Las Vegas skyline landscape
USA-TARO | istock | Getty Images
A version of this article was first featured in the CNBC Property Play Newsletter with Diana Olick. Property Play covers new and evolving opportunities for real estate investors, from individuals to venture capitalists, private equity funds, family offices, institutional investors and large public companies. Sign up to receive future editions directly in your inbox.
John Carrafiell, co-CEO of BGO, a global real estate investment manager with $89 billion in managed assets, takes great pride in the fact that he is sitting right next to the lead data scientist.
Investment strategies have always relied on research and data, no matter what the market is, but artificial intelligence has taken it to a whole new level, transforming investment research models developed several years ago, putting them on steroids.
Carrafiell, who has been in the real estate business for almost 40 years, said he is increasingly irritated by sector research and data methodology. It seemed that everyone was looking at the same information and came up with the same conclusion. The question he said he kept asking himself was, “How do we actually perform?”
The answer was to analyze all his company’s past transactions 20 years ago, using only computer models to remove the human element. What the model found was that outperformance or performance degradation was entirely determined by the local market chosen for investment.
While it may sound trivial considering that real estate mantras have always been “place, place, place,” the outcome was instructing the team to focus almost entirely on the foundations of local markets when choosing future investments, not real estate pricing or national economic trends.
Of course, there are research companies that analyze and rank local real estate markets, but BGO has determined that the results are somewhat random. Instead, it looked at its own past and built a model that accurately backtested what drove its best performance. This model includes all types of local market data points, including demographics and supply trends unique to each location. AI increased the volume and velocity of data in its model.
“We’ve taken thousands of data inputs, many of which are freed from the government, and we have to buy a lot of data inputs that we have to buy from, say, from a communications provider. We’ve found the key,” Carrafiell said. “And we backtest it, so we know it’s accurate.”
BGO has notified its decision to use data science to invest in industrial development in Las Vegas using its partner NorthPoint Development. Other data models suggest that it is not a particularly good investment.
Carrafiell said the “best research” shows investments become mediocre in terms of performance and return.
“But our models were screaming. They explode. They took on rent of $5.88 per square foot. “It doesn’t happen with commercial real estate. That’s not luck.”
The model explained that California’s inland empire is becoming too expensive and analyzes logistical routes. Instead, they discovered that being in Las Vegas can save businesses on a large scale.
“So you had an extra two-hour drive, but you saved 60% on your total cost, and that’s what the model saw,” Carrafiell said. “The tenants we have there serve the entire region. They don’t serve Las Vegas.”
BGO conducted similar analyses for its investments in Florida and the Rust Belt, resulting in significant returns on the investment.
“I think this model has led to a significant increase in our performance,” Carrafiell said.
But he says that while the accuracy of the model is dramatically improved by artificial intelligence, “Boeing can leave Seattle, and the model can’t predict it?”
BGO investment teams focus on upside models of potential characteristics, but their lending teams are focused on negative side modeling because of their risk.
New iterations of the research model on the road include asset allocations to various sectors of commercial real estate. This model suggests an ideally optimal portfolio mix. With the possibility still rising, Carrafiell says he was dialed into the data like never before.
“AI is an enhancer, an accelerator, and it allows for more, but it’s really data science,” he said. “It’s a dedicated data science team of six people sitting next to the CEO and next to the asset management and acquisition team.”
