Tag Archives: Automation

Ubiquitous Guilt: EEOC Disparate Impact Liability

22 Thursday Sep 2022

Posted by in Discrimination, Regulation

≈ 3 Comments

Tags

, , , , , , , , , , , , , , , , , , , , , , , , , , ,

A key part of the Civil Rights Act of 1964 was Title VII, which dealt with employment discrimination. Title VII applied only to intentional discrimination, but it didn’t take long for the Equal Employment Opportunity Commission (EEOC), the agency charged with administering Title VII, to find ways to expand the scope of its enforcement mandate under the law. The EEOC eventually managed to convince virtually all parties, including employers, employees, job applicants, attorneys, and even the courts, that the law prohibited employment practices having disparate impacts on groups protected from actual discrimination under the law. Predictably, this warped reinterpretation created severe distortions to the efficiency and fairness of labor market outcomes .

Another Rogue Agency

On the EEOC’s complete and erroneous reimagining of Title VII, Gail Heriot’s “Title VII Disparate Impact Liability Makes Almost Everything Presumptively Illegal” is a must read. Heriot is a Professor at the University of San Diego School of Law and is a member of the U.S. Commission on Civil Rights. This post attempts to summarize most of the important points in Heriot’s paper, so if you don’t have time for Heriot’s paper, read on. All errors are mine, of course!

Heriot provides an incredible case study on the dangers of regulatory overreach. She first discusses the EEOC’s blatant usurpation of Congressional power:

“It is hardly surprising that EEOC officials would undertake to publish answers to the questions they were hearing repeatedly…. But publishing such ‘guidances’ also had the potential to spin out of control. The temptation would always be to use them to establish what the EEOC staff wanted the law to be rather than what it was. Instead of interpreting Title VII in good faith, guidances would soon become quasi-legislation—disguised as interpretation, but in reality imposing new duties on employers not found in Title VII itself.

None of this should be surprising. It is in the nature of bureaucracy. It naturally seeks to expand its powers, often beginning by occupying niches that are otherwise unoccupied. Over time, a little power often becomes a lot of power. What is surprising is how upfront EEOC officials were about their tactics in accumulating that power.”

Having gone this far, one might be tempted to ask the EEOC what limiting principle they actually apply to determine whether various employment and hiring practices are permissible. Are level of education, industry experience, and tests of physical and cognitive faculties verboten? The answer that is there is no consistent, limiting principle. Instead, Heriot says the EEOC “picks its battles” (see below). She also describes the EEOC’s adoption of a so-called “four-fifths rule”, which is about as arbitrary as it gets. It means the EEOC will challenge an employment practice only if it leads to a selection of any protected group at a rate less than 80% of the most-selected group. That is, the “disparate impact” must be less than 20% to rule out a challenge. This rule appears nowhere in Title VII.

Job Qualifications? You’re Guilty!

Unfortunately, as Heriot takes pains to demonstrate, it’s virtually impossible to identify a hiring guideline or method of employee assessment that does not have a disparate impact. The examples she provides on pp. 34 – 37 of her paper, and on p. 40, are convincing. Furthermore, the EEOC’s “four-fifths” rule hardly narrows the potential for challenge at all.

“Selection rates of less than four-fifths relative to the group with the highest rate are extremely common. Just as everything or nearly everything has a disparate impact, everything or nearly everything has a selection rate that fails the ‘four fifths rule’ for some race, color, religion, sex, or national origin group.”

So the EEOC is allowed to operate with tremendous discretion. Again, Heriot says the agency “picks its battles”, focusing on challenges to screening tools like “written tests, physical strength and endurance tests, criminal background tests [sic], high school diploma requirements, personal credit histories, residency requirements, and a few others.”

This regulatory environment encourages employers to keep job requirements vague, sometimes to the point at which potential applicants might not be sure what the job qualifications really are, or exactly what the job function entails. One upshot is that this makes it harder to detect and prove actual discrimination, and it often leads to more arbitrary decisions by hiring managers, which may, in fact, involve real discrimination, including nepotism and/or cronyism.

Unbiased Intent Doesn’t Matter

Heriot points to a disastrous decision by the Supreme Court that, perhaps unintentionally, helped legitimize the concept of disparate impact as legal doctrine, and as a valid cause of action by plaintiffs against employers. In Griggs v. Duke Power Co. (1971), the Court rejected the premise that an employer’s innocence with respect to their intent to discriminate was an inadequate defense of an employment practice that had adverse consequences to a protected group. Heriot quotes the opinion of Chief Justice Warren Burger:

“… good intent or absence of discriminatory intent does not redeem…. Congress directed the thrust of the Act to the consequences of employment practices, not simply the motivation.”

It’s as if the Court convinced itself that adverse consequences prove actual discrimination, even when there is no intent to discriminate. The Court also emphasized that it’s decision was based on “general deference” to the EEOC! And this was years before the unfortunate Chevron Doctrine (judicial deference to administrative agencies on interpretation of law) was formally established by the Court. Heriot and others assert that the decision in Griggs would have astonished the authors of Title VII.

Heriot also discusses changes in the treatment of “business necessity” as a defense against complaints of disparate impact. It is generally the employer’s burden to show the “necessity” of a challenged hiring practice. “Necessity” was the subject of several Supreme Court decisions in the 1970s and 1980s, but the Court stopped short of requiring an employer to show that a practice was “essential”. In one case, the court shifted some of the burden back onto the plaintiff to show that a practiced lacked necessity. In 1990, there was concern in the Bush Administration and Congress that the difficulty of proving business necessity would eventually lead to the adoption of racial quotas by employers in order to prevent EEOC challenges, though the authors of Title VII had staunchly opposed quotas. While the original hope was that the Civil Rights Act of 1991 would resolve questions about “business necessity” and the burden of proof, it did not. Instead, it can be said that it legitimized disparate impact liability, with conditions. The standard for proving necessity, based on Court decisions, evolved to become more strict with time. There are cases in which courts seem to have left the EEOC to define “business necessity”, as if the EEOC would be in a better position to do that than the business itself!

Inviting Discrimination

Heriot devotes part of her paper to the perverse effects of disparate impact. When employers are faced with prohibitions or the threat of action against a certain practice, whether it be tests of aptitude, strength, or screening on criminal or credit records, they may abandon those devices and opt instead for “informal” proxies. The use of proxies, however, often leads to instances of actual discrimination, whether born of conscious or unconscious bias on the part of hiring managers.

