Bias determines outcomes - in journalism

Usually when dealing with reporting on the mythical gender pay gap, it involves innumerate journalists. They are clouded by their ideological engagement with critical theory, their domain knowledge is limited, and their maths and statistics skills are negligible.

This sequence of reports is different. They are still pushing the myth that there is a gender wage gap which is based on misogynistic bias against women. What is different is that, at least with The Atlantic, the journalist knows that is a myth but is pushing it anyway.

The genesis is a new research report, Still a Man's Labor Market: the slowly narrowing gender wage gap by Stephen J. Rose and Heidi I. Hartman. Sponsored by the Institute for Women's Policy Research.

There are two contesting goals which might be crudely summarized as:

People should be compensated based on their group identities and not on their productivity.

People should be compensated based on their individual contribution to productivity.

The, again, crude characterization would be that the first is a traditional neo-marxist/critical theory view of things, deterministic in nature, centralized structure and led by the vanguard. The belief is that we can plan our way towards our goals through coercive exercise of state power over individuals and groups. Progress and order arises from the exercise of expert decision making by a core body on behalf of everyone else, regardless of human rights, rule of law or empirical evidence.

The second tradition is the classical liberal/Age of Enlightenment position with universal human rights, consent of the governed, rule of law, equality before the law, empiricism, logic and rationalism and a dependence on the innovation and emergent order of lightly regulated free markets. Progress and order arises from the exercise of personal freedom by everyone under common laws and through the freedom of the market of commerce and the freedom of the market of ideas.

Under the former model, the fact that there are group differences in compensation is the only evidence necessary to conclude that the Labor Market is a Man's Labor Market, or a Caucasian Labor Market, or a Cis-gender Labor Market, or a Nativist Labor Market, or any other group identity preferred.

Under this model, the relevant vector of analysis is group averages only. That is the approach taken by Rose and Hartman. If you are a critical theory statist serving Mandarin Class ideological fetishes, it is the only approach that can be taken. With this naive analysis, they conclude:

• Women today earn just 49 cents to the typical men’s dollar, much less than the 80 cents usually reported. When measured by total earnings across the most recent 15 years for all workers who worked in at least one year, women workers’ earnings were 49 percent—less than half—of men’s earnings, a wage gap of 51 percent in 2015. Progress has slowed in the last 15 years relative to the preceding 30 years in the study.

• The penalties of taking time out of the labor force are high—and increasing. For those who took just one year off from work, women’s annual earnings were 39 percent lower than women who worked all 15 years between 2001 and 2015, a much higher cost than women faced in the time period beginning in 1968, when one year out of work resulted in a 12 percent cut in earnings. While men are also penalized for time out of the workforce, women’s earnings losses for time out are almost always greater than men’s.

• Strengthening women’s labor force attachment is critical to narrowing the gender wage gap. Despite considerable progress over the last 50 years, 43 percent of today’s women workers had at least one year with no earnings, nearly twice the rate of men. With high penalties for weak labor force attachment, achieving higher lifetime earnings for women will require strengthening women’s attachment to the labor force. Research has shown that such policies as paid family and medical leave and affordable child care, can increase women’s labor force participation and encourage men to share more of the unpaid time spent on family care.

• Strengthening enforcement of equal employment opportunity policies and Title IX in education is also crucial to narrowing the gender wage gap further. Improved enforcement will help women enter higher paying fields that are now, despite decades of progress, still too often off-limits to women.

The flaw in this approach is glaring. To see it, you only have to play with the group identity to see how misleading it is. Try this example:

Young adults today earn just 20 cents to the typical older person's dollar, much less than the 80 cents usually reported. When measured by total earnings across the most recent 15 years for all workers who worked in at least one year, the age cohort of 15-20 year-olds earnings were 20 percent — less than a fifth — of the 40-45 year-old cohort earnings, a wage gap of 80 percent in 2015. Progress has slowed in the last 15 years relative to the preceding 30 years in the study.

I am guestimating the relative wage differentials between young and old but otherwise that characterization is absolutely accurate. Young workers earn less because they are less productive than workers in their prime earning years of 35-45. As the economy has become more global, less regulated, more complex, the wage premium for experience and knowledge (older workers) has become steeper, meaning that the gap between young and old is wider than it used to be.

For a Classical Liberal, there is no mystery here and no obvious evidence of unfairness or discrimination. More productive leads to greater compensation. Whether there is any bias or discrimination involved is a matter that can only be determined by a legal investigation. The fact that there is a gap is not only no basis for bringing a charge of discrimination, it is no surprise at all.

