Fraud Prevailing Wage Surveys By: Armand Thieblot, Ph.D.

Dr. Thieblot is an adjunct professor at Western Maryland College.


The Davis-Bacon Act, passed in 1931, and similar state prevailing wage laws require contractors on government-sponsored construction projects to pay their workers at least the rates prevailing for similar work in the locality where it is to be performed. The act did not specify how prevailing rates should be acquired or determined. Furthermore, no one—no legislator, official, statistician, or judge—has ever provided a logically defensible definition of the key term, prevailing. This not-inconsequential problem makes even the titles of prevailing wage laws fundamentally deceptive, since they do not ensure continuance of existing (prevailing) private-market labor rates, but rather supplant them with mandatory minimums set typically at the middle or the top of the range of existing private-market rates, sometimes even higher.* Thus, despite their name, prevailing wage laws actually set superminimum wages.

Furthermore, the state laws all differ from one another and from the federal law, and use many different methods to identify, calculate, or select the rates they call prevailing. If presented with the same distribution of actual wages paid in a locality for a particular craft on a particular type of construction, the state and federal laws could produce as many as 13 different amounts to call the prevailing rate for that craft and type of work, each as good as any other.1 Which one is best, or which, if any, is most fairly representative of the actual private labor market rate for that craft and area can’t be determined. Most prevailing wage laws ignore this problem, or simply define the prevailing rate to be whatever the administrators say it is. As a result, determinations may be issued with patently illogical prevailing rates and pass without challenge. Here are some examples of illogical rates found in recent state determinations:

• An Alaska determination identified and set prevailing rates for 95 different types of laborers who differed in title and craft affiliation but who were all unskilled. The average of prevailing rates (derived mostly from union contracts) set for these unskilled persons was 25 percent higher than the rate set for a group of highly skilled persons covered by the same determination—helicopter pilots while actively in flight. (Helicopter pilots in Alaska are typically nonunion.)

• A California determination required nonunion contractors to pay beginning carpet layer’s helpers $33.55 an hour, more than 4-1/2 times greater than the rate ($7.26) it required union contractors to pay for the same work. The union contractors got this break because the job title of Carpet Layer’s Helper is a union title that can’t be used by nonunion contractors—even though both union and nonunion carpet layers use helpers in about the same way. By prevailing rate rules, if a separate title does not exist for them, contractors must pay helpers using the tools of any trade the journeyman’s rate for that trade. Hence, nonunion carpet layer contractors for California public works must pay all their workers, including beginning helpers, the higher rate.

Results such as these have routinely gone unchallenged, partly because the typical process of challenging a determination is cumbersome, and partly because bizarre rates may result even from well-regulated administrative processes that are faithfully followed. Nevertheless, there is adequate reason to question the accuracy of individual prevailing rates and the fairness of the rate-setting process in general. In Oklahoma in 1995, a labor commissioner with previous experience in the construction industry became suspicious that the prevailing rates being administered by her department were unreasonably high.*Upon investigation, she discovered evidence of the following:

• Hundreds of bogus wage survey forms, including multiple forms on single projects, and forms from supposedly different sources filled out in the same, distinctive handwriting.

• Data submitted on wages paid for work on projects that were never constructed, specifically an underground storage tank and chemical building.

• Forms containing phantom data, such as an equal number of nonexistent ghost workers paid fictional, inflated wages added to the number of actual workers paid market rates.

• Forms reporting phantom jobs, such as one using seven asphalt lay-down machines and 21 asphalt-related workers to build a parking lot that was barely large enough to park that many machines, and besides was actually made of concrete.2

If the evidence was true, misstatements and misfilings of this sort would be clear examples of fraud in the determination of prevailing rates--fraud that had increased some individual rate determinations by as much as 162 percent.3 In July 1997, an Oklahoma jury found the first labor union official (who was indicted in a case) guilty of 14 counts of making false statements to the government.4 So the evidence was confirmed.

This was real fraud, but apparently it was not wide spread. The U.S. Department of Labor (DOL), after initially stonewalling requests to investigate the specific charges, eventually did issue revised (generally lower-rate) determinations for the contested areas of Oklahoma, while minimizing the possibility of similar malfeasance elsewhere. But under continuing pressure from Oklahoma and from congressional inquiries, two further investigations were undertaken of fraud in the survey process. One of these, a report issued by the General Accounting Office in May 1996, is difficult to comprehend and of little value.5* The other, released by the inspector general of the DOL in 1997, conducted desk and some field audits of about 10 percent of the wage surveys performed in 1995. It discovered no additional overt fraud, but did find that two-thirds of all wage survey reports subjected to onsite reviews contained "significant errors;" that 5 of the 7 audited surveys had enough errors that they had to be revised; and that 16 percent of the errors found were committed by federal DOL officials involved in the survey process. It concluded that, "If we had conducted more payroll reviews, we believe more exceptions would have been identified and would have revealed more material errors in published wage decisions."6

The inspector general attributed this astonishingly high error rate to inadvertence and inaccuracy, rather then to fraud or deception, of which the DOL was either the victim or the perpetrator. His analysis, however, was based on analysis of the arithmetical competence of DOL officials, the precision of the memories of contractors or third parties providing wage rates, and on other minutia of the survey process. It did not look at the larger question of whether the survey process, itself, might be deceptively biased towards specifying excessive rates—a form of fraud built into the process of rate determination that might be called structural fraud.

