Although the marginal productivity theory of income distribution is a well-established part of economic theory, closely linked to the analysis of markets in general, it is a source of some controversy. There are two main objections to it.

First, in the real world we see large disparities in income between factors of production that, in the eyes of some observers, should receive the same payment. Perhaps the most conspicuous examples in the United States are the large differences in the average wages between women and men and among various racial and ethnic groups. Do these wage differences really reflect differences in marginal productivity, or is something else going on?

Second, many people wrongly believe that the marginal productivity theory of income distribution gives a moral justification for the distribution of income, implying that the existing distribution is fair and appropriate. This misconception sometimes leads other people, who believe that the current distribution of income is unfair, to reject marginal productivity theory.

To address these controversies, we’ll start by looking at income disparities across gender and ethnic groups. Then we’ll ask what factors might account for these disparities and whether these explanations are consistent with the marginal productivity theory of income distribution.

Wage rates in the United States cover a very wide range. In 2013, hundreds of thousands of workers received the legal federal minimum of $7.25 per hour. At the other extreme, the chief executives of several companies were paid more than $100 million, which works out to $20,000 per hour even if they worked 100-hour weeks. Even leaving out these extremes, there is a huge range of wage rates. Are people really that different in their marginal productivities?

A particular source of concern is the existence of systematic wage differences across gender and ethnicity. Figure 19-8 compares annual median earnings in 2013 of workers age 25 or older classified by gender and ethnicity. As a group, White males had the highest earnings. Other data show that women (averaging across all ethnicities) earned only about 72% as much; African-American workers (male and female combined), only 68% as much; Hispanic workers (again, male and female combined), only 57% as much.

We are a nation founded on the belief that all men are created equal—and if the Constitution were rewritten today, we would say that all people are created equal. So why do they receive such unequal pay? Let’s start with the marginal productivity explanations, then look at other influences.

A large part of the observed inequality in wages can be explained by considerations that are consistent with the marginal productivity theory of income distribution. In particular, there are three well-understood sources of wage differences across occupations and individuals.

First is the existence of compensating differentials: across different types of jobs, wages are often higher or lower depending on how attractive or unattractive the job is. Workers with unpleasant or dangerous jobs demand a higher wage in comparison to workers with jobs that require the same skill and effort but lack the unpleasant or dangerous qualities. For example, truckers who haul hazardous loads are paid more than truckers who haul non-hazardous loads. But for any given job, the marginal productivity theory of income distribution generally holds true. For example, hazardous-load truckers are paid a wage equal to the equilibrium value of the marginal product of the last person employed in the labor market for hazardous-load truckers.

A second reason for wage inequality that is clearly consistent with marginal productivity theory is differences in talent. People differ in their abilities: a higher-ability person, by producing a better product that commands a higher price compared to a lower-ability person, generates a higher value of the marginal product. And these differences in the value of the marginal product translate into differences in earning potential. We all know that this is true in sports: practice is important, but 99.99% (at least) of the population just doesn’t have what it takes to throw passes like Tom Brady or hit tennis balls like Roger Federer. The same is true, though less obvious, in other fields of endeavor.

A third and very important reason for wage differences is differences in the quantity of human capital. Recall that human capital—education and training—is at least as important in the modern economy as physical capital in the form of buildings and machines. Different people “embody” quite different quantities of human capital, and a person with a higher quantity of human capital typically generates a higher value of the marginal product by producing a product that commands a higher price. So differences in human capital account for substantial differences in wages. People with high levels of human capital, such as skilled surgeons or engineers, generally receive high wages. In 2013, surgeons earned an average of $233,150.

The most direct way to see the effect of human capital on wages is to look at the relationship between educational levels and earnings. Figure 19-9 shows earnings differentials by gender, ethnicity, and three educational levels for people age 25 or older in 2013. As you can see, regardless of gender or ethnicity, higher education is associated with higher median earnings. For example, in 2013 White females with 9 to 12 years of schooling but without a high school diploma had median earnings 28% less than those with a high school diploma and 65% less than those with a college degree—and similar patterns exist for the other five groups.