Heriot provides a number of examples of the proxy phenomenon, some of which have been confirmed by empirical research. For example, an employer interviewing candidates for a job that requires math proficiency might reasonably use a test of math skill as a key criterion. If such a test is prohibited, the hiring manager might be tempted to hire an Asian candidate, since Asians have a reputation for good math skills. Similarly, an applicant of West European ancestry might be favored for a position requiring excellent grammar skills, absent the ability to explicitly test grammatical skill. Candidates for a job requiring a certain level of physical strength could be evaluated by various tests of strength, but barring that, a hiring manager might be inclined to hire based on gender.

When criminal background checks are prohibited, employers might be tempted to use proxies such as gender and race as a substitute. Likewise, if it’s forbidden to check a candidate’s credit record to gauge reliability, other proxies might lead to discrimination against members of protected classes. Needless to say, these kinds of outcomes are precisely the opposite of what the EEOC hopes to achieve.

As Heriot further notes, the outcomes can be much systematic and destructive than a bit of one-off discrimination in hiring, promotion, pay raises, or task assignment. These may inflict damage reaching well beyond having the wrong people gaining favorable labor market outcomes. For example, an employer might choose to relocate operations to a “safer” or more affluent community, barring an ability to perform criminal background or credit checks. Or businesses might decide to substitute capital for labor, given the interference in their attempts to identify the best job candidates. The difficulty in screening also creates an incentive to automate, just as premature automation is becoming more common with rising wage floors imposed by government.

Killing Jobs and Competition

Like many forms of regulation, however, large firms in less competitive industries are usually better positioned to survive EEOC scrutiny than smaller firms in competitive markets. Indeed, we often see large market players embrace regulation because it gives them a competitive advantage over smaller rivals. In this case, we see large firms adopting their own diversity, equity, and inclusion (DEI) goals. This is not solely related to the threat of EEOC challenges, however. Private lawsuits alleging discrimination or disparate impact are also a concern, as is pleasing activists inside and outside the company. Nevertheless, as Christopher Rufo reveals, there is growing push-back against the corporate DEI regime. Let’s hope it continues to gain traction.

Unconstitutional Executive Discretion

Heriot also dedicates part of her paper to constitutional issues related to the EEOC’s broad discretion in the application of disparate impact to employment practices. For one thing, disparate impact is a direct source of discrimination: when members of “protected groups” are awarded opportunities based on the possibility of disparate statistical outcomes, it means the majority candidates are denied those opportunities, no matter their qualifications. This is outright discrimination, and it’s instigation by a federal agency constitutes an explicit denial of equal protection under the law.

It should be no surprise that many consider disparate impact actions against employers to be denials of due process. Furthermore, when a federal agency like the EEOC exercises broad discretion, the so-called non-delegation doctrine should come into play. That is, the EEOC makes judgements on matters that are not necessarily authorized Congress. Thus, there are legitimate questions as to whether the EEOC’s discretion is a violation of the separation of powers. Granted, the courts have long deferred to administrative agencies in the interpretation of enabling statutes, but the Supreme Court has taken a new tack under Chief Justice Roberts. In some recent decisions, the Court has relied on a new “major questions” doctrine to place certain limits on executive discretion.

Conclusion

Hiring? Creating jobs? Better not get picky about checking your applicants’ skills and backgrounds or you risk liability for contributing to the statistical malaise of one, or of many, protected groups. That’s how it is under “disparate impact” rules imposed by the EEOC. The success of your business be damned!

Gail Heriot’s excellent paper details the way in which the EEOC transformed the meaning of its enabling legislation, expanding its reign over employment practices across the nation. She demonstrates the breadth of disparate impact rules with examples showing that virtually any attempt at systematic screening of job applicants can be held to be illegal. Your intent to hire the most qualified candidate without bias doesn’t matter, under an insane Supreme Court decision that buttressed the EEOC’s authority. As Heriot says, “… everything is presumptively illegal”. She also describes how disparate impact liability leads to employment decisions based on proxy criteria, which often lead to actual (even if unintended) discrimination. Further unintended consequences are the possibility of larger job losses in minority communities and less competition in product and labor markets. Finally, Heriot delineates several constitutional violations inherent in broad EEOC discretion and the enforcement of disparate impact.

One day a court challenge to the EEOC and disparate impact liability might rise to the level of the Supreme Court. Justice Antonin Scalia expected it, but it still hasn’t come before the Court. It should! Another way to do battle against the EEOC’s scourge is to challenge corporations who cow-tow to activists and to the EEOC with their own DEI initiatives. This manifestation of stakeholder capitalism is a cancer on the wealth and productivity of the U.S. economy, resting side-by-side with disparate impact liability.

Missouri Prop B: the Unintended Consequences of Wishful Thinking

04 Sunday Nov 2018

Posted by in Labor Markets, Living Wage, Minimum Wage, Uncategorized

≈ Leave a comment

Tags

, , , , , , , , , , ,

Proposition B sounds really good to many Missouri voters: all we have to do to help low-wage workers is declare that they must be paid a higher wage. That’s the pitch, of course. But voters should hear the cruel truth about the unintended consequences of this well-intentioned and ill-considered proposition on the ballot this week:

  1. Businesses are likely to increase prices to compensate for a higher mandated wage, which hurts all consumers, but especially the poor.
  2. Some low-skilled job losses or lost hours are assured, and they will hit the very least-skilled the hardest. No matter the legal minimum, the real minimum wage is always zero.
  3. Such job losses have long-term consequences: lost job experience that the least-skilled desperately need to get ahead.
  4. The harms will have a disparate impact on minorities.
  5. Large employers can substitute capital for low-skilled labor: automated kiosks to take orders and increasingly sophisticated robots to perform tasks. Again, the real minimum wage is always zero. As I’ve said before on this blog, automate no job before its time. But that’s what Prop B will encourage.
  6. Employers can make other compensatory changes. That includes reduced fringe benefits and break times, increased production quotas, and less desirable shifts for minimum wage workers.
  7. A large share of the presumed beneficiaries of a higher minimum wage are not impoverished. Many are teenagers or young adults living with their parents.
  8. All of the preceding points argue that an increase in the minimum wage is not an effective method of targeting poverty reduction. In fact, the harm it inflicts is targeted at the most needy. 
  9. Small employers have less flexibility than large employers, and Prop B would place them at a competitive disadvantage. To that extent, a higher wage floor is most damaging to “mom & pop”, locally-owned businesses, and their employees. Again, the real minimum wage is always zero.

