Stop Cooking The Books On Set-Asides

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 13
7-93 February 15, 1993 STOP COOKING THE BOOKS ON SET-ASIDES How Statistical Disparity Misleads and Multiple Regression Excels for Assessing Discrimination in Public Works Contracting By Barry W. Poulson, Ph.D. University of Colorado EXECUTIVE SUMMARY The U.S. Supreme Court in Richmond vs Croson (488 U.S. 469, 1989) set new standards for government set-aside or quota programs for minority firms in the award of public works construction projects. To Justify such programs, government agencies must now show: a) a specific finding of past discrimination In the award of public works construction projects; b) evidence that qualified mimority contractors who were willing and able to meet all requirements to successfully do the work received a statistically significantly lower share of these contracts: While statistical evidence may be used to prove past discrimination in the award of these contracts, such evidence must relate to the specific type of contract awarded, and to the relevant pool of qualified minority and nonminority contractors in a specific area covered by ‘a proposed program of race /gender preference. The most widely used methodology used in constructing this evidence is disparity analysis, which compares the share of contracts awarded to minority contracting firms to the share of those firms qualified to do the work in specified industries. However a number of recent court cases have raised serious questions about the use of disparity | What Foidence Will Satisfy studies as not meeting the new standards set in |e Croson Test?......u Page enone eee Drawbacks of Statistical This issue paper will show that simplistic disparity | Désparity Studies .......Page 3 analysis which measures crude ratios is fundamentally ‘i flawed. Therefore. ratlo analysis cannot meet the | Raceand Gender Arent standards set in Richmond vs Croson. ty Vai ne PQ What Multiple Regression Many factors other than ethnicity or gender influence the award of publie works construction contracts. The | Anidlusts Showed : only way to measure the impact of alleged MESIAL vverreernerePOGE iscrimination in the award of these contracts is to standardize for these other variables. Dirmmnien tee Multiple regression models are needed (and will be | Multiple Regression...Page 8 provided later in this paper) in order to have a truly valid framework for the analysis of discrimination in | Redressing Injustice the award of public works construction projects. Spec aman eae (Continued on Page 2) Note: The Independence Issue Papers are published for educational purposes only, and the authors speak for themselves Nothing written here is tobe conserued as necessarily representing the views of the Independence Institute or as an attempt to influence any election or legislative action. INDEPENDENCE INsrITUTE © 14142 Dawer West Parkway #101 * Gotpen, CO 80401 * (303) 279-6536 EXECUTIVE SUMMARY, continued... One study that has applied multiple regression analysis, conducted for the State of Louisiana, concludes that variables other than discrimination have been the major factors determining the award of public works construction contracts, That study shows that in the area of subcontracting there is no evidence of discrimination. Rather the evidence reveals that black-owned firms received preferential treatment in the award of these contracts due to a federal goal program that requires that ten percent of highway construction projects financed with federal funds go to minority firms. ‘The results of the Louisiana study raise serious questions regarding race/gender based government set-aside, bid preference, or quota programs. If there is no specific finding of past discrimination in the award of prime contractor subcontracts, then such programs do not appear to satisfy the more rigorous standards set in Richmond vs Croson. What Evidence Will Satisfy the Croson Test? In Richmond vs. Croson, the U.S. Supreme Court struck down the Minority Business Enterprise (MBE) utilization procurement program s~t up by the city of Richmond. This ruling has brought existing set-aside programs under a wtict-scrutiny, 14th Amendment analysis, and requires a specific finding of past discrimination to establish such programs. ‘The Supreme Court ruled that a state or local government body cannot enact set- aside programs, such as that set up by the city of Richmond, without evidence that relates specifically to the alleged prior discrim:i:tion, In addition to direct testimony regarding specific cases of discrimination, the Supreme Court provided for the use of strictly scrutinized statistical evidence of discrimination: “Where there is a significant statistical disparity between the number of qualified minority contractors willing and able to perform a particular service and the number of such contractors actually engaged by the locality or the locality's prime contractors, an inference of discriminatory exclusion could arise...Moreover, evidence of a pattern of Individual discriminatory acts, can, if supported by appropriate statistical proof, len¢ support to a local government's determination that broader remedial relief is justified.’ 