Out of Business

Measuring the Decline of American Entrepreneurship
July 10, 2012 |

America’s entrepreneurial sector is in deep trouble. Although the mainstream media continues to promote the idea that the nation’s small and upstart businesses are either generally thriving or, at worst, recovering from the sudden blow of the Great Recession, a closer look at the data reveals the exact opposite to be true, with a long-standing decline in the numbers of independent startups per working-age American.

Our self-image notwithstanding, the number of new entrepreneurs and business owners has been dropping – as a percent of the working-age population – for more than a generation, declining by 53 percent between 1977 and 2010. The share of self-employed Americans, meanwhile, has been declining since 1991; by 2010 it had dropped by more than 20 percent.

This is bad for our economy – new businesses are a vital source of new ideas and new jobs. And it is bad politically – the bottom line is fewer independent citizens and a greater concentration of power. Worse, what data we do have may significantly understate the problem. The government has largely failed to adjust how it gathers data to reflect the outsourcing revolution of the last two decades, rendering much of our picture inaccurate at best.

In this report we seek to make clear the extent of the problem, and to do so through a simple and straightforward analysis of available information. We also seek to detail how the problem may be worse than it seems on the surface, which we do by sketching out some of the ways in which government data collection does not account for the new organization of business, and specifically for what appears to be a rapidly growing number of workers and firms that are, for all intents, “captive” to another bigger firm. Such a survey is a vital first step. Before we can begin to understand what is behind this disturbing decline – and what we can do about it – we must first understand the full scale and scope of the problem.

See below for the full report, or download it here.

I.  Introduction

“For all its current economic woes,” the Economist magazine recently asserted, “America remains a beacon of entrepreneurialism.” That idea is at the heart of America’s self-image. Both parties celebrate entrepreneurial small business as the fount of innovation and job creation. Even if America no longer manufactures its own smart phones or computers, we cling to the idea that American entrepreneurs invent most of the new products and services that matter to the world.

Americans also view entrepreneurialism as a vital measure of the nation’s political vibrancy and liberty. We hold that the more independent citizens we have, the more widely will power, responsibility, and voice be distributed. The basic thinking here was best expressed by James Madison, more than 200 years ago, when he wrote that the “greater the proportion” of citizens who are their own masters, “the more free, the more independent, and the more happy must be society itself.”

The number of new entrepreneurs has been dropping for a generation, declining by 53 percent between 1977 and 2010.

Yet how much does this faith in the vigor of American entrepreneurialism stand up to scrutiny? To a casual news consumer, it might seem very well. One well-publicized study claimed that the United States leads all major industrial economies in the percent of the adult population engaged in entrepreneurial activity.[1] “We are in the midst of the largest entrepreneurial surge this country has ever seen,” a CNNMoney article tells us.[2] Or as the Council on Competitiveness has observed, “One of the critical drivers of America’s economic dynamism and flexibility has been the strength of its entrepreneurial economy.”[3]

 

But a closer look at the data behind such assertions raises deep questions about their accuracy. This paper takes a hard look at what we really know about the changes in the size, composition, and health of American small business and entrepreneurialism over the last generation. We find that relevant government data is often incomplete and likely misleading, and, in one important instance, no longer even collected.  Meanwhile even the flawed measures paint a disturbing picture of an economy that is losing entrepreneurial vigor as the rate of independent business formation slows, and of a society in which economic power is ever more concentrated.

Our self-image notwithstanding, the number of new entrepreneurs and business owners has been dropping –  as a percent of the working-age population – for more than a generation, declining by 53 percent between 1977 and 2010. The share of self-employed Americans, meanwhile, has been declining since 1991; by 2010 it had dropped by more than 20 percent.

Worse, evidence suggests that these figures, dismal as they are, actually overstate the ranks of American entrepreneurs, at least as we have traditionally understood the term.  The number of citizens now counted as “self-employed,” for instance, appears to include people performing jobs once done by salaried employees but that have been outsourced to temps, consultants, contract workers, and freelancers. Some of these individuals truly are independent and entrepreneurial. But many more would be better described as contingent workers still subject to the control of a single boss – but now without the regular salary and benefits of a secure job. Such contingent workers accounted for at least a third of the workforce in 2005, the most recent year for which we have data, according to the Government Accountability Office.[4] Similarly, the data on new businesses may include divisions of existing corporations that have been spun off, or firms created to do work outsourced by an existing company. As with the self-employed, existing measures don’t count precisely enough to distinguish between truly independent enterprises and such captive firms. Hence, again, the government apparently ends up counting highly dependent entities as independent businesses.

