Further analysis of "noncontiguous county" determinations suggests the fol lowing additional findings:

-Union wage rates determinations based on rates from noncontiguous counties occur with greater frequency the smaller the population of the county where the construction actually occurs. This is illustrated in Chart 1, which is adapted from Professor Gujarati's data. Gujarati suggests that population may be a good proxy for the extent of unionization and, if this be so, it appears that union wages are imported into nonunion localities by the DavisBacon determinations.

-For each craft, Professor Gujarati calculated the average distance in miles that noncontiguous counties lie from the actual localities in which the noncontiguous wage data are applied. These distances ranged from 72 to 84 miles and Gujarati noted that this suggests that the Labor Department goes beyond a reasonable commuting distance in search of "prevailing" wages.

Summary

This section reviewed the procedure and administration of Davis-Bacon determinations. The available evidence points very strongly in the direction that these determinations are inappropriately high in a substantial number of cases and do not conform with the intent of the act. The question of whether this is an unfortunate consequence of the huge burden imposed on the small number of people responsible for making wage determinations or whether it arises because of other reasons is entirely beside the point. Whatever the reason, the history of Davis-Bacon determinations indicates that these determinations have played an important role in strengthening the position of unionized construction labor, have tended to raise wages in the construction industry, and have spread high wages to various geographical localities irrespective of the wage rates that actually prevailed in those localities.

These problems appear to arise because of shortcomings in the determination of Davis-Bacon wage rates and this has led many investigators, such as the comptroller general, to recommend improvements in the wage determination procedure. It is, however, even more important to examine the validity of the act per se. There are two reasons for this. First, there is no evidence, despite several assurances to the contrary by Labor Department officials, that the determination procedure can ever be significantly improved. The poor record of determinations under the act is, in fact, strong evidence to the contrary. Second, it seems likely that the act would have undesirable consequences even if determination were made in strict accordance with the secretary of labor's directive on wage determination procedures. The following sections take up this latter point in greater detail.

Chart 1

Effects on Wages

The third section dealt with the upward pressure placed on wages by shortcomings in the procedure through which Davis-Bacon determinations are made. It is reasonable to expect, however, that the Davis-Bacon Act (and related prevailing wage legislation) would tend to increase wages even if local wages were accurately reflected in the determination of prevailing wages. This is because the Davis-Bacon Act per se may alter the market structure and the nature of competition in the industry. This possibility has been pointed out and investigated by Ronald Ehrenberg, Marvin Kosters, and Michael Moskow in a preliminary version of a recent unpublished paper.

The analytical basis for this argument is easy to demonstrate in a hypothetical numerical example. The Davis-Bacon Act and related laws tend to make government demand (and government assisted demand) for construction projects relatively unresponsive (or inelastic) to wages. In other words, the act tends to decrease the government's bargaining power by disallowing the possibility of withholding contracts from high wage bidders if these bidders can establish their wage as "prevailing." This tendency is augmented by the bias toward inappropriately high prevailing wage determinations, but it would occur in any situation in which the government is prevented from searching out the lowest bidder. To see the effect of this inelastic demand on market wages, consider the following hypothetical example.

The first two columns in the table below show the amount of demand for man-hours of construction work in the private market at various wage rates. Suppose there are 9,000 man-hours available. Then demand will equal supply and the market will clear at a wage rate of $3.00 an hour. If the government enters this market with a demand for 600 man-hours of labor at any price (column 3), the

total demand (government and private) will be given by the figures in column 4. The market clearing wage for the 9,000 man-hours then rises from $3.00 to $4.00, and this rate will be paid by both the government and the private market. Several comments are in order about the phenomenon illustrated by this example.

