent a year for memory chips between 1977 and 2000. This is unprecedented for a major industrial input.

The Moore’s Law phenomenon also appears to extend from microprocessors and memory chips to high-technology hardware such as computers and communications equipment. BEA figures highlighted by Dale Jorgenson reveal also that computer prices have declined at about 15 percent per year since 1977. (See Figure 4.)

FIGURE 3 Relative prices of computers and semiconductors, 1977-2000.

NOTE: All price indexes are divided by the output price index.

While Moore’s Law appears to predict ever “faster, better, cheaper” semiconductors and computers, it is not a deterministic law of nature, enduring instead by setting the expectations among participants in the semiconductor and computer industry of the pace of innovation and the introduction of new products to market. Before describing the basis of Moore’s Law and what is required to sustain this remarkable phenomenon, we first summarize some of the discussion of the economic implications of Moore’s Law and the challenges they pose to measuring the New Economy.

Measuring the New Economy is a challenge given the fast-changing nature of information and communications technology and the complex and often-invisible roles it plays in economic processes. This means that current data collection methods have to be updated to stay relevant to new products, new categories, and new concepts.

FIGURE 4 Relative prices of computers, communications, and software, 1977-2000.

NOTE: All price indexes are divided by the output price index.

As several participants at the initial conference noted, conventional statistical methods are not adequately adapted to capture what is happening in the economy. Illustrating the challenges facing the federal statistical system, Timothy Bresnahan of Stanford University noted the discrepancy between measures of output in the information technology sector (which he noted are adequate) and measures of output where information technology is used as an input in other sectors (which are not).¹² Shane Greenstein of Northwestern University added that conventional measures of Gross Domestic Product (GDP) provide good data on established channels by which goods and services are distributed, but fail to capture such information about goods and services when there are concurrent changes in the distribution methods.¹³

Illustrating the implications of asymmetries in data availability, Lee Price (then of the Department of Commerce) observed that data on the value of pre-

packaged software (which is more easily measured in terms of both nominal value and price) might not be as important to productivity as custom and own-account software whose value is more difficult to capture—resulting in their under-valuation. He stressed the need to refine statistical methods to better quantify the value of information technology.¹⁴

Several participants at the initial conference also emphasized the problems in valuing information technologies. Kenneth Flamm observed that it is difficult to calculate the percentage of improvement in computers that come from semiconductors.¹⁵ Eric Brynjolfsson of the Massachusetts Institute of Technology (MIT) further noted some hazards in equating price with value for computers, particularly given that many consumers are not price-sensitive, valuing service, brand loyalty, and perceived quality instead.¹⁶ Further to the issue of value, David Mowery of the University of California at Berkeley noted that it is statistically difficult to see the contributions of the semiconductor industry since it is hard to measure the output of “user” industries. He added that the economy outside the computer industry has become “a bit of a black planet” in terms of understanding quality improvements in its products.¹⁷ This value issue was further elaborated at the conference on Deconstructing the Computer.

Additional measurement challenges deal with how well information technologies are integrated and adopted across the economy. There were divergent views on where the United States was on the technology adoption curve at the turn of the century: Some argued that the United States was near the bottom of the S-shaped curve and about to take off; others suggested that the United States was in the middle and thus enjoying rapid productivity gains from the widespread adoption of information technologies; still others believed that marginal productivity gains from information technologies might be declining, signifying that the United States was already near the top of the technology adoption curve. This diversity of opinion, and the contrasting policy actions that it implies, pointed to a need to better measure the distinctive features of today’s economy.

Another major constraint in sustaining the growth in productivity is the rate of technology absorption. Sid Abrams of AT Kearney noted that business organizations often face challenges in reengineering themselves to take better advantage of the technologies available. While the cutting edge of technologies may advance, their potential to advance business productivity may depend on the extent to which executives and others are aware of the possibilities and/or uncertain of the effects of adopting new technologies in their organization.¹⁸ Indeed, as Ralph Gomery of the Alfred P. Sloan Foundation noted in the roundtable discussion that concluded the initial conference, the ability to absorb rapid advances in technology and the cost of re-doing the business organization to take advantage of these advances are, in many cases, more significant for sustaining productivity-led growth than the rate of technological advance. In essence, the question is not merely one of better or cheaper technology, but rather one of how enterprises can integrate productivity-enhancing technologies into the way business is conducted.

Sustaining the benefits of new technologies requires that we better understand the nature of these technologies and the circumstances that promote their development and deployment. STEP’s series of conferences on the New Economy has thus sought to bring together leading economists and also to draw on the knowledge and experience of industry leaders and other experts to describe current trends and their origins, with the challenge to economists to identify data and tools required for measuring and modeling key facets of the New Economy.

Reflecting the centrality of semiconductors to the information technologies, STEP’s conference of September 24, 2001, examined the rapid evolution of semiconductor technologies and a possible modeling strategy that could be used to predict the effects of alternative policy choices for the semiconductor industry.

Participants at this conference noted that semiconductors are the basis of today’s computing, information, and communications technologies. Rapid increases in the power of semiconductors, foreseen by Moore’s Law, and corresponding rapid declines in the price of semiconductor-based information technologies have lead to their swift diffusion across the economy and propelled their adoption across an array of applications.¹⁹

Drawing on his 2001 presidential address to the American Economics Association, Dale Jorgenson reminded the conference participants that the resurgence in the U.S. economic growth trajectory since 1995 is associated with a “relentless” fall in semiconductor prices and coincident with a shift in product cycle for semiconductors from 3 to 2 years.²⁰ Jorgenson drew attention to a series of documented events—summarized in Box B—between an intensifying pace of competition in the market for semiconductor products and the boost in the aggregate growth rate of the U.S. economy.

Given that a disproportionate share of growth appears to be generated by increased efficiencies related to the production and use of information technology (IT), the economic consequences of a two-year product cycle—as opposed to a three-year product cycle—are significant. Jorgenson noted that the contribution of IT to growth from 1995 to 1999 was about 1.3 percent; by comparison, the annual growth of the U.S. economy over the same period was about 4 percent. A third of that is attributed to IT, meaning that 7 percent of the economy accounted for about a third of its economic growth. This is evidence, Jorgenson concluded, that the behavior of prices of IT, and the behavior of prices of semiconductors in particular, are of “momentous” importance to the economy.²¹

Given its importance, how can economists better predict semiconductor price behavior? Participants at the conference highlighted the high sunk costs, steep learning curves, and rapid product cycles found in the semiconductor industry as factors affecting the industry’s cyclicality. To predict price behavior, a successful industry model would have to take the effects of these features into account.

• High Sunk Costs: Sunk costs are costs already incurred that cannot be recovered regardless of future events. In his conference presentation, Minjae Song of Harvard University noted that semiconductor firms face significant sunk costs in building and upgrading of new fabrication plants (often called “fabs”) where a midsized fab today costs at least $1.5 billion to $2 billion to build. In addition, very large research and development (R&D) investments are required to enter this industry—typically as much as 10 to 15 percent of annual sales—with the R&D often specific to a particular market segment.²²

• Steep Learning Curves: Learning curves in semiconductor production are steep—approximately 70 percent. This means that a doubling of output drops unit costs by about 30 percent. In the semiconductor industry, however, these economies are not generated so much by greater labor productivity as by incremental changes to the automated technology. As Kenneth Flamm noted, improvements over the lifetime of a product’s production come from more efficient die shrinks, which increase the chip density of a silicon wafer, and from yield learning, where the number of good chips on a wafer

increases over time as a percentage of the total number of chips that are manufactured.²³

• Rapid Product Cycles: The semiconductor industry is distinctive in its continuous and rapid introduction of new generations of products (i.e., chips) and the dramatic difference in performance from one generation of product to the next. It is also characterized by very large R&D investments—typically as much as 10 to 15 percent of annual sales—with this R&D often specific to the segment of the market that the firm is entering. Over the past 10 years, the industry has produced five to six generations of semiconductors. When a firm puts a frontier product on the market, existing products become non-frontier. For example, when both the Pentium 2 and the Pentium 3 processors were on the market, the Pentium 3 was at the market frontier. Pentium 3 subsequently became the non-frontier product with the introduction of the Pentium 4 processor. According to Minjae Song, this rapid product cycling has meant that stocks of the current frontier product can quickly lose value with the introduction of the next-generation product.²⁴

These features, taken together, affect the semiconductor industry’s cyclicality. Conference participants described a variety of pathways in this regard:

• Drawing Down Inventories: Fast technological change in the semiconductor industry means that a semiconductor firm cannot reserve inventories as a way of smoothing out demand fluctuations if it hopes to remain competitive. Instead, given the short lifetimes of semiconductor products, firms expect that their inventories will lose value, even become obsolete, if held for too long. Considering the need to recoup high sunk costs, semiconductor firms face strong incentives to sell existing stocks of products as quickly as they can. This need to draw down inventories rapidly is thought to contribute to more pronounced industry cycles.²⁵

• Excess Capacity: Attempts to capture the economies of the learning curve can also exacerbate the industry cycle. While, as noted above, the learning economies related to more efficient die shrinks and yield learning help cut costs, the hidden added capacity that results can also contribute to a chip

glut. Faced with excess supply, firms may have to close older fabrication facilities and/or lower prices.²⁶ These measures can add to the cyclicality of the industry.

• Time to Build: Finally, semiconductor fabrication plants take time to build— typically up to 2 years—and lag times between spikes in demand and sale can also play a significant role in the industry’s cyclicality. Unanticipated surges in chip demand may be prompted by shocks such as those related to the mid-1990s boom in the PC market, the subsequent popularity of the Internet, and the rapid expansion (and later collapse) of the wireless communications market. Given that time is needed to build new manufacturing capacity, however, it is possible that demand fades just as the new capacity to meet this anticipated demand comes on stream. These lags between demand and supply, thus, can exacerbate cyclicality in the market for semiconductors.

In all, as David Morgenthaler of Morgenthaler Ventures observed at the conference, technological developments that decrease the cost per function and subsequently expand the depth and diversity of the market do not seem to translate into smoother industry cycles.

Models of the semiconductor industry that reflect its characteristic cyclicality can be a useful tool to predict semiconductor price behavior. In his conference presentation, Ariel Pakes of Harvard University described a modeling strategy that he has developed that he said can capture key features of complex and dynamic industries.²⁷ This model is based on “primitives” that determine each firm’s profits conditional on the qualities of the products marketed, the costs of production, and the prices charged by all firms. This model could then be extended to include additional features of the specific industry being studied. Participants at the conference then examined the Pakes model to see if it could capture the salient features of the semiconductor industry.

A simple, static version of the Pakes model consists of a demand system, cost functions for each producer in the model, and an equilibrium assumption to solve reasonable pricing and quantity-setting decisions. Profits for each firm could then be calculated based on the price, the quality of each product sold, and the firm’s cost function. The hope is that this type of model could be further extended to

consider some dynamic investment decisions that result from those profit estimates and their likely impact on the industry and on consumers.²⁸

The next National Research Council (NRC) conference on the New Economy sought to deconstruct the computer into its components as a way of understanding its sources of growth and to discover how best to measure this growth. To this end, conference participants considered how the Moore’s Law phenomenon of rapidly expanding capabilities applies to the various computer component industries.²⁹

Although Gordon Moore’s initial prediction pertained to changes in the semiconductor capacity, Moore’s Law today more popularly captures the phenomenon of “faster” as well as “cheaper” development across a variety of computer components.³⁰ The conference brought together industrialists from leading computer hardware firms to explain how Moore’s Law applied to their products and described the types of internal measures that industry had developed to track this change.

