Prevailing Wages,Side-By-Side Bids,and School Construction Costs

Abstract

School districts in Maryland may ask contractors to submit two bids for the same construction project: one bid includes the payment of minimum prevailing wage and benefit rates, the other bid does not. Results from a fixed effects estimate of an unbalanced panel of nonunion roofing contractors indicate that the gap between the two bids decreases as the level of bid competition and accumulated contractor bid experience increases. The bid gap is also influenced by a contractor’s eagerness to win a project. Additional analysis illustrates how the sample average bid gap of 9.9 percent disappears under particular bid behaviors and outcomes.

Keywords

wages construction regulation bidding contractors

Introduction

Prevailing wage laws establish location and job-specific minimum wage and benefit rates for construction workers employed on public works projects. Azari-Rad, Philips, and Prus (2005) link the emergence of prevailing wage regulations to other progressive policies that were introduced in the latter half of the nineteenth century, further developed in the twentieth and twenty-first centuries. These policies include child labor laws, the eight-hour day, fair labor standards, collective bargaining rights, workplace safety, and so on. For example, the National Eight-Hour Day Act of 1868 tied reductions in hours worked and prevailing wages. This act required contractors on federally funded public works projects to reduce hours of work, but to maintain the current daily rate of pay in the location where the construction work was conducted. The first state prevailing wage law was introduced in Kansas in 1891. In 1890 and 1900, England and Canada followed American examples by introducing prevailing wage policies in response to concerns over sweatshop industries (England) and interest in pursuing a high-wage, high-skilled growth path (Canada).

Today, prevailing wage standards in the United States apply to construction funded by the federal government, to building activity financed by twenty-seven state governments, and by numerous municipalities.¹ Regardless of the jurisdiction, the primary purpose of modern prevailing wage laws is to prevent large government projects from distorting local compensation standards.² Large projects may attract contractors from areas where wages are lower. Competition between these low-wage, out-of-area builders and local establishments may depress area rates. The wage floor allows all contractors to compete without affecting compensation rates that are determined in local construction labor markets. While research has examined the impact of the wage policy on local economic activity, safety and training in the construction industry, the racial composition of the construction labor force, and the provision of health and retirements benefits, the public policy debate is centered on the impact of the wage floor on the cost of public construction (Duncan and Manzo 2016; Duncan and Ormiston 2017).

The preponderance of research comparing the cost of schools built with and without the payment of prevailing wages finds that construction costs are unaffected by the wage policy. This result contrasts with evidence from the comparison of side-by-side bids. These bids are submitted by the same contractor for the same project where one bid is based on the payment of prevailing wages and the other is not. Comparisons of side-by-side bids suggest that the payment of prevailing wages increase costs by as much as 25 percent. Contradictory results obtained from side-by-side bids and schools built with and without prevailing wage requirements are of more than academic interest. Evidence from side-by-side bids has been used in prevailing wage policy debates in Delaware, Maryland, and Ohio. This paper contributes to the literature and to the policy debate by examining side-by-side bids for school construction in Maryland. Findings illustrate how the request for side-by-side bids is a signal to contractors to alter bids and bidder behavior in ways that influence the gap between the two bids. Instead of isolating the effect of prevailing wage rates on construction costs, the cost “impact” obtained from side-by-side bids is influenced by incentives and bid behaviors stemming from this bid practice. As a consequence, side-by-side bids do not provide an accurate measure of the cost impact of prevailing wage regulations.

Previous Research on Prevailing Wage Laws and School Construction Costs

Research examining the cost effect of prevailing wage regulations has focused on school construction for two reasons.³ From a public policy perspective, taxpayers are particularly sensitive to policies that affect the cost of education. From a research perspective, school construction projects are relatively uniform and numerous. Many of the school studies use project-level data obtained from Dodge Data & Analytics (n.d.). This organization collects and distributes project bid information to the construction industry. Dodge reports the winning bid for a project but does not include change orders that determine final (total) project costs.⁴ Consequently, it is a uniform practice in this literature to use the winning, low bid as the measure of total construction costs. The Dodge data also contain information on project location and bid letting date that allow researchers to determine whether prevailing wage regulations apply. Other detailed project-level data include measures of project size (square feet and number of stories), whether the project is new or an addition, framing and flooring types, and so on. The researcher’s goal in using these data is to measure the cost impact of the wage policy taking into consideration (or holding constant) other project features that may influence costs.

Azari-Rad, Philips, and Prus (2002, 2003) use Dodge data to examine school construction across the United States during the 1990s and fail to find any statistically significant evidence that schools built in states with prevailing wage laws are more costly. Philips (2014) examines new school construction in Kentucky, Michigan, and Ohio when these states enacted, suspended, or repealed prevailing wage policies in the 1990s and finds that there is no statistically significant difference in average square foot costs associated with fluctuations in state-level wage policies. In an analysis of Maryland school construction, Prus (1999) finds that schools built in counties with prevailing wage requirements are no more expensive than the cost of comparable facilities built in counties that do not have the wage policy. Both of the studies by Prus and Philips utilize Dodge data. On the other hand, Vincent and Monkkonen (2010) also use Dodge data to examine school construction across the United States between 1995 and 2004, and find a statistically significant prevailing wage cost effect ranging from 8 to 13 percent. This unique finding may be due to the specification of the statistical model employed by Vincent and Monkkonen who do not include a measure of the business cycle. Azari-Rad, Philips, and Prus (2002) find that doubling a state’s unemployment rate is associated with a 21 percent reduction in school construction costs. As a consequence, if states with prevailing wage requirements also have lower unemployment rates, the prevailing wage cost estimate reported by Vincent and Monkkonen is too high. The studies by Azari-Rad, Philips, and Prus include measures of the unemployment rate and find no statistically significant prevailing wage cost impact. This illustrates how differences in the composition of statistical models may result in inconsistent results, even when the same Dodge data are used.

