Moody's Talks - Inside Economics - Housers on the Housing Shortage

Episode Date: July 16, 2025

Moody’s Analytics Mark Zandi and Cris deRitis are joined by Ira Goldstein from The Reinvestment Fund, Maggie McCullough from PolicyMap, and Jim Parrott from the Urban Institute to discuss their new study that takes a deep dive into understanding the nature of the decade-long housing shortfall. This housing crisis has driven up house prices and rents, and undermined housing affordability. But despite the heightened political attention on the problem, there remains confusion over its true scale and scope. This team of self-avowed housers dissect the shortage down to the census tract and come to some surprising conclusions.To learn more and access the full research paper: https://www.economy.com/bringing-the-housing-shortage-into-sharper-focusGuest: Ira Goldstein, Senior Advisor at The Reinvestment FundGuest: Maggie McCullough, CEO and Founder of PolicyMapGuest: Jim Parrott, Nonresident Fellow at the Urban InstituteHosts: Mark Zandi – Chief Economist, Moody’s Analytics, Cris deRitis – Deputy Chief Economist, Moody’s Analytics, and Marisa DiNatale – Senior Director - Head of Global Forecasting, Moody’s AnalyticsFollow Mark Zandi on 'X' and BlueSky @MarkZandi, Cris deRitis on LinkedIn, and Marisa DiNatale on LinkedIn Questions or Comments, please email us at [e-mail]. We would love to hear from you.  To stay informed and follow the insights of Moody's Analytics economists, visit Economic View. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Welcome to Inside Economics. I'm Mark Sandy, the chief economist of Moody's Analytics, and this is a special bonus podcast. We're going to be talking about a housing paper, a new paper that's coming out, bringing the housing shortage into sharper focus. And it's an important paper because we've got a star-studded cast of housers who participated in the research and in the writing. I'm one of them. But the other one is my sidekick. my co-host on Inside Economics, Chris DeReedies, Chris, good to have you on. Good. You're a Hauser, right? I think so. Yeah. Consider myself one. You do that as a badge of honor, not some kind of.

Absolutely. Absolutely. Right. Yeah, absolutely. And you hail from before Moody's days from Fannie Mae, right? That's right. Would you consider yourself a modeler? Like if I said you were a Fannie Mae modeler, you'd be okay with that?

Credit modeler, yeah. For sure. Okay, okay, very good. Well, you were on board and we had Maggie McCullough, Maggie. Good to have you on Inside Economics. Good to be here. Maggie is the founder and CEO of PolicyMap. I call it PMAP for short.

Greatest, and I should say for sake of disclosure, I'm on the board of PMAP. I'm the lead director. And PMAP, policy map, when you want to describe policy map? real quickly for the listener? Sure. It's a data mapping and visualization application that allows people to, what the goal is to allow people to very easily understand what life is like in neighborhoods around the United States.

And so to that end, we collect data, every public data we can get our hands on, proprietary data that we license, as well as data that we create, like what we're going to discuss today, to really understand what, what life is like in neighborhoods, how it's changing over time, whether it's related to housing, economics, jobs, et cetera. Got it. And policy map's been around for quite some time. I remember you knocked on my door first thinking about it.

It might have been 15 years ago or something. It was 15 years ago. It was. It was. So we started out within a nonprofit reinvestment fund, which is where IRA is. And I was working with IRA at the time. and spun out as our own small business in 2016, I believe. Yeah, very good.

Great company, great people. And Ira, the aforementioned Ira Goldstein. Hi, Ira. Hi, Ira is a legend in the housing world. He started the research arm, which is called Policy Solutions for the Reinvestment Fund. Reinvestment Fund is a CDFI, community development financial institution, one of the largest in the country. And I was on the board for many years at the reinvestment fund, but I got to know IRA very well. And Iri, you're now, you're helping the city of Philadelphia with its housing policy.

In fact, I understand it that you're the one who came up with the 30,000 housing goal that the city has. Is that right? Is that just a rumor? Did you actually come up with that goal? That's definitely a rumor. where we help this figure out how the mayor's promise as a candidate to create or preserve 30,000 housing units should be a portion between the production of units and the preservation of the units and trying to help her sort of understand what parts of the city would be needed to be served under that goal.

Yeah, well, great to have you on board as well. And, of course, our good friend and longtime collaborator, Jim Parrott, hey, Jim. Hey, Mark. Jim, former Obama official back during the financial crisis days in the wake of that and all the things are going on with housing policy, mortgage finance, and, you know, Fannie Mae and Freddie Mac, and now at the Urban Institute and you have our own consulting firm. And you've been on Inside Economics in the past, and I've been meaning to invite you back, but here you are. I only make the bonus cut.

I don't know if that's like, you know, I don't know how to think about that. When you introduce this a bonus track. it occurred to me, you know, I think he said that every time I've been on this show. I don't know whether it's me or you or like, how to do you? That's a reasonable point. No, no, no. I think the first time you were on, that was like, you know, that was like in the flow of the thing. This doesn't quite fit, you know, in like the weekly podcast.

This is something special. This is very special. You're special. Okay, I thought of it that way. That makes me feel better. And of course, this is this group collaborated on the, the, the paper I described, and obviously it goes to the housing shortage, a lot of debate, discussion around that. And so I thought, Jim, we'd start with you. Maybe you could just kind

of frame things for us and give you the listener a sense of, you know, what issues we tackled in the paper and how we went about doing that. But, you know, how should we think about what we did here? Yeah. So it's worth going back to sort of why we did this, at least how I think about it. You guys, folks should correct me on the phone if they came at this with different motivations. But we've been talking about this for years, right? We've been thinking about and explaining the issue to policymakers to the press, to stakeholders, going back, you know, four or five years, I guess, at this point. And in those conversations, two things kept coming up, for me at least.

