In light of widespread complaints that US firms are becoming monopolistic behemoths, it might come as a surprise to learn that the average US enterprise employs just 20.2 workers and the average establishment even fewer (an enterprise or firm consists of the entire company, while an establishment is a separate operation).
However, these size patterns differ significantly by industry. Some industries are inherently small. The five smallest industries in terms of firm size are fishing, hunting, and trapping (3.1 workers), building construction (5.5), real estate (5.9), funds, trusts, and other financial vehicles (6), and repair and maintenance (6.1). The industries in table 3.1 all had average establishment sizes of fewer than three employees in 2012. The vast majority of small establishment industries have few opportunities to get more efficient through size, by means of economies of scale. For example, a commercial photography firm of 1,000 workers using the same techniques would likely be less efficient than the average-sized photography firm of three workers.
Table 3.1 US industries with an average establishment size of three employees or fewer, 2012
Independent Artists, Writers, and Performers
Other Marine Fishing
Shellfish Fishing
Footwear and Leather Goods Repair
Interior Design Services
Record Production
Offices of Real Estate Agents and Brokers
Mobile Food Services
Lessors of Miniwarehouses and Self-Storage Units
Commercial Photography
Home and Garden Equipment Repair and Maintenance
Offices of Real Estate Appraisers
Source: US Small Business Administration, Firm Size Data (Table 2. Number of Firms, Establishments, Receipts, Employment, and Payroll by Firm Size [in Receipts] and Industry), https://www.sba.gov/advocacy/firm-size-data (accessed March 10, 2017).
Moreover, most small firm industries are engaged in local production, selling the lion’s share of output locally (think of barber shops, dry cleaners, real estate appraisers, auto repair services). In fact, just 6 percent of employment in the 100 industries with the smallest average firm size is in traded industries (mostly in fishing and sound recording and production).1 Thus, when policy makers advocate for small business, they are mostly favoring local businesses without national or global competitors, which would exist regardless of government favors. Someone will dry clean our clothes even if dry cleaners don’t get tax subsidies.
At the other end of the size distribution are industries in which the average firm in 2012 had more than 1,000 workers, including hospitals, department stores, casino hotels, nuclear power generators, and airlines. These are industries in which economies of scale and scope enable larger size. For example, companies such as Macy’s, Sears, and Target benefit from a nationwide network of stores. Airlines such as American, Delta, and United benefit from a large network of airline routes so flyers can more easily get from point A to point B. Industries with average establishment sizes of more than 400 workers see similar economies of scale (see table 3.2).
Table 3.2 US industries with an average establishment size of more than 400 employees, 2012
Guided Missile and Space Vehicle Manufacturing
Casino Hotels
General Medical and Surgical Hospitals
Poultry Processing
Guided Missile and Space Vehicle Propulsion Unit and Propulsion Unit Parts Manufacturing
Light Truck and Utility Vehicle Manufacturing
Professional Employer Organizations
Aircraft Manufacturing
Colleges, Universities, and Professional Schools
Source: US Small Business Administration, Firm Size Data (Table 2. Number of Firms, Establishments, Receipts, Employment, and Payroll by Firm Size [in Receipts] and Industry), https://www.sba.gov/advocacy/firm-size-data (accessed March 10, 2017).
Large-firm industries are more likely to sell output outside the nation; that is, they occur in traded sectors. In the approximately 100 industries with an average firm size of more than 175 workers, the share of jobs in traded sector industries was approximately 60 percent. These include industries such as aircraft, tire, motor vehicle, and guided missile manufacturing.2
Trends in Firm and Establishment Size
From 1997 to 2012, average firm size increased 6.6 percent, from 19 employees to 20.2, up from 17.6 in 1987 (see figure 3.1). But this trend masked considerable differences among industries. The average number of jobs in manufacturing firms fell 22 percent, with jobs in apparel manufacturing firms down 35 percent, jobs in textiles down 34 percent, and jobs in computer and electronic products down 23 percent. Likewise, construction firms have gotten slightly smaller. But overall average size would have increased even more had employment not shifted to sectors that employed on average fewer workers per firm. So while services gained employment and their average establishment size increased, their average firm size was much smaller than that of manufacturing firms.
Figure 3.1 Change in Average Firm Size by Industry, 1997–2012 (Employees)
We see similar patterns with the change in size of establishments (see figure 3.2). From 1997 to 2012 the number of workers in the average establishment increased by 2 percent, but this increase masked considerable intersectoral difference, with manufacturing declining and others increasing. Establishments with fewer than 500 employees accounted for 10 percent fewer jobs in 2009–2010 than in 1998, while establishments with more than 500 employees accounted for 20 percent more. There were only two industries in which large establishments lost share: farming, fishing, and hunting and manufacturing. But large construction establishments accounted for 51 percent more jobs than in 1998–1999 and transportation and warehousing accounted for 42 percent more, while information technology (IT) accounted for 27 percent more.3
Figure 3.2 Change in US Establishment Size by Industry, 1997–2012
The perception that firms are getting bigger on average is accurate. True, the share of employment in establishments employing more than 10,000 fell from 16 percent in 1974 to 13 percent in 2006. But this loss was all in manufacturing. In services the share increased from 12.5 to 17 percent. Likewise, the largest 1 percent of establishments employed 58 percent of service workers in 1974 but 62 percent in 1997.
