# Independent Contractors in the U.S.: New Trends from 15 years of

> Briefs, arguments, decisions, and more.

URL: https://www.frixlaw.com/law-library/documents/agency%3Airs%3A2efcb610d31f4c75

## Record

- **Collection:** Agency decision
- **Document type:** Agency decision

## Text

Independent Contractors in the U.S.: New Trends from 15 years of
Administrative Tax Data

Katherine Lim, U.S. Department of the Treasury
Alicia Miller, Internal Revenue Service
Max Risch, University of Michigan
Eleanor Wilking, New York University
July 2019

ABSTRACT
There is growing interest among policy makers and researchers in measuring the prevalence of
independent contractors (ICs), partially due to concern that these workers do not enjoy the benefits
provided to employees. However, identifying IC income is difficult because most existing datasets do
not track it. We make two contributions to understanding changing patterns of IC income receipt.
First, we translate the legal concept of an IC relationship into one that can be used to identify these
relationships in tax data. Second, we use those data to establish several new empirical facts about
individuals who receive IC income and the firms that contract them. We find that the share of workers
with IC income has grown by 1.5 percentage points, or 22 percent, since 2001, pre-dating the rise of
the gig economy and in line with previous estimates of IC growth. Independent contractor income
receipt and its growth are not evenly distributed across workers. The largest share of workers with IC
income are those in the top quartile of earnings who primarily receive wage income. But the fastest
growing group are those in the bottom quartile of earnings who primarily receive IC income. Women
saw more growth in IC income receipt than men, and smaller firms saw more growth in IC labor usage
than larger firms. Together, these trends suggest that the long-run growth in IC labor in the U.S.
cannot solely be attributed to individuals seeking supplemental income, or to the rise of a few online
platform firms, but may represent a structural shift in the labor market, particularly for women.

The views and analysis expressed here are those of the authors and do not necessarily represent the views or policies of
either the Internal Revenue Service or the U.S. Department of the Treasury. We would like to thank the IRS Research
Applied Analytics & Statistics office for supporting this work, as well as the following individuals: Charlie Brown,
Victoria Bryant, Brett Collins, John Guyton, Jim Hines, Ithai Lurie, J.J. Prescott, Shanthi Ramnath, Joel Slemrod, Alex
Turk, and Michael Weber. We also thank participants at the Office of Tax Analysis Conference, the University of
Michigan Public Finance Seminar Series and the National Tax Association Annual Conference on Taxation for their
helpful comments. This research was authorized under the IRS Joint Statistical Research Program and uses de-identified
tax data. All analytical work was completed at IRS and Treasury facilities using official computers. Risch and Wilking
accessed data as IRS employees made possible through the IRS student volunteer program and Intergovernmental
Personnel Act assignments, in support of this research.

1. Introduction
U.S. workers can supply labor to firms either as employees or as independent contractors (ICs):
self-employed individuals who typically provide labor services to multiple firms. The legal relationship
of a worker has a number of important regulatory consequences. Employees are entitled to multiple
protections, including minimum wage and unemployment insurance, to which ICs are not.
Contractors retain greater control than employees over how they carry out their work, and are eligible
to deduct expenses from their taxable income. Recently, there has been growing policy interest in
whether firms and workers are shifting away from traditional employee relationships and toward IC
relationships, and in identifying factors that might contribute to this shift.
Despite these and other serious differences in how employee and IC relationships are treated, it is
difficult to study IC income receipt empirically. Surveys of workers typically do not distinguish IC
income from other self-employment income, and individuals who are contractors might incorrectly
identify themselves as employees due to the often similar nature of their relationship with a firm. Most
administrative datasets also do not separately track those who receive IC income. This is in part due
to legal ambiguity surrounding the definition of ICs, which depends on the extent to which the firm
controls how the work is completed, including whether it provides tools, dictates the timing of work,
and has financial control over the worker.
To overcome some of these challenges, we use data from U.S. tax filings to study recent changes
in IC income. First, we develop a methodology for consistently identifying workers in the tax data
who receive IC income. Conceptually, these are individuals who are being compensated primarily for
providing labor services to a firm, but who are not employees of that firm. In addition to identifying
these workers, we are able to link self-employed contractors with the firms that contract them,
allowing us to track the evolution of firm use of ICs as well. Finally, by linking workers to their
individual income tax returns, we can observe information relevant to the nature of the firm-worker
relationship, such as the degree to which the worker relies on income from the firm, the length of
time a worker has been associated with a specific firm, and whether or not the same worker has
switched classification while working for the firm. These characteristics are important for
understanding how these two types of firm-worker relationships—employers-employees and firmscontractors—have evolved over time.
Our paper makes two contributions. First, we translate the concept of an IC relationship into one
that can be used to identify these relationships in tax data, and we show that applying this definition
has a substantial effect on the measured prevalence of IC income. Specifically, we distinguish
individuals who provide labor services to firms from other self-employed individuals. We do this by
examining the type of information reports individuals receive, the nature and magnitude of their selfemployment deductions, and whether or not they are themselves employers. We find that the share
of workers with IC income has grown by 1.5 percentage points, or 22 percent, since 2001. This
increase is driven almost exclusively by individuals who meet our preferred IC definition. The rise predates the so-called “gig” or sharing economy and corroborates previous findings (Jackson et al. (2017),
Katz and Krueger (2019), Collins et al. (2019)).

2

Our second contribution is to establish several new empirical facts about individuals and firms in
IC relationships. At the individual level, linkage to primary tax filings allow us to identify several
important trends in IC income earners’ sources of labor and household income. 1 First, we find that
the fastest growing group of workers with IC income are those for whom it is their primary type of
income, rather than a supplement to wage income. Second, growth in IC income receipt has been
more rapid among women than men, especially women who are the primary earners in their
households. Third, receipt of IC income has increased the most among tax filers in the bottom income
quartile. While the largest number of IC income earners locate in the top income quartile, growth in
IC income among the bottom of the distribution has significantly outpaced mid and high-earning
households. Taken together, these findings suggest that the growth in IC income has been
concentrated among lower-income workers where it represents a significant source of household
income. Additionally, many of these workers do not have wage earnings and are therefore unlikely to
receive employee benefits from another job.
The linkage between firms and workers in the tax data allow us to study firms’ use of IC labor
over time and across firm characteristics, something that previous work has not been able to do. Our
firm level analysis shows that growth in the use of IC labor, as measured by multiple metrics, is
concentrated among small firms, which we define as those with few employees. We find that the
fraction of firms with at least one IC worker grew by nearly 20 percent, and that, on average, firms are
increasing the number of ICs relative to the number of employees. However, the share of firms’ total
labor compensation paid to IC workers has remained relatively flat, suggesting that, for the average
firm, compensation per IC worker has declined relative to employees. This finding potentially suggests
relatively high growth of lower-skilled ICs, a pattern consistent with the individual-level finding that
IC workers are increasingly likely to come from the bottom of the income distribution.
We find little evidence that firms are increasingly reclassifying existing employee relationships as
IC relationships, but suggestive evidence that firms are hiring more new workers as ICs rather than as
employees. We leverage the panel nature of the tax return data to track transitions into and out of
contracting as the primary source of labor income. Within firms, we find little change in the likelihood
that a firm has employees who transition to IC status, but we do find evidence that more firms have
contractors who transition to employees particularly following the Great Recession.
This paper focuses on ICs specifically because the legal distinction between an IC and an employee
is the relevant distinction for income tax treatment and a variety of regulatory requirements. On
average, workers classified as ICs should have more control over their work process than employees;
however, in practice, there will be overlap between the two groups as many employees have flexible
schedules and a large amount of autonomy in completing their work. There is some concern that
workers classified as ICs are actually misclassified employees. This may occur because there is a great
deal of legitimate legal ambiguity created by the holistic, fact intensive nature of the legal rule defining
contractors. 2 Additionally, there is an incentive for firms to deliberately misclassify employees as ICs
1 Throughout the paper we use the term household generally, but it is identified here by the tax filing unit, which may or

may not coincide with the concept of household used in other data sources.

2 There are two primary sources of this legal ambiguity. First, there is substantial heterogeneity in the standard itself by

jurisdiction. While, in effect, these tests are aimed at understanding the degree of behavioral and economic control, the
language and application of standards by various states and various federal agencies may differ, making it difficult or…

3

to reduce their regulatory compliance costs. 3 Our analysis will not be able to determine whether
individuals who are compensated as contractors are in fact misclassified employees, but we do provide
evidence that firms in industries with historically high levels of worker misclassification are more likely
to hire ICs.
Previous papers that have focused on ICs in particular, or, self-employment in general, have found
mixed results regarding recent trends. Our results on the individual side are broadly similar to those
found in papers using administrative data, which show increases in both the prevalence of ICs and
self-employed individuals since the early 2000s (Abrahams et al. (2018), Jackson et al. (2017), Collins
et al. (2019)). The limited survey data focusing on ICs has also suggested strong increases between
2001 and 2005, and a decline or slower increase since 2005 (U.S. Bureau of Labor Statistics (2018),
Katz and Krueger (2019)).
Our findings do not suggest a simple explanation for why more firms are using ICs and more
workers are receiving IC income. Contractors are a diverse set of individuals who likely have varied
motivations for entering into IC relationships, and we cannot attribute the increase we observe to
changes in firm demand for contractors or individuals’ desire to be contractors. Previous research
finds that self-employment and other alternative work arrangements decline during periods of strong
economic growth, suggesting that at least some of these workers may prefer an employee position
(e.g. Schuetze (2000), Katz and Krueger (2019)). The welfare implications of greater IC income receipt
depends, in part, on whether this is mostly supplementary income, perhaps made possible by
technological changes, or whether this represents the primary earnings for low-income households.
Our results show that over 60 percent of ICs in the bottom half of the income distribution, where IC
growth has been fastest, receive the majority of their labor earnings from IC work. Although it appears
that firms are increasing contractor use through new hires, we also see higher rates of workers
switching from contractor to employee within firms. Further work on the career trajectories of IC
workers could shed light on whether these positions are a stepping stone to earnings growth either as
an IC or employee, or if they represent inferior arrangements for many workers seeking to be
employees. Understanding the role of IC work within a career is particularly important given that ICs
are increasingly female, in the bottom half of the income distribution, and have IC income as a primary
earnings source.

…confusing for employers to understand which standard applies in which context. For example, under the Fair Labor
Standards Act (FLSA), courts have applied an “economic realities” test, with six factors to assess the relationship
between the worker and business. See McFeeley v. Jackson St. Entm’t, LLC, 825 F.3d 235, 241 (4th Cir. 2016) (citing
Schultz v. Capital Int’l Sec., Inc., 466 F.3d 298, 304–05 (4th Cir. 2006)). Contrast this guidance to that offered by the
EEOC manual, which states “[t]he question of whether an employer-employee relationship exists is fact-specific and
depends on whether the employer controls the means and manner of the worker's work performance.” 2 Equal
Employment Opportunity Comm'n, EEOC Compliance Manual, § 2–III, at 5716–17 (2008)). The second source of
ambiguity arises from the nature of the rule itself as a multi-factor balancing test, which does not view any one factor as
decisive. Instead, the arbiter is supposed to weigh the relative importance of these factors or their absence on a case by
case basis. While this nuanced approach may be more accurate than a bright-line rule, it can be difficult, ex-ante, for
employers to apply this nuance consistently to their own relationship.
3 There may also be strong tax incentives for workers to instigate or collude in this misclassification, for example, to
assert inappropriate deductions that will increase after-tax compensation relative to wages.