The first thing to note is that, as long as the supply of man-hours is fixed (in this case it is fixed at 9,000 man-hours), some increase in wages would occur even if government demand were responsive to price. However, the increase in wages would be less pronounced if government demand were not completely inelastic. The second thing to note is that the assumption of fixed supply may be thought of as reflecting union restrictions on worker entry into the industry. As we have seen in Section III, the Davis-Bacon Act strengthens the unions' ability to dominate certain markets, since nonunion contractors are reluctant to bid for jobs at union rates. If the government sought the lowest bidder without predetermining wage rates, additional man-hours would become available and this increased supply would help to moderate the extent of wage increases. Suppose there are no unions and 600 additional workers enter the construction industry at a wage of $3.25 an hour. In this case, the government demand would raise wages to only $3.25 instead of the $4.00 figure of the earlier example. It is interesting to note that when unions limit entry into the construction industry, Davis-Bacon determinations help the unions to raise private construction wage rates. A complete analysis of these issues is somewhat more involved but the main analytical structure of the model is made clear by this simple numerical example.

Ehrenberg, Kosters, and Moskow are developing a statistical model aimed at testing the effect of increases in the fraction of Davis-Bacon contracts on the relative wages of construction workers. Their preliminary statistical analysis, which is described in greater detail in Appendix B, may be summarized as follows: When unionization and construction growth are held constant, increases in the proportion of publicly financed construction in an area results in increases in the union scale wages of journeymen in the construction trades relative to wages of production workers in manufacturing. Ehrenberg, Kosters, and Moskow also find that increases in the proportion of publicly financed construction raises the average wage of helpers in construction trades relative to journeymen in these trades. This confirms Professor Brozen's observation, mentioned above, that Davis-Bacon determinations tend to set very high relative wages for workers in apprenticeship training programs. Brozen points out that this tends to discourage the use of apprentices on public construction projects.

Despite some ambiguities that are not unusual in empirical measurements in economics, these findings are in accordance with the studies of the General Accounting Office and the work of Professor Gujarati. The accumulated evidence points quite strongly in the direction that Davis-Bacon determinations (and determinations of related legislation) exert a powerful upward pressure on relative wages in the construction industry. This upward movement appears to take place directly in public construction and indirectly in private construction through the increased bargaining power which unions derive from the prevailing wage laws.

The Measurement of the Effect of Prevailing
Wage Laws on Relative Average Wages

While the evidence shows that prevailing wage laws have raised the relative wages of unionized construction workers, it is difficult to estimate just how large this effect has been. Data from studies by the General Accounting Office, show that inappropriate determinations ran as much as 100 percent above the "true" prevailing wage. But it is inappropriate to draw any general, economy-wide conclusions from these studies of specific determinations. There are both theoretical and empirical reasons for this which are brought into focus by the algebra in Appendix A. The main problems are summarized here. Suppose the Davis-Bacon Act was either suspended by presidential proclamation or repealed by the Congress. In order to predict the impact of such an action on average wages in the construction industry, at least the following questions would have to be resolved:

Theoretical questions:

What is the supply response of unionized labor likely to be? Will unions accept more unemployment to maintain existing wage levels or will union wage rates be permitted to decline relative to nonunion rates?

What is the supply response of nonunion labor likely to be? How quickly will workers from other occupations move into the building trades?

What is the private sector demand response likely to be? Will there be greater demand for building, houses, and commercial construction? If so, how much? What fraction of public construction expenditures can be switched to the nonunion labor sector?

Would union rates obtain on some contracts in the absence of Davis-Bacon determination? If so, to what extent, and what would the wage rates be? How many workers would be willing to remain in the unionized building trades if the differential between union and nonunion wage rates were reduced? Would state and local governments retain their prevailing wage legislation if the federal law were repealed? What difference would this make?

Empirical questions:

What are the appropriate numerical values of the supply and demand elasticities for construction labor?

What fraction of total construction expenditures is public or publicly assisted? What is the wage differential between union and nonunion labor in the building trades?

What is the fraction of unionized labor in the building trades?

This list could be easily extended, but it should be apparent from just these questions that it is quite difficult to get a very firm estimate on the precise impact