concurring in the judgment:
With genuine admiration for the care with which the majority opinion addresses the principal issues — and I concur in much of what is written — I remain troubled by contextual issues that the majority has not addressed and that indicate an effort by North Carolina to maintain a broad affirmative-action program to assure minority representation and diversity in public contracting, regardless of whether actual discrimination exists.
When we decide cases involving race-conscious and gender-conscious government programs, we must remain especially vigilant in recalling that such programs are presumptively unconstitutional, in violation of the Equal Protection Clause. See Shaw v. Reno, 509 U.S. 630, 643-44, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993). Moreover, to approve race-conscious or gender-conscious programs simply to further a policy of diversity in public contracting perpetuates, not eliminates, discrimination. Only when there is clear evidence of actual discrimination should we approve a state government’s race-conscious or gender-conscious considerations, and only then when the program is designed to remedy the discrimination.
It is noteworthy that the North Carolina program in this case was created without any evidence of discrimination. Rather, it was initiated in the 1980s with a quota system that was the product of policies designed to spread public contracting among minorities and women simply for the purpose of racial and gender representation. The State only undertook to determine whether actual discrimination existed in public contracting when City of Richmond v. J.A. Croson Co., 488 U.S. 469, 109 S.Ct. 706, 102 L.Ed.2d 854 (1989), made clear that North Carolina’s program was constitutionally infirm. As a result, North Carolina has never had — and, indeed, never will have — the opportunity to determine whether discrimination existed in public contracting absent the affirmative-action program’s effects on the market. And even after concluding that discrimination did in fact exist, the State reinstituted the exact same quotas that it had previously used.
To be sure, the program has changed over time, but even the immediate past version of the North Carolina statute required that quotas be satisfied through the participation of a broad array of ethnic peoples, such as Portuguese Americans, Asian Americans, American Indians, and Alaskan natives, regardless of whether evidence of discrimination against those groups existed.
*259In its current form, the North Carolina statute creating the affirmative-action program provides little to no guidance regarding how it is to be implemented, leaving an unnerving amount of discretion to a small number of Department of Transportation administrators. The current statute uses the 2004 disparity study as a predicate to justify race-based and gender-based remedial efforts and directs the Department of Transportation to commission future studies to determine whether future remedial efforts are necessary. Specifically, the statute requires the North Carolina Department of Transportation to “establish annual aspirational goals, not mandatory goals, ... [that] shall not be applied rigidly on specific contracts or projects.” N.C. Gen.Stat. § 136-28.4(bl).
But the statute does not provide any range nor give any criteria for these goals. Instead, it authorizes the Department to act based solely on the Department’s own interpretation of the 2004 study and future studies that the Department itself must commission. As a result, Department administrators, not state legislators, are those responsible for determining which groups qualify as “minorities,” and they are statutorily authorized to make that determination when a group has been “subjected to discrimination in the relevant marketplace” and has been “adversely affected in [its] ability to obtain contracts with the Department.” Id. § 136-28.4(c)(2) (emphasis added). These administrators are also given the power to determine, in their sole discretion, adequate levels of representation in obtaining public contracts for each minority group.
After the Department, in its discretion, adopts “aspirational goals,” it is given the additional authority to decide whether prime contractors have made “good faith efforts” to meet the aspirational goals, which in themselves are not to be “applied rigidly.”
The result is an amorphous exhortation to favor minorities and women in public contracting. I question how it can be concluded that actual discrimination is being addressed and, if so, whether the means used can be sufficiently narrowly tailored so as to comply with the Constitution. I take little comfort from the fact that the only meaningful oversight of this process has been review by the federal courts following years of litigation. While the majority in this case thoroughly scrutinizes whether the implementation of this program is genuinely aimed at fighting actual discrimination in state contracting, the fact that such review is necessary demonstrates that the State and its administrators have few if any standards by which they ensure that their program complies with constitutional requirements.
Our holding in this case is prime evidence of the problem. We conclude that the state officials who used the 2004 study to justify favoring Asian Americans, Hispanic Americans, and women had insufficient evidence of discrimination to do so. That finding hardly instills confidence in the State’s continued application of ad hoc standards purportedly designed to remedy actual discrimination.
Moreover, the arbitrariness of the entire North Carolina structure for determining when to consider race and gender is displayed by the weakness of the statistical data on which North Carolina officials have been willing to rely. Several examples demonstrate this.