Shrinking Manufacturers
While overall firm size is getting bigger, manufacturers have been getting smaller since the late 1960s. The average manufacturing plant employed around twenty workers in the first two decades of the 1900s, but this figure had tripled by the end of World War II, in part because of the adoption of the assembly line and mass production technology in many discrete goods industries (cars, airplanes, appliances, electronics) and process industries (pulp and paper, chemicals, steel) (see figure 3.3) But since 1967 average manufacturing plant size has fallen, dipping below 1925 levels in 2012.
Figure 3.3 Trends in Average US Manufacturing Plant Size (Employment)
We see the same trend for average manufacturing firm size, which fell from around eighty workers in 1972 to around fifty in 2007.4 The same dynamics can be observed in most other industrial nations. In Italy the share of manufacturing workers in plants with more than 100 workers declined from 43 percent in 1961 to 29 percent in 2007, in France from 61 percent in 1962 to 52 percent in 2001, and in Germany from 69 percent in 1962 to 60 percent in 2001.5
Some of the decline may be due to higher productivity in manufacturing than nonmanufacturing. In other words, manufacturers can produce as much or more with fewer workers. But this can’t be the whole explanation, in part because manufacturing productivity has grown faster than nonmanufacturing for more than a century. Rather, new, more flexible production technologies which emerged in the 1960s that allow manufacturing firms and establishments to produce shorter production runs appear to play a key role. In a world where manufacturing meant expensive investments in specialized tooling and machining to make specific parts (e.g., a car door panel), firms had incentives to have very long production runs, necessitating large factories. Changing tooling to make a different part was time consuming and expensive. Moreover, when coordination between different plants and suppliers was difficult (using paper-based forms and telephone communications to coordinate), it was easier to consolidate all production within a massive facility, as exemplified by Henry Ford’s Rouge River plant, where coal and iron ore went in one end and cars came out the other end.6
But with the development of flexible production technology in the 1960s, first numerically controlled (NC) machine tools and then computer-numerically controlled (CNC) machine tools, followed by the availability of computer-aided design tools, firms could lower the costs of shorter production runs. And with the emergence in the 1970s of electronic data interchange (EDI) systems and then in the 1990s of Internet-based systems, it became more economical to link specialized plants together in extended production networks. The rise of just-in-time production networks meant establishments didn’t need to be huge to enable coordination.
Global competition has also led to reduced manufacturing establishment size. With the increased ease and effectiveness of global coordination systems, it became much easier to develop global supply chains, with commodity-based, mass production activities moving to low-wage nations to take advantage of cheap labor (and government subsidies). The establishments most likely to move were the larger ones making more standardized goods.
Nonmanufacturing Firms Getting Bigger
While manufacturing firms have been getting smaller, establishments and firms in most nonmanufacturing industries have been getting bigger. The average service sector establishment increased from twelve employees in 1977 to eighteen in 2014, with firms increasing from thirteen employees in 1977 to twenty-one in 2014.7 Professional services, education, health care, arts, accommodations, other services, and administrative and waste management firms all got bigger, as did mining (by 16 percent), wholesale trade (11 percent), retail trade (14 percent), and hotels/accommodations (7 percent). The changes since 1963 were even more dramatic, with all sectors other than manufacturing seeing increased establishment size: hotels grew 545 percent, construction grew 500 percent, and retail trade grew 355 percent.
In the late 1990s it was widely believed the Internet would do for services what computer technologies had done for manufacturing: empower small firms. Now mom-and-pop firms could access a world market with just a broadband connection and a website. But by the same token, big firms also got access to the mom-and-pop firms’ customers with their broadband connections and better web pages.
The result has been that IT has enabled greater economies of scale for many service industries, in part by letting these firms serve larger regional markets and reducing transport costs between facilities. The internet in fact, enabled firms to get larger. Many services, among them retail, banking, insurance, law, accounting, and securities services, that were once tied to local economies can now be conducted at a distance. IT has enabled an increasing share of information-based services to be physically distant from the customer (e.g., e-banking) or more consolidated (e.g., back office operations) while remaining functionally close.
From 1997 to 2012 the average warehouse and storage company increased in size by six times (from fifteen employees to ninety-eight) as technology enabled warehousing to consolidate and regionalize.8 With the emergence of big box retailers with powerful IT systems to coordinate ordering and inventory management, and more recently online commerce, retail firms got bigger, including electronics and appliance firms, such as Best Buy (33 percent more workers); building material and garden equipment companies, such as Home Depot (19 percent), home and furnishing companies, such as Ikea (10 percent); sporting goods and book and music companies, such as Barnes & Noble (15 percent); clothing retailers, such as Old Navy (35 percent); and general merchandise stores, such as Amazon.com (35 percent). In part because of the increased use of technology and increases in demand, firms in health care got bigger, with nursing homes up 8 percent, ambulatory health care services up 14 percent, and hospitals up 47 percent. Banks saw an average establishment size increase of 30 percent. And with the continued spread of national lodging chains, the average size of accommodation firms grew by 12 percent.
