No. 15-10 Changing Patterns in Informal Work Participation in the United States 2013–2015 Anat Bracha, Mary A. Burke, and Arman Khachiyan Abstract In light of the weak labor market conditions that prevailed in the United States from 2008 until recently, participation in alternative income-generating activities, such as informal side jobs, is likely to have increased during that period. According to the same logic, participation in informal work should have declined more recently, as conditions in the formal labor market improved. However, in recent years technological innovations have created a number of new opportunities for engaging in informal work. Such innovations may have promoted structural increases in informal work participation, and, if so, we would expect informal work participation to remain elevated or increase further even as the economy improves. To test these predictions we designed the Survey of Informal Work Participation, fielded within the Federal Reserve Bank of New York’s Survey of Consumer Expectations (SCE-SIWP). The survey was fielded in December 2013 (Survey 1) and again in January 2015 (Survey 2), on two separate, nationally representative samples. We find that the participation rate increased significantly between the surveys, among both men and women. Differences in participation rates based on educational attainment (among both men and women) and differences based on an individual’s formal wage (among men) became less pronounced between the surveys or disappeared altogether. We hypothesize that recent increases and improvements in the supply of informal work platforms help to account for the higher participation rate in Survey 2, as well as for the fact that the set of informal workers appears more socioeconomically diverse in Survey 2 than in Survey 1. Our results further suggest the existence of two distinct groups—one group of individuals who work informally to offset negative economic shocks, and another group who work informally despite being already fairly well off. JEL Classification: R11, R23 Anat Bracha and Mary A. Burke are senior economists and Arman Khachiyan is a senior research assistant in the research department of the Federal Reserve Bank of Boston. Their e-mail addresses are [email protected], [email protected], and [email protected]. This paper, which may be revised, is available on the web site of the Federal Reserve Bank of Boston at http://www.bostonfed.org/economic/current-policy-perspectives/index.htm. The views expressed in this paper are those of the authors and do not necessarily represent the views of the Federal Reserve Bank of Boston or the Federal Reserve System. This version: October 2015 1 1. Introduction The Great Recession caused the loss of millions of jobs across the United States. Of those able to maintain their employment, many faced reductions in hours that shifted them into part- time status and were often accompanied by a loss of essential benefits such as health care. The job losses also contributed to significant income losses that were not fully replaced by unemployment insurance and other government programs (Rothstein and Valletta 2014). According to the Bureau of Labor Statistics (BLS), the national unemployment rate increased from 5 percent to 9.5 percent from the beginning of the recession (December 2007) to its official end date (June 2009) and subsequently rose to a peak of 10 percent in October 2009. Nearly eight years after the onset of the Great Recession, the unemployment rate has only recently approached its pre-recession range and remains, at 5.1 percent as of September 2015, well above the 4.4 percent rate achieved in May 2007. In addition, a broader measure of labor market slack given by the U6 rate—which includes people working fewer hours than desired—remains at 11.3 percent as of the second quarter of 2015 (seasonally adjusted), more than 3 percentage points above its pre-recession minimum of 8.2 percent. In light of the weak labor market conditions that prevailed in the United States from 2008 until recently, we might have expected that participation in alternative income-generating activities, such as informal side-jobs, would have increased during that period.1 According to the same logic, participation in informal work should have declined more recently, as conditions in the formal labor market improved. However, in recent years technological innovations have created a number of new opportunities for engaging in informal work. Such innovations may have promoted structural increases in informal work participation, and, if so, 1By informal work we refer to any income-generating activity that does not involve a contract between an employer and an employee (except possibly for contracts involving a single task). This definition includes activities that monetize possessions (such as selling used goods or renting out one’s property) as well as activities that monetize free time and skills (such as babysitting). Typical features of informal work are the following: (1) it involves a greater degree of scheduling freedom than a formal job would, (2) the worker is paid on a per-service or per-good basis, and (3) the work does not provide benefits such as health insurance or pension contributions. See