Factor Markets and Labor, Part B: Coalitions and Constraints

In Part A we built one tool and used it on the demand side of factor markets: the value of marginal product (VMP), the value of what one more worker adds. We saw that in a competitive market a worker tends to be paid that value, because paying less lets a rival bid him away and paying more loses the firm money. The hiring rule was simple. Hire each worker up to the point where the value he adds equals the wage you must pay him. Diminishing returns made the demand for labor slope down; substitution and the lump-of-labor fallacy showed that costly inputs get economized on without destroying the total pool of work; and earnings gaps traced to ability and to the cost of acquiring skills.

Now we turn that same engine on the most heated arguments in all of economics. What happens when a group of workers, or a licensed profession, organizes to keep its own pay high by limiting who else may compete? Does a minimum wage help the people it names? Does a competitive market make prejudice cheaper to indulge, or more expensive? Each answer falls straight out of the VMP rule from Part A, with one new ingredient added: political economy, the study of why a policy that hurts the many but helps an organized few gets passed and stays passed. A single lens runs through the whole part. Call it insiders versus outsiders.

Unions Are Legislatively Sanctioned Monopolies That Raise Insiders’ Wages by Restricting the Supply of Labor

A labor union, analyzed plainly, is a monopoly: an arrangement that eliminates competition among workers over the wage at which they sell their labor. That is not name-calling; a former Secretary of Labor and union counsel said as much, calling a union “technically” a monopoly “in the limited sense” that it ends competition among workers for the available jobs. A union does not abolish competition; it redirects it. With wage competition off the table, workers compete instead through seniority, through who gets in the apprenticeship line, through nonwage maneuvering. The rivalry changes form rather than vanishing.

Be precise about the kind of monopoly. A closed monopoly is protected by a legal barrier that keeps rivals out; an open one merely happens to be large and can be challenged by anyone who enters. A union backed by labor law is a closed monopoly, because the law restricts who may compete to supply that labor. A big firm like a steelmaker, by contrast, is usually an open one, exposed to new entrants and imports. In the United States this protection came through labor legislation in the 1930s that sanctioned collective bargaining and created a federal board to enforce it.

How does a union raise its members’ wages? In the end, by restricting the supply of labor competing for the jobs. A craft union does it most directly, by controlling apprenticeships and admission so fewer trained workers chase the work; a wise craft leader, the saying goes, does not need to strike, he controls who gets in. Shrink the supply and the wage of those inside rises, exactly as a smaller harvest raises a crop’s price. This is why the central lens for the whole part is insiders versus outsiders. A coalition raises the pay of its members, the insiders, by limiting the ability of outsiders to compete for the same work, and the outsiders kept out are usually the younger, less-experienced, and less-advantaged workers who could only have competed by offering to work for less.

This lens explains a string of puzzling stances. Organized labor has historically pushed for limits on immigration while employers objected; cutting the supply of competing workers raises members’ wages, which is why workers favor it and employers resist. A union head may champion lower prices for the public while opposing the removal of tariffs that protect his members’ product, because lower domestic prices sell more union-made goods, but cheaper imports would cost members their jobs; the “for the public” framing dresses up a producer interest. Some employers even welcome a strong union, because a union that prices low-wage rivals out of the labor market can knock out the low-cost competing firms that relied on them, protecting the higher-cost incumbent. A strike, where the law bars hiring replacements, is the same logic enforced: it blocks willing outsiders from selling their labor at the open-market wage. When a panel of union, employer, and “public” representatives settles a strike, the group with no seat at the table is precisely the nonunion outsider who would gladly take the job at the going rate.

A union also serves, secondarily, to police an employer’s promises, since a single worker has trouble holding a firm to commitments made over a long career. That monitoring role is real, but it is not the main story for wages and does not change the supply-restriction logic.

Work rules can extract the same insider gain without raising the posted wage. Featherbedding requires an employer to hire workers or preserve tasks that production no longer needs, such as retaining a manual handling crew after premixed materials remove the job or requiring standby performers who will not perform. The rule converts part of the firm’s quasi-rent into insider income while raising the cost of output and excluding outsiders who might use the resources elsewhere. Calling the requirement a safety or craft standard does not settle its effect; ask whether the mandated work adds value at the margin.

