Hey, Wait A Minute

Small-Biz Blarney

What does it take to kill a bad number?

Everybody knows that small business creates all the jobs. Why, only last week during the vice-presidential debate, Jack Kemp argued with evangelical passion for a small-business capital-gains cut to grow the economy faster. “How about the small businessmen and women of America that create 91 percent of all the new jobs?” he asked. A few weeks earlier, during the minimum-wage debate (in which small business was once again singled out for favored tax treatment), virtually every opponent argued that raising the minimum wage could hurt small businesses–which create most of America’s jobs.

This mantra is repeated so often that everybody believes it. But everybody is wrong.

What does it take to kill a bad statistic? More than a Nobel laureate, apparently. Milton Friedman, a conservative economist and proponent of free-market solutions, pointed out in a recent Journal of Economic Literature article that the myth of small businesses’ job-creating potency is one of the most durable falsehoods of America’s political economy. Numerous other economists–and even the occasional columnist–have tried to debunk the myth, but to no avail.

F or example, in his paper titled “On the Size Distribution of Employment and Establishments,” Jonathan Leonard finds that large firms are the main job generators. Three Census Bureau fellows, S. Davis, J. Haltiwanger, and S. Schuh, recently published a report titled “Small Business and Job Creation: Dissecting the Myth and Reassessing the Facts,” in which they assert flatly, “Conventional wisdom about the job-creation prowess of small business rests on statistical fallacies and misleading interpretations of the data.” (.)

Magazines and newspapers have recently run articles like “Debunking the Small Business Myth,” “Small Is Not Beautiful,” “Doing the Small Business Shuffle,” and “The Real Engine of U.S. Economic Growth Might Be Bigger Than Any Believe,” all of which have declared the small-business lobby’s claim “phony.”

Where the Bad Data Come From

Not only politicians but newspaper reporters continue to perpetuate the small-business fallacy, usually without challenge. Take an article in the LosAngelesTimes on July 6, which referred to “98 percent of all jobs since 1989 being created by businesses with fewer than four employees.” Where do these numbers come from? The Office of Advocacy of the Small Business Administration will gladly supply you with a table that might appear, at first glance, to lend support to such claims (see Figure 1).

Look across the second row, for example, which seems to tell you that firms with no employees in 1991 netted an astounding 1,898,600 jobs by 1993.

Figure 1

Net Job Creation by Firm Size, 1991-1993 (Data in Thousands)

Firm Size (Number of Employees in 1993)


*Computed by subtraction within the same size class.

** Firms with no employees, but some payroll expenses, in 1991.

Add in the jobs created by firms with one to 499 employees in 1991 (the next five rows), and you would compute that small businesses (defined by SBA as having fewer than 500 employees) added all but 44,800 of the total 2.4 million net jobs gained by the economy over that period.

Well, what’s the matter with that? The matter, of course, is that the rows broken out by initial firm size tell only part of the story–the happy part. What SBA doesn’t make clear in the table it hands out (though it will tell you if you persist) is that the firm-size rows only include those firms that “made it”–i.e., those that stayed continuously in business throughout the period.

And, of course, it is a fact of America’s dynamic marketplace that while many firms–especially small firms–open for business every year and hire new workers, almost as many close their doors, sending their workers onto the unemployment rolls. Overcounting gains and undercounting losses in this and other ways is what produces the grossly exaggerated picture of small business’s contribution to the U.S. economy. (Economists call the SBA’s technique of sorting firms by class size in a dynamic situation the “regression fallacy.” For a simple example of why it produces a misleading result, click.)

A Quick Fix

To get a better, though still static, picture of what’s really going on, let’s relabel and complete SBA’s table (see Figure 2):

Figure 2

Net Job Creation by Firm Size, 1991-1993 (Data in Thousands)

Firm Size (Number of Employees in 1993)


* Computed by subtraction within the same size class.

