A large coalition of teachers, educational administrators, parents, policymakers, and advocates for racial and social equity have argued for a more racially diverse teaching force. Increasing the number of black, and especially black male, teachers in US public primary and secondary schools is a particular focus of such calls for teacher diversity, as the teaching force remains predominantly white and female. Indeed, the teaching force is increasingly unrepresentative of the students they teach – black teachers constitute only 6.7% of the teaching force while black students constitute 15.4% of the student body (NCES 2017).
The mismatch between student and teacher demographics is important because a growing body of rigorous evidence shows that students, particularly black students, reap immediate academic benefits from exposure to same-race teachers. A seminal paper by Thomas Dee uses data from the Tennessee STAR experiment, in which students and teachers were randomly assigned to classrooms, to show that the math and reading test scores of both black and white primary school students significantly increase in years when they’re taught by a teacher of the same race (Dee 2004).1 That this result was observed in an experimental setting – where students and teachers were randomly assigned to classrooms – suggests that this is a causal relationship and not a spurious result driven by, say, involved parents requesting that their children to be assigned to particular teachers.
No single study is perfect, of course, and Dee acknowledges two limitations to his study. First, the benefits of having a same-race teacher may be unique to the context of relatively disadvantaged schools in Tennessee in the late 1980s – the setting for the STAR experiment. This is not so, as the basic finding that having a same-race teacher boosts current academic performance has since been replicated in other states such as North Carolina (Clotfelter et al. 2007) and Florida (Egalite et al. 2015), and in other settings, such as community colleges (Fairlie et al. 2014) and law schools (Birdsall et al. 2018). Moreover, having a same-race teacher improves non-academic outcomes such as reduced suspensions (Lindsay and Hart 2017) and improved school attendance (Holt and Gershenson 2017, Tran & Gershenson 2018).
Second, Dee (2004) and the subsequent literature examine the contemporaneous effects of having a same-race teacher on end-of-grade test scores, school suspensions, and chronic absence. Such outcomes are mainly of interest because they are seen as proxies for the long-run outcomes of ultimate policy interest, such as educational attainment, which facilitates upward socioeconomic mobility, improves health, reduces criminal activity, and improves labour market outcomes. However, it is not obvious that short-run effects of student-teacher racial match on test scores, suspensions, and attendance should carry over to long-run impacts on educational attainment, as teachers’ effects on test scores fade over time (Cascio and Staiger 2012, Chetty et al. 2014, Jacob et al. 2010).
In a recent paper, we set out to determine whether there are long-run effects of having a same-race teacher in primary school on educational attainment (Gershenson et al. 2018). Evaluating such long-run effects can help policymakers more accurately assess the potential benefits of diversifying the teaching force, given the large positive externalities of increasing communities’ educational attainment, and closing racial gaps in educational attainment.
Specifically, we isolate causal effects by returning to the Tennessee STAR experimental data used by Dee (2004) to see whether random assignment to a black teacher in grades K–3 increased the chances that black students graduated high school, took a college entrance exam such as the ACT or SAT, and ultimately enrolled in college.2 We find robust evidence that black students randomly assigned to a black teacher are five percentage points (7%) more likely to graduate high school and four percentage points (13%) more likely to enrol in college than their same-grade, same-school peers who are not assigned to a black teacher. These effects are large, though similar in size to the long-run impacts on the same outcomes of other primary school inputs such as reduced class size (Dynarski et al. 2013).
Importantly, we replicate these effects by applying a quasi-experimental strategy to more recent data from North Carolina, showing that the long-run impacts of having a same-race teacher are not unique to the STAR schools and cohort. Specifically, we isolate arguably random variation over time in the racial composition of a given school’s teaching force.
These striking results raise the policy-relevant question of what mechanisms are driving these effects. The answer has implications for devising an efficient policy response. While the available data limit our ability to decisively answer this question, we find suggestive evidence that rules out some explanations and identifies some likely channels.
- First, we rule out the explanation that black teachers in the schools serving black students are simply better qualified or more effective in a general sense.
We do this by showing that controlling for observed qualifications such as teaching experience does not mediate the long-run impact of having a same-race teacher, and by showing that white students’ long-run outcomes are not affected by random assignment to a black teacher.
- Second, we propose a model of the education production function that distinguishes between two remaining channels: race-specific teaching skills and role-model effects.
The former may be rooted in black teachers’ ability and/or willingness to teach hidden curricula (Foster 1990) and employ culturally relevant pedagogy (Ladson-Billings 1995), while the latter is more passive and suggests that the presence of a college-educated professional who ‘looks like you’ causes students to update their beliefs about the returns to effort and engagement in school. Critically, our models suggest that under the first scenario, having multiple black teachers should produce larger effects on long-run outcomes. Alternatively, the role-model scenario suggests diminishing returns to having a second same-race teacher, if the necessary information is transmitted by the first teacher and students update their beliefs and behaviours accordingly. We find evidence that both mechanisms are at play, and that the primary mechanism varies across outcomes.
For college entrance-exam taking, there is no dosage effect of having a second black teacher, suggesting that role-model effects drive black students’ college aspirations, if not their actual enrolment decisions. This also suggests that student assignments to existing black teachers could be approached strategically to maximise the number of black students exposed to at least one black teacher. Still, the limited number of black teachers and the uneven distribution of black teachers across schools will present challenges in ensuring that all black students are exposed to at least one black teacher. Moreover, there are likely benefits to white students of having black teachers as well, in terms of racial and social attitudes. This raises the question of how we might increase the diversity and representation of the teaching force, rather than simply managing assignments to existing teachers.