Several other studies use data similar to Dodge to examine the effect of the introduction of prevailing wage requirements on school construction in British Columbia. This wage policy is similar to several strong state-level policies in the United States.⁵ Bilginsoy and Philips (2000) use the CanaData (n.d.) and find that public school bid costs under the wage policy did not differ in terms of statistical significance from the bids for public schools built before the introduction of the prevailing wage requirement. Duncan, Philips, and Prus (2014) examine the effect of British Columbia’s prevailing wage standard by including a control group of private school projects. This difference-in-differences analysis indicates that before the introduction of the prevailing wage policy, the cost of building public schools was approximately 40 percent more expensive than the costs of comparable private schools. The differential between public and private school construction cost did not change after the wage policy was introduced. These authors have also used the British Columbian example to study the effect of prevailing wage laws on the productivity and efficiency of construction. They find that prior to the introduction of the wage legislation, public school projects had from 16 to 19 percent fewer square feet than comparable private structures (given the same project expenditure). This size differential did not change with the introduction of the wage policy. These results indicate that the utilization of labor and other construction inputs were not altered by prevailing wage requirements in a way that significantly affected construction output (Duncan, Philips, and Prus 2006). The authors also find an average efficiency of 94.6 percent for all school construction, regardless of prevailing wage requirements. Average efficiency fell to 86.6 percent for projects covered by the introductory stage of the wage policy. However, average efficiency increased to 99.8 percent for projects that were covered by the expansion of the policy seventeen months after its introduction. This trend in average construction efficiency indicates that the industry adjusted to the initial disruption associated with the introduction of prevailing wage requirements (Duncan, Philips, and Prus 2009). A similar adjustment is observed with respect to construction cost efficiency (Duncan, Philips, and Prus 2012). The studies examining the British Columbian wage policy utilize different sample configurations and statistical methods and uniformly fail to find evidence that prevailing wages increase construction costs.

Atalah (2013a, 2013b) uses data obtained from the Ohio School Facilities Commission (hereinafter, OSFC) to test the hypothesis that prevailing wages increase school construction costs in Ohio. These data are limited to information that identifies the school district, the square foot size of the school, participating contractors, and their bids. The advantage of these data is that there are over eight thousand bids in the OSFC data set. With this information, Atalah is able to compare bids submitted by contractors who are signatories to collective bargaining agreements with bids submitted by “open shop” contractors. While schools were exempted from Ohio’s prevailing wage law in 1997, union rates are the prevailing rates for other construction funded by the State of Ohio.⁶ Consequently, Atalah’s union-nonunion comparison is an indirect test of the impact of prevailing wage and benefit rates, omitting any other administrative costs associated with the policy.

Atalah’s first study compares low bids, adjusted for the square foot size of the school, and finds no statistically significant difference between union and nonunion contractors across most of Ohio (Atalah 2013a). The exception is the southern region of the state where adjusted bids are 51 percent lower for construction based on union rates. This difference is statistically significant below the .01 level. Atalah (2013b) compares the lowest bid costs by trade (plumbing, electrical, etc.) and union status. Results of this study indicate no statistically significant difference in adjusted bids for 72.2 percent (13/18) of the trades, regardless of union status. In 16.6 percent (3/18) of the trades, adjusted bids submitted by union contractors are significantly higher. In 11.1 percent (2/18) of the trades, bids submitted by union contractors are significantly lower. Atalah’s analysis finds that, in general, bids submitted by union contractors are no different than those submitted by nonunion contractors. However, there are cases where union bids are higher or lower than bids submitted by nonunion contractors.⁷

Keller and Hartman (2001) use project cost data provided by the Pennsylvania Department of Education, applicable prevailing wage rates, and total compensation rates from a large nonunion contractor to examine the effect of Pennsylvania’s prevailing wage requirement on school construction costs. By substituting nonunion wages for prevailing wage rates and adjusting for labor costs as a percentage of total construction costs, these authors find that prevailing wages add 2.25 percent to the cost of building public schools. A shortcoming of the method used by Keller and Hartman is that their comparison of prevailing and open shop wage rates ignores the changes in labor productivity and utilization that take place when wages change in the construction industry. For example, Blankenau and Cassou (2011) report that the use of skilled and unskilled workers in the construction industry is sensitive to wage rates. Skilled workers are defined as those with more than a high school degree while unskilled workers have less than a high school education. The elasticity of substitution between these two grades of labor is approximately 9.0. In addition, Balistreri, McDaniel, and Wong (2003) find that capital equipment replaces labor when construction wages increase, though the elasticities of substitution between capital and labor are inelastic in the short and long run. Taken together, the results of these studies indicate that labor productivity and utilization change with wage rates in the construction industry. The method used in the study by Keller and Hartman does not take these changes into consideration when calculating the cost impact of prevailing wages. As a consequence, the estimate reported in this study is too high.

This survey of the literature indicates that the preponderance of research fails to find a statistically significant prevailing wage cost effect. One reason why prevailing wages may not affect construction costs is that labor costs (wages and benefits) are typically a low percentage of total construction costs. According to data from the Economic Census of Construction, labor costs (wages and benefits) represent about 23 percent of total construction costs for the entire U.S. construction industry in 2012 (U.S. Census Bureau 2012). As a consequence, relatively minor changes in labor productivity and utilization as described by Blankenau and Cassou, and Balistreri et al. are needed to mitigate the effect of prevailing wages.

As mentioned above, one reason why researchers select school construction projects to measure the cost effect of prevailing wage laws is that these projects are relatively similar. The attraction of using side-by-side bids is that the cost impact of the wage policy can be measured when projects, contractors, and local and overall economic conditions are exactly the same. Consequently, a comparison of side-by-side bids indicating larger bids when prevailing wages are paid would appear as compelling evidence of the cost of the wage requirements. The purpose of the present study is to determine whether the evidence from side-by-side bids is as compelling as it appears. Data for side-by-side bids in Maryland include information on bid characteristics and bidder behavior that is not available from Dodge Data & Analytics or other sources.⁸ This information is particularly useful in examining how the practice of side-by-side bidding influences bid outcomes, bidder behaviors, and the “measured” prevailing wage cost impact in ways that are unrelated to the fundamental requirements of the wage policy. The effects of the side-by-side bids used in this study arise from the characteristics and provisions of the prevailing wage standard in Maryland.