Gёrxhani (2004) for a literature review discussing the wide range of definitions of the informal sector. Gёrxhani’s discussion, however, predates the recent rise of mobile technologies facilitating informal work. 2 we would expect informal work participation to remain elevated or increase further even as the economy improves. To test these predictions, we designed the Survey of Informal Work Participation, fielded within the Federal Reserve Bank of New York’s Survey of Consumer Expectations (SCE- SIWP).2 The survey was fielded in December 2013 (Survey 1) and again in January 2015 (Survey 2), on two separate, nationally representative samples. Our main motivation in the first survey (Survey 1) was to assess the extent and intensity of participation in paid informal work activities and its determinants, the types of activities engaged in, and the extent to which engaging in such activities helped individuals to compensate for negative economic shocks experienced during the recession and afterwards.3 Our main motivation in the second survey (Survey 2) was not only to follow up on the main outcomes from Survey 1, but also to determine whether the motivations for engaging in informal work, and/or the types of individuals drawn to such work, changed over time as the labor market improved. In Survey 1, 40 percent of respondents (whether female or male) participated in some type of informal paid activity (other than completing surveys). Among men, those employed part time were more likely to participate than those employed full time, whereas among women, employment status was not a significant factor in participation. Among both women and men, and controlling for employment status, individuals with higher formal wages were less likely to participate than those with lower formal wages and informal work was driven mainly by the motivation to earn extra money. Informal work helped a significant share of those who engaged in it to offset negative effects of the recession. In Survey 2, the share of survey-takers who reported participating in informal paid work increased significantly—from 40 percent to 52 percent among men and from 40 percent to 60 percent among women. Earning extra money remained the most widely cited reason for participating in informal work. Among both women and men, participation rates became more equal across education classes in Survey 2. Among women, this equalization reflected in part a large increase in participation among those with high school or less, while among men, the 2The SCE, and thus also the SCE-SIWP, is operated jointly by The Conference Board and Nielsen. 3For a complete description of results from Survey 1, see Bracha and Burke (2014). 3 equalization embedded a large increase in participation among those with a graduate degree. Among men, participation rates became more equal across groups classified based on employment status. As of Survey 2, men from across the formal income distribution are roughly equally likely to participate in informal work, while among women, the negative association between formal income and informal participation remains in force. The Survey 1 finding that informal work (among men) was concentrated among part- time employees—many of them seeking to offset negative employment shocks—suggested that informal participation was indicative of labor market slack. Accordingly, we observe that among men employed part time, the informal work participation rate decreased between the survey periods, while labor market conditions (including the U-6 rate) improved, consistent with the notion that individuals take on informal work in order to smooth income across the business cycle. At the same time, however, informal participation increased between the surveys among highly educated and highly paid men, an outcome that likely reflects the fact that recent technological innovations have expanded the set of informal work opportunities and made it easier to engage in such work. Indeed, among both men and women and in both surveys, more than half of those who report engaging in informal work are performing internet- based tasks. In addition, one of the categories with the highest increase in participation between surveys was “online tasks,” which refers to activities such as rating pictures or copy-editing online. Female informal work participants in Survey 2 were more likely than those in Survey 1 to report both that informal earnings were their main source of income and that informal work helped at least somewhat to offset recent negative employment shocks. Taken together, our results suggest that some individuals continue to seek out informal work in order to offset negative economic shocks, while others engage in informal work—despite already being fairly well off—because it offers an easy way to earn extra cash. Supporting this idea, we observe that informal work participants with higher formal wages also tend to earn higher informal wages. Another important finding from Survey 2 concerns how the Bureau of Labor Statistics would classify those who participate in informal work. We find that the BLS system classifies 4 some informal work participants as “employed” even though they self-report that they are not employed. The same type of