The Risks CRE Investors Aren't Pricing In, with Moody's Analytics' Jeffrey Havsy

Jeffrey Havsy is the commercial real estate industry practice lead at Moody's Analytics, where he focuses on combining property fundamentals with economic, credit, climate, and alternative data to help lenders, investors, and operators make better decisions. He has spent his career at the intersection of economics and real estate, including roles at NCREIF and CBRE. Moody's Analytics covers more than 600 million public and private entities worldwide and generated over $7 billion in revenue in 2025. Jeffrey is based in Needham, Massachusetts.

(02:46) From data scarcity to data curation
(03:39) What lenders and investors miss beyond property fundamentals
(05:34) Tenant credit, crime data, and truck traffic as signals
(09:16) Moody's commercial location score explained
(12:52) Cap rates: overrated, underrated, or misused
(15:39) Moody's MCP launch and what it means for CRE decisions
(18:32) Why Moody's stays AI platform-agnostic
(20:44) Where the human stays in the loop
(23:33) Physical risk vs. ESG
(26:11) The hidden risks in industrial real estate
(30:06) How robotics changes what a warehouse is worth
(31:09) The one macro data point CRE ignores: productivity
(33:10) Collaboration superpower: Abraham Lincoln

🏙️ Learn more:
-Visit Moody's Analytics / Moody's CRE
-Jeffrey Havsy on LinkedIn

🏙️ In-person Events:
-Finance Forum: NYC | 11/3
-State of Office Forum: NYC | 11/4
-Breakthrough Awards Breakfast | NYC | 12/2
-All upcoming Commercial Observer events

🏙️ Virtual Events:
-AI & Real Estate: You’ve Hired an Alien Superintelligence. Now What?: 11/10

-AI & Innovation Forum: 11/18

🏙️ Connect with Tangent:
-Edward Cohen on LinkedIn
-Zach Aarons on LinkedIn
-Tangent on LinkedIn

Creators and Guests

Host

Edward Cohen

Host & Executive Producer at Tangent

Guest

Jeffrey Havsy

What is Tangent: Conversations with Real Estate & Tech Innovators?

A podcast community for real estate and tech leaders, investors, operators and innovators.

A new era is upon us. Tangent💚Proptech unites Proptech founders💡, Real Estate investors and operators🏢, Venture Capital investors, urban leaders🏙 and passionate creators🌎 who are improving our communities and cities. Join us to learn how we can solve present day challenges with innovative technology and greater collaboration.

Join Proptech entrepreneur Edward Cohen, to learn about housing innovation, the future of work and climate tech. We examine these themes through interviews and conversations with Proptech CEO's, real estate owners and operators, and venture investors, and how we can leverage tech and collaboration to help solve the housing crisis, figure out the future of work and how to decarbonize the built environment.

If you would like to get your company’s or organization’s mission featured on Tangent, email us at [e-mail].

Stay curious and always be learning! 🤓

Jeffrey Audio
===

Edward: [00:00:00] [00:00:01] [00:00:02] I'm Edward [00:00:03] Cohen, and this is [00:00:04] Tangent, a [00:00:05] Commercial Observer [00:00:06] podcast. [00:00:07] [00:00:08] [00:00:09] [00:00:10] [00:00:11] Commercial real estate [00:00:12] has more data than [00:00:13] ever. Rent [00:00:14] rolls, vacancy [00:00:15] rates, cap [00:00:16] rates, sales [00:00:17] comps, every [00:00:18] platform in the market [00:00:19] is selling access to [00:00:20] more of it. [00:00:21] And yet, the most [00:00:22] consequential [00:00:23] decisions in CRE, [00:00:24] whether to [00:00:25] lend, whether to buy, [00:00:26] whether to [00:00:27] hold, are still [00:00:28] being made with [00:00:29] incomplete information.[00:00:30]

Because the [00:00:31] data that predicts [00:00:32] what happens to an [00:00:33] asset, crime [00:00:34] trends, insurance [00:00:35] shifts, tenant [00:00:36] credit [00:00:37] deterioration, labor [00:00:38] market changes to [00:00:39] count this [00:00:40] over, [00:00:41] they live [00:00:42] entirely outside of the [00:00:43] commercial real estate [00:00:44] stack. [00:00:45] The firms that [00:00:46] figure out how to [00:00:47] combine those layers are going [00:00:48] to have a structural [00:00:49] advantage over the [00:00:50] ones that don't.[00:00:51]

That is a problem [00:00:52] Moody's Analytics [00:00:53] has been working on. [00:00:54] Moody's [00:00:55] generated over $7 [00:00:56] billion in [00:00:57] revenue in 2025, [00:00:58] and the [00:00:59] analytics segment alone [00:01:00] was over half of [00:01:01] it. They [00:01:02] have data on more than [00:01:03] 600 [00:01:04] million public and [00:01:05] private entities [00:01:06] worldwide. [00:01:07] And today, we do [00:01:08] have the perfect guest to talk [00:01:09] about this.

Our [00:01:10] guest today is [00:01:11] Jeffrey Havsy, [00:01:12] commercial real [00:01:13] estate industry practice [00:01:14] lead at Moody's [00:01:15] Analytics. [00:01:16] Jeffrey, where [00:01:17] does this podcast find [00:01:18] you?

Jeffrey: I am [00:01:19] in Needham today. Thank [00:01:20] you for inviting me. [00:01:21] Uh, I'm excited to [00:01:22] join, and I look [00:01:23] forward. We're gonna have some great [00:01:24] discussion on [00:01:25] data in the real estate [00:01:26] world and, [00:01:27] uh, some other [00:01:28] topics

Edward: Awesome. [00:01:29] So [00:01:30] Jeffrey, you spent your [00:01:31] career at the intersection of [00:01:32] economics, real [00:01:33] estate, [00:01:34] NCREIF, CBRE, [00:01:35] and now Moody's. [00:01:36] So how has [00:01:37] your view of what [00:01:38] commercial real estate [00:01:39] data is [00:01:40] changed over that [00:01:41] time?

Jeffrey: [00:01:42] So it used to be that [00:01:43] data was a [00:01:44] scarcity, right? [00:01:45] People held data close [00:01:46] to their vest. It was [00:01:47] hard to get. [00:01:48] There was incomplete [00:01:49] data. [00:01:50] That is no [00:01:51] longer the problem. [00:01:52] The issue [00:01:53] is not [00:01:54] getting data, [00:01:55] it's [00:01:56] getting [00:01:57] quality data and [00:01:58] understanding [00:01:59] which data [00:02:00] you want to use and [00:02:01] which data is [00:02:02] important given [00:02:03] what you're trying to [00:02:04] do.

So it's [00:02:05] actually more important [00:02:06] now to be able to [00:02:07] [00:02:08] filter ignore or [00:02:09] not use certain [00:02:10] pieces of data [00:02:11] than it is to try to [00:02:12] find [00:02:13] the data [00:02:14] itself

Edward: [00:02:15] Yeah, the filtering, I [00:02:16] mean, I, I think of [00:02:17] filtering as, as [00:02:18] curation [00:02:19] ultimately in terms [00:02:20] of how [00:02:21] much [00:02:22] information we're exposed [00:02:23] to, right? Not only [00:02:24] what [00:02:25] operators out [00:02:26] there, investors out there [00:02:27] subscribe to, [00:02:28] but in terms of [00:02:29] differentiating [00:02:30] signal and noise, [00:02:31] that [00:02:32] curation aspect is more [00:02:33] critical than [00:02:34] ever

Jeffrey: [00:02:35] Absolutely. And [00:02:36] as you talked about in [00:02:37] your introduction, [00:02:38] it's not [00:02:39] just the [00:02:40] fundamentals of real [00:02:41] estate. That's [00:02:42] table stakes, [00:02:43] right? You need that to [00:02:44] begin with. [00:02:45] But to your point, [00:02:46] do you have crime [00:02:47] stats? Are you [00:02:48] using cell phone [00:02:49] data? Are you looking [00:02:50] at truckloads and how [00:02:51] they're being [00:02:52] delivered? What [00:02:53] other data are you [00:02:54] combining [00:02:55] with your real estate [00:02:56] data?