At least 24 earlier posts appear on this blog covering the topic of minimum wages. You can see most of them here. The points above are explored in more detail in those posts.

William Evan and David MacPherson of the Show-Me Institute have estimated the magnitude of the harms that are likely to result if Prop B is approved by voters on November 6, and they are significant. The voters of Missouri should not be seeking ways to make the state’s business environment less competitive.

Voters should keep in mind that wages in an unfettered market reflect the realities of labor demand and labor supply. Wages and other forms of compensation reflect the actual quantity, quality and productivity of available labor supplies. And for unskilled labor, which is often supplied by those who lack experience, a wage that matches their marginal productivity is one that provides that valuable experience. The last thing they need is for tasks requiring little skill to be performed by more experienced employees, or by machines. We cannot wish away these realities, and we cannot declare them suspended by law. Such efforts will have winners and losers, of course, though the former might not ever recognize the ephemeral nature of their gains. And as long as there is freedom of private decision-making, the consequences of such legal efforts will cause harm to those least able to withstand it.

The Comparative Human Advantage

10 Thursday Aug 2017

Posted by in Automation, Technology, Tradeoffs

≈ Leave a comment

Tags

, , , , , , , , , ,

There are so many talented individuals in this world, people who can do many things well. In fact, they can probably do everything better than most other people in an absolute sense. In other words, they can produce more of everything at a given cost than most others. Yet amazingly, they still find it advantageous to trade with others. How can that be?

It is due to the law of comparative advantage, one of the most important lessons in economics. It’s why we specialize and trade with others for almost all of ours needs and wants, even if we are capable of doing all things better than them. Here’s a simple numerical example… don’t bail out on me (!):

  • Let’s say that you can produce either 1,000 bushels of barley or 500 bushels of hops in a year, or any combination of the two in those proportions. Each extra bushel of hops you produce involves the sacrifice of two bushels of barley.
  • Suppose that I can produce only 500 bushels of barley and 400 bushels of hops in a year, or any combination in those proportions. It costs me only 1.25 bushels of barley to produce an extra bushel of hops.
  • You can produce more hops than I can, but hops are costlier for you at the margin: 2 bushels of barley to get an extra bushel of hops, more than the 1.25 bushels it costs me.
  • That means you can probably obtain a better combination (for you) of barley and hops by specializing in barley and trading some of it to me for hops. You don’t have to do everything yourself. It’s just not in your self-interest even if you have an absolute advantage over me in everything!

This is not a coincidental outcome. Exploiting opportunities for trade with those who face lower marginal costs effectively increases our real income. In production, we tend to specialize — to do what we do — because we have a comparative advantage. We specialize because our costs are lower at the margin in those activities. And that’s also what motivates trade with others. That’s why nations should trade with others. And, as I mentioned about one week ago here, that’s why we have less to fear from automation than many assume.

Certain tasks will be automated as increasingly productive “robots” (or their equivalents) justify the costs of the resources required to produce and deploy them. This process will be accelerated to the extent that government makes it appear as if robots have a comparative advantage over humans via minimum wage laws and other labor market regulations. As a general rule, employment will be less vulnerable to automation if wages are flexible. 

What if one day, as Elon Musk has asserted, robots can do everything better than us? Will humans have anywhere to work? Yes, if human labor is less costly at the margin. Once deployed, a robot in any application has other potential uses, and even a robot has just 24 hours in a day. Diverting a robot into another line of production involves the sacrifice of its original purpose. There will always be uses in which human labor is less costly at the margin, even with lower absolute productivity, than repurposing a robot or the resources needed to produce a new robot. That’s comparative advantage! That will be true for many of the familiar roles we have today, to say nothing of the unimagined new roles for humans that more advanced technology will bring.

Some have convinced themselves that a fully-automated economy will bring an end to scarcity itself. Were that to occur, there would be no tradeoffs except one kind: how you use your time (barring immortality). Superabundance would cause the prices of goods and services to fall to zero; real incomes would approach infinity. In fact, income as a concept would become meaningless. Of course, you will still be free to perform whatever “work” you enjoy, physical or mental, as long as you assign it a greater value than leisure at the margin.

Do I believe that superabundance is realistic? Not at all. To appreciate the contradictions inherent in the last paragraph, think only of the scarcity of talented human performers and their creativity. Perhaps people will actually enjoy watching other humans “perform” work. They always have! If the worker’s time has any other value (and it is scarce to them), what can they collect in return for their “performance”? Adulation and pure enjoyment of their “work”? Some other form of payment? Not everything can be free, even in an age of superabundance.

Scarcity will always exist to one extent or another as long as our wants are insatiable and our time is limited. As technology solves essential problems, we turn our attention to higher-order needs and desires, including various forms of risk reduction. These pursuits are likely to be increasingly resource intensive. For example, interplanetary or interstellar travel will be massively expensive, but they are viewed as desirable pursuits precisely because resources are, and will be, scarce. Discussions of the transition of civilizations across the Kardashev scale, from “Type 0” (today’s Earth) up to “Type III” civilizations, capable of harnessing the energy equivalent of the luminosity of its home galaxy, are fundamentally based on presumed efforts to overcome scarcity. Type III is a long way off, at best. The upshot of ongoing scarcity is that opportunity costs of lines of employment will remain positive for both robots and humans, and humans will often have a comparative advantage.

The Tyranny of the Job Saviors

17 Monday Jul 2017

Posted by in Automation, Free markets, Technology

≈ Leave a comment

Tags

, , , , , , , , , , , , ,

Many jobs have been lost to technology over the last few centuries, yet more people are employed today than ever before. Despite this favorable experience, politicians can’t help the temptation to cast aspersions at certain production technologies, constantly advocating intervention in markets to “save jobs”. Today, some serious anti-tech policy proposals and legislative efforts are underway: regional bans on autonomous vehicles, “robot taxes” (advocated by Bill Gates!!), and even continuing legal resistance to technology-enabled services such as ride sharing and home sharing. At the link above, James Pethokoukas expresses trepidation about one legislative proposal taking shape, sponsored by Senator Maria Cantwell (D-WA), to create a federal review board with the potential to throttle innovation and the deployment of technology, particularly artificial intelligence.

