Racial Differences in Student Access to High-Quality Teachers

A troubling finding of past research is that various groups of students differ systematically in their access to high-quality teachers. The forces that lead to these differences lie in the workings of the teacher labor market and the segregation that characterizes schools and neighborhoods in this country. The differences matter because teachers are central to the educational outcomes of students. This policy brief reviews relevant studies, presents new evidence of disparities by race in access to high-quality teachers, and discusses policy options to address this inequity.

Three strands of academic research are relevant to the study of racial disparities in access to high-quality teachers. In the first, numerous studies have documented the disparities themselves and their educational consequences. Whether teacher quality is identified by teacher characteristics—such as years of experience or licensure test scores—or by a teacher's calculated value-added to student achievement, most studies show that white and affluent students end up with the higher-quality teachers at various grade levels (Lankford et al. 2002; Clotfelter et al. 2005; Glazerman and Max 2011; Sass et al. 2012; Goldhaber et al. 2018; Hanselman 2019; Cardichon et al. 2020; Bastian 2021; Mehrotra et al. 2021; Isenberg et al. 2022). An exception is Mansfield (2015), who concludes that high school teachers in North Carolina were fairly evenly distributed across and within schools based on a longitudinal dataset ending in 2007. Moreover, the literature shows that disparities in teacher quality can be consequential for student outcomes. For example, experienced teachers are consistently shown to be more effective in raising the achievement of their students than are novice teachers (see, e.g., Clotfelter et al. 2006; Henry et al. 2012; Ladd and Sorenson 2017).

A second strand of literature documents the racial segregation of schools. Segregation between schools is relevant because racial disparities in access to high-quality teachers between schools and districts simply could not exist if schools were not segregated. The extent of such segregation has been extensively examined (e.g., Coleman et al. 1966; Orfield and Gordon 2001; Clotfelter et al. 2008). Segregation inside schools—and its companion practice of academic tracking—makes possible disparities within schools (see Useem 1992; Lucas and Berends 2007; Kalogrides and Loeb 2013; Kalogrides, Loeb and Be'teille 2013; Grissom and Redding 2016; Clotfelter et al. 2021).

The third strand of literature relevant to racial disparities in access to quality teachers concerns the labor market for teachers and associated patterns of mobility and attrition. An essential element is an understanding of the factors that influence teachers’ decisions to change schools, stay where they are, or leave teaching altogether. These factors include salaries (like other workers, teachers typically prefer positions with higher pay), location (teachers are drawn to their hometowns and states), and characteristics of schools (teachers tend to avoid schools with certain demographic makeups). Jackson (2009, p. 214) succinctly summarizes this last factor: “if teachers prefer working environments with students of a particular demographic . . . , students whom teachers find undesirable will be exposed to teachers of lower quality.” Research on teacher transfers has long documented systematic patterns correlated with the economic and racial makeup of the schools teachers leave and the ones they go to (Becker 1952; Greenberg and McCall 1974). More recent studies of teacher mobility and labor markets uncover similar patterns (Hanushek and Rivkin 2007; Jackson 2009; Goldhaber et al. 2010; Boyd et al. 2011; Clotfelter et al. 2011; Reininger 2012; Engel et al. 2014). Such teacher quality disparities result from the combined effects of attrition, mobility between positions, and hiring for open positions (Goldhaber et al. 2022). One question of particular relevance for the subject at hand is the potential for higher salaries to offset teachers’ aversion to schools serving large proportions of disadvantaged students. Yet, as shown by Clotfelter et al. (2011), teachers with strong credentials are less responsive to salary differences than teachers with weaker qualifications.

Taken together, these three strands of research reveal that disparities in access to quality teachers are baked into conditions and practices that are common in public education, segregation creates racial and economic differences both between and within schools, and the fluid teacher labor market allows teachers to transfer from one school to another, usually with no change in salary. This freedom of movement, combined with prevailing preferences of teachers, as well as decisions about the allocation of teachers within schools, paves the way for the disparities in access we observe.

We performed new empirical work with the aim of determining whether student access to high-quality teachers differs by the race or ethnicity of the students. To do so, we assembled detailed administrative data for the year 2019 for the state of North Carolina that link student and teacher characteristics, categorizing these assignments by grade and subject. To represent schools and subjects at the elementary, middle, and high school levels, we examine five sets of grade–subjects in the state's traditional public schools: fourth grade, seventh-grade English, seventh-grade math, tenth-grade English, and tenth-grade math.¹ We used counties as the relevant local geographic unit because most school districts in the state are countywide and the state's one hundred counties are quite large and best represent the relevant geographic area of school choice and the local teacher labor market.

