How Well Are Federal Employees Really Paid?

Every year the Federal Salary Council and Pay Agent release reports showing federal employees are badly underpaid. According to a table in the Pay Agent report from last December, the national pay gap was 34 percent. In the Washington, D.C., locality area, the gap was 51 percent across all General Schedule grades and all white collar occupations. It’s simply not credible.

To translate that into salary dollars, new graduates starting in Washington at GS-7, Step 1, would be paid $69,680, not the current $45,972.  The average salary for new graduates last year was close to $50,000. That would certainly increase interest in federal jobs.

Does anyone believe the results of the gap analysis? How many people actually read the reports? How many understand how the gaps are determined? Those who have read the reports may have noticed only percentages are reported, not salary dollars.  

Recently at a civil service reform town hall, Office of Personnel Management Director Jeff Pon said that government needed to “collect data” on compensation and “right-size” the pay for different government occupations. He asked the perennial question, “Hey, are we overpaying some occupations and underpaying others?” His question cannot be answered with government survey data, but every HR head needs to be able answer the question.

Significantly, the logic behind the General Schedule and its administration does not recognize that a job can be overpaid if it’s properly classified. But the growing list of special rates confirms they can be underpaid.

Today’s Bureau of Labor Statistics surveys were not planned to generate the job-specific market data that’s needed. It’s telling that when an agency applies for a special rate it is required to support its request with market data.

When off-budget federal agencies need to assess market pay levels, they typically contract with consulting firms that rely on data from surveys they or other firms conduct. The widely used surveys report pay data for benchmark or commonly defined jobs (e.g., Financial Analyst II). Currently, only Defense Department surveys supporting the Federal Wage System and Veterans Affairs surveys under Title 38 report market salary levels.

The use of market data is basic to salary management, which accounts for why there are more than 1,000 surveys conducted each year across the country. A Google search for the phrase “hospital salary surveys” produced 12.5 million hits. Survey data are available for the majority of federal occupations, with pay levels reported for career stages.

Pon has also floated ideas for reducing benefits. That’s understandable—although clearly not popular—for an obvious reason: the federal benefit package is rich and costly. In 2017, a Congressional Budget Office analysis estimated the cost is 47 percent higher than for private sector employees.

Pon’s concern should not be surprising. Every HR head has to be able justify the funds allocated to workforce management. That’s especially true for public employers and not-for-profits. Dollars spent for benefits necessarily reduce the money available for wages and salaries. That’s basic to total compensation planning.

Benefit planning is no different than salary planning. A core question is: What benefits are provided by other employers competing for talent? An additional question for planning federal benefit changes is: What will help agencies recruit recent graduates? Research shows Millennials prefer savings plans because of the portability, transparency and control over investments. The problem now is balancing the promises made to older workers with the need to compete for those recent graduates.

The market pay mystery originated in the mid-1990s when BLS ended its surveys focused on benchmark jobs and salary levels. The switch was to the Employment Cost Index (ECI) survey, a time series focused on labor cost increases. BLS waved its statistical magic wand to combine data from ECI with the Occupational Employment Statistics (OES) program but that does not make it possible to report data relevant to salary planning.

The BLS/OPM approach differs in every respect from the steps described in textbooks and workshops on compensation planning. Instead of simple descriptive statistics, multivariate models are used to produce estimates of pay at each GS level even if jobs do not exist locally. A discussion of the methodology fills 22 pages! Actually the BLS approach is the opposite of what textbooks describe—all survey jobs are “classified” to a GS grade level, which means occupational pay differences are lost in the averaging. At no time are survey salaries matched or compared with federal salaries.

What is generally overlooked is the size of the database used to calculate the pay gap. The ECI is based on a national sample of roughly 35,000 “observations” in government and non-government establishments, sampled from over 5.5 million businesses and close to 100,000 public employers. The OES program includes summary data for millions of workers but it’s aggregated by employers so the validity of the data cannot be confirmed.  

By comparison, the annual Washington-Baltimore area Human Resources Association of the National Capital Area compensation survey provides pay information for over 50,000 local employees. The broad-based national surveys used by businesses include pay data for a million or more employees, with the data reported for benchmark jobs and summarized by industry, location, etc.  

A final data analysis point relevant primarily to managerial and professional jobs is that the BLS data does not include bonus/incentive payments based on company or group performance or on profit sharing schemes. Those payments are integral to understanding market pay levels.

To deepen the mystery, a Congressional Budget Office report last year showed that when the annual GS adjustments are added to “merit, seniority and all other” increases from 2007 to 2016, the compounded increase was 47.9 percent. The comparable number from surveys of salary increase budgets conducted by WorldatWork, the compensation and benefits association, was 33.6 percent. But the gap did not close.  

Going forward, based on Pon’s comments, the pay program will be modified to more closely align salaries with market levels for occupations. That would be consistent with recommendations in several National Academy of Public Administration reports going back almost three decades. Already government has separate pay programs for law enforcement and Foreign Service officers as well as many healthcare specialists, plus the thousands of special rates.  

This mystery can only be solved with the facts. There is a wealth of available market data for this purpose.