One is there's this wild variation, wide and wild variation in estimates, right? So you get on the phone to talk to the press about what kind of problem you've got. And immediately, they're just trying to understand, is it a 4 million unit problem? Is it a 2 million unit problem? Is it a 5 million? How do I think about this? Policymakers ask the same question. And it's a little frustrating to have that level of variation because it really does give you the sense

you know what the hell you're talking about if you really can't come up with a coherent answer as to why it's a two million unit problem versus a five million unit problem that's a wide variation so that's sort of problem one i think and then problem two is why we're talking about this in national terms at all like you've got this tendency when we have these discussions to anchor the conversation in some number whether it's two or five or whatever um and which is and we take that for granted but it's a curious thing because obviously the nation is made up of tens of thousands of markets. You know, some are markets to buy, some are markets to rent, some are in California, some are in North Carolina, some are in Iowa, some are markets for million-dollar mansions,

some are markets for lower income rental. And yet all of those markets, which each have their own supply, demand balance, imbalances, they're all netted against each other in this grand single number. And that's always struck me as a strange way to have the conversation. And much the way, and we use this analogy in the paper, in much the way that if I were deciding, I wouldn't go to the beach this weekend and we're going to look at the weather for the beach, if the AI engine came back and said the average temperature in the western hemisphere is 70 degrees,

that doesn't do me much good in figuring out what to pack for my trip to the beach. In the same way telling someone in the U.S. that, you know, the surplus in the country or the shortfall in the country is X seems to be only modestly relevant to almost any question you could imagine having to answer. So with all that as a background, we then did a deeper dive into the supply demand dynamics at a census track level. And we did that as a way, and you guys can explain it in more detail a second, but we did that as a way to begin to approximate the top of the market versus the bottom versus the middle,

and then we broke out rental versus home ownership. And once you begin to break down to the census tract level, which comes a proxy for income, we are at the top or the bottom of the market, and then you break it out into rent and to own, then you begin to see the nation's sort of aggregate supply demand and balance in terms of the kind of markets that matter. So if you're a consumer going to buy a house or rent a house, you're looking at, you know, a discrete kind of supply, which is either for rent or for own, is for sale for X or versus Y or to rent versus X. Y. We tried to show the picture at a national level that would seem relevant to a consumer that's

actually going to look for housing in a particular market. So that was the challenge. And then one of the things that was interesting, which we might want to get into at some point, is we started this a long time, an awful long time ago. And one of the reason why it took us along to get over the finish line was it was a little counterintuitive. Like we came back, you know, we spit all our data into the great, you know, model machine. And it spit back an answer that took us all a little a back, which is really good. I took that to be, it was very frustrating on the front end, but over time, it was a nice example of the way policy research probably should work. You go into it with some intuitions about what's going on.

The analysis may cut against those intuitions a little bit or forces you to sort of recalibrate a little bit. And so you come out of it realizing that having learned a little bit about the state of the world. that you were trying to come to understand, which I feel like is where we came out with the paper. So does that work as sort of a pre? Yeah, that's a nice synopsis. I mean, now that you mentioned it, when did we start this work?

Well, we started it back when there was another administration running the show. Oh, that's right. What we might inherit by way of administration this time around was very much up in the air. Right. So if you remember back in pre-known, November of last year, you know, there's all this talk of a giant supply-side effort of some kind or another. And given the uncertainties I mentioned before about variation in numbers and the fact that we were talking at a national level at all, it gave me at least the sense that in advising policymakers about how to go big, we're flying kind of blind. You're kind of making stuff up about what it is you were solving for when you couldn't tell them whether the number was seven or three and you couldn't tell them.

anything more granular than at a national level is at a rental problem versus a homeowner issue problem. So it's, this whole thing started when we were getting a lot of asks from policymakers about how to go big on this back when it looked like we might have, you know, a legislature and a president that really wanted to go big. And maybe we still do from a different side of the aisle. But anyway, that's kind of how it started. But this was almost a year ago. I was going to say, I think it's almost a year to the day. Yeah. And then we were finished in the fall, and we were talking about releasing it. And I think it was a combination of election and new administration coverage,

plus the new ACS data was coming out. And we said, why are we going to release this now when the new ACS data is coming out? And we'll talk about it later. But it is our goal, hopefully, to update this annually, which is going to be something cool to see how it changes over time as well. In ACS, American Community Survey. The annual census of your information. Yeah, yeah, I've forgotten about that.

Hey, Jim, just as a teaser, because you make a great point. We did this a number of iterations on this and got some non-intuitive results. Can you just want to, what's the one thing, one result that came out that feels still non-intuitive to you? Is there? I'm not sure it's. well, the result that really caught me off guard is you and I and others, everybody else, when I had read it, a lot of other people are saying this too, you and I have been arguing in print now and all over the place for a long time that the shortfall is deepest at the bottom of the market.

And it's shortest at the bottom of the market because the math doesn't make sense to be. build at the bottom of the market. And so with that in mind, if I had to guess, I would have guessed we would have had this nice linear distribution where the top of the markets oversupplied are kind of okay. The middle of the markets may be a little undersupplied, but not super far off. And the bottom market is just a big mess

because it doesn't make any economic sense to build the bottom market. So that was one assumption. And the other assumption was, Anyway, that was my main operations going in. When we got back, when the teacher gave us back our homework, and that was not the answer the teacher gave us. I mean, I initially assumed that we screwed up, like the data's wrong. There was just something wrong in what we'd done.

And that led to a lot of analytic soul searching around what the data was telling us and why and how to reassess or operating assumptions going in. Anyway, so to me, that was the biggest, oh, wait a minute, what, like, what's going on? Are we wrong or our assumptions wrong? So that was kind of interesting. Yeah, very cool. So just to summarize, national housing shortage, it feels like a coast-to-coast pretty big, developed since the great financial crisis.

a lot of the debate discussion is around the kind of the top line number nationwide how many homes are we short rent homeownership the total shoot match compared to what would be sufficient in a well-functioning housing market and so the paper takes a crack at going through all the different estimates that are out there and you know why there are differences because there are some as you mentioned some big differences between these estimates and then we started drilling down to states, metros, then to the census track. The state

and metro work, I think that was kind of led by Chris. And Chris, maybe I'll just turn to you and you can give us a sense of, you know, what you did. And here, the work that Chris did really set the kind of

methodological framework for doing this at the census track. So we kind of learned how to think about the shortfall, the housing shortfall at a national state and metro area level and use that as a basis for the work we did at a census track. So, Chris, maybe you can take the baton here and give us a sense of what the estimates are at the national level, some of the bottom line numbers, and what you learned about the shortage looking across the country at different states and metro areas.