How are [00:02:57] you combining [00:02:58] it? And [00:02:59] is that leading [00:03:00] you to make [00:03:01] faster [00:03:02] decisions, better [00:03:03] decisions? Are [00:03:04] you able to make quicker [00:03:05] decisions so you can do [00:03:06] more deals or [00:03:07] eliminate bad ones [00:03:08] faster? [00:03:09] But you need to combine [00:03:10] those two [00:03:11] pieces more than [00:03:12] ever in today's [00:03:13] world

Edward: [00:03:14] Commercial real estate data has [00:03:15] always been called [00:03:16] opaque and, and [00:03:17] fragmented. so how [00:03:18] much of that is a [00:03:19] data problem and [00:03:20] how much is an [00:03:21] interpretation of [00:03:22] the data problem?[00:03:23]

Jeffrey: I [00:03:24] think that used to be [00:03:25] the case, and I think [00:03:26] we're kind of living off [00:03:27] that legacy, [00:03:28] but I actually don't think [00:03:29] that's the big problem [00:03:30] anymore. I think [00:03:31] data has become a [00:03:32] lot more complete. [00:03:33] It has [00:03:34] become a lot less [00:03:35] [00:03:36] opaque. In addition to [00:03:37] Moody's, there's several other [00:03:38] players out there in the [00:03:39] market that [00:03:40] have data at the [00:03:41] fundamentals.

[00:03:42] You have supply, [00:03:43] you have demand, [00:03:44] you have vacancy [00:03:45] rates, you have [00:03:46] rents. You have both [00:03:47] asking rents, you have [00:03:48] operating [00:03:49] expenses, and within the [00:03:50] operating [00:03:51] expenses you can understand what are [00:03:52] taxes, [00:03:53] how much is [00:03:54] insurance. [00:03:55] So I think there's a [00:03:56] lot more clarity [00:03:57] than there used to be. It used to [00:03:58] be brokers kept [00:03:59] everything close to their [00:04:00] chest, and [00:04:01] over time [00:04:02] that has [00:04:03] eased dramatically.

So I [00:04:04] think we have a lot more [00:04:05] data, but I also [00:04:06] think we have a lot [00:04:07] better data than we used [00:04:08] to. [00:04:09] But that alone [00:04:10] isn't what you need in [00:04:11] today's world, and I think [00:04:12] that's what people [00:04:13] forget is, [00:04:14] yes, the real [00:04:15] estate data's [00:04:16] important, and [00:04:17] it's gotten much, much [00:04:18] better, [00:04:19] but now there are len- [00:04:20] many more [00:04:21] things. So if you aren't [00:04:22] connecting the pipes [00:04:23] to these other [00:04:24] sources, you're running [00:04:25] into...

You're, you're at a [00:04:26] disadvantage relative [00:04:27] to your competitors[00:04:28]

Edward: So [00:04:29] let's dive into that a [00:04:30] little more. So most [00:04:31] commercial real estate data [00:04:32] platforms stop [00:04:33] at property [00:04:34] level fundamentals, [00:04:35] right? The, the rents, the [00:04:36] vacancies, , the [00:04:37] cap rates, which, we're [00:04:38] gonna talk about that [00:04:39] very [00:04:40] popular, and, somewhat [00:04:41] controversial metric in a [00:04:42] little bit. so what [00:04:43] does a lender, [00:04:44] investor, and operator [00:04:45] miss when they stop at [00:04:46] the property level [00:04:47] [00:04:48] fundamentals?

Jeffrey: So [00:04:49] there are a couple things [00:04:50] there. One, [00:04:51] are they looking at [00:04:52] the economics? Because [00:04:53] that's gonna help [00:04:54] drive [00:04:55] the real estate. I [00:04:56] don't just mean the interest rate. [00:04:57] Obviously, the interest [00:04:58] rate's important, right? Wh- [00:04:59] where's the 10-year [00:05:00] treasury, [00:05:01] or where are [00:05:02] triple Bs if you're using [00:05:03] that as a [00:05:04] proxy to [00:05:05] where cap rates [00:05:06] may be.[00:05:07]

But what other [00:05:08] fundamentals? What kind of [00:05:09] population growth do you [00:05:10] have? ... [00:05:11] And within that [00:05:12] population growth, [00:05:13] what are the [00:05:14] various cohorts [00:05:15] by gender, [00:05:16] by age, by [00:05:17] education, by [00:05:18] income? [00:05:19] Understanding then [00:05:20] who may be [00:05:21] tenants in your [00:05:22] building or who may [00:05:23] be [00:05:24] consumers [00:05:25] or customers [00:05:26] of your tenants [00:05:27] in those [00:05:28] buildings, and all those [00:05:29] things are - [00:05:30] combined with [00:05:31] the real estate [00:05:32] data as part of [00:05:33] that.

And I think you need [00:05:34] to really factor in [00:05:35] that. What is [00:05:36] the tenant credit [00:05:37] of your tenants? Are [00:05:38] you monitoring [00:05:39] that and [00:05:40] understanding it? Are there [00:05:41] bad actors? Have they had [00:05:42] [00:05:43] bankruptcies? All this [00:05:44] information's available. [00:05:45] I think [00:05:46] Moody's does a really good [00:05:47] job, pulling this [00:05:48] together, and that we have all [00:05:49] these businesses.[00:05:50]

And Moody's has spent the [00:05:51] past [00:05:52] year, if not two, [00:05:53] trying to connect those [00:05:54] pipes. And I'm not gonna [00:05:55] say we're perfect, right? We still [00:05:56] have a lot more work to [00:05:57] do, and I think the [00:05:58] industry overall has [00:05:59] a lot more work to [00:06:00] do, but I think [00:06:01] the industry is [00:06:02] taking a step in the right [00:06:03] direction. [00:06:04] Real estate is [00:06:05] always...

[00:06:06] I don't know if it's exactly [00:06:07] last, but it's near [00:06:08] the back when [00:06:09] it comes to adopting [00:06:10] new [00:06:11] technologies, new [00:06:12] methodologies, and new [00:06:13] information. [00:06:14] I think that's [00:06:15] starting to change, [00:06:16] and I think being able to [00:06:17] combine those pieces [00:06:18] together, [00:06:19] whether it's through [00:06:20] Moody's or through [00:06:21] other, , players in the [00:06:22] space, I think [00:06:23] you're seeing [00:06:24] [00:06:25] forward-looking [00:06:26] investors, [00:06:27] lenders, [00:06:28] brokers, [00:06:29] appraisers, whatever [00:06:30] space you are in, [00:06:31] doing [00:06:32] a better job of [00:06:33] that.[00:06:34]

And those are [00:06:35] being able to do, A, [00:06:36] either more [00:06:37] deals, avoid some [00:06:38] of the bad [00:06:39] deals, but make [00:06:40] decisions faster [00:06:41] because of that

Edward: [00:06:42] I think [00:06:43] it's a super [00:06:44] exciting time to [00:06:45] be any [00:06:46] commercial real [00:06:47] estate, , [00:06:48] player, [00:06:49] given what you just said. There's a [00:06:50] more [00:06:51] collaborative world. [00:06:52] There's more access, [00:06:53] affordable access [00:06:54] to [00:06:55] information. so certainly [00:06:56] super exciting. I wanna go back [00:06:57] to your point [00:06:58] about the [00:06:59] macro, because yes, [00:07:00] ultimately real estate is [00:07:01] about [00:07:02] the macro, and [00:07:03] if you bought [00:07:04] a [00:07:05] multi-family building in the New [00:07:06] York City area [00:07:07] between [00:07:08] [00:07:09] 2010 and 2014 and [00:07:10] sold it before [00:07:11] 2019, then you probably [00:07:12] did amazingly, [00:07:13] regardless [00:07:14] of what building it [00:07:15] was.

, So I wanna [00:07:16] ask you just, [00:07:17] So if you bought a building at [00:07:18] the right [00:07:19] [00:07:20] macro time, [00:07:21] and let's call that [00:07:22] low interest rates or let's [00:07:23] call that, [00:07:24] demographics, trends that are [00:07:25] favorable of the [00:07:26] asset class that you're [00:07:27] buying. you get the [00:07:28] macro [00:07:29] right, [00:07:30] but you don't get [00:07:31] the best asset, [00:07:32] whether because there's [00:07:33] more issues [00:07:34] within the building, [00:07:35] it's not in the most [00:07:36] desirable [00:07:37] block, , would the [00:07:38] macro m- [00:07:39] make that [00:07:40] deal a better [00:07:41] deal than it would have been, [00:07:42] while [00:07:43] if you bought the [00:07:44] best potential [00:07:45] asset with the most outcome [00:07:46] but with the [00:07:47] wrong macro [00:07:48] [00:07:49] trends [00:07:50] not, , ultimately make that [00:07:51] a good deal?