• Microprocessors: William Seigle of AMD, a microprocessor manufacturer, compared the Am386, introduced by his company in 1991, with the Opteron, introduced in April 2003. Performance, he noted, had jumped 50 times from 33 MHz to 2 GHz, offering significant improvements in the efficiency of instruction processing, memory hierarchy, and branch prediction.³¹

• Hardware Storage: Remarking on the performance improvements in computer storage, Robert Whitmore of Seagate Inc. noted that performance, measured as input/output transactions per second, had accelerated significantly between the late 1980s and late 1990s. Meanwhile, he noted that the price

of rotating magnetic memory on a dollar-per-gigabyte basis had eroded at an annual compound rate of minus 45 percent between 1995 and 2002. In addition, mean time between failures, a measure of reliability, had grown at a phenomenal compound annual rate of 25 percent from 1977 to 2001.³²

• Software Storage Systems: Mark Bregman of Veritas Software (a company that develops software to help store, access, and manage data) noted the apparent observance of Gilder’s Law, which states that the total bandwidth of communication systems triples every 12 months.³³ Further, he noted that storage devices achieve 100 percent growth in density annually, a reality that translates into better cost at a dramatic rate.³⁴

• Graphics: Chris Malachowsky of NVIDIA documented product performance improvements in graphics from the second half of 1997 to the first half of 2003 at an annualized rate of 215 to 229 percent. Rapid technological advances in graphics technology, he noted, rendered moviemaking chores, previously requiring farms of thousands of machines, to be possible using consumer PCs, dramatically lowering prices.³⁵

Several participants at the conference on computers emphasized the need to develop appropriate categories and performance measures to capture the growth of these dynamic and complex industries. The Brookings Institution’s Jack Triplett underscored this point in his conference presentation, emphasizing that economists need to learn more about the contributions of hardware component technologies to the increase in computer performance.³⁶

Dr. Triplett noted that while the cost of computing today is projected to be about one-thousandth of one percent of what it cost 50 years ago, this estimate still does not account for all aspects of computer performance. An exciting

research agenda, he noted, is to account for the determinants of the great decline in computer price/performance over the last 50 years.

This research agenda is very challenging because it has to account for the qualitative changes in computers. How do we measure, for example, the performance of the computer and its components through time? Dr. Triplett acknowledged that the question is complicated by the dynamism and complexity of the technological change characterizing the evolution of the modern computer. To be sure, the cost, capabilities, and size of a 1952 UNIVAC are significantly different from those of a modern laptop. Since direct comparisons of price are not feasible—the proverbial apples and oranges problem—a key challenge for economists is to adjust their price data for quality differences. Indeed, identifying such “true price change” has long been a goal of price statisticians and national accountants.

One way of adjusting prices for quality differences is to use hedonic price indexes. Developed 40 years ago by Zvi Griliches and enhanced since, this econometric method takes into account an array of characteristics possessed by a product and their functional relation to price.³⁷ Many economists regard hedonic price indexes to be a theoretically promising way of adjusting for quality when measuring the price of computing power through time, while recognizing the need for further development.³⁸

In practice, however, the continued dynamism and complexity of the relevant industries will make the task of developing robust measures of computer performance highly challenging. Rapid supply-driven evolution of products and concepts, as well as changing consumer behavior, keeps the industry in flux, rendering the economist’s task more difficult. Swift technological change can change and, in some cases, even make obsolete the relative importance of particular quality characteristics used in hedonic estimates.³⁹ A further prob-

lem arises because, if things change enough, no price methodology will give accurate estimates.⁴⁰

Illustrating this technological dynamism, Dalen Keyes of DuPont Displays noted that the U.S. display industry sees its future in moving away from LCD (liquid crystal display) technologies and towards Organic LED (light-emitting diode) technologies. OLED display technology, based on a roll-to-roll manufacturing concept, integrates components from the flex-circuitry industry with inkjet printing from the graphic arts industry to get rolls of material that could be “sliced and diced” into displays. Flexible and versatile, OLEDs, he predicted, will possess qualities and applications quite different from today’s displays.⁴¹

Tracing the evolution of technology in the printer industry, Howard Taub of Hewlettt-Packard noted that “we are pretty much at a point where the quality of the image that you can print is about as good as you’re going to get.” As a result, he noted, the quest for “better” had gone on to pursue other dimensions including connectivity and ease of use. He also noted that the computer printer industry is looking to create new markets beyond those for office printing and duplication. New printer technologies, he noted, could enable the production of limited-run custom magazines and advertisements, changing the way consumers think about desktop printers.⁴²

Indeed, for displays and printers, as with other computer components, the use of hedonic indexes to control for quality of a product is likely to be a challenge as continuing rapid innovation changes not just the features of the product but even the concept of the product itself.

Another challenge to developing robust hedonic price indexes arises when—as David McQueeney of IBM put it—“faster, better, cheaper,” collides with the “good enough phenomenon.”⁴³ He noted, for example, that many current models of displays and home computers have crossed the “good enough” threshold for most of today’s home computing needs—the point also raised by Dr. Taub, above. Displays used for everyday desktop home-PC applications have become so good, observed Dr. McQueeney, that “further technological improvements aimed at more pixels per inch could not be detectable to the end user.” Similarly, he noted that disk capacity has become so large that most ordinary users never fill the hard

drive in the 2 or 3 years they normally keep a computer. So, although research and development do not stop at a certain point, their benefits may begin to show up in price reduction and cost-performance reduction rather than in performance measures. “The raw capabilities of technology have in some cases gotten to the point where either the economics of how you sell them and how you ascribe value to them is changing,” he explained, “or you are forced to look elsewhere in the system performance stack to get real improvements.”⁴⁴

Dr. McQueeney also noted that the value of “faster and better” might remain unrealized pending additional developments in technology and finance. Looking ahead to the conference on the Telecommunications Challenge, he observed that there is at present enough fiber capacity to “connect every person in North America to every person in Eastern and Western Europe and to allow all to have a phone conversation at the same time.” He also noted that a tremendous capacity in optical fiber has been installed between various cities and within metropolitan areas of the United States. Yet, “the intelligence needed to light up those fiberoptic networks and make them actually do something useful—the servers, the routers, the switches—is in fact quite expensive,” he stated, “and we’re still struggling with a good investment model that will let us build out that control infrastructure to use the fiber capacity we have.”⁴⁵ This need to realize necessary complementarities was also echoed by Dr. Siegle, who noted that “while micro-processors are important, you can’t make meaningful systems and applications if there are advances in just the microprocessor.”⁴⁶

These conceptual challenges to measuring performance aside, industry experts at the conference described a variety of formal and informal measures currently used by computer component industries to gauge performance. Dr. Whitmore noted that for the hardware storage industry, capacity in bytes, price, performance, and reliability remain the main factors for measurement, although additional metrics are appearing on the horizon. In the printer industry, “faster and better” is measured in terms of printer speed, resolution, reliability, and usability, according to Dr. Taub. Mr. Malachowsky noted that there is a marketing view of performance in addition to internal and external views in the graphics industry. He noted that his company, NVIDIA, measures itself internally on “very engineering-specific, design-specific things” such as bandwidth utilization factors and externally according to particular application benchmarks. Echoing the common theme, Dr. Keyes noted that the display industry relies on an extensive list of technical specifications, including diagonal size of the display, pixel count, and power consumption. Performance measures include luminants

and switching speed, which indicates whether a product could do video-grade displays. Another metric, he added, is the size of the substrate used in the manufacture displays.

Citing these and other performance measures made note of by the participants, Dr. Jorgenson concluded that measuring progress in the computer and computer component industries is not only possible, but that such measurement is increasingly more sophisticated and, in fact, “quite successful.” He recalled that a set of measures for computers and peripherals begun in the late 1960s—grounded in economics research at IBM—achieved incorporation into the U.S. national accounts for the first time in the mid-1980s. These have continued to be in use (while also being enhanced and developed) to the present day. He expressed optimism that similar progress on data measurement and analysis can be made based on what he had heard at this conference—and that this could help improve the economic understanding needed to develop the policies necessary to sustain the New Economy.⁴⁷

Within the U.S. national accounts, software is broken down into three categories: prepackaged, custom, and own-account software. Prepackaged (or shrinkwrapped) software is packaged, mass-produced software. It is available off-the-shelf, though increasingly replaced by on-line sales and downloads over the Internet. In 2003, BEA placed business purchases of prepackaged software at around $50 billion. Custom software refers to large software systems that perform business functions such as database management, human resource management, and cost accounting.⁴⁸ In 2003, BEA estimated business purchases of custom software at almost $60 billion. Finally, own-account software refers to software systems built for a unique purpose, generally a large project such as an airlines reservation system. In 2003, BEA estimated business purchases of own-account software at about $75 billion.⁴⁹

Dr. Jorgenson, in introducing the New Economy conference on Software, noted that while there is sufficient price information on prepackaged software, this category is only thought to make up about 25 to 30 percent of the software

market.⁵⁰ Consequently, he noted, “there is a large gap in our understanding of the New Economy.”⁵¹

Before we can develop appropriate measures of software performance, we first need to understand the nature of software itself. As William Raduchel of the Ruckus Network explained at the conference on software, software comprises millions of lines of code, operated within a stack.⁵² The stack begins with the kernel, which is a small piece of code that talks to and manages the hardware. The kernel is usually included in the operating system, which provides the basic services and to which all programs are written. Above this operating system is middleware, which “hides” both the operating system and the window manager. For the case of desktop computers, for example, the operating system runs other small programs called services as well as specific applications such as Microsoft Word and PowerPoint.

Thus, when a desktop computer functions, the entire stack is in operation. This means that the value of any part of a software stack depends on how it operates within the context of the rest of the stack.⁵³ The result, as Monica Lam of Stanford University suggested, is that software may be the most intricate thing that humans have learned to build. Software grows more complex as more and more lines of code accrue to the stack, making software engineering much more difficult than other fields of engineering.⁵⁴

The way software is written also adds to its complexity and cost. As Anthony Scott of General Motors pointed out, the process by which corporations build software is “somewhat analogous to the Winchester Mystery House,” where accretions to the stack over time create a complex maze that is difficult to fix or change.⁵⁵

This complexity means that a failure manifest in one piece of software, when added to the stack, may not indicate that something is wrong with that piece of software per se, but quite possibly can cause the failure of some other piece of the stack that is being tested for the first time in conjunction with the new addition.⁵⁶ In short, the complexity of software makes measuring software performance very challenging.