The discussion that follows, evaluating the degree of structural fraud in the wage-setting process, is based on a detailed review of a 1995 Davis-Bacon wage determination survey for building construction in Montgomery County, Maryland--one of about 3,200 counties in the U.S. entitled by the rules of Davis-Bacon to its own wage survey. I propose to show that its bizarre results could not be accidental. Then, I’ll give several reasons to be suspicious of any DOL assertions that survey fraud can be corrected through simple means or modest reform. Finally, I’ll suggest alternatives (short of outright repeal, though that possibility should not be discounted) that might lead to more fair administration. I’ll begin the discussion by looking at the outcome of federal wage determinations as a whole.

Statistical Evidence of Fraud is Provided by the Percentage of Union Rates Found in Davis-Bacon Determinations

The definition of prevailing currently found in DOL regulations says the prevailing rate is the one [to the penny] found by survey to be,

"paid to the majority (more than 50 percent) of the laborers or mechanics in the classification on similar projects in the area during the period in question. If the same wage is not paid to a majority of those employed in the classification, the ‘prevailing wage’ shall be the average of the wages paid, weighted by the total employed in the classification."7*

There’s no provision in the regulations for adopting union rates other than by their statistical predominance, and no provision for establishing "safe union territories," other than by the same statistical survey process. Thus, DOL may consider areas to be safe union territory only if a survey once showed the percentage union there to be well over 50 percent in every job category, and if there have been no major changes since.*

In 1995, the DOL acknowledged that 29 percent of all federal Davis-Bacon wage determinations were from what it considered safe union territory (i.e., all job categories had wage rates set at the union scale),8 and noted that a further 23 percent were mixed, containing some but not all union rates. Together, more than 40 percent of all individual job determinations had mandatory minimums at the union rate.

Actual union representation in 1995 was only 18.8 percent in the whole construction industry,9 which leaves an interesting question: How is it possible that over 40 percent of prevailing rate determination categories have had majority union rates when unions were less than 19 percent of the construction population?

The answer, statistically, is "not very" probable. In fact, it is all but impossible to contrive a distribution that will achieve this result. In 1995 there were about 5,135,000 construction workers, according to the Bureau of Labor Statistics (BLS). If we assume that all determinations covered the same number of individuals, and that the 40 percent (of determinations) the DOL acknowledges to be at the union rate were all selected because a scant majority (51 percent) were union, then 1,047,540 union workers would have been needed to gain majorities in the 40 percent of determinations (at the union rate). But only 963,000 were represented by unions that year, which is not enough. This estimate assumes that every union member worked at the union rate. Furthermore, if any of the "safe union territory" determinations had more than 51 percent union, the shortfall would be even greater.

Supporters of the DOL have suggested that because residential construction is known to be overwhelmingly nonunion, makes up a large proportion of the industry, and yet is only one-quarter of the wage determinations, there are enough union rates to form the necessary majorities in other determinations. This is statistically possible in some cases, but barely so. Residential construction makes up between one-third and two-fifths of all construction work.10 If this segment is only 5 percent union, then at the higher number (two-fifths), an equal distribution of union rates in the rest of the construction industry (three-fifths) can produce the necessary majorities by a small amount, but at the lower number (one-third), it cannot (if non-residential construction is two-thirds of the industry).*

There are other factors that reveal greater shortfalls. For example, those already working on public works jobs are supposed to be excluded from new surveys. Because public works jobs comprise about 20 percent of total construction (all done under federal prevailing rates), excluding such workers would eliminate 1,027,000 workers, of whom 410,800 (40 percent) would be union workers. This would leave a balance of 4,108,000 construction workers, 552,200 of them at union rate. Under these assumptions, there must be 838,000 union workers to form a bare majority in 40 percent of the determinations--285,200 more than existed. We would obtain similar disparities by taking into consideration that 80 percent or more of union work occurs in urban areas, which comprise fewer than 20 percent of the counties.

Although it is possible to conceive of situations in which unions with less than 19 percent of the population could have over 50 percent representation in 40 percent of determinations, this would depend on union relative strength being greatest in the smallest trades and in the most remote geographic areas--both of which are contrary to experience. Thus, whether union membership is evenly or randomly distributed throughout the industry, the likelihood of 40 percent of survey rates being union is so remote as to be nearly impossible. The likelihood of an artfully arranged distribution achieving the same result is not much better. From this we must conclude that something is wrong. Either those interested in overstating union presence in surveys are lying to the government, or the survey process followed by the government is skewed in favor of the union rate, or both. The first is properly called real fraud; the second, structural fraud.

