The California Civic Dashboard

What the numbers mean

This is a sensitivity model applied to the dashboard’s project sample, anchored to September 18, 2026. It estimates a share of reported gross project units potentially delivered by a selected year. Completed projects, unknown unit counts, non-housing records and unknown stages are excluded. It does not identify new developable parcels, deduplicate overlapping master plans, subtract demolished homes or separate already-built phases.

City comparisons reflect uneven sample coverage. Unknown affordability is not treated as zero; affordable totals include only disclosed counts. Old records may describe homes already delivered. The model flags records missing a status date or older than two years.

A model you can inspect

Reference delivery times: proposed 6 years, review 4.5, approved 3, construction 1.5. Long-run delivery shares: 55%, 70%, 85%, 95%. Delivery = reported units × stage share × (1 − 2^{−years elapsed / adjusted delivery time}). Elapsed years use the selected year minus 2026.

Each percentage point of financing above 6% lengthens delivery time by 10%; each cost-growth point above 3% adds 4%. Delay adds months. Policy assumptions subtract at most 18 months combined, with a minimum delivery time of 6 months. These coefficients are illustrative, not fitted to historical outcomes. The range varies delivery time and realization; it is not a statistical confidence interval.

Policy & geographic limits

Policy switches test additional acceleration relative to the reference. Legal changes may already be reflected in project records. There is no causal estimate of each bill’s effect. The bond switch assumes approval and funding beginning in 2028; neither is guaranteed.

Transit proximity uses straight-line distance to hand-curated approximate centers for a subset of Caltrain and BART stations. VTA, BRT, station entrances, service frequency, zoning, local alternative plans, hazard exclusions and parcel boundaries are not modeled.

Station references: Caltrain · BART. Bill sources are linked in each card. No automated bill-status feed is connected.

Real public layers

The FEMA snapshot queries all nine Bay Area counties. The September 19, 2026 refresh returned 13,603 hazard polygons across eight counties; no San Francisco polygons were returned by this query. Blue areas are special flood hazard areas; purple areas have a 0.2% annual flood chance. Geometry is generalized to about 3 meters; point screening can be wrong near boundaries. A location with no match may still flood. This is not an insurance determination. Check FEMA’s official map.

NOAA sea-level tiles are live public overlays, relative to mean higher high water. The chosen water level is independent of the simulated year. Blue depicts inundation depth; green depicts low-lying areas not directly connected to the ocean. NOAA viewer and caveats.

Home-value sensitivity, not appraisal

The observed baseline is the dashboard’s July 2026 citywide Zillow index. Reference value = baseline × (1 + assumed annual trend)^years. The supply factor is exp(−sensitivity × extra modeled delivery / 2020 housing stock), capped between exp(−0.3) and exp(0.3). Housing stock comes from MTC / ABAG’s Census 2020 DHC H3 data. West San Jose uses citywide San Jose data.

All-area values are a 2020-stock-weighted average of city indexes where available, not Zillow’s regional index. A flood-exposed example applies your assumed discount to that index; it is not a price for a specific parcel. Interest rates influence modeled delivery only, not buyer demand directly. Gross project units, overlapping plans, stale stock, incomplete coverage and assumed elasticities limit every output. No statistical confidence is assigned to prices.

Energy context

Points show the California Energy Commission’s facilities not flagged as retired in its public snapshot, last updated June 12, 2026. Nameplate MW is not real-time generation, grid headroom, an outage forecast or a service territory. Battery MW describes power, not storage duration. No local home is assumed to be supplied by a nearby plant.

Modeled annual household demand = delivered units × your kWh/home/year assumption. Additional grid connection delay lengthens the housing-delivery assumption. Both are hypothetical. CEC data and conditions of use.