The Region

Chasing the Tail of High-Tech

The impact of high-tech industries, seen as the mainstay of the new economy, proves difficult to gauge.

Ronald A. Wirtz - Editor, fedgazette

Published September 1, 2001  |  September 2001 issue

It's an assumed fact that the nation's economy today is driven by high-tech. Wherever our economic attention turns, high-tech is often the focus, whether it concerns job creation, the stock market, university research or workforce development.

Numerous policy initiatives have been passed on the sole argument that local or state prosperity will be shackled in the new economy without a high-tech development strategy. Woe is the region or state that performs poorly in one of many quasi-scientific indexes published every year that plot and rank high-tech growth. Earlier this year, a headline in the Columbus, Ohio, business press reported the unenviable: "Study finds Columbus lagging in high-tech economy." Oftentimes, it's a simple matter of economic envy. "Dreams of high-tech glory passing [Orange County] by," lamented a headline in last year's Los Angeles Times.

In an oft-cited speech last year, David Kidwell, then-dean of the business school at the University of Minnesota, charged that "Minnesota is not a major player in the new high-tech economy emerging in the 21st century," and argued that "[t]he key to our future economic success is to foster a high-tech renaissance in Minnesota. If successful, this high-tech activity can drive new economic growth—creating new wealth and high-paying jobs—and help maintain our region's high standard of living."

Such rhetoric is hard to disagree with, because everyone knows high-tech is the future, right? Probably, but we're so obsessed with high-tech that a couple of important details are routinely overlooked: First, and most importantly, we don't know (or can't agree on) how to define and draw clear lines around high-tech. A look at almost any two studies of high-tech industries will likely uncover two fundamentally different definitions of high-tech, neither of which might hold water in just a few years because the industry changes so quickly.

This annoying little bugaboo means research to date on the high-tech industry has built only a thin, widely dispersed foundation, which provides little solid footing for further research that might provide better information on and analysis about the industry. As a result, despite our hyperventilation about all things high-tech, there are very little public data on the high-tech industry and its true impact in the economy. In other words, we know high-tech is important in today's economy, but don't ask us how we know. Please.

To most, the semantic hair-splitting involved in defining high-tech is something for economists and policy wonks to settle. As Supreme Court Justice Potter Stewart so aptly said about pornography, a person might not be able define it, but they know it when they see it. But there are real consequences for failing to go beyond such a quaint measuring stick. For example, it's hard to imagine effective policy to encourage "a high-tech renaissance" given that the target is broad, out of focus and fast-moving.

Efforts to define and assess high-tech and other "emerging" industries have engaged academics, business and government agencies—including the Federal Reserve—for years. But the complexities of delimiting and measuring such an amorphous beast have foiled all attempts at capture. The prey remains, largely, at-large.

You decide

No one argues anymore about the importance of high-tech in today's economy. As a 1997 study by the Bureau of Labor Statistics (BLS) said, "High-technology industries are the most important source of strategically transformative products and processes in the U.S. economy. Changes in employment patterns in these industries thus command the interest of researchers, policymakers and the general public."

Business and policy emphasis on high-tech has only accelerated since then, and for good reason. High-tech businesses have a strong reputation for innovation and global competitiveness, which brings with it high average wages and strong employment growth, according to studies by the BLS and the American Electronics Association (AeA).

So what's the problem? In fact, there isn't one if you don't want to quibble about methodologies and other details. But then you also have to be willing to gloss over a little thing called accuracy and be content with old information on a fast-changing industry. A recent internal report on tracking high-tech by the Dallas Fed outlines a few of the problems. Data on high-tech employment is available, but such figures "lag the current state of high-tech by at least eight months ... [and] give us only a partial picture of high-tech manufacturing." Data on high-tech output would be particularly useful but "are even more out of date," the report stated, often running two or more years behind. Numerous contacts also noted that up-to-date regional data is particularly difficult to come by.

That's not to say that past and current research efforts on high-tech are inaccurate, or not useful. Far from it. They have provided the best information to date on the high-tech industry. But such efforts also have what one economist in the Federal Reserve System called "serious data and methodological problems" that ultimately need to be ironed out. That will likely take a lot of heavy lifting.

Much of the problem stems from the fact that "high-tech" is a ubiquitous term, malleable enough to fit almost any economic blank. This makes measuring and tracking the high-tech industry, particularly at the microeconomic level, much more difficult compared to a well-defined, output-based industry like mining. (This means we also have a poor grasp of high-tech's secondary impact as new technologies seep into other industries, but that's an entire discussion in itself.)

