Cognitive scientists refer to something called the “truth effect,” which occurs when people believe something simply because it is widely repeated. This is the case with regard to the claim that small businesses are the font of job creation.1 Ever since David Birch wrote in the late 1970s that small businesses are the job creators, this assertion has taken on mythic proportions, to the point that it is no longer even questioned. President Obama’s budget summed up the view: “Small businesses are the engine of job growth in our economy.”2
This is mostly wrong. But that does not stop small-is-beautiful advocates from continually making the claim. Indeed, job creation is their ace in the hole. Even if small businesses lag large businesses on every other performance indicator, small business advocates can always assert that at least small firms create the lion’s share of jobs. With too many workers still unemployed or underemployed, even after almost a decade of recovery since the global recession of 2008–2009, the claim that the secret to job creation is multiplying the number of small firms is a powerful card to play. But it is a myth.
The Origin of the Small Business Job Creation Myth
Where did this small business jobs myth start? For most of the postwar period, economists believed that large companies created most of the jobs. But this was in part because, in the absence of longitudinal data on individual firms’ employment levels, economists simply counted the number of jobs in each size class in one period and subtracted the number of jobs in the same size class in a prior period to see which size firms created the most jobs. The problem was that this assumed that the firms in one size class in the current period were the same firms as in the previous period. So if a firm had 200 workers in the initial time period but grew to 600 workers in the later time period, it would look as though big firms had created 600 jobs while small firms had lost 200.
This was clearly not the right way to measure job growth. In 1979 MIT professor David Birch decided to study individual firm employment records from Dun & Bradstreet. Birch purported to show that between 1969 and 1976, more than 80 percent of jobs were created by businesses with fewer than 100 employees and more than two-thirds by firms with fewer than twenty workers.
Unlike most economic studies, this became big news. No longer could economists and policy makers simply assume that large corporations were the big job generators. Very quickly the new conventional wisdom became that small firms were biggest generators of jobs. Now a host of small business preferences, from tax breaks to regulatory exemptions to procurement favors, could be justified, not by the old-fashioned argument that small proprietors are the pillars of a democratic society but because small businesses are the most important job generators, the backbone of the economy.
Since Birch’s initial research, some economists have found similar results, but many others have criticized his conclusions, finding different results. Still others have added the wrinkle that it is not small firms that are the big job generators but new businesses. This controversy notwithstanding, Birch’s claim has been endlessly repeated, like an urban myth, getting larger and larger, even being garbled into claims that small business is responsible for all job creation.
Why Small Business Is Not the Main Source of Jobs
Birch’s research has been criticized on a number of grounds. Birch himself said that his earlier results were a “silly number” and that he could “change that number at will by changing the starting point or the interval.”3 Some have criticized his work for a “regression to the mean” bias. In other words, a number of firms classified as large in the base period are more likely to have experienced a recent transitory increase in employment that made them large, meaning that these are precisely the firms that are more likely to contract in the next year, making it look as though large firms lose jobs. Likewise, some firms that are classified as small in the base year may have recently contracted, and because of regression to the mean are more likely to expand in the next year.
Other studies found less compelling or even contradictory results. In an attempt to replicate Birch’s work, the economist Catherine Armington found that from 1976 to 1982, small firms were responsible for 56 percent of new jobs, a far cry from over 80 percent and much closer to their actual share of total jobs.4 Other studies found that large firms created the most jobs. Looking at job creation in manufacturing from 1973 to 1988, the economists Steven J. Davis, John Haltiwanger, and Scott Schuh found that larger firms were more likely to create jobs.5 According to a 2010 study by Haltiwanger, Ron S. Jarmin, and Javier Miranda, large firms more than a decade old with more than 500 workers employed 45 percent of private sector workers and accounted for 40 percent of job creation.6 And a study by American Express and Dun & Bradstreet found that mid-market firms with revenues between $10 million and $1 billion were responsible for 92 percent of the net new job creation from 2008 to the end of 2014.7
More recent research has drilled down and found that it is not small firms per se that create most jobs, only new ones. Haltiwanger, Jarmin, and Miranda found that after controlling for firms’ age, “the negative relationship between firm size and net growth disappears and may even reverse sign as a result of relatively high rates of exit amongst the smallest firms.”8 In other words, it is not the size of the firm that matters in job creation, it is the age. Just as young children grow faster than adults, young firms grow faster than mature ones.