Yeah, yeah, absolutely. So I think it would be helpful to just walk through very briefly, kind of the overall methodology. Yeah, right. Right. So if you're talking about housing shortage or surplus or any kind of good shortage or surplus, basically you're saying that there's something that that market must not be in equilibrium, right? And it's not clearing. And that's clearly the case for housing. We know that there are building constraints, there are lags to construction. So the housing markets are in flux, right? They're constantly adjusting to new demand or supply factors. So key to estimating the shortage or the surplus in a housing market is to come up with some concept of equilibrium. Right. What is the

equilibrium for a housing market, whether you define that as the nation, the state, the metro, or the census tract, or even smaller, really you're coming up with some estimate of what equilibrium is. And the equilibrium conceptually is this idea that we have enough housing to provide for everyone, for everyone in the population, so we're accommodating, we have enough housing to shelter everyone. Plus, we also have to account for mobility. The fact is that most people don't live and die in their homes, they move around, right?

People move for education, for jobs, different points in their life cycle. So there's constant movement. There's dynamism in the market as well. So we always need some portion of the housing market to be vacant, right, to account for this transition that's going on, right? So some portion of the market is awaiting occupancy. Either somebody has moved out and we're waiting for. someone to move in. Someone might be upsizing or downsizing. Someone might be moving far away. So there is

some type of equilibrium vacancy rate, or we posit that there's some type of equilibrium vacancy rate that is present in every housing market. So then if that's the case, what is that equilibrium vacancy rate? If you know that and you know what the housing stock looks like in each market, then you can back into the housing gap, right, whether it's a deficit or a surplus. So to estimate the equilibrium vacancy rate, what we do is look back historically at data, and we look for periods in time where vacancy rates have been relatively stable and house price growth has been relatively stable, right? The idea being that if a market is close to its equilibrium, right, that vacancy rates,

you know, people may be moving in and out of properties, but that vacancy rate overall on net isn't going to bounce around much. So that's a market that's in stasis. And then at the same time, home prices should not be accelerating or plunging in that market as well. We want to look for that stability. So as we did our analysis, what we found really were two periods in time in U.S. history, at least in the U.S. history that we're able to observe, which is after 1980, we found that 1985 to 2000 and 2012 to 2018 were periods of time where, again, you had that relatively stable vacancy rate,

relatively stable house prices. House prices growing pretty much at the rate of income growth. So you have price to rent ratios, price to income ratios that are relatively stable. So that was the key to the analysis is to really identify what is equilibrium. Once we have that, those periods, we calculated the average vacancy rate in each market. So again, whether we're talking about the nation, state, metro, or census tract. And then we compare that to the current or the latest vacancy rate in each market. of those markets, right? So if the vacancy rate today is below the equilibrium, right, what that

historically equilibrium looks like, that implies that there's some type of deficit. We need to add housing to that market in order to get that vacancy rate back up to its equilibrium vacancy rate. And conversely, if the vacancy rate in the market is higher than its historical equilibrium, that suggests that there's some dynamic going on where actually you have some oversupply in that market, at least temporarily. Over time, you'd expect that construction, either up or people move away, people move in, and the market readjust. But again, that's not an instantaneous type of transition. So basically that's the fundamental methodology.

Then there's a lot of statistical work in the background to actually pick up those numbers. As Maggie mentioned, we use the American Community Survey, which is one of the primary data sets we have with broad coverage across geographies with a lot of granularity. So it's great from that perspective. The drawback of that data set is it does have some lag. It's annual data and there is a lag. So our latest data point is 2024, for example. We'll get the more recent data next December, right? So it's not as up to date as say some of the other data sources that we also used in our analysis, such as the census housing vacancy survey, but it has

that granularity that is so valuable. So long story short, that's the broad methodology. If you apply that methodology at the national level. What we found is we have a deficit of around 800,000 units based on the vacancy rate itself, right? And that's split. That's actually been declining, hasn't it? Because the vacancy rate in the last year or two has been rising. Exactly, exactly. So the vacancy rate dropped to a near historic low during the pandemic, and then it's been climbing upward. It's actually seems to be gaining some speed here in the most recent quarter. So as that vacancy rate climbs up, it's still below the equilibrium. right, that historical norm, but it's getting closer and closer.

So the gap keeps shrinking. So it's 800,000 at that national level. There's kind of say, yeah, you know, there's so much nuance here. And obviously we cover some of that in the paper. Some of it, you know, there's just even more than we could cover in the paper. But one of the things to consider is that we're looking at a vacancy rate that's across the housing for rent and for homeownership in an area. but with no more granularity than that.

So you could have a situation, and in fact, we do, where you might have a very high vacancy rate for rental property that is high-end luxury lifestyle rental, and that masks an under supply in the affordable part of the market or the workforce part of the market. So this is a very... I don't want to use... Of course.

Of course, is the right. I was going to say rough, but course is probably a better word for it. It's just a, it's just, if you're trying to understand, you know, shortfalls in the market, this is just one tool and it helps you identify markets you to take a closer look at. Can I add something to that? Yeah. One of the things that I think is cool about the paper, at least what's cool at writing the paper,

was it really does have the effect of, um, of, of taking an aperture and, in shrinking it through the course of the paper. start with the debate about the national numbers, you know, three versus six versus how the rulers do it, how to home builders do it, all that stuff. And then you take the step that Chris just mentioned where you actually regionalize it a little bit and make more sense of how different regions dynamics look given what he's just described. And you begin to see variations in the regions. And you're like, oh, okay, well, you know, it does, you begin to realize intuitively how wrong the national number is in an important sense because you can see

some regions are red because they're lacking and housing at an aggregate level and some reasons are blue but in the aggregate they wash each other out and you've got this nation that looks closer to equilibrium than it really is but then to your point it's like a fractal kind of thing then you go in one step and you look at one region and you realize well that too is just as misleading as a number because it masks or conflates different regions in the region, subregions, and then different markets than the region. So it's kind of nice because it, you don't jump from national to censors track. You go, because there is something that you learn about sand states versus, you know,

Rust Belt states. And so you kind of begin to, you know, see some patterns there that are interesting and tell you something about the different situations. But then you stop and catch yourself and realize that there too, you're oversimple. which then prepares you to take the next step. So it's a nice trajectory, I guess. Yeah, yeah, it's a nice way of putting it. Absolutely.