Studies attempting to define high-tech typically use either an input or output strategy. Input studies identify high-tech companies based on the intellectual investments in a company's core product or process. Most often, such studies define high-tech by the level of research and development and by occupation classifications (for example, average number of scientists and engineers, which acts as a proxy for research and other intellectual inputs). In such studies, the sophistication of the product itself is unimportant—think of the many tape and other adhesive products from research and engineering giant 3M.

Output studies, on the other hand, classify industries as high-tech based on the perceived sophistication and value-added of products and services—like computers or software programming. This study method is less common, mostly because of the subjective analysis required in drawing a line between high- and low-tech.

Both types of studies often use the government's Standard Industrial Classification (SIC) system. A huge public database on private industry, the SIC system classifies different types of companies in increasing detail using two-, three- and four-digit codes (see sidebar for example). Because there are no SIC categories for high-tech, researchers must pick SIC codes that they believe correspond with agreed-upon inputs (high R&D) or outputs (software) that are considered high-tech. Once finalized, employment and various other data can be gathered for each SIC code and then aggregated—voilà, you've just defined and tracked high-tech.

Not so fast, others say. While useful as a starting point, such methodology is nonetheless rife with problems. Probably the most recognized indexes of the high-tech economy come from the AeA, the nation's largest high-tech trade organization and publisher of a series of high-tech progress reports. Cyberstates, for example, narrowed high-tech into three categories: high-tech manufacturing, communication services, and software and computer-related services. It then chose 45 SIC codes at the four-digit level that it believed corresponded to these three broad groupings. [These 45 were listed under 13 different industry groupings, or three-digit SIC codes.]

On its Web site, the AeA acknowledged that in producing its reports, "we found that there is no consensus on the definition of the high-tech industry. ... This means the definition of the high-tech industry varies greatly depending on what combination of products and services are selected from the SIC codes to define the industry."

To illustrate: The BLS has published a handful of studies on the high-tech industry in the last two decades. A 1999 report by the BLS used occupation and R&D intensity to determine which SIC codes represented high-tech. The final list consisted of 29 different three-digit SIC codes—more than two times those of the AeA definition—and included such industries as petroleum, cigarettes, soaps and cleaners, paint, plastics, agricultural chemicals, drugs, aerospace, and motor vehicles and equipment.

There is a small handful of industries—like electronic components, computers and communications equipment—included in most studies of the high-tech industry. But even here, a 1997 BLS study pointed out that "problems are also present even in apparently uncontroversial" classifications. For example, computer and office equipment (SIC 357) includes producers of such low-tech items as staple removers and hole punches, not to mention that most computer makers today "use mass-produced components assembled in highly routine settings with minimal engineering and scientific input," the report said. "[B]ecause the effects of technological change can be seen in almost every industry, these uncertainties are inevitable in any high-tech study, no matter how precise the classifying metric."

The fact that there is little agreement (indeed, even discussion) on the matter has meant that a variety of high-tech indexes and rankings are showing up, each arguably as valid as the next. A high-tech index from, a provider of economic, financial and industry research, had 11 three-digit SIC codes, four of which are not included in either the AeA or BLS definitions. A Brookings Institution review of 14 studies found varying use of 45 different three-digit SIC codes to define high-tech, accompanied by almost 100 different four-digit SIC codes.

Another recent Brookings report summarized the inherent difficulties. "Any classification system is, at best, an imperfect means of describing the activities of diverse and quickly changing business enterprises. ... Any definition that relies on groups of [SIC] codes is likely to only roughly capture the product and market specializations of the firms and areas being analyzed."

SIC codes, the report said, imply "that there is a great deal of homogeneity among these high-tech firms. ... [Rankings] that group inherently disparate firms such as medical devices, semiconductors, telecommunications and software together into a single category of 'high-technology' and attempt to explain their behavior as if they were homogenous units driven by a common set of factors, are likely to be substantially misleading and incomplete."

These little methodological nuances might seem trivial, but they heavily influence where regions and states rank in high-tech studies. For example, the Midwest ranked next to last of nine regions in the concentration of high-tech employment in 1998, according to a 2000 report by the AeA. However, the study did not include either the aerospace or automotive industries, both of which are mature but research-intensive industries. Including both industries in the tabulations boosted the Midwest's rank from eighth to third, and the state of Michigan shot up from 17th to fourth, according to a follow-up study by the Michigan Economic Development Corp.