Maryland’s Prevailing Wage Policy, Side-By-Side Bids, and Contractor Bid Behavior

Prevailing wage rates for construction projects receiving funding from the State of Maryland apply to all twenty-three counties and the City of Baltimore (State of Maryland, n.d.-a). Minimum rates for projects covered by Maryland’s prevailing wage regulation are determined by the following process. The prevailing wage rate is the rate paid to 50 percent or more of local workers in a detailed job classification. If fewer than 50 percent of local workers in a classification receive the same wage, the prevailing wage is the rate paid to at least 40 percent of the local workers in the classification. If fewer than 40 percent of local workers in the same job classification earn the same wage, the prevailing wage rate is the average wage, weighted by the number of workers receiving different wage rates. Between 2000 and 2014, prevailing wage requirements in Maryland applied to school construction projects with a value of at least US$500,000 and when state funding was 50 percent or more of project construction costs. As of July 1, 2014, prevailing wages are required on projects with a value of at least US$500,000 and when state funding is 25 percent or more of total construction costs (Maryland General Assembly 2016).

School districts in Maryland have the choice of opting out of prevailing wage requirements by accepting less than 25 percent in state funding or less than 50 percent prior to July 2014 (Maryland General Assembly 2017). To determine which policy option is most cost-effective, a school district may ask contractors to submit two bids for the same project, where one bid is based on the payment of prevailing wages and the other bid is not. These side-by-side bids allow a school district to determine which pay schedule and wage policy option is most advantageous by comparing the decrease in state funding with the bid-cost savings associated with avoiding the payment of prevailing wages. For example, if the lowest bid based on the payment of prevailing wages is US$1,000,000 for a current school construction project and if state funding is the minimum of 25 percent, the net cost to the school district for this project will be US$750,000, or US$1,000,000 minus that state’s contribution of 25 percent (or –US$250,000).⁹ The school district will compare this bid with the low bid if prevailing wages do not apply. If the low bid without the payment of prevailing wages is US$800,000 and if state funding drops to 10 percent, this district will opt out of wage requirement and accept less than 25 percent in state funding. In this case, the net cost to the school district if prevailing wages are not paid is US$720,000, or US$800,000 minus 10 percent in state funding (–US$80,000).

Pursuing this example further, if state funding were to drop to 5 percent if prevailing wages are not paid on this same project, it would be in the school district’s best interest to accept the bid based on prevailing wage rates. With a 5 percent state funding level, the net cost to the school district if prevailing wages are not paid is US$760,000, or US$800,000 minus 5 percent state funding (or –US$40,000). This is more than the net cost of US$750,000 to the district if prevailing wages are paid (with 25% state funding as illustrated above). Data on different state funding levels based on the payment and nonpayment of prevailing wages for the projects included in this study are not available. So, it is not possible to determine how school districts made decisions regarding the payment of prevailing wages for the projects included in this study.

Based on an examination of 266 side-by-side bids for sixty-seven separate school construction projects, the Public School Construction Program found that, on average, bids based on prevailing wage rates were 11.7 percent higher than bids without prevailing wages. This cost impact is based on the comparison of all bids including the lowest bid for projects built between January 2012 and December 2015. This gap persists when only low bids are considered. For example, for the subset of roof replacement projects, there were a total of eighty-three bids on seventeen roofing projects between 2012 and 2015. When all eighty-three bids are considered, the average difference in side-by-side bids is 9.67 percent. The gap in the lowest side-by-side bids for the seventeen projects is 9.10 percent. The result obtained from the analysis of side-by-side bids is viewed as “incontrovertible evidence” that prevailing wages increase construction costs (Public School Construction Program 2015). Side-by-side bids have been used elsewhere as evidence that prevailing wage laws increase construction costs. The Westlake City School District in Ohio required contractors to submit two bids, one subject to prevailing wage requirements and one bid exempt from the wage policy. An examination of these side-by-side bids suggests an overall construction cost savings of 5.8 percent without prevailing wages (Ohio Legislative Service Commission 2002). The side-by-side bid comparison for the construction of a new emergency medical station in Sussex County, Delaware, indicates that the prevailing wage requirement increased the low bid by about 25 percent (Cape Gazette 2016).

The evidence based on the side-by-side comparisons is at variance with earlier research of Maryland schools. As described above, Prus (1999) finds no statistically significant cost difference in schools built in counties with and without prevailing wage requirements. An important difference is that Prus examines bids on different projects that were covered or not covered by the wage policy. The side-by-side analysis examines different bids on the same project by the same contractor. This is a critical distinction that influences contractor incentives, the disparity in side-by-side bids, and the implied cost estimate of prevailing wages. When school districts request side-by-side bids, they are sending a signal to contractors that some state funding may be sacrificed if significant savings can be promised by avoiding the payment of prevailing wages. Under these circumstances, contractors, particularly nonunion contractors, have an incentive to inflate estimates on prevailing wage bids.

To illustrate, consider a project with one nonunion bidder. Without any competition, both bids, with and without the payment of prevailing wages, will be inflated. If this contractor wishes to avoid the payment of prevailing wage rates and other requirements of the policy including the submission of certified payrolls, apprenticeship registration, arranging benefits that meet prevailing standards, and other administrative responsibilities, the bid based on the payment of prevailing wages will be particularly inflated.¹⁰ Expanding this concept to a more realistic setting with multiple bidders suggests that when bid competition is low and the likelihood of winning is relatively high, the difference in side-by-side bids may be relatively large. A tacit or collusive agreement to increase disparity in side-by-side bids may be made between contractors when bid competition is low.¹¹ This type of arrangement is in the best interest of all nonunion contractors bidding on projects requesting two submissions and may be considered self-reinforcing to some extent. However, in a more competitive situation, the disparity in side-by-side bids may collapse as the likelihood of winning decreases and uncertainty over how other bidders will behave increases.

Other factors may affect the difference in side-by-side bids. For example, contractor experience with bidding on prevailing wage projects, as well as the dual-bid format, may also influence the gap in bids. Those who are new to prevailing wage projects may have greater uncertainty regarding all of the attendant requirements and regulations associated with the wage policy. As a consequence, less experienced contractors may pad these bids accordingly. As experience with this bidding format and the wage policy increases, contractors may reduce the disparity in bids that do and do not require the payment of prevailing wages. This suggests that relatively new bidders will have larger differences in side-by-side bids and that the gap between bids will decrease with accumulated bid experience.¹²

When a contractor is motivated to win a project, regardless of whether prevailing wages are required, it is likely that differences in side-by-side bids are reduced. This outcome may be observed during the peak bid season. For the counties and projects (roof replacements) examined in this study, 41 percent of all projects are open to bidding in March with 48 percent of all bids submitted during this peak month. It is likely that contractors who are very eager to win projects during the peak season submit low bids regardless of the payment of prevailing wages. Several other factors such as a backlog of unfinished work or the desire to work with a particular owner/school district may also influence a contractor’s motivation to win a project.¹³ When a nonunion contractor is not eager to win, both bids may be higher with the bid based on prevailing wages being particularly high. Under these conditions, a contractor’s bid may also be less competitive and finish with a higher ranking/place. This illustration suggests that if a contractor is highly motivated to win a bid, it is expected that the bid ranking will be lower as well as the disparity is side-by-side bids.

