Research and the Community: A Virtual Seminar Series with FRS Advisory Council Members – Housing

Speakers

---

Watch on demand

Download presentation slides (pdf, 821 KB)

Transcript

---

Courtney Falato

Thank you, everyone, for joining us online today for our virtual research seminar titled Research and the Community. Please note that today’s session is being recorded. I’m Courtney Falato, the Senior Vice President of External Engagement at the Cleveland Fed. Hello, and welcome! I know I saw at least a couple of members of Cleveland’s Community Advisory Council registered to attend today’s seminar. In fact, we know that we have registrants from nine Reserve Banks and four time zones here and we’re so glad that everyone was able to make it.

At the Federal Reserve, teams at the 12 Regional Reserve Banks and the Board of Governors support our public service mission to promote a healthy economy and financial stability. We do this through research, education, and outreach.

At the Cleveland Fed, I lead our external engagement function. The job of our team is to be the bridge to the community to learn how people are feeling about the economy. This likely sounds familiar to all of you who support the Fed’s work through your service on the advisory council. Regional teams across the System regularly seek out opportunities to engage with people who live and work in their respective districts to learn about economic challenges and opportunities that they are facing.

Today, we’re trying something new. Some of you have expressed interest in learning more about the work of the Fed. These research forums aim to connect you with more of our community development and economic research experts. The goal? To share recent research on an issue facing communities and to prompt your thoughts on and discussion with each other about possible solutions.

That’s a big ask for 60 minutes, but it’s a start. We’ll ask for your feedback after the event to learn what we can do better.

All right, so let’s get started. The two researchers you’ll hear from today are Erik Hembre, Senior Economist with Community Development Engagement at the Federal Reserve Bank of Minneapolis, and John Mondragon, Research Advisor in Economic Research at the Federal Reserve Bank of San Francisco. Moderating the Q&A is Lauren Lambie-Hanson, special advisor at the Consumer Finance Institute at the Philadelphia Fed.

Please hold your questions after the presentations. During the Q&A, you can use the Raise My Hand feature and Lauren will call on you to introduce yourself and ask your questions, or you can type it into the Q&A panel.

Erik, over to you.

Erik Hembre

All right. Wonderful. Thanks so much. Let me get my screen up here. Great. Hopefully, everyone can see that. Again, my name’s Erik Hembre. I am a senior economist at the Federal Reserve Bank of Minneapolis, and I’m happy to share some research here called “A Rising Tide Lifts All Homes? Housing Quality Improvements for Low-Income Households Since the 1980s.” This is joint work with Michael Collins at the University of Wisconsin-Madison and Sam Wylde at the University of Illinois-Chicago. And I need to put my standard disclaimer out there that this is all my views, not the views of the Federal Reserve Bank of Minneapolis or the Federal Reserve System. And also, this research did receive some funding from the Social Security Administration, and it doesn’t reflect their views either.

The big research question that this paper is looking to answer is, how has housing quality and quantity changed for low-income households since the 1980s?

As somebody who works in housing policy, I find this to be a really important question. Housing comprises the largest component of most households’ budget, so it plays a really vital role in measuring material wellbeing for households and how that’s changed over time. I know that there’s been lots and lots of concern over rising house prices over time and how that might translate into changes for housing for everybody, but again, especially for low-income households. Part of what this paper is trying to answer is, in general, incomes are rising over this time period, are we alongside that, observing people living in smaller, more cramped, worse conditioned housing, or are they living in larger, better-quality housing over time?

That’s a relatively straightforward question to ask, but it can be tough to answer. Most datasets only provide some very basic measures of these types of things, perhaps the number of bedrooms or bathrooms in a house, or you can observe how much people are spending on housing, but not necessarily what they’re getting for that money.

What this research does is it utilizes the American Housing Survey. This is a really unique survey that the census conducts in conjunction with HUD, and it has a long history. So we will be looking at the American Housing Survey from 1985 to 2021. It occurs biannually, so every other year, and it has this unique structure that it’s a panel of housing units over time. So it’s nationally representative. It has about 60,000 observations in its sample, and it’s tracking the same houses over time, not the same people or the same households like other studies, but in general, it’s going to follow the same houses. While at the same time, it will add new houses to the sample as new houses are adding to the housing stock and others are depreciated out of it. But in general, lots of this is following the same house over time.