Last week I mentioned the popular anxiety regarding automation and artificial intelligence in my post on the Universal Basic Income. This anxiety is based on an incomplete accounting of the “seen” and “unseen” effects of technological advance, to borrow the words of Frederic Bastiat, and of course it is unsupported by historical precedent. Dierdre McCloskey reviews the history of technological innovations and its positive impact on dynamic labor markets:

“In 1910, one out of 20 of the American workforce was on the railways. In the late 1940s, 350,000 manual telephone operators worked for AT&T alone. In the 1950s, elevator operators by the hundreds of thousands lost their jobs to passengers pushing buttons. Typists have vanished from offices. But if blacksmiths unemployed by cars or TV repairmen unemployed by printed circuits never got another job, unemployment would not be 5 percent, or 10 percent in a bad year. It would be 50 percent and climbing.

Each month in the United States—a place with about 160 million civilian jobs—1.7 million of them vanish. Every 30 days, in a perfectly normal manifestation of creative destruction, over 1 percent of the jobs go the way of the parlor maids of 1910. Not because people quit. The positions are no longer available. The companies go out of business, or get merged or downsized, or just decide the extra salesperson on the floor of the big-box store isn’t worth the costs of employment.“

Robert Samuelson discusses a recent study that found that technological advance consistently improves opportunities for labor income. This is caused by cost reductions in the innovating industries, which are subsequently passed through to consumers, business profits, and higher pay to retained workers whose productivity is enhanced by the improved technology inputs. These gains consistently outweigh losses to those who are displaced by the new capital. Ultimately, the gains diffuse throughout society, manifesting in an improved standard of living.

In a brief, favorable review of Samuelson’s piece, Don Boudreaux adds some interesting thoughts on the dynamics of technological advance and capital-labor substitution:

“… innovations release real resources, including labor, to be used in other productive activities – activities that become profitable only because of this increased availability of resources.  Entrepreneurs, ever intent on seizing profitable opportunities, hire and buy these newly available resources to expand existing businesses and to create new ones.  Think of all the new industries made possible when motorized tractors, chemical fertilizers and insecticides, improved food-packaging, and other labor-saving innovations released all but a tiny fraction of the workforce from agriculture.

Labor-saving techniques promote economic growth not so much because they increase monetary profits that are then spent but, instead, because they release real resources that are then used to create and expand productive activities that would otherwise be too costly.”

Those released resources, having lower opportunity costs than in their former, now obsolete uses, can find new and profitable uses provided they are priced competitively. Some displaced resources might only justify use after undergoing dramatic transformations, such as recycling of raw components or, for workers, education in new fields or vocations. Indeed, some of  those transformations are unforeeeable prior to the innovations, and might well add more value than was lost via displacement. But that is how the process of creative destruction often unfolds.

A government that seeks to intervene in this process can do only harm to the long-run interests of its citizens. “Saving a job” from technological displacement surely appeals to the mental and emotive mindset of the populist, and it has obvious value as a progressive virtue-signalling tool. These reactions, however, demonstrate a perspective limited to first-order, “seen” changes. What is less obvious to these observers is the impact of politically-induced tech inertia on consumers’ standard of living. This is accompanied by a stultifying impact on market competition, long-run penalization of the most productive workers, and a degradation of freedom from restraints on private decision-makers. As each “visible” advance is impeded, the negative impact compounds with the loss of future, unseen, but path-dependent advances that cannot ever occur.

The Real Minimum Wage Is Always Zero

30 Friday Jun 2017

Posted by in Living Wage, Minimum Wage

≈ 1 Comment

Tags

, , , , , , , , , , ,

Min Wage Denmark

A minimum wage study from Denmark reinforces the findings of the Seattle study released this week by economists at the University of Washington. Both studies conclude that increases in the minimum wage have negative effects on low-earners, at least for large increases in wage floors of the type advocated by “living wage” proponents. Alex Tabarrok provided commentary of both studies this week on the Marginal Revolution blog.

The Seattle study found that both employment and hours worked declined substantially among low-wage workers following the city’s minimum wage hikes. This became particularly clear after the most recent increase from $11 to $13 per hour. The average low-wage worker in Seattle lost $125 per month, according to the findings. The study has generally been praised for its detailed data and careful methodology.

The Danish study took advantage of the fact that the minimum wage rises by 40% on a worker’s 18th birthday. The chart above pretty much boils down the results. Employment drops by a third at age 18. Even worse, Tabarrok notes that after one year, 40% of workers who lose their jobs at age 18 are still unemployed, while 75% of those who keep their jobs at 18 are still employed. The fate of these two groups is likely driven by a gap in talent and skills, and that gap can only expand as the least-skilled are idled.

The minimum wage is a misguided policy that hurts those who can least afford it: low-wage, low-skilled workers. Firms forced to adjust to the higher mandated wage are worse off as well, not to mention their customers, who are likely to face higher prices and degraded service levels. Even those who remain employed at the minimum wage might suffer under less generous job perks and working conditions. Today, large increases in the wage floor can be expected to bring premature automation of jobs.

In the real world, workers of low skill vary tremendously in their actual ability, prior training and discipline to perform in a structured environment. Many of these individuals simply cannot add value over and above the legal wage. Some are simply incapable of understanding the demands of arriving on time and delivering effort over the course of a work day. Hiring firms cannot easily discern these differences up-front from social cues. They might try, however, which could lead to decisions that are unfair to some individuals. An even higher minimum wage makes these decisions all the more difficult and risky, and forecloses opportunities to a broader swath of low-skilled workers, consigning them to dependency on family or the state.

The minimum wage has always had appeal as an exclusionary tactic by higher-paid union workers. Cowed by its ostensible first-order effects on worker incomes, the left latched onto it as a fundamentally just policy. The negative second-order effects are predictable however. The economic evidence has been piling up, while methodological flaws in an earlier, prominent study finding the opposite in the 1990s have been exposed. As a policy, the minimum wage is unjust in its effects on the incomes of low-skilled workers and on their ability to gain valuable work experience, and on businesses attempting to deliver value to their customers.

