of federally financed building projects in selected New England areas disclosed that many of the rates were improperly established at the higher rates negotiated by labor organizations and building contractors rather than at the lower rates prevailing on private construction in the project areas. Also, wage rates determined for certain crafts in connection with a federally-assisted low-rent housing project in Massachusetts were on a level with the negotiated rates normally paid on commercial-type building construction rather than equal to the lower rates paid on similar private housing construction in the locality. Our review showed that these unrealistic determinations were based on inadequate information obtained by the Department on wage rates in these areas, and we believe that the Department has not complied with either its own regulations or the intent of the Davis-Bacon Act that wage determinations be based on the wage rates prevailing for similar construction in the locality. (Italics added.)R The GAO report presented several interesting findings about Davis-Bacon determinations in New England.

1. Wage determinations for power equipment operators on federally financed projects throughout Maine were found to be higher than those prevailing in Maine. The Davis-Bacon rates corresponded to union-negotiated rates in Boston, Massachusetts.

2. GAO noted several cases in which regular employees of nonunion contractors worked at or about the same time on private projects and federal projects and were paid at higher rates on the federal projects. Comparative wages provided in the GAO report showed that employees working on concurrent projects earned wages on federal projects that were from 68 to 221 percent higher than on private projects.

3. GAO found that the Department of Labor included previous Davis-Bacon rates in the information it used to determine prevailing wages on new contracts, thereby carrying any earlier errors forward. In the words of the GAO report:

The [Labor] Department's survey disclosed that the average hourly rate for laborers, while working only on private construction in 1963 was $1.92 or 33 cents an hour lower. The 1962 wage decisions showed that the minimum rate for truck drivers was the negotiated rate of $2.15, but the Department's survey disclosed that the prevailing rate for truck drivers was $1.50 an hour... or 65 cents an hour lower than the negotiated rate. The difference in interpretation of the Department's recent survey stems from the fact that we have excluded from the survey data the wage rates paid on federally financed projects which were subject to prior wage decision of the Department employing the negotiated rate.9

4. The GAO report noted that ". . . as has been shown in this report and in associated reports to the Congress, it has been the practice of the Department to determine the higher negotiated rates paid on commercial type building construction as the minimum rates for federally financed housing construction instead of the

8 Comptroller General's Report to Congress, "Wage Rates for Federally Financed Building Construction Improperly Determined in Excess of the Prevailing Rates for Similar Work in New England Areas" (January 1965).

• Ibid. p. 15.

lower rates prevailing for similar private housing construction in the project areas."10 To estimate the difference attributable to this application of incorrect rates, the General Accounting Office surveyed costs of construction of 29 apartments in Middlesex, Suffolk, and Norfolk counties, Massachusetts that were comparable to a low-rent housing project in Middlesex which carried a Davis-Bacon determination. The GAO did not choose the lowest possible rates but rather applied the Labor Department's determination rule to the survey data. The results are shown in Table 5. On the average over all the reported crafts, the Davis-Bacon determination was 33 percent greater than the GAO survey wage. It should be emphasized that the GAO figures are "prevailing" wages and are quite likely to be above the minimum wage in each craft.

Source: GAO report, op. cit., p. 19. Wage rates are reported to the lowest cent.

(c) Other GAO Studies

The findings of the Quantico study and the New England study were essentially repeated in GAO studies in 1968 and 1970.11 These studies made some other relevant points. First, GAO found that, because of the high rates determined by the Labor Department on the Davis-Bacon construction, open shop, private housing contractors were sometimes reluctant to bid on such contracts. To do so

(a) would disrupt a company's labor force because workers on government jobs would be paid greater hourly rates than those on private jobs, and

10 Ibid. p. 21.

11 Comptroller General's Reports to the Congress "Need for More Realistic Minimum Wage Rate Determinations for Certain Federally Financed Housing in Washington Metropolitan Area" (September, 1968) and "Construction Costs for Certain Federally Financed Housing Projects Increased Due to Inappropriate Minimum Wage Rate Determinations" (August 1970).

(b) would create hardship and morale problems when workers' wage rates decreased after the government job was completed and they returned to work on private construction jobs.

This reaction of nonunion contractors to the excessive Davis-Bacon rates has two undesirable consequences. It limits the competition for government contracts, thereby raising the cost of these contracts. In addition, it suggests that the Davis-Bacon contracts tend to segregate the labor market to the disadvantage of nonunion workers.