The policy change in 2014 that lowered the threshold for prevailing wage coverage to school projects receiving 25 percent of funding from the state may also affect the behavior of contractors and their side-by-side bids. This change made virtually all K-12 projects funded by the State of Maryland eligible for the payment of prevailing wages that exceeded the US$500,000 value threshold (Maryland General Assembly 2016). Under these conditions, nonunion contractors participating in projects requesting side-by-side bids may have responded to expanded prevailing wage coverage by inflating bids based on prevailing wages if they wished to avoid the requirements of the wage policy. This explanation suggests that the disparity in side-by-side bids will be larger after the July 1 policy change.

Side-By-Side Bid Data, Bid Differences by Contractor, and Statistical Model

Data for the study were obtained from the Public School Construction Program, Interagency Committee on School Construction, Board of Public Works, State of Maryland. From January 2012 to December 2015, the Public School Construction Program collected 266 side-by-side bids for sixty-seven school construction projects completed throughout the state. These projects largely consist of renovation work involving a variety of trades and tasks such as carpentry, concrete, demolition, drywall, electrical, flooring, heating, ventalation & air conditioning, masonry, roofing, and so on. As the projects that are included in the overall sample vary and are so different, this study is based on an analysis of roof replacement projects. Roof replacements were selected for this study due to the high degree of homogeneity of this type of work and the relatively large number of projects and bids. Over the period, there were eighty-three side-by-side bids by eighteen different contractors on seventeen roof replacement projects located in Carroll, Frederick, Howard, and Washington counties. As seventy-five of these bids were submitted by ten contractors who participated in at least two projects between 2012 and 2015, an unbalanced panel of nonunion contractors was created for the statistical analysis. Information from Roofers Local 30, United Union of Roofers, Waterproofers, and Allied Workers of Philadelphia, Pennsylvania, was used to identify roofing contractors who are signatories to collective bargaining agreements. The single union roofing contractor included in the master data file bid on only one project over the time period and is not included in the unbalanced panel. According to the information provided by the Fair Contracting Foundation, union contractors are hesitant to bid on projects requesting side-by-side bids due to the uncertainty regarding the outcome and whether prevailing wages will be paid or not.

Table 1 reports data on the lowest and highest differences in side-by-side bids for ten contractors included in the unbalanced panel. To illustrate, consider Contractor 1. In one of the bids submitted by this contractor, the difference between the prevailing wage bid and the bid without prevailing wages was as low as 5.3 percent. In another bid by this same contractor, the difference in side-by-side bids was as high as 30.1 percent. In addition to this variation within contractors, there is considerable variation between contractors. This is evidenced by the average difference in bids (see column 4 in Table 1). For example, the average bid difference for Contractor 1 is 12.7 percent, and 5.5 percent for Contractor 2. The variation between contractors is also revealed by the range in lowest and highest differences. Contractor 5 submitted at least one bid where there was no difference between the prevailing wage and nonprevailing wage bid (where the lowest bid difference is 0.0%). On the other hand, Contractor 6 had one bid where the difference was as high as 42.1 percent (see highest bid difference for Contractor 6). The averages for the seventy-five bids included in the study indicate a mean low difference in side-by-side bids of 3.9 percent, a mean high of 20.5 percent, and an overall average gap in the two bids of 10.2 percent.

Table 1.

Percent Differences in Side-By-Side Bids by Contractor for Roof Replacements, 2012-2015.

Contractor identity	Lowest bid difference	Highest bid difference	Average bid difference
Contractor 1	5.3%	30.1%	12.7%
Contractor 2	1.8%	16.7%	5.5%
Contractor 3	3.4%	33.1%	10.2%
Contractor 4	3.4%	15.4%	11.7%
Contractor 5	0.0%	5.3%	3.2%
Contractor 6	8.9%	42.1%	17.4%
Contractor 7	1.1%	5.7%	3.0%
Contractor 8	8.1%	17.7%	13.5%
Contractor 9	1.5%	26.8%	14.7%
Contractor 10	5.7%	12.5%	9.8%
Overall averages	3.9%	20.5%	10.2%

Source. Public School Construction Program, State of Maryland.

Differences in side-by-side bids may be due to the payment of prevailing wage and benefit rates when contractors plan to use the same workers and production methods on a project. The substitution of skilled for unskilled labor and capital equipment for all grades of labor that typically accompanies wage increases in the construction industry requires time or the entry of contractors with varying skilled workforces and capital intensities (Balistreri, McDaniel, and Wong 2003; Blankenau and Cassou 2011). The resulting changes in labor productivity and utilization may mitigate some of the cost effect of higher wage rates. However, in the side-by-side bid format, the contactor may face inflexibilities that prevent substitutions with increased wage rates passing directly through to bid costs. This may explain some of the difference in side-by-side bids, but the disparities reported in Table 1 are too large to entirely attribute to labor costs. Many of the “highest bid differences” reported in Table 1 are greater than labor costs for this type of construction activity. Information from the most recent Economic Census of Construction indicates that labor costs (wages and benefits) for specialty trade roofing contractors in Maryland are approximately 19.3 percent of total construction costs (U.S. Census Bureau 2012). A bid like that of Contractor 6 that is 42.1 percent higher with the payment of prevailing wages is approximately 2.2 times larger than percent labor costs for these types of projects. If the effect of prevailing wages is isolated from other factors that also influence construction costs, the impact of prevailing wages on bids should be fairly uniform from one project and bid to the next. For example, if prevailing wage rates add 10 percent to the cost of roof replacements, the side-by-side bids should uniformly vary by about 10 percent, depending on wage differences between counties and over time.

