Commercial Banking · the United States
| Measure | Value |
|---|---|
| Net establishment formation, 5yr | -1.45% |
| Standing among all industries | 20th of 287 |
| Reads as | dying faster than it is forming |
| Signal strength, backtested | ρ 0.57 over 4 windows |
Among 287 industries, this one sits at the 20th percentile on net establishment formation over 2019-2023 — it is dying faster than it is forming. That signal carries a measured rank correlation of 0.57 with the following decade's growth across 4 backtested windows, separating top and bottom deciles by 42.6 percentage points. It describes the INDUSTRY, not a business in it, and it was calibrated against establishment growth rather than returns to an operator.
| Measure | Value |
|---|---|
| Operators | 82,411 |
| People employed | 1,625,804 |
| Pay per employee | $107,374 |
| What the work is | Tellers |
| Median wage, that occupation | $39,230 |
| Licensed in | 27 jurisdictions |
| Dimension | Weight | Evidence | Predicts? |
|---|---|---|---|
| O1 · Demand density and recurrence | 25 | evidenced | predictive |
| O2 · Payment evidence | 15 | partial | not backtested |
| O3 · Licence gate favourability | 15 | evidenced | not backtested |
| O4 · Labour availability | 15 | evidenced | not backtested |
| O5 · Incumbent weakness | 10 | no producer | not backtested |
| O6 · Margin structure | 10 | partial | not backtested |
| O7 · Ceiling | 10 | evidenced | DESCRIPTIVE ONLY |
65 of 100 rubric points are evidenced, 25 partially, 10 have no producer. No total score is rendered from that. PREDICTS is a separate question from EVIDENCE: a dimension can rest on a real sourced figure and have no measured relationship with what happens next.
| Revenue band ($000) | Firms | Share | Labour share of revenue |
|---|---|---|---|
| <100 | 53 | 1.3% | 1274.0% |
| 100-499 | 67 | 1.6% | 49.6% |
| 500-999 | 72 | 1.7% | 41.6% |
| 1,000-2,499 | 276 | 6.7% | 31.0% |
| 2,500-4,999 | 478 | 11.6% | 29.6% |
| 5,000-7,499 | 420 | 10.2% | 28.2% |
| 7,500-9,999 | 312 | 7.6% | 28.1% |
| 10,000-14,999 | 499 | 12.1% | 27.2% |
| 15,000-19,999 | 327 | 7.9% | 26.5% |
| 20,000-24,999 | 240 | 5.8% | 26.9% |
| 25,000-29,999 | 168 | 4.1% | 26.0% |
| 30,000-34,999 | 122 | 3.0% | 26.3% |
| 35,000-39,999 | 108 | 2.6% | 26.1% |
| 40,000-49,999 | 143 | 3.5% | 27.2% |
| 50,000-74,999 | 206 | 5.0% | 25.1% |
| 75,000-99,999 | 122 | 3.0% | 24.7% |
| 100,000+ | 505 | 12.3% | 26.9% |
Payroll is 26.8% of receipts industry-wide across 4,118 firms. LABOUR SHARE, NOT MARGIN: Census publishes receipts and payroll and never cost of goods, so gross margin cannot be derived and is not claimed here. Enterprise receipts — a firm counted once at its total size, not per location. Census SUSB 2022.
| Gap |
|---|
| licence titles in this scope — none matched 'Tellers', though it is licensed in 27 jurisdictions nationally. The title differs locally, or the record is held municipally. |
| occupation mix for NAICS 522110 — the labour statistics carry no rows at this level, so everything about occupations here, including the defining occupation and its licence gate, is measured at NAICS 522000, a broader grouping. Establishment counts are for 522110 itself. |
| licence coverage — the register holds 54 of 56 US licensing jurisdictions, and 2,504 of its records carry no jurisdiction at all. A count of 27 is out of 54 searched, not out of 56. The unsearched jurisdictions are not evidence of no licence. |
| bonding_required_usd — no producer, and not estimated |
Stated rather than filled. An absence of match is not an absence of fact.
This document's content hash is below. Anyone can check it at ro.8thwavemedia.com/verify without an account and without taking our word for anything. A single altered character produces a completely different hash, so a document that verifies is a document nobody has revised.
Issued 05 September 2026 at 00:57 UTC · rubric O1.0-20260804
Do not expect sha256sum of this file to match. A document cannot contain its own hash — adding the hash would change it. The hash above covers the report content as issued, excluding this verification panel. That is stated here rather than left to be discovered, because a reader who hashes the file, gets a different answer and concludes the document was tampered with would be wrong for a reason we created.