disagreement in classification occurs to a much lesser degree among people who do not participate in informal work. This result suggests that those who engage only in informal work may be classified by the BLS as employed despite the fact that the individual does not consider informal work to constitute solid employment and may still be looking for formal work. Background and Related Studies: Informal Work and the Peer-to-Peer Economy The number and types of paid informal work opportunities have expanded in recent years, in no small part due to the appearance of new technologies facilitating the so-called peer- to-peer economy.4 Well-known peer-to-peer businesses include Uber, a taxicab-like business that connects drivers with riders via mobile phones; Airbnb, which enables individuals to rent out their home for brief stays; Amazon Mechanical Turk, which offers the opportunity to do basic computing work from home on a fee-for-service basis; and Taskrabbit, which facilitates spot contracting for personal services. All four of these businesses and many others operate through websites and/or mobile applications, and all of them were born relatively recently: Amazon Mechanical Turk was founded in 2005, and the other three in 2008 and 2009. While online platforms have the potential to disrupt existing labor markets, information on the supply of labor to these platforms is lacking. In the past, the BLS has studied “contingent workers,” defined as those working in temporary jobs or jobs not expected to last. However, the last such report was issued in July of 2005, before many of these platforms were created, and the agency currently lacks funding to conduct a follow-up study that might capture new classes of contingent workers (Weber 2014). The U.S. Government Accountability Office (GAO 2015) recently issued a report on contingent work, but the definition of a contingent worker used in 4These activities, or some subset of them, are also referred to as “the sharing economy” and also include cases of “crowdsourcing,” in which actors (including firms) divide a large work task among many individuals operating independently of one another, often using online-based, spot contracting. 5 the underlying surveys would not have captured those supplying labor to most or all online platforms.5 The best estimates of the size of the peer-to-peer economy and its importance in the U.S. economy so far have come from private firms and organizations with a wide range of interests in the topic. PricewaterhouseCoopers (PWC) estimated that, as of 2013, the five core sharing economy sectors together accounted for $15 billion in global revenue.6 Further, they predicted that between 2013 and 2025 these sectors would experience revenue growth rates ranging from 17 percent to 63 percent. The same report estimated that the five traditional rental sectors totaled $240 billion in global revenue in 2013 and would grow at rates between -5 percent and 5 percent between 2013 and 2025.7 A self-reported study of administrative data from Uber, the online platform attracting numerous headlines recently, suggests that the firm grew exponentially during the past few years. In the 18 months ending in January 2015, the number of drivers (referred to internally as “driver-partners”) providing rides through Uber grew from nearly 0 to over 160,000. The report also found that most Uber drivers held formal employment prior to joining Uber and that they were attracted to the platform because it offers flexible hours and stable wages. This analysis by Hall and Krueger (2015) finds that many Uber drivers cite the desire to smooth income fluctuations as a reason for participating. While our own survey does not cover enough Uber drivers to enable direct comparisons with these studies, its findings shed light on broader informal labor market patterns that may also apply to Uber. The study most comparable to our own comes from a July 2014 survey with over 5,000 respondents—commissioned by Freelancers Union and Elance-oDesk—which finds that 34 percent of the national labor force, or approximately 53 million Americans, engaged in freelance 5 Surveys of contingent workers have focused largely on employees at traditional temporary employment agencies, or those who expect formal jobs to end soon. To our knowledge, these surveys have not, so far, incorporated the expanding range of informal work opportunities. 6 The core sectors consisted of: (1) peer-to-peer lending and crowdfunding, (2) online staffing, (3) peer-to-peer accommodations, (4) car sharing, and (5) music and video streaming. See PWC, “The Sharing Economy—Sizing the Revenue Opportunity.” Accessed October 15, 2015 at http://www.pwc.co.uk/issues/megatrends/collisions/sharingeconomy/the-sharing-economy-sizing-the-revenue- opportunity.html. 7 The traditional rental sectors included: (1) equipment rental, (2) B&B and hostels, (3) book rental, (4) car rental, and (5) DVD rental. 