4

2. Background
In this section, we provide additional detail on the legal distinction between IC and employee
relationships and discuss the regulatory consequences of this distinction. In general, there are stark
differences between employees and IC in the treatment of their income under the tax system, the
applicability of labor protections, and their access to social insurance programs. These differences
highlight the need to develop criteria to empirically identify and study IC relationships.
The legal distinction between an employee and an IC originates from common law principles of
vicarious liability, which distinguish circumstances in which the contracting firm is legally responsible
for the actions of their workers (employees) from circumstances in which the firm is not responsible
(ICs). In the U.S., the establishment of an employer-employee relationship depends to a large extent
on the level of control the purchaser of services has over how the work is completed, such as whether
or not the purchaser provides tools, dictates the timing of work, and the extent to which the purchaser
has financial control over the service provider.
These factors have been codified, with minor variations, into state laws as a multi-factor test
weighing the relative importance of several features of the relationship to determine whether the
worker is an employee. A similar approach has been adopted by a number of state and federal agencies
for regulatory purposes. Enforcing the distinction in this context has proved difficult because
evaluating the holistic nature of the relationship between a worker and her firm is factually intensive
and requires significant commitment of agency audit resources. 4
Several features of the tax code depend on worker classification. ICs are treated as sole proprietors
and are entitled to claim "above the line" business expense deductions, and they are not subject to
payroll or income tax withholding. Many other federal regulations intended to protect workers apply
only to employees. Major anti-discrimination legislation, intended to protect workers, such as the AntiDiscrimination Act and Fair Labor Standards Act have this feature, as do several laws that places
requirements on employers for the benefit of employees, e.g. Unemployment Insurance programs, the
Family Medical Leave Act and the Affordable Care Act. While by no means exhaustive, these examples
suggest the vast, and largely implicit, impact that worker classification has on the costs—and
benefits—of a given worker-firm relationship.

4 Consider the guidance provided by the IRS to potential employers in deciding whether a worker is an independent

contractor:
Businesses must weigh all these factors when determining whether a worker is an employee or independent
contractor. Some factors may indicate that the worker is an employee, while other factors indicate that the
worker is an independent contractor. There is no “magic” or set number of factors that “makes” the worker an
employee or an independent contractor, and no one factor stands alone in making this determination. Also,
factors which are relevant in one situation may not be relevant in another.
“Independent Contractor (Self-Employed) or Employee?” Internal Revenue Service.. U.S. Department of the
Treasury. https://www.irs.gov/businesses/small-businesses-self-employed/independent-contractor-selfemployed-or-employee. Accessed July 16, 2019.

5

3. Literature
In this section, we contrast our approach with those taken by recent papers on ICs, and with the
broader literature on self-employment. We highlight differences in study populations and
methodologies that may explain differences in the estimates.
Our work relates to a number of recent studies that document trends in the prevalence of workers
outside of traditional employee-employer relationships. However, the worker population in our
analysis differs from the existing literature in at least two important—yet subtle—respects. First,
conceptually, we construct our preferred definition of IC to adhere as closely as possible to workers’
legal classification. We do this deliberately because the legal dichotomy between employee and IC
determines the tax treatment of income, and a number of other regulatory requirements. By using the
type of form that reports the compensation to the worker to distinguish ICs and employees, we can
identify whether the worker is treated as an IC or employee by the firm for tax purposes. There may
be workers who are legally misclassified according to this definition. For example, some workers may
receive IC income, but, if audited, would likely be considered by the IRS to be an employee. Our
analysis cannot identify these individuals, although we provide evidence that firms in industries with
historically high levels of worker misclassification are more likely to issue their workers Form 1099MISC/Ks. Second, our sample includes all individuals who are subject to information reporting on
income on Form 1099-MISC/K, rather than self-employed individuals more generally.
Previous work can be coarsely divided into four categories based on two criteria; first, by the data
source—collected by survey or in an administrative process, and second, by the population focus-ICs or self-employed. Generally, survey sources show no change or small declines in ICs or selfemployment rates more broadly, while administrative sources show increases in recent decades.
Abraham et al. (2018) combine the two types of data to show that there has been a rise in the types of
self-employment income reported in tax data that would not be well captured in surveys. For example,
IC work that may represent a secondary job, which is not reported in some surveys, or IC work where
the individual may not consider herself self-employed and may either report no employment or
characterize the income to the surveyor as wage income.
Only the Bureau of Labor Statistics’ (BLS) contingent worker survey (CWS) and Katz and
Krueger’s CWS replication have asked explicitly about IC work separately from self-employment more
generally. Both of these surveys only record information on the individuals’ main job, which based on
earnings would only include around half of our ICs. Katz and Krueger’s (2019) preferred estimate
suggests a very small increase in ICs between 2005 and 2015 (0.2 percentage points) while the CWS
suggests a decrease over a slightly longer period, between 2005 and 2017; however, the CWS does find
a large increase in IC use between 2001 and 2005 (0.9 percentage points). The paper with the most
similar approach to ours is that used by Collins et al. (2019), who use administrative tax data to identify
individuals receiving a Form 1099-MISC/K with a specific focus on those working for an online
platform economy firm. They find increases in the number of Form 1099-MISC/K recipients and an
increase in the share of the workforce receiving a Form 1099-MISC/K between 2000 and 2016 of
around 1.9 percentage points. They argue that recent increases after 2013 are driven almost entirely
by online platform economy activity.
6

Another set of studies has focused on an overlapping, but distinct group of individuals
characterized as the self-employed. Conceptually, our population is at once both broader and narrower
than this group; our population includes individuals who would not identify themselves as selfemployed on a survey, and will likely exclude some individuals who would, as shown in Abrahams et
al. (2018). Furthermore, both Abrahams et al. (2018) and Jackson et al. (2017) use filing a Form
Schedule SE as their administrative data measure for self-employment. Our sample will exclude some
of these individuals because they are not issued a Form 1099-MISC/K , or, because they do not meet
our definitions of an IC. Although individuals who receive Form 1099-MISC/Ks are taxed as sole
proprietorships, many do not file a Schedule C or Schedule SE—either because they fall below the
threshold for income tax filing, or because they fail to report the income on the correct form. In
consequence, these individuals will be in our sample but not in the “self-employed” population of
previous papers. 5 Jackson et al. (2017) identify small increases in self-employment, and attribute this
increase to individuals with low-levels of business deductions, which is consistent with our general
findings. Similar to our results, they find that these increases in self-employment pre-date the
introduction of online platform economy companies such as Task Rabbit, Uber, and Lyft.
Finally, a number of papers have focused on a much broader population called “alternative”
workers, which generally include ICs, temp agency employees, workers at contracting firms, and oncall workers. The idea behind grouping these labor arrangements together is that they may share
substantive economic features, such as flexible hours, or finite duration. These papers find mixed
results regarding the growth of such alternative workers, reflecting the sensitivity of findings to the
data source and exact definition of non-traditional work being used. For example, using data from a
survey they administer, Katz and Krueger (2019) find a 1-2 percentage point increase in alternative
work between 2000 and 2015 while the BLS CWS finds no increase in alternative work between 2005
and 2017 (U.S. Bureau of Labor Statistics (2018)). Our paper examines a subset of these workers,
those whom we characterize as ICs.

4. Data
In this section, we first describe our methodology to identify IC income using administrative
tax data, which involves three steps to delineate workers receiving contracting income that is primarily
for labor services provided to firms. First, we identify potential IC labor transactions using Forms
1099-MISC and 1099-K. Next, we distinguish the type of taxpayer based on the recipient identifier
(individuals are identified by SSN, businesses by EIN) and link them to their respective tax filings. 6
Finally, we use information on the amount and nature of deductions to exclude businesses that we do
not consider ICs because they do not appear to receive payment mainly for labor services provided

5 A recent paper by Collins et al. (2019) highlights the differences between the two populations, finding that around 40

percent of Form 1099-MISC recipients in 2016 did not file a Schedule SE and that around 45 percent of those with a
Schedule SE do not receive a Form 1099-MISC, meaning that these individuals will be in our sample but not in the
“self-employed” population of previous papers.
6 Individuals file a Form 1040, while businesses file a Form 1120, 1120-S or 1065.

7

by the owner. Each of these steps results in a significant change in levels of IC workers, but not in
trends.
Next, we provide information on our sampling methodology for the individual and firm level
analyses. A major advantage of using tax return data is that we can link workers to firms, which allows
us to use our individual level definition of ICs aggregated up to the EIN level to provide information
on firm use of ICs. 7 While other sources, like the integrated Longitudinal Employer-Household
Dynamics (LEHD) – Longitudinal Business Database (LBD) infrastructure, provide a panel of firmemployee relationships, sole proprietorships and partnerships, it is not possible to link the subset of
sole proprietors who are unincorporated ICs to the firms that pay them.
4.1 Identifying Independent Contractor Income using Administrative Tax Data
We start with a sample of recipients with positive amounts of non-employee compensation,
which is reported on Form 1099-MISC, box 7. The IRS requires that businesses issue Form 1099MISC to individuals, or other businesses, for services provided by someone who is not an employee
of the issuing business. 8 We start with a 1 percent annual cross section of all recipients (i.e. making no
initial restrictions) for each tax year 2001-2016. However, because we are trying to identify individuals
providing services, we refine our sample to exclude Form 1099-MISC recipients who employ others.
We do this for two reasons. First, conceptually, we consider employer businesses to not be ICs because
their activity rises above merely an individual providing their own labor services to a firm. Second, we
would have no way to determine whether the employee or the owner was providing labor services to
the Form 1099-MISC issuing business. For example, a Form 1099-MISC could be issued to a catering
company with many employees or to a law firm for attorney services.
Form 1099-K was introduced in 2011 as an information report on credit card transactions and
third party payments that exceed both $20,000 and 200 transactions in a year. 9 ICs who receive
compensation in the form of credit card payments may have part or all of their contract income
reported on Form 1099-K rather than Form 1099-MISC. In order to include ICs for whom all of their
contract income is reported on a Form 1099-K, we draw a separate 5 percent random sample of Form
1099-K recipients in each year from 2011-2016. 10 Many Form 1099-K recipients will not be considered
ICs because these forms are issued to any business that accepts credit cards as payment for goods or
services, an issue which underscores the importance of using additional information on recipients to
define ICs. For sampled Form 1099-MISC recipients, we also link to any 1099-Ks that they receive in
7 As we discuss below an important caveat to our analysis is that firms can have multiple EINs so our EIN level analysis

may differ from an analysis at the parent firm level using a different data source such as the LEHD-LBD.

8 In general, these forms are not required to be issued to corporations. Exceptions include fish purchases for cash,

attorneys’ fees and payments by a federal executive agency for services, which can generate a Form 1099-MISC issuance
to a corporation and are reported in box 7.
9 Although these are the legal threshold requirement to issue a Form 1099-K, many firms issue the forms to recipients
with fewer transactions or lower dollar values.
10 An example of an independent contractor that would receive all or most of their income on Form 1099-K as opposed
to Form 1099-MISC would be a ride share driver who receives their payments directly from customers using credit
cards/electronic payments as mediated through the ride share app.