The unique nature of software also poses challenges for national accountants who are interested in data that track software costs and aggregate investment in software and its impact on the economy. This is important because over the past 5 years, investment in software has been about 1.8 times as large as private fixed investment in computers’ peripheral equipment and was about one-fifth of all private fixed investment in equipment and software.⁵⁷ Getting a good measure of this asset, however, is difficult because of the unique characteristics of software development and marketing, as well as the conventions by which it is reported.

According to Shelly Luisi of the Securities and Exchange Commission (SEC), some data about software come from information that companies report to the SEC.⁵⁸ These companies follow the accounting standards developed by the Financial Accounting Standards Board (FASB).⁵⁹ Luisi noted that the FASB developed these accounting standards with the investor, and not a national accountant, in mind. As a result of these accounting standards, she noted, software is included as property, plant, and equipment in most financial statements rather than as an intangible asset.⁶⁰

Given these accounting standards, how do software companies actually recognize and report their revenue? Taking the perspective of a software company, Greg Beams of Ernst & Young noted that while sales of prepackaged software are generally reported at the time of sale, more complex software systems require recurring maintenance to fix bugs and to install upgrades, causing revenue reporting to become more complicated. In light of these multiple deliverables, software companies come up against rules requiring that they allocate value to each of those deliverables and then recognize revenue in accordance with the requirements for those deliverables. How this is put into practice results in a wide difference in when and how much revenue is recognized by the software company, he noted—making it, in turn, difficult to understand the revenue numbers that a particular software firm is reporting.⁶¹

Mr. Beams noted that information published in software vendors’ financial statements is useful mainly to the shareholder. He acknowledged that detail is often lacking in these reports, and that distinguishing one software company’s reporting from another and aggregating such information so that it tells a meaningful story can be extremely challenging.

Although the computer entered into commercial use some four decades earlier, the Bureau of Economic Analysis has recognized software as a capital investment (rather than as an intermediate expense) only since 1999. Describing BEA methodology, David Wasshausen of BEA noted that his organization uses a “commodity flow” technique to measure prepackaged and custom software. Beginning with total receipts, BEA adds imports and subtracts exports, which leaves the total available domestic supply. From that figure, BEA subtracts household and government purchases to come up with an estimate for aggregate business investment in software.⁶² By contrast, BEA calculates own-account software

as the sum of production costs, including compensation for programmers and systems analysts and such intermediate inputs as overhead, electricity, rent, and office space.⁶³

According to Dr. Wasshausen, BEA is striving to improve the quality of its estimates. While BEA currently bases its estimates for prepackaged and custom software on trended earning data from corporate reports to the SEC, it hoped to benefit soon from Census Bureau data that capture receipts from both prepackaged and custom software companies through quarterly surveys. Among recent BEA improvements, Dr. Wasshausen cited an expansion of the definitions of prepackaged and custom software imports and exports, and better estimates of how much of the total prepackaged and custom software purchased in the United States was for intermediate consumption. BEA, he said, was also looking forward to an improved Capital Expenditure Survey by the Census Bureau.⁶⁴

Dirk Pilat of the Organisation for Economic Co-operation and Development (OECD) noted at the same conference that methods for estimating software investment have been inconsistent across the countries of the OECD.⁶⁵ One problem contributing to the variation in measures of software investment is that the computer services industry represents a heterogeneous range of activities, including not only software production, but also such things as consulting services. National accountants have had differing methodological approaches (for example, on criteria determining what should be capitalized) leading to differences between survey data on software investment and official measures of software investments as they show up in national accounts.

Attempting to mend this disarray, Dr. Pilat noted that the OECD Eurostat Task Force has published its recommendations on the use of the commodity flow model and on how to treat own-account software in different countries.⁶⁶ He noted that steps were under way in OECD countries to harmonize statistical

practices and that the OECD would monitor the implementation of the Task Force recommendations. This effort would then make international comparisons possible, resulting in an improvement in our ability to ascertain what was moving where—the “missing link” in addressing the issue of offshore software production.

Despite the comprehensive improvements in the measurement of software undertaken since 1999, Dr. Wasshausen noted that accurate software measurement continued to pose severe challenges for national accountants simply because software is such a rapidly changing field. He noted, in this regard, the rise of demand computing, open-source code development and overseas outsourcing, which create new concepts, categories, and measurement challenges.⁶⁷ Characterizing attempts made so far to deal with the issue of measuring the New Economy as “piecemeal”—“we are trying to get the best price index for software, the best price index for hardware, the best price index for LAN equipment routers, switches, and hubs”—he suggested that a single comprehensive measure might better capture the value of hardware, software, and communications equipment in the national accounts. Indeed, information technology may best be thought of as a “package,” combining hardware, software, and business-service applications.⁶⁸

A further challenge in the economics of software lies in tracking price changes. Drawing on Microsoft Corporation data, Alan White of Analysis Group and Ernst Berndt of MIT presented their work on estimating price changes for prepackaged software.⁶⁹ Dr. White noted that an investigator faces several important challenges in constructing measures of price and price change. These include ascertaining which price to measure because software products may be sold as full versions or as upgrades, stand-alones, or suites. An investigator has also to determine what the unit of output is, how many licenses there are, and when price is actually being measured. Another key issue, he added, concerns how the quality of software has changed over time and how that should be incorporated into price measures.⁷⁰

Surveying the types of quality changes that might come into consideration, Dr. Berndt gave the example of improved graphical interface and “plug-‘n-play,” as well as increased connectivity between difference components of a software suite.⁷¹ Referring to their study, Dr. Berndt noted that he and Dr. White compared the average price level (computing the price per operating system as a simple average) with quality-adjusted prices levels using hedonic and matched-model econometric techniques. They found that while the average price, which does not correct for quality changes, showed a growth rate of about 1 percent a year, the quality-adjusted matched model showed a price decline of around 6 percent a year and the hedonic calculation showed a much larger price decline of around 16 percent.

These quality-adjusted price declines for software operating systems, shown in Figure 5, support the general thesis that improved and cheaper information technologies contributed to greater information technology adoption leading to productivity improvements characteristic of the New Economy.⁷²

How do new information and communications technologies translate into prices and hence consumer welfare? Mark Doms of the Federal Reserve Bank of San Francisco provided the participants in the STEP conference on the Tele-

FIGURE 5 Quality-adjusted prices for operating systems have fallen, 1987-2000.

SOURCE: Jaison R. Abel, Ernst R. Berndt, Cory W. Monroe, and Alan White, “Hedonic Price Indexes for Operating Systems and Productivity Suite PC Software,” draft working paper, 2004.

communications Challenge an overview of what the current official numbers say, and the challenges of coming up with good price indexes for communications equipment and services. He noted that while investment in communications in the United States had been substantial—around $100 billion per year, representing a little over 10 percent of total equipment investment in the U.S. economy—it had also been highly volatile. During the recession of the early 2000s, he noted, IT investment fell about 35 percent from peak to trough (see Figure 6⁷³). Dr. Doms noted that this recession might well be remembered as the high-tech recession, adding that “certainly what happened to communications played a major role in what happened to the high-tech sector.”

Measuring the dollars spent on communications in the United States every year is difficult because technology is rapidly changing. As we noted earlier, a computer costing a thousand dollars today is a lot more powerful and versatile than a similarly priced one of 10 years ago—and this improvement is no less true for communications equipment. Similarly, most long-distance communications 25 years ago was handled through landline phones, in stark contrast to the diver-

FIGURE 6 Annual percent change in IT investment.

SOURCE: Bureau of Economic Analysis.

NOTE: Percent changes based on year-end values.

sity of means of communications in use today. The technology is also in rapid flux. Dr. Doms noted that between 1996 and 2001 alone, there were tremendous advances in the amount of information that could travel down a strand of glass fiber, adding that the price of gear used to transmit information over fiber fell, on average, by 14.9 percent a year over this five-year period.

The fast speed of technological change renders the job of tracking prices (which enables us to see how much better off society is as a result of technological changes) a complex one. Whereas money spent on telecommunications was relatively easier to track 25 years ago when most purchases were of telephone switches, today’s telecommunications equipment includes a wide array of technologies related to data, computer networking, and fiber optics.

Current methodologies for making inter-temporal comparisons in price and quality understate true price declines because they do not fully track these technological changes. While BEA has estimated that prices for communications gear fell an average of 3.2 percent per year between 1994 and 2000—in sharp

contrast to the 19.3 percent fall in computer prices—Dr. Doms noted that more a complete estimate that he had developed shows that communications equipment prices actually fell on the order of 8 to 10 percent over that period.⁷⁴

While this new estimate is a step in the right direction, Dr. Doms acknowledged that more refinement is necessary in measuring telecom prices. Echoing a refrain heard at each of the conferences in the series on Measuring and Sustaining the New Economy, he noted that the job of keeping track of rapid developments in information and communications technologies was growing increasingly difficult for statistical agencies, especially in light of their limited budgets and the rapid development of technology. “Unless the statistical agencies get increased funding, in the future, they are not going to be able to follow new, evolving trends very well,” he concluded.

The second theme of the NRC conferences on the New Economy concerned public polices needed to sustain the New Economy. A major focus of these conferences was on polices to sustain Moore’s Law, the driver of faster and more widely affordable computers and other productivity-enhancing technologies. Participants at the conferences on software and telecommunications also examined the new challenges in globalization emerging from the possibility of sending voice and data at very low costs around the world.

To be sure, the challenge of measuring the New Economy and policies needed to sustain the benefits of the New Economy are two sides of one coin. Better data on what is moving where in offshoring are likely to permit more informed policy debate.

As noted at the outset, Moore’s Law is not a deterministic law but a self-fulfilling prophecy that needs to be sustained if the economy is to continue to benefit from the advantages of faster and cheaper information technologies.⁷⁵ Moore’s Law works by setting expectations about the pace of competition in the semiconductor industry. Each firm, believing its rivals to develop and market

a faster and cheaper product within the 18-month timeframe, steps up its own work—leading, overall, to the faster pace at which new semiconductor products are brought to market. Upholding Moore’s Law, thus, requires keeping up the belief among industry participants that this pace of “faster and cheaper” is sustainable. Continuing this virtuous cycle of expectations requires that each firm in the industry believes that impediments to continuing technological advance can be overcome well in time.

While Moore’s Law is currently forecast to hold for the next 10 to 15 years (not least by Gordon Moore himself⁷⁶), there remain potential technological showstoppers down the road. In the case of CMOS (complementary metal-oxide semiconductor) technology, as explained by Bob Doering at the conference on semiconductors, tunneling problems could arise when a gate insulator gets so thin that it loses its insulating capacity and becomes a new leakage path through the transistor.⁷⁷ This current flow is dominated by quantum mechanical tunneling of electrons through the barrier.⁷⁸

While Dr. Doering noted that continued advances in CMOS device scaling are expected to continue for another 10 to 15 years, Randall Isaac of IBM, also speaking at the same conference, was more pessimistic, observing that progress from scaling could tail off more rapidly.⁷⁹ He noted that the surge in performance, achieved through deep ultraviolet (UV) technologies, is likely not to be sustainable over a long period. He also warned that extreme ultraviolet lithography (EUV), often cited as the next emerging technology, might not prove to be as pervasive as its predecessor has been.

