Inc. sets out to settle once and for all the Question: Who's really creating the jobs?

* * *

For more than 15 years, a debate has raged over small business's contribution to job creation in the United States. One side holds that small companies create most of the new jobs; the other, that small businesses play only a minor role in job creation. We've wondered how trained statisticians could look at the same numbers and draw such dissimilar conclusions, so we brought together leading experts on both sides of the argument to see if we couldn't at least learn why they disagree. What we found was that, as in most discussions of the new economy, even getting the questions right isn't easy, and there are no simple answers.

Inc.'s editor-in-chief, George Gendron, moderated a roundtable discussion among the four participants conducted on the Internet over a four-month period early this year.

Large-Company Experts
Steven J. Davis, professor of business economics at the University of Chicago's Graduate School of Business. Davis specializes in labor issues and macroeconomics. In his forthcoming book,
Job Creation and Destruction (MIT Press, 1995), he argues that large U.S. companies still create most new manufacturing jobs in the country.

Bennett Harrison, visiting professor of political economy at Harvard University's John F. Kennedy School of Government, in Cambridge, Mass. Harrison's book, Lean and Mean: The Changing Landscape of Corporate Power in the Age of Flexibility (Basic Books, 1994), explains how rapid globalization of world markets makes large multinationals the engine of economic growth.

Small-Company Experts
Paul D. Reynolds, Coleman Foundation chair holder in entrepreneurial studies at Marquette University, in Milwaukee. Reynolds has been studying start-ups and measuring the impact of their growth on local and regional economies since 1983.

Zoltan J. Acs, visiting professor of business and public policy at the University of Maryland's College of Business and Management, in College Park. Acs's research focuses on small companies as the prime innovators of the economy. He also coedits Small Business Economics , an international journal devoted to understanding the role of small business in the economy.

* * *

George Gendron: Ever since David Birch first reported on small-business job generation in 1979, a growing body of economic literature and the conventional thinking have held that small companies account for a disproportionate share of all new jobs. Recently, however, other academics have challenged this view, but so far the main effect of the debate has been to confuse, not to illuminate.

I'm asked repeatedly -- by elected officials and policy makers, by business leaders in companies large and small, by marketers -- "When it comes to all this job-generation stuff, what's the truth?" So my first question is, When it comes to all this job-generation stuff, what's the truth?

Paul Reynolds: The data from the United States and other developed countries are pretty clear: each company-size category generates jobs in roughly the same proportion to that category's share of total employment. In other words, firms with 500 or more employees employ roughly half of the U.S. work force and generate half of the new jobs.

Bennett Harrison: A tiny fraction of the companies started during a given time period account for most of the job growth that follows later. This pattern will vary by industry sector and by stage in the business cycle, but the hazard rate for start-ups -- what organizational-studies people call the "liability of newness" -- is huge. Dun & Bradstreet economists think that the small number of start-ups that do grow are born large; the maturation of the 10-person garage shop into a big firm is extraordinarily rare. I think that's right.

Steve Davis: What share of net job creation are smaller employers responsible for? The answer is highly sensitive to the time period, regardless of how you measure growth rates or employer size. To take an extreme and contrived example, suppose that in 1995, U.S. employment rose by 10 jobs, and small-business employment rose by 10 jobs. Then you could argue that "100% of net job creation during 1995 happened in the small-business sector." Or if you wish to be shrill, you could say, "Small businesses accounted for 100% of job creation during 1995!"

Those statements would be true, but they convey a misleading picture of job creation in an economy with millions of new jobs created by both large and small employers.

Gendron: Well then, if you can use net job creation to convey any picture of the economy you want, what's a more useful way to look at the numbers?

Davis: It is the share of gross job creation accounted for by small companies that is closely related to their share of total employment. Small businesses typically account for a larger share of job creation in sectors where they also account for a larger share of jobs. So in one sense, small businesses account for a disproportionately large share of job creation: they show higher rates of gross job creation.

That's the good news. The bad news is that smaller businesses also show higher rates of gross job destruction. Put simply, employment levels are more volatile at smaller employers.

Gendron: Job destruction is always bad?

