U.S. Sen. Mark R. Warner (D-VA) asked the U.S. Secretaries of Treasury, Commerce and Labor to assess existing tax, census, and labor survey data to see if federal officials can generate better information about the size, scope, and growth in the number of Americans participating in the on-demand, or sharing, economy.
U.S. Sen. Mark R. Warner (D-VA) asked the U.S. Secretaries of Treasury, Commerce and Labor to assess existing tax, census, and labor survey data to see if federal officials can generate better information about the size, scope, and growth in the number of Americans participating in the on-demand, or sharing, economy.
U.S. Sen. Mark R. Warner (D-VA) asked the U.S. Secretaries of Treasury, Commerce and Labor to assess existing tax, census, and labor survey data to see if federal officials can generate better information about the size, scope, and growth in the number of Americans participating in the on-demand, or sharing, economy.
fs VW“ | UNITED STATES DEPARTMENT OF COMMERCE
' " | The Secretary of Commerce
ey ‘Washington, D.C. 20280
at
October 15, 2015
‘The Honorable Mark R. Warner
United States Senate
Washington, DC 20510
Dear Senator Warner:
In response to your letter of September 1, 2015, the U.S. Census Bureau (Census Bureau)
agrees that recent advances in technology have changed how workers and employers interact in
the 21st century labor market, Companies such as Uber, TaskRabbit, Airbnb, and others are
often cited as examples of new non-traditional modes of employment and income generation.
How the Census Bureau and other statistical agencies measure these activities has implications
for the accuracy and relevance of our published estimates of employment and earnings, which
affect the quality of any policy analysis that is based upon our published statistics.
The Census Bureau continually strives for the best measurement of this dynamic new
‘economy while recognizing the need to maintain time series continuity and the highest standards
of accuracy and transparency. Our answers to the four specific questions you ask in your letter
highlight how the Census Bureau is actively engaged in new survey efforts and new research
projects intended to increase the accuracy and relevance of how we measure employment and
earnings. Our responses also make clear that certain changes to statute and regulation would
help us provide a better and more unified framework for our measurement of non-traditional
work and income generation,
Enclosed are answers to the four questions you asked in your letter. Please let us know if
anything is unclear or if you would like further detail on any specific point. If you have any
additional questions or concerns, please contact Jim Stowers, Acting Assistant Secretary for
Legislative and Intergovernmental Affairs, at (202) 482-3663.
Sincerely,
Ress
Penny Pritzker
Enclosure1. How does current Census data collection distinguish between workers in traditional
employer-employee structures and those working as independent entrepreneurs?
Census Bureau data can be used to roughly distinguish between workers in a traditional
employer-employee structure and those working as independent entrepreneurs. The current
Census Bureau collections that provide information on this include the Current Population
Survey (CPS), the American Community Survey (ACS), nonemployer statistics, and Survey of
Income and Program Participation (SIP).
As you note in your letter, itis evident from the April 2015 Government Accountability Office
report that definitions matter: “the size of the contingent workforce can range from less than
5 percent to more than a third of the total employed labor force, depending upon widely-varying
definitions of contingent work.”! Many of the key components used to define non-standard work
(such as self-employment, part-time employment, and multiple job holding) are available in the
monthly CPS that the Census Bureau collects for the Bureau of Labor Statistics. Information
about alternative work arrangements, such as independent contractors and on-call workers, was
last collected in the 2005 CPS contingent work supplement.
‘The Census Bureau's ACS identifies the class of worker, which proxies for traditional work
arrangements and independent entrepreneurs. Specifically, the ACS asks working individuals to
identify themselves as working for a private for-profit company, a private non-for-profit
organization, as an employee of either local, state, or Federal Government, as self-employed, or
working without pay in a family business.” One key advantage of the ACS relative to other
household surveys such as the CPS is the large amount of geographical detail available in the
published statistics.
An alternative source of statistics regarding self-employment is the Census Bureau’s
nonemployer statistics. A nonemployer business is one that has no paid employees, has annual
business receipts of $1,000 or more ($1 or more in the construction industry), and is subject to
Federal income taxes. Most nonemployers are self-employed individuals operating
unincorporated businesses (known as sole proprietorships), which may or may not be the
owner's principal source of income. Nonemployer statistics data originate from statistical
information obtained through business income tax records that the Internal Revenue Service
(IRS) provides to the Census Bureau. The nonemployer statistics series is the primary resource
available to study the scope and activities of nonemployers at a detailed geographic level. In
2013, there were over 23 million nonemployer businesses with $1.1 trillion in receipts:
‘The Census Bureau's SIPP is also useful for answering this question. The SIPP collects
information about both traditional and contingent work arrangements, the income associated with
both types of jobs, as well as a host of other information about demographic and job
" This quote is taken from page 3 of “Contingent Workforce: Size, Characteristics, Earnings, and Benefits,” which
is available at hup:/vww.gaonet.govlasses/670/669766.pdl.
