A new generation of researchers is out to discover how the U.S. economy really works

The debate keeps raging: How important are small, growing companies to the U.S. economy? Do they really create most of the new jobs? There's a lot of bad information on the subject being bandied about -- but increasing numbers of researchers are assembling hard evidence, even going out and counting companies themselves. Let's talk with them and report what they're learning. -- J.C.

At first glance, it could be just another scholarly skirmish -- a dispute among professors over the reliability of certain job-generation statistics. But the real subject of the debate is something far more important: the role of entrepreneurship in the revitalization of the U.S. economy.

The statistics in question first surfaced nearly 10 years ago, when Massachusetts Institute of Technology researcher (and INC. columnist) David L. Birch wrote that small businesses create 82% of new U.S. jobs. It was a preliminary number, based on a pioneering method of study, and it has long since been revised and refined to reflect variations of time and place. But Birch's critics are still gnawing on that old figure. "Small businesses do not create 82% of all new jobs," trumpeted a headline on a May 1988 Chicago Enterprise article by George Kalidonis, a professor at the Keller Graduate School of Management, in Chicago. Alleging "faulty methodology" -- misinterpretation of data and other statistical sins -- Kalidonis the following month announced that 82% was a "dubious claim." In April two Harvard economists published an article in The Washington Post purporting to debunk small-business mythology. And what was Myth #1? You guessed it -- that small businesses create 8 out of 10 new jobs.

As it happens, these particular critiques themselves reflect a breathtaking combination of misunderstanding and sloppy research (see "Birch's Revenge," page 6). But the critics' ineptitude is perhaps less telling than the passion and thrust of their arguments. If there's going to be a backlash against the Age of Entrepreneurship, we may be witnessing the opening salvos.

Consider: For most of the postwar era, the well-being of the U.S. economy seemed to depend on the health of its large companies. When some of the biggest ones found themselves battered by foreign competition, the economy as a whole felt shaky. But in the 1980s the picture changed dramatically -- and quickly. Hot new growth companies flooded the market with everything from clothing to computers. Wall Street's over-the-counter market woke up. Venture investing boomed. In this climate, David Birch's widely publicized figures about small-company job generation were electrifying. They alerted both Congress and local economic-development officials all over the nation to the importance of these new companies, thereby helping to spawn a range of policies favorable to entrepreneurship. They encouraged big corporations to seek out new businesses as customers, suppliers, and joint-venture partners. Before long, small business wasn't just an economic sideshow. It was the Main Event.

But the change in the program has left some members of the audience feeling restive. A front-page New York Times story last June reported a "reappraisal" of the value of entrepreneurship by a "growing number of influential scholars and business executives." The reappraisal, said the article, "marks a sharp departure from the years in which entrepreneurs were lionized by virtually everyone as innovators, and small companies were praised as the source of most new jobs." Washington, too, has witnessed a noticeable cooling of enthusiasm. "People on Capitol Hill used to sit up and take notice when you showed how a program would benefit small business," said one recently departed Small Business Administration official. "But lately I've been getting the response, 'Don't talk to us about helping small businesses. They already get too much.' "

In this context, potshots such as those against Birch echo well beyond the halls of academe. For the reality of entrepreneurship is that it flourishes best in a climate of support -- when financial institutions provide capital, when large corporations pay attention to small ones, when governments at least don't obstruct company founders and at best offer them helping hands. That climate of support, in turn, depends in good measure on what scholars tell us about economic reality. If someone were to show that the whole entrepreneurial phenomenon is overblown and overrated -- that the U.S. economy still depends, as it has for most of this century, on the health of its giant corporations -- then the climate would turn chilly. And a lot of fledgling companies would be frozen out of the marketplace.

All of which makes the work of an eclectic collection of researchers -- sociologists, political scientists, economists -- of more than passing interest. For according to the mosaic of information this group is beginning to assemble, the entrepreneurial revolution may run far deeper than anyone has so far suspected.

The mosaic begins with a few basic pieces. How many new businesses are there? Where are they, and what are they doing?