Though, again, modifications can happen to the house, things are changing about the house over time. In addition to really detailed characteristics about the house itself, we’re going to be able to observe housing expenses and household attributes such as age or income or how many members are in the household. To focus on low-income households, we’re going to define low income in two different ways and track them alongside each other for this study. One is going to be if anybody in the house reports receiving social safety net benefits. So this could be from SSI, TANF, or its predecessor, AFDC, or SNAP benefits. But alongside that, we’ll also just measure if the house is in the bottom 20% of the income distribution in any year. So these are similar notions of low-income households, and in general, these are going to have similar trends and gains over time, so I’ll talk about them alongside each other here.

Now, I would argue the biggest component of this project is the focus on housing quality, and I should make it clear that what we are really able to measure here are aspects of poor quality housing. So we’re going to have 35 different indicators in our dataset, which we can track consistently over time. And these are all going to be marked positive only if this poor quality indicator is observed. We don’t get good measures of what you might consider medium or high-quality aspects of housing. So we’re not going to be able to see if you have marble countertops or brick finish out front or a bidet in the bathroom, nothing like that. These are all going to be very basic measures of poor quality and how those change over time. You can broadly categorize these quality measures into four different types. We have exterior features of the house, things like problems with the roof, such as if it’s sagging or there’s a hole in the roof, if there’s visible water leaks on the outside of the house. So external features of the house.

We’ll have interior features as well. If there’s visible presence of rodents in the house, if there’s peeling paint, if there are holes in the floor, number of interior features. We’re going to have things related to appliances, plumbing, and electrical systems, their working condition and presence of. So we’ll be able to see if there’s sewage breakdowns in the house, if there’s blown fuses, if there’s not running water, if there’s presence and working condition of a washer and dryer or dishwasher, these are all things that are captured.

And then we have a couple of measures of neighborhood quality as well. Not a lot, but we use what we can. What we are able to observe here are whether there are any bars on other windows nearby or if there’s visible trash or litter in the street nearby.

Now, we have these 35 indicators that’s a little overwhelming. We want to condense that down into a way to measure quality over time, so we’re going to create a poor quality index. What this is going to do is essentially add these 35 01 indicators together to come up with our poor quality index. And then we do something called Z-scoring, which just means that we’re going to give equal weight to each variable overall. So for instance, a sewage breakdown doesn’t occur very often, but it might be especially disruptive when it does. So something that doesn’t occur very often gets a higher weight, something that occurs more often, such as, say, having a blown fuse in the house gets less weight so that each variable has equal weight within this index. Again, since we’re measuring things of poor quality, please keep in mind that a higher value of the poor quality index is going to indicate it’s a worse quality home.

And then we derive two other indicators based on this PQI to help us summarize quality changes over time. One is an indicator for poor quality housing. So this is going to just represent if you are in the bottom 10 percentile of houses over the entire sample period in terms of overall housing quality. And then alternatively, we have an indicator for whether you are a good quality house. And for that one, it’s going to be any house which doesn’t report any of the 35 indicators. Again, we’ll call that good quality housing. Again, I’m not saying great quality housing or perfect quality housing. These are all very, very low-quality aspects of housing that we can measure here in general. Alongside quality variables, we have 11 quantity variables that we can measure as well too. Again, other datasets, you will similarly be able to see things like the total number of rooms, number of bedrooms, and number of bathrooms.

We’re going to be able to add a number of other features of the house here, which we think are interesting to include as well, such as the square footage of the house, whether there’s a garage, whether there’s a porch, whether there’s a basement, whether there’s a dining room or a fireplace, or whether it has central heating. So these are all housing quantity variables that we observe over time. So these are the three biggest summary takeaway figures from the research.

For each figure, the blue line is going to represent our average for social safety net recipients. The red line is for households in the bottom 20% of the income distribution. And again, each of them is starting in 1985 going to 2021. So our first figure here on the left is the rate of poor quality housing among low-income households. And you can see there’s a steady and large decline in this over time. So for households in the bottom 20% of the income distribution, in 1985, about 30% of them were living in poor quality housing, by 2021, it’s more like 12%.

So that’s a decrease of over 50% over time. This is a pretty large reduction in poor quality over time. Our middle panel here is focused on a quantity measure, which is square footage per person. Again, you can see that it’s rising, although perhaps not as much. Again, for people in the bottom 20% of the income distribution goes from about 800 square feet per person up to about 950 square feet per person over this time period, which is an increase of about 20% over this period. And then lastly, on the right, we have how average housing expenses has changed over time for this group. These are in inflation-adjusted dollars. So 1985 on average, these households are spending about $600 per month, by 2021, that’s up to $1,100 a month. So it doesn’t quite double, but it’s close to doubling over that period. Okay. So again, the broad takeaways here, poor quality decreases considerably for this group, square footage or size of the house increases, I would say, modestly, and housing expenses almost double.