Two points round this out. First, a contrived wage gain is not a stream of easy money forever. Once a privileged position can be passed on or sold, its advantage gets capitalized into the price of getting in, so later holders pay up front and earn only an ordinary return on what they paid. We lack the tool to make this precise, but the verbal version suffices: the gain is already priced into the value of the position today. Present value will make it exact, and that is one of the threads the next topic picks up.

Second, the mirror image of a union, a single dominant buyer of labor, is monopsony. The cleanest example is the military draft. People defended conscription as “cheap” because the budget cost of paying draftees was low, but that figure hid the real cost: the value of what conscripts gave up, forced into service at below-market pay, plus the waste of assigning people poorly. A volunteer force paid by wages reveals the true cost in the budget and tends to lower it, because wages sort people toward the jobs they fit best. And the claim that a lone worker is “helpless,” so a union is essential for fairness, is mostly empty wherever many employers compete. An employer who lowballs a productive worker loses him to a rival who offers more. Competition among employers, not a coalition, is what ordinarily protects the individual worker.

The buyer-side mechanism mirrors the price-searcher seller. Suppose a monopsonist faces an upward-sloping labor supply, so attracting one more worker requires a higher wage. If that higher wage must also be paid to the workers already employed, the cost of the next hire exceeds that worker’s wage: the marginal wage cost includes both the new worker’s pay and the raises for incumbents. The monopsonist hires only until the next worker’s value of marginal product equals this higher marginal wage cost, not merely the posted wage. Compared with competing employers, it therefore employs fewer workers and pays a lower wage. The result needs both ingredients, market power on the buying side and an upward-sloping supply; a lone employer facing perfectly mobile workers cannot create it.

The draft was a monopsony run by the government. Private employers have tried the same trick by agreement. A lone hospital cannot hold intern pay far below what an intern adds, because a rival hospital would bid the intern away. So hospitals did not act alone: their association capped intern salaries, and the cap held because a hospital that broke it risked its top-grade accreditation, the seal it needs to attract students and funding. Notice who keeps the savings. Cheap interns do not lower the price of hospital care by some automatic mechanism; the margin shows up as higher incomes for the people already inside.

College sports ran the same play for over a century, on a bigger stage. The NCAA’s member colleges are employers buying athletic labor, and their amateurism rules were a collective agreement not to pay for it, enforced the same way as the intern cap: a college that paid players risked its athletes’ eligibility and its own standing. With money offers banned, the competition for stars resurfaced in the forms the rules allowed, scholarships, facilities, and no small number of covert payments; rivalry changes form, it does not die. The agreement finally cracked in court. Justice Kavanaugh wrote that the NCAA’s business model “would be flatly illegal in almost any other industry in America,” and once athletes could earn from their name, image, and likeness in 2021 and colleges could pay them directly under the 2025 House settlement (up to roughly $20.5 million per school, with nearly $2.8 billion in back pay), the suppressed wages appeared almost overnight; 310 schools opted in the first year. Recall the cartel lesson from the market-power topic: a cartel survives only where cheating is cheap to detect and punish. An accreditation body that doubles as the cartel’s enforcement arm is exactly that machinery.

Training costs also depend on who can capture the return. Skills useful only at the current employer are specific training: the worker cannot sell them elsewhere, so the employer pays because it captures much of the productivity gain and has reason to retain the trained worker. Portable skills are general training. Because the worker can take them to a rival, the worker pays, often by accepting a lower wage while learning, and later receives the higher market wage the skill supports. Specific training therefore predicts lower turnover and fewer layoffs than general training, because both parties would lose part of a relationship-specific investment if the match ended.

The pattern of who still unionizes says the same thing from another angle. Union membership hovered near a third of the American workforce in the 1950s and has fallen to little more than a tenth, but the fall is almost entirely on the private side: government workers now unionize at close to 40 percent, against under 7 percent in private business. The reason is not sentiment. A private employer who pays above-market wages must cover them out of sales revenue while competing against firms that do not, and such employers shrink or disappear. A government agency pays out of taxes, and taxpayers are a more dispersed and patient source of above-market wages than customers ever are.

