The Institutional Real Estate Inc Sponsorship brochure, Connected-Investor Focused, We connect people, data and insights, sponsorship, events, IREI Products
Issue link: http://read.uberflip.com/i/896341
NAREIM DIALOGUES FALL 2017 25 Rather than traditional hedonic models, which are limited both statistically and by a researcher's predisposition towards "standard" explanatory variables, we then apply assisted machine learning models that rely on (stochastic) decision trees. These models can sift through millions of combinations of thousands of variables, training and testing the model on randomly selected parts of the datasets, leading to precise out-of-sample tests of predictive performance. We find strong evidence on the superiority of automated valuation models (AVM) over traditional appraisals – the median absolute error of the automated model we develop is below 9%, which compares favourably against the accuracy of traditional appraisals, while the model can produce an instant value at every moment in time, at a very low cost. We also find evidence on the importance of using "hyperlocal" information on the location of an asset. While the use of economic and demographic data at the Census tract and ZIP code level are standard practice in real estate modeling, new information layers gleaned from a wide variety of sources, including social media data, police records, and amenities related to economic vibrancy, add significant value to pricing models. The implications of a well-functioning AVM are significant. First, timely estimations of property values are critical for real estate investors and lenders to make informed underwriting decisions, where systematic errors or biases in valuations may have adverse effects on the provision of equity or debt. Second, investors, regulators, and others rely upon appraised values to assess returns on the USD11 trillion U.S. commercial real estate market, and automated property valuations can provide a more accurate reflection of both the real estate stock (i.e. the value of real estate on the balance sheet) and flow (i.e. real estate returns from changes in capital values). Third, automated valuation models can be used for stress testing under adverse economic scenarios, which post-crisis remains a much- needed tool for regulators, banks, rating agencies and investors. Fourth, the availability of an instant, accurate property value may spur financial innovation in the real estate sector, such as automated loan origination by banks, defined-contribution products for private real estate investments, and arbitrage products for commercial real estate (comparable to emerging products for the single-family sector). THESE BUILDINGS HAVE BEEN HERE FOR DECADES! (AND SO HAVE I) For investors in capital goods that stay around for decades, sometimes even centuries, it is tempting to dismiss the emergence of "proptech" as irrelevant for the industry. After all, you may not sleep in an AirBnB, but rather in the Fairmont. And WeWork is probably not for you or your company, preferring a long-term lease with extensive fit-out package and dito incentives. And 3D printing is just for the Chinese, not for high-quality US real estate development. But the long-lived nature of real estate, and real assets more broadly, massively exposes real estate investors to societal shifts and long-term trends. As an analogy: we all understand that small changes in demographics can have large impacts on real estate pricing. When Amsterdam lost (just) 20% of its population in the late 1700s, house prices declined by 80%. Similarly, just 6% of total retail expenditures moved from bricks-and-mortar shops to online channels, leading to the current bloodbath in retail real estate (although the retail landlords want you to believe differently). Rather than denying change, the real estate sector should embrace it. Sure – some roles will change, and disintermediation may happen. Some will resist, comparable to the utility industry resisting renewables, and the proponents of the internal combustion engine resisting electric cars. But think about it differently: online retail means opportunity for real estate. Ridesharing means parking spaces can be repurposed to more effective use, both on and off-street. Big data creates transparency that will allow for further institutionalization of real estate as an asset class, upping the share that institutional investors will allocate. Uncertainty abound, but that's where companies led by innovative, smart executives thrive. Note: The content of this article partially draws upon the GeoPhy Blog and a forthcoming article in the Journal of Portfolio Management, "Big Data in Real Estate? From Manual Appraisal to Automated Valuation.". For more on Google Trends as a leading indicator, I highly recommend the book "Everybody Lies: Big Data, New Data, and What the Internet Can Tell Us Who We Really Are" by Seth Stephens-Davidowitz