Again, since the quality aspect is the one that’s most interesting to me, and I think unique about this paper, I wanted to show a couple of things of how to think about this quality improvement. So what this figure here is showing us is the entire distribution of poor quality housing nationally in 1985. So this is every single house in 1985, keeping in mind that those highest values are the worst quality housing, and I kind of cap the value at 15 here, which is why we have that large mass at 15. So I want to show you how groups are doing in 1985. So where that black line is, that’s where the average housing quality is in 1985, the very average house across the entire United States. The red line and the blue line, again, are for bottom 20% of the income distribution and social safety net recipients.

So on average, this is about the 90th percentile in the poor quality distribution. They’re pretty far away from the average in 1985. But then what I’m going to do is I’m going to use these same attributes, but look at on average these same two groups in 2021 and see how they compare to 1985. And what you can see is there’s a pretty large shift over the blue line and the red dotted line, again, are the 2021 values for these two groups. And you can see for both of these groups, low-income households in 2021 have better quality housing on average than the average house in the entire United States did in 1985. So this is at least one way to visualize or understand the degree of improvement in quality over time. You might also think we’ve got 35 variables here. Maybe it’s driven by some and not others, which ones are improving the most.

So I do break out all 35 quality variables we track here, and I’m going to show you the relative value at the end of the period to the beginning of the period. So if a dot is right on this dotted line here at one, that means nothing has changed about that variable over time. If it’s above it, it means it’s been more common, so it’s worse quality. And if it’s below it, that means it’s less common and it has improved over time. So what we see here is that every single dot is below the dotted line. So for both of our groups, for low-income households and for all 35 measures, poor quality housing has decreased its rate of improvement or decreased its rate of occurrence, so it has improved over time. So it’s both a very broad increase in quality and pretty large. The average decline is a 50% decline in occurrence.

This is another way to understand the quality improvements. It’s pretty broad and large in what we measure here. Flipping that, when we look at quantity, these are 11 measures of quantity. Now we have, again, for both of our groups, for all 11 measures, we have improvements in housing quantity over time. Some are smaller. Again, something like square footage per person or rooms per person are up only maybe between 5% to 15%, but other aspects such as if you have a garage or a dining room have almost doubled over time. The last thing that we want to do is think about how to translate these gains into dollars. So we have a broad and large increase in quantity. We have a more modest but steady increase in quantity. We’ve increased our spending pretty substantially on housing. How do we want to think about the value that houses are getting from that?

So what we do is we take the 1985 housing market, we run what’s called a hedonic pricing model, which just looks at how well these different variables predict how much people are paying for rent. And then we take these same variables, we look across each of the years, and we predict what rent would’ve been in each of those years if you were renting that exact same house in 1985. Ultimately, what we’re going to want to do is say, “How much were these households renting their home for in 1985?” And then when we look at where they’re renting in 2021, we want to say, “How much better would that have been in 1985 for them?” Okay. So this has four charts. It’s probably not worth to go through all four of them, so I’m just going to focus on one. If you look at the bottom right-hand corner, this is for the bottom 20% of the income distribution and focusing on the rental value.

It still includes all homes, but we’re just predicting what the rental price would be for every single home. So what you can see here where that dotted line is at right around $600 per month, that’s saying in 1985, on average, households in the bottom 20% of the income distribution, were spending $600 per month on rent. You see this steady rise in both the blue and the red lines until 2021. And at 2021, if you look where that red line is, that’s going to say that we would predict that the average house that this group is living in in 2021 would rent for about $800 per month in 1985. So that’s about a one-third increase in the rental value of the home that you’re living in. And then we can kind of separate out the relative gain of how much came from the quantity variables and the quality variables. So the distance from that dotted line to the blue line is how much came from the quantity improvements, that’s the size of the home, and then the distance from the blue line to the red line is how much came from the quality improvements.

And you almost see a roughly equal gain in the consumption value from quantity and quality. So even though in numeric terms, the quality improvements have been larger, the value of that to the homes is about equal to the size improvements that we’ve seen alongside it. Okay. So bringing this all together and wrapping up here, big takeaways. So from how we can measure it, we see that housing consumption for low-income households increased somewhere between 34% to 42% between 1985 to 2021. That’s our estimate here.

On the quality side, the improvements were large and broad. Again, all of our measures improved and on average it was a pretty large gain over time for housing quantity or the size per person, this is a smaller percentage increase, but still is valuable to these households.