Chris, I stopped you because there was a second aspect of the methodology you wanted to go into. That's even more complex. If that's right. I really like this. The aperture. This description of the aperture, right? And I think the core methodology, the theoretical methodology is sound. It's just the data limitations or the data.

data that we apply to it. So the vacancy rate analysis is important. I think that it carries a lot of water. But there's another aspect of this problem that is kind of hidden from the vacancy rate because the vacancy rate only captures the share of the active market. We're only looking at occupied and vacant units. It doesn't capture the fact that you have millions, potentially millions of people who are outside of that housing market altogether. They're not expressing themselves, right? So particularly we think of young adults, right? Young adults who may want to buy a home or may want to rent a home, but they simply can't find housing at their price point, right?

It's just not affordable. So what do they end up doing? They end up living with their parents, living with roommates, living with relatives, and therefore they're kind of outsider. They're hidden from the vacancy rate because their housing is now being accounted for by someone else's housing unit. So what we want to also consider are the pent-up demand, for housing that exists. Again, how many households would be formed if there was adequate housing?

And this is a really difficult problem, right? Even more complex. Well, we do know from the data is that there are about 7 million young adults between the ages of 25 and 34 that are living with their parents. And that number has been increasing over time. So we have to imagine that some fraction of them would presumably want to live in their own household, either rent an apartment, buy a home, rent a house, what have you, if that housing was available. So then how do we estimate from this group or from the data that we have, how many additional housing units would we need to accommodate the pent-up household formations? And so for this part of the analysis, what we did is look at the trend in household formations

from 2011 to 2018 to get a sense of how things were progressing. and then we extrapolated that going forward. So we're looking at pre-pandemic trends. Had those trends continued in terms of household formations, what would the theoretical United States look like in 2024? And based on that calculation, what we found is that there's about a 1.2 million household gap between, again, the number of households we would have expected,

given prior trends, and the number that were actually reported by the census for 2024. So that is certainly a substantial number. of additional housing units that we would need, even more than what would be implied by the vacancy rate. And that's a very important aspect to this analysis. The thing is, though, it rests on even more estimates here, right? You could make a case that maybe preferences have changed. Maybe young adults want to live with their parents more than they used to.

Maybe those householder rates have changed. Or maybe there are economic realities like child care or elder care that favor more, multigenerational households today than we had in the past. So you could make arguments that maybe some of the decline, perhaps, in household formations that we've seen from these groups is actually due to preferences. But I think that the data is pretty clear that whether it's 1.2 or a million or 800,000, still a substantial number of additional households that are not fully captured by the vacancy rate. So at the national level, you take the 800K shortfall because of the vacancy rate gap.

you add in the 1.2 million because of the household gap, you get to 2 million. By the way, it feels like a nice round number. I don't, just saying, it's a nice round number. If you want me to, I can give you the decimal points. And just for, so that's what we're saying nationwide is the current shortage of homes, 2 million units. And just for context, if you add up current completions of housing, single family, multi-family, and throw in manufactured housing, we come up to $1.5 million.

So what we're saying is that the shortfall is more than one year's worth of supply. So that's down from where it was at the peak. And if you go back a couple of years ago, it was much higher. But at this point, it's still quite large, quite significant. Is that a reasonable way of thinking about it? Yeah, that's right. That's right. Hey, Ira, this is unfair, but I'm going to throw at you.

You heard Chris's methodology and the results. Anything about that really bother you that you want to tell Chris about? I mean, just, you know, feel free just to tell. Among friends. Among friends. Among friends. Like, what the hell are you doing, Chris? No?

Yeah, no, actually, I think that what Chris described is a very sound approach to being able to sort of take that logic. And then as Jim describes, taking that aperture and sort of closing up a bit and looking right down to sense a track levels. The one piece, there's a couple pieces that you worry about a little bit there. Firstly, the pent-up household estimates are really hard to do when you're talking about census geographies at a census track level, which is, you know, a place like Philadelphia, you're dividing it into 380 pieces. So you have a million and a half people that you're dividing into 380 pieces, that's not a lot of data. And so it makes it very difficult to do the kinds of fine-tuning, I think, that the penthouse will demand analysis could do. I do think it's really

useful as well in the definitional kinds of limitations that you attach to the analysis, that you recognize that there are homes that are now sitting vacant, but they're actually sold in homes that are now sitting vacant that are actually rented. And there are homes that are seasonal that are vacant at this moment, but none of those are necessarily available to anybody else. That to me was a real important innovation and one that we could attach down to the census tract level. And then lastly, and I remember having probably 10 conversations with this group about that other category of census properties, which is other vacant properties. And that takes, on real importance in cities and down the census tracks.

Those other vacant properties are oftentimes the kinds of properties that if you are to sort of make your way through the various neighborhoods of a city, that you would look at a house and be able to see the sky through the front window. These are homes that are so physically deteriorated. They're vacant, and they're neither for sale nor for rent. they're not occupiable. And, you know, in most instances, they're going to be beyond the capacities of being effective housing for somebody. And so that kind of innovation in the methodology of removing the ones that are vacant but not available, seasonally vacant or this other

vacant, from the vacancy rate, I think, makes the estimates well, accurate and well more accurate and usable down to that census track level. Well, you got off the hook there, Chris. I don't know if you got an A, but it felt like at least a B plus, at least a B plus. Those of viewers, the IRA was much harsher, but we edited out the harsh part. So you only have to say. Yeah, truth in podcasting. It's a great point, though, that vacant population, that is amazing.

difference between us and some of the other studies that you cited. A reason why you have some of that disparity is other analysts will consider that vacant population as active, right? That held off the market even though it's a derelict property. And we excluded entirely suggesting that that's not really part of the active housing stock. It could become, right? And that's a whole other conversation of bringing in more of those properties. But in terms of what is actually active, we don't see it as viable. I think we did it both ways initially when we we were sort of feeling this whole thing out. And we looked at some cities and their maps of with that vacant housing captured and

without. And you could see very clearly that the one that had captured it was incorrect. You know. Yeah. Hey, Chris, before I move to Maggie and Ira on the, do a deeper dive on the center track, can you give the listener a sense of the results at the state and metro area level? Oh, sure. So it states.