Another possible explanation for varying side-by-side bids is that, while roof replacements are relatively homogeneous, some may require sheet metal work. Without the payment of prevailing wages, a nonunion contractor would likely have a roofer with suitable experience perform this work with the same rate of pay. But, Maryland’s prevailing wage regulations, like the federal Davis-Bacon Act and most other state laws, set wage rates for workers performing specific jobs. As a consequence, under the wage policy, an employee who splits his or her time between roofing and sheet metal work must be paid the rates for each job classification accordingly. On average, the total hourly prevailing wage compensation of sheet metal workers is 27.9 percent higher than the comparable compensation for roofers (State of Maryland, n.d.-b). This substantially higher rate may appear to explain some of the bid differences reported in Table 1. However, this implication must be tempered by the fact that labor costs are a low percentage of total roofing construction costs. Even if all employees were upgraded to the sheet metal rate, it would affect a relatively small component of total costs and bids. For example, if all roofer labor costs rose by 27.9 percent to the sheet metal rate and labor costs are 19.3 percent of total costs, overall costs would increase by about 5.4 percent (27.9% × 19.3%), assuming that all else is unchanged. The variation in side-by-side bids that cannot be explained by differences in wage rates and the absence of input substitution suggests that factors other than the payment of prevailing wages have an impact on bid differences.

The unbalanced panel of seventy-five bids by nonunion roofing contractors is used to examine the impact of contractor bid behavior on differences in side-by-side bids by estimating the following one-way fixed effects model:

\begin{array}{l} % D i f f e r e n c e i n B i d s_{i t} = β_{0} + β_{1} C o n t r a c t o r_{i} + β_{2} # B i d d e r s_{i t} + β_{3} B i d H i s t o r y_{i t} + β_{4} B i d d e r R a n k_{i t} \\ β_{5} P e a k B i d M o n t h_{i t} + β_{6} 2014 P o l i c y_{i t} + β_{7} R e a l M i d p o i n t B i d_{i t} + β_{8} C o u n t y_{i t} + µ_{i t}, \end{array}

where % Difference in Bids is the difference between the prevailing wage bid and the bid without prevailing wages, divided by the bid omitting prevailing wages (× 100) for roof replacement projects submitted by contractor i in time period t. # Bidders equals the number of contractors who submitted a bid for each of the seventeen projects and measures the level of bid competition for each project. As described above, the nonunion contractors included in the sample have a particular incentive to inflate the bid based on the payment of prevailing wages to avoid the administrative requirements associated with wage policy. As contractors have the greatest ability to increase any bid when the level of competition is low, it is expected that the coefficient for the # Bidders variable (β₂) will be negative. Bid History is the accumulated bid experience of each contractor. This information is collected using the longitudinal aspect of the data set where the number of project bids submitted by each contractor is traced from 2012 through 2015.¹⁴ If nonunion contractors lower their prevailing wage bids as they become more familiar with the requirements with the wage regulations with repeated bid submissions, the coefficient for the Bid History variable (β₃) is expected to be negative. Bidder Rank is equal to the order of each bid submitted by the contractors included in the panel. This variable is a proxy for factors such as project backlogs that may influence bid levels. Contractors, in general, increase bids as their project backlogs increase. The coefficient for Bidder Rank (β₄) is expected to be positive as nonunion contractors, in particular, increase bids based on prevailing wages as backlogs increase. Peak Bid Month equals one for bids submitted in March, zero otherwise. As 41 percent of all projects are open during the peak month of March, and 48 percent of all bids are placed in this month, it is expected that the gap in side-by-side bids will decrease during this most competitive period. This implies that the coefficient for the Peak Bid Month variable (β₅) will be negative. The 2014 Policy is a binary variable equal to one if projects were open to bidding after the July 1, 2014, prevailing wage policy expansion that lowered the state funding threshold to 25 percent, zero otherwise. As nonunion contractors have an incentive to avoid the payment of prevailing wages, they may have responded to the policy change that expanded prevailing wage coverage by inflating bids based on the wage policy. It is expected that the coefficient for the 2014 Policy variable (β₆) will be positive. Because the 2014 Policy variable captures a time component, year dummy variables are not included for a two-way fixed effects estimate. As the effects described above may vary with the size of a project, the Real Midpoint Bid is added as a control. This variable is the inflation-adjusted midpoint between a contractor’s side-by-side bids and allows for the effects of the number of bidders, bid history, and so on, to be measured taking the contractor’s perceived value of the project into consideration. This variable is also a proxy for the size of the project. The typical approach is to use project square feet as a measure of project size, but this information is not available for the side-by-side data. Regardless, if nonunion contractors increase prevailing wage bids on larger projects, the coefficient for Real Midpoint Bid (β₇) is expected to be positive. County is another control variable that takes into consideration regional differences in market and economic conditions. County is a dummy variable identifying projects in Carroll, Frederick, and Howard Counties with Washington County as the reference category. µ is the error term.

Results

Summary statistics for the variables included in the model are reported in Table 2. The average difference in prevailing wage bids and bids estimated without the payment of prevailing wages submitted by nonunion roofing contractors is about 10 percent. The median is 8 percent. Across the ten contractors, the difference in side-by-side bids is as low as 0.0 percent and as high as 42 percent. The number of bidders ranges from two to eight participants per project with an average of 5.3 and a median of 5. The bid history of these contractors is traced longitudinally between 2012 and 2015, and ranges from the first bid to a high of thirteen bids with an average of 4.6 and a median of 4. It is not possible to determine bid history before 2012, so the measure used here is based on the accumulation of bid experience during the period of the study. The bid ranking of any contractor ranges from the first position to the eighth position with an average and median of about third place. Roofing projects are open to bids in six months of the year (January, February, March, April, August, and December). The peak month for bidding on roof replacement projects is March when 48 percent of the bids are placed. Twenty-five of the bids were placed after the policy change in July 2014 that reduced the state funding threshold to 25 percent of construction costs. Fifty of the bids were placed under the previous state funding threshold of 50 percent. The distribution of roof replacements across counties is uneven with 57 percent of projects located in Howard County, 21 percent in Frederick County, 16 percent in Carroll County, and 5 percent in Washington County. The inflation-adjusted midpoint between the bid based on the payment of prevailing wages and the bid omitting the wage requirement is approximately US$1.2 million. The median is about US$2 million with a maximum of over US$3 million and a minimum of approximately US$350,000.