And lastly, housing expenditures roughly doubled, although, again, just a couple of things to keep in mind from this. So even though the expenditure has doubled, the budget share spent on housing for this group has only gone up a little bit. And this is because incomes for this group have also gone up pretty considerably. So even though they’re spending a lot more on housing than they used to, their after-housing budget, what’s remaining after spending on housing has also gone up pretty considerably over time, and they are receiving more bang for their buck. They are spending more, but part of that has translated into getting better quality housing or bigger housing as well. So with that, I’ll go ahead and stop and I will go ahead and turn it over to John.

John Mondragon

All right. Thank you very much, Erik. Let me see if I can share my slides, “Supply Constraints Do Not Explain House Price and Quanity Growth Across US Cities.” I turned everything off, but my Zoom is being very pokey. Okay, here we go. Okay. Thanks very much. This is joint work with Schuyler, who used to be a research associate here at the SF Fed, who’s now in grad school, and Johannes, who’s one of my colleagues. And we have the standard disclaimer. These are our views, not the views of the Federal Reserve System or the Federal Reserve Bank of San Francisco. The basic motivation for this paper is pretty straightforward. Almost certainly you’ve heard about the affordability crisis in housing. So here I’m just showing you one way of looking at that. There are lots of pictures that look like this. So what I have is in blue, a measure of house prices going back to 1975, so it’s 100 in 1975.

And then in green, we have the median personal income. What is the median person taking home? Again, indexed to 100 in 1975. And what you see is these have mostly tracked each other until we get to about the 2000s, and then you start to see some big gaps emerge, right? Really big during the housing boom. The bus kind of compresses it again, but then it expands again. And now after the pandemic, standing here in 2024, this is where we end at 2024, the gap is enormous. So this has drawn a lot of attention, both in research and policy circles, why does this happen and what can we do about it? So if you look at what the recommendations are, they all seem to follow what we call the standard view. This is maybe a complicated-looking figure, but if you think back to your high school or first-year economics class, you’ll be able to follow along.

So the basic idea is that in many cities, and in the country overall, it has become difficult to build housing. This could be due to geographic constraints or regulatory constraints, constraints on zoning, that sort of thing. So we’ve made it difficult to build housing, which means as demand shifts, which is this curve that has the D, if that shifts out and the demand for housing goes up, we’re going to walk along this green line, which is what we’re going to call an inelastic supply curve. What that means is as demand goes up, prices are going to go up a lot, prices are here on the vertical access, but the quantity, how much housing we have, is not going to increase as much. You can see that going from HA to HLB, right? So we’re paying a lot to not get that much housing, as opposed to a world where we have relatively elastic housing supply, and that’s the orange line.

Here, you should think people are often talking about places like Austin or Houston or Atlanta. I’m from North Dakota, so you can imagine North Dakota’s kind of like this. Where then for the same change in demand, you get still an increase in house prices, but you get a much larger increase in housing units, right? The really rapid response of supply has tamped down and tempered the price response. So the policy implication is, “Well, we should do what we can to try and ease these housing supply constraints. We can’t get rid of mountains and oceans and that sort of thing, but maybe we can change the regulatory environment and that should move us, hopefully, from this green curve to the orange curve and so moderate these price increases.” So what we do in this paper, in the very big picture, is we evaluate that view.

Here, I’m just restating what I said to you and I’m doing it now in math where this equation is saying how much housing we get is a simple function of how much prices went up, multiplied by how elastic our supply is. This psi, what looks like a trident, if that’s very elastic, then we’ll get a lot of housing for, say, a 10% pricing in prices. But if it’s inelastic, we’ll get only a little bit of housing for that same 10% increase in prices. So we’re really testing if that psi can explain what’s going on. The way we do this, this is a very, very simple paper, is that we use data for US, what I’m going to call cities, but are really metro areas to evaluate this view. And all we’re going to ask is, “In cities with smaller psi, that is those that are measured to be relatively inelastic, does the same change in income growth predict larger changes in house prices and smaller changes in housing quantities?”

That is, do we see something like that picture I showed you before where the inelastic overregulated places have lots of price growth and little quantity growth, given what we see as the income growth and those places that seem unregulated, relatively less regulated, do they see the inverse view? That’s all we’re going to ask. Let me just summarize the results and then I’m going to show you a bunch of pictures and talk to you really quickly, but I want to make sure you’re going to get the big picture.

In the very big picture, the answer is no. That is from 2000 to 2020, the differences in total income growth, which is both population growth and income growth, or just per capita income growth, that’s just average income growth, or just population growth, they all predict the same differences in house prices, in quantities, and in population growth regardless of these measured housing constraints.