First, as noted by the majority, the data in the study regarding contracts awarded by the Department’s division offices is statistically invalid. While the majority of contract dollars were handed out on contracts from the central office, the division-let contracts represented approximately *260one-third of the total value of state contracts. The program affords privileges to women and minorities in bidding on such contracts, despite the fact that no evidence whatsoever indicates that these contracts were awarded in a disparate manner.
Second, the State asserts that a decrease in subcontracts going to minority-owned and women-owned enterprises during a period when the program was suspended indicates that, absent preferences for minorities and women, discrimination would take hold. But the significance, if any, in the drop in the utilization of minorities and women during suspension of the program is a classic chicken-and-egg problem. If the underlying market for contracting in North Carolina was discriminatory prior to implementation of the program, then the drop in utilization during suspension of the program may have reflected the fact that the program had helped remedy discrimination. If, on the other hand, the underlying market was nondiscriminatory, then the drop in utilization merely reflected the removal of the unfair competitive advantage given to minorities and women. See W. States Paving Co. v. Wash. State Dep’t of Transp., 407 F.3d 983, 1000 (9th Cir.2005).
We cannot resolve this question because no evidence exists regarding whether disparities existed at the time the program first began. It might be surmised that fewer minority-owned and women-owned businesses will receive subcontracts on Department projects if the program is suspended. But it cannot be determined whether this change would reflect a return to a fair, competitive status quo or a discriminatory marketplace. Moreover, regardless of how we interpret the drop in utilization during the period of the program’s suspension, it is rather bold, and perhaps even invalid, to assume that the market will react to suspension of the program in the same manner that it did 20 years ago.
In addition, even if we had reason to believe that the drop in utilization during suspension of the program indicated potential discrimination in the marketplace in the early 1990s, it still could not justify the continued use of race-conscious and gender-conscious policies. Affirmative action is only appropriate to remedy past or ongoing discrimination, not to stave off discrimination that might hypothetically occur in the future. See Hayes v. N. State Law Enforcement Officers Ass’n, 10 F.3d 207, 217 (4th Cir.1993) (“[Ejven when race can be taken into account to attain a balanced work force, racial classifications may not be employed to maintain a balanced work force” (emphasis in original)). Should intentional discrimination appear without this program, future remedial efforts could become appropriate. But we should not ratify the government’s use of race in perpetuity based on mere concern that such a result might occur. Were we to do so, we would send the terrible message that our society has been and always will be divided by race, that racial hostility is an inevitable consequence of a heterogeneous society, and that government cannot escape classifying its people by the color of their skin.
Third, as the majority rightly points out, the evidence regarding state subcontracting practices indicates that women are grossly owerutilized. The State nonetheless contends that women too must be benefited by affirmative action. To justify this “aspiration,” the State relies on evidence of purported discrimination in the private building construction industry without demonstrating how that evidence justifies a need for affirmative action in awarding public contracts from the Department of Transportation. Because this evidence comes from a wholly dissimilar contracting market, it is inappropriate to *261rely on it to draw inferences regarding Department contracting. Moreover, the evidence from the private building sector is at best mixed and therefore fails to justify affording privileges to women.
Fourth, when considering only valid statistical evidence — data that are actually tied to contracts awarded by the Division’s central office — the evidence fails to justify the sweep of preferential privileges afforded to each minority group and women in the program. While each of the minority groups were shown to be underutilized in terms of subcontract dollars relative to their market share, only African Americans were underutilized at a statistically significant level. Cf. Croson, 488 U.S. at 501, 509, 109 S.Ct. 706 (inference of discrimination justifying remedial affirmative action may only be drawn by evidence of “gross” and “significant” disparity). The majority recognizes that Native Americans were underutilized at a confidence level of approximately .15, a result that is not, in most conventional statistical analyses, considered statistically significant.* Similarly, the regression analysis concluding that firm experience, capacity, and the like were not responsible for the shortfall in contract dollars supported a finding that the disparities were a result of race and not other factors only with respect to African Americans. In other words, with regards to Native Americans — as with Asian Americans and Hispanic Americans — the study could not conclude with confidence that the shortfall in contract dollars was not due to mere chance, a lack of firm capacity, or some other race-neutral explanation.
