Search
Sign In
Research Directors Exchange Views on AI, Interest Rates and Market Trends
The NAIOP Research Foundation gathered research directors and Distinguished Fellows for a discussion of the factors influencing commercial real estate’s outlook.
The NAIOP Research Foundation held its National Research Directors Meeting at CRE.Converge in Las Vegas in October. Jennifer LeFurgy, Ph.D., the Research Foundation’s executive director, and Shawn Moura, Ph.D., NAIOP’s senior director of research, facilitated the conversation among research directors from commercial real estate services, data, investment and development firms. Four NAIOP Research Foundation Distinguished Fellows also joined the discussion, which addressed applications for artificial intelligence, the effects of declining interest rates on capital markets and development activity, and how recent trends are shaping the outlook for commercial real estate.
How AI is Shaping Market Research and Talent Management
Several research directors indicated that their firms are adopting AI tools to improve efficiency and productivity but shared that current AI tools still require substantial input, direction and editing from human analysts to produce useful insights. Kim Somers, economic director, real assets and thematics at CDPQ, shared that the global investment group has used natural language models to develop sentiment indicators. One is related to the Federal Reserve “to give us some idea what the next move might be,” Somers said, and CDPQ plans to extend that analysis to predict the interest rate policies of other central banks worldwide. Michael Soto, vice president, research West at Savills, observed that “the granular insights from some of the predictive modeling that’s being developed is the real thing that people want.” Commercial real estate firms are often most interested in trends within a specific market or property type, or submarket/subtype, and AI can be useful in analyzing large datasets to identify when conditions are changing for these narrowly defined markets.
Danny Ismail, managing director and co-head of strategic research at Green Street, shared that the firm has used generative AI tools to assist in drafting market data summaries for tertiary markets to improve efficiency. However, to be effective, human analysts first must clean the data, provide the model with narrow parameters and use the correct prompts, and they still have to edit the resulting summaries. Green Street is also limited by restrictions that vendors place on the use of their data in large language models or by external consultants. Brandon Svec, national director of U.S. retail analytics at CoStar, noted that AI has helped alleviate resource constraints associated with producing insights for 390 markets. He added that CoStar eventually hopes to develop “a dynamic program that can read the data itself and come to some conclusions about it.”
Jack Robinson, managing director, chief economist and head of research at Bridge Investment Group, indicated analysts at his firm are using a ChatGPT interface to enhance internal communications of data analyses and draft summaries. However, he noted that some generative AI models are designed to provide a variety of answers to a repeated query. As a result, “We have a lot of trust issues with the believability factor. … If you are not going to present us with the right answer every time, how do we know which one is wrong?”
Raymond Wong, vice president, data solutions delivery at Altus Group, said although generative AI has taken over some of the work that analysts used to perform, such as compiling and summarizing basic market information, he does not think many of those jobs will be replaced by AI in the near future. Experienced analysts are still needed to explain why observable trends are occurring and offer their intuition on how these trends are likely to play out. “We can see the vacancy rate doing something, but we’ll also qualify it … and that’s only based on experience.”
Meeting participants also discussed the ways AI tools are already affecting employee skill development, particularly for recent hires and those just entering the workforce. Workers who were in college or had just started their professional careers during the pandemic missed out on classroom and workplace interactions that are critical to developing soft skills like teamwork and communication. Generative AI may help augment younger workers’ writing and analytical skills, but it also poses new challenges for managers. Ismail shared that when reviewing the job applications of recent college graduates, AI detection software indicated several had used ChatGPT to draft a required case study. In a few instances, the detection software wasn’t even necessary because the applications contained “the same verbatim response.”
NAIOP Distinguished Fellows in attendance shared how they see students using AI and how they, as educators, are integrating the technology into their instruction. Brian Schwagerl, Esq., clinical assistant professor at the Schack Institute of Real Estate at New York University, observed that generative AI has been particularly useful for international students who can formulate complex ideas but need help translating them into English. Dustin Read, Ph.D./J.D., department head, Blackwood Department of Real Estate at Virginia Tech, observed that when students write using AI, their analyses are stronger and effectively tie together complex ideas, but he worries that many are becoming reliant on the technology and struggle to demonstrate critical thinking skills without it. Mirle Rabinowitz Bussell, Ph.D., associate teaching professor and director of undergraduate studies in the Department of Urban Studies and Planning at UC San Diego, requests that students be transparent when they use AI programs. To demonstrate their critical thinking, she asks that students describe how they use the tools to arrive at a conclusion and evaluate their efficacy.