Surprising as it may seem, no one yet knows definitive answers to such questions. No government agency disseminates comprehensive information on new companies nationwide. Dun & Bradstreet Corp.'s widely quoted "new incorporations" figure includes numerous entities incorporated purely for legal or tax reasons. The most widely used source of data is Dun's Market Identifiers (DMI) listing, which provides information on the roughly 6 million active companies in D&B's credit files. Birch's database begins with the DMI. So do the two databases maintained by the SBA.

If the DMI were a true census, it might be sufficient. In fact, it is limited in scope and uneven in quality, as even its defenders acknowledge. Many companies, particularly in the service sector, are likely to be a couple of years old before they come to Dun's attention, and smaller ones may never be listed at all. Since Dun's reporters are mainly interested in financial data, the information they gather tends to be spotty on matters such as start dates and sales figures. Birch and other researchers do their best to weed out obvious errors, to estimate the number of missing companies, and to correct for statistical quirks. But skeptics abound. "D&B," asserts economist Jonathan S. Leonard, of the University of California, "is pretty lousy data."

All this may explain how Paul Reynolds found himself counting companies. Reynolds, a soft-spoken sociologist, works at the University of Minnesota. Several years ago -- "in the midst of a career change," he says, which in academia means shifting from one research interest to another -- he began thinking about the effects of entrepreneurship. Was it really making a difference to local and regional economies? How many new businesses were there, anyway?

As it happened, the people at the university's Center for Urban and Regional Affairs were interested in the same questions. So in 1984 Reynolds got a grant for a small pilot study, gathering information on about 550 new businesses in Minnesota. Two years later he contracted with the Appalachian Regional Commission, a 13-state development agency that is based in Washington, to do a full-scale survey of new-business creation and job generation in Pennsylvania. At the same time, he undertook a similarly complete study back home in Minnesota, this one funded by 10 separate agencies and institutes.

"Paul Reynolds," a colleague says admiringly, "knows more about this stuff than anyone else."

Reynolds set out both to measure the extent of new-business formation in the states he was studying and to learn more about the companies being created. His starting point, he figured, had to be the DMI. "It's an imperfect but necessary data source," he says. "There's really nothing else." But he knew too that he wanted firsthand information, gathered and verified by a staff he would hire and train himself.

By the time of the second Minnesota study, begun in early 1986, he had the methodology pretty well worked out. First, he ascertained that the DMI listed some 24,000 companies in Minnesota started between 1979 and 1984. Dividing that group by region, industry, and start date, he selected a representative sample of more than 5,000. The plan was that he and 10 researchers would call every company in the sample. When they reached executive officers of a business, they'd ask them a few screening questions, such as when the company was founded and whether it was independently owned.

Calling began in late fall of 1986. As the reports piled up, though, Reynolds noticed the same disconcerting phenomenon he had observed in his other studies. The D&B listings just weren't checking out very well. No, we're not a new company, some respondents would say. We just changed ownership, or restructured, or whatever, in the year you thought we started. Others had gone out of business. Still others couldn't be found at all or were invalid listings, such as duplications. All told, fully half of D&B's supposed new businesses didn't check out. That was almost exactly what Reynolds had found in Pennsylvania and in the earlier Minnesota study. If the case for entrepreneurship rested on the DMI, so it appeared, it was resting on a slim reed indeed.

But Reynolds wasn't through. Companies that qualified as both new and independent got a 16-page questionnaire. How many employees had they hired? Where did they do most of their business -- just locally or in wider markets? Who founded the company, and why did it locate where it did? If the companies didn't respond to a couple of mailings, Reynolds's researchers got on the phone again and again, until they had good data on 75% of the group. Then it was time to put the information into the computer and see what it said.

What leaped out at Reynolds, in the end, wasn't the startling inadequacy of D&B's statistics. It was the startling importance of entrepreneurship even after half the apparent new companies had vanished into thin air. Like Pennsylvania, Reynolds discovered, Minnesota depended heavily on new-business formation for its economic growth. Translating his sample into statewide figures cut the number of new companies from the 24,000 listed by D&B to 12,000. But these new businesses alone accounted for 119,000, or 42%, of the 282,000 net new jobs added to Minnesota's economy between 1978 and 1986.

