Abstract
The articles by Stanberry and by Kilgour on equal pay reminded me that the Equal Pay Act was enacted 50 years ago. I drafted an original version of this editorial in response and sent it to the members of the Advisory Board for comment. They suggested a few changes. Then I received a paper with a somewhat different point of view from Duncan Brown, expressing the view from the United Kingdom (which will appear in the next issue).
His comments raise important issues about the management of compensation. His country along with the United States and others are increasingly concerned about the reality that is an economy with far too many low-pay jobs. Many of these jobs are female dominated. These are public policy issues but clearly worthy of discussion within the compensation community. The truth is that HR policies are at the heart of creating a work environment that is fair and gender neutral. This theme will be addressed in future articles.
The articles published here made me think about how different the world of work was when the Equal Pay Act was enacted. Many women chose to stay home and raise their children. Those who worked were limited to “female” occupations—teacher, nurse, secretary, librarian and a number of service jobs. The career alternatives then available to women had low pay, well below that of male-dominated jobs. But Congress chose to limit the protection of the law to women working in jobs that are essentially identical to those performed by men. The U.S. Congress and the federal courts have thus far rejected the comparable worth/pay equity argument.
Today women in the United States account for more than half of business school applicants and roughly two thirds of college graduates, and women account for more than half of the professional workforce in the United States and a number of countries. To borrow a phrase, women have come a long way. The equal pay issues today are very different, but the pay gap still exists.
As someone who first worked on this problem more than three decades ago, I can say unequivocally that there are valid reasons why the gap has not been closed. Furthermore, at least a portion of the gap is unrelated to conscious gender discrimination. Finally—and this is from someone who feels very strongly that everyone should have equal opportunities—the changes in private sector practices needed to eliminate the gap would be detrimental to employer success.
The gap calculations at the national level depend on “black box” multivariate statistical methods. It’s not that the statistical results and gap estimates are wrong; they are, however, misleading because key variables are frequently not included and/or they are based on data that do not accurately reflect the dynamics governing pay levels. For example, using “degree level” ignores the significant differences in market pay levels for education, liberal arts, business and engineering majors. Similarly, age is not an accurate proxy for experience.
Perhaps the most important variable is the difference related to career choice. It is still true that women are more likely to choose comparatively low-paying occupations. (I am aware this is a chicken-and-egg argument.) Moreover, their career choice is also more likely to channel them into public sector jobs. It’s not possible to ignore the combination of supply-and-demand dynamics and political reality. In today’s environment, with government budget deficits, pay adjustment funds are not readily available.
For employers, the comparable worth argument would require the installation of a defensible job evaluation system and having all employees paid under a single salary system. In the 1980s, women and unions pressured elected officials at all levels of government to enact legislation that would mandate employers to take that step. They succeeded in gaining acceptance for their argument for public employers in a few states—for example, Minnesota and Washington. There are similar requirements for selected employers in Canada and Australia.
One of the barriers in the United States is the obvious fact that private sector employers maintain two (i.e., exempt and nonexempt) or more pay systems. That practice is not required, but it facilitates compliance with the Fair Labor Standards Act. It’s also increasingly common to rely on different pay strategies for different job families.
But from my perspective, the overriding barrier is that the recognized problems in both the design and the administration of traditional job evaluation systems make this unrealistic and unworkable. The existing systems would also perpetuate prevailing internal pay relationships.
First, however, it is important to recognize that pockets of gender discrimination continue to exist. The Stanberry article focuses on one example. My experience suggests, however, that in professionally managed organizations, discrimination in the pay of women in the same jobs as men would today be relatively rare. This editorial should not be interpreted as condoning any form of discrimination. The argument here is intended only to discourage attempts to pass legislation mandating changes in the way pay is managed based on questionable practices.
The Problems With Job Evaluation
Recent surveys show the majority of U.S. companies have jettisoned the formal, quantitative job evaluation systems. However, those systems continue to be important outside of the United States. Furthermore, public employers continue to rely on a variation, job classification, which is even more bureaucratic and costly to maintain. This editorial was drafted for those employers that cling to these outdated systems.
The earliest job evaluation systems date to the 1920s. The Hay system was introduced in the late 1940s. In 1963, larger companies relied on variations of a point factor job evaluation system. (There were and are millions of small employers that have never installed a formal system.) There were virtually no changes until computer-based systems were developed in the 1980s in response to the demands of comparable worth advocates. The logic of those early systems has been continued in all quantitative job evaluation systems.