So, again, this is going to give you the broad regional. aspect of things. And I think from that standpoint, it's fairly consistent. We see Brigger shortages in the southeast, the industrial Midwest, and parts of the southwest. And in contrast, you have California,

actually showing a moderate shortage, not a very large one, and parts of the Midwest that actually look fairly well-balanced. You did have a couple states that were actually in surplus, Alaska, North Dakota, but not by much. And then in terms of the metros, what's interesting there is we found that 75% of metropolitan areas had some degree of shortage, so relatively high, again, when you get to that more granular

aspect of things. There's a range there, of course. Some metros have a very large shortage, and we'll get into that, I think, in the next section. Some others have a more modest shortage. But you have some cities that kind of jumped out, like a newer area. New Jersey, kind of a bit of a surprise, Cincinnati. Yeah, so once you start digging into the cities, and what I think is really cool about the work

that Maggie and Ira did is that now suddenly it comes to life, right? It's just you can see things at the metro area and kind of tell a story, but once you dig into the census track level, now suddenly you say, oh, well, it's really the northeast part of this city or it's the southwest rental market. That's the problem, but other parts of the city are fine, right? So it really gives us this granular detail to enrich the story that I think that, I think the states and metros kind of guide us there, but then the next step really gives us the insight that we need to make those policy choices.

Well, let's go to the census tract. And, you know, this is where all the heavy computing lifting occurred. I mean, a lot of data work here. And kudos to Maggie, Ira, and the policy map team, because I know everyone was involved in all this work. And I'm sure it was highly irritating every time Jim said, you know, that doesn't make any sense. Can you go back and run it? you know, can you please rerun it again?

Everything has to be iterative for it to work, right? So, but it does, it does get a little anxiety producing. I've gotten, you know, I've gotten no Jim very well. We've collaborated on lots of stuff over the years. And whenever, you know, we get down to a place where someone's got to do some, push some pressure on somebody to do more work, I let Jim do it. Like yesterday with my desktop publisher, you know, And it looks so much better.

The result is so much better. But I just felt. He disappeared. I disappeared. And I think, in she your employee? So I'm like, I'm like telling her that she's got to change the font and like, I don't know what I'm doing. I was watching the email exchange and I go, this is great. You really did disappear.

Oh, yeah. He just, I went on vacation. Here I am, micromanaging his desktop. It's so funny. But the product is great. The result is great. But Maggie, why don't you take the baton here and give us a sense of, you know, tell us about census tracks a little bit because I don't think many people know what a census tract is.

And then, you know, exactly how you modified the methodology and let's get into the results. So what I'm, yeah, so I are going to do this together. I think we'll probably talk about the methodology. But at a high level, you know, we looked at 350, we looked at census tracks within 350. cities in the United States. And so that accounted for what IRA, I think like 40 million housing units. And that makes up about 28, 28% of all households in the United States. So it's not every census tract in the country.

It is census tracts that sit within 350 cities. And Ira can speak a little bit more about why we picked 350 and didn't do the entire country. and more about the methodology. And then I'm going to get the fun job of actually talking about the findings. Fantastic. Sure. So we made the selection based on a couple different criteria, but fundamentally what we were thinking about was when policymaking happens at the local level,

investments are made at the local level, they're done ordinarily at the city level. And so what we wanted to do was to sort of think about creating an analysis that a mayor or a housing director or the like might be benefited by understanding what the contours of their place are. And so cities became the sort of the large geography. Within that, we wanted to be able to differentiate what places looked like. Maggie had mentioned that we did about 350 cities. That's right. We made a cut off of about, I think it was about 100,000 as the population.

minimum for the cities to be included. We further made some cutoffs that in some places, census tracks align very nicely with the boundaries of the political geography of a city. Political geographies are not necessarily the same as census geographies. So there are places where the tracks go outside of city boundaries. Because we were mostly concerned about what cities could do and what resources they could bring to bear on these issues of the housing shortage or oversupply. We also made the decision then to only include census tracks that sat at least within 50% of their area was within the city's boundary. And so that didn't lead to losing an awful lot of census tracks across the country, but there are some places where, as I said, a census

track might be sort of partly inside the city, partly outside, and we did that. Why did we make the cutoff? You know, the American Community Survey is something that's done annually, and they survey about three and a half million households per year. To get that data down to the census tract level, the American Community Survey, the Census Bureau rolls together five years of those things, and then starts reporting the data out at a census tract level. We simply made the decision that we thought that cities and the census tracts that comprise them that were very small, we're going to have, in some sense, more noise than signal as it relates to the data.

And that's why we made that cut point where we did. We didn't want to give sort of a false sense of precision that probably didn't actually didn't actually exist with the data as we sort of went down to smaller and smaller cities. When you deal with a metro area or a state or the nation, obviously those errors reduce because the sample sizes are effectively larger. So we made a very, just a practical decision around the geography and the population threshold. Another piece I think too about selecting of the cities was as a political jurisdiction, they have a fair amount of an impact on the housing market in the city, right? So there's land use laws and regulations and permitting and city council gets involved in different things.

