The Federal Salary Council is scheduled to meet tomorrow to “review the results of pay comparisons and formulate its recommendations on pay comparison methods.” This follows comments in the 2017 report from the president’s Pay Agent about “major methodological concerns.”
In the spring, the council reported the pay gap was 32 percent. Somehow despite announced increases averaging only 2.1 percent for 2017 and 1.9 percent for 2018, the gap closed from 34 percent in 2016.
Actually, it's surprising the gap is still that large. When the announced increases are combined with step increases, promotion increases and Quality Step Increases, the totals have been fully competitive. The average step increase, based on the number of employees at each step, adds roughly 1.4 percent to salaries every year. When the annual combined percentage increases are compounded over the past decade, federal salaries, by my calculation, have increased 46 percent. During the same 10-year period, the Employment Cost Index increased 28 percent.
It would be helpful if the Council explained how to interpret the 32 percent gap. With employees spread across 15 pay grades and a growing number of locality schedules, it cannot be that all salaries are that far below non-federal levels. Closing the gap—increasing the average General Schedule salary—$81,246 according to FedScope—by 32 percent produces a salary of $107,103. That’s more than double the average pay of U.S. workers.
That highlights a core problem. The annual analysis has become a convoluted Rube Goldberg contrivance, with a continuously changing sample of employers, the use of two very different Bureau of Labor Statistics surveys, and statistical equations spitting out estimates used in the final gap calculation. It’s a classic black box calculation.
But more to the point is that in the current gap analysis, the salaries of federal jobs are not actually compared with the salaries of non-federal workers. Neither of the two Bureau of Labor Statistics surveys used in the gap estimate was designed to tell us how well federal jobs are paid. Despite what must be millions spent on the surveys and gap analysis, the end product does not and cannot tell us how well a federal job is paid relative to labor market levels. That would require separate analyses.
There are three core problems: 1) the Federal Employee Pay Comparability Act of 1990 requires reliance on BLS surveys; 2) the calculations focus narrowly on the needed increase in GS salary ranges (as someone who was involved in the planning that went into FEPCA, I can state the current methodology was not anticipated); and 3) the GS system no longer meets the need. Simply stated, it’s out of sync with today’s labor markets and with the way compensation programs are managed. The underlying pay philosophy and approach to administering employee salaries originated in the Classification Act of 1923. It’s now an impediment to competing for talent.
No other employer in the country uses a similar methodology. Moreover, no employer relies on the survey data developed by BLS to administer its pay program.
The Critics’ Argument
Analyses by think tanks and the Congressional Budget Office (“Comparing the Compensation of Federal and Private-Sector Employees, 2011 to 2015”) show federal employees are overpaid. Studies by economists going back 40 years have reached similar conclusions. Each study may well be technically solid, but their conclusions are not helpful in assessing federal salaries.
The basic methodology was first used by economist Gary Becker, a Nobel Prize winner. He is credited with introducing the phrase “human capital.” His analyses confirm what we now take for granted—worker pay increases with added education and experience.
That is easily shown today with broader databases and software to facilitate multivariate analyses. In the statistical models developed by the critics, the information available in population surveys—education level, occupation, years of work experience, geographic location, size of employer, and certain demographic characteristics (age, sex, race, ethnicity, marital status) and a so-called dummy variable (1/0) to differentiate public and private are combined to explain employee compensation. Not to get too technical but the dummy variable’s coefficient shows the pay differential between public and private employers.
The analyses are impressive but overly simplistic. Private employers do not pay explicitly for education, age or experience. The equations also assume a college degree is a constant factor in explaining an individual’s salary throughout their career. It also assumes a year of experience has a fixed value, ignoring type of job and employer. Most important, these analyses have no value for salary planning since they ignore factors relevant to pay decisions—an individual’s specific experience, the quality of schools and prior employers, recent performance, etc.
The analyses also ignore the comparative nature of many federal jobs. Realistically, the problems confronting federal employees in many fields are far more complex than in state and local government, have far greater consequences, and impact larger budgets. That is also true for workers in all but a small percentage of the millions of companies in the United States. Population surveys include data from millions of small, simple businesses. The assumption is that salaries should reflect each job’s level of accountability and impact.
The Business Approach to Salary Planning
The Pay Agent’s report highlighted two concerns: “The value of employee benefits is completely excluded from the pay comparisons” since the focus is limited to salaries, and “the comparisons of federal vs. non-federal wages and salaries fail to reflect the reality of labor market shortages and excesses.” Both are valid problems that should be addressed.
The business approach to market pricing is discussed in textbooks and compensation workshops. It’s used with minimal variation across all non-government sectors. The logic is straightforward and easy to understand—employers identify their competitors for talent, compile data from surveys that include those employers, and adjust salaries to sustain the planned alignment with competitor pay levels.
Surveys are conducted annually in virtually every industry. It’s possible to find data for virtually every occupation and geographic area. In the Washington D.C. area the largest is the survey conducted by the Human Resource Association of the National Capital Area that reports data for several hundred local area employers, including government contractors. Most of the surveys would be available at a nominal cost for federal use.
The same logic is relevant in comparing benefits. As a guess, 95 percent of employers provide little more than legally required benefits. But few of those companies are competing with government for talent. The mom-and-pop businesses are not relevant. Plus total compensation should include incentives (the BLS data does not).
For practical reasons, benefit information is normally collected in separate surveys from cash compensation. Linking benefit information with salary data is suspect because the survey participants are not the same.
Benefit analyses are heavily dependent on assumptions about turnover, age at retirement, future earnings, male/female mix, etc. Government’s workforce is one of the oldest. That and low turnover drives up the costs. Any comparisons should be based on employees with similar characteristics.
The GS system is no longer supporting good government. It should be replaced. Its archaic, bureaucratic and costly to maintain. Salaries for new grads and technical occupations are not competitive. That should be confirmed with an independent market analysis. Leaders and employees both need to know salaries are fair and warranted.
If it cannot be totally replaced, FEPCA “authorizes OPM to establish special occupational pay systems for those occupations that it determines should not be . . . subject to the General Schedule.” In 1990 that was intended for the patient care specialists now paid under Title 38. Today it could be the basis for new pay systems for high demand fields. New systems should be based solidly on market data.