For each teacher, we use information on years of experience, certification area, National Board Certification, and teacher licensure test scores.² To represent teacher quality, we defined six dichotomous measures, three indicating strong teacher credentials and three indicating weak credentials. The indicators of strong credentials were: experienced (five or more years of teaching experience), high licensure test score (a score on the Praxis 1 teacher test in the top quartile of all teachers in the state), and National Board Certification.³ The indicators of weak credentials were: inexperienced (two or fewer years teaching experience), low teacher test score (a score on the Praxis 1 test in the bottom quartile of all teachers in the state), and not certified in the area of teaching. For each credential, we measured the probability that a white, black, or Hispanic student was assigned to a teacher with each of these credentials. Then, for each of our six indicators of teacher quality, we compare these probabilities across student racial or ethnic groups.

Taken together, the scatterplots in figure 2 reveal two noteworthy findings. First, differences by race and ethnicity in exposure to experienced teachers are common across the state, with most county groups revealing a higher exposure rate for white students. Second is that, on average, these differences are larger in counties where schools are more racially segregated.

Although figure 2 is based on only one of our measures of teacher credentials, the use of other measures yielded similar findings. That is, we found similar patterns in scatterplots based on our two other measures of strong teacher credentials and opposite patterns based on our three measures of weak credentials. We conclude that white students across the state are more likely to have teachers with strong characteristics and less likely to have teachers with weak characteristics, especially in larger, more segregated counties.

To shed light on which public policies might be effective in reducing the systematic and pervasive racial disparities documented here, it would be useful to know the extent to which the statewide disparities at each grade–subject level are attributable to disparities across counties, across schools within counties, or across classrooms within schools. As noted above, since most school districts are countywide in North Carolina, the most logical local unit is the county. For other states, the comparable local units might instead be local school districts, many of which may be quite small. In any case, the greater is the contribution of gaps across local educational units to gaps in overall teacher quality, the stronger is the argument for state-level policies to reduce them. In contrast, significant contributions to the statewide gaps of differences across schools within counties or across classrooms within schools would suggest the need for policy interventions at the local or school level.

The three segments of each bar represent the portions of each overall racial gap that are attributable to the three mechanisms. For example, for seventh-grade math, 0.015 (1.5 percentage points) of the white–black gap can be attributed to differences across classrooms within schools, 0.044 to differences between schools within counties, and 0.049 to differences between counties. The (somewhat smaller) gap in exposure between white and Hispanic students primarily reflects disparities between schools, and secondarily reflects disparities across counties.

Based on the decompositions shown in figure 3, we draw the following conclusions. First, except for seventh- and tenth-grade math, differences across classrooms within schools contribute relatively little to the overall magnitude of teacher quality differentials. As a consequence, while efforts by school principals to redistribute their teachers and students across classrooms within schools might well be beneficial in some cases, a stronger case can be made for reducing teacher quality gaps with policy actions at the state or countywide level.

Second, the contributions of the other components differ by grade and subject. The figure shows that the largest contributor (about two thirds) to disparities in grade 4 is differences between schools within counties. In grades 7 and 10, the racial gaps are mainly attributable to a combination of disparities between schools within counties and disparities between counties. This pattern reflects a combination of factors, such as differences in the racial compositions of individual counties, racial and economic segregation across schools within counties, and other differences, such as teacher salaries and local living and working conditions, which affect the ability of counties or schools to attract and retain high-quality teachers. Much of the responsibility for reducing disparities of these types would lie with policy makers at both the state and county levels. In addition, the disparities across classrooms within schools that emerge most clearly for seventh- and tenth-grade math would require actions by school-level administrators.⁶

We note that the state government plays a larger role in education finance and policy making in North Carolina than in many other states. In particular, the state government provides 63 percent of total (including federal) K–12 revenue and sets and pays for salaries based on a statewide teacher salary schedule, albeit one that individual counties can supplement out of local tax revenues. Although state-level policy makers in North Carolina may have the authority and power to play a somewhat larger policy role in offsetting racial disparities than their counterparts in many other states, all state governments have significant educational responsibilities. At the same time, education policy makers at local and school levels in every state also have significant roles to play.

Central to the state role is its responsibility for ensuring an adequate total supply of high-quality teachers. To that end, states must invest continually in teacher preparation programs and provide salaries sufficient to make teaching a rewarding and desirable profession. Given that teachers often avoid schools with certain demographic makeups, a shortfall in the overall supply of qualified teachers is likely to be particularly detrimental to students in these schools.

