Skip to content

Methodology

How we source, label and update every number on this site, and what the numbers can and cannot tell you.

Three kinds of numbers

Observed
Figures published by a government agency or other public source, shown as reported. We may add them up (two counties into the Inland Empire) or put them on a common base (an index), and we say so when we do. Observed data still has limits: surveys have margins of error, and agencies suppress small cells and revise recent periods.
Model
Our own calculation from public inputs: an estimate of something no agency publishes directly. Every model is described on the page that uses it, with its inputs and the assumptions that matter most. Treat a model as a careful estimate, not a measurement.
Scenario
A what-if: what the numbers would look like if a stated assumption held (a large earthquake, a price shock). Scenarios are not forecasts. Every chart that uses one carries the “Scenario” label with a dashed-outline marker, and its heading says so. (A dashed line inside a chart is explained by that chart’s own key: it can also mark an average.)

Sources

We use only public data. We never use data from Restavo’s customers, its software or anyone’s private records. Each chart names its sources, each with its own link and how current it is; the end of every study lists each dataset with the date it was retrieved. This list is built from the same files the charts use.

Some useful records have no bulk public download, such as the South Coast AQMD’s Rule 1403 asbestos notifications. We do not scrape sources like that; we will add them if they publish data in bulk.

What “as of” means

“As of” is the latest period the data covers, not the day we downloaded it. Government data arrives with a lag: county employment from QCEW is published about five to six months after the quarter ends, and the latest month of producer prices is preliminary and often revised. A chart marked “As of Q1 2026” describes the first quarter of 2026, even if you are reading it in the fall.

Update cadence

Each dataset is rebuilt from its source on the source’s own schedule: monthly for prices and permits, quarterly for employment, as events happen for disasters and fire perimeters, and annually for housing and insurance tables. When a source revises past numbers, our charts change with it. A study’s “Updated” date is the later of the last edit to its text and the last change in any of its data. “Retrieved” dates are in Pacific time.

Geography

We report by county, using the Census Bureau’s five-digit county codes, for Los Angeles, Orange, Riverside, San Bernardino, San Diego and Ventura counties. Riverside and San Bernardino together are the Inland Empire. Maps use Census cartographic boundary files, simplified for the web. The rest of California will follow.

What the data does not show

Public data counts what agencies can see. Employment figures cover payroll jobs, so self-employed tradespeople and much informal work are missing. Industry codes put a firm in one industry even when it does several kinds of work. We describe what changed and where; when two things move together we say that, and we do not claim one caused the other unless the source does.

Not advice

Nothing here is legal, insurance, health, safety or financial advice. That applies especially to hazardous materials such as asbestos: we report public facts and point to the agencies that regulate them.

Corrections

If a number looks wrong, check the linked source first; if we made the mistake, we fix it, and the study’s “Updated” date changes. You can reach us through restavo.ai.