Methodology & Limitations
Why we publish this
Most advisory firms describe their methodology in general terms and leave it there. We publish ours, including what it cannot do.
The reason is simple. We are asking creditors to act on an analysis that reaches conclusions their own underwriting did not. That is a high bar, and it is not cleared by asserting confidence. It is cleared by showing our work and being specific about where the work is strongest and where it is weakest. If you are going to disagree with us, we would rather you disagree with something precise.
The DRIFT Thesis
DRIFT stands for Distributed Role-Influenced Focus Transition. It describes two forces that act on every commercial real estate credit.
The first is distributed bias at formation. A deal passes through a chain of parties — borrower, underwriter, appraiser, credit officer, property manager — each bringing a role-driven lens to the numbers they touch. No party intends to mislead. Most are unaware their incentive structure is shaping the assumptions they produce. The effects compound before the deal is ever funded.
The mechanism is not our invention. Competitive pressure is the most frequently cited reason lenders themselves give for easing credit standards in the Federal Reserve’s Senior Loan Officer Opinion Survey, quarter after quarter. And the appraisal-independence rules written into post-2008 regulation exist precisely because the industry concluded that transactional pressure on valuation is structural rather than occasional. The pressure is documented. What is rarely done is
pricing it.
The second is focus transition. Once a deal closes, the same parties move to the next transaction. The original deal — and the assumptions baked into it — receives diminishing scrutiny even as real-world performance unfolds against those assumptions. The credit that gets reviewed is the one that has already deteriorated. By then the options have narrowed.
DRIFT holds that a credit’s true risk is best understood not as a fixed underwriting output, but as the product of these two compounding forces.
What BAR Produces
Bias-Adjusted Return is the tool we use to express DRIFT’s effect on a specific credit. It identifies the points in an
underwriting where incentive pressure had the most room to operate, applies adjustments across a defensible range, and shows what the credit looks like under each.
The output is a range, not a revised point estimate. That is deliberate. We do not believe anyone can tell you what a credit is precisely worth once you adjust for bias, and a firm that claims otherwise has replaced someone else’s false precision with its own.
We do not publish BAR’s calibration. The specific adjustments, weightings, and thresholds are our work product, and disclosing them would let any reader reverse-engineer a result without the judgment behind it. Every analytical firm withholds its model. We are telling you plainly that we do too, rather than implying there is nothing to withhold.
What our analysis is, and is not
What it is: independent decision support. A structured second read on a credit from a party with no economic interest in what you decide. A range of outcomes under stated assumptions. A written record of how the conclusion was reached.
What it is not:
• Not an appraisal, and not a valuation. We do not produce opinions of value, we are not engaged under USPAP, and nothing we deliver should be relied on as a valuation for any purpose.
• Not a credit rating.
• Not a fairness opinion.
• Not a recommendation to buy, sell, hold, or extend.
• Not investment, legal, tax, or accounting advice.
• Not a substitute for your own underwriting, your appraiser, your counsel, or your auditor.
• Not a regulatory deliverable. Nothing we produce satisfies an examination, audit, or reporting requirement on its own.
Limitations
BAR's adjustments are calibrated judgments drawn from our experience on the creditor side, not coefficients estimated from a loan performance dataset. We have not yet demonstrated statistically that bias-adjusted ranges predict outcomes better than unadjusted underwriting. We believe they do.
Influence weighting is the softest input in the framework. BAR weights each party’s incentive pull by that party’s actual influence on the outcome. Nobody has measured actual influence across a loan population. Our weights are reasoned estimates, and they are the assumption most likely to be wrong.
Our findings are directional, not point-precise. We can identify where pressure most likely entered an underwriting and
what a reasonable adjustment range looks like. We cannot tell you that competitive pressure moved a DSCR by a specific
number of basis points, and we will not claim to.
A BAR range is not a probability distribution. It is a set of scenarios under stated assumptions. The edges are not tail probabilities and should not be read as confidence bounds.
We analyze the file you give us. We do not inspect properties, re-appraise collateral, audit borrower financials, or
independently verify third-party reports unless a Statement of Work says otherwise. Errors or omissions in the underlying file will flow into our analysis.
The survey evidence supports the mechanism, not the measurement. The Federal Reserve’s lending survey samples
large institutions. We cite it as evidence that competitive pressure on underwriting standards is real and self-reported. It is not a source of our parameters and says nothing about any particular lender.
We see a selected sample. We analyze the credits our clients send us, which are not a random draw from any population.
The firm is new. AM Servicing Advisors was founded in 2026. Our principals bring roughly fifty combined years of institutional experience from Prudential, GE Capital, BNY, GlassRatner, and SitusAMC, but the firm’s own track record is short and we would rather say so than imply otherwise.
How we will know whether this works
A framework that cannot be wrong is not a framework. Here is what would confirm DRIFT and what would refute it.
Confirmation: credits whose bias-adjusted ranges flagged stress ahead of their reported metrics go on to be classified, modified, or charged off at a meaningfully higher rate than credits that did not flag.
Refutation: no relationship between adjusted flags and subsequent outcomes — or ranges so wide they carry no information.
We are building toward that test through a design partner program: a small number of clients contributing anonymized loan-level outcome data against which the framework can be calibrated empirically.
We will publish what that data shows, including if it contradicts us.
Independence and Conflicts
We are compensated by advisory fee only. We take no commissions, no referral fees, no success fees, and no position in any asset we analyze. We do not broker, originate, or service loans. We have no economic interest in whether a transaction closes or a credit is extended.
Where a conflict exists or could reasonably appear to exist, we disclose it before accepting the engagement or we decline it.