8

order to count total contractor income for individuals that receive both forms. Analogously, for
sampled Form 1099-K recipients, we link to any Form 1099-MISCs received. This also allows us to
account for those who receive both forms so we can avoid double-counting this group across samples.
Figure 1 shows that the total number of Form 1099-MISC/K recipients has increased over
our sample period. The number of Form 1099-MISC recipients increased from approximately 18 to
26 million from 2001 to 2016. When including Form 1099-Ks, there are over 30 million recipients in
2016.
4.2 Moving from Form 1099-MISC and 1099-K Recipients to Independent Contractors
To refine the sample to focus on ICs, we match the Form 1099-MISC/K recipients to their
income tax returns to eliminate recipients that are employer businesses or businesses with large levels
of deductions. Specifically, we do not consider businesses with total deductions excluding car and
travel that exceed $10K in 2001$ to be ICs under our preferred definition. As we show below, each
step of the methodology to get from Form 1099-MISC/K recipients to our preferred IC definition
decreases the size of the IC labor force; however, the trend over our sample period is driven by
individuals who meet our IC definition.
We distinguish between two types of Form 1099-MISC/K recipients: individuals and nonsole proprietorship businesses. Individuals are defined as recipients who have a Form 1099-MISC or
K issued to a social security number (SSN) or to a business’s employer identification number (EIN)
that matches with a sole-proprietorship Schedule C. Non-sole proprietorship businesses are recipients
with a Form 1099-MISC/K that is issued to an EIN that does not match a Schedule C EIN.
Individuals make up the majority (82 percent in 2016) of recipients as shown in Figure 1 and Appendix
Table A1. Additionally, we see that there are almost 5 million recipients receiving a Form 1099-K but
no Form 1099-MISC as of 2016, and just under half of these recipients are non-sole proprietorship
businesses. Because Form 1099-Ks may be issued to any individual or business receiving substantial
credit card or electronic payments, we expect that many Form 1099-K recipients are businesses using
these payment methods for transactions underscoring the need to use additional information to
identify ICs.
For individual recipients, we pull information from the household tax return Form 1040 for
those who filed. Additionally, we find information on the income and expenses of the business on the
Form 1040, Schedule C for those who filed a Schedule C. 11 The group of individuals who filed Form
1040s but not Schedule Cs may have reported their Form 1099-MISC/K income elsewhere on their
tax return, or failed to report it. Focusing on individual recipients, the majority (approximately 85
percent in 2016) can be matched with a Form 1040 tax return. Those who cannot be matched with a
Form 1040 may represent non-filers or those who incorrectly provided taxpayer identification
numbers (TINs).
Individuals may not be required to file a Form 1040 if they have very low levels of income or
receive most of the income from social security. In our sample, 99.7 percent of non-filers in a given
11 The Schedule C nominally reports profits or losses from sole-proprietorship businesses, but all Form 1099-MISC

payments received by an individual as business or self-employment income should be reported on a Schedule C.

9

year have filed a 1040 in some other year. The average and median age of non-filers is about 40 years,
and the 25th and 75th percentiles of age are about 28 and 51 years respectively, so it is unlikely that they
mostly represent retirees receiving social security benefits. The median Form 1099-MISC/K earnings
for non-filers is about $4,000-5,000, with an average of approximately $14,000. Therefore, most nonfilers earn relatively small, but not nominal amounts of IC earnings.
We exclude non-filers for the remainder of the analysis as we explore income and demographic
trends in contract labor, but future examination of this group would be beneficial for statistical and
tax compliance analyses. If we were to include non-filers in our analysis, we would calculate an increase
in share of the workforce with IC income of 2.5 percentage points, or 32 percent. By excluding nonfilers, we are conservative with respect to the overall growth in IC labor over this period; however,
non-filers are likely to be relatively low-income, which would support the trends we present in Section
5 showing a large relative increase in ICs in the bottom half of the income distribution.
While the majority of individual filers can be matched to a Schedule C as shown in Appendix
Table A1, a substantial proportion, about 28 percent, cannot be matched. Since Form 1099-MISC box
7 income should be reported on Schedule C, this implies that many tax filers may be incorrectly
reporting their Form 1099-MISC income, or potentially are not reporting it at all, and that many are
not claiming any Schedule C deductions for their contract labor. This mismatch is also documented
using tax data in Collins et al. (2019).
For non-sole proprietorship business recipients, income and expense information is found on
Form 1120 for C corporations, Form 1120-S for S corporations, or Form 1065 for partnerships. We
match the EIN recipients to each of these forms and use information from the business returns on
the number of owners and amount and type of deductions to further characterize whether the owners
may be ICs.
Next, we use the information on the matched tax returns to establish a sufficiently narrow and
consistent concept that can be informative for economic analysis of a coherent group of workers as
ICs. We focus on two definitions of ICs: first a “broad definition” and second our “preferred
definition”, which applies additional restrictions. The broad definition includes all individual recipients
who matched to a Form 1040, and all non-sole proprietorship business recipients that matched to an
S-corporation tax return reporting one owner and no employees. The restrictions on the non-sole
proprietorship business Form 1099-MISC/K recipients are designed as the broadest way to identify
incorporated ICs. 12 Our preferred IC definition uses the deductions claimed by the Form 1099MISC/K recipient as a way of differentiating between small businesses and ICs. In this definition, we
exclude individual Form 1099-MISC/K recipients with greater than $10,000 (2001$) in Schedule C
deductions, excluding car and travel related deductions, and non-sole proprietorship business Form
1099-MISC/K recipients with greater than $10,000 in total deductions reported on their business tax
return. 13 We do not include car and travel deductions on the Schedule C because these are common
deductions for ICs and are becoming even more relevant with the prevalence of contracting through
12 We focus on S corporation owners because partnerships by definition have multiple owners. C corporations generally
do not receive Form 1099-MISCs except in certain circumstances, and we cannot identify their owners in our data;
however, they comprise a small fraction of Form 1099 recipients.
13 Vehicle and travel expenses are separately reported on the Schedule C, but are not similarly on the Forms 1120-S,
1120, and 1065.

10

online platforms. These restrictions are designed to remove businesses that have substantial capital
investments in order to focus on ICs who are receiving income mainly from their labor services. 14
Figure 2 plots time series for these two definitions of ICs. In Figure 1, we saw that between
14 and 18 percent of Form 1099-MISC/K recipients were non-sole proprietorship businesses. Figure
2 shows that the majority of these businesses do not meet either of our definitions of ICs; less than
10 percent meet the broad definition in 2016, and less than 3 percent meet our preferred definition.
As seen in Figure 2, using our preferred definition, the IC levels and trends are indistinguishable
regardless of whether we include non-sole proprietorship businesses. For this reason, going forward
we exclude non-sole-proprietorship EIN recipients from our main analyses.
We find that restrictions based on the level of deductions play an important role in determining
the number of ICs but have little effect on the qualitative time trend. The majority, about threequarters, of the broadly defined ICs have less than $10,000 in deductions. Additionally, the contractors
who meet our preferred definition constitute over 75 percent of the total increase in individual Form
1099-MISC/K recipients over this period. The deductions restriction removes the discontinuous jump
in the number of recipients in 2011 caused by the introduction of the Form 1099-K. We take this as
evidence that the restriction is likely effective in removing businesses from our IC population, allowing
for a more consistent definition of IC labor.
The deduction restriction, though somewhat arbitrary, was selected to allow for a reasonably
large level of deductions for an individual providing labor services while also removing recipients who
are likely receiving compensation for capital investments as well as their labor. Figure 3 explores the
sensitivity to this restriction. Panel A shows car and travel deductions as a share of total deductions
for sampled Form 1099-MISC/K recipients, where each point represents a binned average. For those
with relatively low total deductions, car and travel deductions make up a large fraction of their total
deductions. Because ICs may incur large car and travel expenses, we do not want to exclude those
individuals from the analysis even if the car and travel expenses are relatively large. For businesses
with high levels of deductions, car and travel is a relatively small expense. Panel B shows the
distribution of total deductions less car and travel in our sample. We see that our $10,000 deductions
threshold choice is consequential for how many ICs are included in the definition. The distribution
begins to flatten out at $10,000 so increasing the deductions limit is only slightly more inclusive. For
example, if we increase the limitation to $15,000, 94 percent of those ICs meet the under-$10,000
restriction as shown in Table 1. Finally, Panel C shows time series of ICs for different deductions
limits and shows that time trends are very similar regardless of the level of the deduction restriction
so the trends analysis is not very sensitive to the level of the restriction around this range.
To further understand how the deduction restrictions relate to the IC labor concept, we
analyze the number of distinct payers from which a contractor receives a Form 1099-MISC/K. We
expect that individuals providing labor contract services, on average, contract with fewer firms than
do businesses that may sell goods and services as broadly as possible. This distinction is more relevant
for Form 1099-MISC recipients than Form 1099-K because a recipient could receive one Form 1099K from one payment processor that represents payments from many customers. Table 1 disaggregates
14 See Knittel et al. (2011) as a precedent for using deductions to differentiate small businesses from contractors and large

businesses, and for treating car and travel deductions differently.

11

the different potential IC definitions by the number of distinct payers from which the individual
receives a Form 1099-MISC/K. It shows that ICs that meet our preferred definition have fewer Form
1099-MISC/K payers than those meeting the broader definition, and the profiles are very similar
regardless of whether using the $10,000, $5,000 or $15,000 restriction level. For these definitions, 98
percent of ICs have fewer than five distinct payers.
Our definitions of ICs are imperfect and will certainly include some Form 1099-MISC/K
recipients who are not ICs and exclude some who are. We include recipients that match to a Form
1040 but have no Schedule C. For these individuals, we do not have information on the deductions
or activities associated with the non-employee compensation so they may not be ICs. However, we
believe it is unlikely that these individuals have large deductions associated with their IC income
because they failed to file a Schedule C in order to claim those deductions. Additionally, Form 1099MISC/K information reports are sent directly to the IRS, creating a high risk of detection for a
business owner with substantial business income who would avoid filing a Schedule C in order to
evade taxes.
In our preferred definition, we use information from the Schedule C to exclude owners with
large levels of deductions under the assumption that the deductions are associated with the business
receiving the Form 1099-MISC/K. It is possible that the Schedule C does not reflect the activity of
the individual receiving the Form 1099-MISC/K for two reasons. First, the individual taxpayer may
have multiple sole proprietorships in which case the Schedule C deductions may represent deductions
from various contracting activities and not only the activity associated with the given Form 1099MISC/K. In this case, the deduction restriction is conservative in that we only count an individual as
an IC if they have a less than $10,000 deductions in total. Second, prior to 2007, Schedule Cs did not
identify the individual owner of the sole proprietorship for joint filers. In order to provide continuity
in our methodology, we restrict on the total household Schedule C deductions throughout our sample. 15
Again, the restriction is conservative because ICs with individual-level deductions under $10,000 but
with total household deductions exceeding $10,000 will be excluded based on the deduction
restriction. Finally, in our sample, there are a non-trivial percentage of Form 1099-MISC/Ks that do
not match to an individual or a business entity. These forms could represent instances where an
individual is acting as an IC but either provided an incorrect SSN or EIN or did not file a tax return,
and we would fail to include them in our definition.
For the remainder of the paper, we will report results using our preferred definition of ICs:
individual Form 1099-MISC/K recipients reporting less than $10K in deductions, excluding car and
travel expenses, and with no employees. We will occasionally compare our preferred definition with
the broader definition of individual Form 1099-MISC/K recipients with no deductions restriction in
order to highlight the differences or similarities between these concepts. Additionally, we will generally
group together Form 1099-MISC and Form 1099-K recipients under these definitions.

15 See Appendix Table A1 for information on the share of Schedule C’s that could only be matched to the spouse after

2007; it remains roughly constant at 2 percent.

12

4.3 Employee and Worker Datasets
In order to place our IC results in context, we also create a sample of traditional Form W-2
wage earning employees. We randomly sample a repeated cross section of 0.2 percent of all Form W2 recipients, 2001-2016. In Section 5, we use this dataset to examine changes in the prevalence of IC
income, employee income, and both types of labor income.
To study worker-level transitions between IC and employee status, we draw a 0.2 percent
random sample of all taxpayers for each year 2001 to 2016. For each sampled taxpayer, we link Forms
1099MISC/K, W-2, Schedule C and 1040 information for the subsequent and prior year. In section
6, we study transitions based on the status of the individual from the prior year to the sampled year.
4.4 Construction of the Firm-Level Dataset
We construct a firm-level dataset to study businesses’ use of IC workers using the definitions
developed in Section 4.1. The dataset is constructed by drawing a 2 percent random sample each year
from 2001-2016 from the universe of employer identification numbers (EINs) that issue at least one
Form W-2 in that year. 16 Firms are identified by their TINs, as they appear on Form W-2 (for
employees) or Form 1099-MISC or Form 1099-K (for independent contractors). Linked to each
sampled EIN are all employees who receive a Form W-2 and all workers issued a Form 1099-MISC
or Form 1099-K from that EIN in the sample year. The ideal unit of analysis for our paper is the
level at which labor demand decisions are being made within a firm. Unfortunately, we are not able to
easily construct a firm level concept using the tax data so our analysis will be at the EIN level. For
convenience, we will use the term firm throughout the paper, but an important caveat to our analysis
is that firms can have multiple EINs. In practice, an EIN is likely to be associated with some boundary
within the firm. For example, if one EIN is used for payroll and another for general operations, our
analysis would be unaffected. However, if one EIN is used to pay contractors and another to pay
employees, our analysis would fail to pick up on the substitution between the two types of employment
at that firm.
The sample contains approximately 130,000 EINs per tax year. We measure firm size as the
number of employees issued a Form W-2. The size distribution of firms follows a power law
distribution, with very few large firms in the population. As a result, within each tax year, the sample
contains approximately 3,000 EINs with over 100 employees, or 2 percent of EINs. We discuss our
firm level results in Section 7. Table 10 reports number of observations and means for each of these
variables in each tax year. We build out from this same cross-sectional sample in the analysis in Section
8 (i.e. the cross-sectional sampled EINs form the longitudinal spline for studying work transitions.).
In Section 8, we focus on the extent to which firms have the same workers changing
classification over time. We expand the year-stratified sample of firm EINs used in our Section 7
16 This restriction is partly for convenience: all firms with at least one employee is in the cleaned SOI Databank, which is
already linked to Form W-2 workers. Additionally, this definition mirrors the definition of an employing firm that would
be identified in other datasets and that has analyzed in previous literature.