But some industries shrank. The securities, commodity contracts, and other financial investment and related activities industry fell 27 percent, while the size of insurance carriers shrank by 6 percent. Moreover, in some industries, while technology enabled larger firms, it meant smaller establishments. That occurred in banking, for example, where the widespread use of automatic teller machines allowed branches to employ fewer people. Banks are getting bigger, but average employment in commercial bank branches declined from about twenty employees in 1988 to thirteen by 2004.9
One result is the spread of large, multi-establishment corporations into markets that once were more heavily populated by mom-and-pop stores. From 1977 to 2013, US employment rose by 72.3 percent, while the number of firms rose by just 47.9 percent. In other words, firms were getting bigger. Much of this can be explained by consolidation, as firms could use technology to more easily manage multiple business locations. This is why the number of US establishments rose by 61.4 percent, less than employment growth. In other words, new firm formation lagged.
Why Have New Firm Startup Rates Fallen?
With the tepid and long recovery from the global recession of 2008–2009, a cottage industry has emerged in identifying the reasons for slow growth. One popular answer is the decline in new business formation. It is almost impossible to read business publications without being bombarded by warnings that small firms are starting at anemically low rates, which is keeping America’s growth in the slow lane.
In 2012 Barry Lynn and Lina Khan warned that “the share of the working-age population that is self-employed has been declining since 1994. … Overall, between 1994 and 2009, the share declined nearly 25 percent.”10 This is correct, but the trend goes back much further. Self-employment rates fell from 9.1 percent in 1965 to 6.8 percent in 2005.11
But it was not until 2014 that alarm over dwindling new business formation got widespread media attention, when a Brookings Institution report by Ian Hathaway and Robert Litan found that new business starts had “declined in all fifty states and in all but a handful of the more than three hundred and sixty US metropolitan areas during the last three decades.”12 As their report shows, the trends are clear (see figure 3.4).
Other studies joined in raising the alarm. In 2015 Jason Wiens and Chris Jackson of Kauffman Foundation wrote: “New businesses represent a declining share of the business community. … New firms represented as much as 16 percent of all firms in the late 1970s. By 2011, that share had declined to 8 percent.”13 The next year, John W. Lettieri and Steve Glickman of the Economic Innovation Group wrote that “the U.S. economy has steadily become less entrepreneurial over the past 30 years. That decline turned into a rapid collapse following the Great Recession, when the economy produced hundreds of thousands fewer new businesses than it did in previous recovery periods, marking a lost generation of new enterprise.”14
This decline is seen by many as a crisis of major proportions. Hathaway and Litan warn that the US business sector is getting “old and fat” and that “nothing less than the future welfare of America and its citizens is at stake.”15 Lettieri calls this “the fundamental challenge of our time.”16 Lynn and Khan agree: “The founding generation was right: as America’s entrepreneurs go, so goes America’s prosperity and democracy.”17 Gallup CEO Jim Clifton warns that “this economy is never truly coming back unless we reverse the birth and death trends of American businesses.”18 And John Dearie, executive vice president of the Financial Services Forum, warns that “this is nothing short of a national emergency.”19
We disagree. In our view, the only important question is what is happening to “Schumpeterian” entrepreneurship—the creation of technology-intensive businesses that can get big—not the total number of new small businesses of all kinds, many of them founded merely so that the owner can fulfill personal lifestyle goals. From this perspective, contemporary alarmism over dwindling startups and self-employment is misguided and can even be dangerous, if the mistaken diagnosis leads to harmful prescriptions.
Why have startup rates declined? Few seem to know. According to Hathaway and Litan, “The reasons explaining this decline are still unknown,”20 though a later report from them suggests a few causes, such as slow population growth and increased firm size.21 But by and large, the cause of the decline has remained a mystery. This has not stopped advocates from using the alleged problem to justify their preferred public policies.
The Kauffman Foundation proposes increasing the number of immigrants, removing regulatory barriers, making the tax code simpler, and increasing education levels.22 But immigration for the last two decades—the very period of “entrepreneurial” decline—has been at record highs. Why would it help to add even more immigrants, particularly ones with low levels of skills and education? The tax code is arguably more complicated than it was twenty years ago, but not by much. Besides, any small business can plunk down a few bucks for an easy-to-use tax preparation software package, such as TurboTax. More education? A greater share of Americans are going to college and graduating than ever before, to the point that more than 15 percent of American adults have more education than they need for their job.23
The free market conservatives of the Heritage Foundation blame high taxes, just as they tend to blame high taxes for a variety of economic woes.24 But tax rates are lower today than they were in the 1980s. At the other end of the political spectrum, the progressive liberal Roosevelt Institute blames the decline on increased corporate concentration and profits.25 But, as we show in chapter 11, this does not appear to be the case.
The National Federation of Independent Business warns that small businesses are “burdened by an ever-growing pile of regulations.”26 But new firm formation declined from 1980 to 1992, when Ronald Reagan and George H. W. Bush were in the White House and pushing a deregulatory agenda. It continued to decline after 1994, when the Republicans gained a majority in the House and began to implement their Contract with America to reduce regulatory burdens. Moreover, if regulation is really the problem, why did states with business climates as different as California and Texas have large and similar declines in new business starts?27 According to the Kauffman Foundation rankings, California, a state hardly known for regulatory reticence, outperforms Texas in new firm formation.28
Another theory holds that big banks are refusing to lend to creditworthy borrowers. Barry Lynn and Lina Khan write, “The effects of the radical consolidation in the banking industry that began in the 1980s are equally dramatic. Relatively few bank officers today have the leeway and local knowledge to lend to established local businesses, much less new ventures.”29 But as we note in chapter 10, research shows that bank consolidation has helped as banks are better able to spread risk. However, some have argued that the Sarbanes-Oxley and Dodd-Frank legislation has made it harder for firms to go public and for banks to lend to small business. But that is a different issue from saying that large banks per se are behind the decline in startups.
