How we
keep score.
An editorial description of public data under an experimental method (0.2-experimental). Not an objective fact, a forecast, a grade, or advice.
Two questions, never averaged.
For every scored indicator we ask two separate things: where is it now compared with 2017–19, and which way has it been moving. We show both, side by side, every time.
We picked 2017–19 because it is the last stretch before the pandemic scrambled every series. It is a reference point, not a target — those years had their own strains.
Monthly readings and yearly readings are counted separately and never blended. California and Los Angeles County stay separate. Who pays the income tax is shown as facts and is not scored at all.
The rating stays pending. The method is experimental and not cleared to publish a statewide grade: label meanings are not approved, real-time vintages aren’t archived, Who Pays has no approved direction, and slow domains lag 6–24 months.
Unscored items named. No averaging.
The verdict on every chart
Worse than its own 2017–19 range, and not getting better.
Worse than 2017–19 but improving — or inside the range but sliding the wrong way.
Holding inside its 2017–19 range, or better than it.
Each chart is judged on one named line, with a stated rule for whether up is good or bad (shown on every chart). For the scored indicators the verdict follows the desk’s own level and direction reading. Direction compares the latest value with about a year earlier. The verdict is a quick-read label on one chart — it is not a grade for California.
Definitions
Where the latest reading sits compared with its own 2017–19 range: worse, within, or better. Descriptive, not a health threshold.
Whether the latest releases are moving toward improving or worsening, or show no material change.
The same direction showed up on two consecutive releases, so one noisy month can’t flip the story.
Monthly signals: housing cost, permits, unemployment, job breadth, bond rating. Counted separately.
Yearly signals: budget balance, rainy-day fund, pensions, migration. Their latest data can be 4–16 months old.
A file that is stale past its limit or missing. Named, with the reason. Never counted as neutral.
Every scored indicator, right now
Supplementary · not counted
Facts without a direction
Top 1% share of resident income tax — tax year 2024: 40.22% (preliminary). Concentration is a distributional fact. Higher or lower is not automatically bad.
Net capital gains as a share of wages + gains — 2024: 12.2% (preliminary). A signal of revenue volatility, but it rises in booms. No approved direction.
A Sahm-style calculation on California unemployment (Sep 2026): −0.03, not triggered. The Sahm rule is defined and validated for the national rate in real time. This is an analogy, not the rule, and it doesn’t trigger anything here.
Version history
Strain and direction side by side. Monthly not averaged with yearly. Unknown stays unknown. No invented letter grade on top.
Averaged readings into one label. It let an improving direction cancel an adverse level and treated a missing series as neutral. Retired. Any future method must be shown beside this one, on the same frozen history, before it replaces it.