** Firms with no employees but some payroll expenses in 1991.

Source: Office of Advocacy, U.S. Small Business Administration, from U.S. Department of Commerce, Bureau of the Census Data, prepared under contract (table prepared July 1996).

Focus on the top row, the one optimistically labeled “Net Births” in SBA’s version. These are the net changes (mostly net losses) in job counts produced by firms that either opened or closed during the period. Looking down at the now-complete bottom-row totals, you will get quite a different picture of where the net new jobs are. By this calculation, some 1,673,400 (or almost 70 percent) of the jobs still around in 1993 were added by firms with more than 500 employees in that year–i.e., by “big” business.

Classifying firms by their end–rather than start–size might also seem unfair, since some very successful small companies may end up big by the close of the period, and vice versa. We need a more dynamic model to get a real sense of what’s going on.

Going With the Flow

Getting a good fix on this phenomenon isn’t easy, because most government data are collected to measure monthly employment changes, and there are no current data available on gross job flows by business size. As a senior economist at SBA, however, I was able to specify the preparation of special tabulations by the Census Bureau from business payroll tax reports for the entire private nonfarm economy. These tabulations make it possible, for the first time, to study job flows for the same firms as they migrate from one employment size class to another.

These statistics–never released by SBA and published here for the first time–tell a fascinating story of job creation and destruction. For example, the number of firms with employees increased by 30,000 in the period between 1989 and 1991. This end result, however, was produced by some 1,260,000 firm “births” (a firm reporting no payroll in 1989, but some payroll in 1991), which were offset by 1,230,000 firm “deaths.”

Looking at job flows by firm size (see Figure 3), we see that firms that remained small (including those that entered or went out of business) lost 192,000 jobs during this recession period.

Figure 3

Job Generation by Size of Firm in 1989 and 1991

(Numbers of Jobs in Thousands)


* Net jobs created by boundary crossers = 69,000


Source: Bureau of the Census, special tabulations

Employment gains from firms going from small to large were essentially offset by employment losses of firms that were large and became small. But on net, there was a job gain of 69,000 for “boundary-crossing” firms. At the same time, large businesses (again including those that opened or closed) added 802,000 to their payrolls. Subtracting the small-business losses from the big-business gains gives us a net total of 679,000 jobs added to the entire economy during this period. The analysis thus confirms what earlier studies had suggested: Big business, not small business, is the primary generator of U.S. jobs.

Predictably, SBA’s counts show a totally different (and false) picture (see bottom of Figure 3). By using the beginning-year counting rule, SBA reports that small business generated 557,000, or 82 percent, of the jobs. If the end-year counting rule is used, small business destroyed 872,000 jobs. Neither statistic is reflective of the job-generation process; they measure only economic volatility.

Of course, the period between 1989 and 1991 was one of recession. However, Census Bureau counts show that during the phase of rapid growth between 1982 to 1987, an equal number of firms in each size class reported job gains as losses. This same result held true in the period from 1989 to 1991, demonstrating again that, irrespective of business-cycle conditions, the probability that a firm will add or lay off workers is independent of its size. Analysis of the SBA net-change data for the expansionary period between 1991 and 1993 also reveals that small businesses created jobs in proportion to their numbers–no more, and no less.

A simple economic explanation accounts for these observations. Firms are constantly competing with each other. Both small and large businesses expand when there is a demand for their goods and services that they can meet efficiently. All the economic research suggests that there is no difference in the efficiency of small and large firms once a firm grows beyond a minimum size. It is for this reason that Census Bureau data for the last 50 years show no evidence of any change in the employment size distribution of firms and establishments.

All this might not matter, if the small-business fallacy were not used by its promoters as justification for carving exemptions and preferences for small business into virtually every piece of economic legislation that passes Congress. As the editor of this magazine wrote in the WashingtonPost in August 1993, “Small business is an important part of the American economy. Most small-business owners are admirable, hard-working, patriotic. But the case for tilting public policy in favor of ‘small business’ is based on logical fallacies and lobbyists’ hokum.”