£ The ACS qustonnais is avaible at hp rw senss. gorau
surveys/aes/methodology/ f- Question 41 collects information about class of worker.
‘he CPS as collects formation about class of wake
° Further information about nonemployer statistics, as well as historical data from 1997 to 2013, can be found at
bnto:ZAvww.census.cov/econ/nonemplover/index himcharacteristics.’ A time series of the number of traditional and contingent workers, along with
their demographic characteristics, could be created with data from 1996 through the first half of
2013.
2. Within the category of those working as independent entrepreneurs, is Census able to
distinguish between those who make a full-time living in the contingent work-force versus
those who use contingent work to supplement earnings in traditional employment? What
about those working as independent contractors/self-employed versus those accessing
contingent work and contracts through digital platforms?
‘The Census Bureau is able to provide some information about distinguishing between those
whose self-employment is their primary versus secondary source of income primarily through
the ACS and our Longitudinal Employer-Household Dynamics (LEHD) program.
‘The ACS asks respondents about their income in the past 12 months. The first question (#47a)
asks the respondent for “wages, salary, commissions, bonuses, or tips from all jobs.” The second
‘question (#47b) asks the respondent for ‘self-employment income from own nonfarm businesses
‘or farm businesses, including proprietorships and partnerships.” Analyzing the data from these
‘two questions should allow for a tabulation of how many persons supplement their wage and
salary income with just self-employment income, as well as how many persons make a full-time
living with self-employment income.*
‘The Census Bureau’s LEHD program measures employment covered by State Unemployment
Insurance systems, which is a reasonable proxy for traditional employment.° An ongoing
research project at the Census Bureau merges information about this traditional employment with
information about self-employment (specifically, individuals filing IRS Schedule C, which is the
source data for the nonemployer statistics described above). Merging the microdata from these
two administrative data systems will allow us to answer how many individuals work only in
traditional employment, how many work only in self-employment, and how many have a
‘combination of jobs in each sector, as well as the share of income derived from each sector. One
data issue underlying the merge is that the Census Bureau does not always have the Social
Security Number (SSN) of the person with profit or loss from the business; for a small subset of
households filing jointly, we must impute the self-employment income to a specific person
within the marriage,
‘The research project described above could be improved if we were able to narrow the focus by
distinguishing the types of self-employment underlying the Schedule C. Specifically, it is
feasible to link information on individuals filing a 1099 tax form to the universe of traditional
‘employment in the LEHD. We have not yet started such a research project. It needs to be noted
“Further information about the SIPP can be found at hitp:/www.census.gov/sipp.
* The Annual Social and Economic Supplement (ASEC) to the CPS, often referred to as the March supplement, has
‘different interviewing strategy. The CPS ASEC asks respondents to identify their main job last year (question
#46), and asks about the class of worker of this job (#47E1) and about earnings from this job (#48aa). A later
‘question then asks about earnings from all other employers last year (#49b1d). As such, ihe CPS ASEC does not
allow for an analysis of how much earnings is from traditional wage and salary jobs versus self-employment.
Further information about the LEHD program is at htip:/tchd.ces.census.2ov/.that a 1099-LEHD linkage would have to focus on incidence rather than income, since the
Census Bureau currently does not have the income reported on the 1099 tax filings.
This research project, which links the nonemployer records with records from the LEHD data,
has one other exciting aspect that warrants mentioning. Both administrative data sources are a
universe and collected regularly-the nonemployer records are collected annually, and the LEHD
data are collected quarterly (via state-run unemployment insurance programs). The longitudinal
aspects of these data collections enable us to analyze the mix of traditional jobs and self-
employment jobs aver time, which could be used to help policymakers understand whether the
decision to enter self-employment is the result of choice versus necessity due to economic
circumstances.
‘The Census Bureau is not aware of any surveys or other data collection systems in the Federal
Statistical System that provide information on workers using digital platforms to find
employment and income-generation opportunities.
3. What do existing survey and measurement tools at Census tell us about the demographics of
the shifting workforce? How could these tools be augmented to shed light on emerging
trends about how millennials and other demographic groups work?
‘The main sources of Census Bureau information that can tell us about the demographics of the
shifting workforce are the ACS, the Survey of Business Owners, the Annual Survey of
Enluepreneurs, and the LEHD program data,
‘The ACS provides quite a bit of demographic information about the workforce. The answer to
question #2 on the previous page described how the ACS microdata can be analyzed to provide
information about how much total income an individual receives from all jobs versus from self-
employment, Tabulating this by demographic groups will provide information about the
demographic characteristics of individuals who are earning relatively more income from self-
employment than from traditional employment.