13

analysis to include worker-level panel information. First, for each firm-year in the sample, we
identify all individuals issued a Form W-2 or Form 1099-MISC/K by that firm in that year. Then,
for each individual identified in the previous step, we pull information from Forms 1040, Schedule
C, W-2, and 1099-MISC/K for the previous and subsequent tax years. In each tax year, an individual
is characterized as having one of three mutually exclusive relationships with the sampled firm: (1) an
IC, (2) an employee (3) no relationship. 17 Individuals who receive both a Form W-2 and a Form
1099-MISC/K in the same tax year from the same firm are characterized by whichever form
reported higher compensation. 18 Transitions are then defined as movements from one type of
primary income (e.g., IC) to another type (e.g., employee) in consecutive tax years. We also use the
dataset to study the classification of new hires at firms.

5. Individual Trends in Independent Contractor Income Receipt
In this section, we document growth in the number of workers receiving IC income, by
state, industry and place within the income distribution. We also examine the dependence of the
individual and their tax unit on IC income. We pay particular attention to women ICs, who have
driven much of the overall growth in the IC workforce. Unless otherwise stated, our analysis relies
on our preferred definition of ICs set out in Section 4—individual Form 1099-MISC/K recipients
who are not employers and claim less than $10,000 in Schedule C deductions, excluding car and
travel deductions.
5.1. Long-run Trends in IC Income in the U.S. Economy
We estimate that share of workers characterized as ICs has increased by 1.5 percentage points
between 2001 and 2016. The share of workers with any IC earnings has increased by 22 percent over
this period, while the share of the workers with only Form W-2 earnings has decreased by
approximately 1.5 percent. 19 The majority of the growth (69 percent) occurred prior to 2011, and
individuals who receive the majority of their labor income from contracting represent 78 percent of
the growth. These aggregate results are similar in magnitude to those reported in Collins et al. (2019)
who find a 1.9 percentage point increase in the “1099 workforce”; however, they find that half of that
increase occurred after 2013. The BLS CWS finds an increase of independent contractors of around
17 We only consider income earned from sampled firms. Individuals who receive no information returns, or who receive

only a Form 1099-MISC or Form 1099-K but fail to meet our IC definition are characterized as having no relationship
with that firm in that year. This restriction is consistent with our focus on individuals providing labor services to a firm
and on ICs that are close substitutes for employees.
18 Employees should generally only receive a Form W-2 from their employer while contractors should receive a Form
1099-MISC/K. If the worker changes status during the year, they could receive both forms that year. In the data, however,
around 7.6 percent of firms in 2016 issue both forms to at least one worker. Of those 7.6 percent of firms, the vast majority
(81 percent) issue both forms for at least two consecutive years to the same worker and 40 percent issue the two forms to
the same worker for three consecutive years.
19 The total workforce (ICs plus employees) can be found by summing across the 4th through 6th columns of Table 2.
Shares are derived by taking the total of those with any IC income (the sum of columns 4 and 5) and those with only W2 labor earnings (column 6) and dividing by the total workforce.

14

0.9 percentage points (from 6.1 percent to 7 percent) between 2001 and 2005. Our preferred definition
shows that ICs make up 6.6 percent of the workforce in 2001 and 8.1 percent in 2005. The levels are
a bit higher than those found in the BLS survey, and we see growth between 2005 and 2015 that the
CWS does not. As discussed by Abraham et al. (2018), there has been an increasing divergence
between administrative and survey data in measuring self-employment, which may explain why we
find increases in ICs between 2005 and 2016 while the CWS does not. Additionally, we include ICs
whose main job is likely a wage and salary position. These individuals would not be considered ICs in
the CWS because the survey only asks about main jobs.
The steady increase in IC income receipt from 2001-2016 described in Section 4 contrasts with
slower growth among individuals receiving traditional Form W-2 wages. Figure 4, Panel A presents
the percentage growth between 2001 and 2016 for three groups of workers: those with Form W-2
income only; those with IC income only, and those with both types of income. Incidence of IC income
has grown substantially faster than Form W-2 income since 2001, with the fastest growth among ICs
with no W-2 earnings (50 percent by 2016). In contrast, growth among workers who only receive
Form W-2 income has been much slower, although these workers make up over 90 percent of
workers, as shown in Table 2. Interestingly, we find diverging trends in the years after the Great
Recession with Form W-2 only recipients declining while workers with any IC earnings continued to
grow. These findings suggest that the Great Recession may have intensified the growth of IC income
earners relative to Form W-2 recipients, a finding that is consistent with the evidence in Section 8 that
after the Great Recession firms were much more likely to hire new workers as ICs rather than
employees.
5.1.1 Trends in IC income growth by state and industry
Next, we compare the growth in ICs across industries and states. We find that the increase in
ICs varies greatly across states, from negative growth to over a 90 percent increase (Figure 5). 20 When
comparing the relative growth in ICs to employees, we find that ICs have had larger percentage
increases than employees in almost all states. Though correlated, state-level growth in ICs does not
simply follow employment growth in a state. There are many states where IC growth is well above
average but employee growth is below average, and some states had negative employment growth but
increases in IC labor. This pattern may be partially explained by better economic conditions leading
to lower levels of contracting as argued in Katz and Krueger (2019).
The industries with the most ICs are the “professional, scientific, and technical services”
followed by “other services” and “health care,” and all three of these industries have had large
increases in the number of ICs hired between 2001 and 2016 (Figure 6). 21 There was strong growth
across most industries in ICs with the exception of “Manufacturing”, “Finance and Insurance”,

20 Appendix Figure A1. shows the corresponding levels for contractors in 2001 and 2016.

21 We were only able to match 75 percent of independent contractors to valid industry codes of the employing firms, so
the sum of IC across industries does not equal the total number of IC in the population. Industries are categorized by 2digit NAICS codes.

15

“Public Administration”, and “Wholesale Trade,” where the percentage growth in ICs has been
relatively small. 22
Much of the growth in ICs by industry corresponds with industries that have also been
growing rapidly in terms of employment. According to the Bureau of Labor Statistics (BLS), the largest
growing major industries from 2001 to 2016 in terms of employment were “Education and Health
Care Services” and “Professional and Business Services.” 23 The former includes the industry with the
third largest absolute growth in ICs (health care and social assistance) and the latter includes the
industry with the largest absolute growth (professional, scientific and technical services) and an
industry which doubled in ICs (administrative and support and waste management). In contrast the
industry with the second largest growth in ICs, “other services,” has had almost no growth in total
employment so the growth in ICs may represent restructuring of work in this group of service
industries which includes repair and maintenance, personal and laundry services, private households,
and religious, grant making, civic and professional organizations. It would be useful to analyze
occupations in addition to industry, for example, to know whether the ICs in the professional,
scientific and technical services are the scientists or administrative assistants. Unfortunately,
occupation information is not available in our data, so the survey data discussed in Section 3 are a
better source for analyzing occupational changes among contractors.
5.2 Demographic and Economic Characteristics of ICs
Policymakers’ interest in the rise in IC work often stems from a concern about lower levels of
job protection and benefits. In this section, we explore the demographic and economic characteristics
of ICs and how they have changed between 2001 and 2016. Our results suggest that ICs have varied
experiences and backgrounds, but that the fastest growing groups are relatively more economically
vulnerable populations. First, we find that long-run growth has been fastest among ICs who receive
the majority of their labor income from contracting, while there has been a more recent increase
among those for whom IC earnings are a secondary labor income source. Second, within the broader
distribution of taxpayers, IC earners are disproportionately likely to occupy the top and bottom
quartiles. In levels, the plurality in each tax year are located in the top quartile, but the number of IC
earners in the bottom quartile grew the fastest, implying that an increasing share of the contractor
workforce are in relatively low-income households. Finally, we show a large increase in the growth of
female contractors relative to female employees or male contractors. The relative growth was
concentrated among women whose primary source of labor income is IC earnings, who are their
households’ primary earner, and who are in the bottom of the income distribution. These patterns
suggest that the long-run increases in IC labor provide important sources of household income and
that many of these workers do not have employee relationships with worker benefits and protections
outside of their contracting relationships.

22 Appendix Figure A2. shows the percent change in IC from 2001 to 2016 by industry.

23 BLS statistics are from their website, https://www.bls.gov/charts/employment-situation/employment-levels-by-

industry.htm

16

5.2.1. IC income earners within the U.S. income distribution
Within the broader distribution of taxpayers, IC earners are disproportionately likely to occupy
the top quartile of the income distribution based on adjusted gross income (AGI), as shown in Figure
7. Throughout our sample period, all AGI quartiles saw an increase in IC earners, but the fastest
growth occurred in the bottom half of the income distribution, implying that an increasing share of
the contractor workforce comes from relatively low-income households. Reassuringly, the results are
very similar if we use our broader definition of ICs, which does not include the deductions restriction.
This alleviates concern that this restriction could introduce differential selection that eliminates high
AGI ICs (Figure 7, Panel B).
Next we compare IC and wage earners to understand the relative economic position of IC
earners. The mean AGI and taxable income are higher for ICs than employees in both 2001 and 2016,
but the medians are lower (Table 3). Further, the median AGI of ICs has fallen by over $5,000 in real
terms between 2001 and 2016. Although the median has decreased, mean IC income has increased
for ICs, which could suggest an increase in inequality of earnings among ICs but appears to be
reflective of a structural shift in labor markets as discussed later in this paper. The mean wage and
salary earnings for ICs were almost as high as for employees in 2001, but the median was much lower. 24
In 2016, wage earnings of ICs fell—both in real terms, and relative to employees—commensurate
with the increasing share of ICs with predominately IC income.
5.2.2 IC income relative to individuals’ and households’ other income sources
To investigate the importance of IC income to individuals, we divide ICs into those for whom
IC earnings are the primary income source (IC income exceeds 75% of labor income) and those for
whom IC earnings are a supplemental (IC income is less than 25% of labor income). We define the
share of IC labor income as: total 1099 earnings / (total 1099 earnings + W-2 earnings). 25 Panel A in Figure
8 shows that these are the largest groups of ICs and it is relatively rare for ICs to receive around half
of their income from each source. The number of ICs in these categories has grown rapidly since
2001, with each growing by almost 50% (Figure 8, Panel B). There has been a steady long-run increase
in workers with IC earnings as a primary income source, and the number of workers with IC earnings
as supplemental labor income has grown particularly fast after 2011, coinciding with the growth of
the platform economy. The rapid growth in the two distinct groups of ICs, workers who supplement
existing wage income and workers who rely primarily on IC income, suggests that the aggregate trend
24 Mean wage and salary earnings are calculated including zeros. I.e. those with no wage and salary income are coded as

having zero Form W-2 earnings.