II.  Fewer New Firms

A good way to get a feel for the extent of the decline – and its potential significance – is by examining the number of new “employer businesses” created every year, as compiled by the Census Bureau’s Longitudinal Business Database. The Census Bureau defines employer businesses as proprietorships, partnerships, and corporations that have at least one paid employee. The database relies on information from a variety of sources, including payroll forms that companies file with the Internal Revenue Service. We start with this particular measure because it gives us a good sense of both the health of America’s entrepreneurial sector, and the health of America’s jobs market.

According to the Census Bureau’s raw data, the number of employer businesses created every year has trended slightly upwards over the last thirty years (figure 1).

Figure 1: New Firms

Figure 1 (Source: Longitudinal Business Database)

This by itself belies the idea that we have been living in an age of exceptional entrepreneurial dynamism. But when one adjusts for the dramatic growth in the size of the working-age population over this period, a much darker picture emerges. Between 1977 (the first year captured) and 2010 the working-age population increased by 77 million. Adjusting for this growth, which was largely driven by the maturation of the Baby Boom generation into its prime productive years, new business formation is revealed to have fallen dramatically (figure 2).

In 1977, for example, there were more than 35 new employer businesses created for every 10,000 Americans age 16 and over. By 2010, the number of new businesses had sunk to fewer than 17, a 53 percent drop. While the Great Recession accelerated the trend, it was clearly in evidence before 2007; by 2006, the number had already fallen by 30 percent. In each decade the peaks and troughs of firm births have sunk lower with each economic cycle, never returning to previous highs. The averages across decades also capture the decline: between 1977 and 1989 entrepreneurs created more than 27 new businesses for every 10,000 working-age Americans, compared to fewer than 25 in the 1990s and around 22 in the 2000s.

Figure 2: New Firms/10,000 Workers

Figure 2 (Source: Longitudinal Business Database)

We see a similar decline when we look at the firm entry rate, or the rate at which citizens create new businesses. This firm entry rate, adjusted for changes in the working-age population, fell from 12.4 percent in 1978 to 7.8 percent 2010, a 37.3 percent drop (figure 3). Business birthrates rose and fell at various points over the last generation, but exhibited a long-term secular decline, eventually dropping to less than two thirds of the rate in 1978.

The latest recession has accelerated the drop, to be sure. Yet even before the 2007 financial crash, new business formation rates were down 18 percent from the late 1980s. Strikingly, even during the late 1990s tech boom, entrepreneurship rates were below those in the previous decade.

Figure 3: Firm Entry Rate (Workforce Adj.)

Figure 3 (Source: Longitudinal Business Database)

The number of new firms as a share of all existing firms shows much the same long-term decline. New firms made up more than 16 percent of all businesses in 1977; in 2010 they comprised only 7.8 percent of them (figure 4). Their average share across decades has crept down from 12.3 percent to 10.8 percent to 9.7.

Figure 4: New Firms/Total Firms

Figure 4 (Source: Longitudinal Business Database)

III.  Fewer New Establishments

Another relevant dataset captured by the Census Bureau’s Longitudinal Business Database is the number of new establishments. This data is more expansive than the number of firms, which only looks at uniquely owned businesses. New establishment counts include multiple outlets owned by a single firm – like when Apple opens a new store – as well as standalone operations – like when “mom and pop” open a new independent restaurant.

It is not hard to imagine a situation in which the number of firms falls, while the number of establishments stays the same or grows. This would be the case, for instance, when chain operations replace independents. Yet here too the aggregate numbers show an overall decline. As with firm formation, the establishment entry rate has fallen since the late 1970s (figure 5). Adjusted for growth in the working-age population, the rate at which new establishments are formed has slid from 14.3 percent in 1977 to 10 percent in 2010. From its peak in 1987, it has dropped 22.2 percent. 

Figure 5: Establishment Entry Rate (Workforce Adj.)

Figure 5 (Source: Longitudinal Business Database)

Per worker establishment births, too, have fallen from a high of around 44 new establishments per 10,000 workers in 1977 down to 28 in 2010, following a historic low of 25 in 2009 (figure 6). In this case, as the graph shows, the post-recession drop has been especially steep. It’s worth noting, though, that the number of new establishments has not fallen as fast as the number of new firms (figure 2).

