Appendix A: Estimating The Number Of Workers Affected By Wage And Tax Fraud In The Massachusetts Construction Industry

Introduction

Estimating the number of misclassified independent contractors and off-the-books workers in the construction industry has been a focus of researchers over the last 20 years. But while scholars have used state UI audits as the best direct measure of worker misclassification, there are many reasons to suspect that audits under-represent the rate of illegality. As highlighted in the report, state UI audits do not directly audit labor brokers, are likely to not detect many instances of cash-only payments, and disproportionately exclude small businesses given their ability to evade state auditors. Other government data sources feature similar limitations; after all, contractors and workers often go to great lengths to conceal evidence of illegality from government regulators and data collectors. To make matters worse for researchers, nationally-representative worker surveys—which are the bedrock of most labor market analyses in the United States—do not ask about the legality of workers’ employment situations. Given these data limitations, scholars are unable to assess the extent of worker misclassification using direct evidence.¹

To resolve these issues, scholars have developed an indirect empirical approach to estimate the incidence of wage and tax fraud; variants of this methodology have been featured in every known study of the topic in the last two decades. The foundation of this approach is a comparison between the aggregated results of worker surveys and that of UI payroll records. In effect, researchers have discovered that the number of workers who self-identify as working in the construction industry on Census surveys (i.e., total employment) far exceeds the number of workers employed by tax-paying construction employers as found in UI payroll records (i.e., legal wage-and-salary employment). Although this gap also includes some legitimate self-employed contractors, this difference captures a substantial number of misclassified independent contractors and off-the-books workers.

While this perspective has been foundational to numerous public policy papers on the underground construction economy, this view has been validated by work in academic journals. Most prominently, a 2013 article in the prestigious Journal of Labor Economics strongly advocated that workers who appear in nationally-representative surveys but fail to appear as employees in tax filings are quite likely to be misclassified independent contractors or workers engaged in a cash-only employment relationship.² Further, a 2012 article in Industrial Relations used the methodology advanced in those public policy papers— the aggregate difference between Census surveys of workers and UI records—as a means of estimating the number of informal laborers by industry.³ In sum, this indirect empirical methodology has been developed and applied by scholars as an appropriate means of estimating the scope of payroll fraud on a regional and national basis.

While a statistical comparison of worker surveys and UI records on the aggregate has emerged as the primary method of projecting payroll fraud in construction, there remains one empirical problem for which there has yet to be consensus. As mentioned above, the gap between total employment (i.e., workers self-reporting their industry) and legal wage-and-salary employment (i.e., UI records) features both legitimate, tax-paying self-employed workers and those engaged in payroll fraud. But distinguishing between these two groups— on an aggregate basis—has proved difficult. Multiple studies have recommended different ways of delineating rates of illegality in this group; a full review of those approaches is presented in a 2020 study published by the Institute for Construction Economic Research (ICERES) and co-written by two of the authors of the current report.⁴

To resolve this empirical uncertainty, the authors of this study apply—and further refine— the statistical approach advanced in that 2020 ICERES report. While the authors of the current study believe that this methodology is the best available approach of delineating the degree of illegality in the underground economy at this time, it is acknowledged that there is a nontrivial margin of error in these estimates given the problem of disaggregating legal from illegal self-employment in the data and other concerns of using survey data. This is true in all prior studies on this topic and should not be unexpected in the current report; after all, the authors have effectively been tasked with quantifying the extent of something that employers (and many workers) are actively trying to conceal.

Methodology & Results

The first step in this analysis uses the 2019 American Community Survey (ACS) to establish the total number of people working in Massachusetts’ construction industry. Administered by the Bureau of the Census, the ACS is a foundational tool of economists studying the labor market as it represents the largest annual household survey in the United States; in 2019, the ACS featured survey data from over two million households across the country, including 37,842 in Massachusetts.⁵ While the results taken from any subset of a population is subject to sampling error (i.e., meaning the sample is not entirely representative of the population), the sheer volume of respondents in the ACS minimizes these concerns; in sum, the ACS is considered a “gold standard” data source that scholars and economists routinely use to project labor-market trends in the population.