The AeA and others note that sound data collection is complicated because many new industries widely considered to be high-tech are not fully captured by current government statistics. That's because the SIC system has not been updated since 1987. Many believe the transition to the new North American Industry Classification System (NAICS) will address some of the data problems. But even if it does improve the classification system, it does not (at least not yet) draw the subjective line between what is high-tech and what is not. Nor is it dynamic enough to identify and include emerging, growing technology industries. A salient example of the new system's limitations: NAICS does not recognize or classify biotechnology.

A new breed

To feed the growing interest in the high-tech sector and the new economy, other, less scientifically based indexes are gaining popularity. Many take a more holistic approach, including both traditional measures of high-tech (like employment) as well as indicators of underlying conditions believed to nurture high-tech industry.

The Massachusetts Technology Collaborative, for example, puts together an annual index on the state's "innovation economy." It features 30 indicators on such things as business friendliness (including average pay and CEO ratings), entrepreneurship and idea generation (IPOs and patents), human resources (workforce education and computer access in the classroom) and technology and investment (federal R&D and venture capital).

Similar indexes are produced by the Milken Institute and the Progressive Policy Institute, and tabulated state by state. Milken's "New Economy Index" uses 12 (non-SIC) data sets, ranging from higher education rates to exports to R&D and business startups. The Progressive Policy Institute has a 16-point survey—the third in a series of such indices—it calls the "Metropolitan New Economy Index." Measures (fairly) unique to this effort are jobs in "gazelle" (or fast-growing) companies, broadband capacity and Internet domain names.

These efforts get away from a concrete definition of high-tech and instead attempt to measure those things assumed to correlate with or contribute to a healthy high-tech sector. There is general agreement regarding some of the basic factors that encourage high-tech development, like high levels of college-educated workers. But the sheer variety of indicators used by such studies suggests they more likely reflect the leanings of authors and sponsoring organizations.

The Fed and high-tech

This issue does have important, if indirect, ramifications for the Federal Reserve. As the economy's central pulse-taker, the Fed depends heavily on data to determine the health of the economy. Without any reliable measurement models for high-tech, the Fed arguably cannot provide as clear a diagnosis on the economy's health as it otherwise might.

Some would argue—with some validity—that there's a "so-what" factor to consider. For starters, the Fed is mainly a secondary, rather than primary, gatherer of data on the current economy. It gobbles up reams of research, but the vast majority of it (for example, employment figures) is tracked, tabulated and aggregated by somebody else (like the BLS). To the extent that tracking high-tech is important, the Fed isn't set up to solve that problem directly. What's more, existing data does manage to capture high-tech activity, but it gets categorized and analyzed under more traditional economic frameworks like manufacturing or business services.

But the Cleveland Fed's 2000 annual report argued that good information is important to the Fed's monetary stabilization policy, and data models have not kept up with the shift in the economy. "If the central bank is to conduct monetary policy appropriately, then reasonable management information systems are imperative," the report said. "But the past decade—the New Economy—should have taught us this: The apparatus we currently employ for making sense of the economy-that is, the measurements we employ to distill information about the American economic enterprise into a comprehensible form-are simply inadequate to the task."

The Dallas Fed's report said, "The high-tech sector has been one of the major drivers of growth of the Texas economy in the past decade. In order to track the performance of the high-tech sector in the region we need a dependable and up-to-date measure of high-tech activity."

To gauge opinions and activities on this issue within the Fed, the Minneapolis Fed sent an informal e-mail poll to economists at the 11 other Fed banks, all of whom are involved in compiling the Beige Book, an anecdotal summary of economic conditions prepared for each Federal Open Market Committee meeting. Nine of 11 districts replied, although responses from two districts provided little information on the high-tech tracking issue.

The seven remaining responses indicate there is uneven attention and concern about this issue among district banks. Very few [including the Minneapolis Fed] have formal efforts to systematically track high-tech sectors—again, largely because Fed data gatherers depend on other organizations for much of the raw data that goes into setting monetary policy.

"St. Louis does not do high-tech tracking [via an index or other measure] ... nor have we considered it or discussed it. This is probably because the high-tech sector as narrowly defined is relatively small in our district," said Howard Wall, research officer and regional economics coordinator for the St. Louis Fed. According to economist Tim Schiller of the Philadelphia Fed, "We are currently researching high-tech in our area, so our monitoring efforts for this industry are still evolving."

But the high-tech industry is itself evolving, and at a much faster pace, which complicates matters further. "In our view, the main constraint to tracking the high-tech sector is that it is not clearly defined and changes over time," according to Jason Bram, an economist with the New York Fed.