I don't know [00:07:52] if I asked that in the best [00:07:53] way, but

Jeffrey: Yeah, [00:07:54] no, I understand what you're [00:07:55] asking, right? Yes. So [00:07:56] the macro [00:07:57] trend is going to [00:07:58] drive it, but that does [00:07:59] not guarantee [00:08:00] success. A [00:08:01] poorly run [00:08:02] building, even with the right [00:08:03] macro [00:08:04] trends, will [00:08:05] underperform or [00:08:06] significantly [00:08:07] [00:08:08] underperform A good [00:08:09] well-run [00:08:10] building, [00:08:11] even in a tough [00:08:12] macro environment, [00:08:13] can still [00:08:14] perform extremely [00:08:15] well.

[00:08:16] And as we're seeing in the [00:08:17] current environment [00:08:18] today, [00:08:19] operations [00:08:20] matter more than [00:08:21] ever. I think [00:08:22] AI is part of that, right? [00:08:23] You can manage your [00:08:24] expenses, you can be [00:08:25] better with your [00:08:26] electricity, you can [00:08:27] make a better case [00:08:28] on why your insurance [00:08:29] shouldn't go up as [00:08:30] much and talk to your [00:08:31] insurance company about [00:08:32] that.

You can [00:08:33] anticipate [00:08:34] breakdowns in [00:08:35] mechanicals [00:08:36] and get ahead of the [00:08:37] curve so they're minor [00:08:38] fixes rather than major [00:08:39] fixes. [00:08:40] And that can, [00:08:41] even in a tough [00:08:42] macro environment, [00:08:43] help you [00:08:44] perform well [00:08:45] in that [00:08:46] situation

Edward: [00:08:47] That answers the [00:08:48] question that I was trying to ask. [00:08:49] how... Yeah, it's not, [00:08:50] not, uh, [00:08:51] doesn't guarantee anything, but [00:08:52] certainly helps to have the [00:08:53] macro in your favor. [00:08:54] so [00:08:55] let's talk about and, [00:08:56] and, you know, it can be in the [00:08:57] lens of, [00:08:58] costs, and expenses [00:08:59] as you were talking about, but how [00:09:00] do you combine the [00:09:01] non-commercial real estate [00:09:02] data [00:09:03] with commercial real estate [00:09:04] data in a way that [00:09:05] produces something [00:09:06] a decision-maker can [00:09:07] act on rather than [00:09:08] just analyze?[00:09:09]

Jeffrey: So [00:09:10] one of the things we do [00:09:11] is we have something called a [00:09:12] commercial location [00:09:13] score, and you mentioned [00:09:14] this early [00:09:15] on. What about the [00:09:16] crime? What is [00:09:17] the crime issue, [00:09:18] right? So we [00:09:19] combine that as part so [00:09:20] you can look at and [00:09:21] understand how much risk [00:09:22] there is. [00:09:23] It's not quite a [00:09:24] macro [00:09:25] effect, but it [00:09:26] [00:09:27] definitely could [00:09:28] [00:09:29] impact your [00:09:30] performance.[00:09:31]

And if you take [00:09:32] something like there are all sorts [00:09:33] of news aggregators [00:09:34] out there, we have something called [00:09:35] NewsEdge, and [00:09:36] say, "Hey, wait a second. [00:09:37] If there was a violent [00:09:38] crime in front of my [00:09:39] apartment [00:09:40] building Or I'm [00:09:41] building, I'm thinking of [00:09:42] buying, that [00:09:43] is going to [00:09:44] impact or [00:09:45] potentially could [00:09:46] impact my leasing [00:09:47] velocity, [00:09:48] the rents I get [00:09:49] 'cause people are [00:09:50] scared, [00:09:51] right?

So but just looking at [00:09:52] crime alone, that's [00:09:53] not gonna do it. Was it a [00:09:54] violent crime or was it a [00:09:55] non-violent crime? Did [00:09:56] somebody get ripped [00:09:57] off, [00:09:58] right, because [00:09:59] they, somebody [00:10:00] got access to your [00:10:01] computer and or [00:10:02] did somebody get attacked? [00:10:03] Was it [00:10:04] auto theft [00:10:05] or...

And [00:10:06] all those things [00:10:07] matter, and understanding that [00:10:08] type of crime and how [00:10:09] that may or may not [00:10:10] impact, [00:10:11] and then the building [00:10:12] owner or lender [00:10:13] can respond to [00:10:14] that [00:10:15] accordingly. So [00:10:16] understanding that type [00:10:17] of information. [00:10:18] Nowadays, you can [00:10:19] look and you can say, [00:10:20] "Hey, [00:10:21] what is the truck [00:10:22] traffic?" You can [00:10:23] see how many trucks are [00:10:24] going in or out [00:10:25] of a [00:10:26] warehouse [00:10:27] or a [00:10:28] retail center.[00:10:29]

Now again, it's [00:10:30] masked a little bit, the [00:10:31] data that I'm thinking about, [00:10:32] right? You can't see the [00:10:33] exact retailer, [00:10:34] but you can [00:10:35] understand, hey, wait a second, there [00:10:36] used to be 10 [00:10:37] truckloads going into [00:10:38] this [00:10:39] mall, now there are only [00:10:40] nine. [00:10:41] I better dig a little [00:10:42] deeper and [00:10:43] understand what's [00:10:44] happening [00:10:45] because the [00:10:46] volume may not be there.

They may be [00:10:47] getting less foot [00:10:48] traffic. You can look at [00:10:49] foot traffic via cell [00:10:50] phone. You can [00:10:51] look at, uh, [00:10:52] cars parked [00:10:53] via, uh, satellite [00:10:54] or drone [00:10:55] imagery, right? You can then [00:10:56] compare week over [00:10:57] week or year over year. [00:10:58] So it's that [00:10:59] combination of [00:11:00] [00:11:01] non-traditional real estate data in [00:11:02] addition to, [00:11:03] hey, there's a lot of [00:11:04] supply going up.

[00:11:05] That's gonna impact the [00:11:06] vacancy rate, and that's gonna [00:11:07] impact my rents. [00:11:08] And I think [00:11:09] good [00:11:10] operators, good [00:11:11] investors, good [00:11:12] lenders [00:11:13] use that type of [00:11:14] information to get [00:11:15] ahead of the curve, [00:11:16] to anticipate [00:11:17] things that may be [00:11:18] occurring, or [00:11:19] to be able to react [00:11:20] quickly before a small [00:11:21] problem becomes a [00:11:22] big one[00:11:23]

Edward: [00:11:24] Fascinating. There's so many [00:11:25] angles we can look at it, not [00:11:26] only from l- [00:11:27] d- ground [00:11:28] level, but also from the [00:11:29] air. Uh, [00:11:30] even I'm [00:11:31] sure underground is [00:11:32] gonna come out i- [00:11:33] in a bit with all [00:11:34] these, like, tunnel [00:11:35] and underground [00:11:36] technology coming from the m- the [00:11:37] military side [00:11:38] of things. I [00:11:39] can see that being [00:11:40] part of, uh, the [00:11:41] structure.

Like, are the [00:11:42] foundations [00:11:43] actually what they say they [00:11:44] are? have you seen [00:11:45] anything along those [00:11:46] lines?