The results of the 2019 ACS offer estimates suggesting that there were 218,644 Massachusetts residents who self-reported to working in the construction industry.⁶ But this is only the starting point of the analysis, as a number of adjustments must be made in order to fully reflect total construction employment in the Commonwealth; these adjustments are presented in Table A1. First, the ACS only queries individuals about their primary job, meaning that this number ignores anyone whose second job is in the construction industry. Fortunately, another “gold standard” labor survey—the Current Population Survey (CPS), a monthly survey administered by the Census and the Bureau of Labor Statistics—does ask respondents about their second job. On a national basis, the CPS suggests that including workers’ second jobs would increase total construction employment by 1.99%; this equates to an additional 4,356 construction jobs in Massachusetts.⁷

Table A1. Estimating Total Construction Employment in Massachusetts, 2019

Massachusetts Residents
Total Construction Employment (2019 ACS)		218,644
Number of Second Jobs in Construction (estimate)		4,356
Inflow & Outflow of Workers (2019 ACS)
Residents from Other States Working in MA	16,264
MA Residents Working in Other States	8,012
Net Inflow/Outflow of Workers into/out of MA		8,252
Unauthorized Immigrant Underreporting
MA Immigrant Population (2019 ACS)	1,190,192
ACS Estimated Undercount (Pew)	2.00%
% of MA Non-Citizens in Construction (2019 ACS)	6.09%
Adjustment for Immigrant Undercounting		1,450
Total Construction Employment in Massachusetts		232,702

 

A second adjustment occurs because this study’s focus is on the construction industry in Massachusetts. The results presented in the American Community Survey are based on workers’ state of residence and not their state of employment. As a result, the initial total of industry employment will include Massachusetts residents working in other states while ignoring out-of-state residents who are working in the Commonwealth. Fortunately, the ACS microdata has information on workers’ state of employment, allowing researchers to capture the net flow workers into or out of a state by industry. An analysis of the 2019 ACS microdata revealed that an estimated 16,264 construction workers in Massachusetts are out-of-residents. Meanwhile, the results indicate that there were 8,012 Massachusetts residents who are working in construction jobs in other states. On net, that means that total construction employment in Massachusetts needs to be increased by 8,252 to account for the net inflow of out-of-state residents.

inally, while the ACS is considered the “gold standard” among labor economists, it is not perfect. The Census does an extensive amount of work trying to ensure that the ACS is nationally representative of the population, including authorized and unauthorized immigrants. However, research by the Pew Research Center has suggested that these surveys have been historically undercounting the number of unauthorized immigrants in a region; their most recent estimates suggest that the immigrant population is 2% to 3% higher after adjusting for this undercount.⁸ Factoring this into the analysis is particularly important in construction given the high degree of immigrant laborers in the sector.⁹ Data from the ACS reveals an initial immigration population estimate of 1,190,192; conservatively projecting a 2% undercount and multiplying by 6.09%—the proportion of foreign-born, noncitizens who work in construction in Massachusetts as offered in the 2019 ACS microdata— this equates to an estimated 1,450 additional construction workers who were previously uncounted in the data.¹⁰

After incorporating all of these adjustments, Table A1 projects that total construction employment in Massachusetts amounted to 232,702 jobs in 2019. But as a reminder, not all of these jobs necessarily represented legal employment relationships. To determine how many these were legal wage-and-salary jobs, this study incorporates data from employer payroll records submitted to the Massachusetts Department of Unemployment Assistance (DUA). As part of the required federal oversight of state unemployment insurance (UI) programs, the states provide the U.S. Department of Labor with its UI records; the DOL aggregates payroll records and publishes industry totals by state via its Quarterly Census of Employment and Wages.¹¹ In Massachusetts, this information is also augmented to include public-sector jobs and is published directly by the DUA through the state government’s Labor Market Information portal.¹²

