What is an Automated Valuation Model (AVM)?

What is an automated valuation model (AVM)?

An automated valuation model (AVM) is a sophisticated program designed to estimate the value of real estate property without the need for human input. AVMs use algorithms and statistical techniques to analyze various factors, including the property’s characteristics, comparable sales values and local market trends to produce an accurate property valuation. AVMs are mainly used as a tool to help automate the appraisal process or to assist appraisers in determining home value. Additional use cases include lead generation, loan application, collateral review and more.

How does an automated valuation model work?

The fundamental objective of automated valuation models is to provide fast, more reliable, objective, transparent and cost-effective estimates of a property’s current market value. AVMs accomplish this by combining multiple property-related data sets with mathematical or statistical models to automatically determine an appropriate value range.

AVMs leverage property data – at minimum, the year built, lot size, size and characteristics of the living area, and number of bedrooms and bathrooms – and sales data to compare the subject home being evaluated to other homes on the market. The comparable sales data characteristics include property similarity, location and recent property sales data. An AVM also leverages market trends such as recent home price changes in the area, current market condition, including how the market is trending in terms of supply, demand and inventory.

The AVM calculates this data to then provide a determination of the home value and a confidence score. The confidence score can determine how accurate the valuation is, effectively notifying the user if it is an appropriate valuation of the subject property and the confidence level associated with the value.

Why are AVMs used?

Unlike other similarly large expense items, no two homes are alike. This fundamental idea presents potential inconsistencies when it comes to home valuations, as buyers, sellers and lenders have differing motivations, leading to immense subjectivity when it comes to deciphering how much a property is worth.

This pain point in home valuation combined with advancements in technology and greater access to data, lead to an increased need to ensure unbiased, objective and accurate property valuations. As a result, AVMs became increasingly more available on the market. Over the coming years, the idea of using AVMs has become more widely accepted and adopted, leading to federal regulators finding the need to closely examine them for accuracy and adherence to set standards.

What are the regulations for AVMs?

To produce an accurate property valuation via AVMs, it’s important to have a transparent, objective and credible appraisal valuation process to establish a property value that all parties – lenders, real estate agents, borrowers, brokers and others – can agree on. Thus, it’s critical that a system be in place to challenge and test these assessments when necessary.

This notion eventually led to the Dodd-Frank Act, which amended the Financial Institutions Reform, Recovery and Enforcement Act of 1989 (FIRREA) to establish and codify four AVM quality-control standards which include:

The Dodd-Frank Wall Street Reform and Consumer Protection Act, commonly known as the Dodd-Frank Act, was enacted on July 21, 2010, in response to the financial crisis of 2007-2008. The act aimed to help provide financial stability, increase transparency in how the AVM was produced for all parties involved and protect consumers from financial practices that abused their power during this unfortunate market circumstance.

What is the difference between AVMs and traditional home appraisals?

AVMs, appraisals and Broker Price Opinions (BPO) all begin with public records and/or MLS data to understand property characteristics. All three methods also pull data from a “general market” where homes compete with one another in their neighborhood or surrounding area.

The difference between the three comes from the method in which they gather the comparable selection for each property. Appraisals require an individual to narrow down the pool of comparable sales through application of filters, individual analysis of comparables and review of MLS data and photos. This tedious manual process usually nets around three to six valid comparables which are used in determining the final home value.

An AVM will leverage the comparables dataset to filter outliers and calculate the correlation of the property sale price to its gross living area (GLA). At this point in the process, the amount of comparables may result in anywhere from 20 to 50 homes matching the decided-upon parameters. Then, the model begins to make further adjustments accordingly to narrow down the right comparables.

All three methods provide a reasonable value when properties are similar and comparable data is readily available.

Who uses AVMs?

Lenders, as well as mortgage, capital market and real estate professionals use AVMs for numerous reasons:

What are AVMs used for?

AVMs can be utilized as a valuation tool throughout the loan process to streamline workflow in addition to traditional use cases. Lenders, mortgage, capital market and real estate professionals may use AVMs for:

How are digital valuations and AVMs evolving?

AVMs are enhancing their solution capabilities with real-time, UPS-tracked photos taken at the subject property or by leveraging MLS photos which utilize computer vision to automatically analyze the condition and quality of the property. The AVM is then able to fine tune its results to produce a condition-adjusted AVM and more accurate confidence score. These capabilities help secure an impartial home valuation utilizing more reliable and objective input data.

ICE offers a mobile digital valuation solution, Validate, that combines computer-vision technology, a condition-adjusted AVM and up-to-date property data with borrower-supplied photographs to automatically determine a property’s value and the available equity.

How does ICE measure AVM accuracy?

ICE AVMs are statistical models built to provide objective and transparent property valuations that help support AVM regulations and our own model accuracy measuring standards. Our AVMs are compared to the actual benchmarked recent sales data, including public record and MLS, to produce a forecasted standard deviation (FSD) and a confidence score. We also look at the market from a portfolio and local geographic level to determine which percentage of valuations are within 10% of the benchmark value, which measure as a PPE10 (Precent Predicted Error) metric.

ICE metrics for AVM accuracy:

These model metrics and practices, along with our five valuation pillars, help to secure objective, transparent and accurate AVMs. These pillars are:

What are the benefits of using best-in-class AVMs?

Using the right AVM solutions can secure valuations that are independent of any bias, and are transparent, authentic and credible, helping to provide users with the most accurate comparables and output. This is accomplished through key technological capabilities that provide cost-effective and time-saving benefits, including:

ICE AVMs combine one of the largest, most expansive data sets on the market with advanced modeling techniques, rigorous testing and high-performance technology to provide a single source of AVMs with some of the lowest percentage of outliers in the industry. Learn more about how our AVMs can help solve your valuation needs while delivering exceptional accuracy, hit rates and reliability.