Nevertheless, a widely cited study for the Kauffman Foundation, a foundation devoted to supporting entrepreneurship, finds that all net job growth comes from firms less than one year old—in other words, startups.9 But the problem is that these new firms also destroy jobs as many go out of business soon after they start. As Ryan Decker and coauthors write, “Most business startups exit within their first ten years, and most surviving young businesses do not grow but remain small.”10 And Jonathan Leonard writes, “The obvious pattern, and one that has been largely ignored in previous studies, is that small establishments account for most net job loss just as surely as they account for most net job gain.”11 In other words, lots of new firms hire workers, but most proceed to lay them off soon, when they go out of business. This is why Haltiwanger and his colleagues at the US Census Bureau find that the median net employment growth for young firms is “about zero.”12 According to the Small Business Administration (SBA), just one-third of new businesses survive to their tenth year.13 Indeed, Zoltan Acs writes, “Some industries can be best characterized by the model of the conical revolving door, where new businesses enter but where there is a high propensity to subsequently exit from the market.”14
One study concluded that the smallest firms generate a slightly greater share of new jobs than their share of overall jobs (35.1 percent relative to a 27.2 percent employment share), though “there is stronger evidence that the smallest firms also generate a disproportionate share of gross job destruction (33.9%, relative to the 27.2% employment share).”15 This is why the correlation between the startup rate and the failure rate across industries at the three-digit industry code level is 0.77. In other words, industries that have the highest rates of firm startup also have the highest rates of firm failure.16 Davis, Haltiwanger, and Schuh rightly point out that “a common confusion between net and gross job creation distorts the overall job creation picture and hides the enormous number of new jobs created by large employers.”17
We see this in the fact that from 1993 to 2010, small firms with one to nineteen employees were responsible for 29 percent of gross job creation in the United States but only 15 percent of net job gains. In contrast, firms with more than 500 employees were responsible for 26 percent of gross job gains but 38 percent of net job gains. (Firms with 20 to 99 employees and firms with 100 to 499 employees were responsible for 23 percent each.)18 In fact, overall gross job gains per month were twenty-seven times more than net job gains, reflecting the enormous amount of churn in the labor market, particularly among small and new firms. Decker and coauthors find that that when only continuing firms are considered (leaving out firms that start and then die), from 1992 to 2005 large firms (more than 500 employees) created more jobs at every age than did firms overall. For example, in their eighth to ninth year of existence, large firms created jobs almost three times faster than all firms of the same age (4.5 percent growth versus 1.8 percent). Moreover, for firms older than sixteen years, small firms with fewer than fifty employees actually lost jobs, while large firms continued to grow.19
According to the logic of small business advocates, society should favor firms when they are small, but as soon as they add their 501st employee, they become the object of indifference or even derision. This is as perverse and unhealthy as the attitude of parents who hope that their children will never grow up.
Perhaps the biggest indictment of the small-is-beautiful view when it comes to jobs is the simple fact that in the United States small firms’ share of output and employment over time have been declining for decades. The share of sales accounted for by small firms declined from 57 percent in 1958 to 50 percent in 1982, while the share of workers employed in small firms in 1986 was slightly less than in 1958.20 More recently, since April 1990 private employment has grown by 17.5 million, or 19 percent. About 65 percent of these jobs were in firms with 500 or more employees in 1990, even though at the beginning of the period these large firms employed just 42 percent of workers. Just 19 percent were in firms with between 50 and 499 employees, and 16 percent were in smaller firms.