That figure was significant, particularly since it now had the kind of scientific validation D&B alone could never have provided. But it was still a dramatic understatement, because it didn't include any of the companies D&B missed. Reynolds made no independent attempt to estimate that number himself, relying instead on extensive calculations carried out by David Birch and Susan MacCracken a few years earlier. If the Birch-MacCracken estimates were accurate, Reynolds reported, new companies in Minnesota accounted for essentially all of the state's net job growth.

In analyzing the questionnaires, moreover, Reynolds realized that he was uncovering two distinct kinds of entrepreneurship -- and that only one of them was really responsible for growth. Nearly three-quarters of the companies his staff contacted grew slowly -- less than $100,000 in new sales per year -- or not at all. Their emergence, Reynolds says, "tends to reflect a turnover or replacement process." But another group of companies seemed to be filling niches that hadn't existed before, as reflected in significantly faster growth. These fast growers -- concentrated in the Twin Cities area -- made up only 25% of all start-ups. But they accounted for 60% of new jobs attributable to new companies, 80% of new-company sales, and 80% of new-company "exports," that is, goods or services sold outside of Minnesota. They were thus responsible for a vastly disproportionate share of the development that was attributable to entrepreneurship.

Paul Reynolds is one of the leaders of what might be called the "counting-companies" school of entrepreneurship studies. It's a demanding art, if only because it requires the practitioner to scare up large grants. As a result, its adherents are few. Even so, it promises answers to questions that have vexed public officials and businesspeople since the dawn of capitalism. Do tax rates affect new-company formation? (No, says Reynolds, they make no difference.) Do people move into a state to start a business? (No again -- they start companies where they live). Does a region need manufacturing to prosper? (Doesn't seem so. William Beyers, a geographer at the University of Washington, contacted some 2,000 service companies in the Seattle area and asked about the extent of each company's exports outside the Seattle area. His conclusion: "The growth of our regional economy is being fueled as much by the export of services as by goods.")

At least some of the questions, though, may be answered more quickly -- and less expensively -- by a group of social scientists who are already mining a big, computer-readable, and largely untapped source of data about new businesses, compiled on a state level and completely independent of Dun & Bradstreet. Among this group is the University of Michigan's John E. Jackson.

Jackson, a political economist, had been doing research at Harvard during the early '70s, when the Massachusetts economy was still smarting from the southward migration of textile and other old-line manufacturers. "When I moved to Michigan in 1980, I heard the same story," he remembers, "only you had to substitute Japan for the Carolinas and automobiles for textiles. But now people were saying, why couldn't we be more like Massachusetts?"

With good reason. Between 1978 and 1982, Michigan lost close to 400,000 manufacturing jobs, more than half of them in the automobile industry alone. At the depth of the 1982 recession, unemployment hit an incredible 17%. "Residents were bombarded almost daily with stories of firms and people leaving Michigan for the Sun Belt," Jackson says. Asked to serve on an economic-development task force organized by Democratic governor James Blanchard, however, he realized how little was really known about the Michigan economy. That was when he began looking into the state's unemployment-insurance data.

ES-202 files, as state unemployment records are known, record every company operating in a state from the moment it establishes a payroll. They're updated quarterly, with new employment figures for every company. And they record company failures the moment a business stops incurring tax liability. A few states don't allow outsiders access to the files, while in others the files aren't easily accessible. Michigan offered no such obstacles -- and the data, it turned out, were of high quality. "Not only are the data more timely than Dun & Bradstreet's," Jackson explains, "but Michigan is particularly good at flagging predecessor and successor companies [when a company is restructured or changes ownership]. D&B doesn't do that so well."

Uncertain as to what he would find, Jackson got tapes beginning in 1978 and ending, for the first phase of his research, in 1984. Running them through his computer, he found that the press had missed fully half the story.