Real Fraud v. Structural Fraud

Fraud is defined in Webster’s dictionary as "deception deliberately practiced in order to secure unfair or unlawful gain; trickery." It affects prevailing rate determinations not only by making union rates appear much more prevalent than they are, but also by giving the wage-setting process a veneer of sophistication and accuracy that obscures its true weaknesses. Before illustrating with examples, it is worth noting that the DOL, to some extent, is being defrauded by those who submit tainted wage rate information (e.g., Oklahoma). Nevertheless, DOL seems unconcerned about, even hostile to, the notion that it might be the victim rather than the perpetrator of the fraud. John Fraser, a Deputy Administrator of DOL, emphasized that the Oklahoma "allegations of fraud appear to be ‘isolated’ and do not appear to be ‘part of a larger problem.’"11 Also, Assistant Secretary Bernard Anderson played it down by asking, rhetorically: "If fraud was a major problem, why is it that there has been no suggestion of fraud any place other than Oklahoma City?"12

Without dwelling further on the merits of that question, I shall leave further discussion of overt fraud and misrepresentation to others, and turn to examples of structural fraud, and how the process of prevailing rate determination contributes to a pattern of deception.

Size Of The Rate-Setting Problem

The prevailing rate determination process presents an imposing problem. There are about 3,000 counties in the United States, each of which should probably have a wage determination survey every year for each of four different types of construction—that’s 12,000 determinations per year. Every determination comprises 15 to 100 job categories (sometimes more, such as in Alaska where the determination lists 125 basic rates, plus up to 5 fringes each, covering about 550 separately identified construction job titles), each of which need a statistically adequate number of survey rates drawn from private construction performed within a recent time period. In essence, a full survey process would require contacting tens of thousands of firms each year, gathering 10–20 million wage rate details, each of which would have to be checked, validated and processed.

Despite this challenge, the DOL’s Wage and Hour Division, whose most important function is establishing wage rates, spends only 10 percent of its $100 million budget for surveys and rate determinations. Yet it spends approximately 75 percent for enforcement activities. As a result of the size of the task and the limited funding devoted to it, the DOL is unable to make surveys as frequently as required. The number of full surveys conducted each year is estimated to be no more than 100 to 150, meaning each county can expect a survey once every 20 or 30 years for any type of construction.* In 1995, the year of the inspector general’s audit, the DOL completed a mere 70 survey determinations (about half the usual number) because its survey resources were tied up in the careful resurveying necessary to replace the tainted determinations in Oklahoma.*

The DOL also devotes little attention to ensuring that the surveys it now performs are comprehensive. It uses a commercial service, The Dodge Reports, to identify contractors and subcontractors who did work in the county during the previous year. The DOL sends them copies of forms asking for wages and fringes by trade classification for peak periods. Although Davis-Bacon applies to all construction jobs over $2,000, and is supposed to disregard rates from previous Davis-Bacon jobs, the Dodge Reports are limited to larger jobs, usually over $250,000, and do not differentiate private from public works.* Although the DOL contacts subcontractors identified by respondents, it also accepts data from other sources, such as unions and trade associations. However, because response is not mandatory, firms which have no interest in government jobs tend not to participate. Also, the DOL usually doesn’t do cross-tabulations of individuals, so the same wage rates from the same individuals might be reported by a contractor, a trade association, and a union, and thus be counted three times. This is significant because the DOL considers three responses sufficient to set a rate for a particular trade.

Patterns Of Questionable Process

Details of the whole survey and rate-setting process are complex, but even a limited examination reveals patterns of inattentive or misleading activities that result in unjust determinations. Some of these patterns are well known, but excused as traditional. Other patterns, which may be inadvertent, have surfaced more recently as administrators try to keep up with changing market realities. The patterns are too numerous to catalog, and even close students of prevailing rates are not familiar with them all. A few patterns that seem to have the greatest or most perverse effect are described below.

1. Exact wages and phantom ratesIt is well known that Davis-Bacon relies on exact wage amounts to establish wage determination rates. Since only unions have collective bargaining agreements applicable to all journeymen in a given craft and locality, they are the only ones likely to have any significant number of individuals paid the same rate to the penny. Thus use of exact rates in surveys creates an obvious tilt towards union rates.

This pattern persists even though an increasing number of union (collectively bargained) rates are no longer the same to the penny, and may vary by a few dollars. Consider the case of California carpenters. Carpenters covered by the same local contract in the same counties in California may be paid at more than 12 different basic hourly rates (from $16.75 to $52.86 per hour) depending on their job title and the specific type of carpentry work they do. The different job titles include: carpenter, cabinet installer, floor worker, shingler, pile driverman, heavy framer, dock carpenter, bridge carpenter, saw filer, table power saw operator, pneumatic nailer or power stapler, roof loader of shingles, and stand-by diver--all titles for which separate rates are issued in the state determination. The DOL would classify all of these as "carpenter," since they are all members of carpenters’ unions. But how their wage rate variation would be handled is unknown. That level of classification detail on wage surveys in not available to the public.