Jeffrey: I've [00:11:47] not seen anything, [00:11:48] but I've heard as [00:11:49] people look, especially [00:11:50] in, um, [00:11:51] earthquake zones, [00:11:52] right? [00:11:53] What is the [00:11:54] structure there? So I've seen it more [00:11:55] there, but as you think about, [00:11:56] like, the [00:11:57] boring tunnels or if [00:11:58] you have [00:11:59] subways, what does that do? [00:12:00] And if you have to drill, [00:12:01] like we know in New [00:12:02] York, they- they're [00:12:03] supposed to extend the Second [00:12:04] Avenue subway.[00:12:05]

[00:12:06] Suppose you're under that, how [00:12:07] does that impact it? [00:12:08] What does that [00:12:09] do? So yeah, there's a [00:12:10] lot of stuff there. I've [00:12:11] not done any [00:12:12] work in that area, [00:12:13] but I know it's [00:12:14] of interest to [00:12:15] people, and it's something we'll [00:12:16] definitely be paying [00:12:17] attention to

Edward: [00:12:18] Certainly. I know I, [00:12:19] I recall [00:12:20] that in London it's [00:12:21] becoming a problem, [00:12:22] so, [00:12:23] luxury single family [00:12:24] homes can't really build [00:12:25] upwards, but they can [00:12:26] build, basements. [00:12:27] So they often are [00:12:28] building three, four [00:12:29] levels down. [00:12:30] And then some of [00:12:31] the neighbors are finding [00:12:32] that now their homes are a [00:12:33] little shaky. [00:12:34] so fascinating [00:12:35] new, new world [00:12:36] underground.

[00:12:37] let's, uh, move [00:12:38] on to cap [00:12:39] rates, which [00:12:40] are often the [00:12:41] most cited [00:12:42] metric in commercial [00:12:43] real estate. tell us [00:12:44] how [00:12:45] overrated they [00:12:46] are, and if so, [00:12:47] why?

Jeffrey: [00:12:48] So it's funny. So I think [00:12:49] they're incredibly [00:12:50] overrated, and [00:12:51] yet Moody's [00:12:52] just came up [00:12:53] with... And I have to, [00:12:54] uh, mention my [00:12:55] colleague, Kevin Fagan, [00:12:56] and his team [00:12:57] of Chillar and [00:12:58] Matt. [00:12:59] They did an amazing [00:13:00] job. We now have a [00:13:01] cap rate [00:13:02] estimate for [00:13:03] every building in our [00:13:04] database using the [00:13:05] nearest neighbor [00:13:06] model. it's Alex [00:13:07] Van Daman's, , [00:13:08] modeling, and then [00:13:09] we've worked with Alex to [00:13:10] do that. [00:13:11] I'm not looking at the rent [00:13:12] rolls. I don't [00:13:13] know the details. I [00:13:14] can't tell you, and you [00:13:15] may disagree with my [00:13:16] cap rate, [00:13:17] but what I can [00:13:18] give you is a [00:13:19] distribution, and I think [00:13:20] that's where it's relevant, [00:13:21] right? You and I [00:13:22] can argue is it a [00:13:23] five six? Is it a [00:13:24] five [00:13:25] seven? who knows [00:13:26] who's right? But if [00:13:27] I tell you within one [00:13:28] standard [00:13:29] deviation it's somewhere between a [00:13:30] five five and a [00:13:31] six, [00:13:32] that's gonna [00:13:33] give you comfort, [00:13:34] especially if you're a [00:13:35] lender, right?

You wanna [00:13:36] understand, am I [00:13:37] in the ballpark [00:13:38] with it? And I [00:13:39] think that can be very [00:13:40] valuable, 'cause cap [00:13:41] rates can be done [00:13:42] so quickly and [00:13:43] so easily. [00:13:44] You don't wanna be [00:13:45] overly [00:13:46] reliant on them, [00:13:47] but they can be a [00:13:48] good indication [00:13:49] of pricing, so I [00:13:50] wouldn't totally [00:13:51] dismiss them. And I think people, [00:13:52] yes, they've gotten a [00:13:53] bad rap because [00:13:54] how do you do the math [00:13:55] and what's counted?

Is [00:13:56] it trailing [00:13:57] 12? If it's [00:13:58] forward-looking, is it [00:13:59] recurring revenue or [00:14:00] [00:14:01] not? But I think there's a [00:14:02] lot of really good [00:14:03] stuff in the cap rate that [00:14:04] you can do back of the [00:14:05] envelope very [00:14:06] quickly, and more [00:14:07] importantly, rather than [00:14:08] fighting over that [00:14:09] point estimate, [00:14:10] you can look at the [00:14:11] distribution. And I [00:14:12] think that's one of the things [00:14:13] that Moody's has done is [00:14:14] we've gone out and we've said, [00:14:15] "Listen, we [00:14:16] get the whole thing on cap [00:14:17] rates.

We're [00:14:18] gonna put a [00:14:19] distribution so you can get a [00:14:20] feeling as [00:14:21] to, all [00:14:22] right, I disagree with [00:14:23] your number, [00:14:24] but my number's [00:14:25] still within your range," [00:14:26] whether it's the [00:14:27] 25th, 75th [00:14:28] percentile, whatever it [00:14:29] is. [00:14:30] Okay, I get some, some [00:14:31] comfort or, or it's [00:14:32] an outlier. [00:14:33] If I'm a lender and I [00:14:34] see an outlier, [00:14:35] okay, I better pay [00:14:36] attention.

So I think there [00:14:37] is value there. [00:14:38] I don't think it's... [00:14:39] It, it was [00:14:40] beloved, and then [00:14:41] it became [00:14:42] hated, and it's probably [00:14:43] somewhere in [00:14:44] [00:14:45] between.

Edward: Ah, so funny. [00:14:46] I think the, [00:14:47] blessing and the curse of cap [00:14:48] rates is that they [00:14:49] do, a [00:14:50] tremendous job and [00:14:51] a, an almost too [00:14:52] easy job of, of [00:14:53] telling the story in a [00:14:54] quick way, [00:14:55] on, on what [00:14:56] the conveyor is [00:14:57] trying to tell. Whether [00:14:58] you're a broker [00:14:59] trying to show that [00:15:00] the, the upside of [00:15:01] the property's there, [00:15:02] but you, [00:15:03] you know, [00:15:04] maybe didn't count [00:15:05] the [00:15:06] renovations as much as they [00:15:07] should have, or [00:15:08] you took, you [00:15:09] know, a [00:15:10] better [00:15:11] version, a [00:15:12] more, sharper [00:15:13] version of the, [00:15:14] income that you can [00:15:15] actually expect in the [00:15:16] next year.

Maybe it'll take you [00:15:17] two or three years to get [00:15:18] there.

Jeffrey: Right, or there's [00:15:19] one-time revenue [00:15:20] in that [00:15:21] income and [00:15:22] you didn't realize that was [00:15:23] one-time revenue, [00:15:24] you

Edward: Yeah. [00:15:25] so let's move on [00:15:26] to [00:15:27] [00:15:28] AI, MCP, [00:15:29] and the new data [00:15:30] infrastructure. So [00:15:31] Moody's just [00:15:32] launched an [00:15:33] MCP, that lets [00:15:34] AI agents draw [00:15:35] directly on your [00:15:36] ratings and your [00:15:37] research and your risk [00:15:38] intelligence across [00:15:39] more than [00:15:40] 600 million entities. [00:15:41] So [00:15:42] first, for those who are [00:15:43] unfamiliar, tell us [00:15:44] what an MCP [00:15:45] is and does, [00:15:46] and what [00:15:47] does that change for someone [00:15:48] making a commercial [00:15:49] real estate decision [00:15:50] today?

Jeffrey: [00:15:51] So MCP [00:15:52] just puts the [00:15:53] data in a [00:15:54] format it's [00:15:55] easy to use for a [00:15:56] large language model, [00:15:57] and therefore [00:15:58] you would be spending less [00:15:59] tokens. [00:16:00] Ideally, it's going to be [00:16:01] more cost [00:16:02] efficient for [00:16:03] you to use a [00:16:04] large language [00:16:05] model. And again, it doesn't [00:16:06] matter which one, [00:16:07] whether it's [00:16:08] Claude, [00:16:09] ChatGPT, [00:16:10] Copilot, Gemini, [00:16:11] whatever. [00:16:12] It lets [00:16:13] the [00:16:14] model [00:16:15] access and [00:16:16] use the data more [00:16:17] efficiently, so you [00:16:18] get a quicker [00:16:19] answer. But ideally, [00:16:20] you're also using less [00:16:21] tokens, so [00:16:22] the cost is [00:16:23] lower as part of [00:16:24] that [00:16:25] We've done that not [00:16:26] only for the data [00:16:27] you mentioned, but Moody's is [00:16:28] actually doing that for [00:16:29] every piece of data [00:16:30] we have. [00:16:31] So our economics [00:16:32] data is already [00:16:33] in MCP [00:16:34] format, and [00:16:35] our CRE data [00:16:36] will be in [00:16:37] MCP format, , by [00:16:38] the end of the year.[00:16:39]

So our goal is [00:16:40] to make it available. [00:16:41] We are agnostic as [00:16:42] to which large language [00:16:43] model you [00:16:44] use, but we wanna make [00:16:45] our data [00:16:46] available to [00:16:47] you [00:16:48] so that you can [00:16:49] use your large language [00:16:50] model, whatever AI [00:16:51] or however [00:16:52] you use [00:16:53] AI because of [00:16:54] that. Now, in the meantime, [00:16:55] even though [00:16:56] we don't have all of our [00:16:57] CRE data [00:16:58] in MCP format, [00:16:59] we [00:17:00] have a smart [00:17:01] API.