Far from becoming more important to the US economy, small firms are becoming less important. In fact, from 2000 to 2014, the share of employment in firms with fewer than 500 employees actually fell, from 53 percent to 51 percent. Moreover, the Bureau of Labor Statistics reports that “since its employment low in October 2009, employment in firms with less than 50 workers grew at an annualized rate of 0.8 percent through March 2011. In comparison, employment large firms grew at an annualized rate of 2.1 percent after reaching a low point in February 2010.”21 This is hardly evidence of the increased importance of small firms. In fact, according to the 2011 US Census Bureau’s Statistics of U.S. Businesses, firms with zero to four employees accounted for only 5.2 percent of all employment.22
In contrast, firms with 500+ employees, while constituting only 0.3 percent of all firms, accounted for 51.5 percent of all employment. Most of this employment came from the very largest of firms: those with more than 10,000 employees, while constituting only 0.016 percent of all firms, accounted for 27.8 percent of all employment.23
Finally, research shows that employment change in large firms is a larger driver of the unemployment rate than employment change in small firms. According to Giuseppe Moscarini and Fabien Postel-Vinay, “The differential growth rate of employment between large and small US firms is strongly negatively correlated … with the contemporaneous unemployment rate.”24 In other words, when firms with over 1,000 workers add more workers than firms with workers with fewer than fifty workers, the unemployment rate goes down (the correlation is −0.52). And the converse is true as well. This is why research shows that while small firms create more jobs during periods of high unemployment, they create fewer during periods of full employment. And it is why Moscarini and Postel-Vinay write, “The conventional wisdom that ‘small businesses are the engine of job creation’ finds some empirical support in our data only at times of high unemployment. … This statement clearly fails in tight labor markets.”25
Likewise, Zoltan Acs and Catherine Armington looked at the rate of firm formation between 1993 and 1998 in 394 US metropolitan areas and found that those metro areas that had higher per capita income growth in the prior year had more firm formation in the next year.26 In other words, firm formation was the result of growth, not the cause. And the cause often was the growth of large, export-oriented firms that brought more money into the local economy for spending on small, local dry cleaners, carpenters, and restaurants. In other words, large firms are the driver, small firms the result. Moreover, the authors suggest that one reason why small firms grow more in recessions is that small firms benefit from high unemployment, as that relaxes hiring constraints. In other words, workers who otherwise would want to work at large corporations that pay more and have better benefits now have no other choice but to work at small firms.27 The authors go on to note, “This picture corroborates only in part the common wisdom that small businesses are the engine of job creation: small firms appear to create more jobs as a fraction of their employment only when unemployment is high.”28 The authors found the same dynamics in Canada, Denmark, France, and the UK. For this reason, Scott Shane writes, “Many studies have shown than in places with more unemployment, and in time periods when unemployment is increasing, people are more likely to go into business for themselves than at other times and in other places.”29 For example, between January and December 2009 the number of self-employed Americans remained constant even as the unemployment rate increased from 7.7 percent to 10 percent.
The Startup Jobs Myth
But what about startups, the supposed source of American economic renewal? It turns out that most startups don’t actually create that many jobs either. As Shane shows, “Only 1 percent of people work in companies less than two years old, while 60 percent work in companies more than ten years old.”30 A study of UK startups found that of the more than 560,000 firms estimated to have employed fewer than twenty persons in 1982, 10 percent had gone out of business by 1984, and 88 percent still had fewer than twenty employees at that time. Only 2 percent of the 1982 cohort grew beyond twenty employees during the two years following their startup.31
The majority of small companies actually shed jobs after their first year. One study found that among small companies in their second, third, fourth, and fifth years of business, more jobs were lost to bankruptcy than were added by those still operating (see figure 5.1). This is why the mean number of workers per firm actually goes down every year after a firm is born. According to the SBA, the mean number of workers in a new firm in its first year is 3.07. But by year 5 this figure declines to 2.36, and to 1.94 in year 11.32 Or, as the SBA puts it, “Employment gains from growing businesses are less than employment declines from shrinking and closing businesses.”33 During the depths of the 2008–2009 recession, small businesses were adding an average of nearly 800,000 new jobs a month. But they were shedding an even larger number of jobs per month—about 971,000. In short, small firms create lots of jobs, but they also destroy lots of jobs. In light of increased concern about employment instability, this certainly can’t be a good thing, at least for the workers, half of whom lose their jobs.
