Late last week, Sheldon Friedman, chairman of the committee that oversees the Federal Wage System, defended the 22 percent gap identified in the most recent Pay Agent report. He said federal officials had taken "a more detailed look at the National Compensation Survey [the Bureau of Labor Statistics] did to look at actual job and work level matches. We pulled that out in more detail for lot of relative occupations. Not surprisingly, we reproduced the 22 percent gap." But Friedman didn't provide specific data to support his claim. On Wednesday, OPM Director John Berry said his agency, OMB and Labor, would review the methodology to determine whether improvements are needed.
Which side is correct? The fact is we truly don't know. Both sides rely on analytical methods that even the most astute economists would have trouble understanding. And neither has detailed job-to-job comparisons to support their arguments. The competing analyses are totally different than those used by employers in other sectors. The simple truth is that the federal government is spending roughly $100 billion on the salaries of its white-collar workforce, and we don't know whether the funds are wisely spent.
The last time an in-depth comparative analysis of federal pay was conducted was in 1990 as a background step leading to the passage of the Federal Employees Pay Comparability Act. I was the managing consultant for that analysis. We relied on a database that combined a number of surveys. We determined the pay gap was real and varied by job category and metropolitan area. This conclusion was consistent with previous analyses. I met with federal executive groups in several cities and learned agencies were experiencing serious problems hiring qualified employees. FEPCA was fully warranted. Congress and the White House agreed the gap should be closed over a period of years. But the schedule laid out in the law was never followed.
Although we did not realize it when the law was drafted, a continuing problem is that FEPCA requires the use of Bureau of Labor Statistics survey data. For years, the basis for adjusting General Schedule salary ranges was a BLS survey of "benchmark" jobs. But BLS switched to a new survey in the mid-1990s known as the National Compensation Survey. The President's Pay Agent and OPM have been struggling ever since to develop a credible way to use the NCS data. Their efforts have resulted in a Rube Goldberg methodology that defies understanding.
For a time, BLS tried to convince employers outside government to use the data, but that effort failed. They found that employers are not willing to base salary planning on a market analysis that:
- Does not generate job-specific results.
- Uses a database where 20 percent of the participants change every year.
- Reflects invalid assumptions about wage dynamics in local labor markets.
- Relies primarily on estimated rather than actual pay data.
- Is based on analytical methods that few people understand.
In every other sector, employers routinely rely on industry surveys based on benchmark jobs. There are hospital surveys, technology surveys, state government surveys, association surveys -- every employer can find surveys relevant to its operations. More than 2,000 such studies are conducted every year across the country.
Some of the surveys are private, and results are available only to participants. Many, however, could be assembled for federal use. But BLS complicates that possibility in another way: Its experts have set a high standard for survey design. Statistical validity is not an issue for other employers; the deficiencies in surveys are acknowledged but never discussed in human resources periodicals. Survey quality certainly varies. But in the federal environment, survey data is scrutinized very closely, and any problems become potential ammunition for critics. The old saying, "Good enough for government," is turned on its head in this context: "Good enough for every employer but government."
Still, the resolution of this debate depends less on the quality of surveys and more on what exactly is being surveyed. Federal employees are not going to be convinced unless they see what their counterparts in other sectors are paid. Senior accountants, for example, want to know what other senior accountants are paid. That is the simple reason surveys based on benchmark jobs are so widely used.
There is, to be sure, the argument that federal jobs are more complex and difficult than similar occupations in the private sector, but somehow I doubt whether any of the large federal contractors would agree. There is also the argument that federal workers are older and better educated, but that again depends on the basis for comparison.
Those arguments do, however, highlight a central issue. Employers select surveys based on other employers competing in the same labor markets. But the BLS surveys now collect data from a sample selected from any establishment with at least one employee. The overwhelming majority of those employers are in no way competing with the federal government for talent.
Federal agencies actually compete in very different labor markets across the country. When the Veterans Affairs Department wants to hire nurses, for example, it is competing with hospitals across the country. But the NCS survey was not designed for the kind of job-specific analyses that would help discern how agencies are competing in various markets.
The labor market issue, in my opinion, should be revisited. There's a reason other employers rely on benchmark surveys and market data to adjust their pay systems. Local market surveys have been the basis for adjusting Federal Wage System pay levels for years.
It's clear by now that the current methodology for comparing pay rates is not serving government. It has lost credibility. No one is satisfied. The lack of transparency makes the federal pay system an easy target for critics.
The recently enacted health reform and financial regulation laws make it even more important for federal agencies to be able to hire qualified employees. The wounded soldiers returning from war make patient care specialists vitally important. And the problems on Wall Street demand regulators who can hold their own with some of the best minds in the business world. Federal agencies need a pay system that enables them to compete for talent, facilitates communication with employees, and silences the critics.
Howard Risher is an independent compensation and performance management consultant. He was the managing consultant for the studies leading to the 1990 Federal Employees Pay Comparability Act. He has a Ph.D. in labor and economics from the Wharton School at the University of Pennsylvania.