Jimmy Lo

Projects

Research, data science, housing economics, and AI-related work.

Following the Preference, Missing the Optimum: Compliance Without Optimization in AI Housing Recommendation
AI EvaluationLLM AuditingRecommender Systems

Following the Preference, Missing the Optimum: Compliance Without Optimization in AI Housing Recommendation

Independent research · An audit of 9,945 AI housing recommendations against 3,885 real New York listings with a known best answer

Strict Dominance (Pareto) AnalysisRandomization InferenceEquivalence Testing (TOST)Benjamini-Hochberg Correction
1
Evidence-Grounded Interview Feedback Synthesis Copilot
People AnalyticsAI ToolsInterview Design

Evidence-Grounded Interview Feedback Synthesis Copilot

An AI tool that structures and surfaces evidence from interview debriefs — without scoring candidates or outsourcing the decision

Structured Output ExtractionNLI-based EntailmentLLM-as-JudgePsychometric Calibration
1
Fair Lending in Crisis: Racial Disparities in FHA Mortgage Approvals Before and During COVID-19
Housing EconomicsFair LendingRacial Equity

Fair Lending in Crisis: Racial Disparities in FHA Mortgage Approvals Before and During COVID-19

Harvard Graduate School of Design · Research paper on housing finance, racial equity, and the pandemic's impact on minority homebuyers

Logistic RegressionGeneralized Linear ModelsRegression Discontinuity DesignANOVA
Every Stop Counts: How Taiwan's Pedestrian Yielding Penalty Cut Traffic Deaths by 40%
Causal InferenceTraffic SafetyPolicy Evaluation

Every Stop Counts: How Taiwan's Pedestrian Yielding Penalty Cut Traffic Deaths by 40%

Published in Transport Policy, 2025 · A quasi-experimental study on traffic law enforcement and road safety

Interrupted Time SeriesRegression DiscontinuityQuasi-Experimental DesignPoisson Regression
The Mortgage Lock-In Effect: Quantifying Rate Differentials and Existing Home Sales Suppression Across U.S. Metros
Housing EconomicsMortgage RatesMarket Analysis

The Mortgage Lock-In Effect: Quantifying Rate Differentials and Existing Home Sales Suppression Across U.S. Metros

A panel regression analysis of how the gap between locked-in and current mortgage rates reduces listing activity and transaction volume in 20 major markets

Panel Data RegressionTwo-Way Fixed EffectsEvent StudyTime Series Analysis
Does Upzoning Actually Build Housing? A Difference-in-Differences Study of U.S. Zoning Reform Outcomes
Housing PolicyZoningUrban Economics

Does Upzoning Actually Build Housing? A Difference-in-Differences Study of U.S. Zoning Reform Outcomes

Estimating the causal effect of ADU ordinances, missing middle legalization, and citywide upzoning on permitted housing units across 35 U.S. cities

Difference-in-DifferencesCallaway-Sant'Anna EstimatorEvent StudyPanel Fixed Effects