So you [00:17:02] can access it, [00:17:03] you can dump [00:17:04] it into your [00:17:05] system. It's [00:17:06] just gonna take you a few more [00:17:07] tokens [00:17:08] potentially [00:17:09] to do it because the [00:17:10] data isn't [00:17:11] quite as crisp and [00:17:12] clean for the [00:17:13] large language [00:17:14] model to use. I mean, it's [00:17:15] still good, but it's [00:17:16] not [00:17:17] primed for [00:17:18] large language model [00:17:19] usage. It's not [00:17:20] optimized for that.

[00:17:21] I think people were [00:17:22] worried. It's funny, uh, [00:17:23] for us, that is [00:17:24] a [00:17:25] necessary step. [00:17:26] But if you look at real estate [00:17:27] data, everyone's like, "Well, wait, you're [00:17:28] giving it away." I was like, [00:17:29] "No, I'm not." [00:17:30] Yes, the [00:17:31] history is there, [00:17:32] but [00:17:33] every day, [00:17:34] every month, every [00:17:35] quarter, [00:17:36] there are new [00:17:37] transactions. there are [00:17:38] new buildings being [00:17:39] built.

So [00:17:40] there's new data, [00:17:41] [00:17:42] so if you don't [00:17:43] continue to [00:17:44] subscribe to the [00:17:45] [00:17:46] information, you're not gonna know the [00:17:47] latest cap rate [00:17:48] or [00:17:49] transaction pricing or [00:17:50] price per square [00:17:51] foot or lease [00:17:52] transaction [00:17:53] or new buildings [00:17:54] being constructed. [00:17:55] Again, you may be able [00:17:56] to go hunt those [00:17:57] down, [00:17:58] but you've then just [00:17:59] defeated the purpose of [00:18:00] having the data in [00:18:01] MCP format [00:18:02] if you're gonna have to go [00:18:03] pay to add onto it [00:18:04] yourself [00:18:05] rather than getting all of [00:18:06] that information in [00:18:07] MCP.[00:18:08]

So [00:18:09] I'm not worried [00:18:10] about those [00:18:11] data providers that say, "Well, wait a [00:18:12] second. If they use it, they're never [00:18:13] gonna subscribe." I'm [00:18:14] like, "No, [00:18:15] our data keeps [00:18:16] changing. If they don't [00:18:17] subscribe, they'll lose the [00:18:18] flow, [00:18:19] and that's gonna [00:18:20] be a problem." [00:18:21] So I'm not [00:18:22] worried about that,

And [00:18:23] again, th-those large [00:18:24] language models are gonna be a [00:18:25] little bit different if you're a [00:18:26] lender [00:18:27] versus [00:18:28] you're a, [00:18:29] um, [00:18:30] servicer. But [00:18:31] you can then [00:18:32] automate your [00:18:33] processes. You can automate [00:18:34] your comp set, [00:18:35] you can automate [00:18:36] your [00:18:37] credit [00:18:38] memo, you can automate [00:18:39] your offering [00:18:40] memorandum. You [00:18:41] can automate a [00:18:42] big portion [00:18:43] of the [00:18:44] workflow, but you're [00:18:45] still gonna need a human [00:18:46] in the loop, and I know we wanna [00:18:47] get to that.

[00:18:48] But I think that there's a [00:18:49] lot you can [00:18:50] do that will make you [00:18:51] more efficient, [00:18:52] so the [00:18:53] human in the loop can [00:18:54] have a higher value [00:18:55] add, [00:18:56] ideally either [00:18:57] makes more money [00:18:58] or has less risk[00:18:59]

Edward: Yep, [00:19:00] yep. [00:19:01] so [00:19:02] Moody's is [00:19:03] a [00:19:04] platform-agnostic AI, [00:19:05] meaning you're not [00:19:06] picking a winner [00:19:07] between, say, OpenAI, [00:19:08] Anthropic, [00:19:09] Google Gemini, [00:19:10] Microsoft. so [00:19:11] how does that bet [00:19:12] play out and, and [00:19:13] what does it mean for [00:19:14] commercial real estate [00:19:15] firms trying to figure out [00:19:16] which [00:19:17] AI infrastructure to [00:19:18] build [00:19:19] on?

Jeffrey: Well, [00:19:20] part of that is because a [00:19:21] lot of them have chosen [00:19:22] different things. We also have, uh, [00:19:23] have our data in [00:19:24] Snowflake, [00:19:25] we have it [00:19:26] in, um, [00:19:27] Databricks, right? So, we [00:19:28] wanna meet the customer [00:19:29] where they are, [00:19:30] and [00:19:31] there are different [00:19:32] reasons to [00:19:33] use different, [00:19:34] providers. [00:19:35] I know, [00:19:36] and again, this is m- me [00:19:37] personally, this is [00:19:38] not...[00:19:39]

I found that [00:19:40] Google Gemini [00:19:41] is just better [00:19:42] when I'm trying to do [00:19:43] stuff with [00:19:44] visualization in terms of [00:19:45] pictures or [00:19:46] [00:19:47] images, versus I [00:19:48] like Anthropic, I [00:19:49] like Claude for [00:19:50] other things, and I even use [00:19:51] Perplexity for [00:19:52] some things. [00:19:53] So I found the large [00:19:54] language models [00:19:55] themselves [00:19:56] are today still a little [00:19:57] different depending on what [00:19:58] I'm trying to [00:19:59] accomplish. [00:20:00] Our thought was, [00:20:01] we're not gonna try to [00:20:02] pick the winner. , We have no [00:20:03] idea, right? I mean, it's [00:20:04] changing so [00:20:05] fast. [00:20:06] So we wanna meet our [00:20:07] clients where they [00:20:08] are by [00:20:09] letting them choose whatever [00:20:10] works best for [00:20:11] them, and [00:20:12] allowing them to be most [00:20:13] efficient with the use of [00:20:14] our data so [00:20:15] they can get what they [00:20:16] need and what they're trying to [00:20:17] [00:20:18] accomplish

Edward: I think that's [00:20:19] a positive [00:20:20] sum mentality [00:20:21] in, in growing the [00:20:22] pie together [00:20:23] instead of, [00:20:24] limiting or, or [00:20:25] picking winners, if you [00:20:26] will

Jeffrey: [00:20:27] exactly, right. My job [00:20:28] is to be your partner, [00:20:29] not to force [00:20:30] something down your throat[00:20:31]

Edward: Right. [00:20:32] Right. [00:20:33] so back to your [00:20:34] point about the human in [00:20:35] the loop. So how, how [00:20:36] do you think [00:20:37] about where the human [00:20:38] stays in the loop when [00:20:39] AI is [00:20:40] processing, commercial real [00:20:41] estate risk data and [00:20:42] other [00:20:43] data?

Jeffrey: So [00:20:44] I still don't [00:20:45] think the models are [00:20:46] great at turning [00:20:47] points. I think they will get [00:20:48] better. [00:20:49] But one of the [00:20:50] things that I, I like [00:20:51] using this [00:20:52] analogy, Do you ever see the movie The [00:20:53] Last [00:20:54] Samurai? [00:20:55] it, [00:20:56] it is a Tom Cruise [00:20:57] movie, exactly, [00:20:58] from a long time ago. [00:20:59] Shows my age.

