Automatic Valuation Models: Art or Science?By Mario Gamboa | Tue, 07/05/2022 - 16:00
For a long time, there has been a debate in the real estate industry that attempts to answer the following question: Is property valuation an art or a science?
Some people argue that property valuation is an art since the value is based, in part, on a personal assessment of value. On the other hand, there are those who are convinced that property valuation is a science because it is based on methodologies that use quantifiable attributes and hard data. Over time and the evolution of new technologies, mathematical models called Automatic Valuation Models (AVM) were developed. These could be placed somewhere in between.
AVMs are tools that use statistical modeling techniques to assess properties and employ artificial intelligence and machine learning techniques to determine the fair market value of a property, based on its observable attributes and synthesizing information about comparable properties. AVMs have multiple-use cases; the most common include determining the value of a real estate portfolio, establishing prices and marketing strategies, tracking the evolution of market dynamics, and refreshing the valuations of entire mortgage portfolios. This technology has helped the real estate industry achieve much more accurate results, free from value judgments and personal criteria.
AVMs tackle various problems endemic to the real estate industry. For example, it is not uncommon to experience delays in the receipt of an appraisal, typically about three to four weeks. This delay can be reduced by using modern valuation methods, such as AVMs, as less manual effort is required. It is important to mention that these methods are designed to support professional appraisers; it is, in any case, an extra validation.
The popularity of this technology is growing. In the US, AVMs have been in use for more than 20 years and there are very well-established companies in the sector. Southeast Asian economies are using these technologies as well and, although the same concept applies, the calibration of each model varies depending on the region in which it is used. Countries, including Denmark, Canada, Sweden, South Africa, and Australia, also use these types of models. The Canadian government, for example, has created a database so accurate that they can sell the information to private companies upon request.
The European AVM Alliance, led by private companies from the UK, the Netherlands and Switzerland, promotes a consistent approach to automated valuations for efficient and seamless mortgage lending. The group is very active in educating key audiences on this technology.
One example of how AVMs improve the way data is used for valuation is House Canary, a leading US real estate valuation firm, which conducted statistical studies to measure the accuracy of its automated valuation models. They discovered that when sufficient data is available, its estimates are extremely reliable, thereby providing an objective market valuation, due to the mitigation of personal value judgments.
Intelimétrica’s AVM model has the largest database available in the Mexican real estate market, allowing it to calibrate its models efficiently and constantly. Its proprietary algorithms were originally built from several systems and hundreds of “stacked models,” which in turn are supported by other models that have been programmed to decide autonomously which model is the best for a given use case. Because of that, our models are able to appraise, in a short period of time, more than 150,000 houses located in different cities around the country.
Real estate investment funds exploit AVMs developed by Intelimétrica to help them make business decisions in the housing market that provide the best return on investment. For example, the decision to purchase a housing project of 1,000 houses located in different cities is accelerated when it is possible to value, efficiently and in a short period of time, all properties through the automatic valuation models, instead of using a traditional method in which each one of the houses is inspected.
Banks and nonbank lenders use Intelimétrica’s AVMs to value properties on which they grant mortgage loans to their clients. Professional and independent real estate brokers use this service through the Yals.com portal to understand the values of the properties they commercialize.
Automated Valuation Models have been gaining traction in Mexico and in some Latin American countries. Companies that already rely on advanced digitization models are preferring them over traditional ones. The advantages of these models for researching and accurately evaluating real estate portfolios in short periods of time are the key factors why they are chosen.
The organizational culture of companies that already use AVMs is focused on the effective use of technology and the importance of the intelligent use of data. It is worth noting the commitment of C-level executives to the paradigm shift favoring the use of such models and the ongoing internal communication efforts made in this direction.
The speed, accuracy, ease of auditing, and objectivity provide Intelimétrica’s Automatic Valuation Models with a clear competitive advantage. The models are developed with automated protocols and algorithms that minimize biased valuations, which are common in the traditional approach.
Proptech tools and data are at the core of real estate business strategies. The industry accelerated its transformation during the pandemic and new valuation methods were developed, incorporating technology, to facilitate and accelerate transactions for the benefit of all those involved in the value chain.