Supporting Statement A CPS Contingent Worker_2016

CPS Contingent Worker Supplement Survey

1220-0153

December 2016

SUPPORTING STATEMENT

A. JUSTIFICATION

1. Necessity of the Information Collected

The purpose of this request for review is for the Bureau of Labor Statistics (BLS) to obtain clearance for a reinstatement with change for the Contingent Worker Supplement (CWS) to the Current Population Survey (CPS), which was last conducted in February 2005. The proposed CWS questions focus on contingent workers--those who do not expect their jobs to last or who report that their jobs are temporary--and workers in alternative employment arrangements, such as independent contractors, on-call workers, temporary help agency workers, and workers provided by contract firms (see proposed questionnaire in Attachment A). As part of the CPS, the CWS will collect information about individuals ages 15 and over from a nationally representative sample of approximately eligible 60,000 U.S. households. The CWS is sponsored by the U.S. Department of Labor.

The collection of the CWS in May 2017 will be the first collection of the supplement since February 2005. With the exception of February 2003, the supplement had been conducted every 2 years from February 1995 to February 2005. In order to maintain data comparability over time, the questionnaire will be largely the same as that used in 2005. This will ensure that estimates from today’s economy can be compared with estimates from past collections. However, because new types of work have emerged since the last fielding of the CWS, BLS is adding four new questions to the end of the CWS. These new questions will: (a) explore whether individuals obtain customers or online tasks through companies that electronically match them, often through apps, and (b) examine whether work obtained through electronic matching platforms is a source of secondary earnings (see Section 8 of Attachment A for the proposed new questions).

The CPS has been the principal source of official Government statistics on employment and unemployment since 1940 (over 75 years). Collection of labor force data through the CPS helps BLS meet its mandate as set forth in Title 29, United States Code, Sections 1 through 9 (Attachment B).

2. Needs and Uses

The CWS will provide information on the number and characteristics of workers in contingent jobs--that is, jobs that are structured to last only a limited period of time. The CWS also provides information about workers in several alternative employment arrangements, including those working as independent contractors, on-call workers, temporary help agency workers, and workers provided by contract companies. Although the CWS was fielded every 2 years from 1995 to 2005, there have been no comparable and reliable statistics in recent years to show how the number and characteristics of these workers are changing over time. The 2017 CWS will allow researchers and policy makers to evaluate how the number and characteristics of these workers has evolved. Policy makers also can use these data to inform the design of regulations for different types of workers.

In addition to the contingent and alternative work arrangement questions, which were included in the 2005 collection of the CWS, the May 2017 collection of the supplement will include four new questions. Specifically, two questions will focus on whether individuals obtain customers or online tasks through companies that match them, often through mobile apps. These types of jobs include people using their own cars to drive customers from one place to another, delivering something, or doing customers’ household tasks or errands, as well as online tasks such as data entry, translating text, web or software development, or graphic design. In addition, two questions will examine whether work obtained through electronic matching platforms is a source of secondary earnings. These four new questions will provide researchers and policy makers with much-needed data about a little-studied but emerging group of workers (see Section 8 of Attachment A for proposed new questions).

The Department of Labor will use these data to determine whether existing policies and regulations sufficiently protect workers in contingent and alternative work arrangements, as well as those who obtain work through online intermediaries. The detailed demographic and other information on the CWS will be invaluable in planning Federal programs and formulating policy.

BLS published a summary of the findings from the February 2005 collection in a press release issued in July 2005 (see attachment C).

3. Use of Information Technology

The Census Bureau collects the CPS data, including designing the sample, conducting the interviews, training and monitoring the interviewers, and maintaining a quality control program. These efforts by the Census Bureau keep respondent burden as low as possible.

The CPS and all of its supplements, including the CWS, are collected 100% electronically by using Computer Assisted Telephone Interviews and Computer Assisted Personal Interviews (CATI/CAPI). The data collection instrument is programmed in Blaise, a Windows-based survey processing system developed by Statistics Netherlands and licensed by Westat in the United States. The questions in the CWS were designed to obtain the required information with minimal respondent burden.

