Notorious POD, Ep 32: Musings on AVMs and Appraisers

Musings is a solo version of the Notorious POD in which I think out loud about issues in the economy, technology, and real estate industry.

In this episode, I think out loud about AVMs, appraisers and the future of valuation at least for purposes of lenders. I’ve been researching and investigating valuations and appraisals in the context of Defi lending, and a reader sent me the just published Fannie Mae fact sheet on Desktop Appraisal. This feels like something that is potentially a big deal, so I muse about it.

Notorious POD, Ep. 32: Musings on AVMs and Appraisers

Podcast Only:



8 thoughts on “Notorious POD, Ep 32: Musings on AVMs and Appraisers”

  1. Do you think an appraiser will trust 3rd party inspections for data from untrained people when it’s the APPRAISER takes the entire responsibility for the final opinion? These “data gatherers” have absolutely no consequences for their actions. Lenders do not want actual appraiser trainees (trained by their mentor) to do this work, but will take the info provided by an Uber driver! Strange times indeed…….

    • I don’t know the answer to that. I hope maybe actual appraisers can chime in.

      My current take though is something like this:

      Appraisers will trust third party randoms if the lender is okay with that.

      Lender will be okay with that if Fannie will buy/guarantee the mortgage.

      Fannie has just told lenders that they’re okay with that.

      So if Fannie has approved the loan, then the lender can sell that loan to the secondary market, and then the loan is not their problem anymore. Investors are not relying on the lender or the appraiser at that point; they’re relying on Fannie Mae’s guarantee.

      Now, could some lender get sued and then sue the appraiser for using third party inspections or data from real estate agents? Maybe. Until some judge somewhere rules on that, we won’t know for sure.

  2. Hi Rob! To clarify for your listeners who may be wondering, the new Desktop Appraisal isn’t an AVM. The value is concluded by an appraiser, based on data from the MLS on the subject property and an ANSI-compliant floorplan that meets appraisal specifications for how square footage is calculated.

    This eliminates the need for an inspection and will speed the process up considerably.

    You are correct that this is a very big deal 🙂 It has big implications for the MLS value proposition, lenders, appraisers and more. Love to see you digging in on this subject.

    • Excellent point. Thank you Jeff. Desktop Appraisal is simply Fannie Mae’s software for accepting AVM-based appraisals. And yes, so far, DA requires the use of a licensed appraiser, but without the physical inspection.

  3. Hi Rob… long time reader, first time caller,

    Appraiser here. The biggest news this (DA) announcement makes is the requirement for a “floor plan with interior walls.” This is beyond the typical scope of work for a traditional appraisal report which only requires a sketch to calculate GLA. I see CubiCasa has chimed in, above. In their current marketing they state, “CubiCasa is the only technology that meets all Desktop requirements and can be performed by ANYONE with a smartphone and no training.” I guess if they say it out loud it must be true.

    These 3D interior scanning products are suggesting they “align with ANSI-standards” (which will be the requirement for all GLA reporting come April 1st). Not to get into the weeds about ANSI standards of measurements, but these 3D products produce an interior measurement and estimate the exterior walls in their calculations. Appraisers actually measure the exterior to meet ANSI standards — see the potential problems?

    Why is no-one talking about the relative accuracy of these smartphone measurements?
    Since when is “good enough” and “accurate enough” okay in the eyes FNMA?
    Why did FNMA slide in the word “floor plan” versus the long adopted “sketch” requirement?

    And Rob, don’t get me started on PRIVACY! Not one of these companies, from AMCs to the GSEs to these data collectors are discussing the privacy concerns of the homeowner/future owners. If you scan my daughters room, my safe, etc. etc… Who sees the data? Where is it being stored? Is the data being repackaged and sold? Shouldn’t the consumer know these things? Exhale…

    • Oh man, such good points, such good questions. I’m planning on doing a podcast on this topic with Greg Robertson on our Industry Relations. Maybe we ought to turn that into something with Q&A type of a thing too. Maybe we can look at doing it as a livestream with live Q&A.

      Really really good stuff. Thanks.

    • Andrew I don’t get that part either. I am a real estate photographer and one of the only advantages to using a $3k Matterport camera over a $900 Ricoh Theta is the accuracy of the measurements and scan. They can be off by up to 10%. Neither are substitutes for a measuring tape.

  4. thanks for this podcast – good point of view. There seems to be two definitions of price in these cases. 1) A price based on comps only to project a price for a refinance and 2) A price to confirm a purchase of a property where there is a buyer and seller. Is price defined as what a buyer is willing to pay and a seller is willing to receive or something else? In option 1 there isn’t any agreement of value by the 2 parties – just the computer and some checks and balances. Option 1 is less public and maybe not used in comps for a sale but I could argue these shouldn’t be used in comps for a sale while option 2 has more validity for comp purposes. I’m a Realtor and have always told clients when it comes to an appraisal it is a 30 page checkbox – is the price we have determined OK or NOT OK – it is just a formality to the logic the buyer and seller and both agents have come up with.

    On another note, our brokerage did a study of zestimates last year over a several months – we tracked the zestimate before any of our listing data went public (pre-listing), we tracked it after we went on the market (price is now public) and tracked it again after it sold to see how the zestimate changed in all these cases. I don’t have the numbers handy but the zestimate changed in response to our info becoming public (it should have since it was more data into the algorithm) but it also reinforce that if our sellers sold to the zillow iBuyer model they would have lost significant equity. Maybe the zestimate was biased to their iBuyer model. Maybe my point is that computer prices don’t reflect buyer/seller (good or bad) and should be split out into a different category.

Comments are closed.