by Nils Kok
For decades, if not centuries, the basic process of property valuation has mainly been the same: infer the current value of a building from the combination of the (stabilized) net operating income of a building and a small set of transactions of comparable buildings (“comps”). Of course, a prudent underwriter or appraiser also uses alternative methods, such as a detailed discounted cash flow analysis, and perhaps even a cost approach, but the outcome of these different methods are always reconciled into a single outcome — the market value of a building.
Current valuations are based primarily on manual assessments and processes, and with that are subject to behavioral biases such as anchoring (which means that, in large part, current values are based on past prices or appraisals) and selection bias (which means comp selection is often unconsciously influenced by factors such as how familiar a valuation expert is with a property or area). To some extent, the process of manual valuation has become more automated, increasingly assisted by spreadsheets and underwriting workflow tools, such as Argus (and a host of tech companies aiming to disrupt this field). But better workflow tools do not necessarily alleviate concerns about the influence of subjective, human judgment in the appraisal process. After all, all models rely on assumptions, and “garbage in is garbage out.”
However, the property valuation paradigm is changing. The new buzzword is “AVMs” — automated valuation models. While perhaps mystical to some, the concept of valuations based on automated models is nothing new. Across many sectors — including the medical, manufacturing, as well as the investment management sector — using models to make decisions is now commonplace. Perhaps the closest parallel is the residential real estate market, where mortgage lending is already and predominantly based on home values that are determined by some 8–10 AVM providers, including for example Collateral Analytics, CoreLogic, and HouseCanary. For banks, this means significant reductions in the time needed to make credit decisions and for ongoing portfolio monitoring, leading to cost efficiency, more optimal mortgage risk pricing and, hopefully, better risk management.
In commercial real estate, even more than in residential real estate, there remains a strong predisposition that “every building is unique” and that only a human assessment of a building’s condition and cash flows can lead to accurate valuations. But that human-centric approach stands in sharp contrast to the ever-increasing institutionalization of the commercial real estate sector: the combined CMBS and REIT sectors in the US alone have a value exceeding USD 4 trillion. These are financial assets that are traded globally and priced instantly, but paradoxically the underlying assets are valued infrequently, using archaic methods that have been shown to result in biased outcomes.
The GeoPhy AVM offers the first fully automated valuation for commercial real estate, minimizing “human” intervention to where value-add is highest (and bias is lowest) and maximizing automation where the machine has been shown to be superior to man. The applicability of an AVM to commercial real estate touches not just the traditional valuation or appraisal that is typically done for a transaction, or for investor/regulatory reporting. Many elements of real estate investment and lending could benefit from accurate, efficient, instant assessments of value — ranging from preliminary evaluation of potential acquisitions to portfolio valuation for reporting, and from the assessment of collateral value at underwriting to risk management purposes in work-outs. As an illustration, consider the commercial mortgage-backed securities issued by Freddie Mac, one of the largest lenders in the US multifamily space. These securities are part of the sizeable CMBS market through which many pension funds, insurance companies, and other institutional investors gain exposure to the US commercial real estate market. Under the K-Series program, Freddie Mac issues securities backed by multifamily mortgages with various terms, where the value of the investment is of course contingent on the ability of borrowers to make full and timely payment. One application of the GeoPhy AVM is to accurately assess, at any point in time, the value of the underlying assets in a K-Series deal — this valuation is relevant for CMBS investors at issuance, but especially over the lifetime of a loan, as market conditions and property performance change and the value of the collateral becomes more important.
Although the concept of an AVM is novel to the commercial real estate industry, the GeoPhy AVM is grounded into the fundamental principles of real estate, not least the well-known adage “location location location.” The place where a building is located, including its proximity to jobs, people, and transit, is critical for its value. But what matters in terms of location has changed (and will be changing) over time. For example, we have witnessed the move to suburban office parks in the 20th century, and the reverse in the early 21st century. Tenants have changed from well-established corporates with a credit rating to start-ups and scale-ups without a credit rating, but with a massive footprint. New sources of data have emerged: we no longer have to rely solely on government data to understand the characteristics of a neighborhood or micro-area. And, rather than relying on manual collection or human judgment of such information, we can now automate the collection, aggregation, and interpretation of data that best reflects “location.”
GeoPhy dedicates significant resources to obtain, analyze and interpret detailed local data. To be able to document all the different types of locational amenities that are relevant for property value, we first calculate “catchment areas” for each property, based on GeoPhy Reach technology. These catchment areas are comparable but distinctly different from the traditional notion of a radius around a building — a catchment is based on real walking, driving, or public transit times, using the surrounding road/transit network, as well as natural barriers. A catchment area has a distinct shape that represents how far a person can walk, drive, or get by public transit, within a certain amount of time.
The catchment area is then intersected with eight main contextual data categories that go from the macro-economic level (i.e. nationwide) to the hyperlocal level (i.e. a point of interest or event measured at an exact location). These GeoPhy Value Drivers include:
-Amenities (e.g. retail shops, restaurants, bars)
-Quality of Life (e.g. healthcare facilities, public spaces, parks)
-Financial Health (e.g. mortgage delinquency rates)
-Housing Market (e.g. changes in house price and rents)
-Local Neighbourhood (e.g. employment, population, income)
-Safety (e.g. crime)
-Work / Commute (e.g. traffic density, availability of public transit)
-Macro Economic (e.g. interest rates, stock market)
Using the example above, we can evaluate the presence of amenities in “reach” of the same property, clearly indicating the advantages of using a reach rather than radius approach.
The process of data collection and data enrichment is fully automated in the GeoPhy platform. This platform makes use of a semantic database (as opposed to a relational database), allowing for merging of hundreds of different sources of data and thousands of individual data points. The data in the platform is exposed through the use of an API, which can also be used externally. So, in a matter of milliseconds, the old adage of “location location location” is now objectively quantified, ready for human judgment in the form of a “virtual site visit” or…as input in the GeoPhy AVM.