Figure 5.1 Net Jobs Created by Firm Age, 2000–2013
Source: US Census Bureau, Business Dynamics Statistics (Longitudinal Business Database, Firm Characteristics Data Tables, Firm Age by Firm Size, 1977 to 2014), https://www.census.gov/ces/dataproducts/bds/data_firm.html (accessed March 17, 2017).
Statistics that claim to show that small businesses are responsible for the lion’s share of job creation frequently rely on data in figure 5.1. New firms by definition cannot lose employees from the previous year, and any employees on the payroll are credited as “jobs created.” By contrast, if a firm in its second year goes out of business, this is counted as negative job growth. From 2000 to 2013, only very young and very old firms showed positive net job growth.
The fact that young and adolescent companies don’t on net create many jobs cannot be blamed on regulation and high taxes; if anything, as we discuss in chapter 12, small business is pampered and protected when it comes to the taxes and regulatory burdens bigger firms face. Rather, most small business owners have no desire to grow their firms. Nearly three quarters of individuals who start a business want to keep their businesses small.34 Surveys show that the lion’s share of people who start businesses do so not because they want to be a rich entrepreneur, something that takes enormous dedication and hard work to achieve; rather, most don’t want to work for a boss.35 As Shane found, “One study of a representative sample of the founders of new businesses started in 1998 showed that 81 percent of them had no desire to grow their new businesses.”36
Another study found that 50 percent of small business owners did not start their business principally to make money.37 A Federal Reserve Bank study by Pugsley and Hurst noted the following:
When asked about their ideal firm size, the median response of new business owners is that they desire their business to only have at most a few employees. This is not surprising given that the overwhelming majority of small business owners in the US are skilled craftsmen (e.g., plumbers, electricians, painters), professionals (e.g., lawyers, dentists, accountants, insurance agents), or small shopkeepers (e.g., dry cleaners, gas stations, restaurants).38
This, combined with the fact that so many new firms fail within ten years, is why Shane has found that it takes forty-three startups to end up with just one company that employs anyone other than the founder after ten years.39 And on average, that surviving startup will have just nine employees. As Shane points out, “From 1992 through 2008, the 4 percent of small businesses that had 50 to 499 employees created 30 percent of all net jobs, whereas the 79 percent of small businesses with fewer than 10 employees created only 15 percent.”40
To the extent policy should focus on new business, the focus should be on those that want to and can get big. Recent evidence suggests that only a tiny subset of businesses is responsible for most of job creation. Dane Stangler found that just 5 percent of companies in the United States create two-thirds of new jobs in any given year.41 However, though they tend to be more accurate than studies that claim that small firms create the most jobs, studies that celebrate the role of “high-impact” firms also suffer from several problems.
First, the definition of high-growth companies is based in part on job creation. So a company that is highly productive and that significantly expands sales while lowering prices and increasing wages is likely to be excluded from the universe of high-growth firms because it didn’t create as many jobs as a less efficient labor-intensive firm. As the author of an SBA-funded report on high-impact firms writes, “Many of the earliest definitions were based solely on revenue growth. A limitation of this approach is that it does not take into account employment change. This is an important policy consideration for government.”42 The troubling implication is that productivity is not an important consideration.