Shortfalls of teachers in certain content areas also contribute to racial gaps in access to high-quality teachers. To address that issue, state policy makers can provide fellowships for potential teachers who commit to teaching in content areas, such as math, science, and special education, which are often hard to fill. The relatively small North Carolina Teaching Fellows program, which operated successfully from 1986 to 2011, has shown the potential of a statewide program of this type to recruit highly qualified and effective teachers who remain in the profession longer than other teachers. Because the program placed no limitations on where they taught, however, the Fellows ended up clustered in metropolitan areas rather than in schools and classrooms with high-poverty and lower-performing students (Henry et al. 2012). Hence, some limitations on their placement, perhaps with additional financial incentives for Fellows to teach in underserved schools, would have been useful. That program was reinstated in 2018–19, and the state is now considering greatly increasing the number of Fellows and expanding the number of eligible teacher training programs to cover all regions of the state and to include minority-serving universities to help increase the diversity of the work force (WestEd et al. 2019, p. 62).

In addition to pursuing policies designed to minimize statewide shortages of teachers, state policy makers must also pursue policies designed to offset the greater disadvantages that some counties face relative to other counties in attracting high-quality teachers. For example, using North Carolina again as an example, under that state's teacher salary policy, poor or low-wealth counties, and especially those in rural areas, are at a disadvantage relative to wealthier urban counties who have the means and the desire to supplement the statewide salary schedule. Except to the extent that local salary supplements simply compensate teachers for differences in the cost of living, they add to the host of attractions offered by urban areas, giving wealthier counties a decided edge in attracting and retaining good teachers. In addition, potential teachers, especially those who grew up in more vibrant urban areas, may prefer to avoid counties lacking sufficient suitable housing, far removed from urban centers, or serving primarily disadvantaged students.

State funding for higher salaries in such areas, along with programs to provide housing and other supports for teachers, represent sensible policy options. In addition, fellowships similar to those just described for certain content areas could be targeted specifically toward prospective teachers who commit to teaching in geographically underserved areas. “Grow-your-own” strategies for teachers are also worth considering. Of note, many states are now trying innovative approaches to help train local community members—including career changers, paraprofessionals, or staff currently working in schools—to become teachers. The goal is to help local communities (often in partnership with local community colleges) to produce teachers who understand their local communities and are willing to remain there (Espinoza et al. 2018; see also WestEd et al. 2019, p. 65.).

Policy makers at the county level have the primary, but not necessarily the exclusive, responsibility for reducing racial teacher quality gaps arising from differences across schools within local counties. Given the well-documented tendency for teacher mobility between schools to be correlated with the socioeconomic and racial composition of students, the racial mix of students in a school can affect where teachers end up and who they teach.

Therefore, the most obvious way to reduce racial differences in access to high-quality teachers across individual local schools, but one that has often proven difficult to implement, would be for policy makers to reduce the extent to which schools are racially segregated. If all schools within a county had the same racial mix of students, there could be no school-specific differences in student access to high-quality teachers. Reducing school segregation across elementary schools may be particularly difficult because of the positive value many parents place on having elementary schools that serve local neighborhoods, even if those local neighborhoods are racially segregated. Reducing school segregation at any level would typically require the county to redraw school attendance zones or to pursue policies to reduce residential segregation. Making that remedy all the more difficult is the possibility that some local residents may respond to such changes in ways that could undermine desegregation efforts. These responses may include moving out of the county, transferring their children in other schools within the county, enrolling them in charter schools (Mickelson et al. 2018), or expressing their opposition by voting out the current school committee. On a more positive note, a recent event study of a controlled choice program in Wake County, North Carolina, that reassigned about 25 percent of students suggested that carefully designed school assignment policies can improve school diversity (Domina et al. 2021).

An alternative or complementary approach would be to take steps to reduce the clustering of high-quality teachers in some schools and low-quality teachers in other schools, a pattern that, as previous studies of teacher mobility make clear, typically works to the disadvantage of schools serving large shares of students of color. One potential step would be for state or district policy makers to eliminate the fairly standard practice of giving preference to experienced teachers who wish to move from one school to another when filling teacher openings within the county. State or county policy makers might also try boosting salaries for experienced teachers willing to teach in schools serving large proportions of disadvantaged or minority students. As documented by Clotfelter et al. (2011), however, the pay differential required to induce teachers with strong qualifications to move to or remain in such schools is likely to be quite high. The reason is that, compared with relatively less-qualified teachers, teachers with strong qualifications tend to be less willing to teach in such schools and to be less responsive to salary differentials.

