Tyan Lee

A Policymaker-Facing Recommendation Tool for Homelessness Interventions

Thu, 03 Sep 2026 00:00:00 +0000

Psychiatric Holds and Outcomes Among People Experiencing Homelessness

Wed, 02 Sep 2026 00:00:00 +0000

Empirical MPCs and the Dynamics of Fiscal Stimulus

Mon, 01 Jun 2026 00:00:00 +0000

Bridging Homelessness and Census Data: A CoC-to-Place Geographic Crosswalk

Sun, 01 Mar 2026 00:00:00 +0000

Background

HUD’s roughly 400 CoCs are the primary geographic unit for homelessness data collection and service delivery. However, CoC boundaries do not consistently correspond to any standard Census geography. Historically, therefore, researchers have been forced to pool CoCs to match county boundaries.

Researchers have taken various approaches to overcome this geographic complexity (see, for instance, concurrent work by Schachner and Gannon), but to my knowledge, there is currently no crosswalk that makes Census data available at the CoC level for every CoC in the U.S.

The Crosswalk

Manual inspection of each CoC’s boundaries using HUD documentation reveals that all CoC boundaries that do not align with Census county boundaries appear to be explained by Census place boundaries. I construct the crosswalk by applying county- and place-level inclusions and exclusions to handle CoCs whose boundaries do not follow county lines. Each row represents one CoC-county-place match and contains:

coc_code: HUD Continuum of Care identifier
state_fips: State FIPS code
county_fips: County FIPS code
place_fips: Census place FIPS code

Per-Capita Homelessness

Unlike many other economic and demographic topics, even basic homelessness statistics have rarely been presented in per-capita terms below the state level, because previously it would have been challenging to measure the population of any CoC that does not happen to fall neatly within county lines. With the crosswalk in hand, PIT count data can now be merged with ACS place-level variables, facilitating novel research. In particular, I compute per-capita homelessness at the CoC level.

Racial Disparities in Homelessness

The crosswalk also makes it possible to estimate relationships between homelessness, poverty, and other factors at a granular geographic level, capturing within-state heterogeneity and increasing statistical power. To illustrate this, I merge 2020–2024 ACS data with 2024 PIT counts and investigate how the relationship between poverty and homelessness differs by race.

As national statistics have long indicated, Black Americans experience homelessness at much higher rates than the broader U.S. population. I find that this is true in nearly every CoC. This may be partly attributable to the fact that Black Americans also experience poverty at higher rates than the broader population; again, this is the case in nearly every CoC. With poverty and homelessness rates now available at the CoC level, however, it is possible to confirm that Black Americans experience homelessness at higher rates than other racial groups, even conditional on poverty.

Conclusions and Future Work

The CoC-to-place crosswalk fills a longstanding gap in homelessness research and policy infrastructure.
Researchers can now link PIT counts to ACS housing, poverty, and labor market variables at the CoC level.
Policymakers can obtain Census data for their specific CoC, informing policy decisions.
We are in the process of linking the FBI’s Uniform Crime Reports (UCR) dataset to this crosswalk, allowing us to analyze existing variables in conjunction with crime incidence at the CoC level.
Future work will modify the crosswalk to take into account CoC boundary changes, enabling longitudinal analysis and causal research.

Acknowledgements

I would like to thank Derek Christopher and David Grusky for their extensive support and guidance throughout this project.

Bio & C.V.

Mon, 01 Jan 0001 00:00:00 +0000

During my undergraduate career, I worked as a research assistant in various fields of economics, including homelessness policy (with Prof. David Grusky and Dr. Derek Christopher), education (Prof. Ran Abramitzky), and macroeconomics (with Prof. Adrien Auclert).

My main research interests center around the effects of policy on socioeconomically disadvantaged populations in the United States. Most of my current work is on homelessness: which interventions work, for whom, and under what local conditions.