The second and more serious limitation of studies of high-impact firms relates to how these firms are defined. High-impact companies are defined as firms whose sales double in size over a four-year period. But by definition it is easier for a small pizza shop to double in size than it is for a large company like Apple to double in size. Not surprisingly, therefore, research shows that these “high-impact” firms are mostly small firms. The SBA-funded study found that the average growth of high-impact companies with more than 500 workers was around 125 percent, whereas it was 375 percent for the smallest firms (one to nineteen) employees. It’s not that hard for a firm with three employees to grow to fourteen. (The average size of companies that started with one to nineteen employees was 2.7 at the beginning of the period and 14.4 at the end.) But it’s a lot harder for a firm with 5,000 workers to grow to 17,000 in four years. (For firms with more than 500 workers, the average size was 4,466 at the beginning of the period and 10,102 at the end.)43
Logically, we should be indifferent to whether a company doubles in size from twenty to forty workers or grows from 10,000 to 10,020 workers. In fact, because of the superior performance of large firms, we should be decidedly in favor of the latter. Nonetheless, even with the biased definition the SBA employs, companies with more than 500 employees were still responsible for creating 43 percent of jobs created by high-impact companies.44
Most Small Firms Are Dependent on Big Firms
In the natural world, “capstone species”—large animals or plants—help to shape an environment in which many smaller organisms can flourish. On the North American prairie, bison cleared spaces for prairie dogs and grasses by grazing and wallowing. In tropical rain forests, giant trees provide support for vines and shade for smaller, shade-loving plants. Like capstone species, in many sectors big firms provide “ecosystem services” to much greater numbers of small and medium-sized companies. Large corporations are customers for complex webs of smaller suppliers. Their spending benefits yet other firms, as does the spending of their large number of more highly paid employees. And large firms can help innovations scale up, by buying them from startups or buying the startups. In advanced industrial economies, many small firms and big firms are mutually interdependent partners in a common productive enterprise, not natural enemies engaged in a battle to the death. Moreover, whether America thrives in the global economy is not whether a clothing shop on Main Street sells more pants. It is whether companies that export goods and services and compete in tough international markets do well; whether companies that drive productivity in their operations through the introduction of new technology do well; and whether high-growth entrepreneurial companies, especially ones that develop and commercialize innovations, do well. While they may have Main Street suppliers, these are not Main Street companies. They are “Industrial Street” and “Office Complex Street” companies, the former being manufacturing firms, particularly those competing in international markets, and the latter being technology-based nonmanufacturing companies
Why does job creation or loss by large firms have a bigger effect on the unemployment rate than similar changes by small firms? One reason is, as we showed in chapter 3, that larger firms are more likely to sell goods and services outside of the geographic area they are located in. It’s important to understand the difference between what regional economists refer to as local-serving and export-serving businesses.
Let’s consider the Maytag appliance factory that closed in Newton, Iowa, a few years ago. In a 60 Minutes segment about the suffering of local, small businesses in Newton caused by the closing (the washers and dryers were to be made in Mexico), the host, Scott Pelly, bemoaned the fact that these small companies weren’t getting help: “Three years after the beginning of the Great Recession, with interest rates the lowest they have ever been in history, banks are lending less money to the engines that create jobs.”45
This misses the point. Maytag was an export-serving business, meaning that it shipped products outside the local labor market. Though a small share of the washers and dryers coming off the assembly line were sold to local Newton residents, the vast majority were sold to customers throughout the nation or even the world, who sent money back to Maytag, who paid some of it to its local workers and contractors.
In contrast, the local Newton restaurants, dry cleaners, clothing stores, and barbershops are local-serving, as the lion’s share of their output is sold to Newton residents, including Maytag workers. If one of these local-serving small Main Street restaurants had gone out of business, it would have had no effect on the output of the Maytag factory. Moreover, another restaurant would more or less automatically have expanded or emerged to meet local demand.
But the Maytag factory closure had an immediate negative impact on the local-serving businesses, whose customers (Maytag workers, their suppliers, and the suppliers’ workers) had much less money to spend locally on meals, haircuts, dry cleaning, and other needs and desires. Conversely, if Newton, Iowa, were to attract a large “exporting” company to occupy the abandoned Maytag facility, the health of Newton’s small businesses would immediately revive.
