Technology is good. I love it almost to a fault. One of the skills of using new technologies is understanding where its use is appropriate, cost effective, reliable, etc. One of the technologies that has polarized the appraisal industry is the use of AVM’s aka Automated Valuation Models. Just check how many entries in Google come up using the words avm “real estate”? About 111,000 entries at last count.

Here’s a simple summary of AVM’s in the appraisal industry

AVM’s or Automated Valuation Models are computer programs that use relevant real estate information (sales prices, property characteristics, demographics, etc.) to calculate a value for a specific property. Public records are the primary data source. One weakness in AVM’s is the lack of a physical exterior or interior inspection of the properties utilized. AVM’s in the private sector are primarily used for residential home equity or credit loans.

They also have been popular with homeowners trying to find out a value of their home for a nominal fee [LA Times]

Appraisers seem to be generally against AVM’s because they are viewed as a threat, syphoning off potential business. For example, how much home equity work have appraisers done in the past 5 years? Not much, its pretty much all done by AVM’s.

But lets be honest here, based on the ever loosening lending practices, it doesn’t matter what the result of the AVM search is, the home equity deal is relatively low risk and the deal usually gets done. AVM’s reduce upfront costs and can handle higher volume (I am not addressing costs of defaults and additional exposure for the lender).

Yet there has been resistance to the use of AVM’s in the first mortgage business. Why? Because the accuracy is not there.

Here’s the problem: There’s not a great way to test the results. The data is often flawed. I have had conversations with a number of lenders who indicate that of the 8 to 10 national AVM services, 1 is fairly reliable, 1 is barely acceptable and the remainder are not worth the paper they are reported on. However, an AVM report is still something that can be placed in the files for the regulators.

In a rising market, “rising waters float all boats.” As the real estate market begins to weaken across the country, this is the moment of truth for AVM’s.

Don’t get me wrong, AVM’s are here to stay and have their purpose and function. What I object to are the overpromises and inappropriate use of the technology perpetrated by the industry, especially from appraisal management companies seeking out new revenue streams. Remember, these are the same firms passing off poor quality appraisal reports as the real thing – AVM’s are a lot more complicated to manage.

Remember the AVM mantra: Garbage in, garbage out

Tags: , ,


5 Responses to “Automated Valuation Models (AVMs), A Day Early And A Dollar Over”

  1. pcampbell says:

    Very ineresting article. My question is how do these AVM computor programs obtain their repective State RE appraiser license? I have yet to see a computor sitting in a RE class with a book in front of it or taking a test! Does the person inputting the subject data into the program have a RE appraiser license?

  2. Annemieke says:

    One of, if not THE, biggest problem with AVM’s is that the “comparables” are not always valid. We have had Underwriters come back to us questioning why we used certain comparables and not others and they would provide us with a list showing “average” values in a certain neighborhood.

    Upon close examination of the “comparables” every other one had a financial institution listed as the seller. REO properties are sold under value and therefore are not representative of the Market …. thus are inappropriate comparables.

    In the meanwhile, these AVMs skew the true numbers. Not to mention the fact that AVMs do not tell of the true sf, the condition, etc.

  3. jheron says:

    I ask myself, if I were an invester or a lender with a billion dollar portfolio, what would I rather have as a supporting document for the value of my portfolio: a computerized valuation or an appraisal done by a reliable, experienced certified appraiser? I think most would opt for the certified appraisal.

  4. Rick says:

    “I ask myself, if I were an investor or a lender with a billion dollar portfolio, what would I rather have as a supporting document for the value of my portfolio: a computerized valuation or an appraisal done by a reliable, experienced certified appraiser? I think most would opt for the certified appraisal.”

    Then you don’t understand the realities / economics of this marketplace. Majority of the Lenders sell the paper. Buybacks are very uncommon (so far). Their intent is to maximize the number of loans produced, and maximize their profit on each and every loan.
    AVM = $5-$15 Cost to Review by U/W = $0
    Appraisal Cost = $0 (Customer pays); Review by U/W and/or Reviewer and/or Field Review = $10 – $150 – $350+ depending. “Underlying value? What’s THAT ??”

  5. JM says:

    AVMs are going to severely damage not only traditional appraisers, but also homeowners and some lenders, whether they have a buck and a half portfolio, or multiple billions. The problem is in the data. Yeah, the lenders’ real market is in reselling their loans in huge bundles and their bottom line looks a lot better when they’ve spent only $15/per appraisal rather than a few hundred. But all the complaints here are accurate: the data input computer or person is not a license appraiser; no physical inspection (not even a drive-by) has been done; AVMs don’t care about REO and the usual condition that means, they also don’t necessarily factor in seller concessions (I regularly see them up to $50k here). So AVMs seem to often be wrong on BOTH ends of the value scale, and use “comps” that are not appropriate to the subject because the “human element” has been eliminated.

    The problems for lenders, including their insurors are down the road. The problem for traditional appraisers is staring us in the face.

    JM