Reynolds: There are two basic processes that lead to job creation. One is a natural or base level of turnover among existing businesses: one restaurant folds, and another takes its place, perhaps in the same location. This process accounts for perhaps 75% to 80% of the turbulence among firms. The second process involves the redistribution of jobs across industry sectors: a typewriter-repair shop closes, but a shop that recharges laser-jet printer cartridges starts up.

Trying to determine the relative impact of these two processes is difficult in the short run, partly because the coding of economic activity -- the infamous SIC codes -- lags behind shifts in technology by decades. However, one indicator is the appearance of high-growth new firms, that small percentage of start-ups that receive so much attention for their job-creation potential. These firms reflect a shift in the economic structure -- new products, new processes, and new technology.

Acs: But it's important to remember that no matter what industry sector you look at, small firms never create fewer jobs than their share of employment. And some of those small firms, like Nucor Steel, Microsoft, and Wal-Mart, grow rather quickly into large firms that create lots of jobs.

Gendron: Why should people care about this job-generation debate in the first place? Is this the economic equivalent of a disagreement I've been having for years with a friend over who's the best defensive third baseman who ever lived? Or does the question about where jobs come from actually matter to anyone other than the participants? (For those of you who are interested, Clete Boyer was the best defensive third baseman ever.)

Harrison: Why does it matter? Mostly, I think, for policy-making purposes. Some policies, such as across-the-board subsidies for small companies, are less warranted if small firms are less important to economic growth. But other policies, such as helping small firms become more productive by, say, marrying them to demanding customer firms, may be more warranted. Perhaps the most important policy issue is whether, given this high hazard rate and weak level of job contribution among small firms, a policy to "grow lots of new start-ups" will be astronomically cost-ineffective.

Of course it was Clete Boyer.

Reynolds: The major objective is to have a livable, just society with an acceptable level of economic efficiency and adaptability. And jobs are a good indicator of whether a society is taking care of its members. Voters, policy makers, and social scientists count jobs.

From a policy perspective, the issue is whether you've optimized the capacity of the national economy to generate new firms -- either as replacement businesses or as new types of business. When there is greater international competition and a quickening pace of technological change, then a society that is unable to adapt quickly may become less competitive and face economic decline.

Gendron: OK, the debate does matter, and the policy stakes are high. That said, doesn't it also matter how we define "small business"?

Harrison: The SBA's definition of "small business" as one with fewer than 500 employees is part of the interpretation problem. More than 90% of all firms in the United States employ fewer than 500 employees. I use the Organization for Economic Cooperation and Development cutoff of 100. But whether you use 100 or 500 as a cutoff, you still find that the small number of companies that are large account for a disproportionate share of employment and of sales and output -- precisely the opposite of what the conventional wisdom says about small firms.

Reynolds: The United States should group businesses according to size classes. In Europe and at the Organization for Economic Cooperation and Development, economists distinguish among small, medium-size, and large firms -- medium-size generally refers to firms with 100 to 250 or 500 jobs. Much job growth may be taking place in this intermediate-size group of firms. My impression is that most policy makers conceive of small businesses as being in the 20 to 50 job range or even smaller than that.

Acs: The question of what constitutes a small firm is not a particularly interesting one. No matter how you define small firms -- as self-employed, 1 to 19 employees, less than 100, or less than 500 -- they are different from the giant corporations that have dominated the economy and still dominate in manufacturing. The problems shared by all small firms, however defined, are usually different from those shared by very large firms. This, not the exact number of employees, is the important distinction to make, in my mind.

Davis: I agree with the thrust of comments by Harrison and Reynolds, who argue that the small-large distinction is a crude one. When suitable data are available, it's more informative to pose questions such as "What's the relationship between size and something else?" rather than "What's the difference between small and large?"

The SBA's definition of fewer than 500 employees, is, by the way, ludicrously out of step with the image that most people have of small business. It also happens to be out of step with the notion of smallness reflected in the U.S. legal and regulatory system. Many laws and regulations explicitly treat small and large employers differently. The small-large cutoff usually varies from 10 to 100, with 50 being a common value.

Gendron: Let me pose a hypothetical situation. Let's say you're a fact checker assigned to work on Bill Clinton's State of the Union address, and you come across the line "Small business is the engine that drives the economy and generates 80% of all new jobs." How would you change the line to make it accurate? For example, would it be accurate for the president to say, "Small business creates jobs in approximate proportion to their current distribution in the economy"? -- thereby guaranteeing that you remain a fact checker and not a speechwriter. Can you agree on a line to insert in Clinton's address that would, once and for all, set the record straight?