To those ends, the Massachusetts Department of Unemployment Assistance reports that there were 174,489 legal wage-and-salary jobs among Massachusetts construction employers in 2019. But this does not account for all legal wage-and-salary jobs in the construction industry, as that total must be augmented by construction jobs provided by temporary staffing agencies.¹³ Data from the Bureau of Labor Statistics reveal that Massachusetts employment agencies had 840 construction workers on their payroll as of May 2019.¹⁴ Adding this to the payroll total, this means that there were an estimated 175,329 legal wage-and-salary jobs in the Massachusetts construction industry in 2019.

As expected, the estimate for total construction jobs (232,702) in the Commonwealth far exceeds those on employers’ official payrolls (175,329); in sum, there appear to be 57,373 construction jobs unaccounted for on employers’ payrolls.¹⁵ This differential, however, includes both legitimate self-employed contractors and workers engaged in fraudulent employment relationships. As highlighted earlier, there is no clear consensus in the research on how to best divide these 57,373 jobs into legal and fraudulent categories. While a number of studies have offered different approaches—the pros and cons of which are presented in the 2020 ICERES study—the overarching concern of many of these previously-explored methods is that substantially undercount the extent of illegal employment in the industry. As a result, this report utilizes the methodology developed in the 2020 ICERES report: the use of estimated income underreporting rates by self-employed construction workers as published in reports by the Internal Revenue Service and in academic journal articles featuring IRS-sponsored research.

The authors of the 2020 ICERES study contended that income underreporting rates represent the best available proxy for the degree of illegality among self-employed construction workers. While that study provides more background and justification, the logic is that worker misclassification and off-the-books arrangements are, for the most part, efforts on the part of employers to conceal payments to workers and evade taxes due to the government. To be clear, the decision to report—or not report—income on tax returns is the responsibility of the worker. But employers who rely on cash-only payments—without tax documentation—effectively open the door for income underreporting.

The methodology developed in the 2020 ICERES study purports that between 38.6% and 64.0% of the gap between total employment and legal wage-and-salary employment represents the number of jobs affected by payroll fraud. These proposed minimum and maximum rates were established following the authors’ analyses of IRS-sponsored research on income underreporting, and were validated against the limited direct evidence of payroll fraud available to researchers (e.g., UI audit studies).¹⁶ Applying these thresholds to the 57,373 jobs identified in this study, this suggests that there were between 22,146 and 36,719 construction jobs in Massachusetts that were structured fraudulently in 2019, either through the misclassification of independent contractors or via cash-only work arrangements.¹⁷

Validating the Results

When compared to total employment in Massachusetts’ construction industry in 2019, these results indicate that between 9.5% and 15.8% of the construction workforce in Massachusetts is affected by wage and tax fraud. These results are consistent with other data points on worker misclassification, and may actually be conservative. First, industry-wide results are comparable to studies using similar empirical methodologies in other states, including New Jersey (16%), California (16%) and Tennessee (11%-21%); they are also consistent with the national estimates (12.4%-20.5%) offered in the 2020 ICERES study.¹⁸ Second, the ACS indicates that 17.4% of construction workers in the Commonwealth are “self-employed and not incorporated”; this group is likely to predominantly include misclassified independent contractors and off-the-books workers.

Another way to consider the viability of these estimates is to consider that while there were 57,373 construction jobs unaccounted for in Massachusetts, state data reflects that there were 21,389 construction employers in the Commonwealth in 2019. If the principal of each company is considered to be legally self-employed, then 35,984 jobs remain unaccounted for; this value is nearly identical to the maximum number expressed above. Given that the industry also has a number of legitimate self-employed tradespeople and sole proprietors, the established range—22,146 to 36,719 workers affected by wage and tax fraud—seems consistent with expectations.