At the same time, changing the quality of new recruits into schools serving large proportions of disadvantaged students may not suffice. Also needed are strategies to increase the productivity of all teachers and to promote the retention of the effective teachers already in such schools (Sass et al. 2012; Xu et al. 2015). Hence, a final strategy would be for state or local policy makers to increase mentoring support for inexperienced or otherwise struggling teachers in high-minority schools. When done well, such mentoring can improve the quality of the mentored teachers, induce some inexperienced teachers to remain in the school and thereby to gain experience, and reduce the rates of teacher turnover. That, in turn, would reduce the need to hire new teachers, many of whom are likely to be inexperienced. Further, given the evidence that the quality of the school principal affects teachers’ perceptions of their working conditions and their willingness to remain in a school (Ladd 2011) and that principal turnover increases teacher turnover (Bartenen, Grissom, and Rogers 2019), ensuring that schools serving disadvantaged students have high-quality and stable leadership would help to retain teachers in such schools.

Some policy analysts have suggested that policy makers should consider exploring changes in the institution of teacher tenure as a means of reducing racial disparities in access to quality teachers across schools (Miller and Chait 2008). The logic here would be that the difficulty of removing ineffective teachers under current due process requirements makes it possible for ineffective teachers, many of whom are concentrated in schools with demographic characteristics often avoided by teachers, to remain in such schools. Under a different system for paying and retaining teachers, such as one based more explicitly on merit, such teachers would be easier to fire. We do not recommend such a strategy, however. Not only would it most likely end up as essentially a merit pay system for teachers, which has generated mixed results in the past, it would remove an important, if imperfect, guard against arbitrary treatment of teachers by principals or other administrators that would have to be replaced by other safeguards.

Racial gaps in access to high-quality teachers within schools can occur only when students are distributed unevenly by race or ethnicity across a school's classrooms. Numerous studies have clearly shown the connection between academic tracking and other practices of grouping students into classrooms and various forms of within-school segregation (see, e.g., Mickelson 2001; Lucas and Berends 2007; Kalogrides and Loeb 2013; Clotfelter et al. 2021). Further, the impact of tracking can be exacerbated because of scheduling challenges that arise when students who are tracked for academic reasons into advanced classes for one subject are grouped with similar students in many of their other classes.

One way to reduce tracking in subjects such as math at the high school level would be to make less use of tracking in earlier grades. If all, not just the top, students were made eligible for advanced math in middle school, for example, there would be less justification for tracking of precalculus and subsequent math classes in high school. Successful mixing of students in middle school math courses, however, would necessitate appropriate attention to support both for students and for teachers and, for that reason, might be difficult. Francis and Darity (2021) propose an alternative approach designed to reduce the effects of tracking on racial disparities within high schools. Given their empirical finding that a higher proportion of black students in upper-grade advanced high school math courses increases the probability that black students in the lower high school grades will opt for advanced math courses before they graduate, the authors suggest that a short-term concerted effort to reduce institutional biases among, for example, school counselors, that create barriers for entry into advanced courses among racialized minorities, could have long-term beneficial effects (Francis and Darity 2021). Schools could also reduce the adverse effects of tracking by limiting the number of classes that are subject to academic tracking and by striving to ensure that students who are in one tracked class are assigned to more diverse groups for other classes.

Because racialized minority students, particularly those from less advantaged backgrounds, often have lower levels of prior achievement than their more advantaged peers, an explicit policy of assigning high-quality teachers to classrooms with low prior achievement levels could reduce undesirable within-school racial disparities in exposure to high-quality teachers. One disadvantage of this approach is that it may induce some teachers to avoid the schools serving higher proportions of low-achieving students. A modification within the context of academically tracked classes would be to ensure that all teachers within a school are systematically assigned to teach sections with students at every achievement level each year.

Our finding of pervasive and substantial racial gaps in students’ exposure to teachers with strong and weak credentials indicates that white students in North Carolina are more likely to be assigned to high-quality teachers than are black or Hispanic students. Notably, we find gaps in fourth grade and in both math and English courses in seventh and tenth grades. Further, we find gaps in most counties. Although our empirical analysis focuses on North Carolina, the policy implications are more general and imply that policy makers at all levels—state, district, and school—will have to confront patterns of differential access by racial group to high-quality teachers head-on and develop a variety of policy mechanisms to counter them.

We are grateful to Dan Goldhaber for helpful comments and to CALDER and Duke University for financial support. The views expressed here are those of the authors and do not necessarily represent those of any institution.