489 U.S. at 509 (plurality opinion). ‘The Supreme Court attempted to delineate what types of statistical evidence would meet this standard, For example the Court ruled that evidence relating the percentage of contract dollars going to minority firms and the percentage of minorities in the city’s population was irrelevant, It concluded that the general population is not "the relevant statistical pool” for determining the ‘number of minorities qualified to undertake the particular task", Further, the Court ruled that evidence relating to prime contracting was irrelevant to issues of discrimination in a minority set-aside program for subcontract work. As the summary of the opinion explained: *Reliance on the disparity between the number of prime contracts awarded to minority businesses and the city’s population is ...misplaced, since the proper statistical evaluation would compare the percentage ‘of MBEs in’ the relevant market that are qualified to undertake city subcontracting work with the percentage of total city construction dollars that are presently awarded to minority subcontractors. neither of which is known to the city." (Summary. p. 4133, U.S, Law Week, 1-24-89, synopsizing 488. US. at 5011-02) Finally, the court ruled that the empirical evidence must be drawn from "the relevant market area’. The failure of the City of Richmond to limit the ordinance geographically was one of the reasons that the ordinance was struck down. 488 U.S. at 504-05. Based upon the Supreme Court ruling in Richmond vs Croson, we can conclude that government entities might use appropriate strictly scrutinized statistical data as evidence of discrimination in the award of public works construction profects, but in doing so they must satisfy the following standards. a) Existing set-aside programs in the award of public works construction projects are subject toa strict-scrutiny, 14th Amendment analysts, This requires a specific finding of past discrimination in the award of public works projects to establish such set aside programs. b)_ Statistical evidence may be used to infer past discrimination in the award of public works construction projects. That evidence must show that qualified minority contractors who were willing and able to perform a particular service received statistically significantly lower awards of public works projects compared to non minority contractors. c) The statistical evidence must relate to the specific type of awards, 1.e prime contracting versus subcontracting: and to the relevant pool of minority and non-tinority able, experienced, willing contractors in a specific geographic region covered by the set-aside program. Drawbacks of Statistical Disparity Studies Following Richmond vs Croson, economists have attempted to provide statistical evidence of discrimination in the award of public works construction projects that would pass muster under the new standards. The methodology used in disparity studies usually compares the raw percentage of minority firms in the construction industry to the raw percentage of contract dollars awarded to minority firms by a particular government entity.” ‘The particular groups classified as minorities vary in different studies. Most studies define minorities to include blacks. Hispanics, American Indians, and Asians. Since many of the government set-aside programs for the award of construction projects include women- ‘owned firms, women are also included in the definition of minorities in these studies. Ambiguities in the definition of minorities are revealed in several of these studies. For example the Louisiana case study cited in section five of this report includes French-Acadians. It is not clear that French-Acadians would fall in the definition of minorities used under federal laws relating to discrimination in the award of public works construction contracts. Recent court cases have called into question the use of disparity studies as evidence supporting discrimination in the award of public works construction contracts. The courts have warmed that government ertities relying too heavily on crude percentage calculations may fail to satisfy one or more of the standards set in Richmond us Croson. ‘The 9th U.S, Circuit Court of Appeals ruled against the City of Seattle in a recent case involving the use of disparity studies, The court found that evidence of discrimination in the award of public works construction contracts in one city or jurisdiction can't be extrapolated to argue that such discrimination exists in another. In other words the city failed to satisfy standard (¢) above. In a case which is pending in federal court in Denver. Concrete Works of Colorado challenged a city ordinance that sets aside public works construction projects for minorities. Plaintiffs argued that the statistical data in a disparity study conducted by the city was biased * See, for example, Statistical Supporf for San Francisco's MBE-WBE-LBE Ordinance, prepared by BPA