Should Failing Schools Be Closed? What the Research Says

School closure is relatively common in the United States. An analysis by the Urban Institute found that about 2% of public schools, on average, were closed each year between 2003 and 2013, and these closures were found in urban, suburban, and rural communities.[1] A substantial number of public schools have been closed in Michigan, California, Ohio, and Pennsylvania. Urban school systems that have closed several schools include New York City, Washington D.C., Chicago, Detroit, and Baltimore.

In many cases, schools are closed in response to declining populations or other factors that led to a substantial decline in available resources. In other cases, school boards and elected officials struggle with the decision of what to do with persistently ineffective schools. Should they supply such schools with additional resources and attention to spur improvements? Or is it better simply to close schools where students consistently underperform and to enroll them in others? 

This paper argues that, based on the available research, closing persistently ineffective schools can be a promising strategy for improving the educational outcome of the students who attend them. When considering whether to close a school, policymakers must weigh the interests of that school’s current (and future) students, the students attending the school or schools that will receive the displaced students, and the quality of the schools that displaced students will attend. The paper describes the underlying theory and the empirical evidence related to the effect of school closure on each of these groups, in the short and long run.

It is important for educators and policymakers to continually work to improve existing schools, regardless of their current effectiveness. Even so, the evidence strongly suggests that it is better for school systems to close persistently ineffective schools than to continue providing them with an unending supply of resources.

DOWNLOAD PDF

It may seem obvious that students enrolled in a low-performing school would benefit from moving to another school. In reality, this is not necessarily the case. A substantial and growing body of empirical research has found very mixed effects for the impact of closing a school on its current students. Researchers have found that students displaced by school closures benefited in Michigan, Ohio (two studies), and Louisiana; experienced declining achievement in Chicago, Milwaukee, North Carolina, and an anonymous urban district; and experienced no distinguishing effect in Washington, D.C., Houston, Philadelphia, New York City, and Chicago (in a separate study). In the broadest study to date, researchers at Stanford University’s Center for Research on Education Outcomes (CREDO) used data from 26 states to evaluate the effect of school closures nationally and, consistent with previous research, found widely varying effects.

The most likely explanation for the variation found in the research is that students displaced by a school closure experience a combination of two factors. The effect of a school closure on students reflects two different, sometimes countervailing, factors: the difference between the quality of the new school they attend and the quality of the school closed; and the effect of moving to a new school environment.

Empirical research—as well as common sense—suggests that students make more educational gains in some schools than in others. That at least one school is far less effective than others in a system is the fundamental principle underlying the need to close a school. But that is not the whole story. For current students, the positive effect of attending a higher-quality school is counterbalanced by the disruption caused by the transition itself. Many students find it difficult to adapt to new environments—new teachers, administrators, buildings, social groups, etc.—even when the environment is objectively better than what they left. Transitions between schools can interfere with a student’s learning, at least in the short term. Recent research suggests that nonstructural mobility—when students change schools other than after completing the highest grade taught in their school—has an immediate negative impact on the outcomes of mobile students.[2]

Importantly, the literature also finds that the effect of mobility differs by the reason for the move. Changing public schools within a district (outside the context of a school’s closure) is often a reaction to problems that the students face within the school or at home and is associated with negative outcomes. Across-district moves are more likely to be strategic searches for a higher-quality school, and this sort of mobility is associated with either no effect or a positive impact on student performance. That is, negative transition effects are mitigated and can be outweighed when students move to a higher-quality school.

Thus, for school closure to have a positive effect on its current students, the new school they attend must not only be better, but better by a large enough margin to overcome the negative effect of the transition. For instance, a study of the recent extensive school closures in Michigan found that students who were displaced from the lowest-quality schools experienced the greatest benefit.[3]

The clearest illustration of the balance between transitional costs and quality benefits comes from CREDO’s nationwide evaluation of school closures.[4] The researchers noted the average effect of school closure on displaced students who the following year attended a school that was of inferior, equivalent, or superior quality compared with the closed school. Students who moved to an inferior school lost substantial ground the following year (-0.07 and -0.10 standard deviations in ELA and math, respectively); those who attended a superior school made some gains the following year (0.02 ELA, 0.06 math); and those who attended an equivalent school experienced marginal losses the following year (-0.04 ELA, -0.05 math).

If transitional costs are unimportant, we would expect that moves to equivalent schools would have no impact on student outcomes. Indeed, the size of the negative effect for displaced students who next attended a school of seemingly similar quality matches well with what the more convincing studies have found for the effect of nonstructural mobility itself on student outcomes.[5] That is, in cases where we have reason to suspect that displaced students aren’t actually attending a better school, what remains is the negative effect of the transition itself.

