Recent advances in algorithmic technologies have raised concerns about their impact on social inequality when used in high-stakes decisions in various social domains. In particular, algorithmic technologies have been criticized for being racially discriminatory. Racial discrimination has been extensively theorized as taste-based, statistical, implicit, and/or systemic. I use quantitative models to understand the effects of systemic or structural anti-Black racism, particularly in cases where, like in algorithmic bias, there does not necessarily exist any intention to discriminate. First, I examine the promises and the limitations of algorithmic fairness metrics that aim at racial parity in mortgage loan algorithms. I show that under historic conditions of racial oppression, Black and White potential homebuyers come from different “starting points.” I demonstrate the limitations on the ability of algorithmic fairness methods to close these outcome gaps, and show that in many cases, the optimal outcome that balances equity and efficiency must be achieved with external material interventions that may not necessarily be “colorblind.” Second, I use a heterogeneous-agent stochastic partial equilibrium model from macroeconomics to demonstrate how Black-White wealth gaps are maintained over generations. I model the effects of different policies and evaluate their disparate outcomes on households across the wealth distribution, showing how disparities across multiple domains such as homeownership, education, labor, and investment are related in a structure that reproduces wealth inequality. Finally, I use three causal inference methods: logistic regression, two stage Differences-in-Differences, and robust synthetic control to evaluate the effect of facial recognition technology on homicide clearance rate, both on an ndividual case level and a department level. I examine in particular whether the effects of this technology differ for Black and White individuals, and reflect the promises and the racialized harms of carceral surveillance technologies. This thesis is ultimately centered around ideas of harm and reparations and on the relationship between computing technologies, quantitative methods, and the possibility of racial justice. I theorize structural racism across multiple domains: home-ownership, wealth, and policing, and demonstrate the possibilities and perils of computational methods and technologies.