The Minimum Wage Prices Out the Workers Whose Value Is Below the Floor

Now apply the hiring rule to the most debated labor policy of all. A minimum wage is a legal floor on the wage an employer may pay.

On the boat in Part A, the owner hired a worker only if the value he added was at least the wage. Set a floor above some workers’ VMP and those workers can no longer be hired at a wage that makes sense for the employer. They are not paid more; they are not paid at all in covered jobs. The higher the floor, the more workers it prices out, which is why “the higher the minimum wage, the greater the unemployment among the least-skilled” is sound.

Who are those workers? Disproportionately the young, the inexperienced, and the least-skilled, the very people the policy is meant to help, because their value to an employer has not yet risen above the floor. The damage shows up across countries: where minimum wages are high, youth unemployment runs far above the general rate, sometimes above 20 or even 40 percent against a general rate near 10. In the United States, black labor-force participation was slightly higher than whites’ before the federal minimum-wage laws of the 1930s; black youth unemployment climbed in the era that followed.

Notice, too, who campaigns for a higher floor. Workers already earning far above the minimum, and firms running the latest labor-saving equipment, are among its most reliable advocates. Raising the minimum raises the costs of their labor-intensive rivals, and customers then drift toward the capital-intensive firm and its high-wage workforce. The unionized garment industry pressed hard for minimum-wage enforcement against small immigrant-staffed shops for exactly this reason. The language is concern for the low-paid; the arithmetic is a rival’s costs. It is the licensing story again, wearing a kinder face.

The escape routes confirm the mechanism. Where a floor binds, work shifts toward margins it cannot reach. The share of owner-driven cabs rises relative to employee-driven ones, because a person working for himself pays himself no legal minimum; push a taxi-driver minimum to an absurd level and you would see almost nothing but owner-operators. Self-employment and uncovered sectors absorb the workers the covered sector can no longer afford.

A warning about evidence. Suppose someone surveys firms before and after a minimum-wage hike and adds up the employment changes at the surviving firms. That procedure is broken, because it counts only firms that lived to be surveyed. It misses the firms that closed and, worse, the jobs and firms that never came into existence because the floor made them unprofitable from the start. The disemployment shows up exactly where this method does not look.

A Wage Floor Makes Discrimination Cheap, a Shortage Makes It Costly

The minimum wage has a darker consequence that overturns a common intuition. People assume a wage floor protects vulnerable workers from exploitation. In a competitive market it can do the opposite, and the cleanest case is the most painful. Under apartheid, white-run unions in South Africa backed minimum-wage and “equal pay” rules precisely because they grasped the mechanism: if a black worker could no longer offer to work for less, the employer’s incentive to hire him over a white worker disappeared. Forced to pay the same wage regardless, employers indulged their prejudice freely, even hiring abroad rather than employ the surplus of black workers a floor had created. A wage floor strips less-favored workers of their one competitive weapon, the willingness to accept lower pay, and so it can raise, not lower, the discrimination they face. (Triple the floor and you would only sharpen the effect.)

The general principle: in a competitive market, discrimination is costly to the discriminator. An employer who refuses productive workers from a group he dislikes forgoes the value they would add, and a rival who hires them gains an edge and can undersell him. The market punishes prejudice by making it expensive. History bears this out: before civil-rights laws, black chemists and other skilled minority workers were more often hired by profit-seeking firms than by nonprofits, because the profit-seeker paid a price for prejudice that the nonprofit, insulated from competition, did not. Foreign-owned firms operating where local prejudice ran strong often paid higher wages to hire the workers locals shunned, chasing profit rather than conforming to bias.

The same rule reaches the most-discussed pay gap of all, the one between men and women. An employer who could hire equally productive women at a lower wage and refused to, paying men more for the same output, would be throwing away profit, and a rival who hired those women would gain an edge; so competition steadily erodes a gap that rests on pure prejudice. That does not mean every observed difference in average pay is prejudice. Part of it traces to differences in the cost of employing a worker and to choices workers make: women have on average borne more of the cost of childbearing and child-rearing and have more often chosen occupations whose skills do not depreciate during a career interruption, which shows up as a difference in measured earnings without anyone being underpaid relative to what they add. The honest summary is that competitive markets push against the prejudice component, that the gap has narrowed where competition and women’s qualifications have grown, and that the part of it which reflects voluntary choice and cost differences is not discrimination at all. The size of what remains is disputed, so we assert no number; the analytical point is that the same discrimination-is-costly logic applies here as everywhere else.