Thus, research makes it clear that the effect of school closure for displaced students depends on the context. Closing a persistently ineffective school can and often does lead to educational gains for displaced students. But it is not enough to simply close an ineffective school in a vacuum. When addressing current students, policymakers must also consider the quality of the schools that displaced students will enter. Simply shuffling students to another low-performing school is not just ineffective but can cause harm, at least in the short run.

The decision to close a school can also affect its current students if there is a delay between the announcement and the actual closure. For instance, New York City phased out several of its high schools over a period of several years, during which time some students chose to move to another school and other students remained enrolled. Students who remain in a school designated for closure through the school’s top grade level would not experience the negative mobility effect but would continue to be enrolled in a low-quality school. In addition, we might expect the quality of the school to further decline as its students and teachers leave and because the to-be-closed school is now essentially immune to other accountability concerns. To date, we have very little evidence on the effects of such phaseout strategies. One study of a single school closure in an anonymous district found evidence of a reduction in student test scores in the year following the announcement of the school’s eventual closure.[6] However, remaining in a phaseout high school in New York City appears to have had either no effect[7] or a positive  effect[8] on the likelihood that a student graduated from high school.

That closing a school can harm student outcomes is not a compelling argument against the practice, since the closure might have no effect on student outcomes or a positive effect. School closure is an effective policy from the perspective of current students if it is done thoughtfully to ensure that such students end up in substantially higher-quality schools. Further, the evidence highlights the need not only to close persistently ineffective schools but to replace those seats with seats in more effective environments.

Closing a school also has implications for students already enrolled in the schools that displaced students enter. First, the entrance of new students can affect the quality of instruction by changing peer groups and producing discontinuities in teaching and curriculum development.[9] Though the issue has not received sufficient attention, some studies have found evidence of mild and relatively short-lived negative spillover effects on students attending schools that received students displaced by closures.[10] At least one study found similar effects resulting from mobile students outside the context of school closures.[11]

Closing a school also has effects in cases where collective bargaining agreements ensure that the teachers in the closed school are transferred. A recent study of school closures in Michigan found evidence that the entrance of teachers who transferred from the closed schools explains part of the mild negative spillover effect on nearby traditional public schools. New York City showed that it is possible, if costly, to avoid such spillovers even when displaced teachers are protected by their collective bargaining agreement. Rather than placing them in classrooms in other schools, New York City placed teachers (who could not find another school to accept them) in an Absent Reserve Teacher pool and continued to pay their full salary and benefits (at a total cost of about $150 million per year) to serve as long-term substitutes or contribute to small administrative tasks.[12]

In sum, the negative effects of closing a school for students in other schools within the system appear to be real, but they are also short-lived and not of a large enough magnitude to justify keeping other students in a persistently ineffective school.

The research discussed so far has considered the effect that closing a school has on a district’s current students—and most discussions tend to focus on the immediate impact. But closing a school today also has important implications for tomorrow’s students, especially those who would have been enrolled in the school had it remained open. For this group of students, the effect of closing a school is unambiguously positive.

Only two studies to date have effectively addressed the issue. One, which evaluated the impact of high school closures in New York City, found that future students who would likely have attended the closed school were about 15 percentage points more likely to graduate.[13] In another study, researchers in New Orleans suggested that closing and replacing ineffective schools explains a substantial proportion of the meaningful achievement growth observed within the district since Hurricane Katrina.[14]

In areas with steady or increasing enrollment, the long-run implications of closing a school ultimately depend on whether the seats “removed” are replaced by seats in more effective schools. Closing a school will have little to no effect on later cohorts if students who would have attended a closed school simply enroll in another similarly low-performing school. Thus, an effective school closure policy must also include a plan for opening space in more effective schools.

New York City and New Orleans are prime examples of school systems that have closed ineffective schools and replaced them with better ones. New Orleans is a special case: it closed and then transitioned nearly all its traditional public schools into charter schools following Hurricane Katrina in 2005, and these charter schools tended to be far more effective than the city’s previous schools.[15] New York City—while Michael Bloomberg was mayor and Joel Klein was schools chancellor—closed schools as one part of a strategy to improve the number of quality seats within the district. Their aim was to replace closed large and persistently ineffective high schools with several smaller high schools operating within the same building. Studies have found that students performed substantially better in the smaller high schools and at a lower cost to taxpayers and that the change motivated other public schools nearby to improve as well.[16] In addition, New York City took advantage of the space opened by school closures to expand its highly effective charter school sector.

Instead of closing persistently low-performing schools, why not invest in making them better?

One prime example of this choice was New York City. When he became mayor in January 2014, Bill de Blasio said that he would “move heaven and earth” to improve rather than close schools.[17] He and Carmen Fariña, his schools chancellor, classified 94 persistently struggling schools in the city as “renewal schools” and gave them a host of resources and attention at a total taxpayer cost of $773 million. The result? Some renewal schools did improve, but many did not. Three studies found that renewal schools, on average, either did no better or made only marginal improvements.[18] In any case, four years after the program began, the de Blasio administration either merged or closed 23 of the original renewal schools.