That is, we’re going to use a number of different constraints, and we’re going to see across all of these that we get exactly the same predictions. And this is true not just from 2000 to 2020, but it’s also true from 1980 to 2000 and 1980 to 2020. So this is not something that just happened, it’s something that seems to be a feature of the data, at least for the last 45 years. All we’re doing is looking at correlations and often in economics that’s not going to be kosher, you say, “Well, what about other things moving around?” Income growth doesn’t give you the full picture of demand. There could be things like wealth growth or changes in taste, preferences, that sort of thing. What we argue, and I’m not going to walk you through all the details here, is what we do actually going to be sufficient to tell us if these supply constraints matter under any correlation between income growth and these unobserved demand shocks?

And the basic intuition is that anything that shifts housing demand is going to walk you along one of those supply curves. So it doesn’t matter if there’s other stuff that’s correlated with income growth, that’s still going to help us figure out if we’re on a steep curve or a shallow curve. The thing that could be a problem is if there are unobserved shocks to housing supply. This is a very counterintuitive thing. It’s not something that people have argued is happening in the literature. And what you would have to imagine is that places like San Francisco, where I’m sitting right now, which tend to be viewed as being very inelastic, heavily regulated, as they became richer as they grew, they actually had shifts in their supply curve, booms in productivity of construction that completely offset the fact that they’re relatively regulated and so on.

And we think that’s a very counterintuitive view. It’s completely at odds with the standard view, so we think it’s pretty unlikely. But even if you think that’s pretty reasonable, we do some extra work to say, “Well, what if we look at really specific shocks?” Say the change to work from home, which we argue in another paper, shifted housing demand up, or if we look at use variation that people have used regularly in the literature to isolate shifts in housing demand that are not correlated with supply shocks. And we do that and we, again, find the same results. So what we’re going to argue, and I should have [said a] caveat at the beginning, that our conclusions, this whole paper, is very controversial. So I don’t want you to walk away thinking that everyone believes this, but we believe it pretty strongly. What we’re going to say is that housing supply constraints, at least these measures, do not matter in the way that we expect.

And we think the reasonable takeaway from that is that addressing housing affordability is probably not going to be as simple as we thought as tweaking these regulatory rules that are really the focus of the discussion right now. Okay. I’m going to just race through some empirical results and this paper is actually quite simple. It could just be two pictures, but I’m going to show you over and over again to emphasize what’s going on. This is the standard picture you might see when you look in the literature. On the X axis, we have a measure of how elastic a city is or a metro area. Places that have a very large number are places that are going to be very elastic, like Houston or Austin or so on. And places that have a very small number are going to be places relatively inelastic like San Francisco or DC or New York.

And on the vertical axis, we have house price growth. Lots of people have looked at the data and seen something like this and they say, “Ah, look, these places that are pretty inelastic have a lot of house price growth.” That is these inelastic low number of places have lots of house price growth, this is from the last 20 years, and the places that are relatively elastic have less house price growth. That’s the correlation we might expect if these regulations and so on are driving housing costs. Now, if we live in that very simple world where that’s all that matters, then when we look at quantities, that is growth in housing units, we should see the inverse correlation. These places that were inelastic should have very little growth in quantities and places that are more elastic should have a lot of growth in quantities.

And when we do that, so now on the Y-axis side of housing quantities, we don’t see that. We see basically still a negative correlation, that is, this is a downward sloping curve, not exactly the same as prices. But what it’s telling us is when you put prices and quantities together, that these things are positively correlated, right? That places that tend to have a lot of price growth are also places that tend to be growing in general. They have more housing unit growth, right? And places that don’t have much house price growth are places that are not growing that much. So this is not saying that the standard view is wrong, this is just saying that we need to be a bit more careful about how we’re looking at the data. So that’s what we’re going to do in the paper. We’re going to take measures that people have provided that tell us which places seem to be more regulated, which places seem to be less elastic, and then we’re going to see, does that predict different kinds of pricing quantity growth as we should expect?

What I’m asking is, “Do changes in total income growth predict less price growth in less constrained cities?” Here, I’m taking one of the measures, I’m not going to talk in detail about these, but in the big picture, what you should take away is that we’re going to use four measures and they’re all measured in different ways. So we’re trying to cover the space of how people have approached these questions. In blue, we have the cities that are relatively more constrained. In red, we have the cities that are relatively less constrained. And on the Y axis, we have house price growth. On the X axis, we have total income growth, which again is population and average income growth. And what the test is is if these slopes are different. So if the standard story was working out, what we would expect to see is that the blue line should be very steep.