Because even now men typically have had more years of education than women and Whites more years than non-Whites, differences in level of education are part of the explanation for the earnings differences shown in Figure 19-8.

It’s important to realize that formal education is not the only source of human capital; on-the-job-training and work experience also generate human capital. In fact, there are other factors that also influence wage differences. A good illustration of these factors is found in research on the gender-wage gap, the persistent difference in the earnings of men compared to women. In the U.S. labor market, researchers have found that the gender gap is largely explained by differences in:

For example, in a U.S. Department of Labor study using recent census data, the gender-wage gap fell from 20.4% to 5% once these five factors were accounted for. Moreover, over the past 30 years even the unadjusted gender-wage gap has fallen significantly, from 36.5% in 1979 to 16.5% in 2011, as women have begun to close in on men in terms of these five factors.

But it’s also important to emphasize that earnings differences arising from these factors are not necessarily “fair.” When women do most of the work caring for children, they will inevitably have more career interruptions or need to work part-time instead of full-time. Similarly, a society where non-White children typically receive a poor education because they live in underfunded school districts, then go on to earn low wages because they are poorly educated, may have labor markets that are well described by marginal productivity theory (and would be consistent with the earnings differentials across ethnic groups and between the genders shown in Figure 19-8). Yet many people would still consider the resulting distribution of income unfair.

Still, many observers think that actual wage differentials cannot be entirely explained by compensating differentials, differences in talent, differences in human capital, or differences in job status. They believe that market power, efficiency wages, and discrimination also play an important role. We will examine these forces next.

The marginal productivity theory of income distribution is based on the assumption that factor markets are perfectly competitive. In such markets we can expect workers to be paid the equilibrium value of their marginal product, regardless of who they are. But how valid is this assumption?

We studied markets that are not perfectly competitive in Chapters 13, 14, and 15; now let’s touch briefly on the ways in which labor markets may deviate from the competitive assumption.

One undoubted source of differences in wages between otherwise similar workers is the role of unions—organizations that try to raise wages and improve working conditions for their members. Labor unions, when they are successful, replace one-on-one wage deals between workers and employers with collective bargaining, in which the employer must negotiate wages with union representatives. Without question, this leads to higher wages for those workers who are represented by unions. In 2013 the median weekly earnings of union members in the United States were $966, compared with $797 for workers not represented by unions—more than a 20% difference.

How much does collective action, either by workers or by employers, affect wages in the modern United States? Several decades ago, when around 30% of American workers were union members, unions probably had a significant upward effect on wages. Today, however, most economists think unions exert a fairly minor influence.

In 2013, less than 7% of the employees of private businesses were represented by unions. Just as workers can sometimes organize to extract higher wages than they would otherwise receive, employers can sometimes organize to pay lower wages than would result from competition. For example, health care workers—doctors, nurses, and so on—sometimes argue that health maintenance organizations (HMOs) are engaged in a collective effort to hold down their wages. Yet the sheer size of the U.S. labor market is enormous and the ease with which most workers can move in search of higher-paying jobs probably means that concerted efforts to hold wages below the unrestrained market equilibrium level rarely occur and even more rarely succeed.

A second source of wage inequality is the phenomenon of efficiency wages—a type of incentive scheme used by employers to motivate workers to work hard and to reduce worker turnover. Suppose a worker performs a job that is extremely important but that the employer can observe how well the job is being performed only at infrequent intervals—say, serving as a caregiver for the employer’s child. Then it often makes sense for the employer to pay more than the worker could earn in an alternative job—that is, more than the equilibrium wage. Why? Because earning a premium makes losing this job and having to take the alternative job quite costly for the worker.