Meeting participants broadly agreed that AI skills are not a substitute for interpersonal skills, which remain critical in the commercial real estate industry and are best developed through interactions at the office and with current and potential business partners. Soto urges new hires to canvass their assigned markets and attend open houses to get familiar with properties and how they are marketed, in addition to making contacts with local CRE professionals. “Soft skill development is so important. This is still a relationship business [and] that will never change.”
The Outlook for Capital Markets
Meeting participants said declining interest rates should support commercial real estate investment activity in the coming year, but most expected the Federal Reserve would need to cut rates further before a meaningful effect on transaction volumes resulted. At current rates, high construction costs remain an obstacle to new development. Ismail and Robinson suggested an additional 50 or 100 basis point decline in prevailing rates would be needed before the yield on cost of construction would be high enough to accelerate new multifamily development. (The Federal Open Market Committee lowered interest rates by an additional quarter point in November.)
For U.S. commercial real estate transaction volumes to grow, Phil Mobley, national director for office analytics at CoStar, believes a stable interest rate environment is more important than the level at which interest rates settle. Soto observed that banks have limited interest in originating new loans because they still have distressed commercial real estate loans on their books that they need to work through before they will be able to increase lending. The slowdown in bank lending has left an opening for private investors, who are currently among the most active in the market.
Wong indicated that most of these observations also applied to the Canadian commercial real estate market. Although the Bank of Canada has cut rates by 75 basis points from their recent high, transaction volumes remain stalled due to a persistent bid-ask spread and owners waiting to list their properties until interest rates and cap rates are more favorable. He expected interest rates to decline by another 25 to 50 basis points by the end of the year and is optimistic that a favorable exchange rate should attract foreign investors to Canadian markets in 2025. (The Bank of Canada cut interest rates by 50 basis points on Oct. 23.)
It is difficult to tell how soon declining interest rates will translate into higher property valuations, but Ismail observed that REITs are already issuing new equity and trading above net asset value, indicating that investors expect capitalization rates to decline. The pricing of listed REITs and the spread between real estate yields and those provided by fixed income “are historically reliable indicators of what real estate is going to do over the next three, six, 12 months,” Ismail said. Robinson noted that historically, changes in capitalization rates have lagged changes in interest rates by six quarters on average. However, cap rate increases followed the most recent round of interest rate increases more quickly than had previously been the case, suggesting that they may also come down more quickly than usual.
Office Closer to Reaching a Bottom
Among commercial property types, office buildings continue to face the most headwinds. Mobley said he expected we may still be another 15 months away from a bottom in average valuations nationwide, although some markets like San Francisco have likely already reached a bottom, with limited potential for near-term improvement. He thinks most of the negative impact on demand from increased hybrid and remote work adoption is already reflected in the market. However, “I don’t think we can talk about the outlook for office without talking about the outlook for employment growth, and that’s going to be a big headwind.”
On the supply side, Mobley suggested nationwide availability may have already bottomed out, mostly because the new construction pipeline has dried up. Nontrophy Class A properties face a particularly challenging market, as they are not competitive on amenities with trophy assets and struggle to attract price-sensitive occupiers, who are more likely to remain in Class B space. That said, commodity office properties have recently benefited from a limited supply of new office space. When high-end occupiers cannot find the best buildings in the best locations, many are choosing to renew their leases for up to five years to allow time for new construction. This is especially true for occupiers who would typically make a large investment in tenant improvements in a new building and are not willing “to put $150 per square foot of [their] own money” into a location that is unlikely to meet their long-term needs, Mobley said.
Soto shared that corporate tenants appear to be getting comfortable with longer office leases and that trophy properties continue to outperform the broader office market. However, “even if you’re talking about the trophy segment of the office market … any kind of vacancy or short-term availability is getting severely punished by buyers … and lenders.” Facing lower valuations, occupiers have proved more willing to purchase office properties for their own use, a pattern that has extended beyond corporate occupiers to nonprofits and state and local governments. Soto noted that current high vacancy rates are partly due to a normalization in demand following a pre-pandemic boom. Prior to 2020, coworking operators and tech firms had distorted the office market by leasing more space than was needed to meet immediate demand. “That level of leasing from coworking firms is probably not going to come back,” Soto said, and most large tech firms are still rightsizing their office footprints.
