The Rise of Automated Property Valuations
This question was brought about by the launch of Opendoor in 2014. Opendoor is a company that values houses solely on data. The pace of digital innovation has increased exponentially over a relatively short period of time. The global datasphere is predicted to grow to 175 zettabytes by 2025, and spending on AI and cognitive technologies is expected to exceed $50 billion by 2021, according to the International Data Corporation (IDC). We need to find new ways to join different processes into one system.
As the iBuyer business model continues to grow and evolve, it is bringing to light the many different ways that home prices can be predicted – from traditional methods used by the lending industry to more modern methods like instant estimates on public-facing portals. This growth is providing more liquidity and certainty to an increasingly on-demand economy.
There are three main types of real estate valuation methods: automated valuation model (AVM), broker price opinion (BPO), and appraisal. How has one evolved, while others appear stagnant? What point does an AVM need to reach in order to be more accurate than an appraisal, and therefore be seen as a reliable source? What can consumers do to make sure they’re not being misled by the various “proprietary algorithms” that exist?
It would be helpful to look at the history and current state of methods for objectively determining market value when answering these questions. We can learn more about how technology can help improve human thinking and bring greater efficiency to an ageing profession by studying how it is used.
An Industry in a State of Flux
The Appraisal Institute and The Appraisal Institute of Canada are the most widely recognized professional organizations for real estate appraisers. The Appraisal Institute has over 17,000 members, while The Appraisal Institute of Canada has over 5,400 members. Appraisers in the U.S. and Canada both follow generally accepted appraisal standards, as shown by the Uniform Standards of Professional Appraisal Practice (USPAP or CUSPAP). The requirements for some educational institutions include having a degree, a challenging curriculum, and being mentored before being designated.
The Appraisal Institute states that the number of active members has decreased by 3% every year, with more people retiring in the near future (the average age is around 50). There are many problems in the industry that make it ripe for disruption. These problems include fragmenting, outdated, and overly burdensome regulations.
The proliferation of Appraisal Management Companies (AMCs) is a result of federal regulation that acts as a barrier between lenders and appraisers. Anow and Reggora are two companies working on modernizing the strenuous process of triaging files that comes after a bidding process.
While appraisers typically work independently, they may choose to partner with a private lender. There is no need to work with an AMC. There are too many different groups involved in the market, so it’s hard to keep track of everything and there are lots of inconsistencies.
Intermediaries, complicated workflows, lack of standards, and software that isn’t platform-agnostic create obstacles for appraisers working for organizations with excessively complicated administrative procedures.
Signal for Change
The debate in the U.S. over whether mortgages should require a formal appraisal has been going on for a while, with a majority already exempt by virtue of being below $250,000 (a limit increased from $100,000 in 1994). The final rule for the federal de minimis on residential real estate transactions will be raised from $250,000 to $400,000.
” A residential real estate transaction is defined as a real estate-related financial transaction that is secured by a single 1-to-4 family residential property. For residential real estate transactions that fall below the revised threshold for appraisals, regulated institutions must obtain an evaluation of the real property collateral that is consistent with safe and sound banking practices.
Since 2017, Fannie Mae and Freddie Mac have been offering mortgage products with low loan-to-value ratios that are not subject to traditional reporting requirements. They are instead relying more on their own internal proprietary analytics. The company’s policy allows appraisers who are unlicensed or uncertified, as well as appraiser trainees, to complete a property inspection. They have the right to identify eligible properties and offer waivers at the application stage.
Since brand whitespace provides an opportunity for new companies to develop innovative ways to estimate market value, we are seeing appraisals that are designed to reduce friction, turn around times, and cost for lending institutions.
Factors of Value
You Don’t Know What You Don’t Know
What is the value of other people’s property always a curiosity so that more informed decisions could be made? The challenge is in justifying property values, which can be seen as controversial because they may be based on emotion or lack of insight. It’s not uncommon for lenders and homeowners to have their own, unsubstantiated opinions on property value. For example, they may compare a half-duplex in a quiet suburban neighbourhood to a condominium in the Downtown core, or a concrete house to a wood-frame house. When valuations are displayed on public-facing websites without any explanation of how they were arrived at, there is even more potential for misunderstanding. It’s not just a question of what the numbers are, but why they are what they are.
External factors such as privacy, street noise, crime, topology, conformity to neighbouring properties, proximity to schools and amenities, title to land, orientation, lot dimensions, floor level, view obstructions, physical depreciation, functional obsolescence, government policies, and demographics play a role in preliminary research for a holistic analysis.
Sales comparison is easier with properties that are homogenous. Generally, new construction condos or Vancouver Specials will have many similar sales to compare to, while a custom-built home in a wealthier area might not have as many. The margin of error will be lower for a home that falls within typical parameters.