So [00:21:00] he's, he's hired by [00:21:01] these [00:21:02] villagers to help [00:21:03] them fight against the [00:21:04] samurai.[00:21:05]

And he ends up [00:21:06] switching sides and becomes a [00:21:07] samurai. [00:21:08] But the villagers [00:21:09] then get the latest [00:21:10] technology, which at that point was a [00:21:11] machine gun, [00:21:12] right? This takes place in [00:21:13] the 1800s. [00:21:14] And they end [00:21:15] up defeating the [00:21:16] samurai because the samurai [00:21:17] do not [00:21:18] stay [00:21:19] current with the [00:21:20] technology. Now, [00:21:21] the flip side of that [00:21:22] is, if the [00:21:23] samurai had stayed [00:21:24] current with that [00:21:25] technology, [00:21:26] their use of [00:21:27] tactics and [00:21:28] knowledge would have [00:21:29] made them so much [00:21:30] better.

I [00:21:31] think that's the case. If [00:21:32] you are a CRE [00:21:33] expert, [00:21:34] this is [00:21:35] another tool in your [00:21:36] toolbox and [00:21:37] it can help you with [00:21:38] pattern [00:21:39] recognition, it can help [00:21:40] you with [00:21:41] automating [00:21:42] repetitive [00:21:43] processes. But [00:21:44] there's still something [00:21:45] in your gut [00:21:46] or in your [00:21:47] brain, you [00:21:48] understand, [00:21:49] you know what? [00:21:50] That [00:21:51] property is a left [00:21:52] turn instead of a [00:21:53] right turn, and there's [00:21:54] no traffic [00:21:55] light. It's gonna [00:21:56] be 10% [00:21:57] lower in sales [00:21:58] than this one over [00:21:59] [00:22:00] here. AI at some [00:22:01] point will [00:22:02] probably get there, but at [00:22:03] least [00:22:04] today, [00:22:05] today's real estate [00:22:06] professionals [00:22:07] understand those [00:22:08] nuances that the [00:22:09] models don't [00:22:10] get. And I [00:22:11] think the models [00:22:12] will help you, [00:22:13] but [00:22:14] that [00:22:15] nuance of [00:22:16] understanding the real estate, [00:22:17] walking the [00:22:18] real estate in [00:22:19] addition to the [00:22:20] data, [00:22:21] combined, that is [00:22:22] a much more [00:22:23] powerful source, [00:22:24] and that - leads [00:22:25] to [00:22:26] better [00:22:27] outcomes.

So I think [00:22:28] somebody with the real [00:22:29] estate [00:22:30] expertise that [00:22:31] understands some of those [00:22:32] nuances, [00:22:33] and some of them learn, you're [00:22:34] not able to [00:22:35] verbalize, you just kind [00:22:36] of know it when you [00:22:37] see it, 'cause [00:22:38] you've seen so [00:22:39] much in the [00:22:40] industry, it's [00:22:41] almost automatic in your [00:22:42] brain. It's like driving [00:22:43] a car, right?

You don't think [00:22:44] about putting the key [00:22:45] in or, or putting [00:22:46] your fob or, you [00:22:47] just do it [00:22:48] automatically. Same [00:22:49] thing with real estate. [00:22:50] The data makes you [00:22:51] better, but there's [00:22:52] that [00:22:53] intuition that you [00:22:54] have 'cause you've [00:22:55] seen so much in your [00:22:56] time, whether you're a [00:22:57] lender, a broker, an [00:22:58] investor, whatever the [00:22:59] case may be.

So I [00:23:00] think the [00:23:01] combination of [00:23:02] AI... 'Cause [00:23:03] otherwise, [00:23:04] everybody's getting the same [00:23:05] AI, and we're [00:23:06] all getting the same [00:23:07] outcome, right? [00:23:08] The, the models are all [00:23:09] telling you the same [00:23:10] thing based on the [00:23:11] data. You have [00:23:12] to be able to understand, well, wait a [00:23:13] second, why [00:23:14] is this different? [00:23:15] What is [00:23:16] AI missing? [00:23:17] And I think [00:23:18] that's what's gonna [00:23:19] separate the winners from the [00:23:20] [00:23:21] losers.

Edward: Certainly. [00:23:22] Uh, I think we're [00:23:23] still, people like [00:23:24] to say early innings. I [00:23:25] don't know if we're even [00:23:26] started the game [00:23:27] yet. we will, we [00:23:28] shall see. Um, [00:23:29] talking [00:23:30] about [00:23:31] ESG risk and, [00:23:32] and climate [00:23:33] risk, I mean, just [00:23:34] this week we've [00:23:35] had [00:23:36] [00:23:37] devastating [00:23:38] earthquakes, . I don't know [00:23:39] if earthquakes count as [00:23:40] climate, but they're [00:23:41] certainly nature

Jeffrey: No, the, [00:23:42] we, we [00:23:43] do. We actually have [00:23:44] earthquake as part of [00:23:45] our climate. And [00:23:46] again, because [00:23:47] climate has become [00:23:48] so [00:23:49] politicized, right? [00:23:50] It- it's a term. [00:23:51] We call it [00:23:52] physical risk. [00:23:53] Just as you would look at [00:23:54] interest rate risk [00:23:55] or tenant risk [00:23:56] or anything else, [00:23:57] you should [00:23:58] understand your physical [00:23:59] risk. And it's more than [00:24:00] just, am I in a [00:24:01] floodplain?

[00:24:02] It's what kind of [00:24:03] flooding could I [00:24:04] have on [00:24:05] a [00:24:06] super storm? [00:24:07] Or a, a [00:24:08] non-name storm, [00:24:09] am I in a path for [00:24:10] hail? [00:24:11] And what kind of damage could I [00:24:12] do blowing out my [00:24:13] windows? [00:24:14] Or could a [00:24:15] tornado come through [00:24:16] and hit me? [00:24:17] And understanding [00:24:18] your physical [00:24:19] risk and [00:24:20] the dollar [00:24:21] values that you have, [00:24:22] we have models that [00:24:23] do that.[00:24:24]

So we work [00:24:25] with insurance [00:24:26] companies. We're one of the [00:24:27] biggest providers [00:24:28] of that data, [00:24:29] so we're able to take... [00:24:30] We have over 100 [00:24:31] meteorologists on [00:24:32] staff on [00:24:33] this, So we understand [00:24:34] what the climate [00:24:35] risk is from that, [00:24:36] and then you [00:24:37] apply it to the real [00:24:38] estate, and then you take the [00:24:39] insurance data, and you can [00:24:40] understand what is the [00:24:41] dollar value at [00:24:42] risk [00:24:43] for a particular [00:24:44] building for [00:24:45] different climate [00:24:46] perils. I think [00:24:47] if we get out of the [00:24:48] politics [00:24:49] and [00:24:50] ESG and climate, all the... [00:24:51] and just look at the [00:24:52] physical risk, [00:24:53] yes, [00:24:54] you should be measuring [00:24:55] that the same way you look [00:24:56] at any other risk, [00:24:57] and you're not [00:24:58] doing your fiduciary [00:24:59] duty if you aren't [00:25:00] paying attention to that. [00:25:01] Because there are ways [00:25:02] to mitigate some of [00:25:03] that risk, and if [00:25:04] you aren't doing [00:25:05] that, you are behind the [00:25:06] curve and you're a less [00:25:07] effective [00:25:08] operator

Edward: [00:25:09] Absolutely. I mean, [00:25:10] modeling, [00:25:11] anticipating, taking into account [00:25:12] physical [00:25:13] risk, I actually like the [00:25:14] term, sounds [00:25:15] very [00:25:16] neutral. it's not [00:25:17] about left or right, [00:25:18] it's about [00:25:19] if, there's a [00:25:20] fire, my insurance is gonna [00:25:21] increase. If there's a [00:25:22] flood, my insurance is [00:25:23] gonna increase.