Table 2.

Summary Statistics of Side-By-Side Contactor Bids, Roof Replacement Projects, Fiscal Year 2012-2015.

Variable	M	Median	Minimum	Maximum
% Difference in Bids	9.940 (7.704)	8.0	0.0	42.1
# Bidders	5.293 (1.514)	5.0	2.0	8.0
Bid History	4.640 (2.990)	4.0	1.0	13.0
Bidder Rank	3.107 (1.805)	3.0	1.0	8.0
Peak Bid Month (March)	0.480 (0.503)	—	0.0	1.0
2014 Policy	0.333 (0.478)	—	0.0	1.0
Carroll County	0.160 (0.369)	—	0.0	1.0
Frederick County	0.213 (0.412)	—	0.0	1.0
Howard County	0.573 (0.498)	—	0.0	1.0
Washington County	0.0533 (0.226)	—	0.0	1.0
Real Midpoint Bid	US$1,178,718 (610,602.8)	US$1,955,771	US$345,519	US$3,141,780
N	75

Source. Public School Construction Program, State of Maryland. Standard deviations in parentheses.

Regression results for the fixed effects estimate are reported in Table 3.¹⁵ Because there are a priori expectations regarding the effects of the number of bidders, bid history, contractor bid rank, peak bid month, and the 2014 policy change, the coefficients for these variables are evaluated with one-tailed tests. All other coefficients are evaluated with two-tailed tests. Results indicate that the effect of another bidder decreases the gap between bids that are and are not based on prevailing wage rates by approximately 1.6 percentage points. Findings also support the notion that as contractors gain experience with side-by-side bidding, the gap between the two bids decreases. The coefficient for Bid History reveals that the gap in side-by-side bids decreases by about 1.2 percentage points with each bid experience. The effects of bid competition and bid history are significant at the .05 level.

Table 3.

Fixed Effects Regression Results of Side-By-Side Contactor Bids (with and without Prevailing Wage Rates), Roof Replacement Projects, Fiscal Year 2012-2015.

Variable	Coefficient
# Bidders	−1.558^†† (0.737)
Bid History	−1.242^†† (0.532)
Bidder Rank	1.021^††† (0.257)
Peak Bid Month (March)	−8.005^††† (2.252)
2014 Policy	4.500^† (2.647)
Carroll County	0.625 (0.369)
Frederick County	13.948*** (2.181)
Howard County	4.775** (2.069)
Real Midpoint Bid	−0.0001 (0.0001)
Constant	18.110** (5.828)
N	75
F	211.24
R² (overall)	.423
F-test, all individual effects = 0	4.58

Dependent variable = % Difference in Bids.

Source. Public School Construction Program, State of Maryland. Standard errors corrected for heteroscedasticity in parentheses.

†

Significant at the .10 level (one-tailed test). ^††Significant at the .05 level (one-tailed test). ^†††Significant at the .01 level (one-tailed test).

Significant at the .10 level (two-tailed test). **Significant at the .05 level (two-tailed test). ***Significant at the .01 level (two-tailed test).

Model estimates also support the view that eagerness to win a project affects differences in bids. An increase in bid ranking or place increases the gap by approximately 1 percentage point while side-by-side-bids submitted during the peak month of March are closer by 8 percentage points. Both of these results are significant at the .01 level.

Differences in side-by-side bids increased by 4.5 percentage points after the expansion of the prevailing wage policy in 2014. This effect is significant at the .10 level. As the effect of the policy change is measured by comparing bids submitted before and after July 1, 2014, other factors that changed over this time period may also influence the estimated 4.5 percent increase. One possible influence is the increase in prevailing wage rates over time that would inflate bids if the wage policy applies. However, growth in prevailing wage rates for roofers/waterproofers in the four Maryland counties included in this study was relatively low over the period of the study. Between 2012 and 2015, the prevailing wage and benefit rates for this job classification increased by an average of 3.5 percent.¹⁶ This increase is substantially lower than the 9.2 percent increase in the Producer Price Index for roofing contractors over the same period (U.S. Bureau of Labor Statistics, n.d.). These data suggest that prevailing wage growth in Maryland increased proportionately less compared with overall costs for nonresidential roofing contractors. Also, given that labor costs are a low percentage of total costs for Maryland roofing contractors, the impact of the increase in prevailing wages on total costs is disproportionately low. If wages increase by 3.5 percent and labor costs are 19.3 percent of total costs, the effect of the wage increases is approximately 0.7 percent (3.5% × 19.3%).¹⁷ Consequently, the change in prevailing wage rates is insufficient to account for the 4.5 percent increase in side-by-side bids after 2014.

It is also unlikely that the mere expansion of the policy to projects receiving at least 25 percent in state funding would increase contractor costs and bids. If prevailing wages have a cost impact, it would be measured directly at the level of the project. That is, if a contractor bids on a project that requires prevailing wages and if the contractor expects increased costs as a result, the bid on that project will be higher. The policy change in 2014 would not have an across-the-board impact on project costs and bids. The impact of prevailing wages would still be measured at the project level, regardless of the change in the state funding threshold. Bid costs may increase if the expansion of the policy reduced bid competition. However, the 4.5 percent increase in side-by-side bids after July 2014 is measured with the level of bid competition held constant.¹⁸ The remaining explanation is that the increase in side-by-side bids is due to the reaction of nonunion contractors who are “promising” greater saving without the payment of prevailing wages at a time when prevailing wage coverage is expanding.

Holding all other factors constant, differences in side-by-side bids are larger in Frederick and Howard Counties compared with Washington County (by about 14 and 5 percentage points, respectively). While the impacts for these two counties are significant at the .05 level or lower, there is no statistically significant difference in bids between Carroll and Washington Counties. The estimate for Real Midpoint Bid is essentially zero in terms of magnitude and statistical significance. This finding indicates that the difference between the two bids does not vary with project value and size. The results of the F-test indicate that the null hypothesis that all coefficients equal zero is rejected at the .01 level.¹⁹ The model explains 42 percent of the total variation in side-by-side bids. The F-test implying that individual contractor effects are zero is also rejected at the .01 level.²⁰ This test result indicates that the fixed effects estimate is preferred to an ordinary least squares estimate that does not control for individual contractor effects.