And so the way that housing gets to that cost structure is different, right? So the way housing gets developed in a city compared to its suburb can be very different. and we didn't want this to analyze a suburban census tract that might not be that was you know influenced by the city and we didn't want to look at a city census tract that might be influenced by suburban um uh rules right and i guess the other dimension of this is you uh we classified census tract based on income yes okay so we did describe that yep so we so we had all so we we we looked at all of the census tracks that are within these 350 cities, and we categorized them by income and by city size. And that's where we started to really find some interesting things. But even before doing that, before we categorized, we looked at all census tracks and saw that this was really a renter problem. That was one of the first. findings that we had. So of the census tracks in these 350 cities, the tracks were really home

collectively to about 50% owner units and about 50% renter units. But the need, the renter, there's almost double the need for renter units than there are owner units across those census tracks. So that was like one finding before we sort of went deeper. Then we looked at census tracts by income. And we said, and we used very standard categorizations that the federal government particularly HUD used around income. And we looked at the tracks income and compared it to the city, median city income in which that tract sit. And we decided whether census tracts were either low income, moderate income, middle income, or high income. And we have those categories in the paper, but it's like less than 50, 50 to 80, 80, 80, 120, over 120,000. And when we looked at it,

and then we looked at city size, and we categorized large cities as cities being over a million people. We looked at medium-sized cities, so between 500,000 and a million people, and then small-sized cities, which were 100,000 to 500,000 people. And so that's when things started to get even clearer. And that's when we saw that this rental need, which we had previous like identified, was most acute in the moderate income tract. So if you think about the moderate income track, that is like one rung above a low income tract. That's where we saw the renter need as being most acute. And who would that, what would those people be doing? What are they firemen, police, teachers, nurses? Is that kind of sort of what we're focused on? I'm assuming. I'm,

I mean, I mean, and then the second, and workforce. And then the second one after that is middle. So between moderate and middle, you are going to capture that. Moderate, middle, rental, large cities, that's where the shortage was most severe. In all cities. In all cities. In all cities, that was the case. So it's clear that those moderate and then middle income tracks are getting hit harder with the shortfall than other income category tracks.

and it holds true regardless of city size, like we said. But in the large cities, we did see something very different, which was there's actually a surplus of renter units in the high-income contracts. So you are not seeing, you know, there was some need for high income and low-income in the other cities, but it was very moderate compared to really the need that was in the middle and moderate income tracts. But in the large cities, you saw now a surplus in rental units in high income census tracts, and at the lowest end, at the low income tracks, you started to see a bigger need.

So it was a very different dynamic in large cities. Yeah, and that's very intuitive, right? Because you can look in Philadelphia, we're from a lot of us from Philly. and Jim's from Raleigh, you go downtown, you can see these large luxury apartment complexes going up in the city. And that's the case in big urban areas across the country, you know, Chicago, San Francisco, everywhere. And that's where the vacancy rates have risen most significantly where rents have been weaker. So that's what we're seeing in the census track results is entirely consistent with that. We're seeing that plain as day.

I think we were thinking we would see more in the low income in other cities than we did for sure. Yeah. So that was the real surprise on the rental side to me was the at the low income, it wasn't quite as the shortage wasn't correct. It's that significant. It wasn't that significant. That was the surprise. That was the. That was your surprise.

We're like, oh, goodness. Right. Our math is, is there's some variable in here that's screwing things up. Initially was an impression. Right. So you had these, you classified the census tract into four different income groups, low, moderate, middle, high. You're for rent, for a homeowner by city size, you're saying very clear in the data, high-end rental, high-income rental oversupplied. Moderate middle-income rental, most cities undersupplied.

significantly so significantly so that's where the bulk of the under supply is and then low income not so much it was kind of balanced it was kind of balanced except in big cities except in big it was still it was still sort of balanced but it was higher higher than in other other size cities yes and we looked at owner as well and as you might expect so there was just not there's not the same level of need for the owner units as there is for the renter units um and And we did find that the need was greatest, and it would make sense in some ways in the middle and high income tracks. That's where there was-

I'm totally confused by this. This is where I'm confused, right? Because if you ask me a priori what result we would get, because if you look at the homeowner vacancy rate nationwide, it's pretty low, you know, relative to its so-called equilibrium value. I mean, it got to a record low, not long ago, right? About two years ago was at an old-time record low. And I would have thought that that would indicate a significant undersupply for sale,

for homeownership at the low end, just because builders, you know, we know they're building at the high end because they need a certain return and they can only get that at the high end. Those are price points that give them a return. They can't get that at the low end. So that, if you, to answer my own question, what surprised me, that's what surprised me the most. Right. Does that surprise you? Did that surprise you?

No? When you put it, not really. I don't know. I guess, yeah. No, you were surprised. Okay. I was surprised only because I thought at the low and moderate income levels there wasn't going to be as much need only because it is financially more difficult to begin with.

Like, forget about the vacancy rate, but it's just a, it's a harder, it's a harder hill to climb. Maybe that's the answer. Maybe. And at the middle and higher income levels, it's more. Or maybe a homeranship is not even a possibility for the low income tract. Right. I think, yeah, yeah. Yeah, I don't know.

It's worth exploring further, I would say. But it's before you, on the rental side, it's worth saying a couple things about the bottom of the, of the, the bottom quartile, which is, which becomes intuitive, the more you think about it. It was very counterintuitive, to me at least initially, but as when thought about it, it became. more intuitive over time. One is a data shortcoming point and the other is an explanation, an explanatory point. The data shortcoming point, as we thought about it, it became clear that if you've got a high-income census track next to a low-income census track, and let's say the low-income census track at time one, at our first instance, is way undersupplied, And then at time two, someone comes in and builds a lot of supply.

With our data, it looks like the folks that are living in the low-income tract are now adequately supplied because you've now got a balance in the numbers, at least, in their tract. But what is often happening in those situations is you run out of land in the high-income tract. And so developers are buying cheaper land, a low-income track, in building housing not for the folks that live in a low-income track, but for people that live next door and the high-income tract who don't have enough supplies. So one of the things the data doesn't capture, at least as we've done it, is gentrification, effective. So the data tells a story that's pro that shows low-income

some low-income tracks as being more adequately supplied than they probably are if by that one means adequately supplied for the people who lived there before the new construction came along. So that's what that point. But then the other point, which is more explanatory is, which got me more comfortable with my intuition about all this once we sort of thought about it, is almost all of the subsidy, housing subsidy on the rental side is directed at that bottom quartile. So, Litech and everything else, all the HUD funding, it's all going, guns of blazing at that bottom quartile and God bless it because we need it. And it's still under supply, even with all that

heavy artillery, you know, hitting that bottom quartile. But once you get up out of that bottom quartile, all the subsidy runs out and the market doesn't dig that. Doesn't go down. So in some ways, our initial intuition about a linear picture where the bottom of the market is most deeply underserved and the top is overserved, if you took the subsidies out of it holds true. because the subsidy is only at that bottom quartile, which distorts somewhat that story. So if you take that out and look at what only the market's serving, then it makes total sense. The bottom is most poorly served and the top is over served. Right.