So what determines whether America thrives, including impacts on the unemployment rate, is not whether a clothing shop on Main Street sells more pants. It is whether companies that export goods and services and compete in tough international markets do well. As noted, while some small firms export, big firms are more likely to export.
The majority of US businesses are local-serving. These include 219,986 doctors’ offices, 166,366 auto repair facilities, 151,031 food and beverage stores, 115,533 gas stations, 111,028 offices of real estate agents and brokers, 93,121 landscaping companies, 75,606 nursing homes, 36,246 furniture stores, 28,336 veterinary offices, 15,666 travel agencies, 4,571 bowling alleys, 2,463 amusement arcades, 858 radio networks, and 26 commuter rail systems. When looking at the forty largest four-digit code industries in terms of share of small businesses (with twenty employees or fewer), which collectively constitute two-thirds of all small businesses, all but two industries (consulting services and computer system design) are primarily local-serving, such as restaurants, physician offices, auto repair, insurance agents, and so on.46 In this regard it doesn’t matter for job creation whether a large firm or a small firm satisfies this demand; some firm will because consumers have money to be captured.
Many other small firms are dependent on large firms as their customers or business partners As Bennett Harrison has written, “Many du jure independent small companies turn out in varying degrees to be de facto dependent on the decisions made by managers in the big firms on which the smaller ones rely for markets, for financial aid, and for access to political circles.”47 We see this, for example, in the fact that Boeing, the leading aerospace company, spent $5 billion with US small business suppliers in 2016, representing approximately 50,000 jobs.48 In Europe, for example, 56 percent of small businesses in Denmark are dependent on large firms, with 55 percent dependent in Norway, that is, they are part of a larger enterprise group.49
Finally, at the end of the day, this entire jobs debate hinges not only on what the data show but also on whether one thinks that job creation comes from the supply side or the demand side. Small business advocates John Dearie and Courtney Geduldig point to the supply side when they write, “In the seven other years over the period, older firms also contributed to job creation. But start-ups contributed an average of 3 million new jobs every year. In other words, without new businesses and the jobs they create, net job creation in all but seven years between 1977 and 2005 would have been negative.”50
But this is inaccurate. To understand why, we need only remember that innovation, productivity, and exports are supply-side factors. In other words, while the level of demand in an economy can affect these factors, the major drivers are internal factors within firms—the level of R&D, the level of investment in capital equipment, the development of better business models, and the like. But when it comes to job creation for an entire economy, the determining factors are on the demand side. The key word here is “entire.” The mistake that most small business advocates make when they say that small business is needed for its job creation prowess is that they conflate macro with micro factors. This is a fallacy of composition.
Let’s assume for the moment that the US economy is closed with no trade. Then all the companies sell all their output to US businesses or consumers. Let’s also assume that in the current year, more young people enter the labor force than retire. Now, absent job creation, there will be unemployment because there are now more workers. Do we want to ensure that policies encourage Justin and Ashley to start a new pizza parlor to employ some of those young people? Maybe we should even give Justin the pizza “entrepreneur” a tax incentive so he will take the risk of starting the company in the hope it is successful and he will be able to employ some of those young people.
Or do we want the Federal Reserve Bank to lower interest rates a few basis points, spurring a bit less saving and a bit more spending and investment? As that spending and investment flow through the economy they create more demand, which creates more production, which in turn creates demand for more workers. The young new entrants to the labor market are hired and the economy is back to full employment, at which point the Fed raises the interest rate back to its old equilibrium rate (the rate that balances inflation and unemployment).
In this model it doesn’t matter whether the firms are small or large; firms of any size respond to the demand and hire the new workers. So there is simply no need for favorable policies to help small business in order to create jobs. In fact, as we argue in chapter 12, such policies are downright harmful to economic growth.