25 Form 1099-MISC/K income and Form W-2 income are not, in general, directly comparable income concepts. Form

1099-MISC and 1099-K represent gross income paid to a contractor, but do not necessarily represent net income or
profits, as net income can be lower than gross receipts after accounting for various applicable deductions. Income
represented on a Form W-2 has already been adjusted for some pre-tax deductions and thus more closely approximates a
net income concept, although the employee may still claim additional deductions. Because our preferred definition of
independent contractors only included those with less than $10K in deductions, this distinction makes very little difference
for our results. We have reproduced the results using a conservative net income concept, Form 1099-MISC/K income
minus total Schedule C deductions, and the results are almost identical.

17

masks underlying heterogeneity in the type of work and workers involved in these contract
relationships. The results also likely imply that the overall rise in ICs has differential implications for
the stability of the labor relationships and access to worker protections across these groups.
Table 4 shows that, in general, ICs have relatively low median Form 1099-MISC/K earnings
at the individual level, and perhaps surprisingly the levels are similar across AGI quartiles. IC income
represents a much greater share of household income for lower-income households relative to those
in the top of the AGI distribution (Cols 1 and 4), and this is particularly true for those with IC earnings
as a primary labor income source (Col 5). While some of the low-income households with primarily
IC earnings may be retirees, this is unlikely to be the whole story, as the average age in the lowest
quartile is 43 years. Earners in the bottom quartile are also less likely to claim dependents, and less
likely to be married. While this difference is partly a mechanical feature of measuring AGI at the
household level, it nonetheless highlights that ICs at the bottom of the income distribution are more
likely to be using IC earnings as their primary income source, while for ICs at the top of the income
distribution, IC earnings are more likely to be supplemental – either supplemental income earned by
the primary earner or primary earnings for a secondary earner. 26
5.2.3 Independent Contractor Growth for Primary and Secondary Earners
Next, we explore the growth in IC participation among primary and secondary earners within
a household. We divide ICs into four categories: i) primary earners with the majority of their income
from IC earnings, ii) primary earners with the majority income from Form W-2, iii) secondary earners
with the majority of their income from IC earnings, and iv) secondary earners with the majority of
their income from Form W-2 earnings. “Primary earners” are individuals whose labor income
contributes more than half of total household labor income. “Secondary earners” are married
individuals whose labor income contributes less than half of total household labor income. 27 “IC
Primary” are ICs with more than half of their individual labor income as IC income, and “IC
Secondary” are ICs with a majority of their individual labor income from Form W-2s.
The largest group of ICs is primary earners for whom IC income is a secondary income source,
at approximately 45 percent in 2016 (Figure 9 Panel A). The second largest category is primary
household earners for whom IC income is their primary labor income source comprising almost 33
percent of ICs in 2016. Secondary earners for whom IC income is their primary income source makeup 14 percent of ICs, and secondary earners with secondary IC income make-up the remaining 8
percent. We find that the fastest growth has been among primary earners with primarily IC earnings
(Figure 9, Panel B). The slowest growth has actually been among workers with IC as a secondary
income source, though there has been fast growth amongst this group since 2013 perhaps attributable
26 Categorizing workers by whether the majority of their labor income is IC income shows the demographic distinctions

even more sharply. These workers are more likely to be in the bottom half of the income distribution, and in the bottom
half of the income distribution almost all of household income comes from Form 1099-MISC/K income (Table 4).
27 An IC is a “primary earner” in their household if the sum of their individual 1099 and W-2 income is greater than half
of the sum of total household 1099 plus W-2 income, or (1099 incomei + W-2 incomei)/(1099 incomehh + W-2 incomehh)>0.50
where i indexes the IC and hh indexes their household. For non-married ICs, this will always be true and for married ICs
1099 incomehh = 1099 incomei + 1099 incomes and W-2 incomehh = W-2 incomei + W-2 incomes where s indexes the IC’s spouse.

18

to the rise of the platform economy. The results suggest that the long-run growth in IC labor cannot
simply be attributed to the rise of the platform economy providing new opportunities for individuals
and households to pick-up supplementary income. Over a long horizon, for an increasing number of
individuals, IC labor is their primary income source even though the median worker has relatively low
IC earnings.
Table 5 shows summary statistics for each of these groups. The fastest growing group primary household earners with IC income as their primary labor income source (shown in the second
set of columns) – has substantially larger IC income than the each of the other groups. The median
IC earnings for this group is over $15,000 gross (or $9,500 net of Schedule C deductions), while for
the largest group of ICs, primary earners whose IC income is a secondary income source (shown in
the first set of columns), the median IC earnings is $2,500 gross (or $1,300 net of Schedule C
deductions). At the same time, primary earners with primarily IC labor income are much more likely
to be in the bottom quartile of the AGI distribution, with almost 50 percent of this group in the
bottom quartile and only 17 percent in the top quartile. In contrast, 35 percent of primary earners with
supplemental IC labor earnings are in the top quartile of the income distribution, and only 16 percent
are in the bottom quartile.
The results suggest that policy concerns associated with the “1099 economy” are likely to be
varied. The largest share of ICs receive the majority of their income from wage and salary employment
and are in the top quartile of the income distribution, which mitigates concerns about a loss in labor
protections and fringe benefits. On the other hand, the fastest growth in ICs has been among those
who receive their primary income from contracting. For these individuals the loss in protections and
benefits are more concerning, particularly since the majority of these individuals find themselves in
the bottom quartile of the income distribution.
5.2.4 IC Trend and Characteristic Differences by Sex
We find that growth in aggregate IC labor was disproportionately driven by an increase in
female ICs, and that the aggregate trend cannot be easily characterized by the rise of a single type of
IC relationship. As with ICs in general, the majority of female ICs have contractor income that is
supplemental to Form W-2 earnings, but the largest growth in female ICs was among those whose
primary source of labor income is IC earnings. For the majority of the latter group IC income
represents the primary household earnings source, and these households are disproportionately in the
bottom quartile of the income distribution. This section also explores historically important factors
that determine female self-employment and labor supply more generally including the presence of
children, eligibility of the EITC, marital status, and other household income to shed light on the
potential mechanisms behind the increase in female ICs. We find that, demographically, female ICs
tend to be similar to female employees, but are lower in the income distribution, and that the largest
increases have been in service industries.
Approximately 55 percent of the growth in independent contracting from 2001 to 2016 is
attributable to the increase in female ICs. Women saw a 68 percent increase in the number of ICs
while men saw a 37 percent increase (Figure 10, Panel B). The share of the female workers (ICs plus
19

employees) who are ICs increased from 5.4 percent to 7.5 percent from 2001 to 2016, while the IC
share of the male workforce remained essentially constant over this period. Figure 10, Panel C shows
that the increase in female ICs is similar under the broader definition suggesting that the rise in female
Form 1099-MISC/K recipients does not represent an increase in female owned businesses but is
driven by women providing contract labor services. Additionally we find that over half of the increase
in female ICs is among women who receive the majority of their labor income from IC earnings
(Panels A and C). Though female ICs who earned the majority of their labor income from IC earnings
were only 17 percent of ICs in 2001, this group contributed almost 30 percent of the total growth in
ICs, male and female, from 2001 to 2016 (Appendix Table A.2).
When we compare the demographic characteristics of female ICs to those of male ICs and
female employees, we find that most demographic trends are relatively flat over time and track
relatively evenly between men and women, providing no clear evidence of compositional changes
among female ICs over time along these dimensions. Female ICs are more likely to have children than
employees, but this gap does not grow over time, suggesting that movements to contracting resulting
from a desire for flexible work among mothers is not likely a major contributor to the rise in female
ICs (Figure 11, Panel B). 28 Female ICs are older on average than employees, and the share of women
55 and older participating in the labor force increased from 12.6 to 22.2 percent from 2000 to 2016
(U.S. Bureau of Labor Statistics (2017)). It could be that independent contracting provides a preretirement option for women and that the increasing share of women ICs is related to the increasing
working age for women in the labor market in general. This trend could also be more broadly related
to the finding that, on average, male and female IC are getting older over this period, as seen in Panel
A.
We find that the fraction of female ICs receiving the EITC is growing faster than female
employees. This is consistent with previous research finding that increasing EITC generosity increases
self-employment (LaLumia (2009), Lim and Michelmore (2018)) and that the self-employed may have
a greater ability to target their income to receive a higher EITC (Saez (2010), Chetty et al. (2013),
Mortenson and Whitten (2018)). Further research is needed to understand the extent to which changes
to EITC policies have contributed to the increase in female ICs over this time period.
Next, we examine whether certain industries were responsible for the growth in female ICs.
Figure 12 shows that female and male ICs are concentrated differently across industries and that the
growth in ICs from 2001 to 2016 was differential across industries for men and women. Panel A
shows that there have been large increases in female ICs in the fast growing professional services and
health care industries, industries that started with high levels of female ICs in 2001. Panel B shows
that there is heterogeneity in the increase in female and male ICs across industries. The professional
and other service industries saw substantial growth for both men and women, though with larger
absolute increases for women. Yet, in the fastest growing sector over this period, health care, social
assistance and educational services, the aggregate growth in ICs has been dominated by increases in
female ICs. Also, the increases ICs in retail and trade has been dominated by women, even as total
employment in this sector has been almost flat over this period (BLS). These patterns suggest that
28 We could see an increase in female ICs if the flexibility of contracting work increased relative to wage and salary

employment or if there was a change in preferences among women resulting in a greater desire for flexibility.

20

aggregate service industry growth likely contributed to the increases in IC labor, but that there remains
heterogeneity across industry and between men and women, which is potentially related to differences
in occupations within industry.
Next, we locate female ICs in the income distribution and investigate the contribution of their
IC earnings to total individual and household income, following the analyses described in Section 5.2.1
and Section 5.2.2. As with ICs in general, female ICs have relatively constant median IC earnings
throughout the distribution (Table 6). Even though the median level of net IC earnings is low, around
$2,000, almost 50 percent of female ICs receive the majority of their labor income from contracting,
as do 47 percent of male ICs. Approximately two-thirds of female ICs in the bottom quartile of the
income distribution are primarily IC earners, and IC earnings make-up essentially all of AGI for this
group.
We find that the largest percentage increase in female ICs is among those who receive a
majority of their labor income from IC labor, across all quartiles (“primarily IC income”, Figure 13,
Panel B). Relative to male ICs, there has been more growth across all AGI quartiles for female ICs
and the growth has been more clearly concentrated in the bottom half of the income distribution. This
is particularly true amongst those who receive a majority of their labor earnings from IC income. The
absolute rise in female contracting at both ends of the household AGI distribution, displayed in Panel
A, suggests multiple changes over this time period encouraging women to become ICs.
To investigate differences between male and female ICs by the contribution of IC earnings to
individual and household income, we divide male and female ICs into the same four categories as in
Section 5.2.3: i) primary household earners, primarily IC income, ii) primary household earners,
secondary IC income, iii) secondary household earners, primarily IC income, and iv) secondary
household earners, secondary IC income. Overall we see that the percentage growth of ICs for women
has greatly outstripped that for men in every category, except for secondary household earners for
whom IC income is a secondary income source (Figure 9). Table 7 shows that the two largest
contributors to the total growth in ICs are female primary earners for whom IC income is a secondary
income source, contributing 22 percent, and female primary earners with primarily IC earnings,
contributing 19 percent. These groups also have the largest relative growth; women who are primary
earners in their household and have primarily IC earnings increased by over 90 percent since 2001
(Figure 9). The next largest contributors to the overall trend were men of these same types,
contributing 19 percent (primary earner, secondary IC) and 16 percent (primary earner, primarily IC)
respectively. We do not find evidence that the rise in female ICs is driven by secondary earners, who
perhaps would not otherwise work but are finding new opportunities to enter the labor market
through contracting.
Table 5 shows that female primary earners with primarily IC labor earnings, the group with
the largest relative growth between 2001 and 2016, are also the most likely to be in the bottom quartile
of the AGI distribution, with 51 percent in the bottom quartile and only 15 percent in the top quartile
(second set of columns). Female primary earners with IC earnings as a secondary labor income source,
which make-up the majority of female ICs and have experienced particularly rapid growth since 2012,
are much more equally represented across the income distribution, though they more likely to be in
the bottom half of the distribution than are male ICs of this type (first set of columns).
21

Taken together, we find that the largest share of ICs uses contracting income as a
supplementary labor income source, and that this group has grown fastest since 2012, which is
consistent with previous work (e.g. Collins et al. (2019)). But, we also show that to characterize the
long-run growth in IC labor in the U.S. as individuals seeking supplemental income, may miss a more
structural shift in the labor market, particularly for women. Importantly, the fastest growth has
occurred among those for whom IC income is their primary individual and household income source,
and the absolute growth in IC labor is dominated by an increase in female ICs.