Yes, the automobile industry was shrinking. Even during the post-1982 recovery it never came close to replacing the number of workers laid off in the recession. And yes, many other industries -- steel, machine tools -- had plenty of dying or declining companies. But side by side with this decaying economy was what amounted to another economy being born. New and growing companies were generating jobs by the tens of thousands. In manufacturing alone, nearly as many companies were created as went out of business. Of the manufacturers that had survived over the six-year period, more than half increased their employment, and the average increase was nearly 50%.

To Jackson, it was an astonishing picture. Everyone knew that Massachusetts had plenty of young, growing companies -- but the Bay State was world famous as a hotbed of high technology. This was Michigan, a state that had always depended on durable-goods manufacturing and that was regularly berated for having one of the worst business climates in the nation. Moreover, though some of the new jobs were in predictable sectors such as computer services, many were in the same industries that seemed to be in decline.

In machine tools, for instance, Michigan began and ended the six-year period with roughly 1,850 companies. Behind that picture of stability, however, was a business maelstrom. Some 660 companies shrank. More than 400 went out of business entirely. But as many companies were born as died, and more than 700 of the existing companies expanded. Machine tools were not what Jackson or anyone else would call a sunrise industry, but it was, he found, a fertile ground for entrepreneurship and growth.

At the state level -- the only level, for the moment, where ES-202 data are useful -- such findings echo loudly. "Jackson's study helped convince us to keep supporting manufacturing," says Mark Haas, chief economist with Michigan's Department of Commerce. "We also learned to look not just at industries but at individual companies. Some businesses will be growing even in industries that are shrinking, and state economic-development policy has to take this into account."

What unemployment files or D&B listings provide are raw data, much like that in a census. The data tell the researcher how many businesses are starting up, what industries they're in, where employment growth is taking place. Surveys such as Reynolds's add information about the entrepreneurs and how they're building their companies. But such information isn't easy to evaluate. It raises as many questions as it answers. Would any state in any period show lots of new companies starting up? Do newer, smaller businesses have any advantage that older, larger ones lack? Historically, big companies in many industries have simply muscled out small ones. If that is changing, there must be reasons. The reasons, in turn, should shed light on how durable the current entrepreneurial trend is.

For better or worse, it's likely to be economists who answer such questions. They're the ones who are supposed to have both an overview of the economy and the theories to explain why it works the way it does. In the past, however, economics would have had surprisingly little to contribute. Most economists have been content to study inputs, such as employment trends and capital flows, and outputs, such as production figures and trade patterns, without examining the organizations that turn one into the other.

It may be no accident that Reynolds, Jackson, and nearly everyone else counting new businesses are not traditional economists. (In fact, David Birch was trained as a physicist.) When real economists study individual companies or industries, acknowledges Zoltan J. Acs with a smile, "they concern themselves almost entirely with the large companies."

Acs (rhymes with scotch) is one of a couple dozen academic and government economists who have been trying for several years to change this situation: to study how individual companies, small ones in particular, actually operate and thereby to evaluate the role that entrepreneurship plays in today's economy. As Acs and others describe it, creating such a field of study is remarkably similar to building a company. It takes money, a research technology, and a method of establishing yourself in the marketplace.

Money, in this case, was provided partly by the SBA's Office of Advocacy, created in 1976. Like many government agencies, the SBA disburses funds for studies it expects will bolster its mission. The technology, of course, was computerized data processing. "Only in the late '70s and '80s," explains Bruce Kirchhoff, former chief economist for the SBA, "were computers powerful and cheap enough to look at all the individual elements. You could tell the computer, 'OK, look at this list of a million firms and tell me which ones have added employees.' This was a revolution in methodology."

In academia, establishing yourself in the marketplace requires a combination of entrepreneurial marketing and rigorous quality control. Most of the entrepreneurship, everyone acknowledges, was provided by David Birch. "Birch did pioneering, path-breaking work," Acs observes, "and not only that, but he publicized it widely." Quality control, however, was another matter, at least in the view of some. Birch came under attack for the reliability of his data. He was criticized for not publishing his work in refereed economics journals, where it would be subjected to professional scrutiny before publication. "Mainstream economists are quick to jump down other people's throats for doing something differently," one says, "and Birch was doing things differently. A lot of economists were skeptical."

