Jason Richwine of the Heritage Foundation and I have published a new revision of our working paper “Comparing Federal and Private Sector Compensation” that includes improved estimates of compensation differences between federal and private-sector employees through salaries, benefits, and job security. We made significant improvements in all three areas, using better data and methods to produce more precise estimates.
The basic approach was to compare compensation for federal employees to private-sector employees with the same earnings-related attributes, principally education and experience but also controlling for gender, race, immigration and marital status, region, and broad occupational categories. This approach tries to calculate what a federal employee would earn were he or she to work in the private sector, thereby isolating the federal pay penalty or premium.
We begin with salaries, where we find that federal employees receive salaries around 14 percent higher on average than similar private-sector workers. Less educated federal workers received the largest pay premium and the premium rises as job tenure increases. We supplemented our basic salary calculations with a “fixed effects” approach, which follows single individuals as they shift in or out of federal jobs. This method shows that the same person will tend to earn a higher salary in federal employment than in the private sector. And it debunks the often-heard claim that federal employees could earn more on the outside—in fact, most people who leave federal employment experience a salary cut, while most who enter federal employment receive a pay increase.
But the premium in salaries pales next to that paid through benefits. Federal workers receive 76 percent more paid time off than similar private-sector workers and twice as much supplemental pay, which includes overtime and holiday pay and bonuses. But the biggest difference is in terms of retirement benefits: improved methodology for calculating the value of defined-benefit pensions, data on a hidden $2 billion annual subsidy to the defined contribution Thrift Savings Plan, and calculations of the value of federal retiree health benefits show that overall compensation through retirement benefits is 5.6 times higher for federal workers than in the private sector. Once all benefits are included, the federal compensation premium rises to around 35 percent.
On top of this, federal workers have far better job security than private-sector employees. Once a federal employee has passed his probationary period, his annual probability of being terminated for cause is only around 0.27 percent, while the chance of being laid off is only around 0.02 percent. This job security has value. We found, for instance, that congressional staff—who don’t have the job protections of other federal employees—are given significantly higher pension benefits in explicit compensation for lower job security. And federal employee job security becomes even more valuable when it is protecting a position that pays higher salaries and benefits than an employee would receive in the private sector. Adding the value of job security, the total federal compensation premium rises to 61 percent. In other words, most federal employees would accept significant reductions in pay before they would quit and look for a private-sector position.
Overall, the federal pay premium in 2011 comes to around $77 billion. Over ten years, paying federal employees at market rates could save the budget almost $1 trillion. Overcompensation of the federal workforce isn’t the biggest problem facing the federal budget, but when you’re as deep in the hole as we are, we need to look for savings everywhere.
print this page
Mr. Biggs, to justify the job security premium, you cite the pension benefits of members of Congress as explicit compensation for their lower job security. I’m not aware of a similar explicit compensation for job insecurity in the private sector. Private sector employers don’t compensate employees for job insecurity by giving them higher pension payments. So, aside from the single, isolated case of Congress, there isn’t any empirical justification for the job security premium you claim. Moreover, whatever the federal salary premium is, it would make no difference whether federal employment turnover was 0% or 20%, because the fired employee would simply be replaced by another. “Job security” is just another way of saying that the salary premium is difficult to trim. Placing a dollar value on job security is just double-counting. That observation alone woud make your savings estimate only $570 billion … still substantial, but hardly the $1 trillion you asset.
No, it’s not double-counting. The job security premium means that instead of paying people in salary, you’re effectively paying them in job security – job security is something that’s valuable to workers and can more or less have a value placed on it financially. So in order to equalize total payment between public and private, you can pay them less monetarily and keep job security the same. Additionally and separately, there is an actual wage premium.
Tom, You rarely get explicit (meaning, spelled out) compensating wage differentials in the private sector, for a variety of reasons, one of which is that private sector pay basically responds to supply and demand. Let’s say that Alaskan fishing boats aren’t getting enough job applicants. Partly it’s due to being dangerous work, partly due to being generally unpleasant, partly due to who knows what. All the employer knows is that he’s got to raise pay until he gets the employees he needs. Now, he probably understands the factors that make his job more or less attractive to people, but it’s the market that decides how much to pay. Government is unusual in that things are sometimes spelled out more explicitly. (Why they compensated through pensions rather than some other form of compensation is anybody’s guess.) However, there is evidence of compensating wage differentials in the private sector; more dangerous or unpleasant jobs do pay more than other jobs that demand similar skills. However, as we note in our paper, prior research has shown technical reasons why it’s hard to divine these pay differences directly out of wage data, namely that you need to know a lot more about individual productivity than most datasets allow you to. But there’s no disagreement (literally going back to Adam Smith) that compensating wage differentials exist.
I agree with Emily that we’re not double-counting. Job security has a value even if you’re not getting a pay premium, so they’re different issues. We can calculate a job security premium assuming that a federal employee who loses his job (which pays a salary/benefits premium) then regains his federal job after a period of unemployment, in which case the value job security is only protection against lost wages during the period of unemployment. We can also calculate a second case in which, after losing a federal job, you need to find a private sector position where you lose your federal salary/benefits premium. That gives a different, higher value. We’re very confident that we’re not double counting.
Great post. Thank you for doing all of this work.