Davis-Bacon Prevailing Wage Determinations Need to Be CalculatedScientifically

The Davis-Bacon Act[1] requires contractors on all federal construction projects to pay their work­ers the prevailing wage in the same locality. The law is intended to ensure that the government's buy­ing power does not drive down construction work­ers' wages.

The Wage and Hour Division (WHD) of the Department of Labor (DOL) calculates the prevailing construction wage rates in each county in the United States, and federal contractors must pay these rates. However, the WHD uses a survey methodology that is fundamentally flawed, which means that Davis- Bacon rates often bear no relation to market wages.

The current Davis-Bacon methodology uses an unscientific, self-selected sample that has high error rates, and it takes years to process and publish the results. These flaws cause the WHD to report Davis- Bacon wages that are just one-third of market wages in some counties and more than 75 percent above the prevailing market wage in other counties. Congress has spent tens of millions of dollars to fix the WHD's Davis-Bacon surveys, but the DOL Inspector General (IG) has found even higher error rates after the sur­veys were "fixed."

The government already has an agency that spe­cializes in calculating wage rates around the country. The Bureau of Labor Statistics (BLS) in the Labor Department currently publishes two accurate and timely scientific surveys that report wage rates nation­wide. Both of these surveys produce more reliable estimates than the flawed WHD surveys.

To improve the accuracy of prevailing wage esti­mates, Congress should therefore:

• Direct the WHD to stop duplicating BLS sur­veys and stop using a fundamentally flawed methodology;
• Transfer the resources currently spent on WHD surveys to the BLS to enable the BLS to expand the scope of the National Compensation Survey; and
• Require the WHD to base Davis-Bacon wages on accurate and scientifically valid surveys con­ducted by the Bureau of Labor Statistics.

These reforms would prevent flawed prevailing wage estimates from continuing to harm both work­ers and taxpayers.

The Purpose of the Davis-Bacon Act

The Davis-Bacon Act requires federal construc­tion contractors to pay at least the prevailing wage rates for non-federal construction projects located in the same areas as their federal construction projects. Supporters consider it an important mea­sure to prevent the government's buying power from distorting construction labor markets. In areas where the government is the largest buyer of con­struction services, it could use its negotiating power to lower construction wages.

To calculate the wages that contractors must pay, the Wage and Hour Division surveys construction wages and publishes prevailing wage determina­tions for each county in the United States. Federal contractors must then pay their employees at least the prevailing wage for each class of worker.

Inaccurate Rates

In spite of its purpose, the Davis-Bacon Act does not prevent the government from distorting labor markets in the construction industry, because the WHD's survey methodology reports inaccurate wage rates. Table 1 shows the WHD's prevailing wage determinations for several classes of workers in a number of cities across America and the corre­sponding market wages as determined by the BLS. In many cities, Davis-Bacon wages bear no resem­blance to market wages.

In some cities, WHD wage determinations are more than 75 percent above market wages. In other cities, they are just one-third of market wages. In some states, Davis-Bacon rates are actually below the minimum wage. WHD wage determinations simply do not reflect prevailing market wages, and this failure has serious implications for construction workers and taxpayers.

Harm to Workers and Taxpayers

The Davis-Bacon Act drives down the wages that many construction workers earn. For example,plumbers in Ft. Myers, Florida,earn $16.98 per hour, but Davis-Bacon wages are only $10.96--35 percent below the market wage.

Davis-Bacon wages are minimum wages, so this does not mean that the Davis-Bacon Act imposed a 35 percent pay cut on every plumber working for federal contractors in Ft. Myers. However, by set­ting an artificially low rate (e.g., $10.96 in Ft. Myers), the government encourages contractors to pay plumbers lower wages, and the contractors use the lower wage when submitting bids on federal construction projects. In counties where the federal government is a large buyer of construction ser­vices, its purchasing power can push down wages.

These effects depress construction wages in any county where the WHD issues prevailing wage determinations that are below market rates. In many cities, the Davis-Bacon Act has the very effect that it was intended to prevent.

In other cities, the Davis-Bacon Act has the oppo­site effect, requiring contractors--and thus taxpay­ers--to pay grossly inflated wages. In Trenton, New Jersey, the Davis-Bacon Act requires taxpayers to pay carpenters $35.72 per hour--52 percent above mar­ket wages. In these cities, the Davis-Bacon Act need­lessly inflates taxpayers' costs.

