The Role of Fairness in Wage Determination

by Albert Rees
The Role of Fairness in Wage Determination
Albert Rees
Journal of Labor Economics
Start Page: 
End Page: 
Select license: 
Select License


The Role of Fairness in 
Wage Determination 

A1bert Rees, Pvinceton univevxity

Neoclassical wage theory is based on the premise that a worker's utility is based on his own wage and his own hours of work, without reference to the wages and hours of others. This article reviews anec- dotal evidence that the wages of others are a powerful force in deter- mining worker satisfaction, such that utility goes down when the wages of others go up. The resulting comparisons are a powerful force in determining wage structure, but do not exclude ultimate effect of neoclassical wage determinants.

The neoclassical theory of wage determination, which I taught for 30 years and have tried to explain in my textbook (Rees 1973), has nothing to say about fairness. In this theory, wages are determined by a demand schedule based on marginal productivity and a supply schedule based on the utility function of workers. The arguments of this utility function are real wages and leisure; the wages of worker B do not enter the utility function of worker A.

Beginning in the mid-197Os, I began to find myself in a series of roles in which I participated in setting or controlling wages and salaries. These included serving on three wage stabilization bodies during the Nixon and Ford administrations, as a director of two corporations, as provost of a private university, as president of a foundation, and as a trustee of a liberal arts college. In one of the corporations I serve as chairman of the com- pensation committee.

In none of those roles did I find the theory I had been teaching for so long to be the slightest help. The factors involved in setting wages and

I am indebted to my colleagues Orley Ashenfelter, David Card, Mark Killings- worth, and Sharon P. Smith for helpful suggestions on drafts of this article. They do not necessarily agree with the views expressed.

[Journal of Labor Economics, 1993, vol. 11, no. 1, pt. 11 
O 1993 by The University of Chicago. All rights reserved. 
0734-306X/93/110 1-0002$01.50 


salaries in the real world seemed to be very different than those specified in the neoclassical theory. The one factor that seemed to be of overwhelming importance in all these real-world situations was fairness, and fairness always seemed to be judged by making some kind of wage comparison: with another union, with another employer, or with another person. The comparison was always made upward rather than downward; that is, the reference group or person always earned inore than the union or en~ployee that first suggested the comparison. I cannot recall ever being involved in any dispute or controversy in which any party to the dispute questioned the criterion of fairness as the basis for settling the controversy. Even employers never argued that supply or demand conditions in the labor market should be taken into account. The question was always which wage or salaiy comparison was appropriate, and of course each party would always contend that the comparison was appropriate, and of course each party would always contend that the comparison favorable to its cause was by far the most relevant one.

The purposes of this article are to review briefly the role of fairness in wage determination and to ask how, if at all, this role can be incorporated into formal wage theory. It will rely on anecdotal evidence rather than on any systematic body of data. In this sense, I am afraid, it is not in the tradition of Jacob Mincer, whose work always involves meticulous data analysis.

Another way to summarize this article is to say that it argues that wages are not fully determined by markets. Labor markets are always more or less imperfect and leave a zone of indeterminacy within which there is room for bargaining, both collective and individual, and for discretion on the part of wage setters. This discretion permits development and persis- tence of "firm effects" in wage structures.' There is nothing particularly novel about this argument in its broad outlines. I shall argue, however, that the exercise of discretion in setting wages is both more common and more important than is usually recognized in the labor econon~ics literature and, in particular, that the concept of fairness is a key to understanding how this discretion is used.

The Several Faces of Fairness

Perhaps the best known face of fairness is shown during collective bar- gaining and involves the concept that if workers in one union or group receive a certain wage increase, the workers in another union or group are entitled to the same increase. This concept was called "orbits of coercive comparison" by Ross (1948). Ross wrote that "equitable comparison links together a chain of wage bargains into a political system" and that "coin-

' For recent evidence on the importance and persistence of firm effects, see Groshen ( 1990).

parisons play a large and often dominant role as a standard of equity in the determination of wages under collective bargaining" (p. 50). Depending on traditions in their industry, unions may regard themselves as entitled to the same absolute increase previously won by another union or the same percentage increase. When I was a member of the Construction Industry Stabilization Committee in the 1970s, the construction unions generally insisted on preserving absolute differentials, but when I moved to wage stabilization in the food industry a few years later, comparisons of per- centage increases were more common. In both cases, union leaders were willing to participate in the administration of wage controls that reduced real wages in return for the power to protect established differentials. I have seen a union in the retail food industry allow an employer to close a statewide group of stores rather than agree to a wage increase smaller than that just received by its traditional comparison group.

