Automated valuation models, or AVMs, have been around in residential real estate for quite a while now. The first publicly available AVM was released by Zillow way back in 2006. Interestingly, the goal of creating this valuation algorithm wasn’t to actually help people value their homes but to create controversy. Zillow’s founders later admitted that they thought putting people’s estimated home prices on display would cause a stir and drive traffic to their website. They were right. The first day that the “Zestimate” came out, the website had over a million views. By the fifth day, it was up to 2 million. Since then, Zillow has grown to become the most popular residential real estate website in the country.
Automated valuations didn’t have the same effect on commercial real estate. Valuing a commercial property is much more complicated than a residential one; there are factors like rent roll, lease comparables, and building expenses that are not publicly available. Plus, few people care about the value of their office building or favorite shopping center like they do for their neighbor’s house. But that doesn’t mean that AVMs have no use for commercial properties. In fact, they are already being used by some of the world’s largest valuation service providers.
“We have developed an AVM that is robust enough to help even our largest clients that have properties all over the globe,” said Charles Fisher, Director of Value and Risk Analytics at JLL Risk Advisory. The worry has always been that commercial property valuation is too complicated to be done well by an algorithm. “AVMs can be accurate enough to help asset managers understand which properties pass or fail their internal risk metrics,” Fisher said.
Rather than being a standalone way to conduct commercial property appraisals, AVMs are being used mostly as a quick way to flag properties that might be under or overvalued. They can take in a massive amount of information and provide a suggested valuation for hundreds or even thousands of properties in a very short time. This speed could help some firms stay ahead of the competition, either by making acquisitions of a property before others see its potential or selling a property before a market downturn. “Investment decisions are going to be made a lot faster thanks to AVMs,” said Fisher.
Like any technology, AVMs will only get better over time. Plenty of new technologies will emerge that will help computers become more accurate when assessing a property’s condition and market. As more historical data and contextual information are incorporated into the automated analysis, AVMs will start to get closer to replicating the current appraisal process. “Computer vision technology will eventually be used to analyze pictures of a property to help AVMs understand a location’s uniqueness and the condition of the building itself.
Artificial intelligence and machine learning techniques also hold the promise of improving commercial real estate valuations by finding patterns and insights in large datasets that human analysts may overlook. But, in order for these AI models to be truly effective, they require access to massive amounts of highly structured data covering all of the granular factors that can impact a property’s value – something the commercial real estate industry currently lacks at scale. Some markets might not have access to as much data as others, or there could be a significant time lag between when important new information arises and when the data gets incorporated into the model.
From specific lease terms to qualitative assessments of a building’s condition, there are many nuanced details that trained appraisers take into account but are extremely difficult to comprehensively translate into data formats machines can process. The finishes, amenities, views, and overall ‘pride of ownership’ a property exhibits can significantly influence its market value, but explicitly coding those characteristics into structured data is a major challenge. With this limited availability of detailed data as inputs, AI valuation models risk becoming overconfident in their predictions or overlooking key nuances human experts would identify, introducing potentially serious blind spots. To help combat this, JLL’s AVM comes with a confidence score that will help managers understand just how much they need to double-check the work done by the algorithms.
As Fisher stated, “I think there will always be a need for human involvement in the valuation process…We really believe in the popular theory that AI will replace tasks but not roles. Someone will still need to oversee the valuation role, but they will be able to outsource a lot of their tasks to AI.” While AI and machine learning can automate certain analyses at scale, human expertise remains critically important for validating the results and filling the gaps stemming from data quality limitations or unstated assumptions the models fail to capture.
Valuing a commercial property is not as straightforward as estimating the price for a residential one. That will never change. But that doesn’t mean that the AVM technology that has transformed the residential real estate landscape has no value in commercial real estate. While a human appraisal is still more accurate than one generated by an AVM, the gap is narrowing. As AVMs become more accurate and easier to use, savvy commercial property investors will look to them for a competitive edge in acquisitions and to better understand the properties they already own.