By Michael Clawar & Rachel Levy, Altus Group and Eric Draeger & Gleb Nechayev, Berkshire Residential Investments 

Real estate investing is often referred to as a long-distance race as owners typically hold assets over several years. However, a property’s rent levels at acquisition and right after (the next three to six months) can be as critical to investment returns as rent growth over the rest of the holding period. Consider a hypothetical acquisition of an apartment property with rents 15% or more below market: over a five-year period, it could generate the same investment returns as another property charging fair market rents due to much faster rent growth. There is good reason for the expression that one usually makes, or loses, money on a buy rather than a sale. 

Despite recent advancements in technology and available data, accurately estimating ‘fair’ current rent levels by measuring and explaining market shifts on a typical acquisition timeline remains more art than science. That said, a good model or algorithm can go a long way in taking the guesswork out of underwriting both current rent levels and near-term changes. With greater availability of both high-frequency data at the microlocation level and modeling techniques, data science can enable more precise measuring, predicting and forecasting of rents. These models tie together property characteristics with economic and demographic factors to ‘nowcast’ and forecast rent growth for specific asset classes or even hyperlocal geographies. With explainable AI (XAI), models formerly considered black boxes can give interpretable results to investors, helping explain why certain properties or markets have stronger growth potential.

To download the full article, click here. To download the whole magazine, click here.