Building a Better AVM
Green Street recently introduced an updated version of its Automated Valuation Model (AVM). We are confident it's the most transparent and accurate commercial real estate AVM available today. The Green Street AVM combines nearly 40 years of experience analyzing and valuing properties, producing reliable and accurate valuation metrics and databases, and developing transparent, trusted methodologies into one revolutionary tool.
Andy McCulloch, Green Street's Global Head of Data & Analytics, explains the significance of the AVM and recent enhancements:
AVMs for commercial real estate are not new. What’s so different about Green Street’s AVM? It is really all about accuracy and transparency. Commercial real estate investors have been slower to adopt AVMs than their peers in the single-family residential space. And this is largely because they either don’t trust the output of existing models, or don’t understand how the models work. In designing the Green Street AVM, we wanted to solve for these shortcomings by 1) Building the model on three accepted and recognized valuation approaches: NOI Capitalization, Value Extrapolation, and Sales Comps; 2) Feeding the engine with our high-quality data sets to improve accuracy; and 3) Showing all of our work. The model we created is proving to be quite accurate, with single digit error rates for individual assets, and low-single digit error rates for portfolios of 25 or more assets. And, importantly, our clients can see exactly how we arrived at our valuation estimates.
Why did Green Street decide to build an Automated Valuation Model (AVM)? An AVM was a very natural evolution for Green Street given property valuation is central to what we do, and a main core competency. Our goal as a firm has always been to help commercial real estate investors make better capital allocation decisions: In the public market this is through stock selection of REITs; In the private market this is through sector and market selection. To provide that advice effectively to our clients, we need to be experts in understanding the drivers of property values and knowing – in real time – where assets are trading. This experience and expertise, along with a sizable amount of proprietary data built over the past 35 years, is what makes our AVM possible.
Green Street recently launched a new version of the AVM. What enhancements does version 2.0 include? The new version of the tool really brings two big enhancements. The first is around market coverage. Our initial model was confined to the 50 largest metros in the U.S., but the new model can now value assets in all 384 MSAs for the core property sectors (apartment, industrial, office, and strip centers). The second big enhancement is around the ability for the user to customize the valuation output. Users can now self-select their desired comp set in addition to tweaking the weightings of each valuation approach – both of which were static in Version 1.0 and based on Green Street’s recommendations. While we are quite proud of how accurate we believe the default recommendations are, we still wanted clients to be able to inject their own unique market knowledge and opinions into their property valuations. Said differently, our AVM is not meant to replace the traditional valuation / appraisal processes by providing a definitive “answer” on what a property is worth, but to dramatically speed those valuation processes up. The Green Street AVM can quickly and easily surface all the relevant and proprietary valuation data we have for a given location (e.g., cap rates, market trends, sales comps, etc.) and provide a starting valuation estimate that is both fully transparent and defendable, but one that the user can also fine tune.
You said proprietary data makes the AVM possible. Can you expand on that? Green Street has the benefit of a large team of experienced analysts, who are trained to be valuation experts on sectors, markets, and companies. The work they do supports our best-in-class research platform and associated data products. The “exhaust” – as we like to call it – of all that work is the fuel that drives the AVM. Yes, we have designed a sophisticated and dynamic valuation engine for the AVM that produces instant, automated valuation reports. This of course gives the impression that a machine or computer is doing all the work. In reality, it is actually our large analyst team producing and constantly updating our valuation data sets that allows the AVM engine to produce quality results. It doesn’t matter how “smart” your model is if it is fed with low quality data. Green Street has the luxury of possessing high-quality cap rate data, real-time Commercial Property Price Indices (CPPIs), a property and market grading mechanism, and a deep transaction comp database.
You’re emphasizing transparency in Green Street’s AVM. Why is that important? I mentioned the importance of showing your work. This may sound like something your elementary school math teacher would say, but it is surprising how many models abandon this basic premise. And, they abandon it because the model creators either don’t know themselves what is driving the final output – the black box problem - or because they are embarrassed to tell you what the model is actually using. A model that is trained to find relationships with property pricing will find relationships, no matter how flimsy or truly irrelevant those relationships are. Green Street does use some machine learning techniques to enhance the AVM’s output, but the model is based on traditional valuation approaches where full transparency can be retained. Our reports not only show our estimate of value, but also the results from the three approaches, the build-up of value in each approach, the base and adjusted cap rates, the individual sales comparables, etc. Everything an investor would want to see to have confidence in the result and fully defend their ultimate valuation.
How did you determine the weighting of the three legs in the AVM valuation model? The weightings are based on the confidence we have in each of the three valuation approaches. Each approach has its own scoring system. As an example, in the SmartCompsTM or sales comparables approach, the more robust and recent the transaction activity around a certain property, and the higher the quality of the “matches,” the higher the confidence score. Said differently, if the model is valuing an office building and finds five similar buildings that have traded in the same submarket in the past few months – the confidence score will be high. On the other hand, if it can only find one office building that traded two years ago, on the other side of town, that is also much older and of lower quality, the confidence score will be much lower. For those clients that really want to get into the weeds of the model and our confidence scoring, they can download our whitepaper on the Green Street website. And, as I previously mentioned, the weightings are just a recommendation that clients can now override. As long as the weightings for the three approaches adds up to 100%, users can select any weighting schema they want.
Green Street NAV estimates are known as the gold standard in the marketplace. Are those going to be automated as well? No. Producing a high-quality company net asset value (NAV) estimate is far more than just valuing the operating properties. REITs have land banks, development pipelines, and off-balance sheet joint ventures. There is also very little accounting consistency from one company to the next in things like how costs are allocated between property management and G&A, or between what is a capitalized expenditure vs. an operating expense. All of these things require the expert judgment of a Green Street analyst. This meticulous and manual valuation work, however, creates valuable data and insights that cascade throughout the rest of Green Street’s research process and into the AVM.
Does the public market impact the AVM valuation estimates? The public market does not impact or influence the estimate of value produced by the AVM, which is meant to reflect where properties are being negotiated in the private market. That said, each AVM report we produce provides not just an estimate of current value, but a forecast for future values under different economic scenarios. The public market does influence those forecasts. As anyone who follows Green Street knows, we are huge advocates of all commercial real estate investors closely following signals emanating from the public markets – whether they invest in REITs or not. The public market has shown time and time again that it is predictive of changes in private market values, both at the sector and market level. The public market isn’t always right, but it’s right a heck of a lot more than it’s wrong. In fact, the data is so compelling that any direct investor managing capital who is not paying close attention to the public market is doing themselves and their clients a huge disservice.
Take advantage of Green Street's reliable property and portfolio valuations from the market-leading independent source for actionable commercial real estate intelligence by clicking here.
Learn more about our insights
Our thought leadership helps thousands of clients make better investment decisions every day. Inquire here to learn more about Green Street’s product suite.