Donna Kelley, a professor of entrepreneurship at Babson College, blames millennials, describing them as slackers who have a preternatural fear of failure.30 If so, it suggests that colleges are failing our students, as the number of college entrepreneurship classes has increased twenty times over the last thirty years.31 Moreover, Britt Hysen, the editor in chief of MiLLENNiAL magazine, claims that “60 percent of Millennials consider themselves entrepreneurs, and 90 percent recognize entrepreneurship as a mentality.”32 While some blame millennials, others blame the boomers for being too numerous and too old.33 However, using a regression model of firm entry by state, Hathaway and Litan convincingly rebut this explanation.34
By far the favorite explanation for the decline of startups is that swelling, sclerotic monopolies are thwarting innovative entrepreneurs. Lynn and Khan write, “The single biggest factor driving down entrepreneurship is precisely the radical concentration of power we have seen not only in the banking industry but throughout the U.S. economy over the last thirty years.”35 In his Atlantic article, “America’s Monopoly Problem: How Big Business Jammed the Wheels of Innovation,” Derek Thompson writes that the decline “has coincided with the rise of extraordinarily large and profitable firms that look like the monopolies and oligopolies of the 19th century.”36 Stacy Mitchell of the Institute for Local Self-Reliance writes, “The decline of small business and entrepreneurship is owed, in significant part, to anticompetitive behavior by dominant corporations which routinely use their size and market power to undermine and exclude their smaller rivals.”37 The argument of modern champions of the antimonopolist tradition is simple: concentration went up and startups went down, so the former caused the latter.
To paraphrase former senator Patrick Moynihan, who famously said “Everyone is entitled to his own opinion, but not his own facts,” we don’t disagree with the small firm startup alarmists’ facts, or at least with most of them. We disagree with what they mean. We see the decline in small businesses like Justin and Ashley’s independent pizza parlor as a good thing, reflecting the growth of more efficient and more innovative larger firms.
Moreover, we want to advance a more nuanced argument that the decline in new firms is not so simple. It’s one thing to assert that fewer Justins and Ashleys are opening up pizza parlors or dry cleaners. It’s quite another to assert that there are fewer Sergeys (Brin), Larrys (Page), and Jeffs (Bezos) starting the next Google or Amazon. The first is true, the second much less so.
Should we be concerned that “entrepreneurs” like Justin and Ashley are starting fewer small firms? On the face of it we shouldn’t be, because as we show in the next chapter, large firms outperform small firms on virtually every economic indicator. But if their decline was due to some problem with the business environment or a decline in entrepreneurial drive, then maybe we should be worried, since this might also hamper high-growth startups.
Why Did Startups Decline?
So why have the Justin and Ashley startups declined? The “business startup sky is falling” narrative implies that if we get more new businesses, the economy will grow faster. But what if the economy’s performance has affected the rate of new business formation, rather than the other way around? A Federal Reserve Bank of Cleveland study finds that the rate of aggregate demand, largely after the Great Recession, influences firm startup rates. The authors write: “We find that cyclical factors have contributed to recent low levels of self-employment. … Decreasing demand leads to an increase in exit from entrepreneurship but has countervailing effects on entry.”38
In other words, slower economic growth has reduced the opportunities for new firms. We see this in the fact that the twenty metro areas with the highest rates of new firm formation were all in faster-growing “Sunbelt” states while the twenty with the lowest rates were all in slower-growing “rust belt” states.39 As the US economy over the last three decades has come to more resemble the rust belt than the Sunbelt, it shouldn’t come as a surprise that there are fewer new firms.
But slower US growth is not the only factor. To understand the decline, it’s important to look at startups by industry. When you do that it becomes evident that not all industries have seen declines. While almost 100,000 fewer new firms formed in 2011 than in 2003, in about 40 percent of 290 industries more firms formed in 2011 than in 2003. For example, mining startups increased 30 percent, largely because of the oil and gas shale boom. Educational services startups increased 11 percent as new technology offered new market opportunities. Startup rates in professional, scientific, and technical services; information (e.g., software, telecom, broadcasting); and health care and social assistance industries remained unchanged.
Moreover, some industries that saw declines were ones that felt the impacts of the global recession particularly acutely. The collapse of the construction industry after the housing bust meant that construction startups fell 26 percent and real estate and rental and leasing fell 13 percent. Why start a new construction firm when home building is in tank? Similarly, with the financial crises and the bankruptcy of multiple financial services firms, why start a new firm in the financial services industry, where startups fell 29 percent?
And in some industries where the rate of startups did fall, concentration ratios actually fell or were stable, suggesting that excessive market power favored by antimonopolist narratives was not the cause. For example, in “other services,” startups fell by 24 percent, but the C4 and C8 ratios (the share of sales in industry by the top four and top eight firms, respectively) fell 1 percent. In the wholesale trade and arts, entertainment, and recreation industries, startups declined 16 percent and 14 percent, respectively, but C4 and C8 concentration ratios were unchanged.