Such technological brick walls apply not only to semiconductors but more broadly to computer components as well. For example, Kenneth Walker, of Philips Electronics, noted at the conference on Deconstructing the Computer that while DVD and CD readers had become standard on personal computers, we are starting to reach certain limits in these devices.⁸⁰ Current top-of-the-line CD devices, he noted, rate at 48X to 52X—the equivalent of spinning at about 200 kilometers per

hour. This speed approaches the reigning physical limit for CDs, since operating at higher speeds would cause the disc to shred within the device.

Dr. Walker noted that human ingenuity would extend the scope and pace of improvements for hard disks over the near future. The next generation of improvements, he noted, may be realized not by spinning DVDs faster, but by adopting blue lasers to replace red lasers. Since blue lasers are more focused, more information can be stored on a single disk. Newly discovered ways of writing and rewriting information on disks will also enhance the device’s functionality, he predicted—although these innovations postpone but do not eliminate a reckoning with the brick wall.

In addition to technological impediments, participants at the conference on Productivity and Cyclicality in Semiconductors also reviewed a variety of resource challenges that may jeopardize Moore’s Law. These are summarized below.

• High Costs of Manufacturing: Could the high costs of technical advance be the Achilles heel of the New Economy? Dr. Doering noted that progress on CMOS technology could slow, not because engineers run out of ways to make smaller or faster chips, but because the costs of manufacturing could outstrip the advantages of such miniaturization. Referring to EUV technology, Dr. Isaac noted that at $40 million to $50 million per tool, the economic challenges of investing in such equipment are daunting.

  Dr. Isaac added that the real “fly in the ointment” to computing that is faster and cheaper might well be the cost of power. As engineers place more components closer together, power consumption and heat generation become systemic problems.⁸¹ Though few technologists or economists have factored the cost of power for computing, the energy consumption of server farms is increasing exponentially. To convey a sense of scale, he noted that a server farm uses more watts per square foot than a semiconductor or automobile manufacturing plant.

FIGURE 7 Electrical engineering graduates: bachelor’s degrees earned, 1975-2000.

SOURCE: National Science Foundation, Science and Engineering Indicators 2000, 1975-1987 Engineering Workforce Commission.

• Workforce Issues: Sustaining Moore’s Law will require creativeness and ingenuity in overcoming these technological and economic challenges. George Scalise pointed out that this requires a trained workforce well grounded in the disciplines—such as physics, mathematics, and engineering—that underpin research and manufacturing in the semiconductor industry.⁸² Given this need, he listed some recent trends that appear troubling, including:

  • recent evaluations that place American K-12 students below their foreign peers in mathematics and science;⁸³ and

  • a decline in the number of bachelor’s degrees in electrical engineering awarded in the United States (see Figure 7). While this decline (of about 40 percent over the last several years) seems to have recently flattened out, he said that this trend remains a source of concern.⁸⁴

• Funding for Research: Bob Doering and George Scalise, along with Clark McFadden of Dewey Ballantine LLP, noted that declines in federal R&D funding makes it harder for the semiconductor industry to overcome loom-

ing technological challenges.⁸⁵ They added that the semiconductor industry’s ability to do its own long-term research has diminished with the demise of the large industrial laboratory.⁸⁶ As noted below, several participants called for additional federal investments in research to help maintain the innovative pace of the semiconductor industry.

Changes in the structure of the semiconductor industry may impact the competitive environment associated with Moore’s Law.⁸⁷ Kenneth Flamm noted at the conference on Productivity and Cyclicality in Semiconductors that the growth of the foundry model in semiconductor manufacture might have implications for industry-sponsored research, given that foundry-based companies (often called fabs) often spend a smaller percentage of their sales on R&D than do traditional integrated device manufacturers.⁸⁸ As George Scalise further noted, fabs have also affected the competitive environment by creating a surge in manufacturing capacity. This surge has led to price attrition beyond levels against which many traditional firms that integrate design and manufacture can compete successfully.

According to George Scalise and Kenneth Flamm, these developments highlight the importance of sustaining a variety of cooperative efforts to strengthen the

research base and to propel advance in semiconductor platform technologies.⁸⁹ Positive examples of such cooperative partnerships highlighted at the conference on Productivity and Cyclicality in Semiconductors were:

• The Semiconductor Research Corporation (SRC), whose mission is to provide low-overhead generic semiconductor research and related programs that meet the needs of the semiconductor industry for technology and relevantly educated talent. It currently disburses approximately $40 million per year on directed research carried out in universities by 800 to 900 graduate students worldwide.

• International SEMATECH, a global research consortium, whose role is to develop new manufacturing technologies and methods and transfer them to its member companies, which in turn manufacture and sell improved chips. Member companies cooperate pre-competitively in key areas of semiconductor technology, sharing expenses and risk. Their common aim is to accelerate development of the advanced manufacturing technologies needed to build future generations of semiconductors.

• The Focus Center Research Program, which sponsors a multi-university effort to address major basic research challenges.⁹⁰ This includes the design and test program led by the University of California at Berkeley, the interconnect team led by the Georgia Institute of Technology, the circuit systems and software team led by Carnegie Mellon University, and a materials and devices team led by the Massachusetts Institute of Technology. Each program has seven to eight partners, and funding for the four-year program, which now totals $22 million a year, is expected to grow to $60 million a year over the next few years.

Technology roadmaps are another important mechanism for sustaining Moore’s Law. Providing a graphical portrayal of the structural relationships among science, technology, and applications over a period, a technology roadmap is a tool for firms in an industry to identify potential technical showstoppers and cooperate

in developing (at a pre-competitive level) solutions to these technical challenges. Roadmap strategy areas include technology and product marketing, identifying gaps in R&D programs, and identifying obstacles to rapid and low-cost product development. Moreover, as companies believe that competitive success lies in staying ahead of the Roadmap, the existence of a published Roadmap itself enhances the pace of competition and, hence, the robustness of Moore’s Law.

At the conference on Productivity and Cyclicality in Semiconductors, Kenneth Flamm noted that although the international semiconductor roadmap is often described as a descriptive or predictive process, its role is to coordinate a complex technology with different pieces and multiple suppliers.⁹¹ “What you really have is people identifying potential showstoppers and trying to mobilize people at choke points.” Clark McFadden added that the roadmap is not a “solution” to technological problems but rather a description of various options, challenges, and gaps in charting the future course of a technology. The role of the roadmap, he said, is to communicate information about these options, challenges, and gaps to the industry in a way that suppliers, manufacturers, and customers can appreciate and use.

There are of course limits to the usefulness of roadmaps. As roadmap pioneers William Spencer and T. E. Seidel have acknowledged, roadmaps are expensive and time consuming to develop and are, by definition, out of date as soon as they are written.⁹² As they note, however:

Crediting the Semiconductor Roadmap for the speed of the information technology industry’s recent advance at the conference on Deconstructing the Computer, William Siegle offered two reasons why the road-mapping process is linked to accelerations in the decline of logic cost. First, he noted that making

meaningful improvements in capability requires the coordination of many different pieces of technology, and the Semiconductor Roadmap has made very visible both what those pieces are and what advances are required in different sectors of the industry to achieve that coordination. Second, he noted, as companies believe that success lies in staying ahead of the Roadmap, the existence of a published Roadmap enhances the pace of competition.

In these ways, Roadmaps can help sustain the momentum of “faster, better, cheaper” in industries that produce computer components. While welcoming the development of roadmaps for the different computer component industries—such as that recently published by the U.S. Display Consortium⁹⁴—Dale Jorgenson cautioned that successful models, such as the semiconductor industry roadmap, must be adapted to the operational exigencies of the computer component industry in question.

The next conference in the New Economy series examined the importance of software in the New Economy and the vulnerability of the U.S. economy to software failures and attacks. Software is an encapsulation of knowledge in an executable form that allows for its repeated and automatic applications to new inputs.⁹⁵ It is the means by which we interact with the hardware underpinning information and communications technologies.

The U.S. economy, today, is highly dependent on software, with businesses, public utilities, and consumers among those integrated within complex software systems. Participants at the NRC Conference on Software, Growth, and the Future of the U.S. Economy, examined how this dependence exposes the economy to vulnerabilities in the production and execution of software—major concerns in sustaining the New Economy.

Almost every aspect of a modern corporation’s operations is embodied in software. Anthony Scott of General Motors noted that a company’s software embodies a whole corporation’s knowledge into business process and methods, adding that “virtually everything we do at General Motors has been reduced in some fashion or another to software.”⁹⁶

In addition, much of our public infrastructure relies on the effective operation of software, with this dependency also leading to significant vulnerabilities. As Dr. Raduchel observed, it seems that the failure of one line of code, buried in an energy management system from General Electric, was the initial source

leading to the electrical blackout of August 2003 that paralyzed much of the northeastern and midwestern United States.⁹⁷ Smaller, everyday failures are no less expensive; according to the National Institute for Standards and Technology (NIST), national annual costs of software failures lie in the range of $22.2 billion to $59.5 billion.⁹⁸

Despite the pervasive use of software, and partly because of the relative youth of the science of computer engineering, understanding the economics of software presents an extraordinary challenge. Many of the challenges relate to measurement, econometrics, and industry structure. Here, the rapidly evolving concepts and functions of software as well as its high complexity and context-dependent value make measuring software difficult. This frustrates our understanding of the economics of software—both generally and from the standpoint of action and impact—and impedes both policymaking and the potential for recognizing technical progress in the field.

Given that the infrastructure of the New Economy is based on software, participants at the conference on software considered the vulnerability of this infrastructure and policies that can strengthen this infrastructure.

Software grows more complex as more and more lines of code accrue to the stack, making software engineering much more difficult than other fields of engineering, according to Monica Lam of Stanford University.⁹⁹ This complexity means that the failure of any given piece of software, when added to the stack, may not indicate that something is wrong with that piece of software per se, but quite possibly a failure of some other piece of the stack that is being tested for the first time in conjunction with the new addition. This complexity of software makes it inherently error prone as well as vulnerable to attack.

Indeed, attacks against that code—in the form of both network intrusions and infection attempts—have grown substantially over the past decade, according to Kenneth Walker of Sonic Wall.¹⁰⁰ (See Figure 8.¹⁰¹) The perniciousness of the attacks is also on the rise. The Mydoom attack of January 28, 2004, for example, did more than infect individuals’ computers producing acute but short-lived inconvenience. It also reset the machine’s settings leaving ports and doorways open to future attacks.

The economic impact of such attacks is increasingly significant. According to Kenneth Walker of Sonic Wall, Mydoom and its variants infected up to half a million computers. The direct impact of the worm includes lost productivity owing to workers’ inability to access their machines, estimated at between $500 and $1,000 per machine, and the cost of technician time to fix the damage. According to one estimate cited by Mr. Walker, Mydoom’s global impact by February 1, 2004,

FIGURE 8 Growing attacks against code.

SOURCE: Analysis by Symantec Security Response using data from Symantec, IDC, and ICSA.

alone was $38.5 billion.¹⁰² He added that the E-Commerce Times had estimated the global impact of worms and viruses in 2003 to be over one trillion dollars.

