Mark Twain was on to something when he popularized the line, “There are three kinds of lies: lies, damned lies and statistics.” When comparisons of federal and private-sector pay are based on complex statistical models, no one can prove you’re wrong. But it’s an impasse when both sides rely on black box statistics. Neither side can prove it’s right or the other wrong. The losers are government and federal employees.
There is no resolution in sight. Gap estimates are forgotten days after they are announced. This column was triggered by the latest gap estimate: that federal employees are underpaid by 34.6 percent in comparison with the private sector. That’s an 8.3 percentage point increase in a year. When private-sector salaries increased less than 2 percent, that is not credible.
The U.S. has roughly 6 million employers, and virtually every one monitors pay levels in relevant labor markets to manage compensation. Mom-and-pop employers do it informally. Larger employers rely on salary surveys. Surveys are produced for virtually every industry, geographic area and occupation. One firm in the business of selling survey data has a list of more than 3,500 surveys on its website.
No other employer would rely on the methodology used by the Labor Department’s Bureau of Labor Statistics. The universal practice — except for BLS — is to rely on benchmark job surveys. Yes, BLS collects pay data — the National Compensation Survey and the Occupational Employment Statistics; these surveys no doubt are professionally planned, but neither relies on a methodology useful for pay comparisons.
Benchmark jobs are generic and defined in more or less the same way in many organizations. In benchmark job surveys, data are compiled by job family and three or four job levels. Survey data for chemists, for example, would typically be reported for four levels: entry; intermediate or developmental; full performance/journeyman and senior; and expert.
Employers match their jobs to the benchmarks. It’s not brain surgery, but it’s credible and transparent. This is the common basis for salary planning across all sectors, including state and local government. BLS relied on a benchmark survey until the mid-1990s.
The use of benchmark pay data is not precise and can be criticized. But the BLS methodology, along with the analyses of the think tanks and the Congressional Budget Office, also can be criticized. The basic problem is this: Statistical models cannot confirm that a secretary or accountant or any job is paid fairly. The decided advantage of benchmark job data is that all employees can easily see how their pay compares with the levels in other organizations. When a job is overpaid or underpaid relative to market levels, the evidence is clear. Statistical models are not useful in salary planning.
There are several benchmark job surveys in the Washington area, as there are in every major city. The largest is the annual survey sponsored by the Human Resource Association of the National Capital Area. The 2012 survey includes data from 281 employers for 317,000 employees in 512 benchmark jobs. Participants are broadly representative of public and private employers in the area. The database is far larger than that compiled by BLS. Data are reported by job family and job level, by industry, by size and by location. The survey benchmarks were matched by federal classifiers with government jobs.
Few jobs in the survey paid 30 percent or more below the local averages. In-house attorneys and engineers lead a short list. But most senior-level jobs are within plus or minus 10 percent or so. That includes most technology jobs. Yes, some jobs are paid more than market. As an estimate, federal salaries job by job are roughly 7 percent below the local averages. Other local surveys show a similar pattern.
The General Schedule and the classification system are at the heart of the problem. Government is trying to impose the hierarchy of jobs as it existed in 1949. Then, maybe new engineers might have been paid the same as liberal arts majors. Now engineers commonly start at $60,000 or higher, and liberal arts majors at $35,000, according to the National Association of Colleges and Employers. The average is $50,000. GS-5s currently start at $34,075 in Washington, and the “rest of U.S.’’ locality-pay area at $31,315.
The pay gap no doubt is one of the reasons only 3 percent of recent graduates are planning federal careers. This is at a time when something like 500,000 federal employees are eligible to retire.
The GS system cannot be “aligned” with market rates. Someone at a higher grade level than mine will have to decide what is fair pay for federal employees, but the GS system is not the answer. Government needs to be able to hire the critical specialists in high-demand occupations.
Howard Risher is a consultant and writer on federal pay and performance issues. He was the managing consultant for the studies leading to the 1990 Federal Employees Pay Comparability Act and is author of “Planning Wage and Salary Programs.”