Wage controls reverse the discipline. A floor creates a surplus, more applicants than jobs, and when employers pick from a line out the door, turning some away costs nothing; discrimination becomes cheap. A ceiling does the opposite: it creates a shortage, and an employer scrambling to fill positions cannot afford to refuse a productive worker over prejudice; discrimination becomes expensive. The lesson is uncomfortable for a lot of political rhetoric: the market mechanism critics distrust is the one that makes prejudice costly, and the controls offered as remedies often make it cheap.

Equal-Pay and Mandated-Benefit Rules Override VMP and Shift Who Pays

A cluster of policies tries to set pay by something other than the value a worker adds, and each has predictable winners and losers once you apply VMP.

Comparable worth and single-salary schemes pay people by a job’s rated “worth” rather than what the labor adds in the market. Pay all teachers the same regardless of subject or skill, and you transfer from the scarcer, higher-value teachers, math or science, where outside options are strong, toward those whose skills are plentiful, and from the better teachers toward the weaker ones. The rule does not abolish the value differences VMP reflects; it hides them and reshuffles who is overpaid relative to what they add.

Mandated benefits rest on an idea from earlier in the course: a worker’s full pay is money wages plus the value of every nonmoney term of the job, the parking, the leave, the pension, the security. Employers care about the total, not the label. So mandate a new benefit and, over time, money wages drift down to offset it, because the employer was only ever willing to pay a worker his value all-in. Ask yourself: if a job stripped its benefits, would you demand higher take-home pay, and would the employer pay it? If yes, you were paying for the benefits all along through lower wages. The worker, not the employer, ultimately bears the cost of a mandated benefit, which is why both sides sometimes object to having the mix dictated to them.

The same full-pay logic runs in reverse under a wage ceiling. When a binding cap blocks competition on money wages, compensation spills into nonmoney forms. Centuries ago, maximum-wage laws drove employers to pay above the legal ceiling in food, housing, and clothing. During the Second World War, wage controls pushed American employers to compete for workers by offering health insurance, which is how employer-provided medical coverage became standard and, over the long run, helped entrench a system in which insulated patients and providers face little price discipline and covered medical prices drift upward. A ceiling no more abolishes competition for workers than a floor abolishes the value of labor; it changes the form the payment takes.

Occupational Licensing Restricts Entry to Protect Incumbents, Not Consumers

The union logic, restrict the supply to raise insiders’ pay, reappears in a respectable suit as occupational licensing. A license is a legal requirement to obtain permission before practicing a trade, sold to the public as protection against incompetents. Sometimes that is part of the story. But notice who lobbies for these rules: overwhelmingly the incumbents already in the trade, not the consumers they claim to protect. Licensing laws routinely grandfather existing practitioners, exempting them from the new requirements. If the point were truly to guarantee competence, the people already working would be the first ones tested. Exempting them gives the game away. The real effect, often the real aim, is to raise the cost of entering and thin the ranks of competitors, propping up the incomes of those inside. The same logic explains why a bar association’s fixed minimum fees for routine legal work help established lawyers far more than a beginner hungry for clients.

The mechanism is the labor-market version of the producer protections studied under market power. An import quota keeps foreign sellers out; an occupational license keeps would-be workers out. Both close entry, restrict supply, and direct the resulting rent toward incumbents. The stated objectives differ, but the diagnostic questions are the same: what constraint does the rule create, whose competition does it suppress, and who captures the higher price?

The sharpest illustration is a numerical cap on licenses. New York City has limited the number of taxi licenses since 1937, issuing “medallions” that authorize a cab to operate. The artificial scarcity drove the price of a medallion from $10 when the system began to roughly $80,000 by the 1980s and past a million dollars by 2011. That price is the market’s measure of the monopoly value of being allowed in, created not by serving riders better but by keeping competitors out. The political economy is the same insiders-versus-outsiders story as the union: the gains are concentrated on a small group of incumbents who fight hard to keep them, while the losses, higher prices and fewer choices for riders and lost opportunities for would-be drivers, are spread thin across millions who never organize to oppose them. That asymmetry is why licensing spreads and persists even when it plainly costs the public and shuts out less-advantaged people trying to enter a trade.