The most striking instance I can recall of an institutionalized wage com- parison was disclosed at a hearing held by the Council on Wage and Price Stability in 1975. The business agent of a plumbers' local in the San Fran- cisco Bay area admitted the existence of an agreement known informally as "me too plus a dollar." If another pipe trade local in the area received a wage increase that brought it within $1.00 an hour of the rate received by his members, their employers would automatically pay this new rate plus $1.00. From the perspective of a wage stabilizer, the arrangement was not a pleasant one.

A second striking testimony to the importance of equitable comparisons came when the CEO of a large company called the Council on Wage and Price Stability to con~plain about a wage increase granted by a competing company to the union both companies bargained with. When I told him that the federal government could not help him, he protested that he would have to give the same percentage increase to his whole work force, including his nonunion employees. I suggested that he give the non- union employees the same absolute increase the union had received. "Are you suggesting," he replied "that I give the red-blooded Americans who built this company less than those radicals in the union? Never!"

In determining salaries in nonunion situations, comparisons are usually less compelling, but they are almost universally made. In both for profit and nonprofit organizations, when executives or directors set salaries they refer to surveys of salaries in comparable companies or institutions. (If no such survey exists, they will frequently organize or commission one, often at considerable cost.) They then decide where in the salary range disclosed by the survey they want their organization to be. It is worth noting in passing that most organizations state that they would like to be in the upper half of what they regard as the relevant salary distribution. It is, of course, impossible for more than half of them to be there, and the mac- roeconomic implications of these preferences are therefore more than a bit disturbing

Finally, individual employees who con~plain or file grievances about their wage or salary often express them in the form, "Why is my salary lower than X's?" followed by an argument based on merit or seniority: for example, "Why do I make less than X when I do better work than she does?" or "Why do I make less than Y when I have given this outfit devoted service for twenty years, while he was only hired last year?"

A variation on this theme is to complain that differentials are unfairly small. For example, the chairman of an economics department that is hiring a new Ph.D. at $38,000 for the academic year might get a complaint about salary compression from a full professor with 20 years of satisfactory service whose salary is $52,000. It might be noted that the full professor in this


example is not going to be consoled by information, however accurate, about the market for new assistant professors. He is arguing that under the circumstances he should be paid at least $60,000, and implicitly threat- ening to withhold effort if he is not. This is an example of the general proposition that wage inequities must be remedied by raising the wage that is too low, and not by lowering the one that is felt by others to be too high.

A worker may also express a wage grievance because he feels that some special effort or accomplishn~ent has not been adequately rewarded, which is a grievance not based on relativities. This is an argument usually made by individuals and only rarely by groups.

Incorporating Fairness into Wage Theory

The first question to ask about incorporating the concept of fairness into wage theory is whether it belongs on the demand or the supply side. Employers do not insist on fairness-workers and their unions do. How- ever, the supply involved is not primarily the supply of new workers. A new applicant for a job may consider the wage offered to be unfair in relation to his previous experience, but in this case he will simply decline the offer and the employer will go on to the next applicant. Both parties are making implicit comparisons with the general level of market wages for this type of work, but these are not the highly specific comparisons that give rise to fairness controversies.

Fairness primarily affects the supply of work from experienced, long- service workers, either because a sense of grievance will increase turnover or because workers who feel unfairly treated will withhold eff~rt.~

Of course, in a collective-bargaining situation, disputes over wage structure and wage relativities are a frequent cause of strikes.

In most general terms, what must be involved is that labor supply is

Mark Killingsworth has pointed out to me that most disputes about pay in- equities can be regarded as disputes about the division of the rents from firm- specific on-the-job training.

derived from a utility function that includes at least one argument in ad- dition to real wages and leisure, and this argument must involve the wages of one or more other people. The argument must enter the utility function in a negative way-that is, if B gets a wage increase and A does not, A's satisfaction decreases.

