It is referred to in the New York Times article about the federal pay freeze:
I have been unable to find the actual study (just a lot of summaries) but this article refers to some of the issues*:
The federal workforce is an obvious first target, if one fraught with political risk for a president who relies on union support. Critics have said the federal workforce has been protected from the ravages of the economy. Chris Edwards of the Cato Institute referred to federal workers, in a study in June, as "an elite island of secure and high-paid workers, separated from the ocean of average American workers."
Mr. Edwards found that federal civilian workers had an average annual wage of $81,258 in 2009, compared with $50,464 for the nation's private-sector workers. Average federal salaries rose 58 percent from 2000 to 2009, compared with 30 percent in the private sector, according to his study.
If this is correct then it's impossible to carry out the kind of study Mr. Edwards argues he has completed. The same article continues:
The dispute over how well federal workers are paid zigzags on. An early-October opinion poll by The Washington Post shows the public apparently has been influenced by the news media and possibly by the reports the Cato Institute and Heritage Foundation released this summer. Three-fourths of the Post's respondents think federal workers "get better pay and benefits than people doing similar work outside the government."
But analyses by the Office of Personnel Management continue to show that federal employees are underpaid. OPM's compensation team leader reported to the Federal Salary Council on Oct. 29 that the gap actually widened during the past year and now stands at 24 percent. That is a statistical estimate because many federal employees do not have counterparts in nonfederal organizations.
The fact is, the data relevant to this comparison have not been assembled for almost 20 years. So the public can be reacting only to the vague generalizations that have dominated this debate now for months. Both sides have claimed they have completed analyses confirming their argument, but actual data for actual workers "doing similar work" have not been produced.
Does Mr. Edward's study cover the military compensations and the increases in them over time? The pay freeze will not apply to military pay.
There is, however, a significant discrepancy in the benefit cost data Cato and Heritage report. It is not clear why they differ from the federal budget data. Cato reported the total compensation -- the combination of salary, bonus awards and benefits -- as $123,045 in 2009 -- or 51 percent more than the average salary.
For that year, the executive branch expenditure for benefits, as shown in the current budget, was 36.7 percent and essentially the same in 2010. When that percentage is added to the base salary, the total compensation is $98,455 -- or 20 percent below the Cato figure.
The U.S. Postal Service's benefits were higher, as were the military's benefits. The budget shows the cost of benefits for the entire government was 42 percent of aggregate salaries. Those numbers do not include the additional expenditures for retirees. It would be misleading to include those expenditures in a comparison of total compensation for today's public and private workforces.
The crucial point, of course, is to compare apples to apples and the overall benefits packages. It looks like this is not something that can be done with recent data.
*Thanks to some guy at Eschaton for the link.