I. Introduction

The Davis-Bacon Act—passed in 1931, seven years before the Fair Labor Standards Act’s introduction of a general minimum wage law—requires that prevailing wage rates be paid on federal (and federally-assisted) construction contracts of at least $2,000. Prevailing wages are defined as wage rates paid in the locality of the project for similar crafts and skills on comparable construction work. (Thieblot [1975, pp. 6-18] summarizes Davis-Bacon’s history.)

Local Davis-Bacon rates generally have been higher, the greater the percent of area building trades workers who are union members (Allen, 1983, pp. 717-25). Although union wage rates are not specifically required, former Secretary of Labor F. Ray Marshall and his coauthors stated: “In practice, these minima have been union rates and therefore have been useful to the building trades unions in preventing nonunion wage competition below these minima” (Marshall et al., 1976, p. 240). Indeed, 302 (57 percent) of 530 area wage determinations in effect in October 1976 were taken directly from collective bargaining agreements rather than based on wage surveys (General Accounting Office, 1979, p. 43). Thus, the usual effect of the Davis-Bacon Act has been to favor higher-paid labor and union members on federal construction projects by rendering illegal the employment of lower-wage—virtually always nonunion—labor.

Arguments for and against both Davis-Bacon and similar state and local prevailing wage laws, often called “little Davis-Bacon acts,” mirror those of the much larger minimum wage literature. Proponents have focused on the higher wages received by those employed on covered construction projects, and opponents have emphasized the negative employment effects and higher production costs induced by these increased wages. In addition, commentators have noted benefits to union over nonunion labor, and proponents have argued that Davis-Bacon protects construction workers from wage cutting by cheap labor, achieves stability in construction pay, and counters government monopsony power.

II. Effect on Wages and Costs

Researchers recognize that when construction employment is restricted to high-wage labor with relatively greater productivity, the effect on unit labor costs can be in either direction (Allen, 1987). Metzger and Goldfarb (1983, p. 266) note further that “forcing a contractor to hire more expensive labor may result in fewer units of labor being used, or in the substitution of materials of lesser quality, so that output quality need not rise”; and that the requirement to employ high-wage labor “combined with a competitive bidding process where least-cost bids win contracts, produces the real possibility of perverse output quality results.”

Much of the empirical literature studies the effect of Davis-Bacon on wage rates and project costs. Goldfarb and Morrall (1981) conclude in their survey that Davis-Bacon has raised construction costs, with the size of the estimated wage increase (including federal contractors’ requirement to pay skilled craft rates for certain tasks even when lower-skilled workers are performing them) and the attendant administrative costs varying considerably across studies. Allen (1983, pp. 711-17) provides further criticism of these cost studies, and presents new estimates with wide ranges (pp. 725-34).

III. Employment Effects

My research focused on the employment effects of the Davis-Bacon Act, with particular attention given to the differential impact on skilled workers versus laborers, and on minorities versus nonminorities. Minorities include blacks, Hispanics, American Indians, Asians and Pacific Islanders, and all others not classified by the U.S. Census as white and non-Hispanic. Skilled workers are defined as those whose detailed Census occupations are listed under construction: brickmasons and stonemasons; tile setters, hard and soft; carpet installers; carpenters; drywall installers; electricians; electrical power installers and repairers; painters, construction and maintenance; paperhangers; plasterers; plumbers; pipefitters and steamfitters; concrete and terrazzo finishers; glaziers; insulation workers; paving, surfacing, and tamping equipment operators; roofers; sheet metal duct installers; structural metal workers; earth drillers; and those working in construction trades not elsewhere classified. Laborers are those categorized as laborers, helpers, or apprentices. Supervisors were excluded from the study.

I examined the differential employment of skilled and laborer, and minority and nonminority, construction workers across U.S. metropolitan areas in 1990. My regression equations predict the employment effects of a dollar increase in the mean hourly construction wage rate. Local construction wage rates vary much more than wages in most industries because of wide geographical variation in construction unionization rates, local prevailing wage mandates in only some markets, and the varying regional impact of Davis-Bacon and state prevailing-wage floors.

These regression equations, in conjunction with an estimate that Davis-Bacon regulation is equivalent to an overall hourly wage increase of about 25 cents, yield estimates of the impact of Davis-Bacon on employment across construction worker categories. Before presenting the empirical results, I discuss the theoretical framework. Details are available in Bloch (2003).