6. Worker Transitions into and out of Independent Contracting
In this section, we explore transitions by individuals between primarily earning IC income
and primarily earning wage income over time. We find that there is growth in the number of
workers who consistently receive IC income across years, and in the number of individuals moving
from no labor earnings to IC income. Consistent with our other findings in this section, a majority
of the growth in IC income is among those transitioning into states where all labor income is from
IC labor.
Using our sample of all taxpayers described in Section 4.3, we define a transition based on the
status of that individual in year t-1 and year t. In Figure 14, we show the number of individuals making
transitions between employment statuses each year: “IC any to IC any” are those who had some IC
labor income in year t-1 and some IC labor income in year t; “W-2 to IC only” are those with Form
W-2 income in year t-1 and IC income but no W-2 income in year t; “IC only to IC only” are those
with only IC labor income and no W-2 earnings in both years; and “None” represents having no W2 or IC income in a year.
Our results suggest that the majority of the growth in IC income is attributable to more
individuals entering the labor force as ICs and to an increase in the number of individuals who
continue IC work year to year. Both of these patterns result in IC income being the primary labor
income source for the worker suggesting that the rise in IC income does not simply reflect increases
in the number of people picking-up supplemental labor income as ICs. As Figure 14 shows, there has
been a large increase in the number of people who have some IC income year over year, “IC any to
IC any.” Approximately half of the increase in this group can be attributed to those who that have
only Form 1099MISC/K labor income from year to year, “IC only to IC only.” We also find that,
following the Great Recession, there has been an increase in those transitioning from having no
employment income to having IC earnings. The majority of the increase is among those who transition
from having no earnings to having only IC earnings and no W-2 earnings, “No to IC only.”
Table 8 presents summary statistics for individuals changing worker statuses in 2014. We
also use our sample to identify trends in the incidence and characteristics of transitions between
contractor and employee income received from the same firm. We find an increase in the number of
individuals switching from IC to W-2 labor within their firm, up by 36 percent from 2002, while
there has been a slight decrease (3 percent) in those switching from W-2 to IC workers within their
firm.
22

7. Trends in Firm use of Independent Contractors Labor
In Section 5, we presented trends in individuals’ receipt of IC income, changes economists
typically associate with “supply side” market forces. In contrast, the focus of this section and the next
is on trends broadly (if simplistically) considered relevant to the “demand side.” This section explores
how firm use of IC labor, and the characteristics of the firms that use IC labor, have changed over
time. We define and track measures of both extensive and intensive use of ICs, and find that, while,
in levels, high wage and large firms are more likely to use ICs, and use them more intensively,
throughout the series, growth during this period was driven by small and low-wage firms.
7.1 Cross-Sectional Time Trends
We define two broad measures of IC usage by firms. We define extensive margin IC usage
using an indicator for whether a firm hired at least one IC in a given tax year. It is likely that hiring at
least one worker as an IC represents a significant fixed cost to the firm relative to the marginal cost of
hiring an additional IC, as IC contracting is governed by different tax and labor regulations that require
some consultation, or at least active decision by the firm. Extensive margin usage provides information
on the prevalence of IC labor usage.
We define intensive margin usage as a continuous ratio which captures the firm’s comparative
reliance on ICs relative to employees in a given tax year. We define the “Worker Ratio” as the ratio of
ICs to the total number of workers (ICs and employees). The “Compensation Ratio” is calculated as
the ratio of IC compensation (from Form 1099-MISCs and Form 1099-K) to aggregate worker
compensation (the sum of compensation to ICs and employees). 29 While conceptually similar, the
“Worker Ratio” is informative about the composition of a firm’s workforce, while the “Compensation
Ratio” can be interpreted as the relative allocation of a firm’s labor expenses.
7.1.1

Extensive Margin Usage of Independent Contractors

Extensive margin usage increased for nearly every type of firm, but subtle differences in trends
and levels merit discussion. Figure 15, Panel A plots extensive margin IC usage for two categories of
ICs, those fitting our preferred definition and for a broad definition which includes all individual Form
1099-MISC/K recipients that can be matched with a Form 1040. The trends presented in this figure
yield two insights. First, the trend in IC usage using the broad definition tracks the trend in our
preferred IC definition closely, suggesting that the overall observed increase in extensive use is driven
by hiring of ICs providing labor services, rather than an increase in contracting relationships with
other firms. Second, with our preferred definition, we find that there has been a large increase in the
fraction of firms with at least one IC; the fraction has increased by 5 percentage points, or by 20

29 The compensation from the Form 1099-MISC/Ks represents gross earnings, while Form W-2 wages can more closely
approximate net income; however, this distinction is much less important for our preferred IC definition because those
individuals have low levels of deductions by construction.

23

percent, from 2001 to 2015. The overall fraction of firms using contractors peaked in 2012 and has
been relatively steady since then.
Figure 16, Panel A shows the change in firms’ extensive margin IC usage by quartiles of the
firm’s median wage using our preferred IC definition. 30 Extensive margin IC usage rose consistently
from 2001 to 2012 for firms in all quartiles. By 2015, extensive margin IC usage rose by almost 20
percent for firms in the top three wage quartiles, and by 28 percent for firms with lowest median
wages, relative to 2001 levels. Figure 17, Panel A shows that the increase in extensive margin IC usage
was also shared across firms of different sizes, though relative growth was highest in small firms with
four or fewer employees. Extensive margin use is higher throughout the series for large firms, i.e.
firms with more than one hundred employees. Appendix Figure A3, Panel A plots extensive use by
industry. Most industries exhibit a modest upward trend, excepting manufacturing services, which
increases sharply after 2010. 31
7.1.2 Intensive Independent Contractor Usage Trends
Figure 15, Panel B plots the average worker ratio, defined as the number of ICs divided by the
number of workers (ICs plus employees). As with the extensive margin, it appears that the increase in
the share of contractors per firm is most pronounced when contractors are defined using our preferred
definition, ICs likely to be supplying labor services. In contrast, for our preferred definition, the
average compensation-ratio, defined as total Form 1099-MISC and 1099-K compensation issued by
the firm to ICs divided by total worker compensation (IC Form 1099-MISC/K compensation plus
employee wage and salary compensation), is flat over this period (Figure 15 Panel C). This trend is
notable given that Form 1099-MISC/K compensation is a gross of business expense deductions, and
so might be expected to grow faster than the worker-ratio. The results imply that ICs are lower paid
than employees either because they work fewer hours or because they have a lower hourly
compensation rate.
The comparatively rapid growth in extensive margin IC usage by small and low wage firms
resonates with the trends in intensive margin use. Larger and higher wage firms have higher levels of
IC use, smaller and lower wage firms grew rapidly in the intensity with which they used ICs. Figures 16
and 17, Panels B and C show analogous information but for intensive margin usage. Although higherpaying firms use contractors more intensively throughout the series, low wage firm intensive margin
use, as measured by worker ratio, grew more rapidly. For firms with median wages in the highest
quartile, the worker-ratio grew by approximately 15 percent from 2001 to 2015 compared to firms in
the lowest quartile, which grew by 22 percent. The compensation ratio, in contrast, grew only 1 percent
for high wage firms and 3.4 percent for low wage firms. This could be consistent with several
scenarios: firms could be hiring a larger number of ICs but using each contractor to perform less work;
firms could be shifting employees from part-time or part year to full time; or firms could be increasing
30 Firms are assigned to quartiles based on their median employee compensation within the tax year. Similar results obtain

when assigning firms to quartiles based on their average, 25th percentile, 75th percentile, or 90th percentile employee
compensation.
31 Data limitations caution against inferring too much here. NAICS codes are missing for approximately one-third of the
firm-level sample, particularly among smaller firms, which are growing quickly in IC use.

24

hiring of relatively low skilled IC. Likewise, small firms with fewer than twenty employees saw their
intensive IC usage (worker-ratio) grow by over 10 percent, on average, between 2001 and 2015, about
twice as fast as medium firms (20-100 employees) and a third faster than very large firms (more than
100 employees). The share of labor compensation fell slightly for all but the smallest firms (i.e. those
with fewer than 4 employees.) Overall, the worker ratio was upward trending for most industries, but
more noticeably so for service sectors (Appendix Figure A3, Panel B).
In short, virtually all types of firms were more likely to use ICs, and to use more of them, in
2015 relative to 2001. However, while high wage and large firms use more ICs on both the intensive
and extensive margin throughout the series, the increase for these firms was small during this period
relative to the growth in small and low wage firms. This differential in growth is particularly
pronounced on the extensive margin, where IC use among firms in the bottom median wage quartile
grew at nearly twice the rate as those in the top quartile, and firms with fewer than 4 employees
increased 22 percent relative to an approximately 10 percent increase among firms with more than
100 employees.
7.2 Trends in IC compensation within firm’s wage distribution
The question of where independent contractors fall within the distribution of their payer’s
wage distribution is complicated by the nature of compensation data available in annual tax filings: we
do not observe an hourly wage, but rather, the annual aggregate compensation issued to each worker
(Form W-2 for employee; Form 1099-MISC/K for IC). The worker could have provided services part
year, or part time, and, complicating distributional questions, could be receiving compensation from
several other firms. In addition, compensation issued to ICs is, at least in theory, “gross” i.e. inclusive
of costs that the IC will deduct as business expenses. In contrast, employees are allowed very few
business deductions on the theory that essential expenses will be reimbursed by the employer.
Form-level compensation refers to the amount of cash wages issued to an employee on Form
W-2, or the amount of gross compensation issued to a contractor on Form 1099-MISC. Although
Form 1099-MISC compensation is gross, rather than net, most contractors that meet our preferred
definition do not have substantial business related deductions, rendering their form compensation
closer to a net measure than the average Form 1099-MISC recipient.
We construct a comparative statistic of within firm compensation using the same two percent
cross-sectional sample. Using all Form W-2s associated with a given EIN in a given tax year, we
calculate five statistics for the wage distribution for each firm: the minimum, 25th percentile, 50th
percentile, 75th percentile and maximum. We then assign each contractor to a sextile of the wage
distribution, based on their Form 1099-MISC compensation. Figure 18 plots the average share of
contractors that fall into each of these six bins over time. While the number of contractors vary year
to year, the largest share come from the bottom quartile of the wage distribution and below. This is
potentially consistent with the individual finding that the fastest growing group of workers is located
in the lowest AGI quartile/total labor compensation distribution, where many workers are lowerskilled.