4. Efforts to Identify Duplication

There are no Federal surveys that obtain comparable information on the number and characteristics of contingent workers and workers in alternative employment arrangements, such as independent contractors, on-call workers, temporary help agency workers, and workers provided by contract firms. The CWS is a rich source of information about these workers, providing detailed demographic information about each type, whether they prefer a traditional arrangement, whether they are covered by health insurance and whether it can be obtained through their employer, and their pension plan coverage and participation. In addition, the 2017 CWS will include a small number of new questions about whether people obtain work through online intermediaries who match them to customers.

While no Federal surveys contain comparable information on the full range of information collected in the CWS, some data sources contain information on selected CWS items.

For example, a variety of data sources contain information on workers in the temporary help services industry, including the Current Employment Statistics survey, the Occupational Employment Statistics survey, the Quarterly Census of Employment and Wages, and the County Business Patterns; however, none of these provide information about the industry to which temporary-agency employees are assigned, nor do they provide detailed demographic information about these workers. The monthly CPS collects information on temporary help services, but research has shown that approximately half of those paid by temporary help agencies report their place of assignment when asked industry questions about their employment, which would lead to underreporting of temporary-help agency workers. This underreporting problem likely also affects data on workers in temporary help services collected in other Federal household surveys, notably the American Community Survey and the Survey of Income and Program Participation. The National Longitudinal Surveys include some questions to identify temporary-agency workers, as well as limited information about some other employment arrangements; however, these data are based on small sample sizes and are restricted to specific age cohorts.

Also, the 2012 CPS Disability supplement collected rudimentary information about whether individuals were in temporary jobs, but unlike the CWS, it did not include questions to adequately identify individuals in jobs that were structured to be temporary, as opposed to temporarily being in a job that could have continued. Thus, data from the disability supplement likely considerably overstate the number of contingent workers.

Some non-Federal surveys have also collected limited information about contingent workers and alternative employment arrangements. The Panel Study of Income Dynamics collects information about workers in temporary help services, but like the CPS, many respondents probably report about their place of assignment when asked questions about the industry of their employer. The RAND-Princeton Contingent Worker Survey, collected in 2015 as part of the American Life Panel, contained an abbreviated set of questions from the CWS as well as questions about gigs, HITs (Human Intelligence Tasks), and other small paid jobs. However, this survey had a much smaller sample size than the CWS and also did not include some of the questions necessary to identify alternative arrangements to generate estimates comparable to the 1995 CWS.

5. Minimizing Burden to Small Entities

The data are collected from households; their collection does not involve any small businesses or other small entities.

6. Consequences of Not Collecting

Because the CWS has not been conducted since 2005, there has been a dearth of high quality data in recent years on workers in contingent and alternative work arrangements. With the advent of new business models, there is great interest by researchers and policy makers as to how many of these types of workers there are and whether the labor market has undergone structural change. Law makers need accurate data to design regulations and policies to adequately protect these types of workers. If the CWS is not conducted, there will continue to be a lack of information about how the labor market has evolved and how the number of these types of workers has changed.

In addition, more information is needed about emerging types of work arrangements: specifically, those in which workers are connected directly to customers through websites or mobile apps. The CWS will provide much-needed information on these types of work arrangements. This information--in combination with other information collected in the CWS and the monthly CPS, such as earnings, health insurance coverage, and retirement plan participation--will help guide law makers to determine if new policies or regulations are needed to protect these and other types of workers.

7. Special Circumstances

There are no special circumstances. The CPS data are collected in a manner that is consistent with the guidelines in 5 CFR 1320.5.

8. Federal Register Notice/Consultation Outside the Agency

1. Federal Register Notice:

BLS received seven comments as a result of a Federal Register Notice published in 81 FR 67394 on September 30, 2016. One comment was supportive and suggested no additional changes. BLS responses to the remaining comments are as follows.