Figure 6: New Establishments/10,000 Workers

Figure 6 (Source: Longitudinal Business Database)

IV.  Falling Rates of Self-Employment

Buttressing many people’s faith in America’s entrepreneurial dynamism is a common perception that the workforce is increasingly dominated by the ranks of the self-employed. Flexible work arrangements and freelance gigs have made us a “free agent nation,” in the well-known phrase coined by business guru Dan Pink, in the late 1990s. Yet here, too, official statistics call this perception into question. Let’s begin with the relevant data series kept by the Small Business Administration (SBA). It shows the share of the working-age population that is self-employed has been declining since 1994 (figure 7). The share fell steadily until 2002, stayed level between 2003 and 2006, and has continued to drop since. Overall, the decline between 1994 and 2010 was nearly 25 percent.

Figure 7: Self-Employed/10,000 Workers

Figure 7 (Source: Small Business Administration)

This drop in self-employment is also captured by the Bureau of Labor Statistics, which tracks nonfarm workers. The BLS survey asks workers to classify themselves as wage and salary employees, self-employed, or unpaid family workers. Self-employed workers are further separated into those who have incorporated their businesses and those who have not. (The incorporated are considered paid workers of their own firms and thus classified within wage and salary workers.) According to the BLS, the number of Americans who are both self-employed and not incorporated has fallen significantly as a share of the working-age population, from around 463 per 10,000 in 1991 to just 359 in 2011. This is more than a 22.5 percent decline (figure 8). This fall reverses a long trend in the opposite direction that occurred in the 1970s and 1980s (figure 9).[5]

Figure 8: Uninc. Self-Employed/10,000 Workers (1990-2011)

Figure 8 (Source: Bureau of Labor Statistics)

 

Figure 9: Uninc. Self-Employed/10,000 Workers (1970-2011)

Figure 9 (Source: Bureau of Labor Statistics)

The BLS data show a somewhat different picture when it comes to self-employed persons who have incorporated their businesses. As a share of the working-age population, their ranks grew by 35 percent between 1989 and 2008 before dropping off sharply in 2009 (figure 10).

Figure 10: Inc. Self-Employed/10,000 Workers

Figure 10 (Source: Bureau of Labor Statistics)

On first glance this upward trend paints a positive picture. In contrast with the unincorporated self-employed, the incorporated generally have greater economic impact: they are more likely to hire employees and their owners are almost 20 percent more likely to be working on the business full-time, according to a 2010 article by BLS.[6]

Yet this increase may be largely due to factors unrelated to rising entrepreneurship. Many of these incorporations may instead reflect, for instance, a desire to reduce taxes. As several experts have noted, the rising BLS numbers are at least partly made up of proprietors registering as limited liability companies (LLC), a form of partnership that taxes owners at individual (rather than corporate) rates while still protecting their personal assets. More states began recognizing LLCs in the mid-1990s.

“Most of [this increase] looks to have been driven by existing unincorporated activity deciding to incorporate for tax reasons or other benefits,” notes Ronald Wirtz, an editor at fedgazette, a publication by the Federal Reserve Bank of Minneapolis that has analyzed self-employment data closely.[7] Scott Shane, a professor of entrepreneurship at Case Western University, has described other potential motivating factors, such as the ability to list skyrocketing health care costs as tax deductions, as well as the fact that incorporating has become much cheaper and easier in recent years.[8]

Even if we accept this number without question, however, the total share of the self-employed dropped over the last two decades (figure 11). In 1994 there were roughly 663 self-employed for every 10,000 working-age Americans; by 2011 this number was down to around 573, an 13.6 percent decline.

Figure 11: Total Self-Employed/10,000 Workers

Figure 11 (Source: Bureau of Labor Statistics)

V.  The Rise of the Contingent Worker

As we have seen, official statistics show a pronounced downward trend in the percentage of self-employed Americans over the last generation. Yet there is good reason to believe the decline was actually even steeper, and that the full extent of the drop has been hidden by mis-categorization of captive workers and suppliers as independent companies.