Acknowledging that software will never be error free and fully secure from attack or failure, Dr. Lam suggested that the real question is not whether these vulnerabilities can be eliminated, raising instead the issue of the role of incentives facing software makers to develop software that is more reliable.

One factor affecting software reliability is the nature of market demand for software. Some consumers—those in the market for mass client software, for example—may look to snap up the latest product or upgrade and feature addons, placing less emphasis on reliability. By contrast, more reliable products can typically be found in markets where consumers are more discerning, such as in the market for servers.

Software reliability is also affected by the relative ease or difficulty in creating and using metrics to gauge quality. Maintaining valid metrics can be highly challenging given the rapidly evolving and technically complex nature of software. In practice, software engineers often rely on measurements of highly indirect surrogates for quality (relating to such variables as teams, people, organizations, processes) as well as crude size measures (such as lines of code and raw defect counts.)

Other factors that can affect software reliability include the current state of liability law and the unexpected and rapid development of a computer hacker cul-

ture, which has significantly raised the complexity of software and the threshold of software reliability. While for these and other reasons it is not realistic to expect a 100 percent correct program, Dr. Lam noted that the costs and consequences of this unreliability are often passed on to the consumer.

Addressing this issue, Hal Varian noted that open-source software—which, in general, is software whose source code is freely available for use or modification by users and developers—is one way of improving the reliability of software while introducing plural sources of innovation.¹⁰³ It is different from proprietary software whose makers do not make the source code available to the public. While developing open-source software provides a public good that is predicted to be under-provisioned in standard economic theory, software developers in the real world have many motivations for writing open-source software, noted Dr. Varian, including (at the margin) scratching a creative itch and demonstrating skill to one’s peers. Indeed, while ideology and altruism provide some of the motivation, many firms, including IBM, make major investments in Linux and other open-source projects for solid market reasons.

While the popular idea of a distributed model of open-source development is one where spontaneous contributions from around the world are merged into a functioning product, most successful distributed open-source developments take place within preestablished or highly precedented architectures. It should thus not come as a surprise that open-source has proven to be a significant and successful way of creating robust software. Linux provides a major instance where both a powerful standard and a working reference for implementation have appeared at the same time, noted Dr. Varian. Major companies, including Amazon.com and Google, have chosen Linux as the kernel for their software systems. Based on this kernel, these companies customize software applications to meet their particular business needs.

Indeed, software is most valuable when it can be combined, recombined, and built upon to produce a secure base upon which additional applications can in turn be built. The policy challenge, observed Dr. Varian, lies in ensuring the existence of incentives that sufficiently motivate individuals to develop robust basic software components through open-source coordination, while ensuring that, once they are built, they will be widely available at low cost so that future development is stimulated.

Another major and topical issue concerning software and the New Economy concerns the increasingly globalized labor market for software production. Participants at the NRC conference on software discussed the economic forces that

are driving this trend, and its implications for sustaining the United States’ longstanding advantage in science, research, and innovation.

How is software made and who makes it? Dr. Lam described the software development process as one comprising various iterative stages.¹⁰⁴ After getting an idea of the requirements, software engineers develop the needed architecture and algorithms. Once this high-level design is established, focus shifts to coding and testing the software. She noted that those who can write software at the kernel level are a very limited group, perhaps numbering only in the hundreds worldwide. This division of labor in software production, she said, reflects a larger qualitative difference among software developers, where the very best software developers are orders of magnitude—up to 20 to 100 times—better than the average software developer. This means that a surprisingly small number of people do a disproportionate amount of the field’s creative work.¹⁰⁵

Dr. Raduchel added that as a rule of thumb, producing software calls for a ratio of 1 designer to 10 coders to 100 testers.¹⁰⁶ Configuring, testing, and tuning the software account for 95 to 99 percent of the cost of all software in operation. These non-linear complementarities in the production of software, he said, mean that simply adding workers to one part of the production process is not likely to make a software project finish faster. Further, since a majority of time in developing a software program deals with handling exceptions and in fixing bugs, it is often hard to estimate software development time.

This skew of aptitude in the software labor market means that high-end software firms must look globally to find needed talent, according to Wayne Rosing of Google.¹⁰⁷ Google, he noted, is highly selective. It hired only about 300 new workers in 2003 out of an initial pool of 35,000 resumes submitted from all over the world. While he attributed this high response to Google’s reputation as a good place to work, Google in turn looked for applicants with high “raw intelligence,” strong computer algorithm skills and engineering skills, and a high degree of self-motivation and self-management needed to fit in with Google’s corporate culture.

Google’s outstanding problem, Dr. Rosing lamented, was that “there aren’t enough good people” available to do this high level of work. Too few qualified computer science graduates were coming out of American schools, he said. While the United States remained one of the world’s top areas for computer science

education and produced very good graduates, there are not enough people graduating at the master’s or doctoral level to satisfy the needs of the U.S. economy, especially for innovative firms such as Google.

In addition, recent U.S. visa restrictions mean that Google must hire the engineers it needs outside the country, said Dr. Rosing. Noting that the government in 2004 capped the H-1B quota at 65,000, down from approximately 225,000 in previous years, he said that Google was not able to hire foreign students who were educated in the United States, but who could not stay on and work for lack of a visa. Dr. Rosing said that such policies limited the growth of companies like Google within the nation’s borders—something, he said, that did not seem to make policy sense.

While Dr. Rosing highlighted that the search for talent leads firms like Google to look abroad, Jack Harding of eSilicon noted that manufacturing complexity and business efficiency are often the main drivers of offshore out-sourcing.¹⁰⁸ Speaking at the conference on software, Mr. Harding noted that as the manufacturing technology grows more complex, a firm is forced to stay ahead of the efficiency curve through large recapitalization investments or to “step aside and let somebody else do that part of the work.” This decision to move from captive production to outsourced production, he said, can then lead to offshore-outsourcing—or “offshoring”—when a company locates a cheaper supplier in another country of same or better quality.

Displaying an outsourcing-offshoring matrix (Figure 9), Mr. Harding noted that it was the actually the “Captive-Offshoring” quadrant, where American firms like Google or Oracle open research and production facilities overseas, that is the locus of a lot of the current “political pushback” about being “un-American” to take jobs abroad. Activity that could be placed in the “Outsource-Offshore” box, meanwhile, was marked by a trade-off where diminished corporate control had to be weighed against very low variable costs with adequate technical expertise.

Saving money by outsourcing production offshore not only provides a compelling business motive, it has rapidly become “best practice” for new companies. Though there might be exceptions to the rule, Mr. Harding noted that a software company seeking venture money in Silicon Valley that did not have a plan to base a development team in India would very likely be disqualified. It would not be seen as competitive if its intention was to hire workers at $125,000 a year in Silicon Valley when comparable workers were available for $25,000 a year in Bangalore. Heeding this logic, almost every software firm has moved or is in the process of moving its development work to locations like India, observed Mr. Harding. The strength of this business logic, he said, made it imperative that policymakers in the United States understand that offshoring is irreversible and learn how to constructively deal with it.

FIGURE 9 The offshore outsourcing matrix.

How big is the offshoring phenomenon? Despite much discussion, some of it heated, the scope of the phenomenon remains poorly documented. As Ronil Hira of the Rochester Institute of Technology pointed out at the NRC conference on software, this lack of data means that no one could say with precision how much work had actually moved offshore—a major problem from a policy perspective.¹⁰⁹ Speaking as the chair of the Career Workforce Committee of the Institute of Electrical and Electronics Engineers (IEEE), he noted, nonetheless, that the effects of these shifts were palpable from the viewpoint of U.S. computer hardware engineers and electrical and electronics engineers whose ranks had faced record levels of unemployment in 2003.

What is the impact of the offshoring phenomenon on the United States and what policy conclusions can we draw from this assessment? Whereas some economists believe that offshoring will yield lower product and service costs and create new markets abroad fueled by improved local living standards, some leading industrialists have taken the unusual step of arguing that offshoring can erode the United States’ technological competitive advantage and have urged constructive policy countermeasures.

Among those with a more macro outlook, noted Dr. Hira, is Catherine Mann of the Institute for International Economics, who has argued that “just as for IT hardware, globally integrated production of IT software and services will reduce

these prices and make tailoring of business-specific packages affordable, which will promote further diffusion of IT use and transformation throughout the US economy.”¹¹⁰ Cheaper information technologies will lead to wider diffusion of information technologies, she has noted, sustaining productivity enhancement and economic growth.¹¹¹ Dr. Mann has acknowledged that some jobs will go abroad as production of software and services moves offshore, but nonetheless holds that broader diffusion of information technologies throughout the economy will lead to an even greater demand for workers with information technology skills.¹¹²

Observing that Dr. Mann had based her optimism at that time in part on the unrevised Bureau of Labor Statistics (BLS) occupation projection data, Dr. Hira called for reinterpreting this study in light of the more recent data. He also stated his disagreement with Dr. Mann’s contention that lower IT services costs provided the only explanation for either rising demand for IT products or the high demand for IT labor witnessed in the 1990s. He cited as contributing factors the technological paradigm shifts represented by such major developments as the growth of the Internet as well as Object-Oriented Programming and the move from mainframe to client-server architecture.

Dr. Hira also cited a recent study by McKinsey and Company that found, with similar optimism, that offshoring can be a “win-win” proposition for the U.S. and countries like India that are major loci of offshore outsourcing for software and services production.¹¹³ Dr. Hira noted, however, that the McKinsey estimates relied on optimistic estimates that have not held up to recent job market realities. McKinsey’s 2003 study found that India gains a net benefit of at least 33 cents from every dollar the United States sends offshore, while the United States achieves a net benefit of at least $1.13 for every dollar spent, although the model apparently assumes that India buys the related products from the United States.

These more sanguine economic scenarios must be balanced against the lessons of modern growth theorists, warned William Bonvillian in his conference presentation.¹¹⁴ Alluding to Clayton Christiansen’s observation of how successful companies tend to swim upstream, pursuing higher-end, higher-margin customers with better technology and better products, Mr. Bonvillian noted that nations can follow a similar path up the value chain.¹¹⁵ Low-end entry and capability, made possible by outsourcing these functions abroad, he noted, can fuel the desire and capacity of other nations to move to higher-end markets.

Acknowledging that the current lack of data makes it impossible to track activity of many companies engaging in offshore outsourcing with any precision, Mr. Bonvillian noted that a major shift was under way. The types of jobs subject to offshoring are increasingly moving from low-end services—such as call centers, help desks, data entry, accounting, telemarketing, and processing work on insurance claims, credit cards, and home loans—towards higher-technology services such as software and microchip design, business consulting, engineering, architecture, statistical analysis, radiology, and health care where the United States currently enjoys a comparative advantage.

Another concern associated with the current trend in offshore outsourcing is the future of innovation and manufacturing in the United States. Citing Michael Porter and reflecting on Intel Chairman Andy Grove’s concerns, Mr. Bonvillian noted that business leaders look for locations that gather industry-specific resources together in one “cluster.”¹¹⁶ Since there is a tremendous skill set involved in advanced technology, he argued, losing parts of that manufacturing to a foreign country would help develop technology clusters abroad while hampering their ability to thrive in the United States. These effects are already observable in semiconductor manufacturing, he added, where research and development is moving abroad to be close to the locus of manufacturing.¹¹⁷ This trend in hardware, now followed by software, will erode the United States’ comparative advantage in high-technology innovation and manufacture, he concluded.