In manufacturing, where new firm formation was down 20 percent from 2003 to 2011, the decline was not due to big manufacturers taking market share and crushing the new guy, as evidenced by the average C4 and C8 concentration ratios in manufacturing increasing by less than one percentage point. The most likely reason manufacturing startups fell is because stiff international competition from nations such as China, much of it unfair and based on mercantilist policies such as currency manipulation, intellectual property theft, and government subsidies, made it hard for budding US entrepreneurs to break in.40 Why start a manufacturing firm if you are likely to face predatory competition not from big US firms but from Chinese competitors backed by their state? Unfair foreign competition, coupled with the absence of a national US competitiveness policy, is a major reason why inflation-adjusted US manufacturing output fell more than 10 percent in the 2000s, with over 65,000 manufacturing establishments closing their doors.41 And when output is falling, that is hardly a good time to start a new manufacturing firm.
Between 2000 and 2014 (China joined the World Trade Organization in 2000), the number of small manufacturing establishments (one to four workers) in the United States fell by 14 percent, while the number of large establishments (1,000+ workers) fell by 42 percent. One reason for the difference was the slow movement of smaller firms into the large size class. In 1980, 110 establishments with more than 1,000 workers were three years old or less. That figure fell every decade, to just eighteen establishments by 2014, a decline of 84 percent. In contrast, the growth of younger small establishments fell between 52 and 62 percent. At the firm level, the share of manufacturing workers employed by firms employing 5,000 or more workers fell from 46 percent in 1977 to just 15 percent in 2005.42
Notwithstanding the stiff headwind from competition, a few manufacturing sectors saw an increase in new firm formation, but these sectors were mostly ones that produced products less subject to import competition. For example, bakery and tortilla manufacturing startups expanded by 17 percent, “other food manufacturing” by 14 percent, and beverage manufacturing by 74 percent (in part because of the expansion of craft beer and healthy drink products such as Honest Tea).43 But if the antimonopoly story is right and decreased competition was the cause of fewer startups, we should have seen declining new firm formation in the beverage industry since the C4 and C8 concentration levels increased by ten percentage points over the last decade.
To be sure, the assertion that larger firms are crowding out startups is valid in some sectors. The fact that technology has enabled larger and more efficient firms in many nonmanufacturing sectors means there are fewer opportunities for small and new firms. This is why David Audretsch has argued, “The likelihood of new-firm survival should be lower in industries exhibiting greater scale economies.”44 This is not about predation but about space for new firms in any particular industry.
We see this trend in some industries, especially retail trade. Retail industry startups fell 16 percent from 2003 to 2011, but not because large firms abused their market power to kill startups. Rather, technologies such as software-enabled logistics systems and web-based e-commerce enabled the average retail firm to get larger, meaning there was less market space for startups unless they had something truly unique to offer or local convenience.
For example, it was once not too hard to open a book or music store. When one of us was younger, he used to take a trip after school to the local record store, Pauls (owned by Paul), to sample the latest 45s. Now we get our music from iTunes and Pandora. This is why book and music store startups fell 58 percent from 2003 to 2011, while the total number of establishments shrank by one-third.45 At the same time, the market share of the largest four firms increased from 48 percent to 66 percent. But these changes came about because of technology and efficiency, not predation. Large specialty bookstores such as Barnes & Noble, and then later in the decade online retailers such as Amazon.com, and the rise of e-book sellers such as Apple’s iTunes store and Amazon’s Kindle, meant that more people bought books and music at large brick-and-mortar stores and at online stores because they could save money and have a wider choice of products.
We have seen the same dynamic with hardware stores. Forty years ago someone who was good with tools might think of opening a hardware store. Today they would likely think twice about doing so since big box stores such as Home Depot and Lowes serve this market very well, having gained market share from small, independently owned hardware stores. But they didn’t gain it by predation and unfair practices that crushed the little guy. They gained it by providing a much wider selection of products at a significantly lower price. The typical Home Depot store is around 105,000 square feet (almost the size of two football fields), more than ten times larger than the typical neighborhood hardware store.46 And the big box stores stock upward of 25,000 different products, significantly more than the neighborhood stores do.47 This volume lets them be much more efficient, with sales per square feet of store two-thirds higher than at neighborhood stores and 25 percent more per employee.48
This story has played out in many retail sectors where large retailers have gained market share by providing goods or services that consumers want at prices they can afford. Owner-operator barbershops have been superseded by Supercuts, coffee shops by Starbucks, donut shops by Dunkin’ Donuts, stationery stores by Staples and Office Depot, local pharmacies by CVS and Walgreens. Among the many retail industries that saw increasing average employment size per firm were food and beverage stores (7 percent), furniture stores (10 percent), sporting goods (15 percent), banks (29 percent), electronics and appliance stores (33 percent), and general merchandise stores (35 percent).49
These retail giants have tapped into the substantial benefits of scale, which are passed on to customers. Moreover, large retailers compete directly against each other, spurring innovation, technology investment and adoption, and efficiency. Indeed, technological innovation (particularly computer-based supply chain ordering) has enabled size increases. The effects have nowhere been more dramatic than in those sectors that have always been most congenial to individual proprietorships, such as retail, services, farming, and small manufacturing. These were the sectors and the activities most affected, for instance, by the type of “roll-up” strategies pioneered by financiers like Mitt Romney’s Bain Capital. In the case of the office-supply retailer Staples, Bain’s investment helped propel the company from a one-store operation to a 2,000-store international behemoth. Similar plays resulted in Home Depot capturing a vast proportion of the nation’s hardware business, in Best Buy capturing a vast proportion of America’s electronics business, and in Macy’s capturing a vast proportion of all department store sales. Just one company, Wal-Mart, now has upward of 50 percent of some lines of grocery and general merchandise business—commerce that a generation ago was divided among tens of thousands of family businesses—with shoppers the big beneficiary. And, of course, emerging web-based businesses such as Amazon promise even more competition, efficiency, convenience, and scale.