25

8. Trends in Transitions between IC Income and Wages within Firms
In this section, we explore the dynamics of a given firm-worker relationship over time. As
described in Section 4.2, we expand the firm-level sample used in Section 4 to include longitudinal
information about workers, which allows us to identify transitions between contractor and employee
income received from the same firm. We identify two types of transitions within a firm: a contractor
transition is a transition from earning primarily wages in one year to earning primarily IC income in
the following year; the opposite transition, from primarily IC to employee income, is referred to as
an employee transition.
A rise in the fraction of firms with contractor transitions may indicate that the increase in IC
income we found in Section 5 arises from firms potentially reclassifying employment income as IC
income, perhaps to avoid taxes or other regulatory costs. However, in the aggregate, we see modest
evidence of the opposite trend: an increase in firms with employee transitions. We also find that
firms are more likely to hire new workers as ICs rather than employees, which suggests that the rise
in overall IC income is driven by new hires rather than contractor transitions.
8.1. Trends in Transitions: Main Results
As in section 7, we report both extensive and intensive margin results for worker transitions
within the firm. On the extensive margin, we find that the fraction of firms with at least one
employee transition is consistently higher than the fraction of firms with at least one contractor
transition. Figure 19 plots these fractions separately for our preferred definition of IC and for the
broader set of all Form 1099-MISC/K recipients who file a Form 1040. The trends are largely
similar for these two groups, but the more expansive definition suggests a higher level of
transitions. 32 Interestingly, the two types of transitions appear to be negatively correlated over our
sample period. Employee transitions increased relatively sharply in the post-Great Recession years
while contractor transitions saw a slight decline. Also notable, at the firm-level, transitions appear to
go in one direction: Panel B shows that only 0.3 percent of firms have both types of transitions in
the same tax-year.
Employee transitions are also higher than contractor transitions on the intensive margin.
Figure 20 shows the average fraction of workers making employee (contractor) transitions relative to
the total number of workers at a firm. Echoing the extensive margin trends in Figure 19, a greater
proportion of each firm’s labor force completes an employee transition than a contractor transition,
though both fractions are modest in absolute terms. Using our preferred definition, the average firm
has seen an increase in employee transitions over the period, from 0.35 to 0.43 percent of workers. In
contrast, employee transitions have held steady at approximately 0.17 percent.
Neither figure suggests that there has been an increase in existing employees being reclassified
as ICs by their employers. Indeed, there appears to be a modest increase in ICs becoming employees;

32 Our sample is a representative cross-section of EINS that issue W-2s, so we show statistics on transitions for workers

in the year that the firm with which the worker is associated is sampled.

26

a trend driven primarily by an increase in permanent transitions. 33 One potential explanation for this
trend is that IC relationships constitute a comparatively low-commitment way for a firm to learn about
a worker’s productivity, after which the firm can confer employee status, in order to retain the most
productive contractors. Additional work on IC income receipt over the life cycle and whether it leads
to employee status and future earnings increases could shed light on how this practice affects workers’
earnings trajectories.
Rather than a shift in the status of existing workers, we find that the upward trends in firm IC
use identified in Section 7 are better explained by an increasing propensity to hire new workers as
ICs. 34 We define new hires as individuals who are either an employee or IC (using our preferred
definition) in the year in which the firm was sampled, but who had no relationship with the firm in
the year prior. Figure 21 plots new hires by type, for both extensive and intensive measures. Panel A
shows that the fraction of firms that hired at least one new employee was between 55-65 percent over
the study period; it dropped precipitously during the Great Recession, then recovered, though not to
pre-2008 levels. In marked contrast, the fraction of firms hiring a new IC rose slightly between 2001
and 2016, with a particularly strong increase in the post-recession years. In 2016, 18 percent of firms
hired a new IC.
In Panel B, we show the average fraction of a firms’ total workers who are new employees or
new ICs. The patterns mirror those in Panel A. While the average fraction of workers who are new
ICs has held steady, at around 4 percent, the average fraction of new employees has fallen from 31
percent (in 2001) to 26 percent (in 2016), exhibiting a sharp decline in 2009. Initial classification of
new hires as ICs is also consistent with firms learning about worker productivity before committing
to employee status, as articulated above; however, an alternative explanation is that individuals looking
for new employment are hired as ICs and are overall less likely to enter formal employer-employee
relationships. Further work to distinguish between these explanations would provide information to
help understand the importance of addressing concerns about the lack of social safety net programs
for ICs.
8.2. Trends in Transitions by Firm Size and Industry
Next, we examine transitions by firm size and industry. We uncover substantial heterogeneity,
finding that overall trends are disproportionately driven by small firms, and transitions are
concentrated in industries with previously documented high levels of worker misclassification. Our
results may help identify characteristics of firms with a higher propensity to misclassify workers and
could inform the design of future, targeted protections for contract workers (Harris and Krueger
(2015)).
33 We define “permanent” transitions as those that endure for at least two consecutive years. For example, a permanent

contractor transition is defined as three consecutive tax years in which the worker is an employee in the first year,
becomes a contractor in the second year, and remains a contractor in the third year. In contrast, we consider the
transition “temporary” if the worker reverts to the original relationship or leaves the firm in the third year.
34 Initial classification of new hires as IC is also consistent with firms learning about worker productivity before committing
to employee status, as articulated above.

27

Figure 22 plots employee and contractor transitions by firm-size, binned by the number of
Form W-2 recipients. 35 Panels A and C show the extensive measure of transitions of either type rises
strongly with the size of the firm; however, this could be a largely mechanical effect—larger firms with
more workers can be expected to be more likely to have at least one worker transition. Indeed, when
normalized to 2001 levels in Panel B, growth in extensive margin contractor transitions does not
appear to systematically differ by firm size. The same cannot be said for extensive margin growth in
employee transitions. Panel D shows that the fraction of firms with at least one employee transition
is increasing most in percentage terms for the smallest firms. Panels B and D of Figure 23 tell a largely
similar story for new hires. The fraction of firms hiring a new IC increased while the fraction of firms
hiring a new employee decreased, virtually across the board. Similar to our findings in Section 7, the
trends are most pronounced among small firms.
Given that levels of both worker misclassification and self-employment--of which
independent contracting is a substantial part-- differ substantially by industry (Pew 2015), we anticipate
significant heterogeneity along this dimension. We link sampled EINs to NAICS industry codes at
the two-digit level. 36 As an important caveat, we are unable to identify industry for approximately one
quarter of firms, which may substantially affect our results. 37 Table 9 shows the fraction of firms with
any workers switching statuses in 2016 and the fraction of firms making any new hires of each type of
worker in 2016 by industry. Firms in the information and education sectors are the most likely to have
employee transitions and new hire ICs, while firms in accommodations or food services are the least
likely. This variation could simply reflect higher use of contractors in those industries. However, the
incidence of transitions by industry is closely aligned with the conclusions of worker classification
audit studies, which have found that information and education sectors had relatively high rates of
misclassification compared to leisure and hospitality industries, which had some of the lowest (Carré
and Wilson (2004)).

9. Conclusion
This paper uses administrative tax data to identify a group of workers as "independent
contractors"-- individuals who provide labor services to firms outside of an employment relationship
– and to investigate trends in IC labor usage and characteristics of IC workers over a fifteen-year
period, from 2001 to 2016. We begin by developing our preferred definition of ICs using individual
tax returns and information reports. We focus on individual Form 1099-MISC or Form 1099-K
recipients with less than $10,000 in Schedule C deductions, excluding deductions for vehicle and travel,
who are not employers. We show that the deductions restriction is important for accurately
establishing the size of the IC workforce, by helping to exclude small businesses which may supply
goods or services to firms in transactions that are conceptually distinct from ICs. This restriction is
35 We divide firms into size categories based on their number of W-2 employees. This size measure has the benefit of

not changing across different potential definitions of independent contractors, and it more closely matches the definition
of firm size that can be developed using other data sources such as unemployment insurance records.
36 The variable we link to is contained in CDW’s BRTF table, a longitudinal entity-level database maintained by the IRS.
37 We keep the closest year match to the entity file with a limit of 10 years difference. The match rate generally increases
over time from around 68 percent in 2001 to 79 percent in 2016.

28

particularly important for identifying ICs that receive Form 1099-K. Using our preferred definition,
we corroborate previous studies using survey or administrative data sources, finding that the share of
the workforce with some IC income grew substantially between 2001 and 2016, by 1.5 percentage
points, or 22 percent. We find that much of this growth occurred prior to the rise of the major online
platform businesses after 2010.
We find that the growth of the IC workforce is not associated with a single type of worker or
labor relationship, implying that there are likely a number of contributing explanations for why
individuals are increasingly working as ICs. Approximately equal shares of ICs make almost all of their
labor earnings from contracting or have IC earnings as a small supplement to wage and salary
employment. Both of these groups have been growing over time, with almost 50 percent growth from
2001 to 2016. We find steady long-run growth among those who are primarily contractors and more
recent fast growth among those with IC earnings as supplementary labor income, suggesting that the
trend cannot merely be explained by individuals picking up side contracting jobs to supplement their
main earnings.
Additionally, we find that women account for approximately 55 percent of the total increase in
the number of ICs from 2001 to 2016, a period over which female employment has been relatively
flat. We find that the largest group of ICs earn relatively small amounts of IC income as supplemental
labor income, the majority of which are in the top half of the income distribution. Yet, the fastest
growing type of ICs is those whose primary household income source is contractor earnings, over 50
percent of which are in the bottom quartile of the income distribution. These trends are particularly
pronounced among female ICs.
Next we examine firms’ use of ICs and find a 4.9 percentage point, or 17 percent, increase in the
fraction of firms hiring any ICs between 2001 and 2015. By 2015, ICs make up 11 percent of the
workers at the average firm up from 8.9 percent in 2001. We find that virtually all types of firms were
more likely to use ICs, and to use more of them. High wage and large firms with many employees use
more ICs, but the increase in IC usage was fastest for small and low wage firms. We also find that
over our sample period the fraction of firms hiring a new contractor grew by 1 percentage point while
the fraction hiring a new employee has fallen by 5 percentage points. Again, these trends were
strongest for the smallest firms.
To understand whether workers are being re-classified, we examine the extent to which workers
switch employment types with the same firm each year. In 2016, we find that only 4 percent of firms
have at least one worker changing statuses within the same firm. While the number of individuals
switching status within firm are low, we find that firms in the information and educational services
industries are the most likely to have workers switching statuses, corresponding with audit evidence
suggesting that these industries have relatively high rates of worker misclassification (Carré and Wilson
(2004)).
Together, these trends suggest that the long-run growth in IC labor in the U.S. cannot solely be
attributed to individuals seeking supplemental income, or to the rise of a few online platform firms,
but may represent a broad-based, structural shift in the labor market, particularly for women. We do
not determine whether these trends are driven by an increase in demand for these types of workers or
an increase in supply, but our findings suggest that the growth in contracting may be attributable to
29

multiple factors, and that policy concerns associated with the “1099 economy” are also likely to be
varied. Further research to disaggregate the supply and demand side factors and the implications for
individual career paths and business trajectories associated with these trends will certainly be an
important line of inquiry moving forward.