To be sure, low-performing schools can and often do improve in response to targeted policies—in many cases, without a particularly heavy lift. For example, research suggests that receiving an F-grade on a public report card as part of a formal accountability system—essentially a costless intervention—leads to substantial improvements in student performance within the schools the following year.[19]

Yet the research findings speak to the effect of such policies on average, not for all schools. Even after adopting effective policies, there often remain ineffective schools that do not respond to the interventions or could not improve enough to justify continuing to enroll students.

The choice between closing a school and turning it around presents, in some cases, a false dichotomy: successful “school turnarounds” often require such fundamental changes as to be little different from closing them and starting over. Thus, the more successful school turnarounds identified in the recent literature include removing and replacing a large portion of their teachers.[20] Recent studies have found positive effects in cases where persistently low-performing traditional public schools are transitioned to effective charter school operators[21] or when the practices of No Excuses charter schools—extended learning time, data-driven instruction, frequent teacher feedback, high expectations, etc.—are injected into struggling traditional public schools.[22] The name of the school could remain the same, but when a turnaround constitutes such major changes in a school’s personnel and educational environment, it is better understood as a closure and replacement.

An honest discussion should not focus on whether it is ever appropriate to close a school. Instead, the focus would best be on the conditions under which policymakers should close a persistently ineffective school. No clear algorithm is available, but the evidence does point to some key principles. Decisions about closing a school should take into account the effect it is likely to have on students currently enrolled in the closed school, current students in the schools where displaced students will next enroll, and future student cohorts who will enroll in the school if it remains open. The decision should include an assessment of the difference in the effectiveness of the school to be closed and the effectiveness of the quality of the school that displaced students will attend, and the potential effectiveness of a new school that will open in its place. Finally, the decision should depend on the likelihood that the school can be improved enough to sufficiently close the gap between it and the replacement school at a palatable cost to the taxpayer.

Ultimately, a successful school closure policy is one that replaces persistently ineffective schools with more effective alternatives. Expanding school choice is an appealing way to do so within the framework of a portfolio model of educational improvement.

Though the details can differ, a portfolio strategy provides schools with substantial autonomy to meet educational objectives and holds them accountable for their performance with a combination of formal accountability and parental choice. To at least some degree, this approach brings some essential components of the charter school sector into the broader public school system.

How one views the role of charter schools in the context of school closures depends on one’s frame of reference. Charters draw their enrollment from local traditional public schools. Some view charters as driving part of the enrollment declines within the local school system, and thus one of the causes of public school closures. For those who take a portfolio approach, charters are part of a locality’s public school offerings, and an expansion of the charter sector simply reflects the movement of students within the public system.

Though it is not essential, a universal enrollment system that relies on a deferred acceptance algorithm is a common and potentially important feature of the portfolio approach to educational improvement. Each spring, parents who wish to enroll their child in a different school send to the central district an ordered list of the schools that they would like their child to attend. The district then uses an algorithm that assigns the student to the preferred school and allocates seats randomly in cases where more students prefer to attend a school than there are available seats. The number of districts using this strategy to assign students to at least some schools has been growing and now includes districts such as New York City, Boston, Newark, Camden, Denver, and New Orleans.

One benefit of a choice-based portfolio model is that universal enrollment provides valuable information that can help policymakers identify schools to target for closure. Test scores and other measures are helpful—but limited—tools for identifying ineffective schools. The deferred acceptance choice system provides policymakers with direct information about how parents view the quality of their local schools. If very few parents list a school as a preference, policymakers can use that information as a marker. Since their child’s school enrollment depends on their answer, parental preferences revealed from universal enrollment provide much more actionable information about a school’s quality than does the public demonstrations that often follow notice that a school is designated to be closed.

Choice within a portfolio model also addresses the underlying issue of placing students into seats in higher-performing schools. First, this model allows for the creation of new schooling options that are held accountable by formal accountability systems and parental choice. Second, this system allocates students to the schools that they most prefer in a fair and transparent way that does not depend on administrative decisions. To further counterbalance the negative transitional costs associated with a school closure, policymakers working within a portfolio model might consider granting an assignment preference for displaced students to enroll in their desired school.

No school board or elected official starts out by wanting to close a traditional public school. But no student deserves to attend a persistently ineffective school when there are other options. The research summarized here suggests that policymakers can use school closures to improve educational outcomes for current and future students. But they must keep in mind that closing a school does impose costs on at least some current students. A strategy to close ineffective schools needs to be coupled with a strategy to replace the seats lost with seats in higher-quality educational environments. When thoughtfully administered, that approach will lead to substantially improved educational opportunities for a community’s students.

See endnotes in PDF