Unmasking Inequality in Real-Time: Examining the St. Louis Choice Neighborhoods Initiative using High Resolution Data

• In Event: Innovative Rental Market and Neighborhood Data

Abstract

The primary impediment to advancing social equity in American housing is a structural "blind spot" in the national data infrastructure: the absence of annual, tract-level socioeconomic data. While the American Community Survey (ACS) provides annual estimates for large geographies, it only releases 5-year moving averages (MWA) for Census tracts. This temporal smoothing acts as a "low-pass filter" that masks the neighborhood volatility essential to social science, effectively diluting rapid racial or economic shifts and hiding the precariousness of low-income neighborhoods. Consequently, the impacts of hyper-localized, place-based interventions—such as the Choice Neighborhoods Initiative (CNI)—often remain statistically invisible until the intervention period has passed.This study resolves this precision tradeoff by introducing the "TWDeConv" (Two-Way Deconvolution) framework, an innovation that mathematically recovers latent annual signals trapped within 5-year aggregates. The methodology utilizes a dual-axis optimization: temporal deconvolution via L1-Trend Filtering to isolate annual "shocks" from aggregate noise, and spatial regularization to stabilize estimates by borrowing strength from adjacent tracts. By inverting the 5-year smoothing effect, we transform the ACS from a lagging historical description into a dynamic tool for active policy correction.We apply this framework to the St. Louis Choice Neighborhoods Initiative, a multi-million dollar federal investment in a city marked by deep-seated geographic fragmentation and North-South disparities. Using the recovered annual data, we implement a Synthetic Control Method (SCM) to construct a counterfactual "Synthetic CNI Tract" from non-treated St. Louis neighborhoods that matched the intervention area’s trajectory prior to the CNI. This allows us to isolate the year-over-year "treatment effect" of the intervention on racial composition and median income, determining whether the CNI triggered inclusive growth or rapid displacement.The primary contribution of this study is the democratization of high-resolution temporal data, providing a validated, "discontinuity-aware" framework that eliminates the structural "blind spot" in urban research. By transforming the ACS from a lagging historical indicator into a dynamic signal recovery system, this project enables researchers and policymakers to pinpoint the exact year a neighborhood intervention succeeds or fails, facilitating active policy correction rather than the observation of blurred decadal shifts.

Author

• Yung Chun, Washington University in St. Louis
  Presenting Author