Svec noted strong parallels between today’s office market and the retail market a decade ago. Lessons learned in retail likely also apply to office buildings. Namely, the industry will need to find new use cases for vacant space and demolish obsolete space. “How do we rightsize a structurally disrupted market where demand … doesn’t look like it did precycle?” Svec said.
However, Mark Stapp, Fred E. Taylor Professor of Real Estate and executive director of real estate programs at Arizona State University, pointed out that retail properties are often easier than traditional office buildings to convert to other uses. “You could slide a church in, you could put a bounce-house in [former retail properties]. … You can’t do that with office as easily.” That means a higher proportion of obsolete office buildings will eventually need to be demolished, Stapp said.
Soto observed that the oldest and most obsolete office buildings, such as those from the 1920s, are often the easiest to convert. Schwagerl echoed this, noting that two of the newest corporate headquarters to open in New York City are converted low-rise buildings that had previously housed printing presses in a former industrial area.
Consumer Trends Reflected in Demand
Growth in consumer spending has slowed noticeably in recent months, and shifting consumer behavior appears to be shaping demand for industrial real estate, with varied effects in different markets. Ismail expressed concern that the purchasing power of lower-income consumers has eroded but believes it is too early to tell what the overall impact will be on the economy or demand for real estate. Institutional-quality real estate generally caters to middle- or higher-income consumers, who are doing well. Soto said he sees this divergence playing out in demand for industrial real estate in the Inland Empire. Smaller properties under 300,000 square feet that typically serve small businesses are struggling, while larger properties over 500,000 square feet that serve corporate occupiers are performing well.
Wong observed a contrasting demand pattern in Canada, where small-bay buildings are outperforming the rest of the market. Overall, Canadian industrial real estate shows signs of cooling demand, with flat rent growth, increasing subleasing availability, and availability rates in recently completed buildings more than double what they have been in recent years.
Financial strains on lower-income consumers are also affecting fundamentals for retail occupiers. Robinson noted that consumers are more sensitive to price changes than in the past. He pointed to Walmart earnings reports, which show customers trading down from higher-value to lower-value items.
Svec echoed this, observing that the bottom 60% of consumers are financially stretched, with no room to buy luxury items. “They are focused on necessities, they are focused on health care, they are focused on housing” and are looking for value, he said. This has driven sales growth at Walmart, which is also now attracting customers from higher up the income ladder. While overall consumer spending growth has slowed considerably since the pandemic-era boom, Svec said purchases of goods to replace those that have worn out since 2020 or 2021 are helping maintain stable demand for retail.
Barriers to Entry Boosting Data Center Profitability
Tech firms’ AI initiatives have sharply increased demand for data center space, with many AI models requiring the construction of new data centers that can provide more power and cooling capacity than traditional data centers. While the increase in demand is notable, participants agreed that resource constraints will have a more significant effect on long-term rent growth and price appreciation in the sector by limiting new construction.
Data centers, especially newer ones, require large amounts of electricity to operate. At the same time, electrical grids are in a state of transition toward cleaner energy generation, and demand for electricity is also growing due to electric vehicle adoption, building electrification and the expansion of advanced manufacturing. Physical limitations on the amount of electricity that can be delivered have created a bottleneck in new data center construction. In some Western states, water scarcity further limits the design and location of data centers, reducing potential supply. Robinson noted that data center developers also face a limited supply of sites that have the requisite fiber optic network in place and can meet occupier requirements for network redundancy.
Given limitations in the availability of potential data center locations that can source carbon-free electricity, some Big Tech firms have stepped back from commitments to reduce their carbon footprints. Schwagerl observed that some of the larger data center occupiers are working to meet sustainability objectives by contracting directly with traditional nuclear power generators or with startups that can build new small-package nuclear reactors.
It remains to be seen how much additional data center capacity will be needed to meet future demand from AI applications. Ismail observed this uncertainty might be a risk factor for data centers, which could see falling profitability if the companies currently occupying them struggle to generate a profit from the large investments they are making in AI.
Shawn Moura, Ph.D., is senior research director at NAIOP.