Yeah, I think that's what surprised me was we didn't find shortage at the low end, the bottom, the lowest income census tracks for rent or for homeownership. We didn't find that. And I think the subsidy argument is a very strong one to explain why we don't see it on the rental side. I mean, because we get Litech construction and other subsidy supported supply there. And I think the gentrification argument is the good argument for why we don't see that on the homeownership side. You see housing coming in to those, because that's where you can build at a lower price point because the land. is, you know, less expensive, and you've got more land in some of these areas. And so you get

housing, but it's not, it's not for the people that are living there. It's for people that are going to come in. And we might see that later in. We can't prove that, but that feels like a reasonable, you know, theory as to what's going on and why we don't see that at the law. The only other explanation was the one I think you were alluding to, and I kind of was grabbing onto was, you know, potentially at those low-income census tracks, you just, that, for a home ownership market, it's so thin, so small that it's, you know, it's got all kinds of measurement issues at that point. It's just not, it's not a viable kind of market. It's, it's not really viable to have a lot of single-family housing in those low census, those low-income census tracks. Does that make

sense? Yeah, they do. And I think it's, you know, I think we did agree, it's a combination of all of those factors at play. And as we do this each year over time, it will be interesting to see how they change, particularly the census tracts, the lower income census tracks next to the high income census tracks, if we can start to see the impact of gentrification. Because we are likely underestimating the need in some low-income census tracks, particularly those that sit next to a high one. So all of this data, we mapped.

These 350 cities are available to see on PolicyMap actually. And in the paper, that's not a very good advertisement. Stop, stop, stop. Give me a really good, give me your best advertisement right now. I'm a listener. You know, what am I getting? What can I go? Not my strong point.

Oh, that's it. Come on. Come on. All right. You can go to PolicyMap.com and you can see the results of this study at the census track level. So you can see all of the data for the 350 cities. And as Ira can explain, we did not just the count of the short each census tract and how many units it was short or how many use it was oversupplied for both rent or housing. We also, IRA calculated a severity metric.

So you could compare tracks across cities. So the severity metric has to do with of the shortfall, what percentage of the total rent or housing stock is that? So are they more than 5% short? Are they less than 5? So that you could start to see like how big of an impact this was in that community. So I talk about Philly. So you, and this is in the paper, but we kind of mapped using policy map the tool, this severity measure that Maggie just described.

And you know, there's nobody on the planet that, knows Philadelphia's community is better. I'm not, I'm not making that up. You know them better than anyone because you walk them for years. I mean, literally walk them. And so does the results we get make sense? You know, is it intuitive? You know, once you look at the mapping of the severity measure. Sure. It absolutely does make sense, Mark. And just I'd like to go back one step a little bit on another piece of nuance about understanding what's going on in, particularly more income markets. And I think what many cities, Philadelphia being one of them, experience is not only sort of this shortage issue, but it's an affordability issue. And so there may be units there, but people are

having to spend more and more and more of their income to be able to access those units. And you know, those are the places where particularly lower income people can go because they can't afford the vacant units that are sitting in other parts of the city that are $3,000 a month or $4,000 a month, etc. So to say that there isn't much of a shortage there is accurate on the unit count side, but you don't want to lose the fact that people who are in those places are spending more and more and more of their income to afford what little housing there is in those places. Great point. Great point. So as it relates to Philadelphia, We thought it was extraordinarily accurate.

You know, we, there are parts of Philadelphia. I had to say this before you move on, the word extraordinary, the listener, I don't think I've ever heard him use that word extraordinary to describe any result. Just sidebar. Go ahead, Ira. That's because I'm like very monotonous. This is how you advertise. This is how you advertise.

It's an extraordinary. of that you can take directly to the bank. We need classes in this. Sorry, Aura, go ahead. That's correct. We had a building boom in Philadelphia that was, in some sense, fueled by a very, very generous tax abatement program for new construction and rehabilitation, which changed dramatically just a couple years ago. And there was a rush to pull the building permits and to get the things going,

the units going before you would lose that advantageous abatement. That abatement, by the way, gave you a 100% abatement for 10 years on the value of the improvement. In other words, the house that you built or the apartment that you built. But that changed. And so what we have is parts of the city that were really, in some sense, sort of like overjuiced by that abatement. And now what you see is many of those units are sitting vacant. If you sort of were to take the map of Philadelphia that you see in policymap.com and you pick your city and then sort of drill in, many of the places where you see those vacant units are the places that if you are to walk the streets or drive the streets of the various neighborhoods,

you would find that they are overbuilt. Anecdotally, what you'll find is also that although those were not typically the housing, the areas of the city where housing choice vouchers, for example, that people of modest means would use to subsidize their rent, those are not the places where they would go. All of a sudden, you start to see the articles in the newspaper about developers of these buildings saying, oh, the Section 8 program is a great program.

We love it because what that does is it helps them fill their units with people that they would not otherwise have had access to in those units would sit empty. So when you compare that map, you see the oversupply in those areas that had that dramatic building going on. And you see the under supply And the places that actually in Philadelphia probably grew at a reasonably, you know,

sort of slow but constant rate, but did not see much construction. And the kinds of places those are, by the way, in Philadelphia at least, are, for example, the modest and middle income black neighborhoods of Philadelphia, the neighborhoods of Philadelphia, in sort of the northwest, the neighborhoods in Philadelphia in the lower northeast that have become the places that, you know, second moves of immigrant families have made after they've sort of come in, gotten jobs, built businesses, stabilized and be able to move up. These are places that families have moved to, but there has not been a lot of construction in it.

And these are the places that you see the shortage. Now, again, I think the important point, I think that everybody on this podcast made is that when you sort of aggregate these numbers, you only, almost essentially treat those high income places as fungible for the low income places or the owner units for the renter units. And what we know is in cities, they are nowhere, but particularly in cities, they are not. People are not going to make moves from one part of the city to another. They're not going to go into $3,000 units when they could afford $1,500 units. They are not fungible.

