When the first President Bush signed the Civil Rights Act of 1991, many conservatives complained that it was a "quota bill," as Bush had said of an earlier draft. Congressional Democrats and liberal groups hoped that the legislation would, among other things, help provide access for racial minorities and women to job markets that had been traditionally dominated by white males.

Now comes a statistical study concluding that the 1991 act had precisely the opposite effect, by making employers in traditionally white-male industries marginally less likely to hire minorities and women. Indeed, the study claims, the legislation was part of the reason that "a trend toward greater racial and gender integration in labor markets stopped around 1991."

The study, "The Unintended Consequences of the '91 Civil Rights Act," was published in the Summer 2003 issue of Regulation magazine by Paul Oyer, an economics professor at Stanford, and Scott Schaefer, a management professor at Northwestern. It attributes the (rather modest) negative impact on minority and female job seekers to provisions that greatly increased money-damage awards to job-discrimination plaintiffs.

How could the risk of high damage awards for discriminating against minorities and women make employers more hesitant to hire them? Because employers know that far more lawsuits are brought, and far greater damages are awarded, for claims of discrimination in firing than in hiring. So the risk of being sued for turning down a minority or female applicant is dwarfed by the risk of being sued later for firing the same applicant after giving him or her a try.

"The increases in potential damage awards," write Oyer and Schaefer, "coupled with a decades-long trend toward firing-based, and away from hiring-based, employment-discrimination litigation, means the main impact of the act was to increase the costs to employers of dismissing protected workers.... Because [an employer] feels firing-based costs only if it decides to hire, the costs act as an implicit tax on such hiring. Firing-based protections may therefore lead employers to hire fewer protected workers, not more."

What happened to conservatives' fears that the act's more publicized provisions, which made it easier to bring statistically based "disparate impact" claims, would force employers to engage in quota hiring? These fears proved "largely unfounded," conclude Oyer and Schaefer. Any new incentives for employers to hire more minorities and women were apparently overwhelmed by employers' fears of discriminatory-discharge suits by any of the new hires who might not work out.

Census Bureau surveys show that "industries that historically had low minority and female employment shares were hiring more such workers prior to the legislation, but this pattern did not continue into the post-act period," Oyer and Schaefer assert. While the act did not dent the long-standing increase in labor-force participation by women, "the legislation had a mild negative effect on average black employment and hours worked," which "dropped relative to whites starting in 1992."

The Oyer-Schaefer interpretation is, of course, debatable; the census data they cite might also be explained by broader economic trends, such as increasing demand for highly educated workers. But their conclusions ring true in the context of views expressed privately to me by a high-level executive at a sizable company: While every new hire poses some risk of becoming a candidate for firing, the executive said, African-American employees who don't work out have proved far more likely than others to sue, or threaten to sue. The company usually settles quietly by making substantial payments to avoid the bad publicity that even a bogus discrimination lawsuit can bring.

So when evaluating comparably qualified white and black applicants, the company assumes that it will be easier and less costly to fire the white applicant if he or she does not work out. This is a disincentive to increasing black representation in the company's mostly white workforce.

The 1991 act increased monetary remedies for plaintiffs in three ways. It overruled a 1989 Supreme Court decision by authorizing unlimited compensatory and punitive awards for intentional racial discrimination in firing (as well as in hiring) under the Civil Rights Act of 1866. It authorized such damage awards up to a cap of $300,000, depending on the size of the employer, in lawsuits for intentional gender discrimination and racial discrimination in employment under Title VII of the 1964 Civil Rights Act, which had previously limited monetary awards to back pay. And it gave Title VII plaintiffs jury trials.

The act's unintended consequences were not entirely unanticipated. In a Stanford Law Review article half a year before Congress passed the legislation, Stanford Law School professor John J. Donohue III and co-author Peter Siegelman documented a major shift in the nature of job-discrimination lawsuits—as well as a spectacular increase in their number—since 1970: "While most cases formerly attacked discrimination in hiring, today the vast majority of all litigation suits challenge discrimination in discharge." And although the 1964 Civil Rights Act was extremely valuable in breaking down the flagrant discrimination in hiring then practiced by many employers, the authors wrote, the "dramatic shift to firing cases has greatly increased the likelihood that Title VII will create a drag on the hiring of protected workers rather than the positive inducement it originally provided."

This logic implied that amending the law to expose employers to larger damage awards in firing-discrimination, and other, lawsuits—as the 1991 act was soon to do—would make them at least marginally more hesitant to hire minorities and women.

A modest drag on the hiring of minorities and women might be a price worth paying if the 1991 act were an effective remedy for genuinely discriminatory dismissals. But the census data show that "the overall rate of involuntary job loss for black men did not change," according to Oyer and Schaefer. Rather, some employers responded to the 1991 act by delaying planned dismissals of unwanted black employees until the next round of mass layoffs.

Beyond that, many—and perhaps most—discriminatory discharge lawsuits are bogus. Plaintiffs lose more than half of the cases that go to trial, notes Professor Charles A. Shanor, of Emory Law School, who was general counsel of the Equal Employment Opportunity Commission from 1987 to 1990. Title VII plaintiffs must file their claims with the EEOC before going to court. "Most EEOC employees who have investigated a substantial number of cases will tell you that a substantial percentage are not meritorious," Shanor adds. By some estimates, he says, the vast majority of discriminatory-discharge claims are without merit.

This seems intuitively plausible. Some managers no doubt fire people for discriminatory reasons. But wouldn't a really racist employer simply avoid hiring minority employees, rather than hire them and then look for pretexts to fire them weeks, months, or years later?

To the extent that the 1991 act provides financial incentives for fired employees to sue even when they are not victims of discrimination, it benefits the least-deserving minority and female employees at the expense of minority job seekers as a group, especially those struggling to escape unemployment.

"Once the egregious discrimination is gone," Donohue says, "then litigation-based schemes to bring ever more fairness become more burdensome and of dubious effectiveness. Litigation is a crude weapon—you can't perform surgery with a saber. Unfortunately, Title VII has become a matter of religious dogma for many academics, and certainly for those who benefit—plaintiffs' lawyers and consultants—and therefore immune to any type of critical inquiry."

None of this is to suggest that Congress should eliminate damages for discriminatory discharge. Even if the costs of such lawsuits to minorities and women, not to mention employers, have come to exceed their benefits, the social costs of outright repeal—spreading alarm among minority groups and women, aggravating racial tensions, tempting racist whites to engage in overt discrimination, and more—would be greater still.

But the Oyer-Schaefer and Donohue-Siegelman analyses argue that further expansion of employer liability for discriminatory dismissals, which liberal politicians and groups have long sought, could be bad for minority job seekers. These analyses also suggest that legal reforms such as financial penalties to deter bogus claims would be good for the vast majority of minority job seekers; that journalists should be more hesitant to presume the guilt of employers sued for discrimination; and that we should all be wary of authorizing ever more lawsuits to engineer social reforms.