What happened in 1971? Productivity, Griggs v. Duke Power, and the case for merit systems you can defend

What happened in 1971?

There is a popular question on the internet: what the hell happened in 1971? Charts of wages, productivity, housing, and household finances seem to bend around that time, and many people point to the end of the gold standard in August 1971.

My answer is simpler. In the early 1970s, the growth of total factor productivity, the part of economic growth that comes from working smarter rather than just adding more workers or more machines, broke from its postwar trend. By Crawford Law’s calculation from the Penn World Table, it grew about 1.1 percent a year from 1950 to 1973 and about 0.5 percent a year from 1973 to 1995, less than half the earlier pace. It recovered for about a decade from the mid-1990s, then slowed again. Fit one trend line through everything since 1971, recovery included, and productivity has grown about 0.9 percent a year instead of 1.1. That sounds small. Compounded over nearly fifty years, it means that by 2019 U.S. productivity was about 19 percent below where the old trend would have put it.

When productivity growth slows that much for that long, the predictable result is that it takes more work to maintain the same standard of living. Some of the rise of the two-income household reflects expanded opportunity for women, which is a good thing on its own terms. But in my view, much of it reflects a simple necessity: with productivity growing far more slowly than it used to, many families in 2026 need two incomes for the standard of living one income used to buy.

If you have seen one chart about 1971, it is probably a different one. The best known, from the Economic Policy Institute and featured on wtfhappenedin1971.com, shows output per hour and the pay of typical production and non-supervisory workers rising together after World War II, then separating in the early 1970s, with productivity climbing far faster than pay. That gap is real, though economists still argue about how much of it reflects measurement choices, such as which price index is used, whether benefits are counted, and whose pay is measured, and how much reflects a true shift in who gets the gains.

But that chart answers a different question. It is about how the pie is divided. The chart above is about how fast the pie grows. My argument is that the second problem matters at least as much: when productivity growth slows, there is less to divide, however it is divided.

So the real question is why the productivity trend line broke. Economists have offered many answers: the oil shocks, the end of the postwar catch-up, globalization, regulation, a shift toward services that are hard to make more productive, and productivity gains that our statistics fail to measure. All of those may be part of it. I want to add one more, and it is the controversial one.

Griggs v. Duke Power

On March 8, 1971, the Supreme Court decided Griggs v. Duke Power Co. Duke Power required a high school diploma and passing scores on two general aptitude tests for jobs that had previously been closed to Black workers. Those requirements excluded Black applicants at much higher rates, and the company had not shown they predicted job performance. The Court held that Title VII prohibits not only intentional discrimination, but also neutral-looking practices that disproportionately exclude a protected group and are not shown to be related to the job.

That was the birth of disparate impact. Precision matters here: Griggs did not ban testing. Job-related tests remained lawful. But showing that a test is job-related, called validation, became expensive and uncertain. Many employers concluded it was safer to stop using general aptitude tests altogether and lean instead on degrees, credentials, and unstructured interviews: measures that are often less predictive and no less likely to carry bias.

To be clear about where Crawford Law stands: the firm does not recommend bringing back general aptitude or IQ tests. The better tools are the ones described below, structured interviews and job-related work samples, which predict performance well and are far easier to defend.

My hypothesis, and it is only a hypothesis, is that the timing is not a pure coincidence. When an economy loses some of its ability to match the right people to the right jobs, it loses productivity. I cannot prove that Griggs and what followed contributed to the slowdown. But I think it belongs on the list, alongside oil and globalization.

What disparate impact is, today

Disparate impact works in three steps. First, the challenger shows that a practice, such as a test, a degree requirement, a rating system, or a layoff criterion, produces a significant disparity for a protected group. Second, the employer must show the practice is job-related and consistent with business necessity. Third, even then, the challenger can win by showing an equally valid alternative with less disparity that the employer refused to adopt. Congress wrote this framework into Title VII in 1991. Federal guidelines from 1978 use a rule of thumb, the “four-fifths rule,” to flag a selection rate for one group below 80 percent of the highest group’s rate.

Here is the tension that makes people hate HR. The same statute that creates disparate impact also says:

• Nothing in Title VII requires preferential treatment to fix an imbalance in a workforce.
• Adjusting test scores or cutoffs by race, sex, or other protected group is prohibited.
• Quotas based on protected characteristics are, in practice, unlawful.

So employers are told they may not hire by the numbers, and also that they can be liable if the numbers come out wrong. The Supreme Court acknowledged that bind in 2009 in Ricci v. DeStefano, holding that an employer cannot throw out a test’s results because of their racial outcome unless it has a strong basis in evidence that it would otherwise be liable.