A fourth set of data points on payroll fraud comes from surveys of workers on construction sites; these studies are often conducted in areas known to be rife with payroll fraud and their methodologies typically suffer from concerns about sample size and the representativeness of the sample of workers surveyed. Their results are nevertheless stunning. In a series of studies undertaken in Texas and in six Southern cities in the last 15 years, rates of illegality in workers’ employment relationships ranged between 32% and 41%.¹⁹ More recently, a forthcoming study to be published by the Catholic Labor Network surveyed 79 workers at 24 commercial construction sites in Washington, D.C., and discovered nearly half (47%) were a part of the underground construction economy.²⁰ Taken together, the triangulation of these data points suggests that estimated rates of worker misclassification between 9.5% and 15.8% in Massachusetts may not only be reasonable, but may in fact be conservative.

As a final means of validating the results in this section, results of DUA audits in the main part of this study identified between 11,593 and 13,496 workers misclassified among Massachusetts construction firms in 2017-19. While these results are lower than offered by the indirect method (22,146 to 36,719 workers), this is not unexpected for reasons outlined in the main part of this report.²¹ First, DUA audits do not include labor brokers, contractors without employees, and any other employers who are not in the DUA database. Second, DUA audits are disproportionately completed among the industry’s largest employers. As highlighted in the main text, small employers are more difficult to locate and often do whatever possible to evade DUA auditors—likely in efforts to conceal illegality—meaning that the DUA estimates above are likely undercounting the number of workers affected by payroll fraud. Finally, conversations with DUA representatives note that identifying misclassified independent contractors is typically substantially easier than detecting off-thebooks employment, meaning that an unknown number of workers engaged in cash-only jobs are likely to go undetected in the audit process. Considering that some researchers have suggested that off-the-books employment is far more expansive than the number of misclassified independent contractors using 1099-MISCs—a conclusion that has been echoed by the authors’ many conversations with industry stakeholders in Massachusetts and across the country—it is not surprising that there is a substantial gap between the total number of workers identified by the DUA and the results of the indirect method.²²

Before concluding, this study has been primarily focused on wage and tax fraud amongst blue-collar tradespeople. However, it is important to acknowledge that the estimated rates of 9.5% and 15.8% are established as a proportion of all employment in the construction industry.²³ But sector employment is comprised of both blue-collar tradespeople and white-collar office staff (e.g., salespeople, clericals, engineers). While it is undoubted that payroll fraud occurs with both sets of employees, it is likely that illegality is far more prevalent in the employment of blue-collar workers. As such, these industry-wide rates are likely to appear to undercount the prevalence of fraud among tradespeople.

Given the lack of data comparing blue-collar vs. white-collar payroll fraud in the construction sector, this study proposes two empirical steps to develop a rough estimate of the proportion of payroll fraud among tradespeople. First, the Bureau of Labor Statistics suggests that 25.75% of construction employment in Massachusetts in 2019 came from non-production occupations (i.e., non-tradespeople).²⁴ Applying this rate to the BEA estimate of total employment (175,329), this implies that there were 45,141 legal wage-and-salary white-collar employees in the sector; for calculating a blue-collar rate of illegality, these jobs are removed from consideration (both the numerator and denominator). Second, in the absence of data to guide the process, the authors assume that 95% of the difference between total employment and legal wage-and-salary employment is comprised of blue-collar tradespeople (i.e., 19 off-the-books tradespeople for every one off-the-books white-collar staff member). Putting this all together, the resulting rate of wage and tax fraud for blue-collar workers in the construction industry would rise to an estimated 11.3% to 18.8%.²⁵

Discussion

The authors advocate that the empirical approach described above is the best available methodology to indirectly estimate the incidence of payroll fraud in the construction industry. But this method is far from perfect. Even the 2020 ICERES study that first developed this approach deemed it a “blunt instrument.” None of this is unexpected; after all, it is reminded that the task at hand was to estimate the scope of underground activity for which direct evidence is expressly hidden by both employers and employees. As a result, any study that attempts to estimate payroll fraud using only publicly-available data will feature a considerable margin of error. This report is no different.

