Naturally, I wanted to dig into Keefe's numbers and understand how he did his analysis. I found a paper that does a step by step analysis with public/private wages. It does not address the total compensation (wages + benefits) but it is a good beginning to understanding how the analysis is done.
The paper by John Schmitt can be found here. It is highly readable, with great tables and figures to illustrate the data. This analysis is based on 2009 census data and it covers the US population, there is some information for individual states.
Here is a summary of the paper and the process.
1. Consider the educational level of the workers. Public sector workers are more educated that private sector workers. On page 4, the graph shows that and also a table shows you the ten largest occupations for state and local workers. It is no surprise that teaching professions top the lists.
2. Consider the gender of the workers. More women are public sector workers. On pg 5, he shows you the wage differential for all public workers is positive (+12.8) . Then he breaks it down for women (+19.2%) and men (+11.2%) Are you surprised that women have a larger advantage?
3. Control the data for gender, education level and age. Age is used as the variable that correlates with experience. If we believe that experience matters, you have to control for it. On pg 7, he shows that applying these controls, the wage differential for all public workers drops (it is -3.7%). For women (-1.9%) and men (-6.0%). Do the data reflects a gender bias against men? Are men paid poorly in the public sector compared to the private sector? Does the private sector discriminate against women? Can we know from this data?
4. Up to this point, the data was controlled for three variables, gender, age and educational level. Those controls are not applied in the next analysis. The next table on pg 8 shows the the differential for different wage levels. A wage distribution is done by percentiles. You can see the wage percentile along with the hourly earnings for that percentile and the differential. This chart also shows the differential broken out for men and women.
Let's take some time with this chart. For all employees, as the wage level goes up, the differential goes from positive to negative, (+5.9 to -11.3). This means that higher earners are paid less than their counterparts in the private sector. We assume these are the more educated and more experienced workers. Pg 9 shows this data in graphic form. It makes it clear that the differential changes from positive to negative for low and high wage workers respectively.
A side comment....The data is also broken out by gender. I find the 90th hourly earning percentile result interesting. For men it is $42.31, for women, $33.33. Is this gender discrimination? or could it be that men have more experience or more education? (Remember, this data isn't controled for education or experience, We assume that the higher wage workers are those who have more education and experience.) For the lowest wage workers, women have a much greater positive differential when compared to men. Is this gender discrimination? In the private sector? The public sector? Do we know? Isn't it an interesting question?
5. The appendix give data for individual states. On pg 14 you can see that Wisconsin is right in the middle of the pack with 13.5% of all employees working for the state and local governments. On pg 15, you get the actual numbers broken down to roughly 114,000 state employees and 220,000 local employees. On pg 17 you get the age differential, median worker age for private is 40 and 45 for state and local. Wisconsin public workers are older than public workers which we assumed means they are more experienced. You also get the % of workers with a college degree or more. Private is 27.4% and state and local is 54.9%. Wisconsin public workers are more educated that private workers.
Another side comment, do you believe these numbers? Take a few minutes to look at them and notice that there is no educational differential in DC (56.4% private vs 56.2 public%). Can you explain that unusual result? Does it seem reasonable.
Look at Arkansas, Wyoming and Missippi. The private sector educational % is below 20%. Do these numbers jive with your perceptions of these states?
I found the link to this paper here. This blog post addresses the all too common error of comparing the wages of private sector employees to public sector employees without addressing educational level and experience.
The blog post references the USA Today story that does give total compensation, but fails to control for educational level and experience. This is the only reference in the article to that argument:
"Economist Jeffrey Keefe of the liberal Economic Policy Institute says the analysis is misleading because it doesn't reflect factors such as education that result in higher pay for public employees."
I will return to the Jeffery Keefe (of the liberal Economic Policy Institute) paper with a better understanding of his analysis. I am grateful for the teachers and librarians who taught me to seek out sources of information. The internet puts all this at our fingertips!! They also gave me the confidence to know that I could look at statistical information and evaluate it myself. How does evaluation of government statistics on an important point of public debate makes one a liberal? I think that is name calling and it reflects a bias. I am happy to look at any analysis of this census data or BLS data that refutes that educational level and experience matter.
I have two questions for you to consider.
1. When you are compensated for your work, do you want to be compared to all workers? (70% of American workers do not have a college degree) Do we really believe a college degree is worth more compensation?
2. When you are compensated, do you want your employer to consider your work experience? or should you be compensated in comparison to all workers? (The older you are, the more experience you should have.) Do we really believe experience is worth more compensation?