Note: I believe Don Boudreaux should be credited with the phrase used in the title of this post, which I’m sure I’ve quoted before. For more background on minimum wage effects, see these earlier posts on Sacred Cow Chips. There are 20 posts with that tag, and some are more focused on the minimum wage than others, so keep scrolling!

Embracing the Robots

03 Friday Mar 2017

Posted by in Automation, Labor Markets, Technology

≈ 1 Comment

Tags

, , , , , , , , , , , , , , , , ,

automation84s

Machines have always been regarded with suspicion as a potential threat to the livelihood of workers. That is still the case, despite the demonstrated power of machines make life easier and goods cheaper. Today, the automation of jobs in manufacturing and even service jobs has raised new alarm about the future of human labor, and the prospect of a broad deployment of artificial intelligence (AI) has made the situation seem much scarier. Even the technologists of Silicon Valley have taken a keen interest in promoting policies like the Universal Basic Income (UBI) to cushion the loss of jobs they expect their inventions to precipitate. The UBI is an idea discussed in last Sunday’s post on Sacred Cow Chips. In addition to the reasons for rejecting that policy cited in that post, however, we should question the premise that automation and AI are unambiguously job killing.

The same stories of future joblessness have been told for over two centuries, and they have been wrong every time. The vulnerability in our popular psyche with respect to automation is four-fold: 1) the belief that we compete with machines, rather than collaborate with them; 2) our perpetual inability to anticipate the new and unforeseeable opportunities that arise as technology is deployed; 3) our tendency to undervalue new technologies for the freedoms they create for higher-order pursuits; and 4) the heavy discount we apply to the ability of workers and markets to anticipate and adjust to changes in market conditions.

Despite the technological upheavals of the past, employment has not only risen over time, but real wages have as well. Matt Ridley writes of just how wrong the dire predictions of machine-for-human substitution have been. He also disputes the notion that “this time it’s different”:

“The argument that artificial intelligence will cause mass unemployment is as unpersuasive as the argument that threshing machines, machine tools, dishwashers or computers would cause mass unemployment. These technologies simply free people to do other things and fulfill other needs. And they make people more productive, which increases their ability to buy other forms of labour. ‘The bogeyman of automation consumes worrying capacity that should be saved for real problems,’ scoffed the economist Herbert Simon in the 1960s.“

As Ridley notes, the process of substituting capital for labor has been more or less continuous over the past 250 years, and there are now more jobs, and at far higher wages, than ever. Automation has generally involved replacement of strictly manual labor, but it has always required collaboration with human labor to one degree or another.

The tools and machines we use in performing all kinds of manual tasks become ever-more sophisticated, and while they change the human role in performing those tasks, the tasks themselves largely remain or are replaced by new, higher-order tasks. Will the combination of automation and AI change that? Will it make human labor obsolete? Call me an AI skeptic, but I do not believe it will have broad enough applicability to obviate a human role in the production of goods and services. We will perform tasks much better and faster, and AI will create new and more rewarding forms of human-machine collaboration.

Tyler Cowen believes that AI and  automation will bring powerful benefits in the long run, but he raises the specter of a transition to widespread automation involving a lengthy period of high unemployment and depressed wages. Cowen points to a 70-year period for England, beginning in 1760, covering the start of the industrial revolution. He reports one estimate that real wages rose just 22% during this transition, and that gains in real wages were not sustained until the 1830s. Evidently, Cowen views more recent automation of factories as another stage of the “great stagnation” phenomenon he has emphasized. Some commenters on Cowen’s blog, Marginal Revolution, insist that estimates of real wages from the early stages of the industrial revolution are basically junk. Others note that the population of England doubled during that period, which likely depressed wages.

David Henderson does not buy into Cowans’ pessimism about transition costs. For one thing, a longer perspective on the industrial revolution would undoubtedly show that average growth in the income of workers was dismal or nonexistent prior to 1760. Henderson also notes that Cowen hedges his description of the evidence of wage stagnation during that era. It should also be mentioned the share of the U.S. work force engaged in agricultural production was 40% in 1900, but is only 2% today, and the rapid transition away from farm jobs in the first half of the 20th century did not itself lead to mass unemployment nor declining wages (HT: Russ Roberts). Cowen cites more recent data on stagnant median income, but Henderson warns that even recent inflation adjustments are fraught with difficulties, that average household size has changed, and that immigration, by adding households and bringing labor market competition, has had at least some depressing effect on the U.S. median wage.

Even positive long-run effects and a smooth transition in the aggregate won’t matter much to any individual whose job is easily automated. There is no doubt that some individuals will fall on hard times, and finding new work might require a lengthy search, accepting lower pay, or retraining. Can something be done to ease the transition? This point is addressed by Don Boudreaux in another context in “Transition Problems and Costs“. Specifically, Boudreaux’s post is about transitions made necessary by changing patterns of international trade, but his points are relevant to this discussion. Most fundamentally, we should not assume that the state must have a role in easing those transitions. We don’t reflexively call for aid when workers of a particular firm lose their jobs because a competitor captures a greater share of the market, nor when consumers decide they don’t like their product. In the end, these are private problems that can and should be solved privately. However, the state certainly should take a role in improving the function of markets such that unemployed resources are absorbed more readily:

“Getting rid of, or at least reducing, occupational licensing will certainly help laid-off workers transition to new jobs. Ditto for reducing taxes, regulations, and zoning restrictions – many of which discourage entrepreneurs from starting new firms and from expanding existing ones. While much ‘worker transitioning’ involves workers moving to where jobs are, much of it also involves – and could involve even more – businesses and jobs moving to where available workers are.“

Boudreaux also notes that workers should never be treated as passive victims. They are quite capable of acting on their own behalf. They often act out of risk avoidance to save their funds against the advent of a job loss, invest in retraining, and seek out new opportunities. There is no question, however, that many workers will need new skills in an economy shaped by increasing automation and AI. This article discusses some private initiatives that can help close the so-called “skills gap”.

Crucially, government should not accelerate the process of automation beyond its natural pace. That means markets and prices must be allowed to play their natural role in directing resources to their highest-valued uses. Unfortunately, government often interferes with that process by imposing employment regulations and wage controls — i.e., the minimum wage. Increasingly, we are seeing that many jobs performed by low-skilled workers can be automated, and the expense of automation becomes more worthwhile as the cost of labor is inflated to artificial levels by government mandate. That point was emphasized in a 2015 post on Sacred Cow Chips entitled “Automate No Job Before Its Time“.