All that said, there are a number of methodological reasons that imply that the estimated rates of payroll fraud produced by this indirect method are likely undercounting the true degree of illegality in the industry. The authors of this report acknowledge these issues, however do not make any further adjustments to the projections. This is primarily because the authors appreciate the gravity of the conclusions offered in this study—widespread illegality—and seek to adhere to conservative assumptions and projections in the face of statistical uncertainty and without direct evidence.

Of the three primary methodological reasons to suspect that this study is underestimating the degree of illegality in the industry, two were originally identified in the 2020 ICERES report while an additional factor has been brought to light since the publication of that study. Of the previously-identified reasons, it should be acknowledged that income underreporting rates are better measures of the volume of illegality rather than the number of workers involved. In fact, the authors of this study cannot rule out that all 57,373 jobs unaccounted for in the payroll records are affected by, or are the product of, wage and tax fraud, with the aggregate underreporting rates across all workers set at 38.6% to 64.0%. But not all levels of payroll fraud are the equivalent; for instance, a legally-employed carpenter who doesn’t report to the IRS the $900 in cash earnings resulting from fixing a neighbor’s roof on a random summer weekend is technically engaging in fraud. But that should arguably be counted differently than someone whose primary employment involves cash-only payments and requires weekly trips to a check-cashing operation. As a result, the authors are comfortable with the use of aggregate rates to establish estimates of the relative proportion (or volume) of illegality in the sector.

A second concern is that while income underreporting rates may offer the best estimates of illegality, it is reminded that simply paying one’s taxes does not necessarily mean that a person is working in a legal employment relationship. A worker who is misclassified as an independent contractor may do all the right things in filing their taxes and tacking on their 1099-MISC forms to their tax returns, but that does not make their situation legal. As such, the use of income underreporting rates may undercount the degree of illegality by assuming that all earnings reported to the IRS are acquired in a legitimate and entirely legal employment relationship.

Finally, a 2019 academic paper by Katherine Abraham (University of Maryland) and Ashley Amaya strongly suggests that this study’s estimates of total employment in the Massachusetts construction industry are undercounting the number of people who are actually working.²⁶ In particular, Abraham and Amaya demonstrate that many respondents to large national surveys like the American Community Survey—the foundation of the analysis in this report—fail to report a substantial amount of informal work that they do for pay. Not all of this is nefarious; for instance, a stay-at-home Dad may forget to report the four hours a week that he drives for Uber when answering the survey. The amount of informal work overlooked by survey respondents is substantial; in a 2019 study, Abraham and Ashley Amaya demonstrate that these surveys miss 21.9% of informal jobs (and 13.0% of informal work lasting more than four hours per week).

“Informal” labor is what many scholars call short-term job opportunities not covered by formal wage-and-salary employment structures; this typically includes off-the-books construction employment. As such, if the ACS does not capture all informal construction employment—as Abraham and Amaya’s work suggests—then this would imply that this approach taken in this study is undercounting total employment in the industry. As a result, this would mean that the study is subsequently undercounting the number of workers unaccounted for in payroll records and, as such, underestimating the number of people working in cash-only arrangements in the Massachusetts construction industry. This perspective was further confirmed after Abraham reviewed and publicly commented on the ICERES study when it was presented at the 2021 Labor and Employment Relations Association (LERA) Conference.²⁷

The authors of the current study acknowledge that the use of the ACS may lead to undercounting informal labor and, as a result, the number of workers affected by payroll fraud. However, the desire to maintain conservative assumptions and estimates compel the authors of the current study to not adjust its methodology. The reason for this is that while Abraham and Amaya’s work show that 21.9% of informal jobs are not reported on national surveys, this is an economy-wide number and no industry-specific figures are available. The authors of the current study suspect that rates of undercounting informal work are considerably higher among, say, Uber drivers and babysitters than they are for carpenters. As a result, applying any economy-wide estimate is likely to overrepresent the number of construction jobs unreported by respondents on the ACS. As a result, the authors have decided to forego this adjustment lest it produce overly inflated projections of construction employment and, as a result, payroll fraud.