One of the few explicit expositions of this kind of utility function in the theoretical literature of labor economics is by Hamermesh (1975). He writes, "We assume that the utility-maximizing individual derives pleasure from having a wage above the average at any given level of the average wage of some comparison group that he can readily observe" (p. 425), and then proceeds to set forth a utility function that incorporates this assumption.

The general idea of interdependence of utility functions has also been used elsewhere in the economics literature to create a theoretical basis for charitable giving, which cannot be explained on a theory that holds that people always maximize their own resources, but in that case the utilities are positively related. When C gives to D, C's utility increases because he has made D better off (see Hochman and Rodgers 1969).

What is the difference between these two cases that causes the sign of the effect of interdependence to reverse? In the case of the wage comparison, the person with whom A compares himself is, at least in A's view, someone very much like A, while in the case of charitable giving, D is at least in some dimensions very unlike C. More specifically, A and B will generally have quite similar incomes, while C and D may have very different incomes, C's being higher than D's. In other cases C and D may have roughly similar permanent incomes, but D has suffered some major recent catas- trophe or loss. Finally, in the case of wage comparisons the person being compared with is usually close by-at the next work station or in the next town, while the beneficiary of charity may be half a world away, the victim of an Armenian earthquake or an Ethiopian famine. It is ironic that envy should be a local phenomenon while compassion is often a more dis- tant one.

But it is not enough to say that interpersonal comparisons enter the utility function with a particular sign. One must also add that they can enter in a very powerful way. Let us consider a hypothetical example. You are a professional worker in an organization in which each professional has a secretary. On Monday, you inform your secretary that she has received a pay increase of $20 per week. On Tuesday, she discovers that all the other secretaries in the organization have received increases of $30 per week. Clearly, she will be less happy on Tuesday than she was on Monday. Moreover, it is highly likely that she will be less happy on Tuesday than she was the previous Friday-that is, she would prefer that no one get an increase to getting a smaller one than her fellow workers. In this case, the magnitude of the effect of the comparative wage utility is greater than that of the own wage.

The Role of Demand

I do not mean to argue that the force of interpersonal comparisons by itself determines wage changes. The demand side must still be considered. An employer may resist union wage claims based on equitable comparisons on the ground that he cannot afford to meet them or that the workers are not worth the wages asked. If necessary, he can take a strike, and nowadays can often win it. In the case of wage grievances by individual employees, the employer or supervisor may simply reject the claim and face the con- sequences, even if the result is a quit. If the aggrieved worker stays in the job but conspicuously withholds effort, he or she can presumably be dis- missed for inadequate performance. The main threat to the employer is that the withholding of effort will be too small or subtle to serve as the basis for discipline.

To the extent that wages in different occupations in an enterprise are tied together in an accepted structure that it is costly to disturb, an increase in demand in one occupation can cause an increase in wages in other occupations in which demand is unchanged. This kind of linkage is much stronger in union than in nonunion situations, probably because in a union situation information on occupational rates is available to everyone.

Executive Compensation

The case in which comparisons most nearly determine the whole out- come of salary determination is that of executive pay. Although many executives get very high compensation, their compensation is usually a very small fraction of the total cost of compensation of the firm. The contribution of an executive to the output of the firm is extremely hard to measure, so that equity considerations, except in the case of conspicu- ously outstanding or poor performance, will loom large. Even very poor performance is more likely to result in reassignment or dismissal than in a substandard salary increase, much less a salary cut.

The con~parisons that drive executive pay are those with other executives in the same firm and with executives in the same function at other firms in the industry. For the top positions in large publicly held corporations, such comparisons are unusually easy to make because compensation for these positions is public information.

Those who set wages and salaries do not ordinarily make comparisons between the pay of executives and that of production workers or clerical workers. But if the power of equity considerations is constrained by demand in the case of low-level employees and not for those at the top, the result will be a widening gap in compensation between the top and the bottom. This seems to me to be what has been happening during the 1980s, although I cannot explain why it did not happen sooner. It would not be unusual now to find a total annual compensation for a CEO of $2,000,000 in an


organization where the pay of an entry-level worker is $20,000 or less: a ratio of 100 to one. Perhaps members of compensation committees and boards of directors or trustees will have to put a halt at some point to the game of keeping up with the Joneses and say that enough is enough. Their conscience is almost the only demand-side constraint that exists for ex- ecutive pay.