The Department, perhaps recognizing these infirmities, chose to fill the gaps in its data regarding public contracting by relying on inapplicable and therefore invalid data from the private sector. This demonstrates the structural weakness of the entire program. It is indeed disquieting that the State believes that such a flimsy approach can justify governmental decisionmaking based on race and gender. The majority wisely rejects the State’s approach and properly holds the State to the burden of demonstrating a strong basis in evidence leading to an inference of discrimination in Department subcontracting. Cf. Croson, 488 U.S. at 500, 109 S.Ct. 706. But the fact that North Carolina has come to federal court and unrepentantly relied on dubious — and, in some cases, completely discredited — data to support its conjecture that discrimination is persistent in state contracting might well lead us to conclude that the State is now more interested in seeking a post hoc justification for a program designed to engineer proportional race and gender representation, regardless of discrimination, than in conducting a good-faith, open-minded inquiry into whether remedial efforts are necessary.
*262I do not doubt that North Carolina’s program was conceived with the benign motive of preventing discrimination in public contracting. But its approach of making contracting decisions in favor of minorities and women, based on the flimsiest of evidence, runs the risk of actually creating and perpetuating discrimination.
While I would be more inclined to strike down the entire program as a violation of the Equal Protection Clause, I am willing to allow, as the majority does, North Carolina the opportunity to review and adjust its program in light of what we have written.
Accordingly, I concur in the judgment.
BEATY, Chief District Judge,concurring:
I concur fully in the majority opinion in this case, and I agree wholeheartedly with the conclusion that the North Carolina Minority Business Enterprise Program, N.C. Gen.Stat. § 186-28.4, is constitutional on its face and as applied to African Americans and Native Americans. I add only a few additional observations.
First, I am compelled to note that, in my view, the inequities that are still evident in public contracting with respect to African American subcontractors are a present reminder of past racial discrimination in North Carolina, racial discrimination that included not only enslavement and the denial of full personhood under the Constitution in the past, but that also included legally sanctioned exclusion from the most basic rights of citizenship within not-too-distant memory. See Regents of the University of California v. Bakke, 438 U.S. 265, 387-396, 98 S.Ct. 2733, 57 L.Ed.2d 750 (1978) (Marshall, J., dissenting) (surveying the history of racial discrimination in this country and the Supreme Court’s role in sanctioning that discrimination in Plessy v. Ferguson). In light of this history, the present disparities in public contracting could not be viewed as the result of nondiscriminatory market forces, nor could the statistics showing the exacerbation of those disparities when the Program was previously suspended be viewed as reflecting a fair, competitive status quo. Instead, in the context of history, these disparities reflect the present effects' of past discrimination, and the State has recognized this, as discussed in the majority opinion. In my view, the Program enacted by North Carolina serves the highest interest of the State in attempting to remedy this past discrimination and is fully consistent’with the promise and purpose of the Fourteenth Amendment. Although there is expectant hope that at some point the Program will no longer be necessary to redress the present effects of this history, we cannot deny the history of discrimination or its lingering effects.
Finally, I would note that the majority opinion has held that the Program is unconstitutional as applied to Hispanic Americans and Asian Americans, based on the lack of statistically valid information presented by the State as to these groups. However, as noted in the majority opinion, the State may, in the future, include these groups within the Program if the State undertakes additional study and determines that inclusion of these groups in the Program is warranted. Likewise, the majority opinion has held that the evidence presented by the State would not establish a sufficient basis to justify the Program as to nonminority women, even using intermediate scrutiny. However, the State may very well include nonminority women in the Program in the future if sufficient basis exists, and in making this determination, the State may consider any underutilization of nonminority women that occurs in the absence of the Program.
*263Therefore, I am pleased to concur fully in the majority opinion in this case.
Here I follow the study in examining statistical significance at the .05 level. When a result is significant at the .05 level, it means that the probability of that result occurring by chance is 5% or less. See, e.g., Sherri L. Jackson, Research Methods and Statistics: A Critical Thinking Approach 168-69 (3d ed. 2009). While statistical significance is arbitrary and imperfect, the .05 confidence level is often used in the social sciences as a marker of when a result is a product of some external influence, rather than ordinary variation or sampling error. See, e.g., Earl Babbie, The Practice of Social Research 483 (12th ed. 2010).
We need not decide here whether statistical significance at the .05 level is the bellwether by which we determine whether the State’s statistics are sufficient to support a compelling state interest. Rather, I merely point out that the study’s results regarding Native Americans, Hispanic Americans, and Asian Americans do not reach the level of significance that the study itself sets out as the standard by which one could confidently conclude that discrimination was at work.