One potential problem is that government measures use categories that no longer fit the nature of how more and more Americans actually work. Since the 1990s, for instance, companies have increasingly relied on temporary help to do work that had formerly been performed by permanent salaried employees. These nonstandard arrangements enable firms to hire and fire workers with far greater flexibility and free them from having to provide traditional benefits like unemployment insurance, health insurance, retirement plans, and paid vacations.

The full extent of the drop has been hidden by mis-categorization of captive workers and suppliers as independent companies.

These workers go by many different arrangements and names: temps, contingent workers, contractors, freelancers. While some fit the traditional sense of what it means to be an entrepreneur or independent business owner, most do not. But in government measures, these nonstandard workers are often indistinguishable from them.

In other words, our self-employment numbers blend together different populations with no clear guide as to what kind of work is being counted where. The BLS self-employment numbers, for example, rely on a survey that asks people to self-identify as either working for the government, for a private employer, for a nonprofit organization, or for oneself. Someone who has worked on a freelance basis for a single corporation could identify themselves as either privately employed or self-employed. So too could a contract worker or a consultant. There are few follow-up questions to gauge the actual nature of their work arrangement, even though a taxi driver who owns his car, finds his own clients, and sets his own hours differs meaningfully from an “independent contractor” who drives a company truck and carries that company’s merchandise according to that company’s schedule.

Data that the Census Bureau gathers from the IRS are similarly wanting, because the range of activity capturedby Schedule C filings is so wide. It includes everything from regularly mowing neighborhood lawns for weekend income to selling silk screened t-shirts on Etsy.  Even with the Census Bureau’s minimum and maximum income cutoffs, which exclude proprietors reporting less than $1,000 or more than $1 million, the figures can still include individuals who are earning what qualifies as independent income while still dependent on an employer.

To complicate matters further, we don’t even know how many citizens now fit into this category. In 1995, according to government statistics, there were 39 million of these nontraditional workers, around a third of the U.S. workforce.[9] Over the next decade the government’s statistics showed their number growing by 3 million. But in 2005, the government stopped counting this population in any detail, by halting the Contingent Work Supplement to its main Current Population Survey. In the six years since, we have only a patchwork of proxy measures whose broad categories no longer fit the nuanced variations of nonsalaried work, and thus no longer track it with reliable accuracy. According to one measure, as a share of the total workforce independent contracting may actually have declined over this period. IRS records show that between 1990 and 2010 independent contract arrangements, as measured by the number of 1099-MISC tax forms, grew by only 10 million. The working-age population grew by almost 50 million over this time.

On the other hand, other indirect and direct measures show an increase. The Freelancer’s Union, a group that advocates for more protection and rights for nonsalaried workers, reports that its membership doubled between 2009 and 2011.[10] One BLS article calculates that employment in the temporary help services industry grew from 1.1 million workers in 1990 to 2.3 million workers in 2008.[11] And analysis by EMSI, an economic market and labor research firm, shows that that this population has grown recently too. EMSI’s studies rely on a more expansive measure to count “noncovered” workers, or those proprietors and independent contractors that are not covered by state or federal unemployment insurance and compensation and thus go uncounted by BLS. EMSI analysis estimates that between 2005 and 2010 the noncovered share of the workforce grew from 21 percent to 23 percent, an increase of 4 million workers.[12]

It is not clear why the growth in the numbers of such temporary workers was not matched by growth in the number of self-employed or contract workers. One possible explanation is that these new nonstandard workers filled in for a drop in traditional entrepreneurs and business owners. By this account, the increase in freelancers, consultants, and contractors – who may qualify as self-employed even if their work arrangements are highly dependent – could be masking an even more drastic decline in independent entrepreneurial activity.

VI.  How Outsourcing Gets Mistaken for Entrepreneurship

A good way to understand howcaptive workers and suppliers can be mistaken for independent entrepreneurs and small businesses is through specific examples. Consider, for instance, FedEx’s once routine practice of hiring its truck drivers as independent contractors. Under these agreements, drivers are obligated to lease FedEx trucks, wear its uniform, and deliver its packages along routes assigned by the company.

But were the FedEx contract drivers really in business for themselves? Or was their status as “independent” drivers more of a useful myth that enabled the company to protect itself against unionization, to avoid paying benefits, and to win more business by being able to charge less? (In 2005 it was estimated that this contractor model enabled FedEx to deliver packages at an average rate $1.35 lower than its main rival, United Parcel Service (UPS), which uses unionized, salaried drivers.[13])

Over the years, that question was the subject of protracted legal battles, with rulings going both ways. In some states, for instance, FedEx now requires drivers to incorporate as businesses, and sometimes to “merge” their businesses with other drivers in order to ward off lawsuits demanding they be put on salary.