Economies, San Francisco, 1989. for several reasons. The data was drawn from the metropolitan area, including suburban areas with a small share of minority. and female-owned construction companies, The data was classified by business size, and ignored important time periods which skewed the results. ‘The plaintifis argued that the city’s own study fatled to show an historical pattern of disparity in the share of contracts received by minority and non-minority firms, The challenge to the Denver ordinance was based on the failure to satisfy both standards (a) and (b) above.” In ODonnel vs. District of Columbia. the U.S. Circult Court ruled that data showing a lower percentage of construction work done by minority firms relative to the percentage of minority contractors in the industry was Insuflicient evidence of discrimination in the awa of these construction contracts. The court said that a number of factors, other than discrimination, could ha’ influenced the award of construction contracts. The court suggested that minority firms mi: have been unable or unwilling to do the work for a number of reasons. In other words, the da provided by the District of Columbia failed to show that qualified minority contractors we willing and able to perform the particular service, f.e. the city failed to satisfy standard ( above. Overall, while the case law casting doubt on disparity studies is not yet definitive, sigt increasingly point that way. Such an outcome is foreseeable because, as we shall demonstra below, there are logical and empirical reasons for concluding that the simple ratio analys methodology of disparity studies 1s fatally flawed as evidence of discrimination in the awa) of public works construction projects. To understand why, one merely begins by asking what are the potential sources discrimination in the award of public works construction projects. Race and Gender Aren't the Only Variables One potential source of discrimination is in the bidding procedure. If the award : public works construction contracts is based upon the lowest qualified bidder, then it is hard understand how discrimination could ever occur in determining the winning bid. However, or can think of other obstacles in the bidding procedures. ‘The bidding procedures for the award of these contracts may impose high informatio and transactions costs on potential bidders. Emerging majoriity- or minority-owne construction firms may not have access to information regarding the bidding process, or mz Jack the legal expertise and staff to incur the transactions costs required to participate in tk bidding procedures. ‘The question then is whether these problems are due to their status as minority owne construction firms, or whether these problems are encountered by other firms in the industry It ts conceivable that government agencies responsible for the bidding procedures could plac minority-owned firms at a disadvantage in bidding for construction contracts. If so then thi 4s a government failure, and the appropriate remedy 1s to correct any obstacle originatin within the government agency. It may be that the information regarding the bidding process is available, but the minority-owned construction firms are unable to participate in the bidding process. This ma be because there 4s discrimination in the determination of qualified bidders. To find out, on would have to know what criteria are used in determining qualified bidders and how thos * Editor's Note: Just as this paper was going to press, and too late for inclusion in the above section, Judy Sherman Finesilver ruled against Concrete Works of Colorado in Case No. @2-F-21, decked 2/26/93, Tht decision will be analyzed in a fortheoming Independence Issue Paper. criteria were applied in determining qualified bidders in awarding contracts in the construction industry. If there 1s a problem of obstacles, it would again appear to be a government failure rather than a private market failure, and this would require remedies which standardize procedures for determining qualified bidders for government construction projects. Minority-owned construction firms may be qualified to bid on public works construction projects, but may not be willing to participate in the bidding process. There are many possible reasons why a minority construction firm may be unwilling to bid on public works construction projects. As the Louisiana case study reveals, construction firms are often specialized in different types of construction projects. Many of these firms have no interest in bidding on government construction projects because those projects do not fit their expertise and specialization within the construction industry. ‘The Loutstana case study also reveals that characteristics of the firm, other than minority ownership, can influence both its ability and willingness to bid on government construction projects. For example, small scale firms often do not have the personnel, capital equipment, financing, etc. to compete for the award of public sector construction projects. The problem then is not one of discrimination, but rather