And so the ability to discern where those places are becomes, fundamental to being able to design the kinds of intervention strategies. But I do think to your fundamental question is to its accuracy, we looked at it in Philadelphia. It seemed very accurate. We were able to sort of find subject matter experts in a variety of other cities where we've worked and we're, you know, the people on the team, this team that created these estimates worked and sort of validated what the portrayals were at the census. track level in these different cities. And, you know, to the city, people were expressing that it was an accurate depiction of what they're saying. Yeah, and I think, correct me if I'm wrong, but this is a

blog that you're going to publish or soon publish, that 30,000 unit estimate for the shortfall in Philly, or the goal of rehabbing or building 30,000 new units, very, very consistent with our estimate of the shortfall, isn't it in the city? The 30K? Well, the 30K. Or I'd get that wrong. I thought that was my takeaway from that, a quick read of your blog. Well, the 30,000 was a promise to build or preserve.

Yeah. Preserve, many of the preserved homes are going to be occupied homes. And so the shortage, when we looked at it, there were not 30,000 housing units. that were needed to be built anew in the city of Philadelphia. More particularly in the places that were short, we've estimated that there needed to be about 16 to 17,000. So how do you get from 16 to 17,000 to the 30,000 plan? Well, what we ended up doing was creating a separate way of understanding

how many units needed preservation, and that had to do with condition measures that flowed from administrative data that the city kept around code enforcement and violations and things of that nature. And if you endeavored to serve about 78% of the need for new and about 78% of the need for preserved, you ended up with about 30,000. There you go. That was my read of it. 13,500 new and 16,500 preserved.

Got it. I thought this podcast was going to be short. Never is. This is a very fascinating topic. And I do want to end. Maybe Jim kind of taking what IRIS said about policy making in Philly. What do you think? Because as you said, this work was born out of a policy need, you know, the need to address, you know, the question, how big a shortfall do we have and what kind of policy should we address?

And I don't know, we want to go into, we didn't go into the paper in much depth here because this is kind of the next step. But is there something that you took away from here that you thought is really important on the policy side? Yeah, a couple of things that are kind of related to one another. The first is that targeting is really important. If you just dump a bunch of subsidy into supply without any sort of geographic or demographic targeting, then you're very likely to go into segments of the market that don't need the help. And talking about policies that are that help the overall number in your policies. This, you know, if we turn Yosemite into public housing, we'll have a million new units

over the next, you know, whatever, year and a half. It's all well and good to talk about big numbers, but unless they're numbers in the right place, they're just not going to solve the problems that we need solving at the And at the right income level. At the right income level and the right, you know, on the right side of the rental ownership spectrum, all that. And so that's that big picture takeaway. And then the second sort of next step down from that is the part of the market that is deeply

neglected by both the private sector and by policymakers is the middle, especially the bottom half of that middle. So it's the workforce housing stuff that all politicians. pay lip service to, but that the programs we've got designed currently don't actually get to. So if you're designing a set of policy tools, you've got to find a way to target that middle tranche of consumer, especially those that want to go rent home because right now the market doesn't, the math doesn't make sense for the market to build that housing. and there really aren't any subsidies flowing in from policymakers to improve the math.

And so you need policies, I think, that incentivize, assuming you don't HUD to go and build all this stuff, that incentivize the private sector to go in and build, to fill that gap, to build housing in the middle two quadrants that currently the market doesn't serve and policymakers don't serve. Yeah, that's a very critical result. Maggie, anything you want to add to that? No, I just, no, I agree with both of those points. Yeah, very important. Okay, Chris, anything you want to add there or Ira?

No? I think that's the main maintenance point. Yeah, I think Jim's point is super important. We have, we've sort of figured out as a country a set of subsidy streams that attach to certain income levels. And they work reasonably well, although predict. in some sense very expensive housing for what the costs are. You mean like LI-Tech, low-income housing tax. Like LI-TEC, that's right, and public housing.

The production of those units is very, very expensive, typically far in excess of what it costs to produce regular market rate housing without any subsidy at all. But at least there is a mechanism to produce that. And at the very high end, there are plenty of builders and developers and plenty of circumstances under which, like, yeah, sure, build two, three, four million dollar houses. There's plenty of profit margin in it and and you can do as much as you can possibly fit into the place. But it is those segments of the market from, let's say,

maybe about 60% of a regional average to 100 or 120% of a regional average, where it's too high for the traditional subsidy streams and too low to be the sort of the low hanging fruit for the developers who want to build the, you know, very expensive stuff and make a nice, healthy profit over it. And that's the places that have been, in some sense, neglected. And that, I think, is why we're seeing the shortage that we see in there. Got it. Yeah. Well, guys, I just want to say I really enjoyed this collaboration, you know, even though it took a year to get it across the finish line. It was, at my age, a year goes by pretty fast, just saying. Yeah, very productive. And for all the folks out there, really go check out policy map. You can take a look at that data. You're not,

you're not charging for it. Are you, Maggie? We are not charging. We have a public edition and a subscriber edition, and this is on the public edition. Now, I would say, because I'm supposed to get better at this stuff, right, that by subscribing, you not only have access to this data, but tens of thousands of other indicators when combined can give even greater insights into the communities in which you're working, right? So like in Philadelphia, they did not in policy map, but they're looking at where they're, I think your blog, your draft blog talked about looking at building permits, right? And they're plotting their building permits on top of this data to see are we permitting for places that really need it, right? So those kinds of

activities, that's the kind of thing you could do in a mapping application with not just the data. Okay, but I digress. Perfect. It's all in policy map. I'm sold. And, you know, we are contemplating whether we'll do a webinar. Webinar would be kind of sort of like the podcast, but with some charts and grass, maybe show a demo, but kind of get it down and dirty, you know,

with the methodology and everything. For the, this podcast, I think, was for people who are just slightly nerdy, the webinar would be for those people who are full bore nerdy. You know,

we're thinking about doing that. But we'll see how that goes in the next few weeks. But anything else? Yeah, there you go. There you go. Anything else, guys,

before we call it a podcast? Have a good summer. Have a good summer. All right. With that, the listener, we're going to call it a podcast. Take care, everyone.