Sebastian Olascoaga

City of Boston · MIT

Published Work·Theses

PhD dissertation · MIT · 2023

Housing dynamics in the face of shocks.

Sebastian Sandoval-Olascoaga. Advised by Mariana C. Arcaya. Committee: Atheendar Venkataramani, Albert Saiz, devin michelle bunten.

PhD in Urban Economics & Policy · MIT

This dissertation examines three different shocks with three different causal-inference designs and recovers the same pattern in housing systems each time. Chapter II follows a tornado that destroys a neighborhood, Chapter III a pandemic that lifts an eviction moratorium, and Chapter IV long-run neighborhood change in Greater Boston. In each case the trajectory shows a flat pre-event period followed by a post-event divergence that does not recover. Public policy can shorten the divergence only when its responses are triggered automatically rather than left to discretionary declarations.

Every panel shows a flat pre-shock trend and a post-shock divergence that does not recover.

Each panel is the headline finding of one chapter, both event studies with a flat pre-event period and a post-event divergence that does not recover.

Two-panel small multiples of the PhD dissertation's chapter findings: tornado-affected property values falling for nine years with no recovery (event study), and COVID-19 hazard ratio rising after states lifted eviction moratoria (event study, log scale) CH. II · CLIMATE SHOCK Tornado-affected home prices vs. comparable non-affected homes $0 −$10k −$20k y −5 0 +9 −$23k at year 8 21,174 tornadoes · 1996–2019 · pre-trends p = .60 · FEMA aid had no effect CH. III · HEALTH SHOCK COVID-19 hazard ratio after states lifted eviction moratoria 3.0 2.0 1.0 0.5 wk −15 0 +12 HR 1.83 12 weeks after 509,694 individuals · log scale · pre-trends p = .98 · 95% CI [1.36, 2.46] Each panel uses its own units. Both follow a flat-pre-event, divergence-post-event structure.

Source: Sandoval Olascoaga (2023), Housing Dynamics in the Face of Shocks, MIT PhD Dissertation · Chapter II uses Sun & Abraham (2021) event-study on 21,174 NOAA tornado paths × CoreLogic/Redfin property transactions · Chapter III uses Cox event-time hazards on 509,694 OptumLabs enrollees (Y-axis log scale; 95% CI [1.36, 2.46] at week 12) · 95% confidence bands shown

21,174

Tornadoes (1996–2019) used as natural experiments in Chapter II

−$23,412

Drop in tornado-affected home prices by year 8, with no recovery (Chapter II)

HR 1.83

COVID-19 hazard 12 weeks after states lifted eviction moratoria (Chapter III)

Chapter II · Climate shock

The effects of localized climate shocks on places and people

Working paper · single-authored. Treats roughly 21,000 U.S. tornado paths (1996–2019, NOAA Severe Weather GIS) as exogenous treatment polygons, with a 1 km ring as control. Sun & Abraham (2021) event-study estimator with property-level fixed effects across CoreLogic and Redfin transaction data, Data Axle business records, the Stanford Educational Opportunity Project, and the FEMA Presidential Disaster Declaration Database.

What it finds. Property prices fall by $8,055 two years after the tornado and reach −$23,412 by year 8 (p = .002), with no recovery. FEMA Presidential Disaster Declarations have no statistically significant effect on sale prices. Tornado intensity (Fujita scale) doesn't matter either. A striking null. Property tax collections drop by ~$300 per property the first year and reach −$400 by year 9. School math scores fall −0.26 SD by year 9. Households are 8.8% more likely to live in a different ZIP code one year after a tornado. Small businesses (1 to 2 employees) drop −21% immediately. The "no-recovery" growth path is empirically real, and the federal aid program does not bend it.

Chapter III · Health shock

Eviction-moratoria expiration and COVID-19 infection risk

Published as Sandoval-Olascoaga, Venkataramani & Arcaya, JAMA Network Open (2021). Individual-level Cox proportional-hazards difference-in-differences in an event-time framework on 509,694 OptumLabs Data Warehouse enrollees, 43 states + DC, with the week of each state's moratorium lift as the event.

What it finds. Twelve weeks after expiration, hazard ratio reaches 1.83 (95% CI 1.36 to 2.46). The effect is strongly stratified by baseline health and neighborhood context. HR 2.36 for CCI ≥3, HR 2.31 for high-rent-burden ZIPs, HR 2.14 for high-poverty ZIPs. A null result on individuals' own ZIP-code changes argues the mechanism is community spillover, not personal eviction. This recasts eviction policy as health-equity infrastructure. Full paper page →

Chapter IV

Neighborhood change in Greater Boston

A third chapter applies the same flat-pre-event, post-event-divergence lens to long-run neighborhood change across Greater Boston, combining quasi-experimental analysis with community-based participatory research alongside the Healthy Neighborhoods Study Consortium.

Shock stabilizers should be triggered automatically and shouldn't depend on political cycles or emergency declarations to be activated. From the dissertation.

Cite this dissertation

Sandoval Olascoaga, S. (2023). Housing Dynamics in the Face of Shocks. PhD Dissertation, Massachusetts Institute of Technology, Department of Urban Studies and Planning.

Related workDissertations

Chapter III · 2021 paper

When states ended their eviction bans, what happened to COVID-19?

JAMA Network Open

MA · PSE · 2015

How rich are the richest in Mexico, really?

Advised by Alvaredo & Piketty