Second, GAO found the wage information file of the Department of Labor was inadequate to determine appropriate wage rates. GAO also found that the department had conducted wage surveys in 1957 and 1961 showing that nonnegotiated wage rates prevailed on private construction single-family homes in the Washington metropolitan area. The report goes on to state:

Nevertheless, most of the current wage data in the Department's files at the time of our review consisted of data from negotiated wage agreements submitted by local labor organizations. We did not find in the files adequate current data showing the specific construction projects on which the negotiated wage rates prevailed, the specific construction projects on which nonnegotiated rates prevailed, or the number of workers being paid negotiated and nonnegotiated wage rates.12

Third, GAO also noted that the department's determinations displayed a significant inconsistency. The department determined different classifications and minimum rates for similar low-rent public housing projects in the District of Columbia and the nearby town of Alexandria, Virginia. The District of Columbia project was classified as commercial building construction, whereas the department classified the similar project in Alexandria as residential housing.

Professor Gujarati's Study

As can be readily seen from the above sampling of the comptroller general's studies, a large number of specific cases have been identified in which the Labor Department's Davis-Bacon wage rate determinations were seriously in error on a number of counts. Since, in several cases, the General Accounting Office's review focused on determinations that were known to be in error, it is not easy to decide from these studies alone just how widespread is the phenomenon of inappropriate determinations. Fortunately, Professor Damodar Gujarati conducted a broader sampling study of Davis-Bacon determinations for his doctoral dissertation at the University of Chicago in 1965.13 His study covered 300 counties from the 50 states, the District of Columbia and Puerto Rico. In order to obtain more precise information on Labor Department determinations in cases where unionization was low, the counties were distributed among the areas in inverse proportion to the extent of unionization in these areas.

12 Comptroller General's Report to Congress (September 1968), op. cit., pp. 22-23. 13 Damodar Gujarati, The Economics of the Davis-Bacon Act, doctoral dissertation at the Graduate School of Business, University of Chicago, 1965. A Journal of Business article based on this dissertation has been cited above.

Source: D. M. Gujarati, Journal of Business, op. cit., p. 305.

Notes-Figures in parentheses are percentages. Building construction includes commercial, residential, and building parts of missile construction. Heavy and highway construction includes interstate highways, dams, bridges, etc.

The number of determinations for which wage rates were requested. Not all determinations required rates for all the nine crafts.

†The standard error of the estimate p*.

Professor Gujarati's work sheds some important light on the following questions. (1) What actual locality is used for the purpose of making the wage determinations? (2) What proportion of determinations carry union wage rates? (3) What kinds of data are used to make the determinations?

Gujarati collected data on 372 wage determinations for nine crafts in the sample counties. A detailed breakdown of these determinations is provided in Table 6 which has been reproduced from Gujarati's paper. This table provides several interesting findings:

-The fraction p*, which is the weighted proportion of determinations carrying union wage rates, is close to one in most instances. This means that an overwhelming number of determinations carried union wage rates. Professor Gujarati found it very difficult to measure the actual extent of unionization by craft except in the case of common labor. He estimates that about 59 percent of common laborers are unionized. This compares to DavisBacon union wage determinations of 85 percent and 67 percent for building and heavy and highway construction, respectively.

-Perhaps the most disturbing evidence is the extent to which union wage rates are imported into a locality from noncontiguous counties or from statewide wage data. Professor Gujarati notes that Section I of the Davis-Bacon Act defines the area of construction as the "city, town, village, or other civil subdivision of the state in which the work is performed” and, further, that the legislative history of the act does not indicate that this definition was intended to be construed so loosely as to permit "leap-frogging." The intent of the act was to protect local wage rates, not to raise these local rates by basing determinations on rates from other, higher wage paying areas. Nonetheless, the survey data indicate that a substantial fraction of the DavisBacon determinations uses prevailing wages from noncontiguous counties. According to this data, about 25 to 38 percent of the building construction determinations were based on rates from noncontiguous counties, and 46 to 73 percent of the heavy and highway construction determinations were based on noncontiguous county rates. In some cases the Labor Department went beyond state boundaries for "prevailing" wage data, as was noted in the GAO report on construction in New England.

-Professor Gujarati found that Department of Labor wage surveys were not generally used to determine prevailing wages. Indeed, in his sample of 372 wage determinations, he found that surveys were made for only eight projects (or slightly more than 2 percent). On the other hand, unions were prompt to submit wage rates data and this accounted, in part, for the strong preponderance of union wage determinations.