30

References
Abraham, Katharine G., John C. Haltiwanger, Kristin Sandusky, and James R. Spletzer. 2018.
“Measuring the Gig Economy: Current Knowledge and Open Issues.” NBER Working Paper
24950. http://www.nber.org/papers/w24950
Carré, Francoise, and Randall Wilson. 2004. “The Social and Economic Costs of Employee
Misclassification in Construction.” Construction Policy Research Center Working Paper.
Chetty, Raj, John N. Friedman, and Emmanuel Saez. 2013. “Using Differences in Knowledge across
Neighborhoods to Uncover the Impacts of the EITC on Earnings.” American Economic Review.
103(7):2683-2721.
Collins, Brett, Andrew Garin, Emilie Jackson, Dmitri Koustas, and Mark Payne. March 2019. “Is
Gig Work Replacing Traditional Employment? Evidence from Two Decades of Tax Returns.” IRS
Working Paper. https://www.irs.gov/pub/irs-soi/19rpgigworkreplacingtraditionalemployment.pdf
Harris, Seth and Alan Krueger. 2015. “A Proposal for Modernizing Labor Laws for Twenty-FirstCentury Work: The Independent Worker” Discussion Paper 2015-10, The Hamilton Project.
Jackson, Emilie, Adam Looney, and Shanthi Ramnath. 2017. “The Rise of Alternative Work
Arrangements: Evidence and Implications for Tax Filing and Benefit Coverage.” Office of Tax
Analysis Working Paper 114. https://www.treasury.gov/resource-center/tax-policy/taxanalysis/Documents/WP-114.pdf
Katz, Lawrence, and Alan Krueger. 2019. “Understanding Trends in Alternative Work
Arrangements in the United States.” NBER Working Paper 25425.
http://www.nber.org/papers/w25425
Knittel, Matthew, Susan Nelson, Jason DeBacker, John Kitchen, James Pearce, and Richard
Prisinzano. 2011. "Methodology to Identify Small Businesses and Their Owners." Office of Tax
Analysis, Department of the Treasury Technical Paper 4.
LaLumia, Sara. 2009. “The Earned Income Tax Credit and Reported Self-Employment Income.”
National Tax Journal. 62: 191-217.
Lim, Katherine and Katherine Michelmore. 2018. “The EITC and Self-employment Among Married
Mothers.” Labour Eoncomics. 55:98-115.
Mortenson, Jacob, and Andrew Whitten. 2018. “Bunching to Maximize Tax Credits: Evidence from
Kinks in the U.S. Tax Schedule.” SSRN Working paper. http://dx.doi.org/10.2139/ssrn.2719859
Pew Research Center. 2015. “National Trends in Self-Employment and Job Creation.”
https://www.pewsocialtrends.org/2015/10/22/national-trends-in-self-employment-and-jobcreation/

31

Saez, Emmanuel. 2010. “Do Taxpayers Bunch at Kink Points?” American Economic Journal:
Economic Policy. 2(3): 180-212.
Schuetze, Herb. 2000. “Taxes, Economic Conditions and Recent Trends in Male Self-Employment:
a Canada-US comparison.” Labour Economics. 7(5): 507-544.
U.S. Bureau of Labor Statistics. 2017. “Women In The Workforce Before, During, And After The
Great Recession.” https://www.bls.gov/spotlight/2017/women-in-the-workforce-before-duringand-after-the-great-recession/pdf/women-in-the-workforce-before-during-and-after-the-greatrecession.pdf
U.S. Bureau of Labor Statistics. 2018. “Contingent and Alternative Employment Arrangements.”
https://www.bls.gov/news.release/pdf/conemp.pdf

32

Figures and Tables
Figure 1: Form 1099-MISC and Form 1099-K recipients in the U.S.

Notes: The solid lines represent Form 1099-MISC, Box 7 income recipients and the dashed lines show the sum
of all Form 1099-MISC and/or Form 1099-K recipients. The black series represents all unique Form 1099MISC/K recipient TINs. The blue represents “individual Form 1099-MISC/K recipients”, those with an
SSN or an EIN that can be matched to a Schedule C. The red series represents individual Form 1099-MISC/K
recipients which can be matched to a Form 1040 tax return. The green series represent individual Form 1099MISC/K recipients which can be matched to a Schedule C.

33

Figure 2: Independent Contractors: Broad and Deduction Restricted Definitions

Notes: The solid lines represent Form 1099-MISC recipients and the dashed lines show the sum of Form 1099MISC and/or Form 1099-K recipients. The black series is all Form 1099-MISC/K recipients which are
potential ICs: i) individual Form 1099-MISC/K recipients which can be matched to a Form 1040 or a Schedule
C and are non-employers (do not declare any employment deductions) and ii) potential incorporated ICs,
defined as EIN recipients that match with a business income return Form 1120S, have only one owner and are
non-employers. The blue series shows individual Form 1099-MISC/K recipients and excludes potential
incorporated IC. The red series contains the subset of all potential IC which declare less than $10K in total
deductions, excluding car and travel deductions, and the green series includes only the individual ICs with less
than $10K in total deductions. The differences between the black and blue series and between the red and green
series represent the potential incorporated ICs.

34

Figure 3: Sensitivity to Deduction Restrictions
Panel A: Car and Travel as Share of
Total Deductions

Panel B: Distribution of Deductions

Panel C: Time Series, Various Restrictions

Notes: Panel A is a bin scatter plot of car and travel deductions as a share of total household Schedule C
deductions. Each point represents the average share for those within a given range of total deductions. Panel
B shows the frequency distribution of total Schedule C deductions less car and travel deductions for our sample.
Panel C shows the time series of Form 1099 recipients at various deduction levels. The solid lines represent
Form 1099-MISC recipients and the dashed lines show all Form 1099-MISC and/or Form 1099-K recipients.

35

Figure 4: Independent Contractors Relative to Employees over Time
Panel A: Changes in Composition of the Workforce Relative to 2001

Panel B: Independent Contractors as a Share of the Workforce

Notes: This figure shows Form 1099-MISC/K recipients as a share of the total workforce including ICs and
employees. Panel A shows the growth in workers over time relative to 2001 levels. The black series represents
ICs with no Form W-2 earnings; the red series are those with both Form W-2 and IC income; the blue series
are those with only Form W-2 earnings. Panel B shows ICs as the share of the total workforce,
(ICs+employees), for various definitions of ICs. The red series shows the share of the workforce that are
individual Form 1099-MISC/K recipients matched with a Form 1040; the black series shows the share of the
workforce that are ICs by our preferred definition; and the yellow series shows the share of the workforce
that are the subset of ICs (by our preferred definition) that earn the majority of their labor income from
Form 1099-MISC/K income, or with (1099 income / (1099 + W-2 income))>0.5.

36

Figure 5: Independent Contractor and Employee Growth 2001-2016, by State

Notes: This figure shows the percent change in independent contractors (IC) and employees from 2001-2016
by state. The blue bars show the percent change in IC, where IC are defined as Form 1099-MISC/K recipients
that report less than $10K in deductions on a Schedule C, excluding car and travel expenses. The red bars show
the percent change in employees, defined as those who receive a Form W-2 with positive income. The
horizontal blue dashed line represents the average growth rate in IC across states; the red dashed line represents
the average growth rate in employees across states.

37

Figure 6: Number of Independent Contractors by Industry, 2001 and 2016

Notes: This figure shows the number of ICs by industry of the Form 1099-MISC/K issuing firm for the years
2001 and 2016. ICs are defined according to our preferred definition, Form 1099-MISC/K recipients that
report less than $10K in deductions on a Schedule C, excluding car and travel expenses. Industries are
defined as two-digit NAICS categories are reported on the firm’s income tax return.

38

Figure 7: Independent Contractors in the Income Distribution
Panel A: Number of Independent Contractors
by AGI Quartiles

Panel B: Percent Growth in IC by AGI Quartile
and Worker Category (2001 to 2016)

Notes: These figures show ICs by their position in the AGI distribution by year, categorized by their position
in each quartile of the AGI distribution where the distribution is taken over the universe of taxpayers in each
year. Panel A shows the number of contractors in each quartile by year and uses our preferred definition of
ICs, those with less than $10K in Schedule C deductions, excluding car and travel. Panel B shows the percent
change in the number of ICs in each quartile from 2001 to 2016 for various definitions of IC labor. “Main
Definition” corresponds with our preferred definition; “Broad Def.” are all individual Form 1099-MISC/K
recipients matched with a Form 1040, with no deductions restriction; and the last set of bars shows the change
for Form W-2 employees.

39

Figure 8: Dependence on Form 1099-MISC/K Income
Panel A: Form 1099 Income as a Share of Total
Labor Income

Panel B: Form 1099 Income as a Share of
Total Labor Income (trends)

Notes: These figures explore IC income as a share of total labor income and total household income. Panel A
shows. In each panel we define ICs according to our preferred definitions, individual Form 1099-MISC/K
recipients with <$10K in Schedule C deductions excluding car and travel. Panel A shows the number of ICs
by individual IC income as a share of total income, or Form 1099-MISC/K income / (1099 + W-2 income).
The yellow series are those for whom the IC income share is less than 25% of labor income; the black series
are those with a greater than 75% share; the red series are those with a 25-50% share; and the blue series a 5075% share. Panel B shows trends for these groups, relative to the 2001 level.

40

Figure 9: Independent Contractor Growth by Primary or Secondary Income
Panel A: All Independent Contractors (Levels)

Panel B: Independent Contractors (relative to 2001)

Panel C: Female ICs (Levels)

Panel D: Female ICs (relative to 2001)

41

Panel E: Male ICs (Levels)

Panel F: Male ICs (relative to 2001)

Notes: This figure show the time series for independent contractors by whether they are primary or secondary
earners in their households and by whether their primary labor income source is IC income. “Primary earners”
are primary earners in their household defined as having individual labor income (1099 + W-2 income) more
than 50 percent of household labor income, or (1099 incomei + W-2 incomei)/(1099 incomehh + W-2 incomehh)>0.50
where i indexes the IC and hh indexes their household.. “Secondary earners” are married and with labor
earnings less than half of household labor earnings. “IC primary” are ICs that earn the majority of their labor
income from IC earnings, (1099 income / (1099 + W-2 income))>0.50. “IC secondary” are ICs who earn the
majority of their labor income as Form W-2 earnings. All ICs correspond to our preferred definition, individual
Form 1099-MISC/K recipients with less than $10K in Schedule C deductions, excluding car and travel. Panels
A, C, and E show the number of contractors, and Panels B, D and F show the number relative to 2001 to show
relative growth.

42

Figure 10: Male v. Female Independent Contractor Growth
Panel A: Levels

Panel B: Relative Growth

Panel C: Female IC as Share of the
Female Workforce

Panel C: Male IC as Share of the
Male Workforce

Notes: These figures show time trends in IC labor for men and women separately. Panel A shows the level
trends. The solid series represent female workers recipients and the dashed series males. The black series
represents all individual Form 1099-MISC/K recipients; the blue series represents our preferred definition of
IC, Form 1099-MISC/K recipients with less than $10K in Schedule C deductions excluding car and travel; the
red series shows the subset of our primary IC definition that receives the majority of their labor income from
Form 1099-MISC/K income, (1099 income /( 1099 + W-2 income))>0.5. Panel B shows the number of
contractors relative to 2001 to highlight relative the relative increases in each group. It also includes a series (in
gold) for Form W-2 employees as a comparison. Panel C shows the share of the female workforce (ICs plus
employees) that are ICs, for various definitions. The black shows all individual Form 1099-MISC/K recipients
as a share of the female workforce; the blue series shows the share of the female workforce that are ICs by our
preferred definition; the red series shows the subset of our primary IC definition that receives the majority of
their labor income from Form 1099-MISC/K income. Panel D repeats the exercise for male ICs.

43

Figure 11: Demographic Trends in Female IC
Panel A: IC Women v. IC Men Panel

B: IC Women v. Employee Women

Notes: These figures show trends in select demographic characteristics for female ICs. Panel A compares trends
in female ICs to male ICs and Panel B compares trends female ICs to female employees. In each panel the left
axis represents the share of workers with a given characteristic and the right axis represents the average age for
the workers. “Married” is an indicator equal to one for married individuals; “Children” is an indicator for
claiming a dependent child on Form 1040 in that year; “EITC” is an indicator for receiving the earned income
tax credit. In both panels, the solid series are female IC and the dashed series are the comparison group.

44

Figure 12: Changes in Female Independent Contractors by Industry
Panel A: Women IC in 2001 and 2016

Panel B: Change in IC 2001 to 2016, Men v. Women

45

Panel C: Number of IC by Industry in 2016, Men v. Women

Notes: This figure displays female ICs by industry of the Form 1099-MISC/K issuing firm. Industries are
defined as two-digit NAICS categories are reported on the firm’s income tax return. Panel A shows the
number of female ICs by industry in 2001 and 2016. Panel B shows the change in the number of ICs within
an industry from 2001 to 2016 for men and women separately. Panel C shows the number of female and male
ICs in each industry in 2016. ICs are defined according to our preferred definition, Form 1099-MISC/K
recipients that report less than $10K in deductions on a Schedule C, excluding car and travel expense

[Text truncated at 120,000 characters. The full text is on the page linked above.]

---

Source: Frix Law Library, https://www.frixlaw.com/law-library/documents/agency%3Airs%3A2efcb610d31f4c75. Public record. Not legal advice.