with characteristics of the firm other than minority ownership that do not enable them to successfully compete in the bidding for these contracts. Another potential source of discrimination {s in entry into the construction industry. If there are obstacles to entry into the construction industry, then this 1s a private market failure and not a public sector failure. To answer this question we must recognize that there may be a number of barriers to entry into the construction industry. All construction firms, including minority-owned firms, face barriers in the form of capital costs, financing, technical expertise, etc., that influence their ability to compete. Disparity studies are particularly ill-suited to address the problem of potential discrimination in entry into the construction industry. Disparity studies only measure the share of public works construction contracts awarded to minority firms already in the industry. This 1s a fundamental flaw not only in disparity studies, but also with the government set-aside and quota programs themselves. Setting aside a certain percentage of public works construction contracts may benefit some existing minority- and female-owned firms that are already in the industry. But such programs at the same time create additional barriers to entry for firms attempting to enter the industry, including minority firms. The latter would be placed at a disadvantage vis a vis the minority firms already in the industry who are able to capture the pure rents created by such set-aside and quota programs, To the extent that set-aside or quota programs increase costs in the construction industry, this reduces demand aud therefore reduces profitability for all other firms. This creates an additional barrier to entry of new firms, including minority firms, For example, the Associated General Contractors of California points out that the race/gender bid preference and race/gender subcontractor quotas imposed by the City of San Francisco cost taxpayers tens of millions of dollars annually, due to the manner in which bids must be solicited. AGCC estimates that nine to fourteen percent of all contract dollars are wasted to support such programs. And in spite of such programs, the share of MBE/WBE firms tn the industry declined over the eight-year period after the quotas were introduced to less than ten percent of the firms in the industry.” * AGC of California, San Francisco and North Bay Districts, Letter to Board of Supervisors, City and Couniy of San Francisco, February 25, 1992. What Multiple Regression Analysis Showed in Louisiana Richmond v.s Croson would rule out set-aside programs unless there is clear, strictly scrutinized evidence of discrimination in the award of public works construction contracts among a pool of minority-owned construction firms willing and able to do the relevant work. As the courts have argued, there are a number of reasons why construction firms may be unable or unwilling to compete for the award of these contracts, other than their minority ownership status. Disparity studies that compare the award of public works construction projects to the number of minority-owned firms in a geographic area fall to distinguish between the impact of minority-owned status and other factors that determine the award of public works construction projects. Therefore the evidence provided in disparity studies fails to meet the standards established in Richmond us. Croson. ‘One can expect that the success of a firm in receiving construction contracts is a function of a number of variables, not just the single factor of its owners’ race or gender. ‘Therefore the best way to address the issue of discrimination in the construction industry is to utilize multiple regression analysis. Only by this method can the variables that evaluate minority status of firms be incorporated with other variables to test for the significance of discrimination in the award of public works construction contracts. To date there has been one major study that did use multiple regression analysis to determine race/gender discrimination in the award of public works construction projects. ‘That study was conducted in 1990 at Louisiana State University for the Governor's Task Force on Disparity in State Procurement." The results of the multiple regression analysis in the LSU study can be summarized by saying that a number of variables other than minority ownership of the firm significantly affected the award of construction contracts. firm, 0 variable Itsy vs subcontract award included: © firms that work in the highway, street and bridge construction industries, the number of times the firm submitted bids for prime contract work, and the number of times the firm tried to get subcontract work. variables that h: is Te nit award included: the percentage of the firm's gross receipts from prime contract work, and © whether or not the firm was licensed or worked in the building construction industry. ‘Thus. the study reveals that specialization, expertise. and experience significantly determine the award of these public works construction