A Technological Toolkit for Advances in Real Estate

In early 2023, UCLA Anderson Assistant Professor of Finance Gregor Schubert answered questions about his popular property technology course, Real Estate Trends: Data Analytics and Proptech. His goal — now as it was back then — is to provide his students with an understanding of the latest technological trends impacting real estate markets and executives in the real estate space. He said at the time that the class gives students “an intellectual toolkit with which they can evaluate new technologies they encounter, figure out how those might be affecting real estate markets and cities, and then form their own opinions.”

But in the proptech space, two-and-a-half-years might as well be a millennium. It’s fair to say 2025’s tech trends were perhaps barely on the drawing board in early 2023. Advances and applications in artificial intelligence have changed the game, and so Schubert has evolved his curriculum. It’s a telling example of how UCLA Anderson faculty imbue their courses with both the necessary fundamentals as well as the most up-to-date advances in technology and research.

Schubert’s research encompasses housing finance, urban economics, real estate, labor economics and corporate finance. Most recently, he’s focusing on how urban migration networks affect housing markets. During national boom and bust cycles, he argues, the systematic differences in the degree to which geographic housing markets are affected by common economic shocks can be explained by patterns of migration between U.S. metropolitan areas.

When Schubert joined the Anderson faculty in 2021, he was given free rein regarding the kind of course he wanted to teach. He created a course that responds to some of the big questions in the real estate industry, and how technology and data analytics address those issues. “I looked at what was going on in real time and I structured the course around what I read in the news,” Schubert says. His proptech course is available to students in both the full-time and fully employed MBA programs.

As an undergraduate at Princeton, Schubert minored in theater, and he carries his creative passions into his academic sensibility. If classroom teaching is a type of performance, Schubert hopes to inspire the student audience to think about housing markets in terms of how we choose where we live and why.

What follows is an updated version of the interview conducted in 2023.

Q. Since we published this story two-and-a-half years ago, the ubiquity of artificial intelligence has permeated business, society and the Anderson curriculum. Could you describe how you’ve incorporated AI into this class and how you keep the course material current during a time of technological evolution?

Since I first taught this class, I have added substantial material around the impact of AI on real estate. In fact, one of the exercises in the class is now to build a chatbot that can take over one of the tasks that a real estate agent or other real estate professional might face. I have also added more material to make sure I show the trajectory of how earlier generations of deep learning and AI technology were adopted by real estate firms, as it helps to understand the trade-offs between different AI tools. It does take a lot of updating every year to make sure the references to relevant AI technologies stay current. However, I try to emphasize insights and principles that will guide student decision-making in general when interacting with AI technologies and are therefore useful and relevant, no matter what the latest model release is.

Q. What impact has AI had on real estate markets?

Some of the impact on real estate is the same effect as on the economy in general: marketing, back-office document processing and programming for software applications in real estate all benefit from generative AI making those tasks faster. However, there are specific areas where real estate firms are deploying AI tools more or differently. Real estate platforms have a complicated business that requires matching users to properties such as Airbnb or Zillow, but for a transaction that happens relatively rarely; so the ability to use deep learning and AI to extract preferences from limited interactions, and use image data or click patterns to predict user preferences, is a key part of making these platforms useful. There is also a lot of image data and other large data sets that go into site selection for construction, tracking progress for new development and then interpreting output from building sensors in new buildings. More traditional deep learning and predictive AI are increasingly deployed in these settings in real estate.

On the other hand, real estate is a people business in many ways. In residential real estate, brokers spend their days following up on leads, guiding buyers, advising on negotiations — but many aspects of that interaction are repetitive and standardized from the perspective of the broker. Similarly, property management often involves standard processes, such as routing maintenance requests to contractors, responding to inquiries by potential tenants, sending reminders and updates to existing tenants. Generative AI’s ability to automate these text-based workflows is really obvious to many people working in these areas, and so we have seen a panoply of startups with different ideas for how to use AI to automate these approaches.

Q. When we update the story about your proptech course in another two years, what might be the innovation we’re talking about?

At the risk of sounding like a fool two years from now, I think we will see further progress in deploying robotics and autonomous vehicles in buildings and cities. And how to integrate these innovations into the built environment will be an increasing challenge, as buildings will have to adapt by providing sensors, charging stations and spaces (robot elevators?) that make this deployment feasible. However, in terms of real-world implementation, incorporating even the current generation of AI capabilities into the real world takes time. So, I think we will still be talking about the ongoing progress of generative AI in making firms more productive in two years. The progress in large language models is not stopping.

Q: What is the goal of the proptech course?

The goal of the class continues to be to give students an insight into how current technological trends are affecting real estate markets and the practice of real estate. It gives them an intellectual toolkit with which they can evaluate new technologies they encounter, figure out how those might be affecting real estate markets and cities, and then form their own opinions.