The defense offered for high executive salaries is that they are needed to motivate good performance, and recent research does show significant relationships between executive compensation and the performance of firms (Abowd 1990; Gibbons and Murphy 1990). It is one thing to show, how- ever, that performance-based compensation improves performance and another to show that motivating good performance requires rewards as large as those offered in recent years. The public is also disturbed by con- spicuous cases of large executive bonuses in firms that are losing market share, losing money, or even (as in the case of Drexel Burnham Lambert) about to go bankrupt.

Another result of the importance of comparisons is that what an executive receives depends less on who he is or what he does than on the league he plays in (Frank 1985). A conspicuous example of the possible beneficial effect of changing leagues occurred when Dr. Clifton R. Wharton, Jr., moved from being chancellor of the State University of New York at a salary in the 80 thousands to being president of TIAA-CREF at an initial salary of $500,000. It is true that he was badly underpaid at SUNY, but a good salary for that position at that time would not have exceeded $150,000.

The federal government in 1989 was advertising for a new director of the National Institutes of Health at a base salary below $90,000. The suc- cessful applicant was to administer an agency whose annual budget is $7 billion a year. A con~parable position in the corporate sector, if there were one, would command a salary at least 10 times this high. After the position had been open for several months, it was reported in the press that several distinguished medical administrators from the government and academic sectors had declined to be considered for it (NewYork Times 1989). The post remained unfilled for more than a year. Of course, there are often cases in which executives from other sectors accept government positions at salaries well below their previous ones. Some do so because they are already wealthy, some because their visibility and experience in government will enhance their compensation when they leave, and still others simply because they believe in public service.

The reluctance of Congress to raise its own pay and with it that of federal executives arises because voters are making a very different com- parison. The median voter earns far less than members of Congress. Judging by his own earnings and his view of what congressmen and "bureaucrats" are worth, he concludes that federal officials are overpaid. He therefore tends to vote against candidates for Congress who have supported congressional pay increases.

The Impact of Market Forces on Salary Structures

It often happens that a salary structure that is generally perceived as fair is disrupted when one member of the group receives a very attractive outside offer. Consider the case of an academic department in which the full professors receive salaries between $60,000 and $80,000 for the academic year. The differences within this range are based on a combination of merit and seniority and have not been giving rise to complaints. One of the best of these professors receives an offer from a comparable institution at a salary of $95,000 for the academic year. He announces his intention to accept the offer unless the salary is matched. The university administration informs the department chair that it is willing to match the offer, but that for budgetary reasons it cannot provide any special increases for other members of the department. The other members of the department are now likely to feel unfairly treated whichever option is chosen. If the offer is matched, the salary differentials within the department will be viewed as inequitably large. If it is not matched and the faculty member leaves, the remaining members will feel that they are being paid less than they could earn elsewhere. In the first case, the chair could promise that the differentials would be narrowed in subsequent years, but given typical budget constraints, this process could take a long time. Before normal differentials could be restored, some new outside offer might disrupt the structure again. Under these circumstances, pay equity is a goal that is constantly being pursued but is never reached. In this respect, it is not unlike market equilibrium.

Exactly the same sort of problems arise when a new member is hired into a department at a salary substantially above those already at the same rank because this salary was required to recruit him. In time, the chair can narrow or remove the differentials for those judged to be of comparable quality. Similar adjustments would not be made for those judged to be of lesser quality, perhaps because they are felt to be "past their prime." This can give rise to resentments and grievances from faculty members who do not share such judgment of their own worth. The case where quality is equal is one of a different supply price for marginal and inframarginal units, perhaps because of high cost of mobility. The case where quality is unequal is more complex.

Some universities maintain the same top salaries across different disci- plines even when the outside market pays different salaries; others try to match the market. In the second case, faculty members in disciplines with low outside salaries may feel unfairly treated because they view themselves to be as productive as their colleagues in fields with tighter markets. When salaries are kept equal across fields, administrators will attempt to adjust to the market through earlier promotion in the fields where there are shortages (Bowen 1964).