The second comment, while supportive of the CWS, asked BLS to document its outreach efforts about the proposed new questions and the nature of the feedback. The comment suggested including more information in Section 8b of Supporting Statement A and including an attachment summarizing the primary points of view received in the feedback.

As requested in the comment, BLS updated Section 8b of Supporting Statement A to include the outreach efforts. The new Attachment I describes the outreach efforts and feedback received in more detail.

The third comment, while supportive of the CWS, suggested that some of the questions in the core CWS should be modified for later collections. These suggestions for future collections included:

• Evaluate questions from the 2005 CWS to ensure that they are still relevant, in particular making sure that they could be collected through a self- and proxy-reported survey.

• Specifically asking workers whether their status (employee versus independent contractor, for example) was imposed on them by their employers.

BLS appreciates the feedback about what to possibly include in a future CWS. If BLS should receive funding for a future CWS, and if funding is available for researching possible changes to the survey, BLS will review the questionnaire to see whether changes should be made. This would be a multi-year project, involving considerable research, including a literature review, consultation with subject-matter experts and survey methodologists, development of a questionnaire, and thorough cognitive testing. BLS is particularly cautious in redesigning surveys because improvements in questionnaires need to be weighed against the likelihood that changes could adversely affect the comparability of the data.

The fourth comment advocated revisions to the CWS beyond the additional four questions, with four specific suggestions.

1. Change the term “contingent” to “alternative employment arrangement” and the term “contingent worker” to “workers in alternative employment arrangements.” The comment stated that that the quality, utility, and clarity of the questionnaire would be greatly improved by such a change.

The terms “contingent” and “contingent worker” are terms that BLS uses in its publications to describe workers whose jobs are temporary; “alternative work arrangements” is a term used by BLS to describe different types of workers, such as independent contractors, temporary help supply workers, on-call workers, and workers supplied by contract firms. The term “contingent” appears only once in the CWS questionnaire: in the language that is read to the respondent if they specifically ask the interviewer about why the data is being collected. This optional response reads as follows: “The information you give is important. Answers to the supplement questions will provide objective information to help in the description and measurement of contingent work. "Contingent work" covers a variety of diverse employment arrangements, including temporary work, independent contracting, and on-call work, among others. Whether a worker can expect continuing employment is a defining element of contingent work.”

BLS agrees that this language would be clearer if the term “contingent” were not used. BLS will amend this language to the following: “The information you give is important. Answers to the supplement questions will provide information to help us understand a variety of employment arrangements, including temporary work, independent contracting, and on-call work, among others.”

2. Delete questions and condense the survey screener question in order to establish clear categories of alternative employment arrangements and workers, as well as to reduce respondent burden and confusion.

One major goal of fielding the 2017 CWS is to see how the number of contingent workers and workers in alternative employment arrangements have changed over time. What may appear to be small changes in questions can result in large differences in how people answer them; thus, even relatively minor changes to the core CWS questionnaire could well adversely affect the comparability of the data, resulting in BLS and others being unable to see how the number of these workers has changed since 2005. For this reason, the four new questions added to the CWS appeared at the end where they could not impact the responses to earlier questions. BLS survey methodologists and subject-matter experts believe that deleting questions at the beginning of the CWS and replacing them with a different question—as this comment recommends—would likely adversely affect the data comparability between 2017 and earlier years. Also, BLS survey methodologists have found that, in order to classify respondents into a number of different categories, a series of yes-or-no questions tends to result in more consistent data than does a single question with a number of response options.

In accordance with OMB guidance, all substantial changes to federal survey questionnaires must be carefully tested to ensure that the new questions are clearly understood by respondents. BLS had limited time to make changes to the CWS. In addition to the risk of undermining data comparability, the BLS survey methodologists and subject-matter experts believe that the recommended changes would necessitate a substantial redesign to the survey and would require a considerable amount of time for testing, making a 2017 fielding unattainable.

