The Public-Private Wage Gap, Part II

Filed Under (Uncategorized) by Nolan Miller on Jul 7, 2010

This week, I’m following up on my earlier post regarding the gap between the wages earned by public and private sector workers.  Given the financial problems faced by state and local governments across the country, the idea that overpaid public employees are, in part, responsible for these problems (and that, by extension, lowering their compensation should play a part in solving the problems) has gotten a lot of attention lately.  However, the best evidence we have on the point, not to mention common sense, suggests that the main factual claim underlying this argument – that state and local employees are overpaid on average – is wrong.

The issue of comparing wages across sectors is complicated.  Fortunately, a recent report from the National Institute on Retirement Security provides a nice introduction to the difficulties of the analysis and a comparison of the compensation earned by state/local workers to that of the private sector that paints a very different picture than the one that has been in the newspapers lately.  To be sure, the NIRS study is not without a point of view, and I think that the analysis is off in places, but it is certainly closer to correct than the naïve figures driving the headlines.

First and foremost, we must recognize that government jobs are different than private sector jobs and government workers are different than private sector workers.  In many cases, public jobs (e.g., police, fire fighters) have no private counterparts and vice versa. It is impossible to compare what a police officer earns in the public and private sectors, since there are no private police officers.  In addition, even when jobs have the same basic title, government jobs often have very different responsibilities than their private sector counterparts.  This results in, for example, almost 54% of state and local workers having college degrees, compared to only about 28% in the private sector.

These differences suggest that comparing average wages across sectors, as is done in the USAToday and BLS studies I cited last time, is likely to get the answer wrong due to differences in the composition of the two workforces.  An alternative approach, and the one that is used in the NIRS study, asks the question “given a person’s age, education, experience and other demographic factors, how much more would he or she be paid in the private sector than he would be paid in the public sector?”  Using this “people-based” approach, the analysis comes up with the answer that a typical public employee’s wage is about 11% lower than it would be if that person were employed in the private sector, reversing the basic finding of the newspaper analysis.  (In Illinois, the difference is more like 12 – 13%.)  Even when the study takes into account that public sector employees earn more in benefits than private workers, the gap is still about 7%.  So, the study’s basic finding is that state/local employees have, on average, lower total compensation than private-sector workers.

The NIRS study does a lot to dispel the idea that average state/local employees are earning more than they would in the private sector – getting fat off the taxpayer, as it were.  However, it is not without problems of its own.  The first, which the paper recognizes in a technical appendix, is that the analysis should also take into account that the mix of jobs done by state/local workers differs from that of the private sector.  Controlling for this shrinks the wage gap somewhat.  The second, which Jeff has written about, is that typical government accounting methods undervalue benefits, in particular they undervalue pension benefits.  Properly valuing pensions would further erode the gap.  In the end, I suspect that the gap between state/local (and federal for that matter) workers and private sector workers is quite small and is not systematically either positive or negative.  In other words, markets probably get things about right.

As a final note, here are a couple of economist “smell tests” for wages.  First, when our department tries to hire someone, we are competing against other public and private employers.  When we succeed in hiring someone, we have negotiated to the point where they are willing to accept our offer when compared to their alternatives.  So, almost by definition, the value they place on total compensation at the two places (relative to the demands of the job) should be pretty close.  Of course, since they chose to come here, they must like our packages somewhat better, but if we did a decent job of negotiating, not too much better.  Second, if total compensation in the government sector is systematically higher or lower than those of the private sector, we should see massive flows of workers in one direction or the other.  The fact that we don’t see this suggests that people are, by and large, happy with the jobs they have.  This, again, suggests that the market has wages about right.

The Public-Private Pay Gap and Simpson’s Paradox

Filed Under (Other Topics) by Nolan Miller on Jun 21, 2010

With government budget deficits rising across the country, there has been much ado lately about whether government employees are overpaid.  In many cases, attention has been focused on a few unusual cases – football coaches, university presidents, retiring public employees drawing large pensions, etc.  This is fun, and it fans the fires during a period of growing anti-government sentiment, but it doesn’t do much to clarify the question of whether public sector employees are overpaid.  After all, if you are going to look at the highest-paid public employees, shouldn’t you compare them to the highest-paid private ones?  And, compared to a typical hedge fund manager, the salaries earned by football coaches and university presidents is tiny.

