The economic theorizing above suggests that prevailing wage laws should reduce employment in the affected areas, since at the above-market "prevailing" wage, some workers are not hired compared with the number that might be when employers and workers are free to negotiate wages themselves. This is a testable proposition. The one sector of the economy most affected by prevailing wage laws is construction. In a typical recent year, about one-fourth of all construction in the United States is carried out in the public sector, which faces prevailing wages under the Davis-Bacon Act and in many states, including Michigan, under state prevailing wage laws.4

Job growth in Michigan's construction industry was small in the prevailing wage period, compared with more than quadruple the jobs created during the period when the prevailing wage law was inoperative.

The states can be divided into two categories, the first including the 18 states without prevailing wage laws, and the second the 33 states (counting the District of Columbia) with such legislation. Additionally, 12 states, including Michigan, can be subcategorized as having "strong" prevailing wage laws.5 Using these categories, the number of construction workers per 1000 jobs in 1993 in states with a strong, weak, or no prevailing wage law were compared (1993 is the last full year prior to Michigan having its prevailing wage law ruled inoperative by judicial decision).

Chart 1 shows that construction workers are far more prevalent in states without prevailing wage laws than in those with such statutes. The 18 states without any prevailing wage law (except, of course, for federal legislation) in 1993 had 44.38 construction workers per 1,000 workers, compared with 33.25 for the 12 states having "strong" prevailing wage jurisdictions. Michigan was one of the "strong" prevailing wage states.6 This is compelling evidence that supports the contention that prevailing wage laws reduce employment in the construction industry.

An alternative approach is to examine Michigan's experience for the period before the prevailing wage law was ruled invalid (in December 1994) and then for the period of two and one-half years in which the statute was inoperative (ending June 5, 1997, when the Michigan Appeals Court reversed the district court and reinstated the law).

For this, monthly data on Michigan's aggregate employment, sectoral employment, total unemployment, and total labor force were gathered for the first eight years of the 1990s.7 The growth in employment in Michigan's construction industry in the 30 months prior to the court decision invalidating the prevailing wage law (June 1992 through December 1994) was then compared with a 30-month period during which the law was inoperative, from December 1994 to June 1997.

As Chart 2 shows, job growth in Michigan's construction industry was small in the prevailing wage period (4,000 jobs per year) compared with more than quadruple the jobs created (17,600 per year) in the brief period during which the prevailing wage law was inoperative. This suggests a sluggish rate of job growth during the era of wage restraints was followed by a marked expansion in the job market when those restraints were removed.

This conclusion, however, can be objected to on three grounds. First, there is seasonality in construction employment, and the periods chosen bias the results in favor of the conclusion that job growth expanded during the period with no prevailing wage law, since that period begins in a low employment winter month and ends during the summer, near the peak of the construction season. Second, employment in construction, and in general, is affected by the business cycle, so some consideration must be given to the impact of cyclical movements. Third and finally, weather could conceivably cause some of the observed trend. Suppose June 1992 was a great month weather-wise for construction, while December 1994 was a miserable one (independent of the season of the year). The sluggish growth in construction for the prevailing wage period might be explained by that phenomenon.

In order to control for possible seasonal, cyclical, and weather effects, a series of adjustments were performed using data for eight full years (96 months) from 1990 through 1997. A standard procedure was used to adjust the data for seasonal trends, the effect of which was to lower the reported employment figures for summer months and increase them for winter months when inclement weather makes it difficult to carry out construction activities.8

Graph 2 shows the actual (unadjusted) and the adjusted construction figures by month for the period June 1992 through June 1997. After seasonal adjustment, the 34,000 increase in the number of construction jobs during the era of no prevailing wages is sharply reduced, to 8,833 jobs (job growth was 20,747 in the prevailing wage era, compared with 29,580 in the post-prevailing wage era). Still, adjusting for seasonal factors, employment growth in construction expanded more than 42 percent during the period without the prevailing wage law.

Could that growth be explained by chance patterns in the weather? To largely correct for the impact of weather variation, the data were converted to three-month moving averages. For example, suppose December 1994 were unusually cold and snowy even for December, thus depressing the amount of construction employment. A three-month moving average is used to combine the seasonally adjusted construction figures for the months of October, November, and December. The figures are then averaged to produce December's moving average estimate—dramatically reducing the impact of abnormal weather in the single month of December.

Once the moving average adjustment is made, it turns out that the estimate of construction job expansion in the period of no prevailing wage laws actually increases by over four thousand workers—weather factors kept reported job growth during the December 1994 to June 1997 period lower than it otherwise would be. The estimate of increased job growth, making both seasonal and weather adjustments, rises to 13,017 jobs.

Might the growth in construction jobs be explained by the economic boom that picked up as the 1990s proceeded? The 1990-91 downturn reached a trough (bottom) in the spring of 1991, according to the National Bureau of Economic Research, the organization responsible for dating business cycles. Unemployment, however, actually kept rising into early 1992. Recovery was clearly underway by the fall of 1992, and the Michigan unemployment rate actually fell sharply during the 30-month period of prevailing wage laws from June 1992 to December 1994 (the unadjusted rate went from 9.3 to 4.8 percent), but not in the 30 months without the prevailing wage law (the unadjusted unemployment rate fell from 4.8 to 4.4 percent).

Employment growth, however, was moderately higher in Michigan in the period without the prevailing wage law: 277,800 jobs (seasonally adjusted) vs. 241,948 jobs in the earlier period. Thus some of the greater rise in construction employment in the period without prevailing wages is attributable to slightly more robust general employment growth reflecting cyclical conditions and the passage of time. Accordingly, the estimate of job growth during the suspension of the prevailing wage law was reduced by about 13 percent to account for the moderate general rise in employment growth in that period, yielding a total estimated construction job growth in the era of no prevailing wages of 11,337 jobs (see Table 1).

In other words, after adjusting for seasonal, weather, and cyclical considerations, it is estimated that construction jobs grew by over 11,000 during the no-prevailing wage period. Moreover, this estimate is probably highly conservative for a variety of reasons, the most important of which is that there was considerable uncertainty whether the December 1994 judicial interpretation that rendered the prevailing wage law invalid would be sustained upon appeal. In an environment where employers knew with certainty that there would be no prevailing wage legislation, they probably would have moved even more aggressively to take advantage of the changed legal environment.9 The national data on construction employment in non-prevailing wage states presented in Chart 1 are consistent with the view that the total long-term construction employment impact of a repeal of the prevailing wage law could well be a multiple of three or even four times the 11,000 jobs created during the period in which prevailing wages were ruled invalid in Michigan.

Given the uncertain nature of the 1994 invalidation of the prevailing wage law, the robustness of construction employment growth during the era of no prevailing wages was really quite extraordinary. Once the data are corrected for seasonality and weather (using a moving average), construction employment growth is shown to be much greater after the prevailing wage law was invalidated. As Chart 3 shows, in the first period when prevailing wage laws were operative (June 1992 to December 1994), 78.61 construction jobs were created for each 1,000 increase in total employment. In the era without an operative prevailing wage law (December 1994 to June 1997), there were 116.27 construction jobs created per 1,000 total new jobs, an increase of nearly 48 percent.

Who are the individuals who gain new jobs when prevailing wage laws are ended, as they temporarily were in Michigan from December 1994 to June 1997? Both the economic theory cited earlier and empirical evidence nationally suggest that it is likely that a disproportionate number of new jobs will go to minorities. Removal of wage requirements allows relatively less experienced minority workers a chance to offer their services at lower rates than the artificially determined prevailing wages, obtaining employment and the on-the-job training that leads to higher productivity and earnings in the long run.