Skip to content
MSB Renew

Colorado carries 59 percent of the eight-code signature

Counted on the FinCEN export read 2026-08-21, Colorado holds 13.0% of the money services businesses whose registration renewal falls due on 2026-12-31. It holds 59.2% of the cohort’s strangest feature: the 1,086 registrants who all declare the identical set of eight activity codes. This page prints what one public dataset shows about that concentration, and stops exactly where the dataset stops.

Every count below is computed on cohort-2026.csv, extracted 2026-08-21 from the public FinCEN MSB Registrant Search by pipeline.py, the same file behind the cohort overview. That article reported Colorado’s rank as observed, not explained. This one measures how much of the anomaly sits in one block of filings.

The eight codes, now named

The overview article printed the codes as bare numbers because FinCEN’s legend was unreachable that day. It no longer is. FinCEN appends the key to every bulk export of the registrant list, and we re-downloaded the export on 2026-08-25 (33,288 rows, same public endpoint). The trailing key, titled “NEW KEY-Form 107 (effective on 3/2012)”, names them:

codeFinCEN’s label, verbatim from the export key
401Issuer of traveler’s checks
402Seller of traveler’s checks
404Issuer of money orders
405Seller of money orders
408Check casher
409Money transmitter
413Seller of prepaid access
415Dealer in foreign exchange

So the signature is a business declaring, at once, that it issues and sells traveler’s checks, issues and sells money orders, cashes checks, transmits money, sells prepaid access, and deals in foreign exchange. For scale: 80.0% of the cohort (7,750 of 9,690 registrants) declares one, two, or three activities. 1,089 registrants declare exactly eight, and 1,086 of those declare this same set.

Where the 1,086 are

Every figure in this section is a count of rows in the cohort file, read 2026-08-21:

measurenationalColoradoColorado share
registrants in the cohort9,6901,25813.0%
eight-code signature1,08664359.2%
street address shared with 5+ registrants89659266.1%
both flags at once47635674.8%
at least one of the two flags1,50687958.4%

Read the middle column the other way and it is starker: 51.1% of all Colorado registrants in this cohort (643 of 1,258) carry the eight-code signature. In no other state is that share above a tenth of registrants at comparable volume; the next largest eight-code counts are New York with 109 and California with 48.

Two more shape facts about the 1,086, nationwide: 1,068 of them (98.3%) declare zero branches, and 476 (43.8%) also sit on a street address shared with five or more other registrants. Eight activities, one establishment, and, nearly half the time, a crowded address.

Denver, then one ZIP code

Inside Colorado, still on the same export read 2026-08-21, the concentration continues to narrow. Denver holds 979 of the state’s 1,258 registrants, and 543 of its 643 eight-code registrants (84.4%). Narrow once more: of Colorado’s 592 shared-address registrants, 444 (75.0%) sit in the single ZIP code 80202, in central Denver. The pattern is not spread across a state. It funnels into a few buildings.

Set the block aside and Colorado is ordinary

A useful arithmetic check, computed on the same FinCEN export read 2026-08-21: remove every registrant carrying either flag and re-rank the states.

rankwith everythingwithout the flagged block
1California 1,531California 1,480
2Colorado 1,258Texas 1,186
3Texas 1,232Michigan 443
4Michigan 443Colorado 379
5New York 389Florida 351

Without its 879 flagged registrants, Colorado drops from second place to fourth, at 379, between Michigan and Florida, roughly where a state of its size would sit. This is not an explanation of the pattern. It is a localization of it: the Colorado anomaly the overview article named is almost entirely this one block of filings.

What is observed, and what is only inferred

The observed part is everything above: counts of rows in a public federal register. Our pipeline labels the eight-code set a suspected shell-formation signature. That label is an inference we drew from the shape of the data (maximal declarations, zero branches, stacked addresses), and it should be read as exactly that. It is not a finding of fact about any named business, and this page names none. A legitimate operator that genuinely conducts all eight activities from one office would carry the flag too, and the register gives us no way to tell the two apart from outside.

What the register’s own operator says is worth quoting here. The MSB Registrant Search page, read 2026-08-25, warns in its own voice: “You should not trust a company only because it is listed in the MSB Registrant Search,” and “FinCEN does not approve or endorse any business that has registered as an MSB.” Registration under 31 CFR 1022.380 is a filing obligation, not a vetting. A pattern like this can exist in the register precisely because nothing in the registration process is designed to prevent it.

Method

Source: the public FinCEN MSB Registrant Search bulk export, extracted 2026-08-21; the activity-code key re-read from a fresh export of the same endpoint on 2026-08-25. The cohort is every deduplicated registrant whose computed renewal date falls on 2026-12-31 under the renewal rule of 31 CFR 1022.380(b)(2) and (b)(3) (eCFR text read 2026-08-25, current through 2026-08-24): “Each two-calendar-year period following the initial registration period is a renewal period,” and the renewal form “must be filed on or before the last day of the calendar year preceding the renewal period.” The export does not say whether a given filing was a renewal or an initial registration, which shifts the due date by a year; the pipeline applies the renewal rule uniformly and discloses that choice. Shared-address means an exact normalized street-address match across five or more registrants; the eight-code flag means set equality with the codes listed above. Every figure is a count of rows, with no sampling and no estimation. The full derivation chain is on the method page.

What this page cannot tell you

  • Who files the Colorado registrations, or why. The export carries names, addresses, codes, and dates. It carries no filer identity beyond that, and we did not investigate any.
  • Whether any individual flagged registrant is a shell. The flag is a pattern over the aggregate, not a verdict on a row.
  • Whether the pattern is growing. That takes a second extraction at a later date, and this page is built on one.

Published by Neige AI, Inc., August 25, 2026. See the method and sources.

This page is independent research, not legal advice. It quotes FinCEN’s own rulemaking and the public MSB register with pinpoint citations. Verify anything load-bearing against the Federal Register text itself before acting on it.

Read the same register for your own MSB

The lookup at msbrenew.com/check runs the same public FinCEN export this analysis was built from, for the one registrant name you actually care about.

Check my renewal date
Every MSB Renew post