Concerns about the federal government's buying power driving down wages or about ensuring qual­ity work do not justify requiring taxpayers to over­pay for construction work. In cities where Davis- Bacon rates substantially exceed market wages, the law raises costs and increases the burden on taxpay­ers without providing any public benefit. The repeated inaccuracies in WHD prevailing wage determinations ensure that the Davis-Bacon Act harms the public good rather than serving its intended purpose of preventing the government from distorting construction workers' wages.

Flawed Methodology

Davis-Bacon wages differ from actual construc­tion wages because fundamental flaws mar the pro­cess used to determine prevailing wages. Although the Bureau of Labor Statistics is dedicated to survey­ing labor markets, the Wage and Hour Division cal­culates its own prevailing wages for the Davis- Bacon Act. Unlike the BLS, the WHD does not have expertise in surveying wages.

The Wage and Hour Division surveys construc­tion wages by sending a letter to every construction contractor that it can identify in a survey area requesting its participation. It then sends a paper or an electronic WD-10 form requesting detailed pay­roll information. The WHD conducts limited fol­low-up with contractors who do not respond, sending additional letters requesting participation in the survey. Outside analysts conduct indepen­dent verification of the data to prevent fraud. WHD wage specialists process and clarify the data and then issue final prevailing wage determinations.

This appears to be a sound method of calculating prevailing wages, but audits by the U.S. General Accounting Office (GAO)[2] and the Inspector General have found multiple flaws in the WHD's methodology that lead to unreliable wage determinations.

Scientifically Unsound Surveys.Unlike BLS surveys that estimate the unemployment rate or average wages, the Davis-Bacon survey is not a statistically random sample. The survey is self-reported, which means that only construction firms that take the time to fill out and return the forms influence the result.[3] This introduces considerable bias into the estimates.

WD-10 forms are very detailed, request data in a format that few contractors keep, and can take sev­eral hours to fill out. Many contractors receive the WD-10 form in the mail and promptly discard it. Smaller construction contractors without the staff resources to devote to government paperwork and contractors who do not do business with the federal government frequently ignore the survey and fol­low-up letters. They see no reason to fill out a sur­vey that does not affect them.

Consequently, the WHD does not base Davis- Bacon rates on a random sample of contractors. Although the survey goes out to every contractor that the WHD can identify, wage determinations are based on the self-selected minority of contrac­tors who spend the time and resources to complete the survey.

This self-selection biases Davis-Bacon rates because self-reported surveys are scientifically unsound. Unlike a random sample, self-selected sur­vey responses do not reflect the wages paid by all con­tractors. This is a fundamental and uncontroversial statistical principle. As Nobel Prize-winning econo­mist James Heckman has noted, "wage or earnings functions estimated on selected samples do not in general, estimate population wage functions."[4]

Many common examples demonstrate the unre­liability of self-selected surveys. Internet polls often show candidates winning 80 percent of the vote-- even when those candidates go on to lose the elec­tion. Participants in self-selected Internet polls rep­resent only those politically motivated people who visit that Web site, not all of the voters in an elec­tion. Often, the two bear no relation to each other.

Surveys based on self-selected samples in which contractors choose whether or not to participate are unscientific and unreliable. The IG has criticized the WHD for using this methodology and has recom­mended that the WHD use scientific random sam­ples to estimate Davis-Bacon wages.[5] Until the WHD adopts a sound survey methodology, Davis- Bacon wages will continue to be inaccurate.

High Error Rates. In addition to the unsound methodology, the surveys themselves are plagued with errors. Frequent GAO and IG audits have found continually high error rates in the survey forms submitted to the WHD. In the most recent audit, the IG found that "one or more errors existed in 100 percent of the wage reports" examined.[6] These errors included:

• Misreported wage rates. Some contractors reported one wage rate for a craft when they paid multiple rates to workers of different skill levels in that craft. Others inaccurately reported one rate for all workers.
• Benefits. Contractors often did not report some or all of the benefits that they provided to work­ers. Those who reported benefits had difficulty converting these expenses into hourly rates by occupation.
• Incorrect job classifications. Contractors re­ported apprentice and trainee wages despite instructions not to do so. Contractors included the wages of workers not on the project site. Others reported skilled workers as unskilled and vice versa.
• Wrong projects. Contractors reported wages and benefits for projects not covered in the sur­vey and classified them under the wrong con­struction types.
• Unknown reasons. Accountants often could not explain why they reported erroneous informa­tion to the WHD.

