This past fall, school districts nationwide faced serious teacher shortages that left many schools scrambling to find qualified teachers. Today, halfway through the academic year, many students are being taught by a temporary teacher because their schools could not fill positions in time — in Arizona, for example, more than 1 in 5 teaching positions remained unfilled four months into the school year, and an estimated 20 percent to 30 percent of teachers in urban school systems are hired after the school year starts. Projections suggest that the national teacher shortage is only going to get worse, particularly in hard-to-staff subjects such as mathematics, science and special education.

In response, policymakers have taken steps to boost the supply of teachers. In December, Virginia Gov. Terry McAuliffe (D) passed emergency regulations designed to alleviate what he called the “growing crisis” of a statewide teacher shortage by streamlining education requirements for new teachers. Lawmakers in Arizona, Illinois and Minnesota recently took steps to increase the number of new teachers by lowering the teacher licensure requirements. States such as Oklahoma have staffed classrooms by providing record numbers of temporary emergency certifications. And, motivated in part by a call to ameliorate teacher shortages, New York state recently allowed charter schools to certify their own teachers and dropped literacy tests for teacher candidates.

Although these efforts may prove to be helpful, they fail to address one fundamental root of the problem: School systems need to hire teachers in great numbers only if they don’t retain enough of the well-qualified teachers they currently employ. Unfortunately, 15 years after Richard Ingersoll cautioned about the “revolving door” in the teaching profession, the challenge of teacher retention remains. This revolving door is not only expensive for schools and destabilizing for students, but it also contributes to inequality in educational experiences — students of color and those living in poverty are less likely to be assigned effective teachers.

Recognizing that better information is needed to understand and address this long-standing challenge, we conducted a large-scale study of teacher retention in a diverse set of 16 urban public school districts in seven states that together serve nearly 2.5 million students annually.

We found that on average, just over half of new teachers in the districts we examined remain in the classroom after five years. This finding largely mirrors prior research. What our work newly reveals, however, is substantial variation around this average: While turnover is a challenge in all of the districts we study, it’s a real crisis in some. Our study documented five important trends about teacher retention.

First, across the districts, the share of novice teachers who left their district within five years ranges from just less than half to nearly 75 percent. This is an enormous difference in retention rates. The annual hiring costs in the district with the lowest teacher retention rate would be about $4 million lower if it retained novice teachers at the highest rate we observe. In an era of tight school budgets, these dollars can and should be better spent elsewhere.

Second, even when teachers stay in the same district, they frequently move across schools. In one district, half of novice teachers stayed in the district, but only 1 in 5 remained in the same school for five years. This building-level turnover means that schools still must invest resources to find and train new candidates. And there is good evidence that turnover can hurt students because it causes organizational instability.

Third, after teachers leave the classroom, their likelihood of returning varies widely by district. In half of the districts we examined, it is common for teachers to return after a temporary leave of absence, such as parental leave. In the other half, few teachers returned after going on leave. This suggests that struggling districts may benefit from human resource policies that encourage teachers to return after a leave.

Fourth, we found that few teachers depart the urban districts we studied for other districts in the same state. Thus, these urban districts can’t necessarily point to their suburban counterparts as the drivers of their retention challenges.

Fifth, encouragingly, we found relatively higher retention among more effective teachers. Here again, however, we found considerable variation across districts. These differences imply an additional cost — lower student achievement — in districts struggling to retain their top performers.

Our research revealed no obvious, simple way to improve teacher retention. The differences in retention rates that we saw across districts are not explained by easy-to-observe factors such as student demographics or teacher salaries. But related research shows that teachers leave schools with poor working conditions where they feel they cannot have success with their students, and they stay in schools where they feel supported by their colleagues, their principals and their school culture. Working to build more supportive school environments can both help students and ameliorate the retention crisis plaguing some of our urban school systems.

With teacher shortages on the rise across the country, policymakers must expand their focus beyond policies that only increase the supply of new teachers. While such efforts might act as Band-Aids to solve immediate shortages, they alone will not address the roots of the challenge. Our research shows that some districts are facing turnover rates that are unsustainably high. With teacher shortages on the rise and as states and districts make strides in promoting equal access to high-quality teachers for all students, a focus on teacher retention and attention to what conditions encourage teachers to stay at a particular school must be part of the solution.