Greater productivity is a function of experience, education attainment, personal behaviors (self-discipline, attention to detail, persistence in the face of challenges, etc.), acquired knowledge, skills, values, number of hours worked, persistence of engagement in the labor force, consistency of focus over a career arc, flexibility and adaptability to unpredictable changes, etc.

The fact that productivity varies between groups is not in itself evidence of bias or discrimination. Younger workers have less experience, fewer skills, less education attainment, less capacity to deal with uncertainties, less duration in the marketplace, etc. That is what leads to their lower productivity, not their status as a member of a youthful cohort.

On average. Whatever the group, there are always individual prodigies that defy the average.

And indeed, whatever the group, there are also always individual instances of discrimination for and against. The question is whether those individual instances of positive and negative bias aggregate into a systemically skewed bias.

The point is that the group identity is not driving the differences in productivity, it is their personal attributes, capabilities and choices.

This research is therefore starting from a false foundation and a flawed methodology. We know there are group differences in productivity. Wage is simply a proxy for that.

The foolishness is clarified by another example.

Under the Rose and Hartman methodology, the following scenario is going to generate a shocking degree of market inequality and gender wage gap.

Individual 1 - Middle aged woman with high school diploma works as a part-time school bus driver in her rural county, putting in 20 hours a week on a fixed schedule at minimum wage with three months off in the summer. She reentered the workforce three years ago after taking time off when her children were born. She earns $7,000 per year.

Individual 2 - Middle aged man with a law degree from Harvard works full-time at a prestigious New York law firm, putting in 70 hours a week, working evenings and weekends, with no breaks through the year and a completely unpredictable schedule. He earns $500,000 a year.

She is only earning 1.4% of him!!!!

Is this evidence of a gender wage gap, with bias and discrimination against women?

Of course not. Differences in productivity drive differences in income. Differences in income are driven by hours, profession, experience, etc.

Essentially Rose and Hartman have used a naive methodology to find the largest gender wage gap possible, given the traditional role choices people make with regard to career focus and family obligations.

But it is not entirely bereft of useful findings. Their second point is an interesting one which has long been conceptually understood but the empirical objective research (and therefore the quantification) is less, and is less settled.

The penalties of taking time out of the labor force are high—and increasing. For those who took just one year off from work, women’s annual earnings were 39 percent lower than women who worked all 15 years between 2001 and 2015, a much higher cost than women faced in the time period beginning in 1968, when one year out of work resulted in a 12 percent cut in earnings. While men are also penalized for time out of the workforce, women’s earnings losses for time out are almost always greater than men’s.

A single year out of the workforce causes a 40% drop in income when later rejoining the labor market compared to those who remained in the workforce. But again, Rose and Hartman are obfuscating. This is not a gender issue. It is well known and documented that any person who exits the workforce for any material amount of time both reduces the chances of returning to the labor market and reduces the wage level at which they return. This is true for both sexes. And it is material at way less than a year. Probability of market reentry measurably declines for every month spent in job search.

These are well known facts and they are known to apply to men and women. It would have been interesting had Rose and Hartman reported that there was a differential impact for men and women for long duration market absence. I can easily construct a scenario that a woman reentering the market after a year's absence might suffer a 50% decline in wage whereas a man might suffer only a 30% decline. While plausible, the question is whether it is real. Had Rose and Hartman reported a differential reentry wage level decline, that would be useful information. I do not know whether to attach any significance to the fact that they do not report such a differential.

The final two points in their summary are policy recommendations, ideological ones at that. Those recommendations do not follow from the empirical research. They are grafted on.

But it does reveal the normative objective of the recommendations. By relying on group average rather than individual productivity, Rose and Hartman are essentially saying that everyone should be compensated equally regardless of individual productivity. By comparing a 20 hour a week worker with minimum wage to a 70 hour a week worker with a huge productivity premium, they are essentially saying that they should be both paid the same in order to remove the gap.

If that sounds grossly Marxist, well, it pretty much is. From each according to their ability and to each according to their needs and all that.

I would suspect that Rose and Hartman would deny any desire for vanguard marxism, but what else is there?

The second possibility is that they are pursuing a milder, social justice/critical theory form of state control. They want individual men and individual women to make similar life choices. Equal numbers choose to be firefighters, equal numbers choose to be nurses, technicians, caregivers, doctors, etc. Equal numbers by education attainment, by degree pursued, by hours work, by duration in the labor market, etc.

Rose and Harman are the Professor Higgins of the academic world.

Why can't a woman be more like a man?