Another past post on Sacred Cow Chips called “Robots and Tradeoffs” covered several ways in which we will adjust to a more automated economy, none of which will require the intrusive hand of government. One certainty is that humans will always value human service, even when a robot is more efficient, so there will be always be opportunities for work. There will also be ways in which humans can compete with machines (or collaborate more effectively) via human augmentation. Moreover, we should not discount the potential for the ownership of machines to become more widely dispersed over time, mitigating the feared impact of automation on the distribution of income. The diffusion of specific technologies become more widespread as their costs decline. That phenomenon has unfolded rapidly with wireless technology, particularly the hardware and software necessary to make productive use of the wireless internet. The same is likely to occur with 3-D printing and other advances. For example, robots are increasingly entering consumer markets, and there is no reason to believe that the same downward cost pressures won’t allow them to be used in home production or small-scale business applications. The ability to leverage technology will require learning, but web-enabled instruction is becoming increasingly accessible as well.

Can the ownership of productive technologies become sufficiently widespread to assure a broad distribution of rewards? It’s possible that cost reductions will allow that to happen, but broadening the ownership of capital might require new saving constructs as well. That might involve cooperative ownership of capital by associations of private parties engaged in diverse lines of business. Stable family structures can also play a role in promoting saving.

It is often said that automation and AI will mean an end to scarcity. If that were the case, the implications for labor would be beside the point. Why would anyone care about jobs in a world without want? Of course, work might be done purely for pleasure, but that would make “labor” economically indistinguishable from leisure. Reaching that point would mean a prolonged process of falling prices, lifting real wages on a pace matching increases in productivity. But in a world without scarcity, prices must be zero, and that will never happen. Human wants are unlimited and resources are finite. We’ll use resources more productively, but we will always find new wants. And if prices are positive, including the cost of capital, it is certain that demands for labor will remain.

The Insidious Guaranteed Income

26 Sunday Feb 2017

Posted by in Welfare State

≈ 3 Comments

Tags

, , , , , , , , , , , , , , ,

free-money-gif

Praise for the concept of a “universal basic income” (UBI) is increasingly common among people who should know better. The UBI’s appeal is based on: 1) improvement in work incentives for those currently on public aid; 2) the permanent and universal cushion it promises against loss of livelihood; 3) the presumed benefits to those whose work requires a lengthy period of development to attain economic viability; and 4) the fact that everyone gets a prize, so it is “fair”. There are advocates who believe #2 is the primary reason a UBI is needed because they fear a mass loss of employment in the age of artificial intelligence and automation. I’ll offer some skepticism regarding that prospect in a forthcoming post.

And what are the drawbacks of a UBI? As an economic matter, it is outrageously expensive in both budgetary terms and, more subtly but no less importantly, in terms of its perverse effects on the allocation of resources. However, there are more fundamental reasons to oppose the UBI on libertarian grounds.

Advocates of a UBI often use $10,000 per adult per year as a working baseline. That yields a cost of a guaranteed income for every adult in the U.S. on the order of $2.1 trillion. We now spend about $0.7 trillion a year on public aid programs, excluding administrative costs (the cost is $1.1 trillion all-in). The incremental cost of a UBI as a wholesale replacement for all other aid programs would therefore be about $1.4 trillion. That’s roughly a 40% increase in federal outlays…. Good luck funding that! And there’s a strong chance that some of the existing aid programs would be retained. The impact could be blunted by excluding individuals above certain income thresholds, or via taxes applied to the UBI in higher tax brackets. However, a significant dent in the cost would require denying the full benefit to a large segment of the middle class, making the program into something other than a UBI.

Nathan Keeble at Mises Wire discusses some of the implications of a UBI for incentives and resource allocation. A traditional criticism of means-tested welfare programs is that benefits decline as market income increases, so market income is effectively taxed at a high marginal rate. (This is not a feature of the Earned Income Tax Credit (EITC).) Thus, low-income individuals face negative incentives to earn market income. This is the so-called “welfare cliff”. A UBI doesn’t have this shortcoming, but it would create serious incentive problems in other ways. A $1.4 trillion hit on taxpayers will distort work, saving and investment incentives in ways that would make the welfare cliff look minor by comparison. The incidence of these taxes would fall heavily on the most productive segments of society. It would also have very negative implications for the employment prospects of individuals in the lowest economic strata.

Keeble describes another way in which a UBI is destructive. It is a subsidy granted irrespective of the value created by work effort. Should an individual have a strong preference for leisure as opposed to work, a UBI subsidy exerts a strong income effect in accommodating that choice. Or, should an individual have a strong preference for performing varieties of work for which they are not well-suited, and despite having a relatively low market value for them, the income effect of a UBI subsidy will tend to accommodate that choice as well. In other words, a UBI will subsidize non-economic activity:

“The struggling entrepreneurs and artists mentioned earlier are struggling for a reason. For whatever reason, the market has deemed the goods they are providing to be insufficiently valuable. Their work simply isn’t productive according to those who would potentially consume the goods or services in question. In a functioning marketplace, producers of goods the consumers don’t want would quickly have to abandon such endeavors and focus their efforts into productive areas of the economy. The universal basic income, however, allows them to continue their less-valued endeavors with the money of those who have actually produced value, which gets to the ultimate problem of all government welfare programs.“

I concede, however, that unconditional cash transfers can be beneficial as a way of delivering aid to impoverished communities. This application, however, involves a subsidy that is less than universal, as it targets cash at the poor, or poor segments of society. The UBI experiments described in this article involve private charity in delivering aid to poor communities in underdeveloped countries, not government sponsored foreign aid or redistribution. Yes, cash is more effective than in-kind aid such as food or subsidized housing, a proposition that economists have always tended to support as a rule. The cash certainly provides relief, and it may well be used as seed money for productive enterprises, especially if the aid is viewed as temporary rather than permanent. But that is not in the spirit of a true UBI.

More fundamentally, a UBI is objectionable from a libertarian perspective because it involves a confiscation of resources. In “Why Libertarians Should Oppose the Universal Basic Income“, Bryan Caplan makes the point succinctly:

“Forced charity is unjust. Individuals have a moral right to decide if and when they want to help others….

