"Disparate Impact" Has Bad Impact on Society

Disparate Impact is yet another example of a well-intentioned government  initiative whose side effects are more damaging than its main effects are beneficial.

The Disparate Impact theory of law, being aggressively used by the Obama Administration's Dept of Justice and Equal Employment Opportunity Commission (EEOC,) asserts that even if intentional discrimination is not found, if a racial or gender group is underrrepresented, it's prima facie evidence of discrimination.

For instance, the EEOC and Department of Justice are arguing that employers may be discriminatory if they use credit checks or criminal record as a hiring criterion. For example, Pepsi recently had to pay more than $3 million plus offer jobs and training mainly to African-Americans because Pepsi's background checks would not give full-time employment to applicants with a pending criminal prosecution.

Disparate Impact's intent is reasonable: Because racism is often unconscious and no employer would admit to being racist, the theory says, "Let the proof be in the pudding." For example, let's say 30% of the African-American local population meets the basic requirements to become a manager but only 15% of the hired managers are African-American, that's evidence that the employer is guilty of racial discrimination unless the employer can prove the validity of his hiring criteria.

The problem is that it's very easy for a plaintiff to prove he's underrepresented--the EEOC requires all but the smallest employers to collect and make public those statistics. In contrast, it's very difficult and expensive for an employer to prove that all its employee selection criteria for all positions are predictive of job performance. For reasons too complicated to explain here (restricted range and sample size, for example) it would require a team of expensive psychometricians  to develop instruments sufficiently valid to hold up in court. And without such custom-validated instruments, the employer risks being sued, That's true even if the instrument is a highly validated national test of reasoning (e.g., SAT, LSAT, IQ, etc.) Even though reasoning is critical to all but menial jobs, plaintiffs have been able to argue that tests of cognitive ability are insufficiently related to the job to allow their use if, as there almost is, an adverse impact on underrepresented minorities.

Making employers' burden of proof even more onerous, Roger Clegg, president and chief counsel of the Center for Equal Opportunity, points out that the employer can prevail only if he can prove a “business necessity” for the practice and even then can still lose if the plaintiff can show an alternative with less disparate impact.

Plus, a lawsuit would have to (expensively) resolve such questions as "What constitutes the pool of qualified applicants?" For example, let's say I run a business in Oakland, CA, where I live. Oakland's population is 35% Black,  15% Latino. Using the government's 4/5 rule, if I had fewer than 30% Black executives or 12% Latinos, the EEOC could make me prove my selection criteria were valid. I'd likely respond, "But wait. The pool of people that are qualified are graduates of highly selective institutions such as Stanford or Harvard, and there are fewer minorities in that pool." The EEOC or court could deem that discriminatory unless I proved that that level of employee really was necessary and  that it would be an undue hardship to have to use all the other criteria needed to find people of Ivy caliber and that there was no more race-neutral such screening device. Months or years of expensive, stressful legal arguing could be spent merely on that.

In fact, few Disparate Impact lawsuits are filed, and most of those are won by the employer. But the fear of such lawsuits is significant and increasing because of the Obama Administration's recent actions. So employers are ever more likely to move merit toward the back seat in favor of selection criteria that are less predictive of job performance but that don't reduce the percentage of Blacks or Latinos selected in favor of whites and Asians.

Net, of course, when less predictive hiring criteria are used, we get worse employees and, in turn, worse products and services for all of us. That is hardly a formula for American success in an ever more competitive global economy.

Alas, the Obama Administration is extending Disparate Impact's use well beyond employment. For example, the U.S. Office of Education's Civil Rights division is investigating school districts in which African-American students are disproportionately disciplined. While  it's possible that racial discrimination is causal, it's more likely that, because of, for example, historical, cultural, and socioeconomic factors, African-American kids simply misbehave more often.  But because school districts fear having to defend an expensive and difficult burden-of-proof lawsuit, they're ever more likely to establish racial "targets" for discipline thus having to discipline more White and Asian students than they would have and/or to discipline fewer African-American students. Even more likely, they'll keep in school African-American students who deserved suspension or expulsion to protect innocent students from violence. They'll also reallocate yet more resources from the students with the greatest potential to profit to those students, in the form of yet more counseling and other programs. It seems clear to me that, as in employment, the Disparate Impact theory applied to the schools, will, net, do more harm than good.

As I've written before, decisions should more often be made on a cost-benefit basis, considering all the likely short- and long-term outcomes. I believe that if we did that, most of the redistributive "justice" laws and policies would be rejected or modified in favor of policies that distributed resources more purely on merit.