This Is How One Company Has Been Using AI To Shake Up Residential Leasing

2023 is predicted to be the year that the use of artificial intelligence will skyrocket. The global AI market was valued at $136.55B in 2022 and is expected to increase at a compound annual growth rate of 37.3% from now until 2030. From ChatGPT to self-driving cars, AI is entering all aspects of people’s lives. 

But where is AI in multifamily? The majority of residential landlords and operators have yet to embrace the power of data and AI and its potential to reduce business risk and uncover opportunities. 

This is the message from Elsa Liao, vice president of risk management at fintech company TheGuarantors. AI underlies the value that the company brings to both renters and landlords by underwriting a tenant’s lease. By crunching huge datasets, the company can more accurately predict renter default, thereby better serving both renters and landlords.

The result is that renters are better able to secure a home, while landlords have access to a larger pool of potential renters without the risk of rental income loss, Liao said. Bisnow spoke to Liao about exactly how TheGuarantors' AI model works, and what the future of AI could bring. 

Bisnow: How do you use AI at TheGuarantors?

Liao: We use several AI models for different use cases such as underwriting and pricing. Our underwriting model is what we use to predict the risk of every single renter application and decide if we approve them for an insurance policy.

This model uses two buckets of information created from more than 1,000 data points. On the renter side, we look at factors such as income, liquid assets, credit score, the details in their credit report, how expensive their potential rent is and whether they have roommates. It goes to a very granular level.

The second bucket looks at everything that describes the building’s physical environment: how old it is, the size, the distance to the nearest college and hospital, median house price, and other macroeconomic and multifamily housing trends.

There is a vast amount of information. For example, we’ve seen an inverse correlation between renter default rates and renter satisfaction with their overall living experience, so we factor a number of variables that contribute to renter satisfaction into our underwriting.

Bisnow: How does this compare to the information that most landlords traditionally have to vet a potential renter?

Liao: Landlords use a much more limited set of information. They focus heavily on the credit score, income and, for some, eviction history. These are important factors in our model as well, but they don’t go into the finer details.

Within a credit report there are more than 200 data points, such as how long their history is and what their debt-to-income ratio might be. We use every predictive data point.

Bisnow: How can using this data to create an AI model help a landlord?

Liao: Landlords are currently missing out as they only have superficial information. Using AI will help them to achieve a higher accuracy with rental screening. 

Traditional screening does not do a good job of predicting who will default and who won’t. There are a lot of renters who look bad on paper for many reasons, such as not paying off a medical debt or having insufficient credit information because they are from another country. In a landlord’s eyes, they appear risky, but neither of these issues mean they will necessarily default. 

With only this information, landlords are missing out on many good renters and may therefore limit potential rental growth. At the same time, for renters, it is becoming increasingly difficult to rent a home because the system has too narrow a view. 

AI can do a much better job giving landlords a fuller picture of renters than traditional underwriting. We did a comparison of how effective a FICO credit score is in evaluating renter default risk, compared to using AI. FICO gave a 55% accuracy rate, whereas our model achieves 85% accuracy. 

Bisnow: Eighty-five percent may still not be a high enough accuracy rate for landlords to accept a renter; is this where TheGuarantors steps in?

Liao: We provide two areas of value to renters and landlords. Firstly, we give denied and conditionally approved applicants a second chance to get the apartment. This opens up the pool of people a landlord can rent an apartment to. 

Second, we give landlords more protection for all their renters. We eliminate bad debt by covering the renter if they are unable to pay.

Bisnow: Will AI be used more in real estate in the future?

Liao: As people see incremental improvements in applications over the next few years they will start to see the value of AI. They’ll start to collect data to use for modeling, which will bring forward an AI revolution on a much deeper scale. 

The challenge with AI is that an algorithm is only as good as your data. Right now, there’s a lot of data in real estate that is uncaptured or unutilized. People don’t understand the value of it. They’re financially and operationally savvy, but they’re not savvy in a machine learning sense. 

Bisnow: How else is TheGuarantors looking to use AI?

Liao: As well as underwriting, we also use AI to optimize our customer service resources and plan our staffing accordingly. 

There are so many applications for AI in real estate. Soon, I expect someone will develop an algorithm that can recommend apartments or houses to people using more than location, layout and budget just as how Spotify recommends music based on what you and millions of others listen to. The underlying machine learning logic is the same. 

This article was produced in collaboration between TheGuarantors and Studio B. Bisnow news staff was not involved in the production of this content.

Studio B is Bisnow’s in-house content and design studio. To learn more about how Studio B can help your team, reach out to [email protected].