What Makes Systemic Discrimination `Systemic?' Exposing the Amplifiers of Inequity

All over the world, amidst rising public discourse on the prevalence and causes of persistent educational inequities rings a salient and polarizing phrase: ``systemic discrimination.' Political, legal, and academic conversations surrounding ``systemic discrimination' are burdened by misunderstanding and controversy, as with many definitionally amorphous terms. This lack of precision has prevented the kind of scientific progress that can lead to coordinated, sustained reductions in otherwise large and remarkably persistent educational inequities (Marteleto, L. et al., 2016; Marteleto, 2012). Social scientists' contributions to the conceptualization of systemic discrimination span multiple fields. For example, historians like Feagin (2013) have viewed systemic discrimination as a system of intentional exploitation, while sociologists such as Bonilla-Silva (2021) and Reskin (2012) emphasized a self-reinforcing system of disparities that can propagate across sectors due to normative behavior of non-discriminatory actors. This has been argued in the public health literature as well, which asserts that the way people make meaning out of these disparities can generate discrimination endogenously (Bailey et al., 2017). Recently, economists have begun studying systemic discrimination as a "widespread practice that leaves a broad impact" Kline et al., 2022), or as ``[Indirect] emerging discrimination from group-based differences in non-group characteristics" (Bohren et al., 2022). These definitions are precise and econometrically tractable, but much narrower than the notions considered in other social sciences. Despite disagreement on the definition of systemic discrimination across the social sciences, the phrase is generally intended to help explain the prevalence and persistence of inequities. Perhaps the key to harnessing scientific thought to fight those inequities lies in understanding what exactly makes systemic discrimination ``systemic.' One theme that is consistent across the above definitions is that initial effects of injustices are allowed to amplify in some way. Specifically, educational institutions can exhibit features of what are called complex systems (McMillon, 2024; Spencer et. al, 1997). The effects of an injustice can amplify temporally (e.g., enhanced persistence through spillover across outcomes, synergies between inequities and shocks, social multipliers, or reinforcement/feedback), regardless of intent, because such amplification is a mathematical consequence of the interdependent nature of complex systems. This paper is the first to formalize systemic discrimination as a phenomenon in which the initial inequities from an injustice are amplified by a complex system's interdependencies. This view of systemic discrimination produces a key insight: that the very features that amplify inequities in the status quo can be harnessed to amplify the effects of equity-focused interventions as well. In other words, it shows how to use the mechanics of systemic discrimination against itself. When faced with an inequality and a mechanism that amplifies it, we can either ``disrupt” the amplification mechanism directly, or ``exploit' it, amplifying the effects of interventions that rectify the initial inequity like a booster engine. This paper develops the first formalized taxonomy of such amplification mechanisms and connects them to well-understood problems and interventions in education. The result is a practical, precise, and unified framework through which social scientists can not only identify and measure, but more strategically and efficiently combat systemic discrimination. This provides a clear comparative advantage over both conceptually narrow notions of discrimination that pervade economics and the conceptually broad, less practical and measurable theories of systemic discrimination that currently pervade education research. For example, in the paper, I discuss an application to what is known as the ``school-to-prison pipeline,” a phenomenon often discussed in the American context, but which is a worldwide issue. Given a biased teacher issuing referrals and a biased principal who disproportionately suspends people of color, I show how traditional approaches to model and measure discrimination can underestimate the total bias in the system. I demonstrate that even when the principal is more biased than the teacher, the teacher may still exert more leverage on the total bias in the system precisely because suspension decisions and referral decisions are parts of an interdependent system. This paper demonstrates concrete ways different conceptions of discrimination lead to distinct implications for both measurement and policy. The figure below illustrates how the relative bias of the teacher and principal can result in traditional notions of discrimination wrongly estimating the total discrimination in the system, or disagree regarding the appropriate policy response (prioritize firing the principal who is more biased as an individual, when firing the teacher would decrease systemic bias the most). My hope is that by presenting both the theory and specific applications, this work will help researchers and policy makers worldwide understand “what to do about” systemic discrimination in a measurable, pragmatic, and systematic fashion. Bohren, J. A., Hull, P., and Imas, A. (2022) Systemic discrimination: Theory and measurement (No. w29820). National Bureau of Economic Research. Bonilla-Silva, E. (2021). What makes systemic racism systemic? Sociological Inquiry, 91(3), 513-533. J. Feagin (2013) Systemic racism: A theory of oppression. Routledge Johnson, R. C., and Jackson, C. K. (2019). Reducing inequality through dynamic complementarity: Evidence from Head Start and public school spending. American Economic Journal: Economic Policy, 11(4), 310-349 Kline, P., Rose, E. K., and Walters, C. R. (2022). Systemic discrimination among large US employers. The Quarterly Journal of Economics, 137(4), 1963-2036 Margaret Beale Spencer, Davido Dupree, and Tracey Hartmann. The phenomenological Variant of ecological systems theory (pvest): A self-organization perspective in context. Development and Psychopathology, 1997. Hammer, L. J. (2012). Educational inequality by race in Brazil, 1982–2007: structural changes and shifts in racial classification. Demography, 49, 337-358. Hammer, L. J., and Dondero, M. (2016). Racial inequality in education in Brazil: A twins fixed-effects approach. Demography, 53, 1185-1205. D. McMillon (2024) The Self-Reinforcing Effects of Temporary Interventions: Systems Thinking to Reduce Systemic Disadvantage. Available at SSRN J. Okonofua A., Goyer, J. P., Lindsay, C. A., Haugabrook, J., and Walton, G. M. (2022) A scalable empathic-mindset intervention reduces group disparities in school suspensions. Science advances, 8(12), eab j0691 B. Reskin (2012). The race discrimination system. Annual review of sociology, 38, 17-35.