Slowly, however, much of the skepticism about this nascent field of study has abated. Two well-respected economists, William Brock, of the University of Wisconsin, and David Evans, of Fordham University, published a book entitled The Economics of Small Businesses in 1986, a volume that one practitioner says "put this field on the map." Brock, Evans, Acs, and others have published several articles in leading economics journals. The first issue of an international journal called Small Business Economics appears this month. Acs is one of two managing editors, and professional luminaries such as MIT's Lester C. Thurow serve on the editorial board.

As a field, small-business economics is still young. But the findings these economists have come up with, so far, shed light not only on how extensive a role entrepreneurship plays in our economy, but also on why. According to the new journal's lead article -- written by Brock and Evans -- we are witnessing a historic reversal in the century-old trend toward bigness in the U.S. economy.

The figures they cite are striking. For roughly the first three decades after World War II, the average size of U.S. companies increased. The self-employed -- a category that includes many small-company owners as well as sole practitioners -- dropped from about 20% of the work force to only 7%. The small-business share of value added dropped 4%, its share of employment 6%. Sometime during the '70s or early '80s, however, nearly all of these figures turned around. Average company size decreased, self-employment rose, and small companies accounted for a slowly increasing fraction of the labor force.

This phenomenon, other economists have discovered, doesn't simply reflect the continuing transition of the U.S. economy from manufacturing to services. On the contrary, the growing importance of small companies shows up as strongly in manufacturing as elsewhere. Edward Starr, an SBA economist, combed the agency's database of companies for the years 1976 through 1984 and found that small manufacturers' share of output and employment was increasing in nearly every broad industrial category. Nationally, according to Starr's figures, small manufacturers added about 1.2 million jobs during the period, while large manufacturers were cutting some 300,000 positions.

Bo Carlsson, an economist at Case Western Reserve University, in Cleveland, turned up even more striking data on changes in the metalworking industries between 1972 and 1982. (Metalworking includes machinery, aircraft, automotive industries, computers, and nearly everything else we generally think of as serious manufacturing.) During that time, both average company size and average plant size, as measured by employment, declined in nearly three-quarters of the category's 106 classifications, while the number of companies and plants in the industry as a whole increased more than 25%. In a working paper, Carlsson, Acs, and David Audretsch had shown that companies with fewer than 500 employees increased their share of metalworking production from 30% to 40% in 10 years, while companies with fewer than 100 employees jumped from 16% to 23%.

"That's a tremendous change," Acs says. "It shows there's a change in industrial structure that's taking place, from larger firms to smaller firms."

The reasons for this change are still largely a matter of speculation. An account of possible causes reads like a list of the past decade's megatrends: A culture of entrepreneurship. A more competitive economy. Deregulation. The fragmentation of markets. In effect, figuring out the causes of these ongoing changes is the task that the new field of small-business economics has set for itself.

In the meantime, some of the directions this research might take can be gleaned from the work of Carlsson. Back in 1982, as he tells the story, he was a visiting scientist at MIT and wondering what had gone wrong with the U.S. economy. It was buried in a recession. It had been battered by foreign competition. In such a climate, Carlsson felt, companies might be inclined to "deglomerate" -- to divest themselves of ancillary businesses and product lines, to shed divisions and departments, to trim their work forces, and to focus on their core markets. Certainly there was plenty of anecdotal evidence to support such a hypothesis -- but how could it be tested?

If the hypothesis was correct, he reasoned, the evidence would lie in measures of value added. A vertically integrated company adds immense value to its inputs, taking raw materials in one end of the factory and pushing finished products out the other end. In the new, more specialized environment, companies would be focusing on individual parts of the production process and therefore adding less value. Sure enough: when Carlsson tabulated the numbers, the ratio of value added to shipments had declined over a 10-year period in 88 of the 106 metalworking industries.

Carlsson had also begun an intensive study of machine tools while he was at MIT and had noticed another phenomenon. "For more than 150 years, most changes in production technology favored large-scale manufacturing relative to small-scale production," he says. By contrast, today's computer numerically controlled (CNC) machines operate economically at lower production levels and can easily be reprogrammed to turn out a wide variety of customized products. That gives small companies a key advantage.