6 work8 over the previous 12 months.9 They estimate that total annual earnings from this freelance work amounted to $715 billion. Freelancers in this survey most often reported taking up such work for the extra income and schedule flexibility. To the best of our knowledge, the SCE-SIWP is the only survey of informal work participation that covers a nationally representative sample and is conducted and analyzed by a disinterested party. The survey covers a broad range of types of informal work, including but not limited to, those that are facilitated by internet or mobile platforms, and allows participants to write in unlisted activities. It has the added advantage of being conducted on a recurring basis, which allows us to track changes in informal work participation and its determinants over time.10 Attaching the survey to the established Survey of Consumer Expectations grants us access to the surveying expertise of Nielsen as well as additional information on responding households, including these households’ expectations of economic conditions for the nation at large and for their own household. Therefore, we are in a unique position to be able to comment on changes in the informal participation rate over time and to offer insights into the reasons for such changes. There are also several limitations of our survey that are worth noting. By definition, all individuals responding to our survey (which is conducted online and offers $15 compensation to respondents) are doing an online task for pay. Despite the fact that our sample is nationally representative based on standard demographic and geographic dimensions, it is reasonable to suspect that individuals doing paid online survey work are more likely to participate in other forms of informal work, and, in particular, in online-based informal work, than those not responding to paid surveys would be. This selection effect may cause us to overestimate the percentage of Americans who are engaged in informal work. Given this potential bias in estimated participation levels, we focus the analysis on identifying the determinants of informal work and the motivation for such work. In addition, we have no reason to believe that the 8 Freelance work in this survey is loosely defined as supplemental, temporary, project- or contract-based work, meaning that it largely overlaps with, but does not perfectly coincide with, our definition of informal work. 9 Elance-oDesk, Freelancers Union. 2014. “Freelancing in America: A National Survey of the New Workforce.” 10While two surveys are not sufficient to identify a trend, as future surveys are conducted (beginning in December 2015), our ability to discern trends will improve. 7 selection bias should have increased between the surveys, so changes over time in participation rates should be at least qualitatively robust. The remainder of the paper is organized as follows. Section 2 provides an overview of our data and sample characteristics. Section 3 conducts a graphical analysis of the changes in informal work participation between our two surveys. Section 4 presents a controlled analysis of these changes within a regression framework. Section 5 provides a summary of key findings and offers concluding remarks. 2. Data Overview 2A. Survey Overview, Key Definitions, and Selection of the Analysis Sample Both surveys solicited information on the nature and extent of informal work activities of respondents, the reasons for participating in informal work, and the economic importance of informal work to those who participated. We also collected basic information on individual and household characteristics, such as formal employment status, homeownership status, amount of liquid savings, and household size. We have access to additional demographic and other information (such as inflation expectations and job search activity), based on subjects’ prior participation in the monthly Survey of Consumer Expectations. In Survey 2 we preserved all of the most important questions from Survey 1 and added a series of questions designed to gather additional details about earnings and hours on specific informal tasks, as well as new questions designed to better assess how respondents coped with any negative effects the Great Recession had on their household financial situation. The full texts of the surveys can be found in Appendix: Survey 1 and Appendix: Survey 2. Individuals’ employment status is based on their response to the self-categorizing question “What best describes your current employment situation?” We then classify the responses into one of four employment status groups: (1) employed full time, (2) employed part time, (3) not employed formally but would like a job, and (4) other not working. The first three categories are distinct options in the multiple choice question, while the “other not working” 8 category includes anyone reporting one of the following: that they have no job and are not interested in a job, that they are currently on leave from a job, or that they are temporarily laid- off from a job. The survey respondents are classified as informal work “participants” if they meet both of the following criteria: (1) on a checklist question, they indicate that they engaged in at least one paid informal work task (other than paid survey completion)11 in the previous two years, and (2) on a separate question, they report a positive number of (total) paid informal work hours in a typical month. We require both outcomes to avoid any ambiguity concerning participation, because some individuals may check off an activity on the list of informal tasks and then report zero hours of typical informal work per month.12 Also, the combined criteria allow us to perform Heckman regressions (covered in Section 4) that jointly examine participation and hours. To track movements in informal work participation patterns between Survey 1 and Survey 2, all questions defining employment and informal work engagement from Survey 1 (as well as the control variables in the regression analysis) were repeated either verbatim or nearly verbatim in Survey 2.13 A full draft of Survey 2, can be found in Appendix: Survey 2. We also apply a consistent set of criteria in selecting analysis samples from each set of survey responses. 