So maybe not marxist. Perhaps just authoritarian totalitarians.

There is a lot of this type of research around - undertaken to support an ideological position rather than reveal how things really work.

On the other hand we have some thirty years and more of evidence from across the OECD which consistently find that when you control for education attainment, years in the labor market, hours worked, etc. virtually all labor variances disappear. Men and women earn the same wage for the same work, for the same productivity. Just as the law requires. And just as economic theory predicts.

Economic theory predicts that workers in a free market will be paid based on their productivity and that therefore men and women will be paid the same for the same work. The law requires equal pay for equal work. And the evidence demonstrates that men and women and being paid the same wage for the same work.

There is no wage gap to be solved as revealed in study after study.

So how is the media reporting this Rose and Hartman research that is a predetermined finding arrived at by a known flawed approach?

This is the aspect I find fascinating. From Women May Earn Just 49 Cents on the Dollar: A new study suggests that the gender wage gap is much wider than previously thought. by Annie Lowrey.

That sounds like your typical Ben Rhodes reporter: 27 years old and knows nothing, and I would add, frequently innumerate.

While typically the case, that is not the case here. Lowery is older, more experienced, Harvard educated, etc. She is not a Ben Rhodes journalist. So why does she sound like one?

Lowery leads with critical theory pablum:

Do women earn 80 cents on the dollar compared with men, as is commonly cited? Or is the pay gap just pennies, as one recent survey found? Or do women earn a shocking 49 cents on the dollar, as calculated by the social scientists Stephen Rose and Heidi Hartmann in a new analysis published by the Institute for Women’s Policy Research?

The answer is all of the above. Each number highlights a different aspect of a complicated and nuanced situation—one reflecting not only sexism and bias but the choices made more or less freely by millions of women in hundreds of thousands of workplaces. Rose and Hartmann’s number is lowest because it is in some ways most holistic.

To her credit, Lowery does appear to recognize that there are at least two explanations for differentials - bias or personal choices. Most journalists simply go with the belief that all differentials are only bias.

But look at the framing. Instead of flawed, Rose and Hartman's approach is more holistic. Really? What does that even mean? Personal choices are "more or less freely" made? Again, classic marxism with the sotto voce suggestion that people are making their own choices but that all humans are products of the system, all are blank slates to be remade/reeducated as required by the state. There is no free will or individual preferences to be acknowledged.

It is only six paragraphs in before Lowery actually acknowledges the cruel truth known from much research replicated all across the OECD for thirty years.

Comparing apples to apples and oranges to oranges, women earn close to what men earn: Women in similar workplaces with similar titles and similar credentials make pretty much what their male peers do, whether they are fast-food employees making close to the minimum wage or corporate executives making hundreds of thousands of dollars a year. This has led some publications to argue that the pay gap is far smaller than generally understood, and yet others to argue that the pay gap is a myth.

She is aware that men and women are being paid equally for equal work which seems like it would be shouted from the hill tops for the progress that represents from a hundred years ago.

After acknowledging the research that shows men and women are paid the same for equal work, in the very next paragraph she argues:

Study after study has shown that women do not get equal pay for equal work, nor do they have access to equal work.

Well, which is it? The equal pay studies are produced by multi-discipline groups (sociologists, economists, historians, etc.) in multiple countries under multiple legal systems and, especially in the past decade, with increasingly rigorous methodologies.

The counter evidence she cites tends to emerge from Gender studies type groups, advocacy groups, has weak methodologies, small sample sizes, unrepresentative samples, relies on self-reporting, fails to replicate, etc.

Lowery is displaying artful deception by her framing and presentation, trying to both acknowledge that her argument which she wants to believe is invalid and yet also be honest in her reporting. She does a better job than most Ben Rhodes reporters but there are still patches of basic ignorance.

When women surge into a given field, pay in that field tends to drop, as if women were some kind of industry-wide reputational pollutant.

Yep, it is known as supply and demand. When supply increases for a fixed demand, wages will fall. No need to invoke this industry-wide reputational pollutant nonsense. Just check any Econ 101 textbook.

But then there is this argument.

Moreover, women’s employment patterns are different from men’s, Rose, a labor economist at the Urban Institute, told me. They are less likely to work full-time and to spend years-long, uninterrupted stretches in the labor force. They are more likely to have to take time off to have a child, or to have to work part-time in order to care for family members.