At the Stanford Center on Poverty and Inequality, I evaluated more than 40 quasi-experimental studies and randomized controlled trials of homelessness interventions. That review underpins the tool I am now building at IRiSS, which recommends suites of policies suited to each Continuum of Care’s local context, priorities, and budget. Along the way I built a crosswalk from Continuums of Care to census places, so that census and homelessness data can be used together, and presented it at a convening on unsheltered homelessness in Los Angeles in March 2026. I am also developing a quasi-experimental study on how psychiatric holds affect later outcomes for people experiencing homelessness.

As an Undergraduate Research Fellow at the Stanford Institute for Economic Policy Research, I helped compile and analyze a dataset on U.S. college admissions and student life outcomes for Prof. Abramitzky’s research on higher education and social mobility, and wrote code analyzing PSID and DINA data for Prof. Auclert’s research on inequality and asset demand. My honors thesis, advised by Prof. Auclert, uses Italian household survey data to estimate marginal propensities to consume and studies what they imply for the dynamics of debt-financed fiscal stimulus.

Before that, I spent a summer as a research assistant in Stanford’s Department of Political Science, working on the socioeconomic impacts of climate change in sub-Saharan Africa.

Education

2022–2026: Stanford University, Stanford, CA
Bachelor of Arts with Honors, Economics
Honors thesis “Empirical MPCs and the Dynamics of Fiscal Stimulus,” advised by Prof. Adrien Auclert, awarded the Anna Laura Myers Prize for Outstanding Honors Thesis.
2020–2022: Tanglin Trust School, Singapore

Research Experience

2026–present: Stanford Institute for Research in the Social Sciences (IRiSS)
Postbaccalaureate Researcher
Building a policymaker-facing tool that recommends homelessness interventions suited to a locality's context, priorities, and budget, and developing a quasi-experimental study on the effect of psychiatric holds on later outcomes among people experiencing homelessness.
Jun 2024 – Jun 2026: Stanford Center on Poverty and Inequality
Research Assistant
Evaluated more than 40 quasi-experimental studies and randomized controlled trials of homelessness interventions, and began work on a recommendation tool for policymakers that accounts for each Continuum of Care's local context. Built a CoC-to-place crosswalk that bridges census and homelessness data sources.
Jan 2024 – Jun 2026: Stanford Institute for Economic Policy Research (SIEPR)
Undergraduate Research Fellow
With Prof. Ran Abramitzky's team, compiled and analyzed a dataset on U.S. college admissions and student life outcomes to study the effect of higher education on social mobility; figures appear in “The G.I. Bill, Standardized Testing, and Socioeconomic Origins of the U.S. Educational Elite Over a Century” (2024). With Prof. Adrien Auclert, wrote Python, Stata, and R code to analyze PSID and DINA data on the effect of inequality on asset demand, for “The Race Between Asset Supply and Asset Demand” (2025), an NBER working paper presented at the Federal Reserve's Jackson Hole conference.
Jun – Sep 2023: Stanford Department of Political Science
Research Assistant
Compiled and harmonized large datasets in SPSS, Stata, and R, and ran regression analyses on public-attitudes and climate data to study the socioeconomic impacts of climate change in sub-Saharan Africa.

Presentations

Mar 2026: “Bridging Homelessness and Census Data: A CoC-to-Place Geographic Crosswalk.” Poster, Evidence, Experience, and Pathways: A Convening on Unsheltered Homelessness in L.A., USC Homelessness Policy Research Institute and SIEPR.

Activities

Stanford HAI: Co-founded, with Dr. Lukas Althoff, a Student Affinity Group at Stanford's Institute for Human-Centered AI on the labor market impacts of AI, bringing together economists and technologists for panel discussions and recorded interviews.

Skills

Technical: Python, R, Stata, SPSS, C++, Excel
Languages: English (native), Mandarin Chinese (conversational)

Music

Mon, 01 Jan 0001 00:00:00 +0000