3. If BLS decides not to revise the screener, the comment suggested swapping the order of questions S1 (Some people are in temporary jobs that last only for a limited time or until the completion of a project. Is your job temporary?) and S2INS (Are you paid by a temporary help agency?). The comment also suggested swapping the order of S2 (Even though you told me your job is not temporary, are you paid by a temporary help agency?) and S1SCRI (Could you have continued to work at your job if you had wished?). These changes were recommended as a way to improve the likelihood that employment arrangements are categorized correctly, in addition to providing respondents more context to answer questions.

It is important to note that the CWS questionnaire is highly tailored to respondents’ answers. Therefore, while all eligible respondents are asked question S1, only a subset of respondents receive questions S1SCRI, S2, and S2INS. This is done to provide the most meaningful context for respondents and to reduce respondent burden. Because the question skips are extremely complicated in the CWS, the recommended changes would involve a major reworking of the questionnaire, likely impacting virtually all questions in the survey. As mentioned previously, even minor changes in questions can adversely affect data comparability. BLS survey methodologists and subject-matter experts believe that the proposed changes, which would alter questions early in the survey, would likely result in undermining data comparability across years. There is also insufficient time to redesign and test the CWS if it is to be fielded in 2017.

Additionally, question S1 is intended to capture anyone whose job is temporary, regardless of whether they work for a temporary help agency or not. Previous research using BLS data has shown that the number of temporary workers who are employed directly by an employer is significantly larger than those who are employed by temporary help agencies. With the current order of the questions, the context is appropriate for both groups of temporary workers.

4. The comment also requested the opportunity to review any new questions for the CWS.

Should BLS receive funding to conduct the CWS after May 2017, any changes to the CWS questions would be thoroughly evaluated and tested. A Federal Register notice will describe changes to the questionnaire, and the public will have an opportunity to comment. Also, BLS will likely solicit comments through its Technical Advisory Committee or Data Users Advisory Committee.

The fifth comment called for a broader reassessment and expansion of the CWS that would provide more data to better understand the “fissuring” of work and to bring the questionnaire up to date. It also called for more industry-specific questions for industries in which workplace fissuring is common (for example, higher education).

BLS believes that one of the most valuable aspects of a new CWS collection is that it will be possible to evaluate how the number of contingent workers and workers in alternative employment arrangements have changed over time. A major redesign of the CWS would likely result in the data not being comparable to those collected in earlier surveys.

BLS notes that researchers disagree as to what constitutes workplace “fissuring.” In order to maintain data comparability, BLS believes changes to the questionnaire should be made only after careful research. A broad reassessment and expansion of the CWS would require considerable time and funding, and a May 2017 fielding would not be possible. Such a reassessment would involve considerable research and would include an evaluation of whether changes would likely result in useful information. For example, questions given to a small number of respondents, such as those employed in a specific industry, might not yield reliable statistics due to small sample sizes. In addition, BLS would need to evaluate whether questions about how companies organize their production would be more appropriate for establishment surveys or surveys of companies’ Human Resources experts; household survey respondents may have little understanding about how a company’s workforce or supply chain are structured.

The sixth comment stated that data from the CWS would fail to capture the nuanced picture of independent work because the questions rely heavily on the terms “business,” “work,” and “job,” and those terms no longer mean exactly the same thing to different individuals. The comment requested that new questions be developed for the CWS to identify additional sources of income outside of “jobs” and “businesses,” saying that such questions would help capture supplemental economic activity.

It is well-known among survey methodologists that questions about income are notoriously sensitive and have high rates of nonresponse. For this reason, income-related questions often require more involved testing than questions on less sensitive topics. BLS survey methodologists and subject-matter experts do not believe that questions about supplemental income could be developed and tested in time for a May 2017 fielding of the CWS.