No present measures capture if a business serves only one other company, or overwhelmingly depends on a single other company.

Despite this long history of legal fights over categorization, and drivers’ continued murky status, official government statistics still do not attempt to distinguish these drivers in any way from the upstart entrepreneur who designs a new sports shoe, opens a local restaurant, or founds a bio-tech firm.

Another dramatic example of how today’s official statistics can inflate the number of people we count as independent business owners and entrepreneurs can be found in the sprawling warehouse galaxy in the Riverside and San Bernardino counties of Southern California. Known as the Inland Empire, this region houses the largest concentration of warehouses on earth and serves as the logistical center of America’s leading big box retail companies, including Wal-Mart, Target, Home Depot, and Lowe’s. In all, half of America’s containerized imports pass through this region.

Before the 1990s, most of these warehouses were owned and operated directly by the giant retailers they serve. Yet today most are run through dizzying chains of contractors and subcontractors. Wal-Mart, for example, outsources much of its warehouse management to Schneider Logistics, a subdivision of the third largest trucking and logistics company in the country. Schneider, in turn, relies on workers supplied by any number of staffing agencies. Some of these employ workers directly. Others subcontract the work out a step further, to other temp agencies.[14]

This increasing reliance by big companies on temporary labor has birthed a legion of temp agencies in the region. According to one union organizer, ten years ago the ratio of direct hires to temps was 80 percent to 20 percent in most warehouses; it has now flipped.[15] As was the case with FedEx, decisions to rely on temp workers may allow a big company to pay less in wages and benefits. It also creates, on paper, a new sub-economy of “independent” businesses that are, for all intents, largely dependent on some firm higher up the ladder.

In case after case the story is the same: significant outsourcing by dominant firms has created sub-economies of dependent companies and temp workers. Yet this fact goes undetected by government measures. No present measures capture if a business serves only one other company, or overwhelmingly depends on a single other company. As a result, we have no way of accounting for how outsourcing has played into the makeup of data on new firms. By including these captive firms and individuals in counts of independent activity, official figures are likely masking an even steeper decline in entrepreneurial activity.

VII.  How Did We Miss This Decline?

The fact that entrepreneurial activity has – by many measures – sharply declined over the last generation will surprise many of us. This in turn raises the question of why so few analyses have picked up on the trend, especially given the level of scrutiny the subject generally draws.[16]

Perhaps most surprising is the widespread failure of researchers to adjust for the robust growth of the working-age population over the last generation. A more fundamental, structural reason is that U.S. government databases are designed to count in ways best suited to measuring the health of big businesses and their macroeconomic impact. This was one of the conclusions reached in 2007 by a panel that extensively studied the collection of business data by statistical agencies. This Panel on Measuring Business Formation, Dynamics, and Performance was convened by the Committee on National Statistics in an effort to improve business data systems.[17] In its findings the Panel observed:

Given its historically predominant focus on large and mature businesses, the current federal business data system was designed to provide efficient measures of gross output and net job creation. This is the case because a relatively modest number of well-established businesses account for a large share of the nation’s aggregate economic activity. As it stands, however, the U.S. business data system is inadequate for understanding many of the mechanisms leading to greater productivity and innovation or the dynamics of firm and job creation. The drawback to the current approach is that, when business dynamics vary systematically with business size or age, it can yield less accurate, potentially misleading, measures of changes in economic activity.[18]

This focus on aggregate economic activity in turn determines other data collection practices that tend to hide the ground-level effects; for instance, relying on establishment counts as a base unit of measure, rather than independent firms.

In recent years some government statistical agencies have expressed an interest in finding better ways to measure small and young businesses. Yet in practice, efforts to distinguish between contingent workers and truly independent business owners and entrepreneurs have actually gone backwards. In the early 1990s the government had designed a new survey to gauge the nature of people’s work arrangements, outside the traditional structure of broad categories. Called the Contingent Work Supplement, it was halted by the Bush administration around 2005. Notably, the decision to discontinue this survey came shortly after a GAO report placed the number of the contingent workforce at 42.5 million, or a third of American workers.[19]

Similarly, official methods of measuring business activity have not kept up with the massive phenomenon of outsourcing. This has resulted in inaccurate statistics about the number of truly new firms engaged in actual entrepreneurial ventures. The bottom line is that we have no reason to be self-satisfied about America remaining “a beacon of entrepreneurialism.” To the contrary, we seem to have great reason to be deeply concerned.