Many women do not return to work after having children; when they do, in many cases they return to lower-paying jobs with lower earnings trajectories. Rose and Hartmann’s study found that women’s penalties for leaving the workforce have increased: Women who took a year off from work in the 15-year period starting in 2000 had annual earnings 39 percent lower than women who worked continuously over that time period, a gap that was just 12 percent for women working in the 15-year period starting in 1968.

OK. We know that men and women, on average make different group choices. We also know that when they make the same choices, they make the same wage. All is great.

Except in critical theory world, it is not. That cannot be the answer. We need something that justifies the authoritarian interventions which the Mandarin Class wish to impose unilaterally on free citizens.

Leading to this complete non sequitur.

According to Rose, these facts suggest that the most accurate way to compare women’s and men’s earnings is to take the career-long view. “When you look at all women versus all men over time, the gap is 51 cents,” he said, referring to the 15-year figure.

These facts suggest nothing of the kind if you accept a system focused on universal human rights, consent of the governed, freedom, rule of law and equality under the law.

They only suggest that if you want equality based on group identity (the Hartman Rose approach) rather than personal productivity, then you need to get rid of all that freedom, and universal rights, and rule of law and personal choices. Rose and Hartman do not disappoint. The solutions they advance are indeed statist impositions always recommended by the hard left.

What might help close this wide, long earnings chasm? Rose and Hartmann suggest that paid family leave and child-care subsidies would, by making it more financially viable for women to return to the workforce after having kids and by encouraging both men and women to take time off to care for family members. Stronger rules stamping out pay discrimination would help as well. Laws “that prevent employers from asking about past pay will hopefully spread across the United States or become superseded by national legislation,” they write, “establishing a practice that will end the perpetuation of pay discrimination from one job to the next.”

But these policy recommendations involve trade-offs.

Nations with national policies with generous paid family leave and child-care subsidies encouraging/forcing individuals to take time off also have more rigid labor forces with lower income, less innovation, greater national debt, etc. You can choose the subsidized goodies but those subsidies cause a reduction in productivity and innovation and eventually you no longer can afford the subsidies (see the reforms in Scandinavia in the 1990s).

Stronger rules stamping out pay discrimination is kind of a red herring. Today, most large and medium-sized corporations have enterprise-wide ERP systems. It is far harder to hide purposeful and systematic wage discrimination. When discovered, hungry lawyers swoop. To find systemic wage discrimination, you have to get down into very small businesses and even then, variances are as often as not a product of faulty market knowledge.

It is interesting Rose, Hartman, and Lowery do not acknowledge at all one of the most striking features of countries who do have the policies they recommend - Scandinavian countries, the Netherlands, France, etc. They are textbook examples of egalitarian policy, women-friendly policy. And what is the consequence? Women earn less (because they enter low compensation fields more), are less represented in leadership positions, women are less represented in STEM etc. than in the US. More specifically In virtually every field of competitive endeavor in the US, women are between 15-30% of the leadership, 15-30% of the highest compensated, 15-30% of the business executives, 15-30% of the award winners, 15-30% of the STEM workers, 15-30% of the political leaders, 15-30% of the lawyers, etc. In those European countries? 5-15%.

Rose and Hartman have a series of off-the-shelf postmodernist policy recommendations with little or no bearing on reality and with no track record of success in any other country. In order to justify these a priori recommendations they adopt a methodology which gives them a shocking number but which tells you nothing about the prevalence of active discrimination.

Lowery is enough of an honest journalist that she returns to the rub of the issue in her final paragraph.

Even then, of course, a gap might remain. For a variety of reasons women do different work than men, enter different fields than men, work less than men, have different work schedules than men, leave work more often than men, and, yes, are simply paid less for the same work. But public-policy changes would give women more control over their working lives, and would help foster a more equitable workplace. And that would be good for everybody.

We know that the best evidence all indicates that when comparing apples-to-apples in the labor market, most OECD countries demonstrate that companies are operating in compliance with the law against gender discrimination and that men and women earn the same amount for the same work.

Lowery knows this and is reporting on singular research with an inherently and obviously flawed measure which does not compare like-to-like. So Lowery is informed, at least moderately numerate, understands the counter evidence. She knows this and yet we still get this headline Women May Earn Just 49 Cents on the Dollar: A new study suggests that the gender wage gap is much wider than previously thought and we get an argument that is comparing findings from gold-standard research to a blatantly purposeful report intended to support a predetermined position and treating the flawed research as if it were superior.

All kudos for Lowery trying to slip in some facts that allow a careful and informed reader to reach a different conclusion than the misleading headline. Still, it is a massive piece of cognitive pollution, intended to support a political or ideological position, not an effort to find the truth.

Disappointing.