For example, printing and TV broadcasting used to be considered high-tech, but no longer. If e-commerce is considered high-tech, then Wal-Mart would qualify because of its online merchandising and state-of-the-art inventory and information systems.

"If some aspects of a firm's business seem high-tech and others seem low-tech, as in these examples, how do you treat it?" Bram said. "This is why we feel that defining the sector is the biggest hurdle. If what you are really looking for is industries with strong future growth potential, then your best bet would probably be to come up with a set of rapidly growing industries using BLS growth projections."

Anecdotally speaking

The motivation for districts to track any information on a region's economy "is to provide data to the [district bank] president for FOMC meetings," said Yolanda Kodrzycki, assistant vice president and economist with the Boston Fed. But with little hard data available, districts are trying new strategies to fill the information gaps.

The Dallas Fed created its own index for high-tech output in three areas: computers, telecommunications and electronic components. Though the index is still in the experimental stage, "we have a measure where we know whether [high-tech] is adding to [district] growth," said Mine Yucel, Dallas Fed assistant vice president.

To improve its district Beige Book summary, Kodrzycki said the Boston Fed has added additional contacts in high-tech fields with a significant presence in the district, like computer makers, semiconductors, software and IT. "It's sort of a natural to beef up coverage of those areas."

Rather than invest heavily on internal research on high-tech, Kodrzycki and others at the Boston Fed are involved in outside efforts with similar research goals. She is on the advisory boards of Massachusetts Technology Collaborative and Mass High Tech, the latter of which is a regional journal that tracks and publishes various high-tech indicators. Such external efforts are a "really low-cost way to develop measures that are useful to us" at the Boston Fed, she said.

At the Atlanta Fed, "We're beginning to devote more efforts to tracking the high-tech sector in the district, particularly in Atlanta, in part because state officials in the Southeast are devoting considerable resources to attracting and retaining jobs in that sector," said Madeline Zavodny, an Atlanta Fed economist. In the past, she said, "we have primarily relied on newspaper accounts and reports from our branch directors for information on the high-tech sector. We hope to develop other methods but aren't yet sure how to best track this sector." An analyst with direct experience in the high-tech sector was recently hired and "is now helping us figure out how to best track it and to develop contacts for us."

At times, high-tech's volatility can make it difficult to find and maintain reliable industry contacts, said Yucel of the Dallas Fed. The industry is "very secretive," she said, adding there can be "great difficulty establishing contact for even a few years" with the same individuals.

But by the same token, "being the Fed puts us in a special position" to attract leaders in the high-tech industry to a district bank's board, according to Harvey Rosenblum, senior vice president and research director for the Dallas Fed. "They give us an entree we might not otherwise get" regarding conditions in the high-tech sector.

Rosenblum also pointed out that high-tech is not the first industry to pose measurement problems. When he was at the Chicago Fed years ago, Rosenblum said, researchers typically looked only at mature industries with easy-to-measure output. While useful, he said such efforts "were not measuring anything that was new to the economy." So he helped create what he called "the shadow Beige Book" that looked specifically at new businesses and industries.

In a similar vein, tracking high-tech was not as important as the underlying strategy to seek out and understand new and changing segments of the economy. "This is something the Fed is always working on," Rosenblum said. The Dallas Fed's high-tech index "is part of trying to keep a handle on it."

And efforts to better track high-tech will likely continue to engage Fed economists, academics and business people—just a step behind the industry they are trying to capture.

Code Name: High-Tech

Many studies on high-tech use the government's Standard Industrial Classification (SIC), which breaks down industries and industry groupings into increasing detail using corresponding two-, three- and four-digit codes. Various data on private industry-for example, employment and sales—are collected by the government at each SIC level, which many researchers use to analyze economic and business trends.

The industries selected as high-tech so qualify only because a researcher or study has determined them to have some correlation to what is commonly believed to represent high-tech. All the industries listed at right have been considered high-tech in at least one previous study.


28 Chemicals and pharmaceuticals
     281 Industrial chemicals
            2812 Alkalines and chlorine
            2813 Industrial gases
     282 Plastics material and  synthetics
            2822 Synthetic rubber
            2823 Cellulose manmade fibers
     283 Drugs

35 Industrial machinery
     351 Engines and turbines
     353 Metalworking machinery
     357 Computer and office equipment
            3571 Electronic computers
            3578 Calculating and
                     accounting machines
     367 Electronic components and accessories
            3671 Electron tubes
            3674 Semiconductors


Related Links