That is, as these places grow, they have lots of house price growth. And the red line should be relatively shallow. As these places grow, they have less house price growth. But the slopes that you see are actually the same. There’s a gap here, and if we have time in the Q&A, we can talk about that if you’re interested. It doesn’t mean what people intuitively think it means, but we’ll keep going. That’s one measure. If you look at another measure, again, the slopes are the same. As you move total income growth around, you get exactly the same changes in house price growth across these same groups of cities. This is the third measure that we look at that really focuses on regulations. And then there’s the fourth that looks at a measure of land values, and the slopes are the same across all of these.

So we’re going to say no. You find exactly the same changes in house price growth, whether you do this with income growth, per capita income growth, population growth. But what also matters is what happens with quantities. Maybe prices are moving the same, but you’re getting very different quantities. That could happen. It’s very simple to write down a framework where that’s what happens. Here, we’re going to look at quantities. On the Y axis, we have growth in housing units. And on the X axis, we have, again, that same real total income growth. And what you see is that these curves are nearly identical. They are identical. Statistically and economically, they’re identical. They lay right on top of each other. There is no difference as these places grow with how many housing units they’re going to be adding.

And this is going through all the measures. This one has a slight tilt to it, but that’s all driven by a small number of observations that are not growing. And there’s really good reasons for thinking that those places should not be informative for this question. So if you ignore those, which we don’t do initially, but if you do, you again see that these slopes are exactly the same. So that’s a very small and, we think, spurious difference. And again, here with the fourth measure, these curves are exactly the same. And what this is again saying is that even though we have these measures of some places have these mountains and some places have these regulatory constraints, at least as best we can measure, when you group them and you say, “As they grow, do they grow differently?”, this is telling us no. They grow in the same way for the same growth that we see.

So the answer is no. You see the same changes in house quantity population growth in more and less constrained cities. And then we can say, “Is this just a statement about the years from 2000 and 2020?” And I’m not going to walk you through all the results. The paper is extremely long. You can look if you’re interested, but I’m going to say, “No, we find the same results from 1980 to 2020.” There are some interesting differences, but nothing that looks like the standard story. And we again find the same results from 1980 to 2000. So this is not a new thing. It seems to be a feature of the data for at least the last 45 years.

And then we can say, “Well, what if we isolate specific kinds of shocks?” So what we do in the paper, as I said, is we look at changes in remote work.

So this is the post-pandemic or the pandemic period and we can see what that does to housing demand and we can see which places are affected by it. And we can see that, do more regulated places have more permitting or less permitting, more house price growth compared to less regulated places. And we can also look at things that people have done in literature and say, “If we know that employment is going to grow in this place because this industry is really booming nationally, do we see differences in the price and quantity growth and so on?” And we do both of those things and we get exactly the same results. So this is all coming together to say to my conclusion that supply constraints do not explain differences in the growth of house prices and house quantities relative to income growth. And we think that this is actually very, very inconsistent with the standard view.

Before I wrote this paper or even thought about it, I thought that these differences in slopes would be a function of the regulatory environments and so on, but it seems very clear that that’s not the case. It suggests, and this is consistent actually with what other papers have found looking at upzonings and things like that, that relaxing these supply constraints is actually pretty unlikely to affect house prices or quantities. And that’s, I think, a takeaway from our paper as well as other papers that have looked in very different ways at what happens when you change the regulatory environment in a city. And our paper brings up a lot of interesting questions. Is this something just about the US? We have work ongoing that tries to look in other areas and says, “No, it’s actually like what most rich or middle-income countries look like,” and then some more technical things.

But what I want to finish with is this last picture. It’s the same thing as I showed you before, except I’ve added another line. Before, we just had the blue line, which was house prices, and the green line, which was median personal income, and now I’ve added average personal income. When you add in the average and you don’t look just at the median, you actually see that house prices have not grown out of line with how average incomes have grown. One thing that I think our paper suggests is that part of what’s maybe going on in the housing market and why it looks like an affordability crisis is not that the housing market is behaving in an unusual way, that it’s become dysfunctional because of regulations or so on, but that people’s labor market experiences are actually very, very different, more different than they have been, say, for the 40 years that preceded this point, and that that might be showing up in the housing market dynamics and having these implications that we’re all talking about.

So with that, I will conclude and I will turn it over to Lauren. Thank you very much.

Lauren Lambie-Hanson

All right. Thanks, Erik and John, for those great presentations. We’ll now turn to the audience discussion. We invite all of you to ask our speakers questions or offer comments based on your expertise and the perspectives from your communities. To do this, please use the Raise Hand feature, which you can find by clicking on the Reactions button at the bottom of your Zoom application. We’ll call on you and help you come off mute. We’d be grateful if you could briefly introduce yourself, telling us your name and your organization. If you’re not comfortable coming off mute, but you have something to say, that’s okay. Just submit your question in writing through the Q&A panel, which you can also find at the bottom of your screen.