The impact of these migrations is likely to be amplified: Yielding market leadership in software capability can lead to a loss of U.S. software advantage, which means that foreign nations have the opportunity to leverage their relative strength in software into leadership in sectors such as financial services, health care, and telecom, with potentially adverse impacts on national security and economic growth.

Finally, Mr. Bonvillian pointed out that “manufacturing matters” even in the New Economy. Referring to the work of John Zysman and others, he noted that advanced mechanisms for production and the accompanying jobs are a strategic asset, and their location makes the difference as to whether or not a country is an attractive place to innovate, invest, and manufacture.¹¹⁸ For the United States, the economic and strategic risks associated with offshoring, noted Mr. Bonvillian, include a loss of in-house expertise and future talent, dependency on other countries on key technologies, and increased vulnerability to political and financial instabilities abroad.

With data scarce and concern “enormous” at the time of this conference, Mr. Bonvillian reminded the audience that political concerns could easily outstrip economic analysis. He added that a multitude of bills introduced in Congress seemed to reflect a move towards a protectionist outlook.¹¹⁹ After taking the initial step of collecting data, he noted that lawmakers would be obliged to address

widespread public concerns on this issue. Near-term responses, he noted, include programs to retrain workers, provide job-loss insurance, make available additional venture financing for innovative start-ups, and undertake a more aggressive trade policy. Longer-term responses, he added, must focus on improving the nation’s innovative capacity by investing in science and engineering education and improving the broadband infrastructure.

What is required, in the final analysis, is a constructive policy approach rather than name-calling, noted Dr. Hira. He pointed out that it was important to think through and debate all possible options concerning offshoring rather than tarring some with a “protectionist” or other unacceptable label and “squelching them before they come up for discussion.” Progress on better data is needed if such constructive policy approaches are to be pursued.

New telecommunications technologies—the subject of STEP’s fifth conference—have contributed significantly to the New Economy. These contributions include the advantages of new product capabilities for businesses and consumers as well as new, more efficient forms of industrial organization made possible by cheaper and more versatile communications. Thus, while the telecom sector accounts, by some measures, for about 1 percent of the U.S. economy, it is estimated to be responsible for generating about 10 percent of the nation’s economic growth.¹²⁰ A key policy question, therefore, is how to sustain or improve on this multiplier of ten, even as new technological innovations are ushering a major shift from a vertical model to a horizontal model of production and distribution in the communications and entertainment industries.¹²¹ This task of adapting policies and regulations regarding the communications industry to new realities is made more challenging given its long legacy—one that goes back past Alexander Graham Bell to Benjamin Franklin, the first postmaster of the United States.

Moore’s Law, which in its modern interpretation anticipates the doubling of the number of transistors on a chip every 18 months, has spurred the modern revolution in digital technologies for over 40 years.¹²² It is likely to continue for

another 10 to 20 years, according to experts in the semiconductor industry.¹²³ This pace of ever faster and cheaper semiconductors and semiconductor-related technologies is likely to continue to have significant impacts, not least on communications technologies. As William Raduchel noted at the conference on telecommunications and the New Economy, the endurance of Moore’s Law means that “the most powerful personal computer that’s on your desk today is going to be in your cell phone in twenty years.” Technologies for display, storage, and transmission of data are also expected to show rapid improvement, he added, though their rates of improvement are likely to abate sooner than that of semiconductors.¹²⁴

Raduchel predicted that enhanced digital sampling, skyrocketing storage capacity, and expanded packet switching technologies will change the way we will work, communicate, and entertain ourselves in the future.¹²⁵ Faster computers mean that digital sampling for recording, playback, looping, and editing of music will improve to the point where it is nearly error free, changing the way music is heard and distributed. Advances in storage capacity and speed will lead to new products (as already previewed with today’s iPods and TiVos) that will likely challenge existing business models of how music and video entertainment is packaged and distributed, and ultimately consumed. In addition, advances in packet switching, where information is commoditized for transmission, will likely mean that “radio, television, classified information, piracy, maps, … any-thing” can be moved around a communications infrastructure with no distinction as to what they are. These developments, in turn, will require greater attention to the issue of standards that can allow for coherence as well as future growth and innovation.

These advances in capturing and distributing information and entertainment in commoditized packets build on the concept of the stupid network—where the intelligence is taken out of the middle of a communications network and put at the ends—a design principle that has already guided the development of the Internet.¹²⁶ According to David Isenberg, such an end-to-end network allows for diversity in the means of transmission—including varieties of wired and wire-

less technologies—with this diversity creating greater robustness against the failure of any one element. As we see next, enhancements in packet switching capabilities are already making such novel technologies as Voice over Internet Protocols (VoIPs) and Grid Computing technically and commercially feasible for widespread use.¹²⁷

• VoIP (Voice over Internet Protocol): In Internet telephony, voice is broken into digital packets by a computer and conveyed over the digital network to be reassembled at the other end. The voice network of the future will run over the Internet Protocol, according to Jeff Jaffe of Lucent Technologies. Since this technology has a completely different capability than traditional landlines when it comes to voice quality, cost, and reliability, he predicted that it will bring about a generational change in voice communications.

Louis Mamaokos of Vonage (a company that has introduced VoIP to commercial markets in the United States and elsewhere) cited two sources of opportunity that arise with VoIP: One is through sharing infrastructure, which comes from chopping up audio into packets and transmitting it over an existing packet-based network, which yields significant cost advantages compared with traditional telephony. But equally powerfully, he contended, are opportunities that come from using software to provide a variety of services for the consumer. For example, by marrying it with the computer, phones could be programmed to control who can call through and when.¹²⁸

• Grid Computing: Grid computing, which allows users to share data, software, and computing power over fiber optic networks is expected to be another major development in information and communications technology. Mike Nelson of IBM likens grid computing to a utility supplying electricity, noting that logging onto the Grid could provide a user access to far more computing power than is possible from a single computer system.

A widely known (but limited) instance of the concept of Grid computing is the current SETI (Search for Extraterrestrial Intelligence)@Home project, in which PC users worldwide donate unused processor cycles to help the search for signs of extraterrestrial life by analyzing signals coming from outer space. The project relies on individual users to volunteer to allow the SETI project to harness the unused processing power of the user’s computer. About 500,000 people have

downloaded this program, generating an amount of computing power that would have cost $100 million to purchase.

Grid computing is likely to have fewer nodes that are tied together than in the SETI case, said IBM’s Nelson, but because the size of the machines can be larger—including large servers, storage systems, and even supercomputers—high levels of computing power can be generated. Further, since the systems involved in Grid computing will be more tightly coupled and more general purpose, they can be far more versatile. The next step in Grid computing, he predicted, is the “Holy Grid” where everything is connected to everything, running common software, able to tackle a wide range of problems. With the advent of such a grid, both small and large companies would be able to buy the computing power they need and get the software they need over this grid of network systems as needed on a pay-as-you-go basis.

In IBM’s view, a part of the larger vision of Grid computing includes autonomic computing, where integrated computer systems are not only self-protecting, self-optimizing, self-configuring, and self-healing, but also come close to being self-managing. Another important component of this vision is pervasive computing, where sensors embedded in a variety of devices and products would gather data for analysis. These sensors will be located all around the world and the data they generate will have to be managed through the Grid. As Nelson predicts, “Soon we will have trillions of sensors, and that is what we really rely on the ‘Net for.”

The predicted arrival of Grid computing means that firms in the computer industry have an enormous stake in the future of telecommunications networks.

With the Grid, the future of computing lies in complex network-based technologies, such as Web services, which tie together programs running on different computers across the Internet, and utility computing to provide computing power on demand. With telecommunications firms becoming more dependent on information technology, and vice versa, the two industries are likely to become ever more closely intertwined.

While these and other emerging technologies offer alluring prospects for a more vibrant and productive future, a major focus of the STEP conference on telecommunication technologies concerned the regulations that condition the speed at which these technologies and others can be adopted as they become available. As Dr. Jorgenson pointed out in his introductory remarks, the issue of regulation is particularly germane to telecom, which is regulated at both the federal and state levels. Broadband regulation, in particular, was identified by several conference participants as a bottleneck to realizing the benefits of new information and communications technologies in the new “wired” and “wireless” economy.

Broadband, which refers in general to high-speed Internet connectivity, already supports a wide range of applications ranging from email and instant messaging to basic Web browsing and small file transfer, according to Mark Wegleitner of Verizon.¹²⁹ In the near future, he said, improved broadband networks can lead to true two-way videoconferencing and gaming as well as VoIP. The future of broadband, he predicted, includes multimedia Web browsing, distance learning, and telemedicine. Beyond these applications, he noted, rests the possibility of immersive gaming and other types of information and entertainment delivery that comes with high band output combined with high-definition receivers.¹³⁰

Can we indeed arrive at this promising future? Charles Ferguson of the Brookings Institution noted that while many foresee what a “radiant future” should look like, there exists an enormous gap for many between this vision for broadband-based technologies and the lack of adequate high-bandwidth access to a broadband network.

Indeed, as many conference participants pointed out, the United States is falling behind other nations in access to high-bandwidth broadband.¹³¹ Jaffe drew attention to the reality that the United States had fallen far behind other leading nations in broadband penetration. Isenberg underscored this point, reporting that the International Telecommunications Union (ITU) had, in fact, ranked the United States in thirteenth place in 2003 and that the United States had likely since fallen to fifteenth place in broadband penetration. Citing the ITU figures for 2003, Ferguson reported that the penetration of digital subscriber lines (DSLs) in the United States was 4.8 per 100 telephone lines, in contrast to South Korea where the penetration rate is 27.7 per 100 telephone lines. He noted that the United States had also fallen behind Japan and China in the absolute number of digital subscriber lines.

Acknowledging that this low figure for DSL is explained in part by the fact that a majority of U.S. residential broadband connections are through cable modems, Ferguson nevertheless contended that this fact did little to change the overall picture. In the first place, he explained, when business connections were included, the percentage of total U.S. broadband connections provided by cable was relatively low. In the second place, even in the residential market the percentage of connections provided by cable had been holding roughly constant, as had the cable system’s growth rate in respect not only to connections but also to bandwidth levels.

Ferguson observed that bandwidth constraints rather than computer hardware frequently dominate the total cost of adoption of a new network computing application. Personal computers were adequately powerful and relatively inexpensive, he noted, but given bandwidth constraints, deploying a high-performance, high-quality videoconferencing system or other applications could nonetheless prove extremely expensive.

Adding his own negative assessment of the U.S. competitive position, H. Brian Thompson of iTown Communications noted that while (what is commonly called) the Information Superhighway is capable of handling very high capacity in its fiber optic network, and while most desktops and laptops could function at between 1 and 3 gigabits per second, the problem was that there was often less than 1 megabit of connectivity between the two. This weak link—the broadband gap—was illustrated schematically by Thompson at the conference. (See Figure 10.)