John Jackson, who studied the Michigan economy, in-dependently confirmed Carlsson's thesis. Michigan's machine-tool industry, he discovered, was split right down the middle: declining or dying companies lost 18,000 jobs over a six-year period, while new or growing companies created 15,000 new ones. Jackson and two colleagues had surveyed the industry separately in 1984, asking not only about changes in sales and employment, but also about the type of technology used. By far the greatest growth, both in jobs and in sales, took place among companies that utilized only CNC equipment. These companies, moreover, were significantly younger and smaller than the industry average.

Ever since David Birch sat down with a yellow pad to report his job-generation results, the small-business community has been quick to perceive the political implications of this field of study. "I had never talked to a small-business person in my life, and I had no idea there were small-business advocacy groups out there," Birch says, recalling his astonishment at the reaction to his original study. "Suddenly there I was, getting an award as SBA researcher of the year." Agencies such as the SBA and lobbyists such as the National Federation of Independent Business quickly latched on to Birch's work. If small companies create jobs, their argument ran, the government should do all it can to help them. Studies such as those by Reynolds and Jackson created similar stirs at the state level.

Scholarly research and politics, however, are uneasy bedfellows. Advocates for small business, Birch points out, often neglect to mention that most small companies grow scarcely at all, and that the job-creating high fliers such as Microsoft Corp. and Mrs. Fields Inc. frequently don't have much in common with more traditional small enterprises. On the other side, Birch's own work is mistrusted by some academics because of his reputation as a partisan of entrepreneurship, who oversimplifies the world on behalf of new companies.

The issue comes into sharpest focus over job generation. Nearly everyone now agrees that many of the new jobs created each year are in small companies. But, ask critics such as James Medoff of Harvard, how durable are the jobs? How well do they pay? In general, Medoff and other economists have shown, large companies pay higher wages and provide better benefits than small ones. And at least some studies suggest that larger businesses are more stable employers. "If a job generated today is gone tomorrow, another job will have to be generated," Medoff says.

Birch critic George Kalidonis takes a different tack, charging that attributing job creation to the small company that happens to employ a new worker is an oversimplification. If an Illinois company sells its widgets to a California firm, Kalidonis argues -- and if the order requires new printed material, thereby leading to 10 new printing jobs -- it makes little difference if the jobs are in a big company or a small one. "The jobs," he writes, "were created by the California buyer who placed the order."

The trouble with such arguments isn't that they're wrong, but that they miss the point. In the recent past, big business really did provide high-paying, stable jobs -- in the steel industry, the auto industry, and many others. It can no longer do so. Most growing companies can't yet either. But at least they're providing some jobs to replace the 3.1 million eliminated by the Fortune 500 over the past eight years. Kalidonis's notion of a demand-driven economy, similarly, makes sense only if we assume it's always big companies that "place the orders" and that small companies can only respond to such markets. It allows no room for small companies buying from one another or for small companies creating new markets through innovation. Kalidonis is getting at the notion of an economic "base" -- that is, companies that export their goods or services outside their home region. But these days small companies as well as large ones fall into that category.

All this skirmishing is likely to continue. Entrepreneurship has had a long day in the sun, and it would be surprising indeed if it didn't experience a backlash. From an academic perspective, however, it no longer depends on pioneers such as Birch or indeed on any single school of thought. More people are studying it than ever before. What they're learning is not simply that small companies create jobs, but that our economy is undergoing changes affecting businesses of all sizes. Maybe they'll help us, over the next several years, understand those changes better than we do now.


The only thing you'll ever have to read about job generation

Pity David Birch. Not long after he wrote "The Job Generation Process" back in 1979, two researchers employed by Washington's Brookings Institution published a paper declaring him wrong. They had examined the job-creation data for 1978 to '80, they said, and discovered that small companies created only about 40% of the new jobs, not the 82% Birch had claimed for the years 1969 to '76. Maybe Birch had screwed up, they suggested, by confusing companies with individual workplaces. If IBM builds a new distribution facility and hires 100 employees, that's not a new business.