This creates a problem in prevailing rate administration that is not well known. Davis-Bacon rate surveys are supposed to uncover existing rate-paying practices, from which an administrator is to select the prevailing rate by applying a decision rule, and then require that rate to be paid on future work as listed in the determination. Actually, the rate that shows up in the determination may not have been reported at all in the survey.

A common example of how this works is provided by the Davis-Bacon determination of wage rates for Montgomery County, Maryland, issued November 15, 1995. This determination was based on a survey, conducted in October 1994, of work performed in the county between July 1993 and August 1994. Exhibit 1 compiles the survey results found for piledrivermen. (Additional information on other work performed in the county was also accepted, which explains why the dates on projects in Exhibit 1 were from outside this time range.) Notice that there is little room for confusion about wage rates. The basic rate of all 10 piledrivermen was identical, and the fringe amount was only minutely different (perhaps the result of a transcription error). The question is: Based on this survey, what rate appeared on the wage determination?

These results, though surprising, are common. Whenever union rates are found to be prevailing in a survey, but a current union contract has different (usually higher) rates, then the actual prevailing rates (as determined by the survey) are discarded in favor of the phantom amounts listed in the union contract in effect when the determination is issued. The Montgomery County survey found rates for a net of 33 crafts, setting 15 of them (some questionably) at the union rate, 17 at an average rate or at a nonunion majority rate, and 1 in an indeterminate way (apparently union, but with 0 fringes). But when the determination based on this survey was issued a few months later, only the 17 average rates were presented as calculated. All 15 of the union rates were substantially increased (by up to 16.7%) because of phantom rates.

Although the survey data show that only two of the union wage rates changed during the 14-month survey period, every union rate in the determination is said to have increased in the three months between the time the survey was compiled and the determination was issued. Furthermore, until the next full survey in Montgomery County is performed (in perhaps eight or ten years), updated determinations will be issued with additional increases in the union rates to reflect current contracts.

A further complication occurred in the survey returns for heat and frost insulation workers. The DOL went so far as to adjust the rates reported by as much as necessary so that the "union rate" could be said to prevail—even though no more than 30 percent of survey participants received the same union rate, and 5 different union rates were reported. The results are complied in Exhibit 2.

2. Wages determined for every (union) trade, but not for (nonunion) helpersWhen the Davis-Bacon Act was passed by Congress in 1931, construction workers were either laborers or mechanics in one of about a dozen trades. Laborers were workers who were not allowed to use the (journeyman) tools of a trade, so a person would not typically start out as a laborer and then advance to being a mechanic (or journeyman). If he wanted to be a mechanic, he would enter the apprenticeship program for his chosen trade, and may eventually achieve journeyman status. Unions, which dominated the industry, controlled entry into the apprenticeship programs, and thus the flow of new qualified journeymen. Semi-skilled categories were rare then, but as technology changed, the need for additional levels of skill between beginner and journeyman increased. Thus, union bricklayers added their tenders, electricians their groundsmen, plumbers their pipetradesmen, pipefitters their utility assistants, and tile setters their finishers. Some, like elevator constructors, even called their helpers, helpers. Although these semi-skilled categories are proliferating, they do not all exist in every trade and jurisdiction, though some show up in most prevailing rate determinations.

As the nonunion segment of the industry grew, it was hampered by lack of access to apprenticeship programs and consequently to a flow of new, qualified journeymen. Thus nonunion contractors turned to using on-the-job training and the semi-skilled "helper" categories as stepping stones to journeyman status; these have become standard in nonunion construction. Helpers are now among the most populous labor categories. A report prepared by the Delaware Department of Labor in 1987 found that 14 percent of all nonsupervisory construction workers in the state were classified as helpers--more than any other trade except carpenters, electricians and painters.13

Nevertheless, nonunion helpers have had great difficulty in gaining recognition by prevailing wage laws. In Delaware, for example, despite the numbers, no wage rate determinations were issued for helpers at the time the report was prepared (or since). Nor have helper rates been issued elsewhere (except briefly by the federal government in a few surveys in 1992). Because of the lack of recognition, nonunion contractors working on prevailing rate jobs must choose either not to use helpers or to pay them journeymen’s rates.

The DOL has been under a legislative requirement to supply rates for helpers for half a generation, since 1982, but has found a succession of excuses for failing to do so. The most recent, voiced by Assistant Secretary Anderson, is patently absurd:

We learned that in the great majority of cases . . . no separate and distinct occupation defined as ‘helpers’ were found to be prevailing in the local labor markets.14

That this is a lie is evident by the Montgomery County survey and determination for sprinkler fitter helpers, whose job duties are easily defined and as well circumscribed as any others in the construction industry. Survey rates for sprinkler fitter helpers are shown in Exhibit 3.