The results reported in Tables 2 and 3 can be used to illustrate changes in side-by-side bids as the regression equation is solved with a given value of one variable, holding all other variables at their averages. For example, consider changes in the overall average gap in roof replacement bids of 9.9 percent (as reported in Table 2) when accumulated bid history changes from its average value of 4.64 bids to the maximum number of thirteen bids. With the thirteenth bid, the difference between bids based on the payment of prevailing wage and tenders that do not adhere to the wage policy collapses to –0.4 percent, holding all else constant. Similarly, if the number of bidders is at its maximum value of eight competitors, bid rank equals first place, and bids are submitted in the peak month of March (with all other variables held at average values), the average gap in side-by-side bids vanishes as the average falls from 9.9 to –0.6 percent. While these illustrations do not take into account the confidence intervals of the coefficients or the standard error of the estimate when solving the regression equation, these exercises illustrate the extent to which the difference in bids that are based on the payment of prevailing wages and comparable bids that do not include prevailing wages vary with changes in the bid behavior and outcomes.

Conclusion

Side-by-side bids from school construction projects have been used as evidence that prevailing wage laws are associated with increased building costs. These findings are at variance with the preponderance of the research reporting no cost effect of the policy based on comparisons of school construction projects that are and are not covered by prevailing wage requirements. This study finds that it is important to consider the incentives that side-by-side bidding creates for nonunion contractors and how these incentives influence the measured cost impact of the wage policy. Results from a sample of public school construction projects in Maryland illustrate how the gap between bids that require and do not require the payment of prevailing wages decreases as the level of bid competition and accumulated contractor bid experience increases. The disparity in side-by-side bids is also influenced by a contractor’s eagerness to win a project. Additional analysis illustrates how the average gap between the two bids of 9.9 percent disappears under particular bid behaviors and outcomes. Because the gap in side-by-side bids varies with bid and bidder characteristics, these data do not provide an accurate measure of the cost impact of the wage policy.

The data used in previous research typically do not contain information on bid and bidder characteristics. This study contributes to the literature by showing that these factors are important when examining the cost effect of prevailing wages. The results of this study are based on a sample on nonunion contractors and do not reveal information on how union contractors would respond to the side-by-side format. This study is also based on a sample of roofing contractors located in Maryland. Results may differ for other building types and regions.

Footnotes

Acknowledgements

The author is indebted to David Lever, former Director of the Public School Construction Program of the State of Maryland, for providing the bid data used in this report and for useful insights. The author would also like to thank Kimberly Glassman, Executive Director of the Foundation for Fair Contracting, for providing additional information that was also used in the report.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Author Biography

Kevin Duncan, PhD, is a professor of economics at Colorado State University–Pueblo. He has examined the effect of prevailing wage laws on construction costs and productivity, construction worker poverty and reliance on public assistance, minority employment in the construction industry, and the economic impact of the wage policy. He has provided testimony and research related to construction labor market policy to state legislatures in Colorado, Hawaii, Kentucky, New Hampshire, Ohio, and Vermont.

References

Atalah

Alan

. 2013a. “Comparison of Union and Non-union Bids on Ohio School Facilities Commission Construction Projects.” International Journal of Economics and Management Engineering 3 (1): 29-35.

Atalah

Alan

. 2013b. “Impact of Prevailing Wages on the Cost among the Various Construction Trades.” Journal of Civil Engineering and Architecture 7 (6): 670-76.

Azari-Rad

Hamid

Philips

Peter

Prus

Mark

. 2002. “Making Hay When It Rains: The Effect of Prevailing Wage Regulations, Scale Economies, Seasonal, Cyclical and Local Business Patterns Have on School Construction Costs.” Journal of Education Finance 27 (Spring): 997-1012.

Azari-Rad

Hamid

Philips

Peter

Prus

Mark

. 2003. “State Prevailing Wage Laws and School Construction Costs.” Industrial Relations 42 (3): 445-57.

Azari-Rad

Hamid

Philips

Peter

Prus

Mark

. 2005. “Introduction: Prevailing Wage Regulations and Public Policy in the Construction Industry.” In The Economics of Prevailing Wage Laws, edited by Azari-Rad

Hamid

Philips

Peter

Prus

Mark

, 3-27. Aldershot: Ashgate.

Balistreri

Edward

McDaniel

Christine

Wong

Eina Vivian

. 2003. “An Estimation of US Industry-Level Capital–Labor Substitution Elasticities: Support for Cobb–Douglas.” The North American Journal of Economics and Finance 14 (3): 343-56.

Bilginsoy

Cihan

Philips

Peter

. 2000. “Prevailing Wage Regulations and School Construction Costs: Evidence from British Columbia.” Journal of Education Finance 24 (Winter): 415-32.

Blankenau

William

Cassou

Steven

. 2011. “Industry Estimates of the Elasticity of Substitution and the Rate of Biased Technological Change between Skilled and Unskilled Labour.” Applied Economics 43 (23): 3129-42.

CanaData. n.d. http://www.canadata.com/products.html.

10.

Cape Gazette. 2016. “Sussex Council Approves New Paramedic Station.” August 29. http://www.capegazette.com/article/sussex-council-approves-new-paramedic-station/114563.

11.

City and County of Denver. 2017. “About Prevailing Wage.” Denver Auditor’s Office. https://www.denvergov.org/content/denvergov/en/denver-auditor/prevailing-wage/about-prevailing-wage.html.

12.

Department of Legislative Services. 2014. “Review of Maryland’s Procurement Policies and Structures.” http://mgaleg.maryland.gov/Pubs/BudgetFiscal/2014-Procurement-Structures-Policies-Practices.pdf.

13.

De Silva

Dakshina

Dunne

Timothy

Kosmopoulou

Georgia

. 2003. “An Examination of Entrant and Incumbent Bidding in Road Construction Auctions.” The Journal of Industrial Economics 51 (3): 295-316.

14.

Dodge Data & Analytics. n.d. http://construction.com/.

15.

Duncan

Kevin

. 2015a. “Do Federal Davis-Bacon and Disadvantaged Business Enterprise Regulations Affect Aggressive Bidding? Evidence from Highway Procurement Auction.” Journal of Public Procurement 15 (3): 291-316.

16.

Duncan

Kevin

. 2015b. “The Effect of Federal Davis-Bacon and Disadvantaged Business Enterprise Regulations on Highway Maintenance Costs.” Industrial & Labor Relations Review 68 (1): 212-37.