 

---

¹ In addition to state UI audit studies, the best source of direct evidence of wage and tax fraud has resulted from surveys of workers on job-sites. The most prominent studies have been published by the Workers Defense Project, whose on-site surveys revealed rates of worker misclassification—including off-the-books workers— of 38% in Texas and 41% in six Southern Cities (Workers Defense Project, 2009, 2013). While these types of studies offer the most direct evidence of widespread illegality, there are two concerns that undermine on-site surveys as viable options for organizations and researchers interested in assessing illegality. First, one cannot be sure that the samples are representative of the broader industry labor force, meaning that the results may or may not accurately reflect the true rate of illegality in the industry. Second, engaging in surveys featuring a sufficient number of workers typically requires a substantial amount of resources that is prohibitive for some researchers and their affiliated organizations.

² Abraham, Katharine G., John Haltiwanger, Kristin Sandusky, and James R. Speltzer. 2013. “Exploring Differences in Employment Between Household and Establishment Data,” Journal of Labor Economics, 31(S1), S129-S172.

³ Bohn, Sarah, and Emily Greene Owens. 2012. “Immigration and Informal Labor,” Industrial Relations, 51(4), 845-873.

⁴ Ormiston, Russell, Dale Belman, and Mark Erlich. 2020. An Empirical Methodology to Estimate the Incidence and Costs of Payroll Fraud in the Construction Industry. Accessed at: http://iceres.org/wpcontent/uploads/2020/06/ICERES-Methodology-for-Wage-and-Tax-Fraud.pdf.

⁵ The Census conducted 2.06 million interviews while contacting 3.54 million households for the 2019 American Community Survey. For year-over-year data on the number of addresses selected and interviews conducted by the Census, see: https://www.census.gov/acs/www/methodology/sample-size-and-dataquality/sample-size/index.php.

⁶ The number of Massachusetts residents working in each industry in 2019 can be found here: https://data.census.gov/cedsci/table?t=Class%20of%20Worker%3AIndustry&g=0400000US25&tid=ACSST 1Y2019.S2407&moe=false&hidePreview=true.

⁷ The full estimate (1.99244%) of second-job holding in construction is drawn from a national perspective given that the sample size of Massachusetts construction workers in the CPS is deemed by the authors to be insufficient as a means of predicting a stable rate of second-job holding in the Commonwealth on a year-to-year basis.

⁸ For a full conversation on the undercounting of unauthorized immigrants in the ACS and CPS, see: https://www.pewresearch.org/hispanic/2018/11/27/unauthorized-immigration-estimate-methodology/

⁹ According to the research from the Pew Research Center, the construction industry features the second-highest rate of employment of unauthorized immigrant laborers, behind only agriculture. For more, see: https://www.pewresearch.org/hispanic/2018/11/27/unauthorized-immigrant-workforce-is-smaller-but-with-more-women/

¹⁰ The 6.09% employment rate comes from the authors’ analysis of 2019 ACS microdata and represents the proportion of foreign-born non-citizens living in Massachusetts who work in the construction industry. The 1,450 estimate is derived by multiplying the estimated number of immigrants in the Commonwealth (1,190,192) by the undercount rate (2%) and the proportion of similar residents in construction (6.09%). As a reminder, the 1,450 value is only the amount projected to be undercounted and thus represents a fraction of all unauthorized immigrant laborers in Massachusetts’ construction sector. Data on the number of immigrants in Massachusetts is found here: https://data.census.gov/cedsci/table?q=immigrants&g=0400000US25&tid=ACSDP1Y2019.DP02&hidePreview=true.