The DOL survey found rates for 14 sprinkler fitter helpers working 12 different jobs during the survey period, but the survey summary indicated that "insufficient data" were available to set a rate, so none was included when the determination was issued. The nonunion employers who had supplied 70 of the combined 84 rates for sprinkler fitters and sprinkler fitter helpers would have to forego using helpers or pay them nearly twice their customary average wage on prevailing rate jobs.

By contrast, consider the rates for piledrivermen, already seen in Exhibit 1. Only 10 rates from 2 jobs were found, but this was considered enough. Similarly, sheeting ironworkers (8 rates, 1 job), boilermakers (15 rates, 1 job), reinforcing ironworkers (18 rates, 1 job), drywall finishers (6 rates, 3 jobs), tile finishers (3 rates, 3 jobs), tile setters (3 rates, 3 jobs), pipelayers (4 rates, 4 jobs), screedmen (11 rates, 4 jobs), backhoe operators (8 rates, 6 jobs), Gradall operators (3 rates, 3 jobs), loader operators (8 rates, 8 jobs), roller operators (6 rates, 4 jobs), and elevator constructors (5 rates, 4 jobs) all had rates issued on the basis of less information in the survey than was available for sprinkler fitter helpers.*

In the Montgomery County determination, one can also find a rate for a category of plumbers rather charmingly defined as one that works on,

Apartment Buildings over 4 stories (except hotels), schools, colleges, and speculative office buildings, strip shopping centers, churches, water coolers, room air conditioning units, appliances, packaged ice machines, and light commercial refrigeration and/or air conditioning systems serving a single business in a single story building and not to exceed 5 h.p. or tons, self-contained package unit up to and including 5 h.p. or tons.

Such plumbers are to receive $14.52 per hour, whereas plumbers doing all other plumbing work are to get $21.63. Since the survey did not uncover any workers (zero) who fit the above definition, this phantom category must have been lifted from a plumber union’s collective bargaining contract and added to the determination by a gracious DOL, despite the job title’s rather startling lack of prevailing status in the local job market. It’s one of a burgeoning number of union semi-skilled job categories popping up in determinations around the country. In Montgomery County, similar categories were included for lower level steamfitters, and for a subset of boilermakers called "small boiler repair mechanic."

Finally, there is the matter of (union) elevator constructor helpers. The Montgomery County survey did suggest a prevailing rate for this union semi-skilled category despite few survey returns (5 rates from 4 jobs). Yet when the determination was issued, the elevator constructor helper category was not listed. Nevertheless, a Maryland state prevailing wage specialist suggested that elevator constructor helpers could undoubtedly continue to be paid their "customary" helper rates, even on prevailing rate jobs, since "elevator constructors always use helpers." The same specialist seemed puzzled that nonunion contractors, who are "always using" carpenters’ helpers, might also want to pay them customary rates.

3. Miscalculations or deliberate misinterpretationsOne should never ascribe to malice what can adequately be explained by simple human error or stupidity. Nevertheless, and with due respect to the DOL inspector general’s views to the contrary, some mistakes appear too deliberate to be excused. Consider, for example, the calculation of fringe benefits for average-rate portions of the Montgomery County determination. As illustrated by Exhibit 4, the fringe benefit amounts required to be paid (for three different job categories) seem to bear no resemblance to survey results.

If the fringe benefit amounts found in Montgomery County resulted from arithmetical error, then the DOL has major obstacles to overcome in properly training its staff to prepare accurate wage determinations. But if the phantom amounts resulted from deliberate, arbitrary change, it is difficult to ascribe a motive for it other than mischief.

Inexplicable sloppiness (or arbitrariness) in the determination process can produce results that are either too high or two low, and examples of both are found in the Montgomery County survey. With respect to elevator constructors, special holiday and vacation provisions in the union contract were added to the determination, even though there was no evidence that they were present in the survey or referred to by respondents. For electricians, the survey fringe benefit amount was changed to an amount plus a percentage which totaled the same. (Perhaps the percentage is meant to apply to overtime as well as to base wage, but this is not specified. No explanation is given for the change.) In two cases (structural and ornamental iron workers, and laborers), fringe benefit amounts seemed to be decreased from what the survey indicated. There is one rate (for glaziers) that has no apparent reference or justification at all. Without access to the union contracts, we could not check the rates for the 15 union categories whose wages and fringe benefits were increased; but, as mentioned earlier, all 15 were raised to phantom amounts—perhaps as found in a union contract current at the time the determination was issued.

