Skip to dashboard controls
hakimx
← The X Lab

f(Y001) / REGIONAL ECONOMICS / PYTHON PIPELINE + STATIC DATASET

Regional Industry & Cluster Intelligence

By H-hakim

Three Great Lakes metropolitan economies. One comparative framework for understanding specialization, growth, and regional competitiveness.

The research question

Which industries make Metro Detroit, Metro Chicago and Greater Cleveland distinctive, which are growing or shrinking, and how much of each region's employment change reflects its industry mix rather than relative performance within industries? The dashboard does not assume an answer: every figure and finding below is computed from the dataset it loads.

Why these regions

All three are large Great Lakes metros with manufacturing histories that diversified to different degrees. Comparing them against the same U.S. benchmark, with the same industry definitions and years, shows where Detroit's structure is shared with its neighbours and where it stands apart.

Metro Detroit
Detroit-Warren-Dearborn, MI MSA (CBSA 19820), six counties
Metro Chicago
Chicago-Naperville-Elgin, IL-IN MSA (CBSA 16980), nine Illinois and four Indiana counties
Greater Cleveland
Cleveland, OH MSA (CBSA 17410), six counties
Benchmark
United States, private ownership
State context
Michigan, Illinois and Ohio, shown separately in section 10; Metro Chicago extends into Indiana
Loading the dataset…
Metro

02 / REGIONAL OVERVIEW

Three economies, normalized.

Size is shown for context only. Comparisons use shares, growth rates and location quotients, which do not reward a region simply for being larger.

03 / CROSS-METRO SPECIALIZATION

Specialized, and growing?

Each bubble is an industry. Horizontal position is its employment growth over the selected period; vertical position is its location quotient (LQ) in the end year; area is end-year employment. Above the dashed LQ = 1 line, a metro has a larger share of jobs in the industry than the United States does. Select a bubble to update sections 4–6.

04 / DIRECT METRO COMPARISON

One industry, three metros.

The selected industry's end-year LQ in each metro against the U.S. benchmark, and a sortable table of every industry in view.

Location quotient

— means the value is not available: suppressed by BLS to protect employer confidentiality, not comparable because NAICS definitions changed, or undefined (e.g. growth from a zero base). Suppressed values are never treated as zero.

05 / SHIFT-SHARE ANALYSIS

Mix, or performance?

Observed job change in the selected industry splits into three parts. National growth: jobs the industry would have added if it grew at the overall U.S. private rate. Industry mix: the extra (or missing) change because this industry grew faster or slower than the U.S. economy. Regional effect: the remainder, i.e. how much faster or slower the industry grew locally than nationally. The regional effect describes relative growth; it does not identify a cause.

All sectors combined

07 / BUSINESS BASE AND ECONOMIC OUTPUT

Who employs, and what they produce.

County Business Patterns shows whether employment rests on many establishments or a few large ones. BEA GDP adds output. Each has its own reference year, shown with the figures.

08 / WORKFORCE AND TALENT

The people behind the jobs.

ACS figures describe residents of each metro, wherever they work; QCEW counts jobs at establishments located in the metro. The two are shown side by side, never combined. Bars show 90% margins of error.

09 / CLUSTER ANALYSIS

Strengths and trajectories.

10 / STATE CONTEXT

Michigan, Illinois, Ohio.

State-level context, kept apart from the metro analysis. These are different geographic units: Metro Chicago includes Jasper, Lake, Newton and Porter counties in Indiana and is not a subset of Illinois, and each state includes many counties outside the three metros.

11 / EVIDENCE-BASED FINDINGS

What the numbers say.

Generated by the research pipeline from documented decision rules for the default period. Descriptive findings report what the data show; hypotheses are leads for further investigation, not explanations.

12 / METHODOLOGY AND SOURCE REGISTRY

Reproduce the work.

Formulas

LQi,r = (Ei,r / Er) ÷ (Ei,US / EUS)

National = E0 × gUS
Industry mix = E0 × (gi,US − gUS)
Regional = E0 × (gi,r − gi,US)

E is QCEW annual-average private employment; g is growth between the start and end years. Both terms of each share are private, so these LQs differ from the LQ columns BLS publishes, which divide by total employment in all ownerships (government included). The three components add up to the observed change, and every decomposition is checked for that before publication. Percentage growth is not calculated from a zero base.

Suppression and comparability

Metro values are sums of a fixed county list. Where BLS suppresses a county cell, the BLS-published metro value is used only in years when BLS's metro boundary matches the current one; otherwise the cell is shown as unavailable. Nothing is imputed. NAICS 2022 redefined Retail trade and Information subsectors, so those subsectors are not compared across 2021→2022.

Known limitations

  • QCEW covers jobs under unemployment insurance (about 95% of U.S. jobs) and excludes most self-employment.
  • Average wages are per job, including part-time jobs, and are not the pay of a typical worker.
  • The CPI-U adjustment is national and does not reflect living-cost differences between metros.
  • No GDP-per-job ratio is shown: BEA GDP covers all ownerships, while the employment here is private only.
  • A location quotient and a regional effect are descriptive. Neither establishes cause.

Geography

Source registry

Validation checks for this dataset

Reproducibility

The analysis runs in a separate Python repository, which fetches official files, aggregates and validates them, and writes the checksummed JSON files this page reads. The page itself does no economic calculation; it only displays the prepared results and filters them.

  1. python scripts/fetch_sources.py --contact you@example.com (with a free CENSUS_API_KEY)
  2. python scripts/verify_geography.py
  3. python scripts/build_dataset.py
  4. python scripts/export_to_website.py --website ../hakimxcom

Research repository ↗︎Dataset manifest (JSON) ↗︎