These errors occurred both because of contractor confusion about how to fill out WD-10 forms and because of sheer carelessness.[7] For example, few contractors normally calculate the hourly costs of their employees' benefits, and they made mistakes calculating them for the survey.

Although these errors are not systematically planned, they can noticeably affect final wage deter­minations. This is especially true in counties where only a few contractors return the surveys. In these counties, errors will not tend to balance out. The last audit to evaluate the effect of these errors on final wage determinations found that they led to inaccu­racies that varied from overstating wages by $1.08 per hour to understating them by $1.29 per hour.[8]

Out-of-Date Wage Rates. In addition tothe unscientific methodology and error-ridden surveys, the Wage and Hour Division takes an average of 2.3 years to process and update Davis-Bacon rates after the survey period ends. The WHD spends four-fifths of this time processing and correcting the data, not collecting information.[9] By the time the WHD publishes prevailing wage rates, they are already several years out of date.

Following publication, the WHD waits many years before updating the rates. The WHD is nowhere close to meeting its long-term goal of sur­veying construction wages in every county in Amer­ica every three years. This would require processing 10 times as many WD-10 forms as it currently pro­cesses.[10] Already out-of-date wage determinations remain effective for years before being updated. Some rates take decades to update. One wage deter­mination in North Carolina has not been updated since 1978.[11]

Changing labor market conditions and inflation mean that the workers' wages several years ago are not the same wages that they earn today. Even if the WHD used a scientifically sound and error-free methodology to estimate wages, the long delays in processing and updating Davis-Bacon rates ensure that they differ starkly from prevailing market wages.

Failed Attempts to Reform

After GAO and IG audits in the 1990s found widespread errors in WHD wage determinations, the government spent tens of millions of dollars to reengineer the wage determination process. An IG investigation after these reforms found that they were not effective.

On-site Data Verification. Since 1999, the WHD has employed an accounting firm to verify the information received. The WHD selects a sam­ple of contractors based on those submissions that have the greatest impact on the final wage determi­nations, and the firm compares payroll records to completed WD-10 forms.[12]

Redesigned Survey Forms. The WHD began sending new wage survey forms to contractors in December 2000. These forms are easier for contrac­tors to understand, provide more room to report benefit information accurately, and can easily be scanned into computer databases.[13]

Online Survey Forms. Since September 2002, contractors have been given the option of submitting surveys online instead of filling out paper copies.[14] The electronic form is simpler for contractors to complete, eliminates difficulties with unintelligible handwriting, and cannot be submitted until the con­tractor has completed all of the necessary forms.[15]

Automated Survey Database System. The WHD has created the Automated Survey Database System (ASDS), a new computer database to pro­cess surveys and issue wage determinations. It is designed to use the latest survey technologies and end the practice of wage specialists manually enter­ing wage data into the computer. The system sim­plifies analysis of wage data, highlights surveys that need clarification, and calculates prevailing wage rates.[16] The WHD began using the first phase of the system in 2002, but glitches and contractor errors have marred the installation, and the ASDS is still not fully installed.

After the WHD made these reforms, a follow-up audit by the IG found that error rates had become even higher than before[17] and that wage determina­tions still take years to process. The $22 million spent to improve wage determinations has done lit­tle to improve the accuracy of Davis-Bacon rates.

BLS Surveys Are a Better Alternative

The Wage and Hour Division estimates of con­struction wages across the United States are unreli­able and inaccurate, and attempts to improve their reliability have proven fruitless. WHD prevailing wage estimates are still error-ridden and differ starkly from prevailing market wages, despite tens of millions of dollars spent to reengineer the WHD's methodology.

This is not surprising. The WHD enforces fed­eral laws regulating wages and many working conditions, such as minimum wages, prevailing wages, child labor, overtime, and the Family and Medical Leave Act, but has no institutional exper­tise in surveying wage rates. The Bureau of Labor Statistics in the Department of Labor has that pri­mary responsibility and already conducts highly accurate wage surveys.

Congress spends over $500 million annually on the BLS. Among other surveys, the bureau conducts two nationwide wage surveys that estimate occupa­tional wages across America: the National Com­pensation Survey (NCS) and the Occupational Employment Statistics (OES). Both surveys are con­ducted in a timely manner and updated annually.

The WHD's prevailing wage surveys duplicate, albeit very inaccurately, these existing surveys. Either the OES or the NCS, with only slight modifi­cations, could be used to estimate local prevailing construction wages. This would allow the WHD to focus on enforcing the Davis-Bacon Act instead of on estimating prevailing wages.