Like, it's [00:25:24] just,

Jeffrey: [00:25:25] Exactly. It's just common [00:25:26] sense, and, uh, [00:25:27] we just have to get the [00:25:28] politics out of it [00:25:29] so you can [00:25:30] focus on what [00:25:31] matters, which [00:25:32] is what is my [00:25:33] risk and how much could it [00:25:34] potentially cost me? [00:25:35] There's a great [00:25:36] story I have on [00:25:37] that, which is, um, [00:25:38] I know of a, an [00:25:39] investment they, they [00:25:40] had... They bought a a two-building [00:25:41] office park, [00:25:42] and there was this giant [00:25:43] lake, with one of the [00:25:44] fountains And [00:25:45] they learned, they [00:25:46] realized that if they [00:25:47] sloped [00:25:48] the land a little [00:25:49] bit more, [00:25:50] they could change [00:25:51] the runoff [00:25:52] from heavy [00:25:53] [00:25:54] rainstorms, and they had less [00:25:55] flooding, [00:25:56] and therefore, they were able to [00:25:57] lower their insurance [00:25:58] costs just by [00:25:59] understanding your physical [00:26:00] risk, it wasn't [00:26:01] climate, it wasn't [00:26:02] ESG.

It was [00:26:03] increasing your [00:26:04] returns by doing the right [00:26:05] thing.

Edward: [00:26:06] Right, right. [00:26:07] Isn't, isn't it wonderful when those [00:26:08] two align?

Jeffrey: [00:26:09] Exactly.[00:26:10]

Edward: Yep, [00:26:11] great. let's, uh, [00:26:12] move on [00:26:13] to, asset [00:26:14] class, that sometimes it's [00:26:15] including within [00:26:16] commercial real estate even though [00:26:17] industrial real [00:26:18] estate is, on [00:26:19] its own [00:26:20] standing as an asset class, [00:26:21] and not only thanks to [00:26:22] the rise of [00:26:23] e-commerce years [00:26:24] ago, but also [00:26:25] now with data [00:26:26] centers being a [00:26:27] sort of specialized [00:26:28] industrial real [00:26:29] estate asset [00:26:30] class.

[00:26:31] but, it's been a, a [00:26:32] consensus trade in many [00:26:33] pockets of commercial real [00:26:34] estate for years. [00:26:35] tell us about some of the [00:26:36] risks that the market is [00:26:37] not pricing in right [00:26:38] now for [00:26:39] traditional [00:26:40] industrial real [00:26:41] estate.[00:26:42]

Jeffrey: Absolutely. I'm glad you asked, 'cause [00:26:43] this is one of [00:26:44] the things, that I don't [00:26:45] think the market is fully [00:26:46] pricing the risk. [00:26:47] We are going through [00:26:48] such a [00:26:49] change. We [00:26:50] are seeing [00:26:51] a [00:26:52] reallocation of [00:26:53] distribution [00:26:54] networks because of the [00:26:55] [00:26:56] change of what's going [00:26:57] on with tariffs [00:26:58] and China [00:26:59] [00:27:00] nearshoring, [00:27:01] some onshoring [00:27:02] and production within [00:27:03] the US [00:27:04] where new factories [00:27:05] are going, that it's [00:27:06] changing our [00:27:07] distribution network.[00:27:08]

You combine that [00:27:09] with [00:27:10] driverless trucks that are [00:27:11] coming, they're already being [00:27:12] tested. [00:27:13] In addition [00:27:14] to the modern [00:27:15] warehouse, we have [00:27:16] to think in cubic [00:27:17] feet as we've gone from 32 [00:27:18] to [00:27:19] 36 to even [00:27:20] 40 feet. And [00:27:21] again, there aren't as [00:27:22] many 40-footers. [00:27:23] I get it, [00:27:24] but they are coming and [00:27:25] people are building more [00:27:26] of them.

You [00:27:27] combine the [00:27:28] robots, you combine the [00:27:29] new racking [00:27:30] systems, you think in [00:27:31] cubic feet, [00:27:32] the throughput is [00:27:33] that much more [00:27:34] efficient. So even if [00:27:35] the demand for [00:27:36] industrial goes up, [00:27:37] the need for [00:27:38] more space may [00:27:39] not [00:27:40] because you are more [00:27:41] efficient. And then what [00:27:42] happens to those [00:27:43] older [00:27:44] buildings, [00:27:45] 32 or even [00:27:46] 36 square feet, that [00:27:47] don't have the [00:27:48] power [00:27:49] or the [00:27:50] clearance or whatever, [00:27:51] or the...

to [00:27:52] be able to [00:27:53] get that modern [00:27:54] throughput that's [00:27:55] [00:27:56] necessary, the price of those, [00:27:57] they become not [00:27:58] functioning obsolete, but [00:27:59] functionally [00:28:00] obsolete. And we've [00:28:01] seen that with [00:28:02] retail and the Great Recession, [00:28:03] we've seen it with office [00:28:04] buildings from [00:28:05] COVID, and I think [00:28:06] it's coming for [00:28:07] industrial, and I don't [00:28:08] think the market is [00:28:09] fully aware or fully [00:28:10] pricing that [00:28:11] in.

And then [00:28:12] power's a huge issue, [00:28:13] right? We see that with data [00:28:14] centers. Do [00:28:15] you have the power to [00:28:16] run your modern [00:28:17] warehouse?

Edward: [00:28:18] I really haven't [00:28:19] seen this being spoken a [00:28:20] lot out there, and [00:28:21] truly, uh, [00:28:22] under the hood [00:28:23] risk here. [00:28:24] so [00:28:25] much of, [00:28:26] progress and [00:28:27] growth and scale [00:28:28] comes from, like, [00:28:29] speaking the same language, [00:28:30] having the same concept, [00:28:31] standardization. [00:28:32] I always think [00:28:33] about, when [00:28:34] [00:28:35] containers were [00:28:36] standardized for shipping, [00:28:37] the, the 20 by [00:28:38] 40 feet, you know, [00:28:39] container.

It's something that [00:28:40] we don't [00:28:41] think about. Of course, [00:28:42] we take it for granted. But that [00:28:43] was a [00:28:44] huge, huge, [00:28:45] innovation, [00:28:46] just standardizing [00:28:47] across the world so we can just [00:28:48] trade in [00:28:49] mass, to [00:28:50] get, economies of scale [00:28:51] and get [00:28:52] affordable products. [00:28:53] so the [00:28:54] fact that now [00:28:55] what is needed [00:28:56] for [00:28:57] warehouses, for [00:28:58] distribution is, [00:28:59] is changing, [00:29:00] and we're not taking into [00:29:01] account how [00:29:02] to power them, how to [00:29:03] operate them [00:29:04] in a, in an AI-powered, in [00:29:05] a robotics [00:29:06] way.

Yeah, I wonder [00:29:07] if there'll be-- [00:29:08] how many of those [00:29:09] will be [00:29:10] the office [00:29:11] buildings, of a few [00:29:12] years ago [00:29:13] that, can't find [00:29:14] tenants because people don't [00:29:15] need to be in the office as [00:29:16] frequently in a [00:29:17] location that isn't as [00:29:18] desirable with [00:29:19] no amenities.[00:29:20]

Jeffrey: [00:29:21] Exactly. And then [00:29:22] also on the last [00:29:23] mile, right? We haven't talked [00:29:24] about [00:29:25] drones or [00:29:26] self-driving cars or [00:29:27] trucks, [00:29:28] and how that may [00:29:29] impact where you wanna [00:29:30] locate [00:29:31] for that type [00:29:32] of distribution. [00:29:33] And then how is [00:29:34] 3D [00:29:35] printing going to impact [00:29:36] things? So there's a lot [00:29:37] of change coming to [00:29:38] the industrial [00:29:39] market, maybe [00:29:40] not tomorrow, but in [00:29:41] the next couple [00:29:42] years, and we don't [00:29:43] know how it's gonna play [00:29:44] out.

I don't think [00:29:45] people are accounting for all [00:29:46] of that [00:29:47] risk

Edward: I [00:29:48] don't know if everyone has [00:29:49] seen the [00:29:50] robotics Olympics that [00:29:51] they're playing out in [00:29:52] China. I mean, there's some [00:29:53] very funny videos, [00:29:54] some inspiring [00:29:55] videos, and also [00:29:56] some somewhat scary [00:29:57] videos. so, and a lot [00:29:58] of the industrial buildings were [00:29:59] designed for humans [00:30:00] and forklifts [00:30:01] naturally. So how does the [00:30:02] shift to [00:30:03] robotics change what a [00:30:04] building is [00:30:05] worth?