[1] The study notes, “This analysis shows that high-expectation entrepreneurial activity … is highest in the USA with approximately 1.6% of adults involved.” Global Entrepreneurship Monitor, High-Expectation Entrepreneurship 2005, http://www.gemconsortium.org/docs/download/268

[2] Phaedra Hise, “Everyone wants to start a business,” CNNMoney, February 1, 2007, http://money.cnn.com/2007/01/22/magazines/fsb/entrepreneurship.boom.fsb/index.htm

[3] “Where America Stands: Entrepreneurship,” The Competitive Index, February 2007, http://www.compete.org/images/uploads/File/PDF%20Files/WhereAmericanStands-Entrepreneurship.pdf

[4] This estimate by the GAO includes part-time workers, which it includes in its definition of “contingent.” “Employment Arrangements: Improved Outreach Could Help Ensure Proper Worker Classification,” GAO, GAO-06-656, July 2006.

[5] In 1994 BLS redesigned the Current Population Survey, the instrument it uses to gather employment data. After the redesign, significantly higher proportions of the total employed were classified as incorporated self-employed.  This change affects the direct comparability of the data – but, for our purposes, the relative deflation of pre-1994 numbers only makes the decline since then starker. For a brief description of the redesign and its impact, see Steven F. Hipple, “Self-employment in the United States,” Monthly Labor Review, September 2010.

[6] See note 5.

[7] Ronald Wirtz, “Self-employed: To be, or not to be,” fedgazette, 26, No. 1 (January 2012).

[8] Scott Shane, “More of the Self-Employed Incorporate,” February 15, 2010, http://smallbiztrends.com/2010/02/more-of-the-self-employed-incorporate.html

[9] This estimate by the GAO includes part-time workers, which it includes in its definition of “contingent.” “Employment Arrangements: Improved Outreach Could Help Ensure Proper Worker Classification,” GAO, GAO-06-656, July 2006.

[10] “We don’t think the government is counting right,” said Gabrielle Wuolo from the Freelancer’s Union. The organization has been a forceful critic of how workers are categorized by government surveys, which it says are still premised on the single stable job model and no longer fit how a growing share of the country now works. The group says its members are proof that more and more people are entering the “gig” economy, juggling various freelance and temporary work assignments rather than reporting to a single job or employer.

[11] Tian Luo, Amar Mann, and Richard Holden, “The expanding role of temporary help from 1990 to 2008,” Monthly Labor Review, August 2010.

[12]. Joshua Wright, “Data Spotlight: Independent Contractors on the Rise,” EMSI, April 29, 2011, http://www.economicmodeling.com/2011/04/29/independent-contractors-other-noncovered-workers-on-the-rise

[13] Dean Foust, “The Ground War At FedEx,” Bloomberg Businessweek, November 28, halt.

[14] Dave Jamieson, “The New Blue Collar: Temporary Work, Lasting Poverty And The American Warehouse,” Huffington Post, December 20, 2011. http://www.huffingtonpost.com/2011/12/20/new-blue-collar-temp-warehouses_n_1158490.html

[15]Warehouse Workers United organizer Sheheryar Kaoosji, as quoted by Dave Jamieson in “The New Blue Collar: Temporary Work, Lasting Poverty And The American Warehouse,” see note 15.

[16] Case Western professor Scott Shane is a notable exception. Author of The Illusion of Entrepreneurship, he has examined entrepreneurship rates on a per capita basis and has drawn attention to the uniform downward trends.

[17] The Panel was funded primarily by the Ewing Marion Kauffman Foundation.

[18] John Haltiwanger et al., “Understanding Business Dynamics: An Integrated Data System for America’s Future,” Panel on Measuring Business Formation, Dynamics, and Performance (Washington, DC: National Academies Press, 2007), 2.

[19] “Employment Arrangements: Improved Outreach Could Help Ensure Proper Worker Classification,” GAO, GAO-06-656, July 2006.