Reynolds: Language for Clinton's -- or Newt's -- political speech: New and small firms play a vital role in the growth and adaptation of the economy, helping the United States maintain a competitive international posture. In the process, these firms provide career opportunities for many. When all businesses have a fair chance to succeed or fail, we will know that our economy is maximizing its potential for change and growth. The federal government should ensure that new and small firms are not prevented, by unfair practices or inappropriate regulations, from demonstrating their value in the marketplace.

Four sentences, but a beginning.

Davis: If I ever run for office, I'll hire Paul Reynolds as my speechwriter.

Gendron: As far as I know, no one has yet invented a more efficient means of moving assets from unproductive to more productive parts of the economy than the start-up. So it makes sense that during times of rapid economic change, you want to foster an environment that's conducive to people's launching new businesses. Ben Harrison and Paul Reynolds seem to disagree over this point.

Can policy encourage new-business formation? Should it? If so, how? If not, why not?

Reynolds: The overwhelming factor that encourages new business formations is the perception that there is a market to be served.

The most governments can expect to do is either encourage firm formation once others decide to start a new business or try to eliminate inappropriate barriers.

Two barriers present in the United States that are not present in other advanced economies are the methods of providing health and retirement benefits. Start-ups may have a better chance to succeed if federal and state governments take more responsibility for citizens' retirement and health care. This reform would allow fledgling companies to attend to running a business instead of dealing with social-welfare issues.

Harrison: George, you suggest that Paul and I disagree on whether "during times of rapid economic change, you want to foster an environment that is conducive to people's launching new businesses." No, we don't disagree. I have no problem with market incentives for new-firm formation. Let private business start up as many new companies as it wants.

My concern is with using public policy to subsidize this behavior, especially given the high rate of expected failure among start-ups, which means that you have to support the creation of a dozen businesses to retain just one. The opportunity costs of the tax breaks, subsidies, and so on are too high.

Davis: Policy should encourage the allocation of capital and human resources toward their most productive uses. If that involves channeling greater resources toward start-ups, so be it. If it involves channeling greater resources toward large existing firms, so be it.

Gendron: Clearly, there is great concern about job security in today's economy, especially as large companies downsize.

Acs: We are starting to envision the economy as one in which we are in a permanent process of restructuring. In both technology and strategy, we rely less on permanent structure and more on newly created structures.

Harrison: Amen to Zoltan Acs, on continuous, if uneven, change.

Reynolds: Downsizing is overemphasized. This focus reflects a preoccupation with the largest corporations and no understanding of the important role played by the medium-size firms.

Gendron: If downsizing is not the right issue on which to focus, then what is?

Reynolds: Well, what seems to have changed is the increase in turbulence -- births and deaths -- among firms. This change is perhaps more critical than shifts in the size or structure of firms. High turnover among firms and jobs can cause real dislocation and readjustment problems for the individuals involved. Further, it makes it difficult for individuals and families to count on firms as a stable source of health and retirement benefits.

Davis: I'm not aware of any evidence that supports Paul Reynolds's claim that turbulence -- higher rates of job creation and destruction -- has increased over time.

I'm skeptical about the usefulness of turbulence measures based on the turnover of firms and establishments for four reasons. First, the meaning of "firm" is ambiguous. Is a sole proprietor with no employees a firm? What if family members are the only employees? What if the firm is wholly owned by another firm? What if the firm is partly or mostly owned by a franchisor? Second, measuring firm birth and death is hard to do in a reliable, consistent way. Younger and smaller enterprises, the very ones that turn over the most, experience so many changes in their lifetimes that trying to track them accurately is extremely difficult. Third, changes in tax and legal codes are likely to alter firm behavior in a way that renders firm-based measures of turbulence less valid. And finally, employment- or job-based turbulence measures are intrinsically more interesting than firm-based measures because they inform us about the share of the work force affected by turnover. This brings us back to the issue of employer size. It may be true that almost all firms have fewer than five employees, but so what? We care about these firms not because there are so many of them but because they provide employment for a nontrivial share of the work force.

Reynolds: Comments regarding problems with measuring turbulence among firms are correct, but that doesn't mean that doing so may not have some real value conceptually. Approximate answers to important questions may have some value compared with precise answers to trivial questions.