This transition from local Justin and Ashley stores to chain stores has been going on for more than a century, starting with stores like A&P and Sears, and has occurred largely independently of government policy. If anything, government policy has leaned into this wind, providing big subsidies to the mom-and-pop businesses, and in some communities actively restricting the large stores.
Yet for the antimonopolists, this is a decidedly bad thing. As Thompson writes, “Today, in a lot where several mom-and-pop shops might once have opened, Walmart spawns another superstore.”50 This is supposedly a bad thing because it is stifling “entrepreneurship.” But this fetishization of small business misses the point. New business formation is not an end; it is a means. The end is more and better goods at lower prices for consumers, not maximizing the number of owners of small, inefficient businesses.
What Kinds of Firms Are Starting?
To assess whether the overall slowdown in new firm formation is a problem, we also need to look carefully at the types of firms that are being started. As Antoinette Schoar writes,
It is crucially important to differentiate between two very distinct sets of entrepreneurs: subsistence and transformational entrepreneurs. Recent evidence suggests that people engaging in these two types of entrepreneurship are not only very distinct in nature but that only a negligible fraction of them transition from subsistence to transformational entrepreneurship. These individuals vary in their economic objectives, their skills, and their role in the economy.51
Justin and Ashley, the owners of the new local pizza parlor, are not likely to be very much like Sergey Brin and Larry Page or Jeff Bezos. Schoar also found that
[The] founders of venture-backed startups in the majority were previously employed at larger technology firms such as Microsoft, Intel, or similar firms. An alternative group of founders of transformational entrepreneurs were serial entrepreneurs who had previously started a high-growth firm. In contrast, almost none of them were running small subsistence businesses before they started a high-growth business.52
Nor does there appear to be any correlation between startup numbers and economic growth. A study of entrepreneurship data by Catherine Fazio and coworkers notes that “quantity-based measures of entrepreneurship have little relationship to GDP growth. Yearly fluctuations in counts of firm births appear to hold little relationship to medium-term measures of economic performance.”53
To be clear: starting yet another small bookstore or pizza parlor is not entrepreneurship; it’s small business. In other words, Justin and Ashley are not likely to be Sergey and Larry and Jeff. And what Justin and Ashley do or do not do has little effect on economic growth. If Justin and Ashley don’t start that pizza parlor, then Brianna and Jalen or someone else will.
Indeed, using the term “entrepreneur” for someone who opens a conventional small business such as a pizza parlor uses the term in a sense completely different from that of Joseph Schumpeter, the economist who pioneered modern innovation theory. Schumpeter famously wrote,
The function of entrepreneurs is to reform or revolutionize the pattern of production by exploiting an invention or, more generally, an untried technological possibility for producing a new commodity or producing an old one in a new way, by opening up a new source of supply of materials or a new outlet for products, by reorganizing an industry and so on.54
He did not say the function of an entrepreneur is to start a business. Only certain kinds of business founders are entrepreneurial. In other words, the startups that really matter are those that are able to exploit market opportunities through technological and/or organizational innovation.55 As one study of firm formation trends stated, “Poorly performing economies seem to have too many subsistence entrepreneurs and too few high-growth transformational entrepreneurs.”56
This difference in the kind of new firm startups is why dire claims that the sky is falling on new business formation can exist parallel to claims that we are living in a time of robust innovation and entrepreneurship, with Silicon Valley and other tech hubs throughout the nation enjoying frothy and dynamic innovation. As Silicon Valley venture capitalist Marc Andreessen tweeted, “There’s too much entrepreneurship: Disruption running wild!” “There’s too little entrepreneurship: Economy stalling out!”57 A big reason for this contradiction is that the above studies (which are endlessly quoted) don’t differentiate between Justin and Ashley, on the one hand, and Serge and Larry and Jeff on the other. The real question is what has happened to the entrepreneurship exemplified by Sergey and Larry and Jeff, and why.