¹¹ To examine employer payroll records using the Bureau of Labor Statistics Quarterly Census of Employment and Wages (QCEW) program, see the QCEW Data Viewer at: https://data.bls.gov/cew/apps/data_views/data_views.htm

¹² For more, see: https://lmi.dua.eol.mass.gov/LMI/EmploymentAndWages.

¹³ This step is necessary because household surveys ask workers to identify the industry in which they make the most money; from the authors’ analysis of worker responses to that question, it is suspected that many temp workers in construction occupations identify themselves as working in the construction industry and not the employment staffing industry.

¹⁴ Data from the BLS does not offer information on the destination industry of workers employed by staffing agencies, so the authors are using the number employed in construction occupations as the best proxy for that value. There would seem to be some margin for error around this number (840), as it could be that construction industry employers could be using staffing agencies to help in their front office; this would mean that 840 is too low. On the other hand, those in construction occupations could be working in other industries, meaning that 840 could be too high. Given uncertainties surrounding this issue, the authors decided to stick with 840 as the best approximation. For more on occupational totals within the employment agency industry, see: https://www.bls.gov/oes/current/oes_research_estimates.htm.

¹⁵ Using the difference between worker surveys and payroll records to assess the number of jobs unaccounted for within a specific industry relies on the critical assumption that the industry codes (a) are 100% compatible across data sources and (b) are identified and coded correctly on surveys and payroll records. In terms of the former, these concerns are minimized—if not entirely erased—given that “construction” is a distinct industry code in both worker surveys and payroll records, producing a one-to-one connection between the industry code of the ACS (Census=770) and that used in payroll records (NAICS=23). Things are less certain for the second issue, as there is research identifying that it is not uncommon for workers’ occupations to be miscoded on surveys; we presume similar findings may occur for workers’ industries. Similarly, our conversations with DOR representatives in the course of this research revealed that employers’ industry classifications may evolve over time without the companies updating their industry code from their initial time of registration with the state; further, it is possible that construction employers may be strategic in how they classify their industry code on state and federal forms. While these issues are likely to affect the employment estimates in some way, there are no known credible assessments of the net effect in construction. As a result, we assume the net effect is zero and that the data utilized in this study are accurate, a presumption bolstered by the fact that the data sets are extracted from two government agencies—the Census Bureau and the Bureau of Labor Statistics—that represent the gold standard for large-scale data collection in the United States.

¹⁶ While the 2020 ICERES study offers considerable more detail, the minimum rate was established by the authors’ analysis of a 2016 article in the academic journal Public Budgeting & Finance that found that self-employed construction workers substantially underreported their income on tax forms when compared to their answers on national surveys; the article also showed that a substantial amount of self-employment income by construction workers was incorrectly submitted to the IRS as wage-and-salary income, further highlight potential payroll fraud in the sector. Meanwhile, the maximum rate was established via a 2016 IRS study demonstrating that 64% of self-employment—across the entire economy—was not reported on tax forms. To be clear, the 2020 ICERES study also identified possible rates between these two extremes and also explored rates less than 38.6%, however these were discarded after it was shown that the results were inconsistent with the known data points on payroll fraud from state UI audits and public policy papers that surveyed small numbers of workers. For more, see: Ormiston, Russell, Dale Belman, and Mark Erlich. 2020. “An Empirical Methodology to Estimate the Incidence and Costs of Payroll Fraud in the Construction Industry.”; Alm, James, and Brian Erard. 2016. “Using Public Information to Estimate Self-Employment Earnings of Informal Suppliers,” Public Budgeting & Finance, 36(1), 22-46; Internal Revenue Service. 2016. “Federal Tax Compliance Research: Tax Gap Estimates for Tax Years 2008-2010.” IRS Publication 1415.

