Ghost Rates: Why MRF Files Are Full of Numbers Nobody Pays

Open a machine-readable file expecting a clean list of current rates and you'll find something messier: a lot of numbers that no longer describe anything anyone pays. Terminated contracts. Duplicate postings. Rates from arrangements that lapsed and were never cleaned out. We call these ghost rates — and they're one of the main reasons a raw MRF lookup can hand you a confident, precise, wrong answer.

If you've read what an MRF actually is, you know these files are machine-generated and comprehensive. That's exactly why the ghosts accumulate.

What a ghost rate is

A ghost rate is a rate that exists in the published file but doesn't reflect a live, current contract. It comes in a few flavors:

• Terminated — a contract that ended, but whose rates linger in the file.
• Legacy — old rates from a prior arrangement that were never superseded or removed.
• Duplicate — the same rate posted multiple times, sometimes with small inconsistencies that make it look like several data points when it's one.

The common thread: a number can be real in the file and dead in practice. Its presence is not proof of life.

Why the files fill up with them

Ghost rates aren't a scandal; they're a byproduct of how the files are produced. MRFs are generated automatically, at enormous scale, to satisfy a comprehensive-publication requirement. When the incentive is to publish everything, the result is a file that errs toward including too much rather than too little — and cleaning out every stale or duplicate line isn't what the format optimizes for.

So the file grows comprehensive and noisy at the same time. The live, current rates are in there. So is a layer of residue around them. From the outside, the two look identical: they're all just rows.

How ghost rates mislead a negotiation

Here's where it bites. You're preparing a payer conversation, you look up a peer's rate for your service, and you find a number that supports your case. If that number is a ghost — a terminated or legacy rate — you've just anchored your entire ask on something the payer stopped honoring some time ago.

A payer who knows their own contracts will dismantle that in a sentence. Not because your data was fabricated, but because it was dead, and you couldn't tell. Anchoring on a ghost rate is arguably worse than having no number at all: it gives you false confidence and a position that collapses under the first informed push-back.

Recency: stale files are a market problem

There's a related, quieter problem: staleness. MRF data circulating in the market is often old. A rate from years ago can misrepresent where a payer stands today, and plenty of the data floating around isn't refreshed on any reliable cadence. "It came from an MRF" tells you nothing about when.

This is why recency has to be treated as a first-class question, not an afterthought. A number without a sense of how current it is carries a hidden expiration date you can't see. Working from current data — and being honest about what can't be confirmed as current — is part of reading these files responsibly rather than trusting whatever happens to be lying around.

Separating live signal from artifact

None of this makes MRF data useless. It makes raw MRF data unfinished. The job is to separate the live signal from the artifacts — to resolve rows to the real entities behind them, filter or flag what's terminated, duplicated, or legacy, and keep what can actually be stood behind.

That filtering is inseparable from confidence scoring: part of scoring how much to trust a rate is detecting when a rate shouldn't be trusted at all. The rates that survive are real and current; the ghosts get filtered or flagged rather than quietly presented as fact. For the short version of how this fits together, the rate-data FAQ walks through it.

The MRF gave everyone the raw material — ghosts and all. The advantage goes to whoever can tell the living rates from the dead ones.

See the live signal for your rates

If you want rates you can actually stand behind — resolved to your facility, filtered of ghosts, and scored for how firmly to lean on each one — that's what TierBench is built for. See which payers we hold rate data on, or book a 20-minute call to talk through your payers.

---

← Back to all posts