Researchers who have tried to differentiate between the two types are MIT professors Jorge Guzman and Scott Stern. They looked at trends in high growth entrepreneurship for fifteen large states from 1988 to 2014 and found that “in contrast [to] the secular decline in the aggregate quantity of entrepreneurship … the growth potential of startup companies has followed a cyclical pattern that seems sensitive to the capital market environment and overall economic conditions.”58 In other words, while the Justins and Ashleys are starting fewer firms than before, the Sergeys, Larrys, and Jeffs are starting almost the same number. Moreover, Fazio and her colleagues found that when they compared the original “birth dates” of firms that achieved successful exits (defined as an IPO or acquisition at a multiple of the firm’s valuation within six years) relative to overall firm births, again, they could find no apparent relationship.59
Indeed, in another study Guzman and Stern found that even after controlling for the size of the US economy, the second highest rate of high-growth entrepreneurship occurred in 2014.60 This research indicates that the entrepreneurial potential (successful startups as a share of GDP) by founding year hit its low point in 1990, peaked in 2000 at almost twice as high, fell after the dot-com bust, and then rose to 2007, fell again with the global recession of 2008–2009, but then bounced back to almost record highs by 2014. As Fazio and colleagues note, “Quantity-based measures document a troubling, three-decade-long decline in the U.S. rate of entrepreneurship. … Conversely, outcome-based measures indicate that the rate of entrepreneurship is rising. Early-stage angel and venture capital financing of new ventures has been on a significant upswing over the past several years.”61
The contradiction that Marc Andreessen tweeted about becomes clearer when one digs into the startup data. The studies that warn of a decline in startups rely on Census data that ask people if they started a business, regardless of whether it is “subsistence” or a “transformational” business. This explains why Kauffman’s “2015 Index on Startup Activity, State Trends” finds that the two most entrepreneurial states are Montana and Wyoming.62 Real high-tech powerhouses such as California and Massachusetts rank just fourteenth and thirty-fourth respectively. As Fazio, et al. write, this “mismatch between index rankings and top hotspots of entrepreneurial activity (like Silicon Valley and Kendall Square) signals strongly that, to the extent that trends in entrepreneurial growth potential are being captured, they have been swamped by the effects of more local or regional businesses.”63
Technological Cycles and High-Growth Startups
As Guzman and Stern note, the number of truly entrepreneurial, high-growth startups is a small fraction of the size of the number of startups overall. But despite their relatively small size, they play an oversized role in the economy in bringing new innovations to the market. For many of these startups either produce technology (e.g., Tesla) or use new technology to create new business models (e.g., Uber), or both (e.g., Google). This means that the rate of technology-based startups is in part a function of the overall rate of technological innovation. As Acs and Audretsch write, “Fundamental changes in technology often result in shifts in the firm-size distribution.”64
As we have observed earlier, these fundamental changes do not proceed evenly but rather in waves, with some periods being more active than others. In fact, technological innovation appears to follow a pattern of repeating S-curves, with waves of technology emerging and then plateauing before the next new wave. This is what Joseph Schumpeter argued when he wrote that “each of the long waves in economic activity consists of an ‘industrial revolution’ and the absorption of its effects.”65 He went on to state, “These revolutions periodically reshape the existing structure of industry by introducing new methods of production—the mechanized factory, the electrified factory, chemical synthesis, and the like.”66
Schumpeter’s key insight was that innovation is not a regular process bringing steady incremental improvements but rather a discontinuous process that leads to waves of innovations. He noted that “these revolutions are not strictly incessant; they occur in discrete rushes which are separated from each other by spans of comparative quiet. The process as a whole works incessantly, however, in the sense that there is always either revolution or absorption of the results of revolution, both together forming what are known as business cycles.”67 One reason technology changes in waves is because the prior technology system establishes firmly committed ways of doing things that are not easily disrupted. Existing systems must become exhausted before institutions look to whole new approaches.
Proponents of Schumpeterian technology long-wave theory postulate five waves to date: (1) the first Industrial Revolution, launched by the steam engine in the 1780s and 1790s; (2) the second revolution, of iron, in the 1840s and 1850s; (3) the third revolution, based on steel and electricity, in the 1890s and 1900s; (4) the fourth revolution, based on electromechanical and chemical technologies, in the 1950s and 1960s; and (5) the fifth revolution, based on IT and communications technology, that emerged in the 1990s and peaked in the last few years. According to this long-wave periodization theory a sixth wave will emerge, but not before an intervening period of relative stagnation perhaps as long as twenty years; a period we appear to be in now.
This may explain why John Haltiwanger, Ian Hathaway, and Javier Miranda find that the high-tech entrepreneurship rate fell from 55 percent in 2000 to 38 percent in 2011.68 They write,
In the post-2000 period, the high-tech sector is experiencing a process of economic activity consolidation, away from young firms and into more mature firms. The high-tech sector looked different than the rest of the private economy did during the 1990s, when the share of young firms was declining in the overall economy but rising in high-tech. In the early 2000s, entrepreneurial activity in the high-tech sector began declining sharply during what is well-known as the dot-com bust.69
But rather than be seen as a serious flaw in the US innovation system, certainly at least part of this decline reflects the fact that as technology systems become more mature, the number of new entrants generally falls. As techno-economic paradigms mature, the most successful startups based on the latest “general-purpose technology” tend to eliminate most of their competitors and dominate an industry, on a national or global scale. Despite what the “big is bad” proponents claim, firm size is largely determined by the economics of production, and that is determined in part by the nature of production technologies. As Blair has stated, “It is technology which largely determines the relationship between the size of plant and efficiency.”70 The pace of entrepreneurial, technology-based startups is therefore related to the phase of the technology long-wave cycle.