The medallion price shows what happens when a privileged position can be bought: the rent is capitalized into the entry price, and the buyer earns only an ordinary return. When the position must be qualified for instead, the same rent gets eaten in the qualifying. About a quarter of American workers now need a government license to do their jobs, up from under five percent in the early 1950s, and medicine shows where the toll goes. The typical indebted medical graduate leaves school owing about $205,000, then works several more years of residency at a first-year stipend near $65,000, having first beaten out most of the 54,699 applicants who competed for 23,440 medical-school seats. Stack up the tuition, the debt, and the years of forgone earnings, and the license’s apparent monopoly rent mostly disappears: the entrant nets roughly a normal return on the whole undertaking. The windfall was collected once, by whoever already held a license when the gate closed. That is why incumbents defend these restrictions so fiercely; repeal would not merely trim their income, it would hand them a capital loss.

Much Unemployment Is Productive Search, and Lowering Its Cost Lengthens It

One more labor-market fact deserves a place here, because it shapes how everything in this part gets measured: the unemployment spell itself. Wage offers for the same worker are not identical across employers, because finding out who pays what is costly. A worker whose job ends and who grabs the first offer he stumbles into may throw away years of higher pay to save a few weeks of looking. Time spent finding and comparing offers produces something real, information about the best job available, and that makes a search spell an investment, not idleness.

An economy can hide this search by assigning everyone a post. A military or centrally directed workplace may report that everyone has a job while keeping people in tasks whose output is worth less than what they could produce elsewhere. The unemployment is disguised as low-productivity employment. Open unemployment makes the search visible and gives workers freedom to compare assignments; a zero measured rate can therefore coexist with worse matching and more wasted labor.

How long should the search run? By the same marginal rule that has governed everything else in this course: keep searching while the expected gain from one more inquiry, a better offer weighted by the chance of finding it, exceeds the cost, chiefly the wages given up by not working. When the expected gain slips below that cost, stop and take the best offer in hand. The lowest offer worth taking now rather than searching on is your reservation wage.

Change the cost of searching and you change how long people search, for ordinary marginal reasons. Unemployment insurance and welfare payments replace part of the income lost while looking, so covered workers rationally hold out longer for better matches. That is not an accusation of laziness; it is the same marginal rule operating at a lower price of continued search. The rule also says who can afford to be choosy: a worker in a household with a second earner can run a longer search than one whose rent is due on Friday.

Information also helps explain why firms often lay workers off instead of cutting everyone’s wage immediately when demand falls. Workers cannot instantly tell whether the decline is temporary or permanent, general to their occupation or peculiar to one employer. A wage cut might signal that the worker’s best outside wage has fallen, or it might be an employer trying to exploit workers who have made firm-specific investments. Resisting the cut while accepting a layoff can therefore be rational while people learn which story is true. Unemployment becomes “involuntary” in the strong sense only when the worker’s belief about available wages proves mistaken; until the search resolves the uncertainty, the observed refusal does not show that the worker knowingly rejected an equally good job.

Once you see search, the measured unemployment rate stops being one thing. It lumps together workers producing information between jobs, workers priced out by the wage floor we met earlier in this post, would-be electricians and bricklayers standing outside the licensing and apprenticeship gates who report themselves as unemployed, and seasonal workers who take jobs only at peak wages. Four different situations, four different remedies, one number. When someone quotes “the” unemployment rate, it is worth asking which of the four they have in mind.

One thread ran under every case in this part and connects back to our earlier property-rights topic. The people who attend hardest to the market value of a resource are its owners: the boat owner in Part A who beached the boat rather than overpay, the employer here who loses money by indulging prejudice. When a resource is owned, the owner keeps the gains from using it well and bears the losses from using it badly, so ownership concentrates the incentive to attend to market value while its absence dilutes that incentive.