Information about income can be obtained from another supplement to the CPS, the Annual and Social Economic Survey (ASEC). This supplement contains questions about earnings from jobs or businesses, as well as detailed questions about income, including but not limited to child support payments, Social Security payments, disability income, veterans’ benefits, pension income, retirement income, annuity income, income from property, regular financial assistance, and unemployment compensation. The supplement also contains questions about any other income the respondent may have, such as income from foster child care, alimony, jury duty, Armed Forces Reserves, severance pay, hobbies, or any other source.

About three-quarters of CWS respondents will have completed the ASEC in a prior month. Both the ASEC and the CWS data will be publicly available, allowing researchers to link the files. This will give researchers a rich data source to investigate respondents’ incomes in addition to information about contingent work and alternative employment arrangements.

The final comment stated that the four new CWS questions as currently designed would fail to measure the size and characteristics of the “gig or on-demand economy” or of “independent work.” The comment expressed concern that the new CWS questions, while intending to fill an information gap, would “paint an inaccurate and incomplete view of the gig economy and independent workforce” that would ultimately impede the ability of federal and state lawmakers to make sound policy decisions. The comment recommended revising the questions to “solicit responses that provide a more holistic picture of the independent workforce that is not limited to electronic platforms.” To that end, the comment suggested that the questions should focus more on the nature of the work completed, should not use the terms “main job” and “second job,” should expand the reference period from the prior week to the prior year, and should address workers’ reliance on earnings and whether they engage in the independent workforce by personal choice or financial necessity.

BLS notes that there is currently no standard definition of the “gig economy” or of the “independent workforce.” In fact, there is great disagreement among researchers as to what should be included in these definitions. The primary goal of the CWS is to measure the number and characteristics of contingent workers and workers in alternative work arrangements and to see how these have changed over time. The survey collects data about many more types of workers than simply those who engage in electronically-mediated work, and BLS feels that the data will be useful to those studying the gig economy or independent work. Many researchers find the CWS data to be invaluable, as the CWS provides information about independent contractors, on-call workers, day laborers, temporary help supply workers, and workers in jobs that are not structured to be permanent—types of workers who are included in many definitions of the gig economy and independent work. The four new questions will add to this information by providing data about electronically-mediated work, another type of work frequently included in definitions of the gig economy and independent work. In addition, the 2017 CWS microdata will be publicly available, ensuring that researchers can decide for themselves which types of workers to study. The CWS is designed to be very flexible and to facilitate research; for example, BLS routinely publishes three different definitions of contingent workers based on these data.

Additionally, the changes to the new questions requested in this comment were substantial, though BLS notes that some of changes requested are already in the CWS—specifically, independent contractors, on-call workers, day laborers, temporary help supply workers, and workers in jobs that are not structured to be permanent are asked whether they would prefer to be in a different type of work arrangement and the reason why they are in their current work arrangement. BLS survey methodologists and subject-matter experts believe that the number of topics about which the comment recommended collecting information could not possibly be addressed in 4 simple questions.

2. The following people have been in continuous consultation concerning the development of the survey:

U.S. Department of Labor

Molly Irwin

Acting Deputy Chief Evaluation Officer

U.S Census Bureau

Lisa A. Clement

Survey Director, Current Population 

  and American Time Use Surveys

Associate Directorate Demographic Programs

In addition to the above, a statement soliciting comments for improving CPS data is prominently placed in all Census Bureau publications that cite CPS data. A similar statement is included in the technical documentation that accompanies the microdata files. Furthermore, the advance letter provides respondents with an address at the Census Bureau to which they can submit general comments on the survey, specifically those regarding respondent burden (see Attachment D).

In the development of the four new questions, BLS did considerable stakeholder outreach. This is described more fully in Attachment I.

9. Paying Respondents

The Census Bureau does not make any payments or provide any gifts to individuals participating in the CPS.

