A Quality-Based Look At Seniority-Based Layoffs
** Also posted here on “Valerie Strauss’ Answer Sheet” in the Washington Post
Eliminating seniority-based layoffs is a policy idea that is making the rounds these days, with proponents making special appeals to cash-strapped states and districts desperately looking for ways to save money while minimizing decreases in the quality of services. Mayors, editorial boards, and others have joined in the chorus.
There’s a few existing high-quality simulations that compare seniority-based layoffs with one alternative – laying off based on teachers’ value-added scores (most recently, one analysis of Washington State and another using data from New York City; both are worth reading). Unsurprisingly, the simulations show that the two policies would not lay off the same teachers, and that the seniority-based layoffs would save less money for the same number of dismissals (since the least experienced teachers are paid less). In addition, the teachers laid off based on seniority have lower average value-added scores than those laid off based on those value-added scores (as would inevitably be the case).
Based in part on these and other analyses, critics have a pretty solid argument on the surface: Seniority makes us “fire good teachers” simply because they don’t have enough experience, and we can fire fewer teachers if we use “quality” instead of seniority.
To be clear: I think that there is a sound case for exploring alternatives to seniority-based layoffs, but many of the recent arguments for so-called “quality-based” layoffs have been so simplistic and reactionary that they may actually serve to deter serious conversations about how to change these practices.
The first issue is that critics of seniority-based layoffs often conflate the need to cut costs and the need to get rid of incompetent teachers. This is a distraction. Districts must address the problem of incompetent teachers, whether or not they are in budget crisis. Similarly, districts in budget crisis need to figure out how to deal with it, whether or not they have incompetent teachers. These are separate issues, and we should be careful not to let one drive the other.
The second (and more practical) source of confusion pertains to how current seniority-based layoffs actually work. In reality, they rarely proceed strictly by seniority. Although there are variations from place to place, there is usually a formula of sorts, which accounts for subject area licensure, special skills, location, and other factors, in addition to experience. For example, the aforementioned Washington State simulation (which used actual layoff notices issued [but repealed] in 2008-09) found that experience was the strongest predictor of being laid off, but there were also strong effects associated with having an advanced degree and being certified in a high-needs subject. In fact, about one in three teachers who would have been dismissed under seniority were in their third year or higher, while over 10 percent had five or more years of experience. In other words, termination decisions were made according to multiple criteria (though they tended to be based on need).
So, those who rail against seniority-based policies – using oversimplified characterizations such as “last in, first out” – sometimes appear rather out of touch with the reality of how these policies actually work, and how much they vary by location. The public (and perhaps some policymakers as well) may not be aware of these details, and we should be clear about the current situation before deciding whether and how to change it.
Third, all of the outrage against seniority seems way overblown. It has for decades been considered a fair and impartial way of proceeding, in both the public as well as private sectors (though it is far more common in the former, and among unionized employees in both sectors). In education, this policy also has some research backing: Even by the narrow measure of student test score growth, experience is among the few proven signals of teaching quality (see here, here, here, here, or our summary here), to say nothing of the possibility that experience matters more when it comes to other student learning outcomes (including, by the way, reducing attrition; experienced teachers are less likely to leave the profession).
In short, seniority is definitely imperfect, but it is hardly outrageous to use it as a proxy for quality. There is a reason why districts have long agreed to use it in layoffs and other decisions, and why virtually every OECD nation uses it in determining teacher pay. Let’s cool it with the “seniority hurts kids” rhetoric.
(Side note: In both the New York and Washington simulations, seniority-based layoffs did not affect higher-poverty schools more than “quality-based” terminations, as some previous analyses have found, with much fanfare.)
The fourth and final issue with the anti-seniority uproar is probably the most important for policy purposes: The implication that there is a better alternative that is willfully being overlooked. Most critics simply attack the existing policies without specifying a replacement, which is really the most important element here. New evaluation systems would seem a viable option, but in most places, this work is still very much in progress. Completing it should be our focus, rather than rushing to eliminate seniority before we have a solid method with which to replace it. Until we do, any unqualified argument that we should select based on "quality" amounts to little more than a talking point.
I have not yet heard anyone state publicly that we should conduct layoffs based solely on value-added scores, but assuming these people exist, their policy preference raises a few important questions and issues that bear on the debate. First off, and most obviously, one has to wonder how they plan to account for the fact that the vast majority of teachers do not get these scores (i.e., they teach untested grades/ subjects). How are they to be selected?
But let’s say we did it – we chose who to lay off based on value-added estimates alone. Putting aside all issues of fairness and applicability, the achievement benefits are not quite as grand as you may have heard. The estimates from growth models, due mostly to random error and measurement error, are highly unstable over time. This means that the teachers who are “saved” from being laid off by a “quality-based” policy (i.e., they would have been laid off under a seniority system) will be a fraction as effective in future years compared with the initial layoff year (and, even in the initial year, many teachers will be fired and retained erroneously).
For example, in the aforementioned New York simulation, the teachers who would have been fired based on seniority (but were retained in the layoff based on value-added) were only about one-third as effective (relative to their fired colleagues) two years after the simulated layoff (though the difference was still significant). Even if you accept the estimates at face value, the achievement payoff declines rapidly over time because the measure is subject to high degrees of error. This suggests an obvious point, but always an important one: The idea of "quality-based" layoffs sounds great in an editorial, but, in practice, measuring "quality" is tenuous even when you predefine it (e.g., in terms of value-added). Note also that, even in the initial year, the simulations show that layoffs would have to be rather large to have any appreciable effect on overall student achievement (though some layoffs may in fact be large, and many individual students would be affected either way).