Audrestch argues that technology cycles create either entrepreneurial regimes in which new technology opportunities emerge and business models are not yet set, or routinized regimes thereafter when the business models are more set and a few firms emerge as the winners. He writes that in the entrepreneurial regime,
New entrants have a greater likelihood of making an innovation and … are less likely to decide to exit from the industry, even in the face of negative profits. By contrast, under the routinized regime the incumbent businesses tend to have the innovative advantage, so that a higher portion of exiting businesses tend to be new entrants. Thus, the model of the revolving door is more applicable under technological conditions consistent with the routinized regime, and the metaphor of the forest, where the new entrants displace the incumbents—is more applicable to the entrepreneurial regime.71
In summary, new-firm startup activity tends to be substantially more prevalent under the entrepreneurial regime, in which small enterprises account for the bulk of the innovative activity, than under the routinized regime, in which the large incumbent enterprises account for most of the innovative activity.
We see this technology-driven dynamic in the disk drive industry in North America, in which the number of new firms increased from just a couple in the early 1960s when the technology was invented to around eighty-five in the late 1980s. But as the technology began to stabilize and economies of scale got larger, coupled with stiff foreign competition, the number fell to around twenty-five in the mid-1990s.72
Likewise, in the 1980s and 1990s there were numerous entrants in the computer industry, as such companies as Compaq, Dell, Gateway, Leading Edge, Apple, HP, IBM, AT&T, NCR, and others tried to become key players. But as computer innovation slowed and scale economies increased, fewer startups emerged. In the computer and peripheral equipment manufacturing sector, startups fell 45 percent, while the industry actually became less concentrated, with the C4 and C8 ratios declining 7 percent and 8 percent, respectively, largely owing to foreign competition.
We see this pattern in manufacturing. Francisco Louçã and Sandro Mendonça studied the largest 200 manufacturing firms from 1917, 1930, 1948, 1963, 1983, and 1997, or a total of 543 distinct firms for the entire period. They found that only twenty-eight firms were present in all six years—“persistent giants,” Louçã and Mendonça call them—that were largely founded at the turn of the twentieth century or resulted from mergers during that century.73 The persistent giants included such firms as Alcoa, Amoco, CocaCola, DuPont, Deere, Procter & Gamble, Ford, and GM. But more than half of firms appear only once. The number of new firms entering the top 200 was constant over time but the rate changed, with peaks in 1930 (eighty-eight firms) and 1997 (eighty) but lower numbers in the middle periods of 1948 (sixty-two); 1963 (seventy-two), and 1983 (forty-two).74
Using a Schumpetarian long-wave model in which technological change is cyclical, with strong periods and weaker ones, Louçã and Mendonça found that peaks of firm entry into the top 200 are associated with periods of technology transition from one wave to another. It is during these transition points, marked by the emergence of combinations of radical innovations, that new technology systems replace older ones. If this is true, it suggests that today’s relative decline in startups and firm churn could be more related to the stage in the long-wave cycle we are in, with more entry likely to emerge in another decade or so as the sixth wave begins its take-off phase.
In some industries, innovations, often based on federally supported R&D, emerge, often producing a swarm of entrants seeking to capitalize on the opportunities. We see this today as companies such as Tesla (electric cars) and Google (self-driving cars) are getting into the auto business. Indeed, in industries where technology is still evolving and has not matured, we see more startups as entrepreneurs take risks to try to find new opportunities in new markets. For example, startups in the pharmaceutical and medicine manufacturing industry are up 38 percent over the last ten years as entrepreneurs try to capitalize on biotech innovations such as genetic tools. Likewise, startups in the engine, turbine, and power transmission equipment sector are up 87 percent, in large part because of innovations in renewable energy and natural gas power.
In other proverbial “buggy-whip” industries, entry declines because the product is superseded by technological change. For example, startups fell 47 percent from 2003 to 2011 in the magnetic and optical media industry, largely as magnetic tape and CDs were replaced by electronic storage. Similarly, with the rise of cloud computing and the maturing of the data processing and hosting industry, startups fell by 47 percent. Indeed, the last years of the 1990s led to a flourishing of startups as tens of thousands of firms competed to be one of the winners that would emerge. But the market would naturally not support tens of thousands of firms, so there was a shake-out. For example, scores of companies competed to be the dominant Internet search firm but only a few, among them Google and Microsoft, survived.
This suggests that those who blame the slowing rate of innovation on the decline of startups have the causal direction backward. Startups declined because innovation slowed, providing fewer opportunities for new entrants to pursue. Restoring the pace of technological innovation, in part by reversing the budget cuts for federal R&D and expanding tax incentives for business investments in innovation, will help create the opportunities that will lead to increased startups.
In summary, technology is the key driver of firm size. In manufacturing, flexible production technologies are enabling efficiencies with smaller firm size. In nonmanufacturing sectors, IT is enabling greater efficiencies with larger firm size. And because of that, in many of these sectors there is simply less space for the cycling in and out of small startup businesses that was so prevalent before the 1990s. To the extent that larger, more efficient firms are using technology to gain market share and displace many small, inefficient startups, many of them doomed to quick failure; regardless, we should be celebrating this change, not bemoaning it.75
Moreover, to the extent there is any decline in real entrepreneurship, of the kind that produces technology-driven, fast-growing startups, much of that is explained by where the contemporary economy is located in the Schumpeterian long-wave cycle. If a new cycle emerges, perhaps sometime toward the end of the 2020s, we are likely to see a resurgence of tech-based entrepreneurship.
