The Baltimore Situation Should Worry Every Lender
by Tony Pistilli
By now, most people in the industry have seen the reporting on the Baltimore foreclosure crisis: Hundreds of properties allegedly bought at distressed prices, loaded with liens and deferred maintenance, and then appraised and financed at dramatically inflated values. Renovations claimed on paper that were never actually done. Appraisals that used condition ratings and comparable properties that did not hold up to scrutiny under review.
What was most striking was not the scale, it was how preventable it looks in hindsight. The signals were there. The question is whether anyone’s collateral review process was built to catch them. For lenders, it is a question worth considering.
What fraud actually looks like at the property level
DSCR fraud is when someone misrepresents a property’s condition, value, or rental income to get a loan approved that would not have been approved with accurate information. The most common methods are:
» Inflated appraisals based on misrepresented condition
» Renovation claims for work that was never done
» Using comparable properties that do not honestly reflect the subject property’s condition and/or quality
The most common one is misrepresenting the property’s condition. A property in genuinely poor shape gets assigned a C3 or C4 condition rating when the photos tell a completely different story. Because most properties cluster in that range anyway, a one point misclassification does not immediately raise flags, but it can justify an adjustment worth tens of thousands of dollars on the appraisal.
Then, there are renovations that never happened. An “as-repaired” value gets assigned based on work that was never completed. Without access to the property’s listing history, the reviewer has no easy way to check whether the property’s actual condition matches the renovation claims in the report.
The third pattern is “cherry-picking” comparables. The appraiser selects properties that are of better condition or quality than the subject, then downplays the difference to make the numbers work. The subject looks more valuable by comparison, and the inflated value becomes defensible on paper.
What all three have in common is that they are very hard to catch when you are only looking at what the appraiser chose to include in the appraisal.
The problem with traditional appraisal review
A thorough reviewer with enough time can pull up photos of comparable properties, check prior listings, and find the inconsistencies. The problem is that amount of time does not exist. Quality control teams reviewing loan portfolios across the country simply cannot do that on every file, and fraudsters know it.
This is exactly the gap that property intelligence tools are designed to fill, not by replacing the reviewer’s judgment, but by giving them information the appraiser did not provide.
When you can automatically pull a property’s full listing history, including interior photos from every time it appeared on the market, you are no longer limited to the picture the appraisal presents. You can see what the property actually looked like six months ago, before and after when renovations were supposedly completed. That history either supports what the appraisal says, or it does not.
What consistent condition scoring actually changes
One of the ongoing challenges in appraisal review is that condition and quality ratings are judgment calls. The Uniform Appraisal Dataset (UAD) system, the standardized rating scale that Fannie Mae and Freddie Mac require for appraisals, describes six condition quality ratings. But two appraisers can look at the same home and land on different ratings, and research has shown that the same appraiser will sometimes rate the same property differently when they encounter it again later.
Computer vision condition scoring works differently. It applies the same criteria to every property, every time. When computer vision scores a subject property and its comparables, the same model runs on all of them. That consistency makes it much easier to spot when an appraiser’s rating looks out of line with what the property actually shows.
It does not override the appraiser. What it does is point reviewers toward the files that deserve a closer look, so they are not guessing at where to focus their time.
What lenders can do right now
Waiting for a crisis to force changes is part of how the Baltimore situation happened. There are practical steps available today that do not require rebuilding the underwriting process from scratch.
» Build listing history checks into the review process for any loan with an “as-repaired” value. Before closing, the subject property and its main comparables should be checked against what they looked like on the market.
» Use independent, computer vision derived condition and quality scores for comparables, not just the appraiser’s ratings. The appraiser’s assessment matters, but it should not be the only input. A computer vision score for each comparable either confirms what the appraisal says or flags something worth looking into.
» Make photo evidence a requirement for renovation claims. If a loan’s value is built on improvements that were supposedly completed, there should be photos showing those improvements, not just a written description saying they happened. A picture is literally worth a thousand words — and can save thousands of dollars.
None of this will stop fraud entirely. But it makes the schemes much harder to pull off, and it creates documentation that supports action when fraud is found.
The bigger picture
The patterns described are the ones that come up most often in production, but fraud looks different depending on the market and the loan type. The schemes that succeed are the ones where the people doing the collateral review rely entirely on information supplied by the people who benefit from the outcome.
What property intelligence tools do is bring in objective information from outside that loop. Listing records, interior photos, independent condition scores, none of these come from the appraiser, broker or the borrower. They tell a story that nobody in the deal wrote.
Most people in the DSCR space are operating honestly. Keeping it that way means taking the verification and validation gap seriously before the next “Baltimore” shows up, not after.