4. Use all available information, even if suspect. The final matter to be considered in this series is the process of collecting wage rate data for determination surveys, and specifically, why the union rate has a much better chance of being accepted, regardless of the actual state of the local labor market. In surveys, Davis-Bacon administrators will take statements showing wage rates paid from any interested source, so long as they include various bits of project information. The administrators are not required to certify the validity of this information, and the information does not have to be from someone who actually paid or received the rate. Therefore business agents (whose job it is to do this sort of thing) can submit wage rate data for many employee–job combinations served by the union hiring hall without much difficulty, even though the inspector general’s report singled out third-party data providers as being especially error prone.15 On the other hand, nonunion employers generally have access to fewer such combinations, have little individual stake in the outcome (especially if they do no Davis-Bacon work), have little financial incentive, and assume some risk to submit rates at all. Thus the whole rate-gathering process is an example of structural fraud: if it is easier for union rates to be submitted, it also becomes easier for them to be accepted as prevailing.

Somewhat less well known is the degree to which prevailing rate specialists are gullible, accepting bizarre rate submissions without any proof. Here are some samples from the Montgomery County survey, not all of which may be improper, but all of which should cause suspicion of the adequacy of the rate setting and review process:

• Project #294, said to be of $30,000 value, showed 4 carpenters, 1 sheet metal worker, and 50 electricians working in its peak week—a most unusual combination.

• Project #197 reported employing simultaneously 2 union glaziers at each of the following hourly rates: $13.50, $13.75, and $14.00, all plus $1.61 in fringes. Another project, #80, reported simultaneously employing 1 union glazier at $13.00, 1 at $13.75, and 2 each at $14.00, $14.75, and $15.00, all plus fringes of $1.16 (not $1.61). These are unusual variations to be found on single jobs at union rates.

• Project #275 separately identified 68 structural ironworkers and 68 ornamental ironworkers, but both groups had the same rates of pay and fringes. These were probably the same individuals, since structural and ornamental ironworkers are commonly joined.

• Project #240 had 16 roofers paid at 13 different rates ranging from $8.00 to $16.50. Although nonunion, it is unusual to have this much wage variation and this many roofers on a single job.

• For three Sears store renovations, done successively by what appears to have been the same crew going from one job to another, there were about 20 carpenters, 2 electricians, 3 carpet layers, and 4 resilient floor layers. All were paid at consistent union rates, all likely worked for the same general contractor, and all seemed to have had their wage rates tallied successively for each assignment. Realistically, such persons were not participating in the labor market for each job and should not have been counted repetitively. (At least two rate calculations might have changed from majority to average if apparent duplication had been eliminated.)

• Project #42, said to be of $900,000 value, employed a single (one), nonunion laborer in its peak week.

• Nine jobs showed total employment of 2 to 5 sheet metal workers, and 10 jobs showed total employment limited to 1 to 6 sprinkler fitters and helpers. This probably demonstrates more about which trades participate actively in surveys than it does about wage rates paid construction workers in the county.

• Project #275 was so large ($275 million) that it dominated the survey, providing 21 percent of the total number of survey wage rates and having a controlling majority in 7 job categories--3 of them exclusively. Such a project is fundamentally different from a $200,000 school repair job or the typical building project. Davis-Bacon administration calls for different rates for different types of construction; but this job, a "resource recovery facility," was not differentiated from schools and office buildings, whose wage requirements will nevertheless be influenced by it for years.

Conclusions and Recommendations

There can be little question but that wage determinations as now performed by DOL are shot through with fraud, at least with structural fraud. They produce a proportion of union-level prevailing rates that is clearly impossible to arrive at by honest means. (Even if the DOL blames this outcome on nonunion firms’ unresponsiveness to surveys, it must still reconcile the outcome to its own mission—which is to fairly reflect local prevailing wage rates in its determinations.) Nor can there be any question that the error rate in survey determinations (as partially illustrated here and discussed in the inspector general’s report) would be unacceptable if provided to a government agency rather than by one. So long as the DOL can grade its own exam papers and "solve" its extensive problems by "trying to do better," these errors will probably remain—and probably remain hidden. Clear documentation of deliberate fraud as brought to DOL’s attention by Commissioner Reneau of Oklahoma was insufficient to motivate DOL to take its problems seriously. Thus it is unlikely that this analysis of structural fraud will elicit a change in attitude or application from the DOL—even if it were repeated a dozen or a hundred times with similar results. After all, the facts of the helper controversy have been known for decades, and the DOL has been under a legislative mandate to determine rates for helpers for about 15 years, but continues to refuse to do so.16

My principal purpose has been to expose and draw attention to the problems discussed herein, many of which are buried too deeply in arcane prevailing rate administration details to have received much notice. I hope I’ve sparked further evaluations with a view towards uncovering and correcting fraudulent practices, both overt and structural. As for solutions, the nonunion sector has traditionally sought to repeal prevailing wage laws and has had measurable success at the state level during the past dozen years (and is to be commended), but if Davis-Bacon is retained, then other recommendations that could improve the accuracy of the wage setting process are in order.