Then there are the savings, which is, lately, the big selling point of this policy. While it is true that “quality-based” layoffs would result in fewer dismissals to achieve the same level of budget cuts, one must, as Bruce Baker points out, put these "savings" in proper perspective. Even if we could terminate, say, 10-15 percent fewer teachers using value-added instead of seniority (as was the case in the Washington simulation), this doesn't save money per se (typically, you have to cut a certain amount), it only saves teachers. The benefits might show up, for example, in outcomes such as class size.
From that perspective, it is noteworthy that, as is the case with achievement, seniority-based layoffs would have to be rather large to have an appreciable overall effect on class size, relative to "quality-based" layoffs (though, once again, many individual students and classrooms would affected). Of course, every little bit helps (especially if you're one of the teachers who might be fired or one of the students who might have had a smaller class), but the idea of using “quality-based” layoffs as a means to solving budget gaps is unrealistic and misleading. I would also reiterate that, in my view, layoff policies are serious educational decisions, and any budgetary implications should be viewed as entirely secondary.
So, again – I fully support exploring alternatives to seniority-based layoffs. I'm far from certain that they are the best option, and the idea of laying off based on more nuanced, multidimensional measures is clearly preferable (to whatever degree layoffs are ever preferable).
But the case against seniority is misleading and overstated, while its alternatives remain unclear and their benefits sometimes exaggerated. The goal of this particular debate – finding the least destructive way to conduct layoffs – is one we all share, but using simplistic, misleading arguments to stir up outrage is counterproductive, and it risks pressuring states and districts into hasty, ill-considered policy decisions. Let’s also keep our eye on the ball here: Better measures of teacher quality that will yield benefits regardless of the fiscal environment.
What a thoughtful and nuanced piece. Does your institution have any recommended guidelines for what districts ought to do with, as you put it, incompetent teachers?
Does there exist a comprehensive inventory of teacher quality that you would endorse, factoring in both observations of teacher performance and objective data points?
Thanks for your comment and kind words.
We do not have a set of specific recommendations for evaluations. In general, we support a concerted effort to develop good evaluation systems that use multiple measures, with full participation from teachers. The measures that comprise these systems – and how they are configured – is the focus of a lot of work that is currently underway all over the nation. The systems might include test-based growth measures, if they are used carefully and commensurate with their limitations (see here: http://shankerblog.org/?p=1383). Other components under discussion include principal and peer observations, student portfolios, lesson plans, and others.
Unfortunately, of course, there is no “correct” system – these efforts are still in progress (and have been for a long time in many places). It will be some time before we begin to get an idea of what works, and this will vary by situation/location.
Interesting points and a good topic.
Another big question is raised by using value-added scores as a replacement model: New teachers cannot be measured by value-added scores as there's no prior data by which to measure them.
I believe the LA Times analysis also only rated teacher for whom there were at least three years worth of data in the same classroom, because less than that would have created an even wider margin of error.
So that means any teacher with less than four years at the same school could not be measured up against more senior teachers in a value-added layoff model.
Thanks for taking up this timely issue - you make many good points.
Washington State is considering a bill that would lay off teachers based on their evaluation ratings, which I think makes more sense than VAM scores, which tend to be unstable and subject to factors that may be outside of teachers' control (such as the classes they're assigned to teach). I like this approach because it sharpens the utility of performance-based layoffs: If you have a two-tier (satisfactory/unsatisfactory) evaluation system, the only people who get laid off are people who have already been rated unsatisfactory, who may also in some stage of a rather protracted (potentially stalled) dismissal process.
Assuming the evaluation is fair (which, granted, is a big assumption, but people do not have an intrinsic right to an evaluation they agree with), I think this is the most sensible way to conduct layoffs when they're necessary.
Even if it doesn't save much money, keeping satisfactory people while laying off unsatisfactory people only makes sense. I'd feel differently if it were a high-stakes VAM system, though.
A characteristically brilliant dissection of the current discussion on seniority and layoffs!
In Illinois, we are currently facing an onslaught against seniority and while the unions are proposing our own reform of it, we are encountering strong opposition. Why? Because we are proposing exactly what you point to--allowing districts to locally bargain around this highly complex issue, and to place credentials first, followed by performance together with seniority. But we insist that performance should be used after the state's new revised (improved?) evaluation system is in place. The so-called reformers want to base release on evaluations under the current eval system (which a year ago they called broken). Go figure.
Your analysis is very helpful to our work here. Thanks.
Thanks for the comment, Mike.
As I mentioned in the post, it would appear that advanced degrees are already figured into layoff decisions to some degree (at least in Washington). Put differently, teachers with advanced degrees were less likely to receive RIF notices, all else being equal. This means that having a degree was an explicit part of the “layoff equation,” and/or that administrators are choosing, when possible, to retain teachers who have degrees.
Board certification was not included in these two analyses (I’m sure this is due to data availability), and I’m not aware of any papers in which it is included (which may just mean I haven’t read them). I would imagine it is a factor in many places, but I am not certain.
I hope that answers your question, and thanks again.