Atlas Growth Intelligence

Atlas Growth Intelligence (urban-growth-signals)

Core question. Is this place growing, declining, stabilising or transforming, and what evidence supports that?

Keep three claims apart:

• A wealthy place is not necessarily growing.
• A growing place is not necessarily emerging.
• A pattern is not a forecast.

Findings to report

1. Growth. Population, households, real income, housing stock or jobs are expanding (or contracting).
2. Momentum. Growth is speeding up or slowing down compared with an earlier period.
3. Structural transition. The composition is changing even if population is stable. Examples: affluence upgrading, tenure shift, education upgrading, commercial intensification.
4. Emergence. Several independent dimensions point to the place becoming a different kind of neighbourhood or commercial centre.
5. Anomalies. The place behaves unlike its peers, or its signals diverge. For example, home values rise while jobs fall.

Procedure

1. Resolve the geography.

   • A ZIP: area_change(action="growth_profile", zipcode="11530").
   • "Around" a place or address: get its point first (find_geography(action="geocode", query=...)). Then call growth_profile with latitude, longitude, mode (walk or drive) and minutes (1 to 30).
   • A bare point: the ZIP that contains it.

2. Profile. area_change(action="growth_profile", ...). Read these fields of result:

   • classification: growth_state, momentum, transitions, growth_character, emergence_stage;
   • signals: growth, momentum, relative, structural, convergence, divergences, transitions;
   • sufficiency;
   • evidence_detail: dated records by dimension.

3. Emergence. For a place: area_change(action="emergence", zipcode=...). For discovery, such as "find emerging ZIPs in Dallas": area_change(action="emergence", metro="Dallas", limit=20), optionally with state, stage and min_confidence.

4. Compare. area_change(action="growth_compare", zipcodes=["11530", "11237"]). It compares ZIPs only; profile each catchment on its own and compare ZIPs with ZIPs.

5. Explain each headline finding before stating it, especially anything surprising: area_change(action="growth_explain", topic=..., <same geography>). Topics are:

   • growth_state, momentum, growth_character, emergence;
   • a transition: affluence_upgrading, commercial_intensification, …;
   • a divergence: values_up_jobs_down, …;
   • any metric name.

   Report the competing explanations it returns.

6. Context (optional).

   • housing_market(action="rent_history", zipcode=...) for the rent path.
   • housing_market(action="affluence_context", zipcode=...) for value levels.
   • crime(action="context", zipcode=...) for reported crime; it is context, never a growth signal.
   • area_profile (population, income_distribution, ...) on a context from context.resolve for current composition.
   • places(action="count", context=...) for today's commercial inventory, which is not a history.

Rules for the answer

• Give every number with its period, geography and source, all of which the records carry.
• Records with status not_significant, insufficient_history, unavailable, suppressed or withheld are not evidence of change. Say so. Never fill them in.
• Records with status proxy are county series. Call them county context; they never prove local change.
• A transition with significance: not_tested (tract-weighted catchment values carry no margin of error) is a direction, not a significant shift. Say so.
• Affluence comes from real income and from the constant-dollar share above $200k (2024 dollars). Never infer it from a nominal income bracket: inflation alone moves households across a fixed dollar line.
• Remote work (remote_work_shift) is a national pattern. Report it, but never let it make a place "transforming".
• Convergence counts independent dimensions only: demographic, economic, housing, commercial, mobility, urban fabric. Rents and home prices together are one dimension.
• Say "emerging" only when the emergence stage says so, or when at least three dimensions converge with a supported transition. Otherwise say "growing", "transforming" or "stable".
• Peer-relative statements use the returned percentiles: 70th or above is above peers, 30th or below is below.
• Do not forecast or claim causes. The engine (growth_v1) is descriptive and analytical, not backtested. Use "consistent with" and "suggests", and give the competing explanations.
• Name the standing gaps when they matter: there is no history of places opening or closing (commercial change comes from Census ZIP Business Patterns), and demographic acceleration needs a third ACS period.

Output format

Start with a short verdict, then this block, then the evidence and caveats.

skill: urban-growth-signals
location: <name>
geography: <ZIP nnnnn | drive 10-minute catchment around …>
classification:
  growth_state: <growing|declining|stable|transforming|insufficient_evidence>
  momentum: <accelerating|decelerating|steady|mixed|insufficient_evidence>
  growth_character: <label or null>
  emergence_stage: <stage or n/a>
signals:
  - name: <e.g. Affluence upgrading>
    type: <growth|momentum|structural_transition|emergence|anomaly>
    direction: <positive|negative|flat|mixed>
    evidence: [<metric (period, source)>, ...]
    status: <supported|not_significant|not_tested|insufficient_history|unavailable|national_pattern>
confidence: <sufficiency level>; predictive use not validated
gaps: [<...>]