What I wanted to measure

The guiding question is whether median rent climbed faster near a major park investment than in a similar tract without that story. A full answer needs matched neighborhoods, a clear construction date, and controls for things like crime reporting bias and school catchment narratives. This section lays out the indicators I use, why they wobble, and what careful claims can look like.

American Community Survey (ACS) basics

The Census Bureau publishes median gross rent (contract rent plus utilities where tenants pay them) at city and tract levels. Tract-level estimates update on a rolling basis and come with margins of error that get wide when the tract is small. Ignoring the margin of error makes it easy to read noise as a trend.

I treat ACS as a flashlight, not a verdict. It shows whether rent stress in an area is high compared with the rest of the region. It does not prove that a new median strip or a park opening caused a lease increase on a specific block.

Median rent and ACS tables

To compare areas near Lake Merritt and Fruitvale, median gross rent can be drawn from ACS Table B25064 at data.census.gov for selected census tracts and five-year estimate windows. Example tracts often used for illustration (verify boundaries in TIGERweb): Alameda County 4033 (Lake Merritt south shore / downtown adjacency) and 4060 (Fruitvale BART area). Report the survey years, cite the table, and include margins of error for tract-level cells before claiming change over time. Renter share and unit mix (ACS Table B25003 and related tables) affect how median rent should be read, not only the park timeline.

Evictions and investor purchases

Eviction counts from court data miss informal moves: cash for keys, harassment, roommates leaving quietly when the rent jumps. Investor share metrics from deed records also lag and sometimes hide LLC shells. I still look at them because they show whether housing near a park is shifting into portfolio ownership, which changes how tenants experience a green upgrade.

Evictorbook (Anti-Eviction Mapping Project) and Urban Habitat link Oakland eviction data to ownership networks. Their public synthesis (2022) reports that corporate landlords accounted for about 25% of Oakland evictions across a recent five-year window while owning only about 8.9% of multifamily buildings—order-of-magnitude evidence of concentration, not a story about every small landlord. The city’s Rent Adjustment Program open data captures formal notices; it misses buyouts, informal pressure, and many no-fault paths.

Foreclosure-crisis research on Oakland (often citing James Yelen’s UC Berkeley thesis and Participedia summaries) describes tens of thousands of foreclosures between 2007–2012 and high investor share of resales—useful background for why corporate portfolios show up in eviction statistics today, but not a substitute for current deed-level work.

Data ethics

I kept geography at aggregate levels. I did not try to map individual households. Where I talk about tenant experience I lean on reporting from community groups and public comment, not on guessing someone’s income from a dot on a map.

What I can say without lying

Green amenities can correlate with rising rents because they sit inside a housing market that already treats housing as a commodity. They can also sit next to organizing wins that lock in affordability. Correlation is not causation, but silence is not neutral either. If a city leads with park renderings and trails behind on enforcement against tenant harassment, that order signals who the plan imagines as the public.

Causation checklist

  • Time order: did rents or investor activity move before or after the park headlines?
  • Geography: is pressure regional (Bay Area-wide) or localized to one tract?
  • Confounders: transit, crime narratives, schools, interest rates, remote work, state law.
  • Direction of causality: Reibel et al. (2021) find park funding may track into already-gentrifying areas—green investment and displacement pressures interact; the arrow is not always “park first.”
  • Displacement is often informal; eviction counts are a floor, not a total.
  • Scale: a signature waterfront bond hits land markets differently than a pocket park (see Anguelovski et al., 2018).
A clean chart still needs an honest methods note. This page leaves out plenty; the limits matter as much as the trend lines.