contracts, Firms that are focused on prime * Lunn, John and Huey L. Perry, An Analysts of Dispartly and Posstble Discrimination in the Loulstana Construction industry and State Procurement System and lis Impact on Minority and Women Owned Firms Relative to the Public Works Arena, Final Report for the Governor's Task Force on Disparity in State Procurement, Department of Economic Development, 1990. contract work, and that specialize in the building construction industry are less likely to bid for and receive subcontract awards. ‘The purpose of the study was to standardize for variables other than ethnicity and gender in order to determine if discrimination in the award of public works construction projects existed in Louisiana. Separate dummy variables were introduced in the regression ‘equations for firms owned by women, blacks, French-Acadian, and “other” which includes Asian, Hispanic, and American Indians. The coefficients on ethnicity and gender, other than black, were consistently insignificant. These results confirm that the award of construction contracts was determined primarily by factors other than race and ethnicity. The only coefficients for ethnicity and gender that were significant in the regressions for all years was a positive coefficient for black-owned firms. Thus, the study revealed that, blacks received favorable treatment in the award of subcontracted public works construction contracts. When the data was analyzed by individual years, the coefficients on ethnicity and gender, other than black, were again consistently insignificant. For prime contracts awarded the coefficient for black-owned firms was negative and significant in only one year, the 1989 data subset. The coefficient on subcontracts awarded black- owned firms was positive and significant in two years, the 1987 and 1989 data subsets. Analysis of the data in the Louisiana study shows that few minority firms, including black-owned firms, were involved in prime contract work. The study concludes that the results of the analysis of prime contracting are somewhat fragile. Barriers to entry appear to be greater in the prime contract construction work compared to subcontract work, but these barriers to entry appear to be due primarily to factors other than ethnicity and gender. Minority-owned firms. including black-owned firms, participated extensively in subcontract construction work. The study concludes that the results for subcontracting are more robust. Evidence that blacks received favorable treatment in the award of subcontracts for public works construction projects ts not surprising in light of the federal goal program, ‘That program, which has been in place since the early 1980s, requires that ten percent of highway construction projects financed with federal funds go to minority and women owned firms. ‘The Louisiana study also conducted regressions for the award of construction projects that did not receive federal funding, and therefore were not subject to the federal goals program. In none of these regressions were the varlables for minority- and female-owned firms significant. The study concludes that blacks received favorable treatment in subcontract awards that were subject to the federal goals program, but not in the award of contracts that were not subject to the federal goals program. (However, the Louisiana study is somewhat ambiguous because the State of Louisiana had its own set-aside program that mandates certain levels of involvement of minority- aad women-owned firms.) ‘The study concludes that minority- and women-owned firms were not being discriminated against in the state-funded programs even in the absence of the enforcement of state law. Further, the study concluded that the evidence of preferential treatment of blacks in the award of subcontracts subject to the federal goals program, but not in the award of state-funded subcontracts indicates that the state set-aside program has not been enforced. Proposed Models for Discrimination Analysis Using Multiple Regression A. All Projects ‘This first set of multiple regression models provides a framework for the statistical analysis of race/gender discrimination in the award of total, prime, subcontracts, and other construction projects, including both private and public sector projects, C(total)it = ao +bit(Fit) + cit(Dit) +eit where Citotal)it is a measure of total construction contracts that a firm 1 receives in time period t. (Fit) ts a vector of variables representing characteristics of firm { in time period t that are expected to influence the Itkelthood that the firm will receive construction contracts. (Dit) is a vector of dummy variables that indicate whether the firm is minority owned ao, bit, and cit are parameters estimated in the model eit Is an error term C(prime)it = ao +bit(Fit) + cit(Dit) +eit (1.1) where C(primeit 1s a measure of prime construction contracts that a firm { receives in time period t C(subjit = ao +bit(Fit) + cit(Dit) seit where C(subjit is a measure of sub contracts that a firm 1 receives in time period t Clother)it = ao +bit(Fit) + cit(Dit) seit