Robots – or, more accurately, computer-generated models – will increasingly replace humans in conducting home appraisals according to recent reporting from the Wall Street Journal. The article quotes Jeremy Sicklick, the CEO of appraisal technology company HouseCanary, who said, “You’re seeing the beginnings of the machines outperforming humans in terms of accuracy.”
While the promise of automatic appraisals may be alluring, there are many drawbacks of this technology becoming the primary valuation method. And HouseCanary’s recent actions show there’s more than meets the eye.
Jeremy Sicklick raised eyebrows in July when he asserted at an industry conference that “probably about 70 percent of properties and transactions will be evaluated in an automated way.” Among the problems that this trend could pose includes the loss of an appraiser’s judgment informed by human senses and years of experience, increased risk for investors due to weakened credit quality of mortgage-backed securities, and the challenge of squeezing highly differentiated properties into a one-size-fits-all model.
Indeed, Sicklick acknowledged that distinct properties are not quite ideal for models designed by his company. But those issues aside, the story of HouseCanary raises an entirely different host of issues about automated valuation models – particularly if the producers of these models fail to adhere to ethical standards.
When Sicklick informed the industry audience that 70 percent of appraisals would be model-generated, a co-panelist rejoined that companies like HouseCanary are simply making money off the data produced free of charge by agents – real-life humans in the field. However, HouseCanary’s recent product launch expanded this profiting off data produced by others to also include the profiting off data owned by others.
HouseCanary’s launch of ComeHome, an analytics platform for residential properties, was immediately met with anger from multiple listing service executives who discovered their data to be stolen. ComeHome had incorporated the private data owned by these numerous listing services without properly licensing the information. After enough executives complained about the rip-off, HouseCanary finally deactivated all its property listings on the service.
Automated valuation models raise questions about the proper use of the data involved. But more concerning is efforts to deliberately mislead consumers by painting over the flaws of the technology – a strategy that HouseCanary has opted to pursue according to former employees.
Appalled by the company’s willingness to misrepresent its capabilities, several former executives at HouseCanary stepped forward to reveal that the company was selling sham products. HouseCanary’s former director of appraisal experience said that the company produced “wire frames and apps that didn’t function,” developed an automated valuation model that was an “overlay” of a competitor’s model, and “lied” to clients to conceal this information.
HouseCanary even offered high-paying consulting agreements with “no specifically-defined scope of work, nor any minimum or maximum number of hours per month” to former employees to keep them mum about the company’s inadequate appraisal technology.
This is only known thanks to the emergence of several high-ranking whistleblowers. It serves to reason that additional red flags about the valuation technology could exist that haven’t yet been uncovered.
As one of Sicklick’s panelists pointed out at the conference, automated valuation models are not a magical formula (despite HouseCanary’s characterization of its own as a “magic machine”). For the foreseeable future, and as HouseCanary has proven, people should be wary of over-relying on this technology. There have been too many problems – and deliberate efforts to obscure their existence – to buy what’s being sold.