Forcing people to help others who can’t help themselves… is at least defensible. Forcing people to help everyone is not. And for all its faults, at least the status quo makes some effort to target people who can’t help themselves. The whole idea of the Universal Basic Income, in contrast, is to give money to everyone whether they need it or not.”

Later, Caplan says:

…libertarianism isn’t about the freedom to be coercively supported by strangers. It’s about the freedom to be left alone by strangers.“

Both Keeble and Caplan would argue that the status quo, with its hodge-podge of welfare programs offering tempting but rotten incentives to recipients, is preferable to the massive distortions that would be created by a UBI. The mechanics of such an intrusion are costly enough, but as Don Boudreaux has warned, the UBI would put government in a fairly dominant position as a provider:

“… such an income-guarantee by government will further fuel the argument that government is a uniquely important and foundational source of our rights and our prosperity – and, therefore, government is uniquely entitled to regulate our behavior.“

The Looting Wage and Its Ultimate Victims

15 Wednesday Feb 2017

Posted by in Living Wage, Markets, Minimum Wage

≈ 1 Comment

Tags

, , , , , , , , , , , , , , ,

img_3920

Like children asking their peers to exchange quarters for nickels, advocates of a “living wage” hope that the government and voters will agree that workers should be paid by private employers at a rate the activists deem appropriate, regardless of skills. (The “living wage” is left-speak for a very high minimum wage.) Even worse, those advocates actually believe that such a trade can be justified. Or do they? The simple economics of the claim is undermined by assertions that a living wage is simply a matter of social justice. But social justice cannot be served in this way unless one’s definition is so bound up in virtue signaling that you don’t know the difference. It’s even too charitable to say that the left’s definition of social justice is simply bound up in the present and the short-term interests of specific groups. The unfortunate truth is that the “living wage” sacrifices the very well-being of a large number of individuals in those groups, now and in the future. Here’s why:

Suppose the government mandates a “living wage” as well as a series of measures intended to neutralize all of its unintended consequences. These measures would include a complete prohibition of involuntary terminations, investments in automation, price hikes, movement of capital abroad, and immigration. The measures must also include subsidies for failing employers. Just imagine the burden of compliance costs related to these measures, and the complex task of carving out exceptions, such as the allowable price hike in the wake of an increase in the cost of raw materials. What about the additional workers who would enter the labor force to seek employment at the higher wage? Should they be prohibited from doing so, or should employers be required to hire them, or should they be subsidized to hire them? And how will taxpayers afford all of these government subsidies?

Clearly, the situation described thus far is not sustainable. Both the initial wage hike and many of the other steps, ostensibly intended to cushion the blow on various parties, represent flagrant abridgments of private property rights, or rather, property takings! Of course, the real intent is for private parties to pay for the “living wage”. Presumably, employers are to pay the costs, especially large employers and their wealthy investors, like you when the value of those shares in your IRA, pension or 401k plan begins to tank. The reality is, however, that the unintended consequences will spread the cost in a variety of unpleasant ways.

Those in the coalition for living-wage legislation have not given much thought to the reverberations of such a change. At the most basic level, some people cannot command a high wage because they lack higher-order skills. Some have not learned the importance of reliability, of making sure they arrive at work by a specific time every day. Some have not learned the importance of concentrated work effort, of demonstrating that effort and avoiding excessive slack time. Some communicate poorly, or fail to comport themselves in a manner that commands trust. Some have a sketchy work record, presenting a risk to prospective employers. Living wage advocates assert that all of this is irrelevant, but it means everything to an employer.

How would employers attempt to to survive under a living wage? One doesn’t have to think too deeply to realize that wage floors lead to a loss of jobs for several reasons. Those lacking the skills to justify the higher wage will be out the door. Some employers will fail, finding it impossible to pay the hike in their labor costs or to pass it along to their cost-conscious clientele. The living wage is likely to lead to premature automation of many tasks otherwise requiring unskilled to more moderately-skilled workers. The capital investment needed to automate any manual process may well become worthwhile given such a shock to wage rates. Moreover, while some in the living wage movement complain that U.S. employers seek-out lower wage rates abroad, the living wage itself would lead to more of this substitution. The living wage also creates opportunities for those willing to work illegally at sub-minimum wages, including many undocumented immigrants. By driving a larger wedge between the wages of other home countries and the U.S., the living wage creates an incentive migrate In pursuit of the enlarged set of black-market opportunities for labor.

So just imagine having the government mandate a wage that is nearly double the market-clearing level. The quanity of labor demanded declines and the quantity supplied increases, leaving a surplus of workers at the mandated wage. The demand for labor declines still more as the weakest firms close shop. And it declines still more over horizons long enough to enable investment in automation and relocation of production to foreign shores. Add to the mix an expanded flow of workers from abroad. Not all of these surplus workers, native and immigrant, would be willing to take “underground” work at a rate below the living wage, but some will.

So, which of the measures listed in the second paragraph would mitigate the costs imposed by the living wage? In reality, none of them would succeed without spreading the cost more widely. Prohibiting involuntary terminations? Businesses will fail and/or prices will rise. Prohibiting investment in automation? The same. Prohibiting price hikes? Business failures, terminations, and premature automation. Prohibiting movement of capital abroad? An outright revocation of property rights and a distortion of incentives for productive investment, which would also discourage the movement of capital into the country, not just out.

Are there measures that could make the “living wage” a sustainable outcome? Yes, but they cannot be accomplished immediately by decree. Indeed, doing so would thwart the achievement of the objective. In short, productivity must increase. While productivity is multi-dimensional, education, training and work experience all foster improvement in a worker’s ability to add value. Unfortunately, our system of public primary and secondary education has been unsuccessful in producing graduates who can compete in the labor market, even at today’s minimum wage. Wholesale reforms are needed, but even the best educational reforms will take time to come to fruition. In the workplace itself, apprenticeship programs could provide under-skilled workers an avenue toward greater competitiveness at higher wages. Again, apprenticeships may only make economic sense to employers at a legalized sub-minimum wage, as Australia allows.

Second, productivity is dependent on the quality and quality of the capital invested in a business. The key to improving this capital is profitability. It’s ironic that living-wage advocates fail to see that their proposal runs directly counter to steps that would contribute to  productivity and wages. Instead, they seem intent on killing the geese that lay golden eggs! Far better to allow those eggs to be transformed into new capital assets that can enhance worker productivity and justify higher wages. Some jobs will be replaced by automation, but capital and new technology tend to create new kinds of jobs and inevitably boost worker productivity. (See “Will Automation Make Us Poor?” by Aaron Bailey.) Employers will still have an interest in seeking out, if not developing, new talent. The automation should take place as part of a more natural evolution, not one prematurely hastened by unrealistically high wage mandates.