Two-Tier Wage Structures

During the 1980s, a number of unions for the first time negotiated two- tier wage structures in which newly hired employees were paid less than previously employed workers doing exactly the same work. Such schemes were introduced by employers whose markets were under severe compet- itive pressure from imports or from nonunion firms as an alternative to reducing wages of present employees. The unions accepted these schemes as the lesser of two evils. Such wage structures were particularly common in the airline industry and in the retail food industry. Some two-tier wage structures were designed to be temporary, but others were considered per- manent.

The persistence of two-tier wage structures will afford a test of the importance of the concept of fairness in wage determination. Those who believe that this is an important concept would predict that two-tier struc- tures cannot be permanent; they will eventually cause too much resentment by the workers receiving the lower rates. They will therefore be negotiated away when financial conditions in the industry improve or when the lower- paid workers become a majority in the bargaining unit. By 1989, we were beginning to see a number of settlements in which two-tier structures were eliminated or the wage differential between the tiers was reduced.

Homogeneous Labor

Earlier in this somewhat rambling article I have suggested that neo- classical wage theory fails to deal with the issue of fairness because it uses too simple a utility function. There is a second difficulty in the theory that is almost as important. This is the assumption that labor is homogenous or that all workers have equal productive capacity. It is, of course, perfectly consistent with neoclassical theory to have two or more classes of labor, each homogenous within itself and each treated as a separate factor of production. It requires only a minor modification of the theory to consider the case in which individual workers differ in ability but can be unambig- uously ranked (Rees 1973). Neither of these extensions of the theory goes far enough to reach the case actually facing the large firm. Such firms determine wages or salaries for hundreds or even thousands of occupations. In some of these, there will be a standard rate, but in many wages or salaries will be set individually for each employee. At this extreme, the assumption of homogenous labor breaks down utterly, and salary deter- mination will necessarily involve an element of arbitrary decision making or of bargaining.

Let me conclude by expressing a worry. I am concerned lest thi~ a .i ' . be interpreted as some kind of recantation, and I do not want to bc

17: 3

interpreted as saying that I now think that neoclassical theory is wrong. It is not, but it is incomplete. The most it can do is to set the stage for players such as personnel directors, union officers, and individual workers.

It cannot write their lines. It matters a great deal whether the demand for labor in an industry or occupation is rising or falling and whether or not a particular occupation is in short supply or in oversupply. These are the forces with which neoclassical theory deals. But given any configuration of these forces, the individual players usually still have substantial room for maneuver. In their maneuvering, they constantly struggle to preserve fairness as they see it. What makes this struggle so interesting to watch or to participate in is that each of them sees fairness differently.


Abowd, John. "Does Performance-based Managerial Compensation Affect Corporate Performance?" Industrial and Labor Relations Review 43 (February 1990) : 52-73.

Bowen, William G. "British University Salaries: Subject Differentials." In his Economic Aspects of Education: Three Essays. Princeton, N.J.: Princeton University, Industrial Relations Section, 1964.

Frank, Robert H. Choosing the Right Pond, chap. 1. New York and Oxford: Oxford University Press, 1985.

Gibbons, Robert, and Murphy, Kevin J. "Relative Performance Evaluation for Chief Executive OfTicers." Industrial and Labor Relations Review 43 (February 1990) : 30-5 1.

Groshen, Erica L. "How Wages Are Determined." In Economic Commen- tary. Cleveland: Federal Reserve Bank of Cleveland, February 15, 1990. Hamermesh, Daniel S. "Interdependence in the Labor Market." Economics 42 (November 1975): 420-29. Hochman, H. M., and Rodgers, J. D. "Pareto Optimal Distribution."

American Economic Review 59 (September 1969): 542-57. New York Times. (September 8, 1989), p. A12. Rees, Albert. The Economics of Work and Pay, p. 79. New York: Harper

Sr Row, 1973. Ross, Arthur M. Trade Union Wage Policy, chap. 3. Berkeley: University of California Press, 1948.

  • Recommend Us