It was in the late 1970s that I first heard the phrase “comparable worth.” I was managing a consulting practice in Manhattan, which is not important other than that my contacts included several individuals then leaders in the compensation community. It also gave me the opportunity to develop contacts among the more prominent comparable worth advocates. We were friends but had some great arguments.
To keep it brief, I conceived the model for computer-based job evaluation systems and developed the prototype in conjunction with Michael Graham, who was then at Bausch & Lomb and now is on this journal’s advisory board. When computers control pay levels, gender can be explicitly omitted as a pay determinant and also tested for significance. It was and is a viable alternative to those 50-year-old systems, but the problem was few practitioners or employees—who presumably should understand the rationale governing their pay—had an adequate understanding of the basis for those systems, multiple regression analysis.
The time invested in the development of that model and subsequent experience also made me very much aware of the weaknesses in traditional systems. One problem in particular has been exacerbated by the recession. They require HR specialists to administer them. The U.S. Bureau of Labor Statistics keeps track of employment by occupation. In 2007 there were 110,000 “Compensation, Benefits, and Job Analysis” specialists; today there are 85,600.
Traditionally job analysts interviewed (or observed) job incumbents and drafted the position descriptions (PDs). Early in my career, I recall reading an article that discussed the time involved in drafting PDs. The estimate was 12 hours, which included (a) the time to interview both incumbents and supervisors, (b) the time to draft the PD, (c) the time for managers to review it, and (d) the final editing and approval time. For employers that frequently reorganize or redefine jobs, that’s a never-ending cycle.
As a point of contrast, when people discuss their jobs at social occasions, it takes no more than a couple of minutes to tell someone what we do.
A related problem is that job analysts as well as incumbents “see” the job through their eyes and draft PDs using their vocabulary. At best job documentation is suspect. And as anyone who has dealt with these issues knows, the words and phrases used can influence evaluation decisions.
No questionnaire or other data-gathering strategy can confirm the facts. So employees and managers commonly exaggerate the importance of the job—it’s to their benefit to do so.
Studies years ago confirmed the obvious—jobs whose description of duties and responsibilities is inflated are evaluated higher. That should not be surprising. With the disappearance of job analysts and the reliance on prewritten or incumbent written PDs, the problem is no doubt worse today—but hardly anyone is paying attention.
There are also problems with the typical process to evaluate jobs. It’s a myth that job evaluation is objective. Anyone who has been involved in the meetings of a job evaluation committee knows horse trading is central to the process. Managers know they will “win some and lose some.” Studies have confirmed that more powerful managers can control the decisions. That’s not necessarily bad because everyone coming out of those discussions is likely to agree with the decisions—but it’s very possible for bias of one form or another to creep in.
The argument that the process is based on “job measurement” is also worthy of comment. Proponents of point factor systems would contend that jobs are evaluated using interval scales, but that’s a stretch since the intervals are arbitrary. Some would argue that those are ordinal scales, which are far less rigorous. Moreover, combining points is also arbitrary. That is far below the standard commonly accepted as measurement.
These considerations should not be surprising when the “compensable factors” are abstract and vague, and when the process requires the comparison of obviously very different jobs. For example, the decisions to evaluate the “accountability” of an engineer and a marketing analyst—especially when no one involved has a hands-on knowledge of the jobs—would be difficult to verify.
The critics in the 1980s argued that the compensable factors in existing job evaluation systems under valued the duties and requisite skills of female-dominated jobs. They argue that job evaluation systems should reflect a balanced approach to male and female job characteristics. They want to see factors like accountability and authority balanced with issues like “ability to listen,” “caring and tending” and “mental stress” that are important to women’s jobs. Their basic point is that job evaluation should be gender neutral.
Job evaluation systems can be designed to be gender neutral. The experience in the 1980s proved that is possible. However, the traditional job evaluation methods, primarily today’s point factor systems, have their origins almost a century ago. There is probably no other area of human endeavor that has changed less over the years. From strictly a business perspective, the problems and added cost outweigh the possible benefits.
That is reflected in a 1990 Ed Lawler book, Strategic Pay. His discussion covers both the advantages and disadvantages of job evaluation. He devotes 3 pages to the advantages and 13 to the disadvantages. The disadvantages are as follows:
Promotes bureaucratic management style
Implicitly specifies what not to do
Reinforces the job hierarchy
Depersonalizes value orientation
Fosters internal focus
Impairs strategic orientation
Discourages organizational change
Encourages point grabbing
Erodes honesty/credibility
Inflates pay system operating costs
Fails to encourage skill development
Makes promotions too important
Rewards wrong behavior
The world of work has changed a lot since 1990, but nothing on his list would be less valid today. If anything, organizations are far more dynamic now and more concerned about being able to respond quickly to market developments.