The practical result, in my view, is that disparate impact pushes employers toward quota substitutes: informal pressure to make hiring, promotion, ratings, pay, and layoffs come out close to proportional, without ever saying so. Nobody ever admits to a quota. No company would put one in writing, and no manager would say the word out loud. But anyone who has sat through enough hiring, promotion, and layoff meetings knows they exist in practice: numbers watched as targets, candidate slates that “need” a certain mix, and decisions quietly revisited when the totals look wrong. And the math is unforgiving. Most decisions involve small numbers. Slice a company’s data by location, level, job, decision, and group, and some disparity will appear somewhere by chance alone. At some point, at some level, in some location, every employer will be out of step, and exposed to a claim that its own data appears to support.

I do not mind. That is job security for employment lawyers. The government does not mind either. When nearly every company is technically exposed, regulators are in a position of “show me the company and I’ll show you the violation,” which gives them leverage to make companies behave as they want. But the people who actually work mind. HR professionals mind too, because they do not enjoy having to talk out of both sides of their mouths.

The landscape is shifting. A 2025 executive order directed federal agencies to abandon disparate impact enforcement as far as the law allows, and the EEOC stopped investigating charges based only on disparate impact. But disparate impact is still in the statute, private plaintiffs can still sue, Texas has its own version in the Labor Code, and a future administration can reverse course. The risk is lower today. It is not gone.

Real discrimination destroys productivity

Let me be crystal clear. Crawford Law will fight tooth and nail to root out discrimination based on anything other than merit, wherever it appears and whoever it targets. That is not a policy preference. It is black-letter law: Title VII, the Age Discrimination in Employment Act, the Americans with Disabilities Act, and Chapter 21 of the Texas Labor Code make it unlawful to hire, fire, pay, promote, or otherwise treat someone differently because of race, color, religion, sex, national origin, age, disability, or another protected characteristic. And as the Supreme Court confirmed unanimously in 2025, those protections apply to every employee under the same standard. Judging people on merit is exactly what the law demands.

None of this is an argument for tolerating discrimination. The opposite. Discrimination based on anything other than merit is deeply harmful to productivity, because it puts the wrong people in the wrong jobs and wastes talent the economy needs. One of the most important economic studies of the last decade found that better allocation of talent, as barriers fell for women and Black Americans entering occupations they had been excluded from, accounts for roughly 20 to 40 percent of the growth in U.S. output per person from 1960 to 2010.

Read that carefully, because it cuts both ways, and that is the point. Opening jobs to everyone made the country richer. Matching people to jobs on anything other than ability, whether that is prejudice or a quota, makes it poorer. Real discrimination has to be rooted out completely. Inclusion and talent development are essential, because talent is spread across every group and most of it goes undeveloped. And the goal has to be selection on merit, measured well.

The ideal everyone actually shares

A society that is blind to color, gender, sexual orientation, and every other characteristic that has nothing to do with doing the job well is an ideal that people across the political spectrum share, even when they argue about how to get there. A great company does not have to pick a side in that argument. It can pursue the ideal directly:

1. Understand the biases. Find where bias can enter: who gets referred, how resumes are screened, who interviews, how ratings are calibrated, who gets stretch assignments.
2. Design them out. The best-known example is orchestras that began auditioning musicians behind a screen, which, by one well-known study, increased the odds that women advanced. Researchers still debate the size of the effect, but the principle is sound: when you cannot see the person, you judge the performance. Structured interviews, work samples, scored criteria, and blind review of early-stage materials apply the same idea at work.
3. Use selection tools that work. The most recent large review of hiring research found that structured interviews and job-related work samples are among the strongest predictors of performance, and they generally produce smaller group differences than many alternatives. Validate them for your jobs.
4. Develop talent everywhere. Recruit from more places, train and mentor broadly, and build paths that let anyone move up. See fishing in more ponds.
5. Keep the data, and have the courage to defend it. Monitor outcomes through counsel. When a disparity appears, find out whether it reflects bias, which must be fixed, or a small sample and a valid, job-related process, which can be defended. A company that has validated its tools and documented its decisions is in a position to defend a merit-based outcome even when a small sample does not match what a disparate impact claim would demand.

Getting the trend line back

I believe smart, merit-based systems, built to eliminate bias and backed by the data and the courage to defend them, are part of how this country gets productivity back toward its pre-1972 trend line. Merit has to be the forcing function in every business, not office politics, not prejudice, and not a numbers game. See treating your stars like stars and incentive pay done right.

Crawford Law views that mission as essential, and good for everyone: for the employees who deserve to be judged on what they do, for the HR professionals who deserve to say one thing and mean it, and for the companies that need every bit of talent they can find.