‘The Census Bureau publishes demographics of business owners from the Survey of Business
Owners and Self-Employed Persons (SBO).’ The SBO provides the only comprehensive,
regularly collected source of information on selected economic and demographic characteristics
for businesses and business owners by gender, ethnicity, race, and veteran status, Businesses
eligible to be selected to complete the survey include al! nonfarm businesses filing Intemmal
Revenue Service tax forms as individual proprietorships, partnerships, or any type of
corporation, and with receipts of $1,000 or more. The SBO covers both firms with paid
‘employees and firms with no paid employees. The SBO data have been collected every
five years since 1972, for years ending in “2” and “7” as part of the economic census, which
would enable analysis of the demographic trends of business owners.
An exciting new survey effort that will supplement the SBO is the Annual Survey of
Entrepreneurs (ASE). The ASE, which responds to users’ requests for more frequent and more
7 Further information about the SBO can be found at hitp:/www.census.gov/econ/sbolindex. himtimely data, will provide an updated socio-economic portrait of America’s business owners in the
years between the Survey of Business Owners. Data from the inaugural 2014 ASE are
tentatively scheduled for release in summer 2016. The ASE has an innovative modular design
that enables the Census Bureau to supplement yearly baseline questions with special questions of
interest. For example, the 2014 ASE collection will ask special questions related to innovation.
‘The proposed module for the 2015 ASE collections concems management practices, which could
include questions about the use of nontraditional employer-employee structures.
Several of the research projects described above in the answer to question #2 (the nonemployer-
LEHD linkage and the 1099-LEHD linkage) could produce time series statistics on the
demographics of who is in traditional employment, who is in self-employment, who is in 2 1099
work arrangement, and who is in a mix of these various types. The possible demographic
characteristics are age, gender, race and ethnicity, and education.
4. Some surveys today indicate a significant percentage of the American workforce will receive
some income through non-traditional sources by 2020, What tools-statutory or otherwise
would help Census keep pace with the measurement needs of the changing workforce?
The most important tool to help the Census Bureau keep pace with the measurement needs of the
changing workforce would be to broaden and deepen the scope of administrative data available
to the Census Bureau. One strength of the Census Bureau, above and beyond the surveys we
conduct, is our use of administrative records (administrative records are defined as person-level
or firm-level data collected for non-statistical purposes such as tax collection). For example, the
Census Bureau has a multi-decade history of using IRS administrative data on businesses as the
foundation of our business universe (the business universe is a list ofall businesses in the
country used for the Economic Census and other business surveys), and we have a long history
of using IRS administrative data on individuals to assist us in measuring the quality of our survey
data and helping to create imputations for missing data in our household surveys. Our use of
administrative records occurs with a commitment to handling data responsibly and protecting the
confidentiality of the data we collect,
‘As mentioned several times above, the LEHD data is # key building block for linking together
information about traditional employer jobs and other types of employment and earnings. The
LEHD is the Census Bureau’s frame of private sector jobs in the United States. The data
underlying the LEHD come from the State Unemployment Insurance systems, with
Memorandums of Understanding in place with 50 different states plus the District of Columbia,
‘enabling the Census Bureau to collect and use the state-level administrative data, The Census
Bureau's ability to efficiently collect and process the LEHD data would be substantially
enhanced by changes in data sharing legislation that would allow the U.S. Department of Health
and Human Services to share the National Directory of New Hires database with the Census
Bureau,
The ability of the Census Bureau to develop statistics on the level and the growth of non-
traditional forms of employment would be enhanced by greater access to several items of IRS
‘As mentioned above, information on income reported on the 1099 tax filings would enableus to answer questions about whether 1099 workers are exclusively self-employed, whether they
use this self-employment income to supplement earnings from traditional employer jobs, and the
shares of income that they receive from each. Furthermore, information on the organization
issuing the 1099 would enable us to determine how many nontraditional (“gig”) jobs the
individual holds in a given year. The Census Bureau's ability to obtain access to information on
1099 tax filings requires the development of agreements between Census and IRS, and access to
some other tax data may require changes in Department of the Treasury regulations. Census and
IRS collaborate on statistical uses of Federal tax data along a number of dimensions.
As mentioned above, allowing the Census Bureau to obtain and use all the Social Security
Numbers on the IRS Schedule C filings would greatly enhance our ability to assign self
‘employment earnings to a specific individual rather than a specific household. It would also be
helpful to get the net profit or loss information, in addition to the receipts information, that is on
the Schedule C. Access to these items, which are negotiable with IRS, would assist us in
providing more accurate statistics on the levels and trends of self-employed persons, their
demographic characteristics, how individuals mix traditional and self-employment jobs, and the
income derived from the resulting mix. Furthermore, these more accurate data combined with
the work history of self-employed individuals would allow research into why they became self-
employed.