One new approach might be formulated by combining the appeals process noted in the General Accounting Office report ("Process Changes Could Raise Confidence That Wage Rates Are Based on Accurate Data") with the error rates reported by the DOL inspector general (in "Inaccurate Data Were Frequently Used in Wage Determinations Made Under the Davis-Bacon Act"). The inspector general found an overall error rate of 65 percent of the survey forms it reviewed, and five of seven determinations inaccurate enough to require replacement. In short, it found prima-facia evidence of wide-spread error (without even considering most of the things we have been discussing in this paper).

The GAO report suggests that " . . . Labor’s appeals process can serve as an additional internal control to guard against the use of fraudulent or inaccurate data in the wage determination process . . . (p. 16)." It notes that any interested party can launch an appeal and submit a formal request for reconsideration to DOL. Even so, it has proven difficult in the past to appeal rate determinations beyond internal DOL review, and even internally, interested parties could contest a wage determination successfully only by presenting "evidence demonstrating that the survey wage rates do not reflect the pattern of wages paid in a particular area" [GAO, p. 11]. Thus, previous challengers typically had to conduct and verify their own surveys. Since the inspector general’s report shows such a propensity for error in the existing DOL wage setting process, I wonder if an unbiased judge might not put the shoe on the other foot; that is, by requiring the DOL to demonstrate the validity of any new wage rates that it publishes, requires and enforces, rather than forcing challengers to prove the rates are invalid.

As they operate now, Davis-Bacon and other prevailing wage laws perpetuate fraud by government on government (and taxpayers). This adds needless billions a year to the cost of public works--according to one estimate, it is equivalent to a handout of $5,000 per year to every individual union member in public works construction. The country gets very little for this investment, and very little of it trickles down to benefit workers who might need the protection that public works projects are designed in part to provide. Rather, billions go to support a labor elite, the highest-paid workers in the construction field, whether they are working on government contracts or not. The wage rates that Davis-Bacon type determinations specify are a sham, and the government appears incapable of equitably administering them in their current form.

Exhibit 1

Survey Compilation for Piledrivermen

Montgomery County, Maryland

Project Date # Rate Fringes
207 Oct. 95 4 17.71 3.38
290 July 95 6 17.71 3.36
Totals & Averages   10 17.71 3.37
Majority rate, per survey     17.71 3.36
Rates set by determination     ? ?

Source: DOL Survey 94–MD–022


Exhibit 2

Survey Compilation

for Heat and Frost Insulation Workers

Project Date # Rate Fringes
80 Sept. 95 1 18.68 5.43
81 Dec. 94 5 20.68 5.43
83 Nov. 94 2 18.68 5.43
86 July 95 1 14.00 0.42
    1 17.25 0.42
113 Dec.94 1 18.68 5.43
184 June 94 1 18.68 5.43
    1 20.18 5.43
262 June 94 3 19.18 5.43
275 May 95 1 19.68 5.43
    2 20.47 5.43
280 Dec. 94 3 19.18 5.43
    1 20.18 5.43
285 April 94 2 18.68 5.43
286 Nov. 94 2 18.68 5.43
287 July 94 1 19.68 5.43
288 May 94 1 18.68 5.43
310 July 94 1 19.68 5.43
327 Jan. 95 1 18.68 5.43
Totals & Averages   33 19.24 5.13
Majority Rate, per survey     n/a 5.43
Average Rate, per survey     19.24 n/a
Rates set by determination     19.58 5.68

Source: DOL Survey 94–MD–022

Exhibit 3

Survey Compilation for Sprinkler Fitter Helpers

Project Date # Rate Fringes
73 Sept. 94 1 10.00 2.10
80 Sept. 95 1 9.50 0.00
123 Oct. 94 1 8.50 1.04
135   1 7.50 0.49
166 June 94 1 7.50 0.49
195 Aug. 95 1 10.00 1.11
    1 8.00 0.18
    1 11.00 1.75
212 Nov. 94 1 8.00 0.00
271 June 95 1 10.00 1.11
285 April 94 1 6.76 0.18
314 Aug. 94 1 10.00 1.75
330 Sept. 94 1 7.50 0.00
336 Nov. 94 1 8.50 1.14
Totals & Averages   14 8.77 0.79
Average rate, per survey     8.77 0.92
Rates set by determination     none none
Nonunion contrator would pay:     15.94 2.75

Source: DOL Survey 94–MD–022

Exhibit 4

Survey Compilation for Fringe Benefits for

Pipelayers, Backhoe Operators and Loader Operators

Pipelayer Backhoe Operators Loader Operators
Project # Fringes
31 1 0.00
86 1 0.00
196 1 2.40
196 1 4.16
Project # Fringes
31 1 0.00
86 1 1.24
86 1 0.00
115 1 1.64
196 2 3.68
221 1 0.00
285 1 1.24
Project # Fringes
31 1 0.00
31 1 1.24
77 1 0.00
86 1 0.00
196 1 0.00
196 1 2.51
221 1 0.00
273 1 0.00
Average (majority): 164 1.44 (0.00)
Per determination: 2.36 2.17 2.17

Source: DOL Survey 94-MD-022


{A} In the case of Davis-Bacon, the new minimum rates are the weighted average of all rates found by survey (a middle rate) unless a majority are paid the same rate to the penny, which would typically be the case only for union rates. That’s how the union rate, which is usually the maximum, can become the new minimum rate. Even more surprisingly, as will be explained herein, it is not uncommon for minimum rates set by Davis-Bacon surveys to be higher than the highest actually paid anyone covered by the survey.