The sales of similar properties that have occurred in the past 60 to 90 days are used to help determine the value of a property. An appraiser will first check for 1-2 sales in the same building as the subject property (if it’s a condo), which is in line with standard lender requirements. If no such sales can be found, the appraiser will look to the same street, and then expand out to the surrounding neighbourhood as necessary.
Conditions of sale, such as stigmatization and special assessments, can also make it difficult to find arm’s length transactions. The final estimate of value is weighted more towards sales that didn’t need as many adjustments to be made.
Back to the question at hand…
Can we use technology to create a system that works like the human brain to come up with a fair market value for a property?
Although property data is abundant, it is fragmented. Despite the slow adoption curve, decentralization is still destined to happen. Reducing redundancy is akin to the programming principle of DRY (Don’t Repeat Yourself):
In order for a system to be effective, every bit of information must have a single, clear representation that is considered to be the most authoritative.
– The Pragmatic Programmer
As more data becomes available in a machine-readable format, many industries are finding that manual tasks such as filtering, copying and pasting, calculating, and form filling are becoming outdated. Doing the same boring, repetitive tasks over and over can lead to what is called decision fatigue. This is a perfect example of when it would make sense to use automation. The effectiveness of human intervention is decreased by a high cognitive load and a susceptibility to error. Machine Learning can sift through large amounts of data much more effectively than any person, considering new data points as soon as they are available in great detail.
Open Standards and the API Economy
Data portals and APIs that use the RESTful protocol provide access to key datasets like property parcels, tax records, streets, transportation, city services, schools, crime, etc. to help researchers understand how effective these services are. The City of Vancouver offers Open Data APIs that have evolved to offer a variety of export types according to the JSON Schema vocabulary standard.
As innovation in different areas flourishes, the need for integration and standardized increases. Cleaning and labeling big data is a huge undertaking – it’s painstaking, costly, and extremely complex. The Real Estate Standards Organization has created standards that are accessible to everyone in the MLS community in order to make data sharing more efficient.
Zillow’s BridgeAPI, which is a combination of their acquisition of Vancouver-based Retsly and Bridge Interactive, is certified by RESO and provides one of the more modern single point of entry solutions. Its API makes it easier to transport public and transactional records on 148 million properties throughout the U.S.
Turning Point in Canada
Historical property data had been notoriously difficult to acquire, compile, and normalize?—?hidden behind walled gardens of old. The lack of transparency around sold prices and the number of days a home spends on the market has been a problem for Canadian homebuyers who are trying to do their own research.
The Competition Bureau was successful in their anti-competition case against TREB, which resulted in increased access to property data for both Canada and the U.S. This was a costly 7-year legal battle, but it was worth it in the end.
We need to stop stalling and start enacting policies that match up with what the law says. This way, we can start making progress and allow new ideas to come into the field of real estate services.
– Matthew Boswell, Interim Commissioner of Competition
Ubiquity of Technology
The platforms of Amazon Web Services and Google Cloud have introduced a pay-for-use business model and entry-level tiers, which has made it possible to build and scale any project at a lower cost.
The training of ML models has become simpler and more accessible due to open source libraries like NumPy, SciPy, scikit-learn, and pandas. Data-driven visualization can be done using d3js and Jupyter notebooks. The toolkits SageMaker, AutoGluon, and Tensorflow allow for deep learning solutions through minimal coding. The development of Neural Networks opens up many use cases applicable to real estate:
- Image classification (qualitative analysis)
- Object detection (degree of affixation)
- Tabular prediction (comparative analysis)
- Regression models (predictive analytics)
The Future of Housing
Smart Real Estate is a new type of real estate that is based on technology. These new types of real estate platforms facilitate the operation of real estate assets. Buildings are being fitted with sensors that can monitor different aspects of the building, such as the level of activity inside, how energy efficient it is, or whether it needs renovations.
According to a 2019 Altus survey, BIM?—?digital representation of physical and functional characteristics?—?is considered one of the top 3 technologies most likely to cause maximum disruption in real estate development firms. The software is suitable for property valuation, so that creating a “digital twin” could provide real-time access to the most recent version of a survey or floor plan for pre-construction analysis. The use of technology such as LiDAR for 3D models or drones for geospatial imagery allows for the physical world to be explored in ways that were previously unimaginable.
Power to the People
There are many companies that are interesting to watch develop, but the following are particularly noteworthy. They have the capital, network effects, and greatest potential to revolutionize AVMs:
This means that half of all Zestimates are within 4.5% of the actual selling price. Zillow held a $1 million global data science competition in order to improve home valuation. The competition was won by a team that had not even met in person, which is common in today’s distributed workforce environment that is facilitated by tools like Github and Slack.
