I will protect your pensions. Nothing about your pension is going to change when I am governor. - Chris Christie, "An Open Letter to the Teachers of NJ" October, 2009

Wednesday, July 15, 2020

A Response To Critics Of My Research On NJ Teachers

Earlier this week, the Sunshine Sunlight Policy Center put out a critique of my report, published by the New Jersey Policy Perspective, on the shrinking numbers of teacher candidates in preparation programs in New Jersey. That report references an earlier report I wrote for NJPP on the NJ teacher workforce; that report looked at, among many other things, teacher compensation. SPC has also criticized that report, focusing on the sections on teacher wages. 

There's a small echo chamber for this sort of stuff here in New Jersey: a couple of bloggers will pump it up and add their thoughts, "reform" groups will pass it around social media, etc. What's curious about this last round of criticism, however, is that several people contacted me to say they had received news of SPC's report through email, even though they never signed up to receive messages. People receiving unsolicited copies of criticisms of my work is a new one for me -- I can only assume I really touched a nerve.

I don't have a problem with anyone critiquing my work; in fact, I regularly write reviews of research from think tanks for the National Education Policy Center. But it's a rigorous process, with a clear format and several layers of editorial review. That keeps a reviewer from just flinging anything they think of on the wall and hoping something sticks.

I'm not going to go through and do a point-by-point rebuttal of SPC's criticisms. Many are, to be frank, silly: arguing, for example, that I made a point about a limitation of the data in an appendix as opposed to the main text is nit-picking of the highest order.

But let me quickly get to the main issues:

- Teacher wage modeling. SPC makes no objection to the functional form or the covariates included in my wage models. They do object that my data does not disaggregate private and public school teachers -- a fact I point out and write about in my report. I have to wonder if SPC would have even made its objection on this point had I not brought it up.

SPC tries to use a graph I included, which shows the difference in reported teacher wages between two different data sources, to determine the gap between private and public school teacher wages. But the graph is not showing this; it's showing the difference between two data sources: the NJ Department of Education's salary data, and the IPUMS data I used for the report.

SPC suggests substituting the NJDOE salary data for the IPUMS data would narrow the wage gap I show for teachers. But substituting one data source for another, just for teachers, is wholly invalid. The IPUMS data are survey data; the NJDOE data are salaries reported from all school districts. The reported salaries from a survey could be easily biased downward for all respondents (some people reporting take-home pay instead of before-tax earnings, for example). And there is no way to know just how many private school teachers are included the survey.

When you do an analysis like this, you go with the best data you've got, use it uniformly, and note the limitations and possible biases. I've done that. Trying to ding me for it is petty.

- Other research. SPC makes a big deal about how other research comes to conclusions that are different than mine. Again: I'm the one who brought up all the other research and compared it to my own. I admit in my report that the latest iteration of the Economic Policy Institute's teacher wage gap model shows a significant change for New Jersey teachers, with a much smaller wage gap than they found the year before. How can I be "burying" it when I'm the one who brings it up?

It is odd to see such a large swing in one year, so, yes, I consider the latest EPI report an "outlier." We'll see what their future research shows. This is what social science is: looking at a problem from various angles with different data and comparing the results. What is not valid is what SPC does when it tries to prove a teacher wage advantage: pulling out results from various reports that align closest with its predilections, then combining them even though they use different methods and different data. That's an undergraduate mistake and a bad indulgence in confirmation bias.

- Teacher benefits. Central to the issue of teacher compensation are whether and to what extent benefits, such a health insurance and pensions, make up for the wage gap. I am very clear in my report: I make no attempt to determine whether this is the case. I cite other work that has made the attempt, and I cite research on the generosity of NJ teacher pensions and benefits relative to other states. SPC says: "Citing irrelevant research obfuscates rather than illuminates." I choose not to underestimate the intelligence of my readers and include these citations; I have faith that they are perfectly capable of putting these references into the proper context.

- "Underpaid" teachers. The issue of estimating the worth of benefits brings up a larger point: how can we model all of the aspects of teaching as a career -- pay, benefits, personal satisfaction, etc. -- to determine whether teachers are underpaid? The answer is that we can't. I have known teachers who have told me they would work for free if they had to because they love the job so much. If New Jersey could find enough of these people, it could slash taxes and our schools would still thrive.

Obviously, however, that's not how the real world works. The relevant policy question is whether teacher compensation is high enough to continue to attract enough qualified workers into the profession. It is wholly germane to that discussion to note that teachers make less in wages, on average*, than similarly educated workers, even when holding other factors constant. It's especially relevant when some NJ policymakers constantly talk about reducing teacher benefits, as those benefits are helping to close this wage gap.

- Teacher candidates. SPC admits there is "...a real and worrisome decline in the number of teacher candidates, and that New Jersey’s education system needs a large pool of qualified teacher candidates." Yet that is the primary conclusion of my report.

I'll say it again: SPC agrees with the primary conclusion of my report.

Their objection, then, isn't with my data or analysis-- it's with my contention that this decline is likely linked to teacher compensation and working conditions. Is it so far fetched to assume that compensation might be involved here? I include several citations in my first report that show that compensation does, in fact, influence workers' decisions to go into teaching. Is SPC really trying to deny this? Are they arguing that compensation doesn't affect people's career choices?

SPC argues that I have not shown a direct causal link between teacher policies in New Jersey and the decline in candidates. It's an odd argument to make when SPC also contends that millennials' attitudes toward work are affecting teacher recruitment, but offers no causal evidence to support their contention; apparently, only my arguments have to withstand their dictated level of scrutiny.

