Teacher Supply and Demand: Getting the Numbers—and the Solutions—Right
On September 15, the Learning Policy Institute (LPI) released a report on teacher supply and demand that examines the data behind shortages that are emerging in a number of areas around the country. Our goal was simple: to clarify the nature of emerging shortages and their impact, and to offer evidence-based strategies to guide the responses of educators and policymakers.
We began by examining the real-world indicators of current shortages. These include the more than 40 states reporting severe shortages in subject areas like mathematics, science, and special education, and the hiring of substitutes and individuals without credentials by the thousands in states ranging from Arizona, California, Oklahoma, and Utah to Florida, Indiana, North Carolina, and South Carolina. We also noted surveys of districts by the American Association for Employment in Education (AAEE), in which 56% of respondents reported that finding enough candidates for open positions in 2014–15 was a “big challenge.” These rates were nearly double those from the previous year.
We set out to understand why districts were reporting these difficulties when teachers were being laid off only a few years earlier. We titled the report “A Coming Crisis in Teaching?” with a question mark signaling the fluidity of the problem—that what now feels like a crisis in many states may be resolved by policies states and districts enact to boost supply and reduce attrition.
We also noted that we expect the labor market to respond to these current indicators of shortages, with more new people entering in response to reports of available jobs. Thus, projections only reflect what might happen if conditions do not change, even as we know they will.
We define shortages as an inadequate number of qualified individuals willing to offer their services for available jobs under prevailing wages and conditions. From this perspective, the key issue is not whether there will be enough warm bodies to enter teaching. The key issue is whether there will be enough well-qualified individuals willing to offer their services in the specific fields and locations that currently lack an adequate supply—and whether this will happen on its own, in response to the market, or will require policy interventions.
The question of potential shortages is clearly a hot button issue. Commentaries have been posted on The 74 and the Bellwether Partners websites questioning the size of current or potential shortages, arguing that significant shortages are unlikely to emerge, and questioning particular data points in our analysis. We address these here.
1. State or national labor markets: We agree with the points made by Dan Goldhaber and Roddy Theobald in an opinion piece in The 74 that teacher shortages are, for the most part, not national in nature, and that there are, in fact, 50 separate labor markets for teaching. We emphasize in our report that the current shortages differ from state to state, across districts of different types, and among subject areas. The appendix of our report and our interactive online map demonstrate that the conditions influencing supply and demand—factors like compensation, working conditions, and turnover rates—are very different across states, as is the equitable distribution of qualified teachers. At the same time, 25% of teachers cross state lines in their careers, and many leave the profession because of barriers to transferring their licenses and pensions, so solving these national labor market problems would be helpful.
2. Assumptions about demand: We also agree that various assumptions can change shortage projections, and we carefully explain the data underlying all of the assumptions in our model in an appendix to our report. Goldhaber and Theobald argue that demand projections are highly sensitive to shifts in demand-side policies. They note that if, for instance, the pupil-teacher ratio remained at 16-to-1 instead of decreasing to 15.5-to-1, as we modeled based on National Center for Education Statistics (NCES) estimates, it would reduce the needed teachers by 100,000, which they argue would essentially wipe out the entire projected shortage in 2021.
It is true that demand estimates are sensitive, as they should be! Pupil-teacher ratio changes can have noticeable impacts on demand, so it does not make sense to ignore those impacts, but to model them in. We incorporated the estimates from NCES about trends in pupil-teacher ratios, which would return districts to levels closer to those that existed before the recession. But if districts did not try to restore positions or programs lost in the recession, the effects would be less stark than Goldhaber and Theobald suggest. We model these changes in pupil-teacher ratios to occur slowly over about eight years—not landing in a single year—thus the much smaller annual share of these 100,000 teachers would modestly offset the predicted shortage in 2021.
In fact, even in the years where the projections show the greatest increased demand due to pupil-teacher ratio reduction and student enrollment growth, more than 85% of demand is still caused by attrition. If we assume—in contrast to NCES projections—that pupil-teacher ratios remain at recent highs of 16-to-1, and student enrollment remains flat, we still would need significantly more teachers than we predict will be available if other conditions do not change.