Gendron: Let's expand the discussion of job security to include the nature of work itself. We've heard much in the past year or two, particularly from the secretary of labor, about all the "bad" jobs being created in the United States, mostly in the service sector and mostly, we're told, in small companies.

What's your definition of a "bad" job? Since bad is a relative term, what's your definition of a "good" job?

Reynolds: Most of the job comparisons between small and large firms are unable to show that jobs in small firms are -- as a rule -- societally unacceptable. While they do show that some features, like pay, benefits, and tenure, of jobs in large firms are more desirable, they do not suggest that rational adults should, for their own good, be prevented from working for a small firm.

The problem in defining a good versus a bad job comes from the wide range of perspectives of those doing the work. Some workers may consider a job acceptable even if it is dull and demeaning because the pay and benefits are high enough to offset these problems. They are satisfied given the alternatives. Others may find a job with much risk and few benefits quite satisfactory because it offers challenge and potential for great rewards -- psychic and material.

From society's perspective, good jobs minimize the possibility that individuals will try to violently overthrow the status quo, and they allow for adaptation and change in the system.

I have been teaching a course on entrepreneurship research to part-time M.B.A. students for the past three years. Each class has about 25 young adults who hold full-time management positions. When we get to the discussion of firm survival, I ask the students to indicate whether they expect to have the same employer and the same job five years in the future. Much to my astonishment, virtually no one says yes, although a minority -- 10% to 20% -- say they are not sure; most students -- 80% to 90% -- unequivocally say no. Very anecdotal but dramatic evidence that norms about "good" jobs may be changing at a conservative school -- Jesuit Marquette -- in a conservative region -- Milwaukee.

Acs: For me a bad job is one that does not pay enough to move above the poverty line ($10,000 to $15,000 a year for a family of four), does not pay benefits, and offers little job security. A good job is one that pays a living wage ($30,000 a year with benefits) and carries some amount of job security.

Gendron: Assuming you can meaningfully define a bad job in the first place, is it accurate to say that the proportion of bad jobs is increasing? How do we know this?

Acs: It's not that the number of bad jobs is increasing but that good jobs that do not require a high level of education and skill are becoming harder and harder to find.

Davis: Zoltan makes an important point that I would like to amplify. Over the past 20 years, and especially since 1980, the opportunities for someone with a high school education or less to obtain a well-paying job have greatly diminished. Economic data show a sharp drop in wages for less-educated males, a finding confirmed in numerous studies and now widely recognized by academic researchers.

The data for the 1980s also show a decline in consumer spending among households headed by less-educated men, still another sign that this portion of the population has been experiencing a lower standard of living.

Reynolds: In the sense that a larger share of jobs do not provide full-time work, full health and retirement benefits, guaranteed employment for the duration of an individual's work career, and fulfilling work, the proportion of unacceptable jobs may have increased in the past 20 years.

Gendron: If the system is creating more "unacceptable" jobs now than in the past, why is this happening and what, if anything, can we -- policy makers, that is -- do to encourage the creation of more good jobs?

Reynolds: One appropriate role for local, state, and federal governments would be to enact regulations that prevent the creation of unacceptable jobs. Beyond this, the most important thing to do about bad jobs may be to separate the provision of health and retirement benefits from the job itself. Some cooperative mechanism could provide minimal health and retirement benefits to all individuals; this would be a second role for governments. Guaranteed access to health care and retirement programs would allow individuals to focus on finding fulfilling work instead of being forced to take a job just because it offers desirable benefits.

Constant churning and turbulence in the economy suggest another appropriate role for government: to help individuals and firms make the transition from one set of job requirements -- skills, training, and information requirements -- to another.


How to Read Between the Lines

George Gendron: [Below] is a news item that appeared in the December 1994 edition of the Small Business Advocate, published by the SBA's Office of Advocacy.

What, if anything, does this item tell us that is meaningful and trustworthy about what was happening in the economy during the period in question? Does it tell us anything? Would we need to know more even to evaluate its contents? If so, what information would we need that we're not given?