11If someone checks off only “survey completion” among the checklist of informal tasks, we do not consider that person an informal work participant. We rule out such individuals because, by virtue of participating in the SCE- SIWP, all of our respondents do paid survey work; therefore, if we included survey work among the qualifying tasks, we would (or at least should) observe a 100 percent participation rate. In fact, not all respondents mark paid survey work on the checklist, but that issue is not relevant here. 12This combination of responses does not necessarily represent a direct contradiction, because someone might have engaged in a task in the previous two years, but if that individual is not currently engaged in the task, he or she may consider zero hours to be appropriate for the “typical” month. 13The checklist question that provides the first criterion for informal participation differs in two respects between Survey 1 (question 29) and Survey 2 (question 27). The checklist in Survey 2 contains an additional item not included in Survey 1, “Driving for a ride sharing service like Uber, Lyft, or Sidecar.” Both checklist questions include an item labelled “other,” within which respondents could fill in something not on the list. Also, in Survey 1, respondents were presented with a list of tasks and were asked to check the box next to each task they engaged in during the past two years, while in Survey 2 respondents were presented with a list of tasks and asked to mark either “yes” or “no” for each task (regarding engagement in the task during the past two years). In Survey 1, no one wrote in any form of driving in the “other” tasks. In Survey 2 only six respondents marked “yes” for the “driving...” item, and, among these, only one individual marked only the “driving…” task. Therefore, only this last person might have been classified as a non-participant in Survey 1, and we conclude that the addition of the “driving” item to the list of tasks cannot account for the increased participation in Survey 2 compared with Survey 1. 9 To focus on adult respondents of working ages, we restricted the analysis sample to those ages 21 and over, and we removed anyone who indicated that he or she was retired. Any respondent supplying incomplete information on questions about his or her informal work participation, informal hours, informal earnings, formal employment status, formal earnings, formal hours, age, race, gender, educational attainment, homeownership status, or economic expectations was also removed from the analysis sample. We dropped two respondents from Survey 1 and one from Survey 2 for seemingly erroneous reporting of informal income or hours.14 Finally, from each survey we removed five respondents who had (individual) annual formal incomes of $600,000 or greater, a cutoff that is further justified in the regression analysis. Taken together, these restrictions reduced the sample size in Survey 1 from 1,218 to 778, and in Survey 2 from 1,220 to 701. Combining analysis samples from Survey 1 and Survey 2 results in a total of 1479 subjects, of whom 732 are women and 747 are men. Although the monthly Survey of Consumer Expectations is a rolling panel survey, participants are retired after a maximum of 12 months. Accordingly, there is no overlap between the sets of respondents to our two surveys. The rolling sample is maintained by Nielsen to approximately represent Census demographics along selected dimensions. Nielsen does not use hard quotas to maintain representativeness. Sample weights are used to adjust for remaining differences between the sample characteristics (along the selected dimensions) and the national population characteristics. Therefore, each of our (weighted) samples constitutes a distinct, representative sample of the U.S. population. 2B. Sample Characteristics For each survey, summary statistics were calculated separately for women and men using survey weights to match American Community Survey national population estimates based on income, education, region, and age.15 Table 1 shows summary statistics across all variables of interest for women in both surveys; Table 2 presents the corresponding statistics for 14 One subject in Survey 1 reported informal income almost identical to his formal income, while another reported formal income 87 times the amount of her formal income. In Survey 2 a subject reported 500 hours of informal work in the typical month. 15 Survey 1 was weighted based on 2012 ACS population estimates, Survey 2 was weighted based on 2013 ACS population estimates. 10