 No, the right question is not whether there are a few public employees that are paid a lot.  There are.  And, arguably, there’s nothing wrong with a person being paid a lot.  What would be wrong is if the person is paid dramatically more in the public sector than he or she would be in the private sector due to inefficiency or corruption in the public system.  And, if it were true that the average public employee’s total compensation were higher that could be received in the private sector, this would suggest there are problems with the entire system.  Despite the occasional highly-paid private employee, there seems to be little evidence that total compensation for public employees, on average, is higher than total compensation for those workers were they employed in the private sector instead.

 Two recent stories have pushed the idea that government workers are overpaid.  The first was a USA Today story from earlier this spring that argued that, focusing on occupations that exist both in government and the private sector, in 2008 the average federal employee earned $67,691 while the average private sector employee earned $60,046.  Benefits for the average federal employee were $40,785 per worker while private sector employees received $9,882 per worker on average.  So, federal employees earned a higher salary and more benefits, and total compensation for a typical federal employee was about $38,500 higher than in the private sector.

A BLS report released last week tells a similar story regarding the gap between state and local worker total compensation and that of the private sector.  The report finds that total compensation for state and local workers averaged about $39.81 per hour (in March 2010), while private worker total compensation averaged about $27.73 per hour.  So, the total cost of an average state/local worker is about $12 more per hour than a typical private worker.

Of course, these comparison cannot possibly be right.  It simply doesn’t pass the smell test.  I’ve known people who left the private sector for government work, and they never did so for higher salary.  Shorter hours, yes.  Better benefits, yes.  More job security, yes.  But better pay?  If the average federal employee got almost $40,000 more in total compensation each year than he or she could earn in the private sector, we should expect to see people clamoring to leave their private sector jobs for federal jobs.  The same is true for state/local workers.  We should see a line of private-sector workers trying to move into lucrative government work.  We don’t really seem to see either of these.

The most likely cause of the suspicious numbers is the fact that government employees, as a group, are different than private sector employees.  If government employees are, on average, more educated and more experienced than private sector employees are, on average, then this could account for the difference.  To take a concrete example, suppose the private sector and public sector each hire 10 employees.   Ignore benefits cost to make things simple.  The private sector pays lawyers $120,000 and clerks $40,000 each and hires 3 lawyers and 7 clerks.  The average cost of a private employee is $64,000.  The public sector, on the other hand, pays lawyers $100,000 and clerks $30,000 and hires 5 lawyers and 5 clerks.  Despite the fact that the government pays lawyers less than the private sector and clerks less than the private sector, the average cost of a government employee is $65,000 per worker in this example.   Even though the government pays both high-skill (lawyers) and low-skill (clerks) workers less than the private sector, the average compensation cost in the public sector is higher due to the fact that it employs a greater proportion of high-skill workers!

The previous paragraph illustrates a phenomenon known as Simpson’s Paradox.  I first learned of Simpson’s paradox in the context of airline delays.  To take a simplified version, compare two airports, Seattle and Phoenix.  (I’m making all these numbers up.)  Suppose that when it rains in Seattle the typical flight is delayed 15 minutes and when it doesn’t rain the typical flight is delayed 0 minutes.  In Phoenix, on the other hand, when it rains the typical flight is delayed 30 minutes and when it doesn’t rain the typical flight is delayed 5 minutes.  Now, suppose I were to tell you that the average flight out of Seattle is delayed longer than the average flight out of Phoenix.  How would you make sense of that?  Your response might be “well, that’s because it rains all the time in Seattle!” and that would be the right answer.  Suppose it never rains in Phoenix.  The average delay is 5 minutes.  If it rains 2/3 of the time, the average delay in Seattle is 10 minutes, and 10 is more than 5.  In fact, as long as it rains more than 1/3 of the time, average delays in Seattle will be longer than average delays in Phoenix, despite the fact that Seattle has shorter delays both on days when it rains and days when it doesn’t rain!

Enough for today.  Next time I’ll write about this report, which does a better job with the data and claims to show that government workers are underpaid.  It isn’t perfect either.