This is how the Department of Labor enforces other prevailing wage statutes. Prevailing wages for the Foreign Labor Certification program and the Service Contract Act are calculated using OES esti­mates. It makes little sense for the WHD to spend millions of dollars annually to produce inaccurate estimates of prevailing wage rates when the BLS already produces highly accurate estimates.

Surmountable Challenges

The WHD has previously considered using BLS data to estimate prevailing wages. In 2001, it con­cluded that BLS surveys would be both more accu­rate and more timely than the current process but that difficulties in calculating employee fringe ben­efits and the geographic scope of the BLS surveys made reengineering the WHD's methodology the more attractive solution.[18] However, the IG has since found that the reengineering effort did not improve the accuracy of Davis-Bacon rates and that these difficulties could be surmounted.

Fringe Benefits Already Calculated. The chief obstacle to using BLS data is calculating hourly fringe benefit rates as required by the Davis-Bacon Act. The OES calculates highly accurate hourly wage rates but does not cover employee benefits. The NCS covers employer spending on benefits and could be used to estimate hourly wages and benefits for construction employees by occupation and type of construction. The WHD determined that the NCS does provide the information necessary to enforce the Davis-Bacon Act in the areas that it sur­veys,[19] but the NCS provides local wage informa­tion for only 154 metropolitan and non-metropolitan statistical areas. It does not cover the entire United States. Consequently, neither survey directly provides all of the prevailing wage rates necessary to enforce the Davis-Bacon Act.

This does not mean that the Department of Labor could not use scientifically reliable data to calculate prevailing wage rates. The Inspector Gen­eral determined that transferring funds from the WHD to the BLS would enable the BLS to expand the scope of the National Compensation Survey to cover local wages throughout the United States.[20]

Estimating Local Rates. The second main prob­lem with using the BLS surveys is that the BLS does not conduct these surveys on a county-by-county basis. The Davis-Bacon Act requires that contractors pay the prevailing wage "in the civil subdivision of the State in which the work is to be performed."[21] The WHD has interpreted this to mean that they cannot use BLS data. Both the OES and the NCS are statistical samples of workers in metropolitan (and non-metropolitan) statistical areas (MSAs). MSAs are calculated on the basis of commuting patterns, and most MSAs include several counties.

However, this statutory language does not present an insurmountable difficulty to using scien­tifically valid surveys. Since MSAs are calculated on the basis of work commuting patterns, occupational wages within MSAs tend to be similar. Scientifically valid surveys of MSAs would estimate prevailing wages at the county level more accurately than the current unsound method does. Additionally, the WHD does not conduct all wage surveys at the county level. Many Davis-Bacon wage rates are cal­culated for economically similar counties in the same state, not individually. The WHD has previ­ously applied wage rates from one county in a state to other noncontiguous counties in that state.[22]

If the Department of Labor has the authority to conduct multi-county wage surveys and apply rates from one county to other counties that do not share a common border, then it has the authority to use scientifically valid surveys at the MSA level.

What Congress Should Do

Congress should:

• Direct the Wage and Hour Division to stop dupli­cating Bureau of Labor Statistics surveys and stop using a fundamentally flawed methodology.
• Transfer the resources currently spent on WHD surveys to the BLS. This would enable the BLS to expand the scope of the National Compensation Survey to produce construction wage estimates that fully meet the Davis-Bacon statutory requirements.
• Require the WHD to base Davis-Bacon wages on accurate and scientifically valid BLS surveys.

Conclusion

The Wage and Hour Division's methods for calculating Davis-Bacon wages are scientifically unsound. Davis-Bacon rates are calculated using a self-selected sample instead of a statistically random sample. They take years to process and even more years to update, meaning that contractors must pay out-of-date wage rates. The survey forms confuse contractors and investigators, and the most recent audit of the WHD found errors in every wage report that it examined.

Unsurprisingly, Davis-Bacon rates bear little cor­relation to market wages. In many cities, they are below market rates, while in other cities, they are well above market rates. This hurts both workers and taxpayers. Furthermore, a decade of efforts to reengineer and improve the flawed wage determina­tion process has failed. Rather than allowing the WHD to continue duplicating the work of the Bureau of Labor Statistics, Congress should require the WHD to use BLS wage surveys to calculate pre­vailing wages.

James Sherk is Bradley Fellow in Labor Policy in the Center for Data Analysis at The Heritage Foundation.