3. Assumptions about supply: Goldhaber and Theobald also point out that the potential supply of teachers depends both on how many newly prepared teachers are available and the labor market conditions that attract them and others into teaching—and that future labor market conditions can differ from past conditions. We agree with these points and describe how we combined data and evidence from different labor market contexts to create ranges of reasonable estimates for new entrants and re-entrants. (See the report’s Appendix A for details.)
Picking up on this point about the supply of newly hired teachers, Chad Aldeman suggests in a Bellwether blog that our estimates predicting the size of potential shortages that might occur without policy interventions underestimate the supply of potential teachers. Aldeman inaccurately states that we count only new graduates recently entering from teacher education programs as our source of supply. In a doctored version of one of our graphs, Aldeman mislabels our supply line as “estimated supply (of new graduates),” having added the words in parentheses, which were not in our figure.
In fact, our estimates of supply include those who are entering from preparation programs immediately (“recent graduates”), those who enter after a pause of up to 4 years (“delayed entrants”), and those who enter from the reserve pool of teachers who have taught previously but are not currently teaching. In figure 9 we show the upper and lower bound of potential teacher supply sources associated with a range of entry and re-entry rates based on several kinds of data described in our methods.
Aldeman offers data from NCES illustrating that recent graduates comprised about 32% of the new hires in 2004 (74,500 out of 236,407 newly hired) and about 37% of the new hires in 2008 (92,500 out of 247,964 newly hired). His point is that recent graduates are, in some years, a relatively small share of all hires, despite the numbers suggesting there are many more recent graduates. He then suggests that our figures are “way out of proportion.”
In a conversation in which one of us sought to discover what Aldeman’s confusion was about our methods, we learned that he “eyeballed” our graphs and thought the total demand number in one graph was close to the number of teacher preparation completers shown in another, and assumed that we had used the number of recent graduates as the number of new teachers demanded. Since we did not do that, the question arose in our conversation: How does the number of candidates leaving teacher education—the potential supply—dwindle to the levels of new graduates our estimates identify as actually likely to enter and NCES data confirm is the case?
The answer is that we based projected entry rate estimates on studies of actual entry and re-entry rates, so that we could make estimates of likely supply, which turn out to be quite accurate.
Here is a quick example: In 2009, federal Higher Education Act (HEA) data show 233,000 completers of teacher education programs. As we note in our appendix, not all completers become credentialed. Since most states have several licensing tests needed to enter teaching, each of which typically has a fail rate of at least 10%, some of these candidates will not receive credentials. Most of the tests are taken before graduation and some are taken after. If we generously assume that 90% of those who complete a program will become credentialed, the total is now about 210,000. During this era, studies find that about 75% of newly credentialed teachers entered the profession within 4 years of graduating from college (about 157,500). If we estimate that 2/3 of this group of teachers enter immediately (105,000), and the remainder enter as “delayed entrants” (52,500) over the subsequent several years —a category that NCES counts separately but Aldeman ignored—the entry numbers for new graduates (105,000) are close to those Aldeman cites from NCES for 2008 (92,500). We did not include in our model the share of newly graduated teachers who go into private schools, which employ about 13% of all teachers. Adding that factor would bring these very close statistics even closer together.
What about all those other teachers who were part of the potential supply? Well, some decided to start families or go into different fields based on personal preferences or labor market conditions. Recently, those conditions precluded the hiring of many new teachers; looking ahead, many more are likely to be hired. This possibility is taken into account in our estimates, as our projections include new graduate supply ranges (including delayed entrants) from 75% to 90%, based on data from different labor market eras. Some prepared in fields where they were not needed (e.g. elementary education or social studies, which often have surpluses when others have shortages), or preferred only to teach in well-heeled districts with fewer vacancies. This signals the importance of understanding labor markets more fully and creating targeted incentives to shift the decisions of entrants about how they will prepare and where they will teach.
Finally, on the accuracy of our projections, we can also compare our statistics with those produced by NCES: Using the same source of NCES data Aldeman quoted and updating it to 2012, we find that, between 2004 and 2012, the proportion of new hires in a given year who were recent graduates entering teaching averaged about 32.5%—very close to the numbers Aldeman cites—while the proportion of delayed entrants (ignored by Aldeman) was about 25.4%. Together these sources accounted for roughly 58% of new hires, and the remaining 42% were re-entrants. These numbers are close to the middle of the ranges we used in our estimates—in other words, just about right on target.