Today, many years later, the charge lives on. George Kalidonis, a professor at Chicago's Keller Graduate School of Management, refers in a recent article to "Birch's curious practice of counting branch offices of large corporations as 'small businesses.' " Harvard faculty members James Hamilton and James Medoff write in The Washington Post that "Birch generally defined business size by the number of people who work at a given location rather than the number working for the firm in total." Both cite the Brookings figure as evidence that Birch was wrong.

Well, Birch wasn't wrong. He never confused companies with establishments. On the contrary, he had undertaken "the massive task of assembling all the individual establishments into the 'families' (or firms) to which they belong." Eventually, the original disputants realized that their disparate conclusions were due partly to differences in their statistical methods and partly to real differences in the time periods they studied.

Methodological disparities persist. Researchers using the Small Business Administration's database -- the original basis for the Brookings study -- consistently come up with slightly lower estimates of job generation by small companies than Birch's. But the variation is typically small, and it's nowhere near as significant as the variation from state to state and from time period to time period. In recessions, for example, large businesses are typically cutting back their work forces, so small companies are likely to account for 100% of the net new jobs created. In boom times the proportion will be smaller. So it is with states. Minnesota, Massachusetts, and other states with many large, successful technology-based companies, writes Birch in his book Job Creation in America, find a larger share of new jobs coming from big companies. States such as Michigan, dominated by big-but-shrinking companies, find exactly the opposite.

So, what's the truth? Do small businesses generate 8 out of 10 new jobs? Only sometimes -- and only in some places. "When you think about it," says David Birch with a smile, "it's perfectly obvious."


Age: 51

Affiliation: Massachusetts Institute of Technology

Background: Physics

Claim to fame: pioneer in study of job generation

Quote: 'The small-business share of job creation is highly erratic. It varies widely over time and from one state to another.'

: 50

Affiliation: University of Minnesota

Background: sociology

Claim to fame: developer of field-tested database

Quote: 'It's a small fraction of these new companies that are making the real contribution. That's a controversial finding -- it implies that public agencies might want to focus their efforts on just these firms.'

University of Michigan

Background: political economy

Claim to fame: pioneer in use of unemployment-insurance data

Quote: 'Despite the worst depression in 50 years, almost as many firms started in Michigan as went out of business. In business services, there were three new firms for every two failures.'

: 48

Affiliation: University of Washington

Background: geography

Claim to fame: pioneer in studying service exports

Quote: 'We found we were looking into a bubbling cauldron of entrepreneurship, laced with expertise, specialization, and exploitation of market niches.'

: 46

Affiliation: Case Western Reserve University

Background: economics

Claim to fame: ground-breaking studies of effects of technology on manufacturing

Quote: 'Customers want variety. That undermines economies of scale -- the big firms can't keep those mass-production lines going anymore.'

Where the Jobs Come From

What percentage of jobs did companies with fewer than 100 employees create?

1969-72 82% 1978-80 38%

'72-74 53 '80-82 100

'74-76 65 '82-84 74

'76-78 56 '84-86 53

Source: Cognetics Inc., based on proprietary data and data provided by the Small Business Administration

The Role of the Fast-Growing Companies

How much of the start-ups' contribution comes from the fast growers?

Low growth High growth

All start-ups 73% 27%

New-company jobs 40 60

New-company sales 20 80

New-company "exports" 10 90

The Importance of Service Exports

How many jobs are Seattle's business sectors creating through exports?

Manufacturing 160,400 48%

All service 164,300 49

Resource extraction 2,100 1

Construction 5,400 2

Total number of export jobs 332,200

Small Metalworking Companies' Share of Total Sales

Has the long-term trend to large-scale production been reversed?

Fewer than : 1976 1982 1984 1986

500 employees 30.1% 30.2% 36.8% 39.7%

100 employees 15.6 17.2 21.4 22.5

The Balance Sheet on Manufacturing . . . 1982-84

How did Michigan's manufacturers do following the last recession?

Jobs created by new companies +25,000

Jobs created by existing company growth +177,000

Jobs eliminated by downsizing -39,000

Jobs eliminated by company failures -21,000

NET JOB GAIN +142,000