So a worker who happens to be observed performing poorly and is therefore fired is now worse off for having to accept a lower-paying job. The threat of losing a job that pays a premium motivates the worker to perform well and avoid being fired. Likewise, paying a premium also reduces worker turnover—the frequency with which an employee leaves a job voluntarily. Despite the fact that it may take no more effort and skill to be a child’s caregiver than to be an office worker, efficiency wages show why it often makes economic sense for a parent to pay a caregiver more than the equilibrium wage of an office worker.

The efficiency-wage model explains why we might observe wages offered above their equilibrium level. Like the price floors we studied in Chapter 5—and, in particular, much like the minimum wage—this phenomenon leads to a surplus of labor in labor markets that are characterized by the efficiency-wage model. This surplus of labor translates into unemployment—some workers are actively searching for a high-paying efficiency-wage job but are unable to get one, and other more fortunate but no more deserving workers are able to acquire one.

As a result, two workers with exactly the same profile—the same skills and same job history—may earn unequal wages: the worker who is lucky enough to get an efficiency-wage job earns more than the worker who gets a standard job (or who remains unemployed while searching for a higher-paying job).

Efficiency wages are a response to a type of market failure that arises when some employees are able to hide the fact that they don’t always perform as well as they should. As a result, employers use nonequilibrium wages to motivate their employees, leading to an inefficient outcome.

It is a real and ugly fact that throughout history there has been discrimination against workers who are considered to be of the wrong race, ethnicity, gender, or other characteristics. How does this fit into our economic models?

The main insight economic analysis offers is that discrimination is not a natural consequence of market competition. On the contrary, market forces tend to work against discrimination. To see why, consider the incentives that would exist if social convention dictated that women be paid, say, 30% less than men with equivalent qualifications and experience. A company whose management was itself unbiased would then be able to reduce its costs by hiring women rather than men—and such companies would have an advantage over other companies that hired men despite their higher cost. The result would be to create an excess demand for female workers, which would tend to drive up their wages.

But if market competition works against discrimination, how is it that so much discrimination has taken place? The answer is twofold. First, when labor markets don’t work well, employers may have the ability to discriminate without hurting their profits. For example, market interferences (such as unions or minimum-wage laws) or market failures (such as efficiency wages) can lead to wages that are above their equilibrium levels. In these cases, there are more job applicants than there are jobs, leaving employers free to discriminate among applicants. In 2011, with unemployment over 9%, the Equal Employment Opportunity Commission, the federal agency tasked with investigating employment discrimination charges, reported that the complaints from workers and job-seekers had hit an all-time high, the most logged in the agency’s 46-year history.

In research published in the American Economic Review, two economists, Marianne Bertrand and Sendhil Mullainathan, documented discrimination in hiring by sending fictitious résumés to prospective employers on a random basis. Applicants with “White-sounding” names such as Emily Walsh were 50% more likely to be contacted than applicants with “African-American-sounding” names such as Lakisha Washington. Also, applicants with White-sounding names and good credentials were much more likely to be contacted than those without such credentials. By contrast, potential employers seemed to ignore the credentials of applicants with African-American-sounding names.

Second, discrimination has sometimes been institutionalized in government policy. This institutionalization of discrimination has made it easier to maintain it against market pressure, and historically it is the form that discrimination has typically taken. For example, at one time in the United States, African-Americans were barred from attending “Whites-only” public schools and universities in many parts of the country and forced to attend inferior schools.

Although market competition tends to work against current discrimination, it is not a remedy for past discrimination, which typically has had an impact on the education and experience of its victims and thereby reduces their income.

The main conclusion you should draw from this discussion is that the marginal productivity theory of income distribution is not a perfect description of how factor incomes are determined but that it works pretty well. The deviations are important. But, by and large, in a modern economy with well-functioning labor markets, factors of production are paid the equilibrium value of the marginal product—the value of the marginal product of the last unit employed in the market as a whole.

It’s important to emphasize, once again, that this does not mean that the factor distribution of income is morally justified.