4. Are there really shortages (and have there been in the past)? A theme in these blogs and another posted by Michael Antonucci on The 74 is that the shortages may not be real, or that the degree of projected shortages may not materialize. We strongly disagree on the first point: Data from states and districts, from surveys and credentialing offices, makes it clear that there are currently not enough qualified teachers offering their services in the fields and locations where they are needed in all parts of the country. Our projections estimated shortages in the 2015–16 school year of about 60,000, which is in a range very close to the number of hard-to-fill vacancies reported from states in the past year. (California alone issued nearly 8,000 permits to teachers who were not certified for their jobs and hired large numbers of substitutes in still other classrooms.)
On the second point—that projections of future shortages may change—we agree. We noted in our report that current shortages—which are clearly real—could grow if current trends continue, while we also noted that current trends could and should be disrupted by both labor market responses and policy interventions.
Critically important in the definition of shortages is the issue of quality. We can always find ways to fill vacancies in the teaching force. But that does not mean shortages are solved. In the early 1980s, for example, when salaries had been dropping for a decade and districts could not fill their vacancies, the average quality of teachers reached a nadir in virtually every respect. In addition to the fact that untrained teachers were hired and teachers were assigned out of field, college admissions standards were lowered for teacher education programs: Teachers, on average, had poorer academic records and test scores than other college graduates, and many teacher education programs lowered their expectations to accommodate their clientele.
Antonucci suggests that the fact that a growing number of teaching slots were filled during the 1980s means there was no shortage. But warm bodies are not the same as well-qualified entrants. That “crisis in teaching,” which he now treats as a mirage, did indeed occur, and shortages were largely addressed in the beginning of the decade by lowering the quality bar for teaching to one of the lowest levels it has ever reached.
Similarly, California demonstrated when it reduced class sizes during the late 1990s that a state workforce can grow without actually meeting the demand for qualified teachers. As the workforce expanded, more than 40,000 uncertified teachers were hired and placed largely in high-minority and high-poverty schools, filling vacancies but not resolving shortages.
Returning to the 1980s “crisis,” an equally important part of the story is that the labor market was ultimately rebalanced through policy—in some states more productively than others. Policymakers raised average salaries for teachers by nearly 100% between 1980 and 1990. In addition, standards for teacher education programs were strengthened, and states like Connecticut and North Carolina, which undertook large-scale, comprehensive reforms, turned shortages into surpluses of much better trained, better supported, and better (and more equitably) paid teachers, while substantially raising achievement and reducing achievement gaps. (See p. 56 in our report.)
The question before us now is how we will address the imbalances clearly plaguing many schools and districts—especially those serving the neediest students. To answer this question, we need to track available data while encouraging policy moves that make a difference.
5. What should be done? The most important question is what to make of these data. We agree that the goal should not be just to pump out more new teacher education graduates. As we argue in our report, imbalances in the teacher labor market vary by region, subject area, and student population. Incentives should be applied strategically to recruit teachers to the fields and locations where they are needed. Goldhaber and Theobald suggest some useful starting points. We review a number of others from research on successful recruitment and retention strategies.
Furthermore, boosting supply is not necessarily the only, or even the most important, strategy for solving shortages where they exist. Attrition is a key aspect of the problem. The 8% annual attrition rate for teachers in the United States—which is about twice the rate found in high-achieving countries like Finland, Singapore, and nearby jurisdictions like Ontario, Canada—is responsible for more than 90% of the current annual demand for teachers and a projected 85% or more in the years to come. The lion’s share of this attrition (more than two-thirds) occurs before retirement. Simply reducing the number of teachers who leave each year—by strengthening preparation and induction, and by ensuring more supportive and collegial working conditions—would help resolve many of the shortage problems.
We are heartened to have begun a robust conversation on strengthening the teacher workforce with many who have deeply considered our analysis, and we are optimistic that this conversation can inspire policies that improve educational opportunities for students around the country.