For instance, there is a growing number of startups in property technology. As professionals, these students will need to evaluate their economics, the potential for their technology and what the implications are for the pricing of different real estate sectors and different geographies. How might different neighborhoods or cities change in price as a result of, for instance, remote work technology? Or how would widespread prefabricated housing affect different parts of cities? I’m trying to guide the students and prepare them for when they encounter these new trends and how they would be affecting real estate.

Q: UCLA Anderson’s curriculum, in many areas, reflects the evolution of data analysis over the last decade or more. Could your proptech course have even been taught 10 or 20 years ago? Was the course in any way created in response to student interest as it relates to that evolution?

First off, I think this class would have looked very different 20 years ago. Because of the overarching trends toward broader availability of high computing power and data storage, it is now common to access large data sets online and do pretty complex computations on our home computers in a way that 20 years ago would not have been easily feasible. This trend has impacted the type of business models that are feasible and the way a lot of real estate companies operate.

Now it’s pretty common practice that real estate companies will hire analysts who understand how to work with large data sets. They will try to extract information in the form of automated valuation models and use machine learning to put some valuation on properties instantaneously — in a way that goes beyond old-school methods, like an appraiser’s manually trying to figure out what something is worth.

These techniques are now basically part of a standard tool kit. Almost any real estate company that has any interaction with technology — Zillow, Airbnb, Opendoor, BlackRock — they all have large staffs working with these large data sets and seeking to translate data into actionable insights.

And because the tracking of real estate data has changed so much over the last 10 to 20 years, it’s now essential that students have some understanding of how these things work. Because, in their professional lives, they will encounter lots of companies that throw around words like automated valuation models, machine learning and natural language processing based on this tech. They need to have a strategic understanding of how these tools are used, what is good about them and bad about them, and the potential pitfalls of how they work.

I think as a result of that, there’s demand on the student side to acquire new skills and learn to be knowledgeable about them.

Q: Are there any prerequisites or prior knowledge expected of the students before they take this course?

The classes explicitly target students who are not coming in with a programming background, nor a strong background in engineering or data science, but rather are interested in the topic from the perspective of a manager who will be interacting with data scientists or analysts who are running these analyses and who will report back to them. As a result, the only background required is a basic familiarity with introductory economics and statistics concepts.

Q: Do you bring your own research and scholarship into the class? What about case studies?

I think you need some sort of intellectual tool kit that comes from the academic side. We use demand and supply analysis to understand the equilibrium of market forces that bring about real estate prices. For that, I bring in some of my own research, which is mostly concerned with how housing markets react to shocks.

I try to complement that with real-world examples of companies that are actively using these tools or are encountering the kinds of technological shifts we’re talking about in the class. Every class has sort of a balance between theoretical insights about how the world works, and some empirical examples of current trends the students should know about. Then, usually a third of the class is reserved for a case study where we talk about well-known companies that are using technology in their real estate investing. I also bring in guest speakers, usually founders or executives at companies that are in the property technology space, to allow the students to interact with someone actually using these tools on a daily basis and who can tell them from experience how they will be useful.

Q: Did you learn anything from the first time you taught it that shaped the second iteration?

Because these are new trends, we don’t really have existing textbooks or standardized theories. I tried to fill in the gaps, for myself and then for my students, about how one would respond systematically to the big issues of our time that are in the news and seem to be shifting real estate. I basically came up with a list of these “big topics,” things like climate risk, remote work, the rise of data science and analytics in real estate practice. Then for each of those topics, I structured a module based on my academic background, case studies and the insights of some leading practitioners and other scholars.

One takeaway for me from the first time teaching this class was that some topics that are of academic research interest — such as what happened during the great housing boom and bust cycle of the 2000s — are important to an academic audience, but are not particularly relevant to my students or to current trends. I also realized that a lot of the cases we were covering in class, and a lot of the discussions among the students, seemed to come back to the use of automated valuation models and, in particular, machine learning. In the second iteration of the course, I have therefore added a whole week on that topic.

Q: Students are always interested to know what they have to do to pass the class. What are the basic requirements?

There are three main components. There are empirical exercises in which student groups work together using data and try to tackle some business issues with these new data sets and try to get some insights into when particular technologies for data analysis work well and when they don’t.

The second component is a final exam that is very similar to a case discussion. I don’t want the students to come out of the class memorizing facts about the world. As a professor, you insist that they become critical thinkers about technological trends. So the exam tests them on evaluating a new property technology startup or an interesting news story about property technology.

Then the third component is class participation.

Q: Many professors mention class participation as part of their student evaluations. Why is it so important?

I think the class wouldn’t be as exciting if it were just three hours of listening to my voice.

What is really exciting is the students’ coming in with their own backgrounds and experiences and bringing them into the discussion. They’ve often read and seen things in their professional practice that make really enriching contributions to what we’re talking about in class. They may have experience either directly or indirectly with people working with these new technologies.

That experience brings new perspectives, especially given that these are all new and emerging trends that we’re talking about. There is no settled wisdom on what these trends are doing, and it is actually really important for us to talk to each other and form the opinion as a group, because ultimately that’s probably some of the best information we can get.