Here is what we know: we have fewer teacher prep enrollees and candidates per 1000 students than in the last ten years. During that time, teacher benefits degraded (thanks to Chapter 78). There's strong evidence teachers were already behind on wages compared to similarly educated workers. Research shows teacher compensation affects decisions to enter the profession.

SPC can dance around all they want, but those are the facts; if they find them inconvenient, that's on them.

- What about the rest of it? If you only read SPC's criticisms you would think all I've reported on was teacher compensation and the decline in teacher candidates. In fact, that was only a part of these reports. I actually consider other sections to be more important.

  • The teaching workforce does not look like the student population. The majority of teachers are white females, yet that's not the case for NJ's student population. We've got to get more people of color into teaching, and yet the number of Black and Latinx teacher candidates is falling.
  • Teacher wages are not similar across districts of different socio-economic status. It appears that the most affluent districts are willing to pay a wage premium to experienced teachers with advanced degrees. This is a question of equity that needs to be addressed.
  • SPC says I don't talk about barriers to entry into the teaching profession. In fact, I discuss it specifically with regards to racial bias, which is a serious problem. I also reference recent work by Drew Gitomer on EdTPA, a problematic hurdle for student teachers.
SPC doesn't address these issues of race and class as related to teacher compensation and recruitment. I would urge them to leverage some of what appear to be their substantial resources toward these topics; perhaps SPC's staff could convince their funders that they are important.

* One of the bloggers who joined in on criticizing my work has made a big deal about comparing "mean" and "median" wages between teachers and other workers. If anyone wants to get into a whole discussion of the validity of using a quantile regression method in teacher wage modeling, be my guest. I'd just note the data is censored at $250K, so it's really not that big of a deal. (Yes, I should have noted that in the report.)

UPDATE: This is incorrect: the data is censured in the aggregate when reported, but not the micro data I use. So yes, non-teacher data skews upward; I regret the error. However, my point in the next paragraph stands. (Also, after checking again, I can report there are no 7-figure income figures in the data; highest observation is $720K.)

It's also odd to argue that I'm overestimating the teacher wage gap because the top of the pay distribution for other workers is much higher than it is for teachers. The fact that a teacher in the highest pay quantile will never make what a similarly positioned lawyer makes doesn't much help the argument that teachers don't suffer from a wage gap.

Key and Peele explain it better than I can:


StateAidGuy said...

I think it's pretty basic why median compensation, not mean compensation, is the more accurate representation of what people are earning, since mean income includes high-income outliers. Since 24% of NJ's income is earned by the top 1%, using mean incomes inflates what the _typical_ person in the private sector is earning.

You're right, no teacher, even the best of the best, is going to make a seven figure income, but teachers' lowest incomes are higher too. According to the TPAF valuation report, 1st year teachers in NJ average $58,000 in salary. There are a lot of college grads who earn less than that, especially now in the Coronavirus Recession.

Again, very few people the private sector have seven-figure incomes but those incomes are included in the calculations of averages and skew them upwards.

The difference between median and mean income is not trivial. The US's mean household income is at least $24,000 per year higher than the median income. NJ's mean household income is $30,000 a year higher than the median.

The point of this is that the inclusions of private school teachers in the teacher salary (who are significantly lower paid) and high-income outliers among the data for private sector college grads have the effect of making the teacher/non-teacher pay gap look larger than it is between public school teachers and typical private sector college grads.

Duke said...

Jeff, read the post. The data are censured. Those 7 figure salaries aren't in the data. And again: how would you propose I specify a quantile regression?

And once more: I am the one who brought up the private school teacher salaries potentially being part of the data. There's not a doubt in my mind SPC wouldn't have even considered the issue if I hadn't brought it up. I addressed it as best I could. If you think I've been dishonest... well, I find that sad. You've been a keen observer of NJ school finance in the past. Maybe someday we can have good faith arguments again about this stuff.

Duke said...

I updated the post because my last comment was incorrect. The data are censured, but only in aggregate reports. The microdata I use is not. But I stand by my main point: the highest paid teacher never comes close in salary to the highest paid non-teacher. So a median difference arguably masks another pay disparity between teachers and non-teachers.

StateAidGuy said...

"It's also odd to argue that I'm overestimating the teacher wage gap because the top of the pay distribution for other workers is much higher than it is for teachers. The fact that a teacher in the highest pay quantile will never make what a similarly positioned lawyer makes doesn't much help the argument that teachers don't suffer from a wage gap."

I think a difference between the public sector and private sector is that the private sector is higher risk, higher reward. When someone starts a business, odds are, that business will fail, and even if it lasts, it might have years of low profits during recessions. Yes, if the business succeeds the owner might become wealthy, but the odds are always against that happening. Even if the business comes out with a quality product, the business might still fail. For instance, some industries, like tech, are notoriously winner-take-all.

A difference between teachers and college-educated private sector workers is that private sector workers include people who have been promoted and who perform a different task than they did when they started out their careers.

Teachers enhance their skills and they should get salary increases as they gain experience, but, by definition, a teacher is still a teacher even after 20 years, whereas someone in the private sector might have a managerial role that comes with a higher salary because it carries more responsibility.

When a teacher becomes a "manager," ie, assistant principal, principal, assistant superintendent, or superintendent, the ex-teacher indeed gets a higher salary, but that person is no longer considered a teacher and so the higher salary isn't baked into average salary data for teachers that you are using.

If you were going to argue that education "managers," like superintendents, are underpaid, then I would agree with you. Roger Leon's base salary is $260,000, and he doesn't have a lot of job security. A CEO running a business as complex and large as the Newark Public Schools would get way more than that.

Anyway, so I don't think the presence of high-income private sector outliers and their skewing of salary data, indicates that NJ teachers (as a class) are necessarily underpaid.

DeShawn Reed said...
This comment has been removed by a blog administrator.