In his remarks at the conference, Mark LaJoie of Time-Warner Cable cautioned that national aggregations showing the United States in thirteenth place worldwide do not tell the whole story. Differences in regulatory climate, the history and condition of infrastructures, the way in which products are used, as well as population densities are all factors influencing measures of broadband penetration. High-density cities like Tokyo and Seoul were likely to have higher levels of penetration, as do similar urban areas in the United States, he said, and added that while the infrastructures in Europe and Asia were newer, U.S. cable and telecom firms were making significant investments in expanding broadband capacity.

Agreeing that there are many ways to spin the numbers on broadband deployment, Mark Wegleitner of Verizon nonetheless acknowledged that “we aren’t leading in what we have to perceive as one of the key technologies for any national economic environment going forward.” He noted that his company, Verizon, was spending $12 billion annually on improving the broadband infrastructure—including expanding fiber to the home—thereby helping the United States catch up with other leading nations. At the same time, he predicted that “bandwidth demands are just going to grow and grow and grow,” as new applications come into use.

If broadband can serve as an engine for the nation’s future growth and competitiveness, as emphasized by several participants at the conference, a lack of an adequate access to the broadband network may lead to a loss of this economic opportunity.¹³² Assessing the impact of the broadband gap, Charles Ferguson noted that the “local bandwidth bottleneck” is having a substantial negative effect on the growth of the computer industry and of various other portions of the information technology hardware and software sectors. While conceding that computing an estimate of this impact in a rigorous way would be extremely difficult, he nevertheless asserted that “you can convince yourself easily that this effect is something on the order of one-half of one percent—or even up to one percent—per year in lost productivity growth and GNP [Gross National Product].”

Commenting on the national security implications of the broadband gap, Jeff Jaffe reminded the audience that the 9/11 Commission had recommended that the nation’s digital infrastructure be prepared to deal with simultaneous physical and cyber attacks. In the case of a national emergency it will be important for first responders and other individuals to communicate effectively with each other, and

a high-bandwidth, interoperable system is essential for this task, he said, adding that such a network is still not in place today.

While many of the participants at the conference concurred that the United States faces a broadband gap, views varied as to the reasons for as well as solutions to this situation. Some suggested that the broadband gap has emerged because some telecom and cable companies have been reluctant to provide adequate interface between the user and the fiber optic cable networks. Others suggested that the broadband gap arose from the consequences of federal and state regulations.

• Flawed Market Motives of Telecom and Cable Companies: What is holding back high-bandwidth broadband penetration in the United States? Dr. Isenberg noted that the rise of the stupid network makes it difficult for the telephone or fiber company to sell anything other than commodity connectivity. In the new inter-networked model, it was the Internet Protocol’s job to make all that was specific to a single network disappear and to permit only those things common to all networks to come to the surface. Since the Internet ignores whatever is specific about a single network, including features that had formed the basis of competition for the telephone or cable companies, these companies have little to sell beyond access, he argued, and therefore faced little incentive in providing the public access to high-bandwidth broadband. The result, he said, was a crippled network with far less bandwidth available than technology would allow or than is available in other technologically advanced countries.

Ferguson suggested that flawed markets were behind the high cost of securing adequate bandwidth in the United States. He noted that both the telephone and the cable companies had “severe conflicts of interests” and that they largely avoided competing with each other. Even competition for residential markets was “quite restrained, and much less substantial than you might suspect.”

The conflict of interest for the telephone companies is “fairly obvious,” Ferguson asserted. Incumbent businesses were providing very expensive voice and traditional data services. Very rapid improvements in price/performance of bandwidth would undercut their dominant businesses in a major way. The same was true of the cable system: It provided video services that could easily be provided over a sufficiently high-performance Internet Protocol network.

• Consequences of Unbundling Network Elements: In the discussion following the second panel, Kenneth Flamm noted that more than one speaker had spoken of a tendency to dismantle some of the opening up of the local loop that had been the centerpiece of the 1996 Telecommunications Reform Act. The Act required incumbents to make parts of its network available to competing

operators, in particular the “local loops”—the wires that run from telephone exchanges into homes and offices.¹³³

The 1996 Act sought to promote competition by asking incumbents to share this part of their networks with rivals—technically known as “local loop unbundling” (LLU)—given that the expense for competitors to build their own networks would be very high in the short term. In practice, however, most incumbent operators saw unbundling as robbery, according to Thompson. This meant (as The Economist describes it) that “the incumbent must, in effect, give its rivals a hand as they try to steal its business. Not surprisingly, most incumbents find procedural, legal and technical reasons for being slow about it.”¹³⁴ Though intended to promote competition in the short run, local loop unbundling may have inhibited investments in alternate infrastructure that competitors might otherwise have made over the longer term. And because it forced incumbents to share their networks with rivals, this may have also deterred them from investing in new equipment. An unintended consequence of the 1996 Telecommunications Act may well have been to inhibit investment needed to provide high-bandwidth broadband access over the local loop, although the issue of whether mandatory unbundling increases or decreases the roll out of broadband network access remains an open empirical question.

Even so, one of the authors of the Telecommunications Act of 1996, Charles Thompson, conceded that the concept of unbundled network elements, introduced in that legislation was moribund—that he “would be the first to put flowers on the grave of unbundled network elements.”

• Outdated Standards and Regulatory Uncertainty: Outdated standards and a regulatory uncertainty may be retarding progress in addressing the broadband gap, according to some conference presenters. On the issue of standards, Peter Tenhula of the Federal Communications Commission (FCC) acknowledged that wireless technology regulation was still being governed by a 90-year-old spectrum management regime rather than one “rooted in modern-day technologies and markets.” Such outdated regulations, he noted, fail to capitalize on technological advances in digital technologies such as those that allow for greater throughput of information, interference management, and spectrum sharing.

Regulatory uncertainty is also holding down the installation of fiber all the way to the curb, noted Dr. Jaffe. Clear regulation is needed, he stated, to

encourage sufficient near-term investment in fiber infrastructure. This regulatory environment may have been further clouded in recent years by increasing federal concerns about infrastructure protection, disaster recovery, and emergency services in the wake of recent concerns about terrorism. According to Jaffe, vendors such as Lucent face uncertainties in developing new products at a time when regulatory imperatives are very slow to come out.

Another important source of regulatory uncertainty is the patchwork of local regulations issued by individual municipalities. Cable infrastructure is often governed by city-specific franchise agreements, while telephone companies and other broadband providers may in some cases prefer statewide or even national authority as a means towards greater regulatory simplicity and predictability.

In addition, as Verizon’s Wegleitner observed, prevailing uncertainties in updating regulation make it difficult for his company to invest in the development of an effective broadband network. Incremental rulemaking in the transition from the old regulatory regime to a new one often creates ambiguities, with investments of millions or even tens of millions of dollars hinging on the interpretation of words that, while written only a few years before, were already technically obsolete. “It is that interpretation that is going to determine the path forward of the network’s evolution.” This “unnecessarily complex regulatory environment” did not make sense in that it discouraged investment.

Thompson objected, however, arguing that large telecom and cable companies are not passive recipients of federal and state regulation and that, moreover, the current regulatory environment is greatly affected over the years by the power of incumbents on all sides. To the extent that incumbents influence regulation, the current uncertainty in regulation may well reflect the uncertainties that major cable and telecom providers are facing in coming up with a viable business model that allows profits in an arena that has been transformed by new technologies. Lisa Hook, recently of AOL-Broadband, noted in this respect that firms in the broadband industry were struggling at the service layer to find business models and revenue streams based on new technologies that would justify the investment needed to make nearly unlimited bandwidth widely available.

According to IBM’s Nelson, the Internet revolution is less than 8 percent complete, with many new applications still to be enabled by future technologies like the Grid. Realizing this vision of the next-generation Internet will require both new technologies as well as significant investment, he cautioned, as it will entail providing whole neighborhoods with gigabits-per-second networks that are as affordable and reliable as they are ubiquitous. “Getting there is going to require more intelligent, more consistent policies than we have today,” he declared. Participants at the conference considered a variety of means by which

the nation could close the broadband gap, of which some key approaches are previewed below.

• Directed Government Incentives: Ferguson suggested that the nations that were ahead of the United States in broadband penetration shared two characteristics. The first was that their governments are “much more heavily involved in providing incentives and/or money and/or direct construction of networks than is the case in the United States.” The second was that their Internet providers are under government pressure to improve their price and performance. For example, he said that the Chinese government had made it clear to the country’s principal telecommunications providers that broadband deployment was a major national priority. The situation was similar in Japan and Korea, adding that government encouragement in Canada and the Scandinavian countries had also enabled those countries to surge ahead of the United States in high-bandwidth broadband penetration.¹³⁵

For the United States, Ferguson recommended a variety of policy measures to bridge the broadband gap. Initiatives could include subsidizing the deployment of municipal networks and offering investment incentives to public and private providers. Putting more pressure on incumbents to open up their networks so that there is an open architecture broadband system that is more analogous to the structure of the Internet is another avenue.

• Faith in Efficient Markets: In contrast to this more policy-driven approach, Verizon’s Wegleitner noted that broader technical, financial, and regulatory improvements would reduce uncertainty and allow markets to function efficiently. While admitting that current challenges resisted simple solutions, he put forward what he called a short answer to the problem: “Let the markets rule.” By this, he envisioned the Internet of the future as an interconnection of commercial networks such as Verizon’s rather than the confederation of commercial providers that it is now. He added that the future requirements for services offered customers via broadband would be of such quality and scope that only an interconnection of commercial networks could provide this service.¹³⁶ To make this network of the future possible, Wegleitner recommended further development of appropriate standards for communication protocols and a new way of levying tolls on customers for use of the infrastructure that belongs to companies like Verizon, combined with a light regulatory touch.¹³⁷

• Networks in the Hands of Customers: In the discussion that followed the first panel, Jay Hellman, a real estate developer, observed that there exist business opportunities both in laying fiber to the home and in making sure it functions. He likened the duo of fiber and services to a public roadway where service companies like FedEx and UPS competitively ply their fleets. It was desirable, he added, that the street be accessible to as many competitors as possible. He also added that his own frustration with the capacity offered by existing providers had prompted him to start his own small telecommunications company. Responding to this comment, David Isenberg noted that the development of technologies that allow customers to create their own networks and that create opportunities for individuals to provide service innovations was important to sustain innovation and provided a broader, more generic solution to the broadband challenge.

• Municipally Owned Fiber: Thompson proposed a different approach, recommending the development of non-profit public-private partnerships at the local level to stimulate the development of broadband to the home. These partnerships would serve as a utility, lighting fiber but not providing any service on that fiber except those municipal services that the town or community base chose to provide. The network would be open to any and all service providers with an Internet Protocol basis—be they telephone companies, cable companies, software companies, or others providing on-line entertainment—and it would be used by all under the same terms and prices. Communities could build this network, just as municipalities build and maintain roads and sewers, he added, citing the case of Ireland where, Thompson said, such partnerships have been successfully developed to provide broadband access.

While separating the network access component from retail services may help municipal providers of network infrastructure, more needs to be learned about the feasibility of this idea in the United States, including whether customers want to buy their services in this way. The issue of whether the municipal provision of infrastructure will in fact lead to more competition for broadband access also remains to be studied.

