Putting an accurate value on a building is core to the real estate industry – and machine learning and predictive analytics tools that leverage the power of big data are increasingly becoming part of the process.
An Automated Valuation Model (AVM) is an algorithm-based computer program which provides estimates of a real estate property market value at a specific point in time, whether now or in the future.
They’re becoming more prominent across real estate as machine learning modelling techniques get more sophisticated and the sheer volume of data on variables ranging from property characteristics to consumer behavior amplifies the need for consolidation and analysis.
For investors and owners buying and selling buildings, these new forecasting tools allow for a better understanding of the drivers behind the headline figure, which in turn enables them to make more informed decisions.
AVMs are already a common tool in the US, Canada, Australia and several European countries such as the UK – and their use is growing.
At present they’re a feature of the residential market, mainly because the sector is more liquid and transparent than its commercial counterpart and has larger transaction volumes that help to generate more accurate estimates.
Yet that is changing as institutional investors start to leverage AVMs to assess risk and opportunities within the commercial real estate industry – whether for multifamily developments or new multi-use projects. Strategic decisions around site selection and portfolio optimization are increasingly driven by the information from computer-based modelling.
Elsewhere, AVMs are also helping to evaluate existing loans, assess refinancing deals and perform transaction valuation in loan negotiations.
However, even as they become more widespread, and deliver data-driven results, AVMs remain a controversial topic within the industry.
"There is no consensus around the use of AVMs to get a thorough property value assessment. Many real estate professionals reckon that AVMs need to be audited in order to determine how reliable and accurate the model outputs are. Another critical aspect is the quality and transparency of data and the specifics of the area to which the AVM is applied."
Claire Leblanc from JLL
AVMs are nevertheless evolving quickly as the mix and the quantity of available data expands exponentially.
"They’re likely to get even more mainstream along with the advancements in data mining and analytics. The potential of AVMs lies in their ability to determine the future value of an asset given different market conditions by simply changing assumptions and varying estimates. This could lead to predictive forecasts for all types of real estate and dynamic property valuations, which would be an invaluable tool to all parties in real estate transactions."
Tanguy Quero, Head of Online Business at JLL France
And as technology advances, the AVMs of the future may be able to develop an investment strategy instead of just determining the outputs of a given one.
“Many startups are driving innovation within this fascinating field. JLL has started that journey with the successful launch of AVMs in Spain and Finland. These models have huge potential for the real estate industry by bringing together man and machine to give greater clarity and confidence to buying and selling decisions for owners and investors in Europe and beyond.”
Charles Boudet, CEO at JLL France