Pension funds need better data on rental trends. Could AI be the solution?
Two industry experts have taken on the mantle of exploring how artificial intelligence and machine learning could predict rent prices in Canada.
Aaron Pittman and Erkan Yonder believe the implications for pension funds investing in real estate are significant. The steady rise in rents ensures reliable cash flow, making rental real estate a strong long-term asset.
But as it currently stands, Pittman underscored institutional investors are at this alone.
“The genesis of turning the tide here, I think, comes from the government, but for investors in the private sector and on the development side, we know that they're not going to do it,” said Pittman, portfolio manager at Equiton. “We need more stock in the country, and we need better permitting turns to get shovels in the ground.”
“We need institutional investment because there's not enough capital,” he added. “The government is not going to pile enough capital into the system to solve the problem, so we need the large institutions to get on board.”
With pension funds already holding vast real estate portfolios, AI-driven rent forecasting could help refine their investment strategies, especially as Pittman sees a major shift in the way rent prices are behaving.
“Even as we increase the number of completions, rents are still projected to go up. That’s a complete disconnect from classic economics,” he said, emphasizing the assumption has always been that adding supply should temper price increases.
“We're so far behind the curve in Canada, demand has outstripped supply to such a degree that even as we put on additional stock, the rents will continue to increase,” he added.
Pittman emphasized a need “for approximately, 1500 cranes in Toronto just to level the rents. And we know that's an impractical number.”
He also pointed to low vacancy rates in the country. The extremely low vacancy rates, sometimes measured in hours rather than days or weeks, indicate a very tight market with limited room for clearing.
“We're seeing vacancies at around 1 per cent in a lot of the major centres in Canada. In a healthy, functioning rental or even real estate environment, there's a clearing process that slows the velocity through the system. Right now, we've just completely thrown that out the window,” he said.
Yonder agreed, highlighting the structural issues in Canada’s housing supply. The associate professor of real estate and finance at the John Molson School of Business at Concordia University pointed out that while immigration policies have been aggressive, there’s been no corresponding supply-side policy to match.
“If we bring in people, I don’t think there’s a problem… but if you have an immigration policy, then you should build up more shelters for people,” he said. “If you don't have enough space, that creates a mismatch between demand and supply, which drives the prices and rents upwards.”
This is where AI’s role comes in. Predicting rent prices with traditional models often falls short due to the complexity of the market. AI, however, is proving to be a more effective tool. With real estate deeply ingrained in Canadian investment culture, AI is quickly becoming a crucial tool for institutional investors.
And while pension funds have historically relied on traditional valuation models, Yonder believes the ability of AI to process vast datasets could give institutional investors an edge.
“AI gives some kind of complexity that you don’t understand, but it gives you a better prediction than conventional models,” he said, noting prediction accuracy improves by 20 to 30 per cent on average.
He also acknowledged that predicting rent 10 years into the future is “a difficult task,” but AI helps to capture long-term patterns.
These patterns point to one key message Yonder and Pittman have found in their research. Unless drastic policy shifts occur, rents will continue to rise. Yet, the research suggests that a two-bedroom apartment in Vancouver could hit $7,750, while Montreal’s could exceed $4,000.
“If we do things in the same way that we’re doing, we are heading to a very critical level when it comes to affordability,” said Yonder.
So, how can pension funds avoid to getting to this critical level? Yonder and Pittman assert pension funds need to invest more in data analytics. Yonder pointed to the example of using immigration data.
“We know that immigration impacts the real estate prices and rents in Canada. Then the question becomes how we can make immigration data useful for our real estate investment decision making process and that's the key thing,” he said. “You can use AI to create an AI driven immigration factor that could be helpful with our real estate decision making. It’s all about making meaning out of the data.”
By investing more in data and building teams to effectively utilize data and AI models, pension funds can make more sophisticated, long-term investment decisions in the real estate market, underscored Yonder, particularly as real estate is less liquid.
“The transactions in real estate occur much less frequently than the stock market,” said Yonder. “It's difficult to value real estate as opposed to stocks because it's less efficient. If it’s less efficient, data can help much better than other markets.”
“AI is already here so we should find ways to incorporate AI and data science into our real estate decision making,” said Yonder.