17.

Duncan

Kevin

Manzo

Frank

. 2016. “The Economic, Fiscal, and Social Effects of Kentucky’s Prevailing Wage Law.” https://illinoisepi.files.wordpress.com/2016/12/kentucky-report-duncan-and-manzo-2016-final.pdf.

18.

Duncan

Kevin

Ormiston

Russell

. 2017. “Prevailing Wages: What Do We Know? ” Institute for Construction Economic Research. http://iceres.org/wp-content/uploads/2014/10/prevailing-wage-review-duncan-ormiston.pdf.

19.

Duncan

Kevin

Philips

Peter

Prus

Mark

. 2006. “Prevailing Wage Legislation and Public School Construction Efficiency: A Stochastic Frontier Approach.” Construction Management and Economics 24 (6): 625-34.

20.

Duncan

Kevin

Philips

Peter

Prus

Mark

. 2009. “The Effects of Prevailing Wage Regulations on Construction Efficiency in British Columbia.” International Journal of Construction Education and Research 5 (2): 63-78.

21.

Duncan

Kevin

Philips

Peter

Prus

Mark

. 2012. “Using Stochastic Frontier Regression to Estimate the Construction Cost Efficiency of Prevailing Wage Laws.” Engineering, Construction and Architectural Management 19 (3): 320-34.

22.

Duncan

Kevin

Philips

Peter

Prus

Mark

. 2014. “Prevailing Wage Regulations and School Construction Costs: Cumulative Evidence from British Columbia.” Industrial Relations 53 (4): 593-616.

23.

Jofre-Bonet

Mireia

Pesendorfer

Martin

. 2003. “Estimation of a Dynamic Auction Game.” Econometrica 71 (5): 1443-89.

24.

Keller

Edward C.

Hartman

William T.

2001. “Prevailing Wage Rates: The Effects on School Construction Costs, Levels of Taxation, and State Reimbursements.” Journal of Education Finance 27 (2): 713-28.

25.

Kim

JaeWhan

Chang

Kuo-Liang

Philips

Peter

. 2012. “The Effect of Prevailing Wage Regulations on Contractor Bid Participation and Behavior: A Comparison of Palo Alto, California with Four Nearby Prevailing Wage Municipalities.” Industrial Relations 51 (4): 874-91.

26.

Ohio Legislative Service Commission. 2002. “S.B. 102 Report: The Effects of the Exemption of School Construction Projects from Ohio’s Prevailing Wage Law.” Staff Report No. 149, State House, Columbus, Ohio.

27.

Sheng

Philips

Peter

. 2012. “Construction Procurement Auctions: Do Entrant Bidders Employ More Aggressive Strategies than Incumbent Bidders?” Review of Industrial Organization 40 (3): 191-205.

28.

Maryland General Assembly. 2016. “Prevailing Wage Law—Applicability to the University of Maryland System, Morgan State University, and St. Mary’s College of Maryland.” Fiscal and Policy Note, First Reader, Department of Legislative Services. http://mgaleg.maryland.gov/2016RS/fnotes/bil_0000/hb0860.pdf.

29.

Maryland General Assembly. 2017. “Procurement—Prevailing Wage Rates—Applicability to Sheet Metal Fabricators.” Fiscal and Policy Note, First Reader, Department of Legislative Services. http://mgaleg.maryland.gov/2017RS/fnotes/bil_0000/hb1210.pdf.

30.

Ohio Laws and Rules. 1976. “Chapter 4115: Wages and Hours on Public Works.” http://codes.ohio.gov/orc/4115.

31.

Philips

Peter

. 2014. “Kentucky’s Prevailing Wage Law: An Economic Impact Analysis.” http://www.faircontracting.org/wp-content/uploads/2014/02/Kentucky-Report-2014-Philips.pdf.

32.

Philips

Peter

Mangum

Garth

Waitzman

Norm

Yeagle

Anne

. 1995. “Losing Ground: Lessons from the Repeal of Nine ‘Little Davis-Bacon’ Acts.” http://www.faircontracting.org/PDFs/prevailing_wages/losingground.pdf.

33.

Prus

Mark

. 1999. “Prevailing Wage Laws and School Costs: Ana Analysis for Public School Construction in Maryland and the Mid Atlantic States.” http://www.faircontracting.org/PDFs/prevailing_wages/Prevailing%20Wage%20Law%20and%20School%20Construction%20Cost%20in%20Maryland.pdf.

34.

Public School Construction Program. 2015. “The Cost of School Construction: A Comparison of the Monarch Global Academy and Conventional School Facilities.” http://www.pscp.state.md.us/Reports/Monarch%20Final%20Report%2010-28-15.pdf.

35.

State of Maryland. n.d.-a. “Compliance Frequently Asked Questions (FAQs)—Prevailing Wage for State Funded Construction Contracts.” Department of Labor, Licensing, and Regulation. https://www.dllr.state.md.us/labor/prev/prevwagefaqs.shtml.

36.

State of Maryland. n.d.-b. “Prevailing Wage Information Rates.” https://www.dllr.state.md.us/PrevWage/web/content/PWRequestRates.aspx.

37.

U.S. Bureau of Labor Statistics. n.d. “Producer Price Index by Industry: Roofing Contractors, Nonresidential Building Work.” https://research.stlouisfed.org/fred2/series/PCU23816X23816X.

38.

U.S. Census Bureau. 2012. “Construction: Geographic Area Series: Detailed Statistics for the State.” http://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?pid=ECN_2012_US_23A1&prodType=table.

39.

U.S. Department of Labor. 2017. “Dollar Threshold Amount for Contract Coverage.” https://www.dol.gov/whd/state/dollar.htm.

40.

U.S. Department of Labor. n.d.-a. “Davis-Bacon and Related Acts.” http://www.dol.gov/whd/govcontracts/dbra.htm.

41.

U.S. Department of Labor. n.d.-b. “The Davis-Bacon Act Protecting Wage Equality since 1931.” http://www.dol.gov/whd/programs/dbra/Survey/conformancefaq.htm.

42.

Vincent

Jeffrey

Monkkonen

Paavo

. 2010. “The Impact of State Regulations on the Cost of Public School Construction.” Journal of Education Finance 35 (4): 313-30.