Fifty Sets of Rules Is the Easy Part. The Categories Themselves Are Contested.
Every lender operating across state lines knows the requirements differ. [4] That is not the part that causes operational trouble.
The trouble is that the classifications used to organise those differences are themselves inconsistent between authorities — so a rule set built on one classification will disagree with a reviewer working from another, and both will believe they are correct.
Start with what is settled
Some state distinctions are unambiguous, and these are the easy ones to encode.
Community property. Nine states: Arizona, California, Idaho, Louisiana, Nevada, New Mexico, Texas, Washington and Wisconsin. [1] The list is stable and universally agreed. (Alaska is an opt-in jurisdiction, which is a genuine edge case rather than a disagreement.)
The operational consequence is concrete: in a community property state, a non-borrowing spouse may need to execute certain documents even though they are not obligated on the debt, and the absence of that signature is a defect that no amount of borrower-side review will surface, because the borrower's own documents are complete.
Foreclosure procedure. Judicial foreclosure is the only route in close to half the states. [2] This does not change what documents an originator collects, but it changes the security instrument used, the notice provisions inside it, and the loss-severity assumptions an investor applies to the loan.
Even here the picture is not binary. Several states sit in both categories depending on circumstance — Oklahoma and South Dakota are non-judicial unless the homeowner requests judicial; New Mexico and the District of Columbia appear on both lists; Hawaii uses both routinely. [2] A rule encoded as "state X is non-judicial" is wrong in exactly the cases that matter.
Where the classifications actually conflict
Now the harder category.
Ask how many states require an attorney at a real estate closing and you will get different answers from different authorities. An earlier version of this article quoted a specific three-way breakdown. Those figures have been removed rather than re-sourced: they could not be traced to an identifiable authority, and published lists use materially different definitions of what "requires an attorney" means. [3] Replacing one unsourced count with another would not help. The useful observation is why such counts disagree at all.
"Requires an attorney" conflates several distinct things:
- an attorney must be physically present at closing
- an attorney must prepare or review the deed and security instrument
- an attorney must certify title
- attorney involvement is customary and effectively mandatory in practice, without a statute
compelling it
- the requirement applies only to certain transaction types or counties
A state can satisfy one of these and not the others. Counting it as "attorney required" or "title company permitted" is a modelling decision, not a fact — and different sources make it differently.
This is the part that produces disputes between a QC finding and a reviewer's judgement. The finding says the file is missing a required attorney certification; the reviewer says that requirement does not apply in that state; both are working from a defensible classification, and neither is being careless.
Getting it right comes first. Being able to show it comes second.
The instinct is to resolve this by getting the classification right — commission an authoritative survey, encode it, move on. That instinct is correct, and this article previously undersold it. Determining the applicable law accurately is the primary obligation. A meticulously versioned record of the wrong interpretation is still the wrong interpretation, and version history is no defence to having misread the statute.
What a records discipline adds is a second, separate obligation: being able to demonstrate later which authority was relied on, over what scope, from which effective date, on what reasoning, and who approved it. Both are required. Neither substitutes for the other.
The reason the second one needs deliberate attention is that two things move underneath it.
The underlying law moves. State legislatures amend recording requirements, notice periods, licensing and disclosure obligations on their own schedules. A classification correct in January may be wrong in September, and there is no single feed announcing it.
The classification is a judgement that can legitimately be revised. When a lender's counsel concludes that a state's requirement is narrower than previously modelled, the rule changes — not because anyone was wrong, but because the interpretation was refined.
Neither of those makes the correctness question go away. The first question is always which reading of this state's requirement is right, on what authority, from what date. The second, and separate, question is which reading was applied when this file was reviewed, and can that be shown later. A file needs both answers; this article is about the second because it is the one most often missing, not because it outranks the first.
A review performed in March under March's understanding of a state requirement is defensible. The same review becomes indefensible if, in September, it can only be described using September's rule set — because it then appears the file was assessed against a standard nobody applied at the time.
What this implies for a review programme
State rules should be data, not practice. Where jurisdictional requirements live in reviewer knowledge or a regional team's habits, genuine legal differences between states become indistinguishable from inconsistent local practice. Encoding them makes the first explicit and the second visible.
Rule sets need versions with dates. Not for tidiness — because the answer to "why was this file cleared" has to be reconstructable against the standard in force at the time, and state requirements are among the fastest-moving inputs to that standard.
Ambiguous classifications should be recorded as decisions, not facts. Where a state's attorney-involvement requirement is modelled a particular way, that modelling choice and its rationale belong with the rule. A reviewer who disagrees is then disagreeing with a documented decision rather than with an unexplained finding, which is a far more productive argument.
Multi-state consistency is measurable only if the rules are explicit. A regional lender with branches across several states cannot otherwise distinguish legitimate jurisdictional variation from branch-level drift. Both look like inconsistency in aggregate reporting.
The underlying point
Multi-state lending is usually described as a volume problem: more states, more rules, more to remember. Framed that way, the answer is a bigger checklist.
It is better understood as a knowledge problem and a provenance problem. The rules are numerous, some are contested, and they move. A multi-state review is defensible when the interpretation applied was correct and the record shows which version of it was applied to a given file, under which authority, from which effective date.
One further precision: the applicable rule usually turns on a specific date in the loan's history — application, note, consummation or acquisition, depending on the requirement — not on the month the review happened to be performed. A versioning scheme keyed to review date will attach the wrong rule to loans near a transition.
That is a records property, not a knowledge property, and the distinction determines whether a review survives being questioned two years later.
Sources
- The nine community property states are Arizona, California, Idaho, Louisiana, Nevada, New Mexico, Texas, Washington and Wisconsin; Alaska permits community property by election. IRS Publication 555, Community Property sets out the same list for federal tax purposes. State-law consequences for a mortgage file are separate from the tax treatment and must be taken from state law.
- Judicial and non-judicial foreclosure availability is set by state statute, and several states permit both depending on the instrument and the circumstances. Take the position for any given state from that state's statute or your own counsel; general 50-state summaries disagree with one another and are not a safe basis for an encoded rule.
- The attorney-involvement breakdown cited in an earlier version of this article could not be traced to an identifiable authority, and published lists use materially different definitions of "involvement". It has been removed rather than re-sourced.
- Federal settlement-procedure requirements that sit alongside state rules are at Regulation X, 12 C.F.R. Part 1024. State requirements operate in addition to these, not instead of them.