Lawler could have added that job evaluation decisions are often unpopular and at times contentious. Even if HR wins an argument, it loses. Playing the “pay police” role diminishes HR’s influence.
I know from experience in the 1980s that one of very real problems is that job evaluation impedes organizational change. When a merger or reorganization is announced, employees know that jobs will be redefined. They also know some will be paid more, some less—and their concern prompts resistance. Plus it takes the time of everyone to document and reevaluate all the jobs.
I also know job evaluation increases the payroll. To recall the graphs from that hated college statistics course, with regression lines, roughly half the points are about the line and half below. With a so-called policy line “fitted” to market rates, some jobs are paid above market—that’s inevitable. Interestingly, as high-demand, high-pay jobs increase in importance, it tips the regression line, which could reduce the pay of jobs at the lower end.
There Are Alternatives
Employers need a rational process to determine pay levels. Job leveling is a phrase that has come into common use. Ideally it should be transparent, readily explained and defensible. Ideally it should be based on data or a simple logic that is widely understood. Employees need to believe that they will be treated fairly.
It is also important in light of the significant decline in the HR specialists available to manage pay systems that the process requires minimal staff support. Computers of course need to play a role, but the experience in the 1980s demonstrates that the “answers” cannot be locked in and computer generated.
Market pricing now is the dominant practice. Relying on market data provides a simple, easily communicated logic. (The problems with surveys would warrant a separate article.) Market pay levels, of course, reflect whatever bias exists to hold down the pay of women. But that only makes it more important to use surveys and market data intelligently.
Survey data of course are not available for every job. Years ago the American Management Association developed and marketed a simple framework for overcoming this problem and addressing internal pay relationships. It’s based on a matrix with salary grades down the side and organizational functions across the top. A committee uses available market data to assign jobs to one of the boxes in the matrix. They can move jobs up or down to reflect the appropriate internal relationships. That essentially combines the use of market data with whole job ranking. The nonbenchmark jobs are then slotted into the boxes using job ranking. It’s simple, credible and easy to maintain.
Research again years ago confirmed whole job ranking can produce essentially the same answers as the more “rigorous” and costly application of a point factor system.
The design of job evaluation systems depends heavily on decisions that have little, if any, empirical justification. New systems are often designed to perpetuate existing internal pay relationships. The selection of factors and weights is based on tradition and intuitive judgment (e.g., of course job knowledge is important). Once we acknowledge that the design choices are largely arbitrary, it opens the door to relying on simpler methods including negotiated grade assignments. Job evaluation committees could be allowed to develop their own methods, assuming all parties—including women—are represented.
Perhaps the most radical alternative is a pay program based on salary bands since it shifts the focus from jobs to career ladders and individual capabilities. The idea was actually introduced 30+ years ago as an alternative to the problems and costs associated with job evaluation in an anticipated merger. A federal research lab is often credited with the idea. In 1980 the proposed merger of two naval labs prompted the commanding officer to tell his HR director that he was not willing to wait for all the jobs to be reclassified.
A research lab of course has a highly educated workforce, so the decision was to adopt a program model similar to those common with college faculties (e.g., full professor, assistant professor). In higher education, promotions and salary increases are commonly based on professional accomplishments and recognized expertise. The banding idea provides considerably more flexibility that the traditional, bureaucratic model. Nevertheless, it’s been highly successful and copied in a number of smaller federal agencies.
The banded-program model extends that pay philosophy to other sectors. Banded programs often rely on competency-based pay and career ladders. When combined, these ideas emphasize individual growth and development and introduce a reward philosophy that “fits” knowledge organizations.
As economies recover, the so-called ‘war for talent’, triggered by aging workforces and the ever-increasing importance of professional expertise, will make the demand for the best qualified workers an organizational priority. Critics sometimes question the idea of market-based pay systems, but their concerns thus far have been overridden by the urgent need for talent. To borrow another statement, “the only sustainable source of competitive advantage is people.”
In this environment, it is difficult to imagine any employer consciously discriminating to hold down the pay of women or any other talent pool. Individual expertise is a business necessity.
This only scratches the surface. The firms and independent consultants represented on the advisory board no doubt have developed their own program variations that are better answers than the outdated approaches from the past.