Jeffrey: I think [00:30:06] that's a problem, [00:30:07] right? I think if you don't [00:30:08] have the ability [00:30:09] to have those super [00:30:10] flat [00:30:11] floors to [00:30:12] run... I mean, not the-- [00:30:13] Some of those robots that we've [00:30:14] seen running in [00:30:15] China, you know, they're [00:30:16] falling down, but there, there are [00:30:17] other robots that are [00:30:18] based on wheels or other [00:30:19] types of [00:30:20] treads, but they need really [00:30:21] flat floors, and they [00:30:22] need the paints on the [00:30:23] lines to keep them [00:30:24] aligned, and they need the [00:30:25] power, [00:30:26] and they need to...

[00:30:27] You understand, wait, [00:30:28] how do I keep [00:30:29] the humans out of [00:30:30] the way of the [00:30:31] robots so there [00:30:32] aren't collisions and [00:30:33] injuries? [00:30:34] And if [00:30:35] you haven't [00:30:36] designed it that way, [00:30:37] you almost have [00:30:38] to take [00:30:39] everything out [00:30:40] and redesign [00:30:41] it versus it's [00:30:42] probably easier to [00:30:43] move [00:30:44] what happens to that old [00:30:45] warehouse, then that [00:30:46] needs to be redesigned or [00:30:47] repurposed [00:30:48] to combine [00:30:49] human [00:30:50] vers- and tech- [00:30:51] other types of [00:30:52] technology.

so I think [00:30:53] that's an issue g- uh, [00:30:54] forward that hasn't been [00:30:55] addressed.

Edward: [00:30:56] Mm-hmm. if you had [00:30:57] to identify one [00:30:58] macro data [00:30:59] point the commercial real [00:31:00] estate industry is not [00:31:01] paying nearly enough [00:31:02] attention to, [00:31:03] what would it [00:31:04] be?

Jeffrey: [00:31:05] That, that is a tough [00:31:06] question. [00:31:07] My thing has always [00:31:08] been, [00:31:09] productivity, so [00:31:10] I, I'm gonna stick with [00:31:11] that. And I don't think we [00:31:12] pay enough [00:31:13] attention to [00:31:14] the changes in [00:31:15] productivity, 'cause that's [00:31:16] gonna impact both [00:31:17] wages and labor [00:31:18] force [00:31:19] and the hiring.

[00:31:20] So I think people need to [00:31:21] pay more attention. Now, it's [00:31:22] a very noisy [00:31:23] number, so I would look [00:31:24] at it as a trailing [00:31:25] 12 [00:31:26] moving average rather [00:31:27] than that [00:31:28] particular quarter, 'cause it [00:31:29] does jump around. But I [00:31:30] think productivity is [00:31:31] probably the one that I [00:31:32] would pay attention to[00:31:33] [00:31:34]

Edward: Productivity. [00:31:35] Interesting. [00:31:36] And there's naturally a [00:31:37] lot of talk about [00:31:38] productivity gains [00:31:39] because of AI, [00:31:40] right? And is [00:31:41] doing the same [00:31:42] amount of work but in [00:31:43] half the [00:31:44] time, does that mean we're being [00:31:45] more productive? Is [00:31:46] if the other [00:31:47] half of the time we're not [00:31:48] doing anything or we're [00:31:49] just having fun and [00:31:50] making ourselves more [00:31:51] happy so that we can do [00:31:52] better work?

[00:31:53] Like, it's gonna be a [00:31:54] philosophical [00:31:55] fundamental question as well on, [00:31:56] on what [00:31:57] productivity means and how [00:31:58] to measure [00:31:59] it and prioritize [00:32:00] it.

Jeffrey: [00:32:01] Exactly. you have [00:32:02] both the quality of [00:32:03] life in addition to the [00:32:04] quantity of the [00:32:05] output

Edward: [00:32:06] So last [00:32:07] but not least, [00:32:08] collaboration [00:32:09] superpower. If you could partner [00:32:10] with one person, [00:32:11] historic or [00:32:12] living, real or [00:32:13] fictional, who [00:32:14] would it [00:32:15] be?

Jeffrey: I always thought [00:32:16] Abe Lincoln [00:32:17] and [00:32:18] just his ability [00:32:19] to bring [00:32:20] together so [00:32:21] many disparate [00:32:22] parties, [00:32:23] right? the cabinet of [00:32:24] rivals. And I [00:32:25] always found that [00:32:26] fascinating, and especially [00:32:27] given in today's [00:32:28] world, right, where we're [00:32:29] so [00:32:30] polarized, the [00:32:31] ability to [00:32:32] work with [00:32:33] all sorts of different [00:32:34] groups and have [00:32:35] them [00:32:36] get along, [00:32:37] that to me is [00:32:38] a, a, a true [00:32:39] superpower that I've [00:32:40] always been amazed with, [00:32:41] and I, I would [00:32:42] love to be [00:32:43] better or at least learn how to [00:32:44] do that better.

[00:32:45] so that, that would be [00:32:46] what I would [00:32:47] choose.

Edward: Love [00:32:48] that. And to do that [00:32:49] better, we need to [00:32:50] tell, better stories, [00:32:51] and we do that with better [00:32:52] data. So you're [00:32:53] certainly in, in the right [00:32:54] place and, and the right [00:32:55] person to do that. [00:32:56] where can Tangent [00:32:57] listeners and [00:32:58] commercial real estate [00:32:59] lenders, servicers, [00:33:00] investors, and operators [00:33:01] learn more [00:33:02] about Moody's [00:33:03] Analytics and connect [00:33:04] with you?

Jeffrey: [00:33:05] so you can always find [00:33:06] me on LinkedIn, [00:33:07] Jeffrey [00:33:08] Havsy. And then [00:33:09] Moody's [00:33:10] CRE [00:33:11] and Moody's [00:33:12] Economics [00:33:13] are two places [00:33:14] where we always w- have a [00:33:15] tremendous amount of [00:33:16] content that we put out [00:33:17] there. We're writing [00:33:18] blogs, [00:33:19] white papers, [00:33:20] data. we're happy to [00:33:21] have you come visit [00:33:22] us at, at Moody's [00:33:23] CRE or, or [00:33:24] Moody's [00:33:25] Economics

Edward: Today [00:33:26] we covered why [00:33:27] having more [00:33:28] commercial real estate [00:33:29] data is not the same as [00:33:30] having better [00:33:31] information, and what it [00:33:32] takes to combine [00:33:33] property fundamentals [00:33:34] with the economic, [00:33:35] credit, and [00:33:36] environmental signals that [00:33:37] predict what happens [00:33:38] to an asset before [00:33:39] it shows up in the [00:33:40] numbers. [00:33:41] We've talked about why [00:33:42] cap rates are one of the [00:33:43] most manipulated [00:33:44] metrics in the industry, [00:33:45] where AI [00:33:46] deployment goes wrong [00:33:47] when humans stop [00:33:48] checking the outputs, and [00:33:49] what asset [00:33:50] level [00:33:51] physical risk [00:33:52] questions nobody is [00:33:53] asking yet.

[00:33:54] And we got into the [00:33:55] industrial real estate [00:33:56] risks not showing [00:33:57] up in most [00:33:58] underwriting models today, [00:33:59] from [00:34:00] functional obsolescence [00:34:01] for robotics [00:34:02] to the [00:34:03] geographic reshuffling [00:34:04] that [00:34:05] self-driving trucks will [00:34:06] eventually force. [00:34:07] [00:34:08] Jeffrey, thanks [00:34:09] so much for coming to [00:34:10] Tanium today

Jeffrey: [00:34:11] Thank you for having [00:34:12] me. It was a great [00:34:13] time, and I look forward to [00:34:14] doing it again in the [00:34:15] future [00:34:16] [00:34:17] [00:34:18] [00:34:19]

Edward: Muchas [00:34:20] gracias. [00:34:21] If you learn something [00:34:22] new or enjoy the [00:34:23] conversation. [00:34:24] Text a friend the [00:34:25] link to this episode [00:34:26] right now.

[00:34:27] [00:34:28] You can find [00:34:29] links, resources, and [00:34:30] ways to connect with [00:34:31] us in the show [00:34:32] notes [00:34:33] [00:34:34] [00:34:35] [00:34:36]