The living wage is a prescription for failure and a death-knell for the private economy. It will fail the least-skilled workers and even some semi-skilled workers who cannot compete for jobs at the living wage. It will automate jobs before the natural time dictated by the market-driven process of technical evolution. It will lead to higher prices, which drive down the real value of any wage gains that workers manage to capture. It will lead to business failures, especially among small businesses. It will offer false hope to unskilled immigrants. It will reduce capital investment among smaller firms struggling to meet the higher wage bill. It may well lead to a slew of even more destructive public policies, such as business subsidies and other price controls. And it will create dependency on the state. The living wage is a destructive policy and ultimately a prescription for the death of self-sufficiency. It  cannot foster real social justice.

Pay and Productivity

05 Sunday Jun 2016

Posted by in Living Wage, Markets, Minimum Wage

≈ 1 Comment

Tags

, , , , , , , , , , , ,

allaboutproductivity

Remember, the real minimum wage is zero, and state-imposed unemployment is not justice. In the private economy, wages rise with productivity, and that’s true across workers at any point in time, for workers over time, and for workers in different industries over time. Don’t think so? Contrary to the blithe pronouncements of Barack Obama and reports by the union-backed Economic Policy Institute (EPI), there has been no divergence in productivity and pay since the early 1970s. This is shown convincingly by James Sherk in “Workers’ Compensation: Growing Along With Productivity“. Sherk’s work shows that hourly productivity increased by 81% since 1973, while average employee compensation increased 78%. In contrast, the EPI has claimed that productivity grew 91% since 1973, but  employee compensation grew just 10%.

How did the EPI (and Obama) reach such a faulty conclusion? Sherk breaks the error into three major parts: 1) comparing the pay of a subset of workers to the productivity of all workers; 2) excluding the pay growth of the self-employed; and 3) inconsistent adjustments of pay and productivity for inflation.

The link between wages and productivity is immutable in a market economy. The state can attempt to short-circuit the relationship, but such intervention comes at the cost of dislocations in resource utilization and damage to well-being. Veronique de Rugy discusses Sherk’s findings and emphasizes the folly in thinking that the government can somehow divorce the pay of workers from their underlying contribution to the value of output:

“One of the assets of the American economic model is a relatively flexible labor market, especially when compared with labor markets in many European countries. It explains some of the consistently lower U.S. unemployment rates and higher economic growth. Unfortunately, this flexibility is increasingly threatened by government policies that would increase the cost of employing workers.“

Populists and statists share some destructive tendencies, such as a fixation on increasing the cost of labor to employers. The current debate over a “living wage” of $15 per hour involves more than doubling the minimum wage in many parts of the country. This is so far out of line with the productivity of low-skilled workers as to make absurd claims that it won’t have a serious impact on their employment. There are employers who won’t be able to survive under those circumstances. There are others who will have to scale back operations. Employers having access to capital in industries such as car washes and fast food know that automation is more than viable as a substitute for low-skilled labor. And new labor-saving innovations are inevitable when creative entrepreneurs are confronted with an obstacle like high-cost labor: necessity is the mother of invention. But premature automation is not an obvious consequence to living-wage advocates. And that’s to say nothing of the futures destroyed when low-skilled workers are denied opportunities for work experience.

The connection between wages and productivity is part of a well-functioning economy and it is just as alive and well in today’s economy as ever. The “right” wage cannot be determined by central planners, bureaucrats or legislators apart from productive reality, and the adverse consequences of their attempts to do so cannot be wished away. Only markets that price the real value of productive contributions can put resources such as low-skilled labor to their best use, avoiding the waste inherent in regulation that is always ignorant of dynamic preferences and resource availability.

 

Automate No Job Before Its Time

28 Monday Dec 2015

Posted by in Price Controls, Technology

≈ 6 Comments

Tags

, , , , , , , , ,

This interactive chart from the McKinsey Global Institute (not the one above, as good as it is…) shows occupations at risk of automation, and it should give warning to those asserting that a substantial increase in the minimum wage is in the interests of low-wage workers. It shows the extent to which various jobs can be automated under existing technology. The salient facts here are that a large number of workers earn less than $15 per hour, that most of those workers perform jobs that can be automated, and that further advances in technology will increase the potential for automation beyond what’s shown in the chart.

A simple truth that must be understood is that wage rates are strongly associated with the skills and productivity required for particular jobs. Denial of that fundamental rule cannot help anyone, and will almost certainly harm many. Low skill requirements are less highly-compensated because they add little value and are easily satisfied.

As Don Boudreaux points out, innovation is often spurred by economic forces. A mandated wage minimum, which is a price floor creating artificial surplus conditions, magnifies incentives for greater innovation. In addition to the substitution away from low-skilled labor (or domestic labor) that can be expected, there are many other margins along which employers can economize in the face of such government edicts: higher expectations for productivity, fewer benefits, fewer breaks, fewer niceties in the workplace, and less flexibility over hours and days off. These things matter greatly to employees and employers. A wage law can make for an unpleasant work environment.

Those who suffer most from minimum wage decrees are the least skilled, whose jobs are the most vulnerable. Economist David Neumark notes that “The Evidence Is Piling Up That Higher Minimum Wages Kill Jobs“, despite claims to the contrary (gated… Google “wsj NeumarK”, select the December 15, 2015 link).

Lest anyone decry the technologies that could replace these workers, recall that the substitution of capital for labor over time has led to the great gains in productivity that have elevated wages and income over time. Many jobs that are commonplace today (and were not even imagined in earlier times) would not exist if not for advances in technology. Likewise, there will be jobs that are commonplace in the future that do not exist today, and we won’t have the power (nor will the government) to anticipate those jobs until the enabling technologies come to fruition and early adoption. These kinds of changes are never without difficulty, as workers bear significant costs of adjustment in the short run, including the acquisition of new skills. However, wage floors force an even earlier and contrived adoption of technologies, which harms low-wage workers most severely. Far better to allow an unfettered and natural process of free choice, technological diffusion, price adjustment, and growth to take place.