Theoretical Effects

The proportion of construction workers who are minorities (relative to minority labor force representation) should be positively correlated with the scale of construction activity, and negatively correlated with construction wage rates. As construction activity expands, both the number of construction contractors and the quantity of labor employed within pre-existing firms should increase. Particularly in booming economies, both dimensions of construction growth require that firms generate relatively more hires from those not working and those employed in lower-paid non-construction jobs. Both groups are likely to be disproportionately minority. Minority unemployment rates have typically been double those of whites. Furthermore, the representation of minority workers is generally higher in lower-paid and less-skilled occupations. For example, the 1990 Census reveals the following minority representations in the U.S. experienced civilian labor force: managers, 13.1 percent; professionals, 15.6 percent; technicians, 19.6 percent; sales, 17.5 percent; office and clerical, 21.7 percent; craft workers, 18.5 percent; operatives, 30.3 percent; laborers, 31.9 percent; and service, 32.5 percent.

Higher wage rates have both a scale and a substitution effect. The scale effect refers to the discouraging effect of higher wages on production, and therefore on employment. The substitution effect of an increased wage reduces the employment of those workers whose wage is increased in favor of substitutable workers whose pay is not increased (or is increased by a lesser amount), as well as non-labor factors of production. The scale and substitution effects of higher wages reinforce each other in discouraging employment.

Imagine, for example, that a project can be completed in the same amount of time and at the same level of quality by either two plumbers and one helper, or one plumber and three helpers. If plumbers earn $50 per hour and helpers $15, then the one plumber-three helper combination, costing $95 per hour, is cheaper than the two plumber-one helper combination, costing $115 per hour. If a $30 minimum hourly wage is imposed, the laborer wage is now doubled but the plumber wage is unaffected. The two plumber-one helper combination at $130 per hour is now cheaper than the $140 one plumber-three helper combination. This same reversal results even if the plumber wage is raised to $55: the two plumber-one helper combination now costs $140 per hour, and the one plumber-three helper combination costs $145. In the last example, the relatively greater wage increase for helpers reduced their employment as compared with that of plumbers. Under either minimum wage scenario, the employment of plumbers has increased at the expense of the employment of helpers.

Because minorities are disproportionately represented among the relatively unskilled laborers, this shift to higher skilled labor accompanying increased wage rates should adversely affect the employment of minority construction workers. In 1990, 32.0 percent of the 1.26 million construction laborers (including helpers and apprentices), but only 22.2 percent of all 5.71 million nonsupervisory construction workers, were minorities. Blacks constituted 13.0 percent of laborers and 8.5 percent of all construction workers; Hispanics, 16.7 percent of laborers and 11.7 percent of all construction workers.

Minority employment may also be affected by unionization, apart from union-based higher wages. Minorities were traditionally viewed as less likely than whites to be union members because of: (1) less access to information about vacancies in unions in part because of fewer friends, relatives, and neighbors with union experience; (2) poorer training as youngsters and therefore lower qualifications to enter union apprenticeship programs; and (3) active discrimination on the part of union officials. Recent affirmative action efforts on the part of unions have countered these tendencies.

Indeed, aggregated data for the years 1986-1994 reveal that 23.8 percent of all, 24.0 percent of white, and 21.8 percent of minority nonsupervisory construction workers in the construction industry were union members. Part of this mild racial disparity results from the lower unionization rate for laborers, who, as noted above, have a relatively high representation of minorities. The unionization rate was 20.3 percent for laborers (19.9 percent white; 22.1 percent minority), but 25.0 percent for the remainder of nonsupervisory construction workers (25.3 percent, white; 21.7 percent, minority). In contrast, Ashenfelter (1972, p. 451) notes that, in 1967, more than half of white and about a quarter of black skilled construction workers, plus 28 percent of white and 35 percent of black laborers, were unionized. These data illustrate that the pronounced narrowing of the racial gap in construction unionization rates in the 1970s and 1980s was produced mainly by a sharp reduction in the union membership rate of skilled whites.

If individuals interested in the relatively well-paying construction jobs but not employed in construction—those employed in lower-paying jobs, the unemployed, and nonparticipants in the labor force—are disproportionately minority, then these measured unionization rates understate the racial unionization gaps based on this expanded supply of potential construction labor. Also, given the many allegations of hiring hall discrimination in referred hours of work, construction union members not employed in the building trades may be disproportionately minority (unions typically allow their nonworking members to take jobs in other industries). In any event, because only one of four construction workers is a union member, much of the potential employment effects of higher wage rates are likely to occur in the nonunion sector.

