Technology has transformed our day to day lives and one of the most mind-boggling aspects is the amount of data we consume and generate each day. In 2012, the world entered the Zettabyte Era (that figure is 1 with 21 zeros, amount of data generated to date) and it is projected to reach 175 Zettabytes by 2025 (IDC report “Data Age 2025”). To put that data generation in context, in just two days in 2010, the world created as much data as mankind created from the dawn of civilization until 2003. The amount of data being generated is not only “big”, but is also happening at an incredible speed which results in approximately 90% of the data being “unstructured”.
At this scale, the information is not in the form of simple relational data tables and thus requires a totally different approach leveraging big data technologies. These technologies are data processing tools, methods and approaches that help extract meaningful insights from this unstructured and non-traditional data points akin to extracting metal out of ore. This wealth of data has led organisations across various industries to use these technologies to make critical business decisions, including the Real Estate industry, that has, for too long, utilised an antiquated combination of intuition and traditional, retrospective data to make decisions. The transformation in this industry can be clearly highlighted in the rapid adoption of automated valuation models (“AVMs”) for property appraisals across the globe.
An AVM is a program that automatically analyzes mammoth amounts of varied data points to estimate the current value of the property. AVMs typically use advanced techniques, such as machine learning models, to scrutinise different data points for a given property while taking into consideration complex non-linear relationships from traditional and non-traditional data sources that would be almost impossible for a human brain to handle (thanks to the unlimited cloud based computing power). New data is continually being updated, often intra-day. The algorithms are also updated to reflect changes in market trends. A McKinsey study from 2018 on the same matter concluded that almost 60% of predictive power in real estate can come from non traditional variables (eg proximity to hotels, cafés in the area etc) rather than traditional variables (market performance, property performance and features). Similarly, a study conducted by Yuvoh Analytics using its AVM models concluded that the density and quality of holiday home properties can be one of the top 5 drivers of property values in select regions of the Greek real estate market.
Even though AVMs use a large amount of data and cutting edge technology to make predictions, the real advantage of AVMs are speed and cost. It is incredibly time consuming (and hence expensive) for lenders and investors to get a manual appraisal using the traditional approach and often not realistically feasible when valuing a significant number of properties at once. AVMs can be the clear solution for such cases.
Thus, AVMs are a valuable tool for banks and investors. They can benefit from its use in a wide range of important and labour/cost intensive processes – bulk revaluations of non-performing assets, underwriting or audit of new origination loans, risk management, regulatory reporting, negotiation of delinquent assets or key management information packs for decision making. The innovation in big data and artificial intelligence is the way forward and firms must consider these tools as integral to their current underwriting, portfolio review and research processes. Early adopters of this technology can bring about transformational changes to their internal processes and cost structures and stand to gain significant information advantage over their peers in all areas of the real estate market.