where C(other)it is a measure of other contracts that a firm { receives in time period t The foregoing models are designed to capture discrimination against minority-owned firms in the award of total, prime, subcontracts, and other construction contracts. ‘The hypothesis tested here is that there are barriers to the success of minority firms in the award of construction contracts, when standardized for other factors that influence the success of firms, other than minority owned status. ‘The coefiicient on the dummy variable for minority owned status is expected to be negative. B, Public Projects Only ‘The following multiple regression models provide a framework for the statistical analysis of race/gender discrimination in the award of construction projects, i.e excluding private sector construction contracts. PWO(totalyit = ao +bit(Fit) + cit(Dit) + dit(Rit) + eit where PWC(totaDit is a measure of total public works construction contracts that a firm receives in time period t (Rit) 1s a vector of variables that represent rules government agencies impose on firms in the award of public works construction projects. PWC(prime)it = ao +bit(Fit) + cit(Dit) + dit(Rit) + eit where PWC(prime)it is a measure of prime public works construction contracts that a firm { receives in time period t PWC(subjit = ao +bit(Fit) + cit(Dit) + dit(Rit) + eit where PWC{(sub)it is a measure of subcontracted public works construction contracts that a firm { receives in time period t ‘The foregoing models are designed to measure discrimination against minority owned firms in the award of public works total, prime, and subcontracted construction contracts, ‘The hypothesis tested here is that there are barriers to the success of minority firms in the award of public works construction contracts, when standardized for factors that influence the success of firms other than minority owned status. Note that in addition to the variable for firm characteristics we include a variable for restrictions that the government agency imposes on firms in competing for these public works contracts. Thus not all minority firms are qualified to bid on public works construction projects. Firms that do qualify may not meet other requirements such as financing, bonding etc. Note also that Richmond vs. Croson requires refinements in the model to determine discrimination that might arise in the award of contracts to prime contractors, or in subcontracting by these prime contractors. ©. Testing the Effectiveness of an Existing Set-Aside Ifa set-aside or quota system is used In the award of public works construction contracts then the question is whether these programs are an effective policy to reduce discrimination in the construction industry. It is also possible to address this question rigorously, using multiple regression analysis. The following models are modified to reflect the presence of set-aside or quotas for the award of public works construction projects. PWC(total)'it = a’o +b'it(Fit) + c'it(D'it) + d'it(R'it) + e’it where PWC(total)it is a measure of public works total construction contracts that a firm 1 receives in time period t in the presence of set-asides or quotas for the award of contracts to minority firms PWC(prime)'it = a'o +b'it(F'it) + c'it(D'it) + d'it(R'it) + e'it where PWC(prime/'it is a measure of public works subcontracting that a firm { receives in time period t in the presence of set aside or quotas for the award of contracts to minority firms PWC(subj'it = a’o +b'it(F'it) + c'it(D'it) + d'it(R'it) + e'it where PWC(sub]'it is a measure of public works subcontracting that a firm i receives in time period t in the presence of set-asides or quotas for the award of contracts to minority firms ‘The foregoing models test for the significance of discrimination in the presence of set-aside or quota programs in the award of public works construction projects. In this case minority status is expected to have a positive effect on the award of these contracts. ‘The coefficient on the dummy variable for minority owned status is expected to be positive. 1D. Reasons for Declining to Bid A firm may choose not to bid on public works construction projects, That decision may reflect perceptions of discrimination or potential discrimination by minority firms. However, many factors may also influence the firm's decision to bid on public works construction projects, other than either minority-owned status, or perceptions of discrimination. ‘The following models provide a framework to estimate the impact these variables on the decision to bid, with and without the presence of set-asides or quotas in the award of public works construction projects. Bit = ao +bit(Fit) + cit(Dit) + dit(Pit) + eit where Bit is a dummy variable set equal to unity when a firm bids on a public works contract, and set equal to. when a firm does not bid on public works contracts Pit is a variable capturing perceptions of discrimination or potential discrimination in the bidding process Biit = a'o +bit(F't) + cit(D'It) + AIK(P'IL) + e'it where Bit is a dummy variable s.. .

You might also like