New Data Show Smallest Firms are Nation's Greatest Job Creators
New data from U.S. Bureau of the Census show that the nation's smallest firms -- those with 0-4 employees -- created virtually all the net new jobs between 1989 and 1991.1,2,3

Tabulations prepared by the Office of Advocacy, based on the Census data, show that most of these jobs were created through expansions by businesses that were on average at least two years old. Firms were classified by their size in the base year, 1989.

Chief Counsel for Advocacy Jere W. Glover said, "These figures clearly show that microbusinesses, not large businesses, create the jobs in our economy. And these are businesses that have staying power.4,5,6 They don't just start up and disappear."7,8

The Advocacy analysis of industries -- services, wholesale and retail trade, agricultural services, mining, construction, manufacturing, transportation, and finance, insurance and real estate -- reveals that microbusinesses were the ones that added employees while firms in most of the larger size categories lost workers.

Some key findings of the analysis include:

· The smallest businesses (0-4 employees) created 2.6 million net new jobs. Large companies (500+ employees) created 122,000. All other business size classes lost jobs.

· In manufacturing, only the smallest businesses (0-4 employees) added jobs.

· The greatest growth was seen in the services industry. During the 1989-1991 period, the services sector added 1.9 million jobs. All but 475,000 came from small firms (those with fewer than 500 employees). Approximately 1.1 million came from firms with 0-4 employees.

"The disturbing aspect of these new data is that all the other small business size categories -- those with between 5 and 499 employees -- showed net job losses during this period," said Glover.9 "Without the growth and expansion of the firms with 0-4 employees, the recession would have been much worse, and there could have been over 2 million more unemployed workers."10

1 "The data were developed and provided by the Bureau of the Census, not Dun & Bradstreet, which suggests that they are based on Social Security files. They will therefore exclude all businesses that do not file FICA payments, including sole proprietors and the self-employed." -- Reynolds

2 "This claim is another instance of SBA hokum about small-business job creation. The study underlying it classifies all firms born after 1989 in the 0 to 4 size class, regardless of how many employees a firm had in 1991. Gross job creation at new firms is large relative to changes in net employment over the same period. The SBA could not have found anything other than the 'result' that 'microbusinesses' accounted for a large portion of net new jobs." -- Davis

3 "The SBA claim may be hokum, but how much hokum is there in it? It is true that the SBA classified all firms born after 1989 in the 0 to 4 size class, regardless of how many employees a firm had in 1991. However, when it classified firms by the correct firm size category in 1991, it was still the 0 to 4 category that created jobs." -- Acs

4 "There's no way to evaluate this comment about staying power. Assume at least half, perhaps three-quarters, of businesses survive for five years." -- Reynolds

5 "What's the basis of this claim about 'staying power'? The SBA study of census data follows firms only from 1989 to 1991, so how can the study draw any inferences about long-term staying power? Every careful study I know of finds that smaller and younger firms offer less job stability than their larger and older counterparts." -- Davis

6 "What is wrong with short-lived firms if they make an honorable exit and have served a good purpose during their lifetimes? The potential for economic adaptation of the entire system may depend on the potential for higher levels of firm turbulence. God knows, those running the major established business firms don't seem to be able to adapt to change." -- Reynolds

7 "To the extent that a significant share of these are basically unsound self-ownership boutiques, shops, or consultancies triggered as a last resort by the recession and the growing number of displaced workers seeking to find something to do with themselves, we can predict that their longevity is likely to be short." -- Harrison

8 "I am saddened by the image presented by Ben that new firms, 'microbusinesses,' reflect the desperate acts of those unsuited to wage work. Most people trying to start new firms, over 80%, have full- or part-time jobs or are running another business." -- Reynolds

9 "It is possible that the net loss in intermediate-size categories (5 to 499 employees) represents a shift from existing, larger firms that are becoming obsolete and that are being replaced by newer, smaller firms in the same industries." -- Reynolds

10 "This is a ridiculous statement. If for some reason there had been no expansion by firms that the SBA classifies in the 0 to 4 category, then other firms would have expanded more. Suppose some economic or policy change drove mom-and-pop grocery stores out of business. Would people stop eating food and shopping at grocery stores? Of course not. They would shop at larger grocery stores, and employment at larger grocery stores would expand. This is not meant to belittle the economic contribution of new businesses (or small grocery stores, for that matter), but there is no basis for the suggestion that their elimination would have meant 2 million more unemployed people." -- Davis