The winning team developed a system that mimicked the neural circuitry of the brain, leading to the building of accurate predictive models to improve the algorithm that changed the world of real estate.
At least one homeowner requests an offer from Opendoor every 60 seconds. That scale is inconceivable without data science and automation. Opendoor tries to improve the accuracy of their valuations by using multiple models and taking the weighted average of their estimates. They use both automated and human-led valuations in order to scale rapidly while maintaining accuracy.
HouseCanary has been collecting data on transactions of homes in the US for four decades. This data allows them to provide modern, end-to-end valuations for people who invest in real estate or work in that field. This company’s predictive analytics are accurate to within 2.8% for transactions as of July 2019.
Predicting an Uncertain Future
In an on-demand society, customers are trained to expect speed and efficiency. This means that businesses need to be able to provide what the customer wants, when they want it. In the real estate industry, where large sums of money are required to finance projects, the cost and size of new technologies can be prohibitive factors preventing their widespread adoption. It’s not that appraisal professionals want their trade to become obsolete. They just want to find out how technology can be incorporated into their work more quickly. Valuations done by machines rather than by people have the potential to make property transactions simpler and faster, while providing more and better information. More and more businesses are adopting data transparency policies, and those who don’t may find themselves at a disadvantage. Technology is constantly changing, so companies need to be proactive in order to stay ahead of the curve.
A pandemic has occurred since I started writing this article. The world changed in an instant. While the industry awaits further clarity on what the future looks like for in-person real estate transactions, many professionals who require entry into a property are wondering what will happen to their livelihoods. It’s possible that people will become more okay with digital data collection as it’s seen as less of an invasion than if someone were to physically come into your house. In the current climate, it is more important than ever to embrace innovation and move forward, as illustrated by Fannie and Freddie’s decision to adopt alternative appraisals in response to the coronavirus. Nowadays, almost every step of a real estate transaction can be done online. When considering the world’s largest asset class, it is profound to think about. The value we have emotionally tied to our homes is greater than any number, making it difficult for intelligence to predict with certainty what will happen tomorrow.
BUILDING A SMART FUTURE
Property valuations are important for real estate owners, investors, lenders and insurers around the world. The valuation and appraisal industry is in the process of changing the way it operates, including the technology it uses, the services it offers to clients, and the role of the valuation professional.
The technology, data, and client expectations are continuing to evolve which is causing new demands to be placed on the business industry for accurate and reliable property valuations. The profession of teaching has become increasingly complex in recent years, with the added layer of difficulty posed by the COVID-19 pandemic.
1/ EVOLVING CLIENT REQUIREMENTS
After the start of the Global Financial Crisis in 2008, tighter regulation was put in place for many industries, especially the finance industry. This has resulted in a wider range of financial institutions, such as pension funds and insurance companies, seeking more comprehensive reviews of their investment portfolios and those of their clients to ensure compliance.
The financial services industry regulators have demanded that banks and institutional investors have a closer scrutiny of their real estate investments. This has lead to valuations playing a critical role in improving sophistication around risk and mitigation and enhancing the financial stability of markets and industries.
2/ THE ROLE OF THE VALUER
The ongoing automation of various elements of the valuation process, the expanding importance of data analysis and interpretation, and the changing needs of clients all have implications for valuation professionals around the world in terms of their skill set, day-to-day duties and overall role and responsibilities.
“Valuation professionals of the future will require even stronger analytical skills to understand increasingly complex data sources and accurately assess the various factors contributing to property values.”
Public policy, the opportunity to reduce expenses, demand from tenants, and a societal move toward corporate social responsibility are all reasons why the construction of energy-efficient buildings is increasing rapidly around the world. Although many traditional investors have not yet adopted sustainability as a key investment criterion, many global investors are now formally including sustainability in their new investment strategies or investment committee processes.
The importance of the office environment in attracting both tenants and talent has caused workplace wellness to become a pressing issue for corporate real estate in recent years. The COVID-19 pandemic has caused landlords and occupiers to implement wellness measures more quickly and to introduce additional hygiene and health-related protocols.
As occupiers increasingly focus on employee health and wellness by specifying buildings with sustainability and wellness features, demand will be strongest for properties with indoor air quality, ventilation systems, and other indoor environmental features that improve employee comfort.
4/ BIG DATA
In the future, valuation professionals will use Big Data collected by technology powered by the Internet of Things to more accurately judge the performance of real estate assets in the present and future. As we enter the next decade, energy efficiency and foot traffic will become more important factors to consider. The data that will be most useful will be information on crime, environmental hazards, education, and transportation.