We Should Probably Stop Citing EPI’s “Teacher Wage Gap” Data

When writing about teacher pay, it’s tempting to cite the Economic Policy Institute’s work on the “teacher wage gap.” As of 2019, EPI authors Sylvia Allegretto and Lawrence Mishel purport to find the teacher wage penalty hitting an all-time high of 21.4 percent.

It’s a compelling statistic that aligns with much of what we hear from teachers across the country. But the underlying methodology is flawed in four key ways that have become increasingly apparent over time:

Flaw #1: It’s measured in weekly wages. 

Allegretto and Mishel are quite transparent that they are measuring the weekly wages of teachers. On one level this makes sense, because teachers need current wages to pay for their current expenses, everything from mortgage payments to groceries to childcare.

But teachers aren’t paid like everyone else. Teachers work plenty hard during the school year, but they typically only teach for 10 or 11 months of the year. As others have pointed out, differences in weekly wages don’t account for the differences in time spent across an entire year.

Still, I wouldn’t classify this as a fatal flaw in the data as long as they are accompanied by appropriate caveats. Other flaws, however, start to compound the issues.

Flaw #2: EPI’s benefit calculations are incomplete. 

To EPI’s credit, they do attempt to factor in the relative advantage teachers have in benefit spending. They’ve improved that methodology over time, and their main analysis attempts to combine the teacher salary gap and the teacher benefit advantage into one overall picture of total compensation.

However, as I pointed out in The Hill earlier this spring, EPI’s benefits methodology is still inaccurate. The comparison group they use includes teachers–that is, EPI is comparing public school teachers to a larger group that also includes teachers, and in fact teachers are largest sub-group within that comparison group. This decision biases the EPI results and makes the teacher benefit advantage appear smaller than it really is. Once you correct for that, total teacher compensation has not budged relative to other professions in at least 10 years.

Moreover, EPI’s calculations take the current benefit spending at face value. If states and districts are under-counting their pension benefits, as many economists believe, then more realistic assumptions would drive the teacher benefit advantages even further away from their private-sector peers.

I would probably rate this as a mid-level concern, but the problems are starting to mount.

Flaw #3: EPI’s state comparisons drop the benefit advantage entirely. 

In their national figures, Allegretto and Mishel carefully combine the teacher wage gap and the teacher benefit advantage to come up with a figure for the total compensation gap. Even though I disagree with EPI’s calculations on benefits, they at least deserve credit for attempting to balance out the changes in salary with the changes in benefits.

But then the authors proceed to drop benefits entirely when they report state-level data. These figures have been endlessly repeated in state and national media stories, but it’s simply irresponsible on EPI’s part to even include them absent more context. In many states, rising benefit costs fully account for flat or declining teacher salaries. EPI seems to recognize this in their national figures even as they ignore it in their state-level section.

Up to this point, I think the EPI wage gap figures might still be worth using given the appropriate context and caveats. But the next problem undermines them entirely.

Flaw #4: EPI’s wage gap methodology assumes credentials matter equally in all settings. 

At first blush, EPI’s wage gap methodology might make sense. They calculate the teacher wage gap as “penalties that remain after controlling for education, experience, state, and other factors known to affect wage levels. Generally, we express the teacher wage penalty as a percent disadvantage—how much less, in percentage terms, the average teacher earns relative to a similar college graduate in another profession.”

But this calculation starts to unravel if you know anything about teacher credentials in the education sector. That is, researchers have concluded that Master’s degrees and advanced credentials do not translate into better teaching performance, and yet school districts across the country have tied teacher pay to teacher credentials. As a result, teachers are now some of the most credentialed professionals in the country: 57 percent of public school teachers have a Master’s degree, up 10 percentage points in just the last 15 years.

As Andrew Biggs and Jason Richwine write in a new piece for National Affairs, teachers are now in roughly the 95th percentile when it comes to educational credentials. Under EPI’s assumptions, teachers would have to be paid accordingly in order to be paid fairly. But Biggs and Richwine find some alarming conclusions when they apply the same EPI methodology to other occupations:

EPI’s own pay-gap methodology leads to some other conclusions that are, to put it delicately, less intuitive. Using the same Census data and the same basic techniques that EPI applies to teachers, we find that registered nurses are “overpaid” by 29%. Meanwhile, telemarketers deserve a big raise, as they currently suffer a 26% salary penalty. Aerospace engineers are apparently overpaid by 38%, but “athletes, coaches, and umpires” are paid 21% less than their skills are worth. Photographers should consider going on strike, as they make 16% less than comparable workers. Firefighters are moochers by contrast, taking in 25% above their rightful salaries.

These stats should make anyone question the EPI comparisons. Worse, EPI’s findings of a rising teacher wage gap have been compounded by the changes in the teaching profession. As teachers have gotten more and more advanced degrees, teaching as an occupational group has moved up the credentials ladder. That does not mean, however, that those Master’s degrees would (or should!) be equally valued in the private sector.

To be clear, none of this confirms that teachers are under- or over-paid. Mike Antonucci has read the same studies and concluded that the narrative is more important than the underlying data. But to my mind, I’d still rather have more information about whether schools are able to recruit and retain high-quality teachers, whether higher salaries can solve those issues and for whom, and what happens to teachers who leave the profession. The research consensus so far–which Biggs and Richwine cite–suggests that ex-teachers do not earn higher wages in their next job. A recent paper on ex-teachers in Texas suggests that consensus may be wrong, and it may be skewed downward by people who leave the workforce entirely. Moreover, these types of analyses confirm that different types of teachers have different employment opportunities available to them.

That type of information would allow us to adjust compensation structures accordingly. It’s tempting to look at the sorts of synthetic comparisons that the EPI report constructs, but what we really need to know is if teacher compensation structures–in their unique labor market context–are set up to attract and retain a high-quality teacher workforce.

–Guest post by Chad Aldeman