Achieving representation in the teaching force would require roughly doubling the share of black teachers. The pipeline is simply not there to make this happen in the near future (Putman et al. 2016), so we would have to recruit more college-educated and college-enrolled black candidates into teaching. Supposing this is feasible, however, we must consider two unintended consequences of such a policy.
- First, this would necessarily reduce racial representation in other professions that require a college degree that have similar representation problems and in which there are large, documented benefits to representation, such as law (Birdsall et al. 2018) and medicine (Alsan et al. 2018).
- Second, because teaching is a relatively low-paying occupation, moving black college graduates from more lucrative fields to teaching, at current teacher-pay levels, would exacerbate racial gaps in earnings and intergenerational mobility.
We perform simple back-of-the envelope calculations to show that the median earnings gap between black teachers and black, college-educated non-teachers is about $4,000. Thus, hiring an additional quarter million black teachers to achieve representation would amount to one billion dollars in lost wages for black professionals annually.
What, then, should be done in the short run? There are promising light-touch, low-cost interventions that could improve white teachers’ effectiveness with non-white students. For example, the empathy workshops piloted by Jason Okonofua et al. (2016) show great promise, as do well-designed implicit bias workshops and teacher training programs that emphasise culturally relevant pedagogies and maintaining high expectations for all (Dobbie and Fryer 2015). Finally, role models need not be teachers, as a diverse group of professionals could visit schools and engage with students to reach a similar goal. Recent experimental work, for example, finds that one-time, one-hour visits from female scientists in high-school science classes increased the likelihood that female students apply to selective science majors in college (Breda et al. 2018).
A policy response along multiple dimensions, including making every effort to recruit, support, and retain talented black teachers, and providing relevant training to white teachers, coupled with a bit of creativity in how students are assigned to classrooms and exposed to possible role models, can potentially close stubborn racial gaps in educational attainment.
Alsan, M, O Garrick and G C Graziani (2018), “Does diversity matter for health? Experimental evidence from Oakland,” NBER, Working paper w24787.
Birdsall, C, S Gershenson and R Zuniga (2018), “The effects of demographic mismatch in an elite professional school setting,” Education Finance and Policy forthcoming.
Breda, T, J Grenet, M Monnet and C Van Effenterre (2018), „Can female role models reduce the gender gap in science? Evidence from classroom interventions in French high schools,” Paris School of Economics, mimeo.
Cascio, E U and D O Staiger (2012), “Knowledge, tests, and fadeout in educational interventions,” NBER, Working paper w18038.
Chetty, R, J N Friedman and J E Rockoff (2014), “Measuring the impacts of teachers II: Teacher value-added and student outcomes in adulthood,” American Economic Review 104(9): 2633–79.
Clotfelter, C T, H F Ladd and J L Vigdor (2007), “How and why do teacher credentials matter for student achievement?” NBER, Working paper w12828.
Dee, T S (2004), “Teachers, race, and student achievement in a randomized experiment,” Review of Economics and Statistics 86(1): 195–210.
Dobbie, W and R G Fryer Jr (2015), “The impact of voluntary youth service on future outcomes: Evidence from Teach For America,” BE Journal of Economic Analysis & Policy 15(3): 1031–1065.
Dynarski, S, J Hyman and D W Schanzenbach (2013), “Experimental evidence on the effect of childhood investments on postsecondary attainment and degree completion,” Journal of Policy Analysis and Management 32(4): 692–717.
Egalite, A J, B Kisida and M A Winters (2015), “Representation in the classroom: The effect of own- race/ethnicity teacher assignment on student achievement,” Economics of Education Review 45: 44–52.
Fairlie, R W, F Hoffmann and P Oreopoulos (2014), “A community college instructor like me: Race and ethnicity interactions in the classroom,” American Economic Review 104(8): 2567–2591.
Foster, M (1990), “The politics of race: Through the eyes of African-American teachers,” Journal of Education 172(3): 123–141.
Jacob, B A, L Lefgren and D P Sims (2010), “The persistence of teacher-induced learning,” Journal of Human resources 45(4): 915–943.
Holt, S B and S Gershenson (2017), “The impact of demographic representation on absences and suspensions,” Policy Studies Journal.
Ladson-Billings, G (1995), “Toward a theory of culturally relevant pedagogy,” American Educational Research Journal 32(3): 465–491.
Lindsay, C A and C M D Hart (2017), “Exposure to same-race teachers and student disciplinary outcomes for black students in North Carolina,” Educational Evaluation and Policy Analysis 39(3): 485–510.
NCES (2017), Digest of education statistics, National Center for Education Statistics, Washington, DC: US Department of Education.
Okonofua, J A, D Paunesku and G M Walton (2016), “Brief intervention to encourage empathic discipline cuts suspension rates in half among adolescents,” Proceedings of the National Academy of Sciences 113(19): 5221–5226.
Putman, H, M Hansen, K Walsh and D Quintero (2016), “High hopes and harsh realities: The real challenges to building a diverse workforce,” Brookings Institution.
Schanzenbach, D W (2006), “What have researchers learned from Project STAR?” Brookings Papers on Education Policy 9: 205–228.
Tran, L and S Gershenson (2018), “Experimental estimates of the student-attendance production function,” IZA, Discussion Paper 11911.
 See Schanzenbach (2006) for an engaging overview of the Project STAR experiment.
 We are deeply grateful to Susan Dynarski and Diane Schanzenbach for sharing data on STAR students’ college enrollments, which was initially used in their study with Joshua Hyman of class size’s impact on college enrollment (Dynarski et al. 2013).