I’ll give you a moment to find that button. In the meantime, I’m going to go ahead and ask a question, that’s my prerogative as moderator. John, I’ll start with you. Your study is really timely, and I think it’s fair to say it’s provocative. A common narrative these days that we hear, which you alluded to, is that homes are less affordable because of building restrictions and zoning rules and if we relax those things these problems will go away. Could you talk a little bit more about this? Are you saying that local regulations like these just aren’t relevant for housing affordability?

John Mondragon

Yeah, that’s a great question. I think what we’re definitely saying is that these do not explain why some places have become so expensive and other places have not. A couple of caveats or important things to keep in mind, is that the level of aggregation that we say, which is just a fancy way of saying, “Are we thinking about cities or are we thinking about neighborhoods?” will matter a lot. I think it’s obvious that in certain neighborhoods certain things are not allowed to be built and that has an effect on what that neighborhood looks like and so on. But when you zoom out to this metro or city level, it seems pretty clear from the way these data are behaving and other studies that have looked at upzonings that there’s lots of scope, actually, for building all sorts of different things.

It just happens in certain neighborhoods. You can’t build something in this place, but you can build really tall dense stuff in this other neighborhood, and those are enough of a substitute. I think that’s very important that we’re talking about the city level, which I think there’s good reasons to think about that. But in the big picture, yeah, I do think that our results suggest that there is not much economic bite from these kinds of policy interventions. At least we don’t see strong evidence of that in the data. And for those who are curious, there’s a really nice survey from Yonah Freemark of the Urban Institute that looks at when cities upzone, what happens? And in general, you actually don’t find much of an effect. Sometimes it goes in the wrong way. Usually, there’s just not much going on. So that’s kind of where I sit now, if that answers the question.

Lauren Lambie-Hanson

Thanks a lot. I see we have an audience question from Frank. Frank, we’re going to help you come off mute if you could introduce yourself.

Frank Wells

Great. Thank you so much. Frank Wells, I’m president of Bright Community Trust. We’re a nonprofit housing development organization based in Florida, do some data policy convening work nationally. Absolutely fascinating discussion today. John, I would love to hear your thoughts about a couple of things. One, do you think this would’ve looked different 100 years ago, not 20 years ago? Because by and large, post-World War II, America builds in a kind of suburban development pattern, and I wonder if that looks different in a very different development universe that far back.

And then I would also love to know if you have thoughts about changes in, can I call it housing consumption, people turning what would’ve been an accessory dwelling unit, garage apartment that would’ve served the local workforce, becomes an Airbnb short-term rental, more people own more vacation homes and we’re seeing those units remain vacant. And I think that’s a little more prevalent, than would’ve been 20, 30, 40 years ago. So, love your thoughts on those things. Thank you.

John Mondragon

Yeah, those are great questions. The honest answer is we haven’t done the analysis in the 1920s, so I can’t really claim to know. But what I do know is if you look in the 1920s, people are actually already talking about the housing crisis. And I think that cities and things look very different, but they’re actually encountering a lot of the same problems that we’re talking about today. And I think that suggests something deep about what we’re picking up. And I can’t say much more than that. I think the regulatory environment is so different in the 20s, but they’re experiencing lots of similar problems where lower-middle-income households are having a very hard time finding adequate housing in cities where production industrialization is taking place.

For the second question, we haven’t looked at that specifically, but I think in the big picture, what you say I’m pretty sympathetic to, which is that housing consumption is largely a function of wealth and that as you get richer or as the returns to doing different kinds of housing change, then the market is going to respond to that. So in some places, say, Barcelona or Portugal, these have become very attractive to tourists globally. So a lot of the housing stock, it’s more profitable to orient it towards that kind of housing consumption. But on the flip side of it, I think going back to Erik’s paper, I think what that’s also demonstrating is that there have been really big changes to the quality of the housing stock as the US has become a richer country.

That can explain, I think, a lot of why we’re paying more, in certain ways, why we keep spending so much on housing because we live in much, much better housing than we did even 40 years ago, and especially if you’re thinking 100 years ago. So that’s kind of the big picture.

Frank Wells

And I think both of those are interesting observations. I appreciate you drawing that line to Erik’s research also. That observation, I think, also syncs up with what you said at the end of your presentation about the difference between average incomes versus median incomes and how that changes. So a lot to think about. Thank you.