10. Assurance of Confidentiality

The Census Bureau will collect the CWS in compliance with the Privacy Act of 1974 and OMB Circular A-130. Each sample household will receive an advance letter (Attachment D) approximately one week before the start of the initial CPS interview. The letter includes the information required by the Privacy Act of 1974, explains the voluntary nature of the survey, and states the estimated time required for participating in the survey. Field representatives must ask each respondent if he/she received the advance letter and, if not, must provide a copy of the letter to each respondent and allow sufficient time for him/her to read the contents.

Also, interviewers provide households with the pamphlet "The U.S. Census Bureau Respects Your Privacy and Keeps Your Personal Information Confidential," which further states the confidentiality assurances associated with this data collection effort (Attachment E).

All information given by respondents to Census Bureau employees is held in strict confidence under Title 13, United States Code, Section 9 (Attachment F). Each Census Bureau employee has taken an oath to that effect and is subject to a jail penalty and/or substantial fine if he/she discloses any information given to him/her.

11. Justification for Sensitive Questions

No sensitive questions are asked in this supplement.

12. Estimate of Respondent Burden

The estimated respondent burden for the CWS is 7,050 hours. This is based on an average respondent burden of approximately 9 minutes for an estimated 47,000 respondents. The actual respondent burden is dependent upon the characteristics of each respondent.

The overall annualized dollar cost to respondents for collection of the CWS is $98,700. This estimate assumes a wage rate for all respondents of $14.00 an hour, the median hourly earnings for workers paid by the hour in 2016.

Estimated Annualized Respondent Hour and Cost Burden

Activity	Number of Respondents	Number of Responses per Respondent	Total Responses	Average Burden per Response (In Hours)	Total Burden Hours	Average Hourly Rate*	Total Burden Costs
Contingent Worker Supplement	47,000	1	47,000	.15	7,050	$14.00	$98,700

13. Estimate of Cost Burden

1. Capital start-up costs: $0

2. Total operation and maintenance and purchase of services: $0

14. Cost to the Federal Government

The total cost of the May 2017 supplement is approximately $1 million. This cost is to be borne by the Department of Labor and largely represents the charge by the Census Bureau for conducting the CWS. Census activities for the CWS total approximately $0.9 million and include programming the questionnaire (the 2005 questionnaire was programmed in software that is no longer used), developing interviewer training materials, collecting data, processing survey microdata, and developing public use files. The remaining $0.1 million is for the BLS staff to prepare a news release and publish estimates, as well as develop specifications, test the collection instrument, and evaluate data quality.

15. Changes to the Respondent Burden

The supplement is administered primarily to households in which there is an employed person, although a small number of the unemployed are also eligible. Total respondent burden for the collection of the 2017 CWS is estimated to be 7,050 hours, an increase of 1,200 hours from the previously approved collection. The increase is due to an increase in the overall sample size, as the 2005 CWS was administered to three-fourths of the CPS sample rather than the full CPS sample. In 2017, the number of eligible persons per month is approximately 47,000. Also, as a result of the new questions, there is an increase in the estimated average response time for these households, from 8 to 9 minutes since the last approval.

16. Time Schedule for Information Collection and Publication

The supplement is expected to be collected in May 2017 during the week containing the 19^th of the month. Processing of this supplement will begin the month following the collection. Survey results will appear first as a news release several months after the CWS is collected.

These news releases will be published in electronic and paper formats. The electronic news release will be posted on the BLS webpage at www.bls.gov/cps. Additionally, the Census Bureau will release a public use version of the microdata after the publication of the news release.

17. Request to Not Display Expiration Date

The Census Bureau does not wish to display the assigned expiration date of the information collection because the instrument is automated and the respondent, therefore, would never see the date. The advance letter sent to households by the Census Bureau contains Census’ OMB clearance number for the CPS and Census’ version of the failure to comply notice. (See Attachment D) Copies of this advance letter are stockpiled by the Census Bureau for use as needed; changes to the letter would make the current inventory of letters unusable.

18. Exceptions to the Certification

There are no exceptions to the certification.

9