The scale and substitution effects discussed above also apply to the skill mix of construction workers. Thus, higher wages should be negatively correlated with the relative employment of laborers as opposed to skilled workers. Furthermore, increased wage rates likely reduce total employment in the construction industry. Unionization and construction employment are negatively correlated across metropolitan areas, in part because of the very strong positive correlation between unionization and wage rates. Non-wage effects of unions on construction employment may be either positive (if, for example, unions increase construction activity by lobbying successfully to replace publicly-owned structures that could have lasted years more) or negative (if the presence or growth of unions discourages construction activity).

Empirical Results

My primary regression model implies that a $1 increase in construction hourly wage rates would induce a 146,000 nationwide job loss in construction occupations—39,000 laborers and 107,000 skilled workers. This reduction in jobs represents 110,000 lost jobs for minorities (28,500 laborers; 81,500 skilled) and 36,000 lost jobs for whites (10,500 laborers; 25,500 skilled).

These empirical results, in conjunction with a conservative estimate that the impact of Davis-Bacon is equivalent to a 25 cent increase in the overall construction mean hourly wage, imply that elimination of the Davis-Bacon Act would increase the total number of construction jobs nationally by more than 36,000. This increase would reflect a disproportionate gain of 27,000 jobs for minorities (7,000 laborers; 20,000 skilled), plus a gain of 9,000 jobs for whites (3,000 laborers; 6,000 skilled). Economy-wide employment increases are lower than those for the construction trades because some additional construction jobs would represent shifts from other occupations.

Because skilled workers outnumber laborers by more than 3.5:1, laborers would benefit disproportionately from repeal. Thus, in addition to removing the inefficiencies induced by wage floors, repealing Davis-Bacon would increase the relative construction employment of minorities and laborers. The disproportionate gains for minorities and laborers reflect the considerable positive equity impact of Davis-Bacon repeal. A complete equity analysis must also take into account the wage increases enjoyed by minorities and laborers employed on Davis-Bacon projects. Repeal of state prevailing wage laws would be expected to produce similar effects.

These employment estimates somewhat understate the expected increase in construction jobs resulting from Davis-Bacon repeal. First, I have used a conservative estimate of the wage effect. Second, the Davis-Bacon Act affects employment in Census occupational classifications other than construction. Third, repeal of Davis-Bacon could considerably weaken unions and induce a further reduction of construction wage rates.

IV. Conclusions

Higher wage markets are associated with lower construction employment and, most striking, substantial reallocations of employment across race and skill groups, to the detriment of minorities and laborers. These wage effects imply that repealing Davis-Bacon and state prevailing wage laws would have particularly salutary effects on construction employment opportunities for minorities and laborers.

Research on this issue can be continued along at least two avenues: study of the employment effects of state prevailing wage laws, and replication of my study with the soon-to-be-forthcoming Census 2000 data. Differential effects across time and space might be correlated with the extent of discrimination, suggesting a decline in racial redistributive effects as discrimination diminishes over time, and as opportunities increase for both full-time construction employment and secondary occupations for construction workers.

Allen, Steven G. “Much Ado About Davis-Bacon: A Critical Review.” Journal of Law and Economics, October 1983, 26(3), pp. 707-36.

________. “Can Union Labor Ever Cost Less?” Quarterly Journal of Economics, May 1987, 102(2), pp. 347-73.

Ashenfelter, Orley. “Racial Discrimination and Trade Unions.” Journal of Political Economy, May-June 1972, 80(3), pp. 435-64.

Bloch, Farrell. “Minority Employment in the Construction Trades.” Journal of Labor Research, Spring 2003, 24(3), pp. 271-91.

Goldfarb, Robert S. and Morrall, John F. III. “The Davis-Bacon Act: An Appraisal of Recent Studies.” Industrial and Labor Relations Review, January 1981, 34(2), pp. 191-206.

Marshall, F. Ray, Cartter, Allan M., and King, Allan G. Labor Economics: Wages, Employment, and Trade Unionism (3rd ed.). Homewood, Ill: Richard D. Irwin, 1976.

Metzger, Michael R. and Goldfarb, Robert S. “Do Davis-Bacon Minimum Wages Raise Product Quality?” Journal of Labor Research, Summer 1983, 4(3), pp. 265-72.

Thieblot, Armand J., Jr. The Davis-Bacon Act. Philadelphia: University of Pennsylvania Press, 1975.

U.S. General Accounting Office. The Davis-Bacon Act Should Be Repealed. Washington: U.S. General Accounting Office, April 27, 1979.