AVMs may greatly increase risk in residential property space

Automated valuation models are altering the landscape of residential property valuations by increasing the amount of automation whilst lessening the need for chartered surveyors to provide as much input through the process.

While this may seem a natural progression, as machine learning technologies improve their ability to carry out an ever-increasing number of complex tasks competently, businesses remain concerned about the quality of output AVMs provide. Despite growing acceptance of these technologies in the residential property space (banks are among the most enthusiastic adopters), in some cases, the need for human judgement is indispensable.

Although there are clear benefits to using AVMs as part of residential property valuations, they may not always provide a clear enough picture of the risks involved in any given transaction.

Besides the most obvious benefits of slashing the costs required for both lenders and buyers and cutting down on the manpower needed for a high volume of transactions, in theory, AVMs also provide sound, unbiased data in which all of the parties involved in a transaction can trust.

However, a valuation based solely on mined data or statistics lacks the judgement of a seasoned professional and relying on AVMs to do the job alone could be riskier than realised. Instead, a better purpose may be to use them as a ‘sense check’, a hybridised approach in which a surveyor utilises AVMs as a supplementary benchmarking tool to accelerate his tasks.

The increased use of AVMs, overall, will make them more accurate in the long run, as the growth and availability of Big Data helps residential property professionals pinpoint cycles and detect trends. Indeed, proponents of AVMs argue that AVMs have been stress-tested and re-calibrated over time to create optimised solutions. For others, there is a lack of transparency that may affect the accuracy of the models used.

In essence, AVMs are “black boxes” from the user perspective, based on a trove of data which has been fed into the algorithm and then recalibrated based on that input. For this reason, there is no real understanding of which data the machine is relying on to draw its conclusions, just how the algorithms work, and even the reliability of the results. Likewise, it may be unclear what the margin of error is. In addition, because the models are non-standardised, it can be hard to compare between providers as a large variety of models exist.

Additionally, many European AVM providers remain reluctant to share their methodologies. This lack of transparency is further complicated by the fact that no legislation exists to regulate AVMs, so there is no avenue to examine a given methodology for determining its accuracy.

It is also debatable whether algorithms can ever be a match for professional knowledge. After all, individual experts possess an unrivalled degree of insight about local markets that is simply unachievable at this time for AVMs.

In some areas, AVMs could pose an even greater risk. For example in bridging loans, an area which has seen a substantial increase of specialist providers springing up to fill the gaps in the banking ecosystem, as banks increasingly exit these markets due to higher risk. AVMs are a boon to these companies, providing a competitive advantage as a result of their ability to process property valuations speedily and at a low cost.

Employing AVMs, for more complex bridging loans, could drastically increase risk. As valuation fees plummet, thanks to the cost advantage AVMs create, a ‘race to the bottom’ could emerge. This would place pressure on valuers’ margins and transaction times, thereby increasing risk.

Regardless, liability will ultimately rest with surveyors, in the same way it would have had AVMs not been used in the first place. However, if surveyors are instructed to use AVMs by banks, it is unlikely that a court would deem that a chartered surveyor’s duty of care had diminished if automation was deployed to reduce costs, even if cost savings were passed on through lower fees.

While the use of AVMs by surveyors is recognised as being an area for continued growth, surveyors will always be ‘on the hook’ for an incorrect valuation, even if the liability really lies with the AVM provider. To minimise exposure to potential claims, attention must be paid to the terms of engagement with clients and the basis of valuation method used needs to be made clear.

As house price growth starts to decelerate and further interest rate rises become a distinct possibility – a combination which has historically been the key cause of negligence claims against surveyors – AVMs should be used with greater care.

Adequate professional indemnity cover, crystal clear client terms is essential to ensure that potential lines of recovery from the provider remain open in the event of negligence on their part. Broad professional indemnity cover is available for these risks, with few exclusions.

While technology will undoubtedly become increasingly advanced, making AVMs more accurate and reliable, is important to remember that they are not a replacement for professional expertise and should be thought of as a tool to enhance and support that expertise.