{B} In Oklahoma at that time, state law required that prevailing rates for state work be adopted without adjustment from the federal Davis-Bacon rates. This requirement was tested in court shortly after the investigation began, and resulted in the Oklahoma statute being declared unconstitutional.

{C} Indicative of the lack of comprehension exhibited by this report is the breathless statement (accusation?) found on its page 12: "The use of inaccurate data could also lead to lower wages for construction workers on federal projects than would otherwise be prevailing. Industry association members and officials told us [GAO] that in several parts of the country, employers, especially nonunion contractors, paid wages on their private projects below the prevailing wage levels specified by the Davis-Bacon Act in their areas[!]" [Emphasis added.] It is small wonder that the rest of its analysis is worthless and most of its recommendations vacuous. The report does contain (in its second appendix on page 39) a fascinating example of wage compilation from a 1994 survey (94-SD-200). This wage compilation shows that the DOL apparently felt it had sufficient data to establish a prevailing rate for scraper operators based on 3 wage reports in the survey and for electricians based on 4. However, it also shows that the DOL felt it had insufficient data to establish a prevailing rate for carpenter helpers (27 in the survey, the third largest group), carpenter laborers (25 in the survey), cleanup laborers (9 in the survey), drywall tapers or drywall finishers (7 each in the survey) or roofers, insulators, or applicators (4 each in the survey). All of these survey rates were simply disregarded. This was also characteristic of the 1995 Montgomery County, Maryland, determination discussed later in the text. It is interesting to note that although nonunion helper categories are always discarded by DOL for "insufficient data," DOL’s survey form, WD-10, invites respondents to use carpenter’s helpers as a likely employee description.

{D} The Davis-Bacon law, passed in 1931, required prevailing rates but did not suggest how they might be estimated, acquired, or determined. The DOL established an administrative definition four years later, followed by procedural instructions (e.g., survey of rates to the penny, rates established by county separately for each class of worker in four classes of work) that remain in effect, largely unchanged. The only fundamental changes came in 1962, when fringe benefits were added, and in 1984, when the 30-percent plurality rule for rate selection became a 50-percent plurality rule. The details of the survey process are all arbitrary, as is the rate selection method, because "prevailing" has no statistical basis or objective meaning.

{E} There has been no substantive effort by DOL to conduct surveys to verify the "safe union territory" areas since the Davis-Bacon Act was amended in the 1980s—changing the 30-percent rule to a 50-percent rule. Not only did this rule change make it less likely that the union rate would be specified, but the amount of unionization in the workforce also declined dramatically. In the decades before the change, unionization rates exceeded 60 percent in all construction fields except home building. Since then, however, the rate of unionization in construction has fallen to below 20 percent. Yet, the DOL continues to assume that many, if not most, of the areas previously established as "safe union territory" continue as such.

{G} For those interested, the numbers look like this: Assume a construction industry composed of 5,135,000 persons, of whom 963,000 are union members. If 33 percent of the industry is residential work (as opposed to building, highway, or heavy work), and residential construction is 5 percent union, then residential work uses 1,694,550 persons, 84,727 of them union, leaving 3,440,450 persons for nonresidential work, 878,273 union. Assuming that 25 percent of all wage determinations for residential construction are purely nonunion, then 53 percent (.4/.75) of the remaining determinations would need union-rate majorities so that 40 percent of all determinations would be at union rate. If workers were equally divided among all remaining job categories, then 929,954 union workers would be needed to form a majority (51 percent) among 53 percent of 3,440,450 workers. But there are only 878,273 union workers available, so the 40 percent union-rate outcome would be statistically impossible. The corresponding numbers at 40 percent residential are 832,794 needed and 860,300 available, so the outcome is statistically possible at this level, but not by much.

{H} The Montgomery County, Maryland, survey used here replaced a previous survey that was 9 years old. Although I have not been able to verify this, I was told that some parts of the country have not had an actual survey performed since 1935.

{I} On this basis, with existing resources, it appears DOL could accomplish about 10 careful surveys a year, leaving them about 11,990 short of par.

{J} Oddly, current DOL survey forms specifically ask for wages paid on both private and prevailing rate jobs, although it is under legislative constraint to not consider those paid under Davis-Bacon.

{K} See the third footnote early in the text.