¹⁷ The use of 38.6% as a minimum threshold of income underreporting is supported by data on 1099-MISC filings in Massachusetts that is discussed in Appendix B; in sum, data from the Massachusetts Department of Revenue reflects that 32% of non-employee compensation listed on 1099-MISC forms in the state’s construction industry is never reported by workers on income tax returns filed with the DOR. Given that offthe-books payments are likely to exhibit much worse rates of income tax reporting, the use of 38.6% would seem to be consistent with the concept of a baseline minimum income underreporting rate across all workers in the state’s construction sector.

¹⁸ For more, see: Cooke, Oliver, Deborah Figart, and John Froonjian. 2016. “The Underground Construction Economy in New Jersey”; Liu, Yvonne Yen, Daniel Flaming, and Patrick Burns. 2014. “Sinking Underground: The Growing Informal Economy in California Construction”; Canak, William, and Randall Adams. 2010. “Misclassified Construction Employees in Tennessee.”

¹⁹ While rates in Massachusetts are estimated to be lower than the national average, this is consistent with limited survey data showing that payroll fraud is especially rampant in the Southern part of the United States. For more, see: Workers Defense Project. 2013. “Building a Better Texas: Construction Conditions in the Lone Star State”; Workers Defense Project. 2009. “Building Austin, Building Injustice”; Theodore, Nik, Bethany Boggess, Jackie Cornejo, and Emily Timm. 2017. “Build a Better South: Construction Working Conditions in the Southern U.S.”

²⁰ For more, see; Sinai, Clayton and Ernesto Galeas. Forthcoming. “The Underground Economy and Wage Theft in Washington, D.C.,’s Commercial Construction Sector.” Catholic Labor Network.

²¹ The data that comprise the foundation of the indirect statistical method—the worker surveys from the American Community Survey (Census) and UI payroll records from the Massachusetts Department of Labor— are both based on worker employment at a specific moment in time. Meanwhile, UI audits may examine employment records over a longer time horizon, especially if there is evidence of misclassification. Especially given that some employment relationships in the construction industry are short-term, it is likely that the UI audits are identifying more cases than would be recognized in the ACS/BEA analysis.

²² As an example of a study offering a statistical case that off-the-books employment is far more extensive than the number of misclassified independent contractors, see: Liu, Yvonne Yen, Daniel Flaming, and Patrick Burns. 2014. “Sinking Underground: The Growing Informal Economy in California Construction.” https://economicrt.org/publication/sinking-underground.

²³ This assumes that 95% of workers affected by payroll fraud in the construction industry are tradespeople.

²⁴ For the occupational breakdown of the Massachusetts construction industry, see the Occupational Employment Statistics series of the BLS at: https://www.bls.gov/oes/2019/may/oes_research_estimates.htm.

²⁵ To demonstrate how the authors reached this number, start with total employment (232,702) and legal wage-and-salary employment (175,329) and subtract off the estimated number of legal white-collar jobs (45,141) from both numbers. That still leaves the same 57,373 difference identified earlier in the paper; applying the minimum and maximum rates of illegality, this leaves estimates of 22,146 and 36,719 workers affected by payroll fraud. Multiplying both numbers by the assumption that 95% of affected workers are tradespeople, this would result in estimates of 21,039 to 34,883 of blue-collar workers affected by payroll fraud in the industry. Dividing these numbers by the estimated number of all blue-collar workers in the sector—for the minimum rate, the calculation would be (21039) / (232702-45141-0.05*21039)—results in rates of payroll fraud for tradespeople that are 11.3% and 18.8%.

²⁶ For more, see: Abraham, Katherine, and Ashley Amaya. 2019. “Probing for Informal Work Activity,” Journal of Official Statistics, 35(3), 487-508.

²⁷ This perspective was documented in a PowerPoint delivered to the authors of the current report.