Automated valuation models (AVMs) that use algorithms and big data are routinely used to determine the value of residential real estate. Although this is still not widely the case in commercial real estate, technology is gradually being developed to accurately value these assets. Richard West, general manager of LightBox, spoke to Scotsman Guide about the future of AVMs in commercial-property deals.
How commonly are automated valuation tools used in commercial real estate deals?
It’s definitely used far less than [with] single-family residential. It’s usually for simple commercial real estate assets, and they tend to be owner-occupied or nonleased, smaller properties.
Can you describe how this technology works to assess values?
The main thing that these automated valuation models do is ingest large amounts of data, such as assessor records at the county level. They also ingest general databases — things like market analysis and statistics — which would include inventory levels. So, for example, how many office buildings are there in a market? What is the total square footage of office buildings in the market, the vacancy rates and in various locations overall?
They would also pull in things like demographics, population density [and] flood-zone maps. You could go on and on. There’s a long list of publicly available and privately available general databases that they’ll pull in. And then the second thing they’ll pull in is transaction data, so that would be the market details, such as specific rental-rate transactions.
So, the AVM will take all of that generalized information, and then run it through an algorithm and come up with an estimate of value for all the buildings that they’ve targeted in an overall area. They’ll do a pass of the entire market and come up with an estimated value for all the buildings in that market. When a specific value is requested, it might update the algorithm for that one building based on all the other calculations that have already been done.
What property types are suitable for this technology?
In general, what would produce a higher confidence level from an automated valuation model would be simple properties. And I’ll define that as property types like multifamily, where you have less-complex lease terms, less-complex income and expenses to analyze. Another broad category would be nonleased, owner-occupied buildings, because you don’t need to consider the lease terms.
What are the barriers to this becoming mainstream?
One barrier would be the quality of information. An example of that would be if I’m trying to pull a dataset together that accurately describes sales in a market. There’s quite a few nuances, such as how long was the property on the market? How many buyers put bids in on the property? What was the motivation of the seller? How did the story behind the actual transaction influence the sale?
Another barrier in that data-quality category would be the lease comparable. It’s really important to understand all of the terms in the lease. Examples of those terms would be the rental rate, escalations over the period of the lease options and whether the tenant may have a tenant-improvement allowance. And there are other terms that could impact the overall transaction, but the absence of that quality of data is a big barrier.
Do you believe appraisers will almost always be needed in commercial real estate deals?
I believe appraisers will continue to be involved because there are so many variables in a commercial real estate transaction, and great appraisers are very adept at analyzing and validating data. The market will continue to demand that because of the benefit, even if it’s more expensive and it’s slower than an automated valuation model. ●