Automated valuation models (AVMs) are a cost-and-time-efficient solution for valuing lower-risk properties in home equity lending, portfolio valuation, pre-valuation decisions, and more. While AVMs aren’t the right tool for every property, there’s been a definitive shift in the industry towards more AVM use cases and their acceptance as a rigorous valuation methodology. The expectation is for this trend to continue.
It’s important to ask the right questions to ensure you’re getting the most from your AVM provider. Simply put, you want to know if you’re working with a marketing-grade or lending-grade AVM. Once you know how the AVM is engineered and how it will perform in a rapidly-changing market, you can make an informed decision on whether the AVM can provide the degree of accuracy you need.
Financial institutions should look for a highly-accurate, lending-grade AVM that combines quality data, valuation expertise, and machine learning. Additionally, it’s beneficial to test the AVM yourself to ensure the provider is not leveraging loopholes to game the test.
Known for their speed, accuracy, and reliability, lending-grade AVMs are prevalently used in property valuations. While many situations still require the input of a highly-skilled appraiser, AVMs offer another tool in the valuation toolbelt when the situation is appropriate. If an AVM fits into your lending workflow, we recommend asking these questions to ensure you’re working with a lending-grade AVM.
1. Does the AVM combine quality data, valuation expertise, and machine learning?
The combination of all three creates a highly-accurate, lending-grade AVM. Marketing-grade AVMs often don’t employ all three of these components, and many AVMs claiming to be “lending-grade” are still missing components of the formula. Ask your provider if the AVM utilizes these key factors in its model. If it doesn’t, you could be working with a marketing-grade AVM — these simply don’t provide the level of accuracy or expanse of property data that lenders need to make smart investments.
2. Does the AVM use data from multiple listing services (MLS)? If yes, how much coverage does it have?
An AVM is only as good as the data it’s built upon. Simply having a large amount of data isn’t enough, or even ideal. What matters more is the accuracy and timeliness of that data. The MLS is widely considered the gold standard for current, accurate, up-to-date property information.
A highly-accurate, lending-grade AVM should be built upon a comprehensive and high-coverage set of ethically-sourced MLS information, in addition to public records and other sources. Clear Capital’s AVM has 93% coverage of the United States and is widely considered one of the most robust AVMs on the market.
3. Does the AVM employ a data governance program to select the most reliable and accurate data for each property?
By doing so, in the event of data mismatch or inconsistency, the AVM can select the most reliable and accurate data source for each property.
4. Is the AVM built by experts with an existing and comprehensive understanding of the best way to complete a property valuation?
While an AVM can quickly process an immense amount of data much faster and more accurately than a human, it can only be as good as the human minds who built it. The best AVMs are built by humans with an existing and comprehensive understanding of the best way to complete a property valuation. Lending-grade AVMs are informed by appraisers or companies with “boots on the ground” valuation experience and many years in the industry. To account for market cycles, the more years of experience the better.
5. Does the AVM prevent and mitigate bias in its valuations?
It’s important to know if your AVM uses any factors that contribute to appraisal bias in its model. For example, an AVM shouldn’t employ any data that’s representative of or a proxy for race, color, national origin, religion, sex, familal status, or disability to determine a property value. In addition to controlling data inputs, an AVM should have continuous oversight to ensure ongoing compliance with Fair Lending Laws.
6. Does the AVM use the sale price, contract price, or refinance appraisal as a benchmark for testing?
While there are pros and cons for each of these benchmarks, the ability to game the test results is the biggest concern. Most AVMs have access to sale prices, which allows for gaming of the test. Contract prices are more difficult for an AVM to game, since most AVM providers do not have access to these benchmarks and they’re not usually reflected in the MLS. Refinance appraisals are not easily gamed, as these properties are not typically for sale, so the AVM does not know what they are listed for, what they will sell for, or any MLS information. Most AVM providers do not have access to these appraised values, so their model is forced to be blind to the benchmark.
7. Does the AVM provide Mean Absolute Error (MAE) or Median Absolute Error (MdAE) when reporting absolute error?
The MAE and MdAE provide insight into the model’s accuracy that other error metrics may not communicate. MAE is determined by calculating the percent variance between each AVM and benchmark, taking the absolute value of each, then averaging them over the whole test/sample set of benchmarks. In other words, it accounts for the outlier predictions, which are important when judging an AVM for day-to-day use.
MdAE uses the median (instead of the mean) so it hides the instances where the AVM was really far off; i.e., the outliers. This is the most common measure for marketing-grade (not lending-grade) AVMs since it makes an AVM appear more accurate than it actually is.
8. How do they calculate forecast standard deviation (FSD)?
FSD (forecast standard deviation) is a statistical measure that scores the likeliness that a valuation is accurate. It’s important to understand how FSD is calculated to ensure the methodology aligns with your risk management practices.
Since every AVM vendor typically builds a proprietary model to produce the FSD statistic, the communication of this measure can vary between AVMs. Right now, there is no standardized way to produce an FSD in the valuation industry. This means a 0.07 FSD from one AVM can have a very different measure of accuracy or confidence from another AVM. This makes it difficult to use multiple AVMs together or to apply a framework for how to use AVMs in general.
9. What address standardization technique is used by the AVM provider?
A substantial number of “misses” — i.e., where the AVM cannot calculate a value estimate — are not the result of failures in the model itself, but rather a failure to properly identify the address within the AVM’s database. Addresses can be represented in many different ways, and even the slightest variations are confusing to an automated system. Knowing which address standardization technique is used by the AVM provider helps both the financial institution and the AVM provider ensure the best possible performance.
10. How often does the model get updated with fresh data?
An AVM is only as good as its data. Confirm if the AVM is updating fresh data automatically, in near-real-time, as this provides more accurate results.
Try a lending-grade AVM for free
Whether you’re originating, servicing, or refinancing loans, getting accurate valuations in a rapidly-changing housing market is crucial. This leading, lending-grade AVM uses the most comprehensive real estate data in the nation.
ClearAVM™ is an online valuation tool that combines field-gathered data on 139+ million properties nationwide with valuation expertise and machine learning to deliver valuations that are up-to-date with market trends. Intuitive and easy-to-use, ClearAVM delivers accurate reports in just a few seconds.
About Clear Capital
Clear Capital is a national real estate valuation technology company with a simple purpose: build confidence in real estate decisions to strengthen communities and improve lives. If you want to learn more about our services, reach out to our team!
Our goal is to provide the industry with a complete understanding of every U.S. property through our advanced, intuitive field valuation services and analytics tools, and improve appraisal workflows with our platform technologies.