• The Wireless Wildcard—A Silver Bullet?

Wireless broadband access can be a third tier that competes with cable and DSL, according to David Lippke of HighSpeed America.¹³⁸ In this way, wireless broadband can help overcome the limitations associated with traditional wired

broadband access. While wireless broadband has been in limited use so far due to relatively high subscriber costs and technological limitations such as problems with obstacle penetration, rapid advances in technology are likely to overcome such challenges. Moore’s Law applies to wireless no less than other forms of telecommunications, he noted, predicting that wireless data rates would reach all the points through which traditional telecom had passed.

In particular, scientists and engineers working on the upcoming WiMAX standard have resolved a number of problems that had bedeviled existing wireless protocols such as WiFi. The prospect of reaching gigabit speeds was now being mentioned, and other quality-of-service issues as well as lower costs of installation are being addressed. To the extent that these predictions are realized, the WiMAX protocol may well offer an effective wireless solution to the broadband gap, especially for smaller towns and communities across the United States.

The move from analog to digital information and communication technologies is ushering a major transformation disrupting how telecom, cable, and music and video entertainment companies, among others, do business. Because analog solutions were all that existed until recently (except in some fields of computing), these industries each matured into separate industries, with separately evolved business models and regulatory frameworks. In the digital age, however, basic technologies like digital sampling and packet switching enable the commoditization of voice, data, and images into digital packets that resemble each other. These packets can be sent over the Internet with no distinction as to what they are, to be reassembled at the intelligent ends of the network.

Drawing on these observations, William Raduchel noted at the conference that the information and communications technology revolution will usher the end to stovepiping as service and content providers shift from vertical integration to a greater reliance on horizontal platforms. This change, he noted, will give rise to a variety of public policy issues as individuals and businesses in the economy restructure to take advantage of the potential offered by new technologies.¹³⁹ He also noted that the speed of change is likely to be such that the economy may not be able to adjust to it readily. Among the issues to be addressed is the challenge to intellectual property rights and the question of regulation, which is expected to be very challenging.

The potential and implications of the move from analog to digital information and communication technologies were discussed by several of the conference’s participants. Key points from these discussions are summarized below. As in any

conference that includes a variety of perspectives, some of these policy recommendations are mutually contradictory, and evidence may be required regarding their efficacy.

Raduchel sees the Internet as having two complementary aspects—it is both a physical set of networks as well as a protocol known as TCP/IP. At present, the physical network can only support movies and other applications at low bit volumes and is often not cost-effective—although this can be expected to change as technology improves and the broadband gap is overcome. The significance of the Internet Protocol, he said, is that it makes all networks look the same and allows interoperability. It was for this reason that the telecommunications world could be expected to move to one set of interconnected webs, he said, predicting that “five to ten years from now, we will be online all the time.”

This convergence is challenging the traditional business models of firms in these industries. How would telecom companies, for example, deal with new technology that makes cell phones work perfectly everywhere or with much cheaper VoIP service? The next decade, warned Dr. Raduchel, would be marked by “lots of dislocation” as firms attempt to adjust to new technological and commercial realities.

According to Mr. LaJoie, the convergence of data, voice, video, wireless, public networks, and private networks in an end-to-end infrastructure was changing the terms of competition across industries. Where there was once a big separation between what the telecom and cable industries did for example, “now everybody is in everybody else’s business.” While cable television, Internet, Cellular, WiFi, and satellite transmission businesses were once distinct, LaJoie believes that they are all destined to overlap and offer similar kinds of products, suggesting with some optimism that the economic rewards that will arise from this competition would be what drives continued innovation, the advent of new services, and increased broadband connectivity.

The potential end of stovepiping also poses new challenges for consumers. Many consumers, faced with a proliferation of Internet services, operating systems, and devices will want a service that is easy to use and integrated, predicted Ms. Hook. She noted that companies like AOL Broadband see a market opportunity as aggregators, packaging a variety of content and communications services over the Internet and protection against viruses and spy-ware that are easy to launch and use.

In addition to disruption in the business models of firms that deliver a digital signal is the disruption to business models of firms that provide the content. Indeed, the music and entertainment industries are among those that are also undergoing a fundamental shift in the digital age. Andrew Schuon of the International Music Feed television network noted that while the public’s desire to consume music has never been greater, with new technologies allowing users to take an entire music collection with them anywhere they go, the key problem for content providers is how to make money selling music in the new medium—given that technology already available has allowed consumers to share music and other content with each other for free. At present, he noted, legitimate downloads account for only a few percent of all downloads from the Internet.

He noted that technology developed for building legitimate services makes it now possible to protect intellectual property, to monetize it, and to track licenses while, at the same time, creating a good experience for the consumer. However, this technology has to catch up with consumer expectations that have developed in the absence of such constraints: “If you steal the content, you can do anything you want with it—put it into any portable device, put it on as many computers

FIGURE 11 Vertical silos to horizontal layers.

as you have, use the content as you see fit.” The challenge for the music industry is to find a way to get the consumer to pay for its product while at the same time being more creative than the illegitimate download sites. The music industry, Mr. Shuon said, has to offer the modern customer the flexibility to use the content in the way they want to, in addition to offering superior content and a fair price.

Steve Metalitz, of the law firm Smith and Metalitz, agreed that developing a legitimate market for copyrighted materials over broadband—for entertainment, services, software, video games, research and reference works—was indispensable for the long-term viability of these industries. Acknowledging that piracy will continue to be a problem, he added that the challenge for the future of broadband is to achieve a relatively low level of piracy and a very high level of legitimate products. Addressing this challenge requires:

• developing legitimate markets for copyrighted materials over broadband,

• providing greater security for delivering content to an end-user including measures to ensure that the income-generating potential of material going into the pipe did not vanish forever,

• creating a usable legal framework to protect the technological measures used to control access to copyrighted material in the network environment,

• focusing enforcement of piracy problems on organized criminal groups as well as dedicated amateurs who play a role in making the system insecure, and

• improving public education to make consumers aware that certain types of file sharing are illegal and of the need to secure permission to avoid copyright infringements.

Cooperation, Mr. Metalitz concluded, is needed among providers of network services along with better communication with policymakers to advance these objectives.

According to Peter Tenhula of the FCC, the challenge for regulation concerns the migration from decades of regulatory stovepipes towards a new vision of a variety of applications and services (covering voice, video, and data among others) that are provided over multiple and competing telecommunications platforms (including cable, satellite, DSL, and power lines). For this idea to work, content or service providers need a choice of mechanisms by which they can reach their customers. Rather than preserve the artificial vertical integration that had existed for decades and had created silos that grew up over the years, Mr. Tenhula suggested that it made better sense to let the natural layers fall as they might. (See Figure 11.) Replacing sector-specific communications regulation with a layered regulatory model, he added, would better complement the networked characteristic of the New Economy.

The FCC’s agenda, he said, was to guide and propel the journey from a slow, conventional analog world to a digital world with significant opportunities for faster, more reliable, higher-quality information and communications, with the overall goal of providing substantial benefits for American consumers.

Concluding the series of conferences on the New Economy, Dr. Jorgenson noted that the New Economy had witnessed a huge shift from a vertical model to a horizontal model in the computer, semiconductor, and communications industries. In this new model, he said, most of the interesting innovations were disruptive. The challenge for businesses in this changing environment was to figure out how to make money, which was hard given that consumers were both clever and unpredictable. It was “too bad,” he said, that the consumer ends up carrying away most of the welfare, which then cannot be delivered to shareholders. But in another respect, he added, the fact that “consumers emerge over and over again as the big winners … was a great thing about the New Economy.”

Dr. Jorgenson characterized the policy issues in the telecommunications challenge as particularly difficult. While many economists are prone to offer private property as an answer to policy dilemmas, the presence of common property in the form of the digital communications infrastructure made matters more complex, he noted, adding that a way had to be found of maintaining common facilities within a market-based approach. The transmission of property such as data, software, and music across this network also raised questions about its protection, while ensuring privacy for users. Taken together, these issues provide a robust agenda for further study and consideration about the New Economy— which, he noted, has been a central aim of the National Academies’ Board on Science, Technology, and Economic Policy.

FIGURE 12 Productivity growth over the business cycle: 2001 recession compared with averages of earlier recessions.

SOURCE: U.S. Department of Labor, Bureau of Labor Statistics.

NOTES: Productivity series are normalized to equal 1.0 at the beginning of each recession. The 1973-2000 line represents average productivity growth over the four recessions during that period; the 1947-2000 line represents average productivity growth over the nine recessions during that period.

The New Economy is alive and well today. Recent figures indicate that since the end of the previous recession in 2001, productivity growth had been running about two-tenths of a percentage point higher than in any recovery of the post-World War II period.¹⁴⁰ (See Figure 12.) The challenge rests in developing evidence-based policies that will enable us to continue to enjoy the fruits of higher productivity in the future. It is with this aim that the Board on Science, Technology, and Economic Policy of the National Academies has undertaken a series of conferences to address the need to measure the parameters of the New Economy as an input to better policymaking, and to highlight the policy challenges and opportunities that this New Economy offers.

This ambitious series was begun in the midst of a tremendous economic boom, and although economic conditions have changed since then, the basic structural dynamics underpinning the New Economy have remained intact. Faster and cheaper computing power and communications capabilities continue to have a momentous impact on productivity growth in the United States and around the world. Understanding the basis and dimensions of this New Economy is important if we are to develop the economic policies required to ensure the nation’s future prosperity and growth.

STEP’s series of conferences on the New Economy have given momentum to this task. STEP’s first conference on Measuring and Sustaining the New Economy showed that technology is the main source of the development denoted by the term “New Economy,” and that the key technologies center on semiconductors.

The second conference addressed semiconductors specifically, dealing—as described by Moore’s Law—with the speed at which semiconductor technology develops. At that conference, Robert Doering of Texas Instruments and other leading authorities in the field projected that semiconductor development would continue at that accelerated pace for at least another decade or so while highlighting what needs to done to keep Moore’s Law on track.

The topic of the third conference in the series was computers. That conference brought to light that the industries that manufacture computers and computer components are also driven by a Moore’s Law phenomenon and that they have developed internal metrics to gauge rapid technological developments.

The fourth conference of the series examined developments in software measurement, the vulnerabilities affecting the nations’ complex software infrastructure, as well as implications of the offshoring of software production abroad.

The final meeting on the telecommunications challenge described a huge shift from a vertical model to a horizontal model of production made possible by inexpensive computing and communications. Low-cost and rapid data and voice transmission is transforming the competitive strengths of national economies by ushering the rapid globalization of research and production. How we adapt our laws and regulations to capitalize on these new technological opportunities will determine the future of the United States’ security and economic preeminence in the world.

Taken together, the work sponsored by STEP under the rubric Measuring and Sustaining the New Economy has produced what Dale Jorgenson described as

the most detailed and comprehensive picture available to date of what is known as the New Economy. This undertaking provides the basis for further research on the dimensions of the New Economy and policies that can enhance the benefits of the New Economy.

This page intentionally left blank.