Lauren Lambie-Hanson

Thanks. I see we have a question from Julie. Julie, if you could introduce yourself.

Leonard Adams

Hi, Lauren. This is Leonard Adams. I received a popup on my screen that said, “The host wants you to speak.” And then you said Julie, so I’m not sure if it was me or not.

Lauren Lambie-Hanson

Oh, this is maybe a quirk of Zoom. We have you identified as Julie, but, Leonard, go ahead and tell us what organization you’re with and we’ll roll with that.

Leonard Adams

Okay. Let’s roll with that. Hello, everyone. I’m Leonard Adams, president and CEO of Quest Community Development Corporation in Atlanta, Georgia, AKA Julie. I don’t know where that come from. But my question to the panelists, first, thank you for your data, I was just curious, Quest as a developer of affordable and supportive housing for the last 25 years, really using HUD as a guide to rental, for-sale homes, single family homes, rental raise income limits, I was just curious if your data has been overlaid or layered with the HUD data and if there’s any findings of similarities or larger gaps as it relates to how HUD positions its regulations to us as affordable housing developers and those parameters that we should fit in and are the numbers off? Are they fairly in the same boat or where are we compared to your data and maybe government data?

Erik Hembre

Thank you, Leonard, for that question. You’re definitely going to be a lot more familiar with lots of the HUD regulations around this than I am. Would you mind maybe just giving me an example or two so I’m in the right framework in my head of the types of things you’re thinking about?

Leonard Adams

Yeah, Erik, I was just thinking, say for a for-sale home right now, for an affordable 80% AMI homeowner, the max sale price HUD will dictate to us if we use federal dollars what that amount would be. So say right now that is 285, 120 max sale price. So I’m just thinking as your data is being collected, is that right? Is that a fair number? Is it too high? Is it too low? We can’t build the house for the 285. So we know we can’t build the house for 285, but is that for-sale number true to the environment?

Erik Hembre:

I’ll give you a perhaps unsatisfying answer, but I’ll do what I can here. I know HUD uses the American Housing Survey for several things. One thing I know that they use it for is they use it to determine fair market rents across different metro areas, which is one of the main ways that say Section 8 housing vouchers determine how much they could spend on rents. So in some ways, it sets those bars. And then I know that HUD also gets these estimates from census on what area median incomes are everywhere, and they set these AMI thresholds based on that. So that’s not exactly the same survey, but very similar. I think the type of number you’re quoting there on where they’re going to put a max price on, that is coming from survey data like this, although not this exact survey.

I’m going to add something in that you weren’t quite asking. In that, I don’t know if HUD is using the data that I am in the same way at all. I’m guessing they don’t. I’m not aware of it in terms of how they’re setting the regulation, but it is true that one thing I can’t separate or I didn’t discuss in my paper is how much of this increase in quality over time is driven just by demand, by the fact that people are wanting to spend more on housing over time, which I think is a reasonable assumption, but it’s also true that building code regulations very likely have been becoming more stringent over time. So part of this can also be regulatory. It could be that it’s more difficult to build lower quality housing than it used to be, even though that’s going to be more affordable.

So part of the reason we might see this increasing trend in higher quality housing is a combination of people spending more on it, but also it might be tougher to build housing that had these same characteristics that we used to have. I know it’s not quite what you were asking, but I don’t know of any way in particular that HUD determines its regulations from the data itself.

Leonard Adams

Thank you. Thank you so much.

Lauren Lambie-Hanson

All right, we see some more questions coming in. Colleen.

Colleen Dushkin

Hi, good afternoon. My name is Colleen Dushkin and I am the VP of Culture and Strategy for Cook Inlet Housing Authority, which is a tribally designated housing entity serving the largest community in Alaska, and that is Anchorage. I also serve on the San Francisco Fed Community Advisory Committee. Erik, I wanted to touch a little bit on what you had just said, and that’s kind of the building standards, which yeah, absolutely I would believe impacts the price point. I’m curious, and to build on what Leonard had just said, a couple of things.

Number one, we see in affordable housing the insurance implications being huge. We don’t have any regulatory framework around the insurance industry right now to a sense that you can’t increase by 200%, those type of things. So are the hypotheticals considered and what are the affordable housing side as far as multifamily, what is that impact and the regulations that are guided by HUD and those impacts that have increased the cost to develop, what are those impacts on the spectrum? We’re talking about the spectrum of single family and having more affordable multifamily. How does that spread across and has that been considered also?

Erik Hembre

Oh, great. Well, thank you for the question, Colleen. That is an important question. I’m trying to think of the right way to connect it to what I was measuring. So it’s true that some aspects of the housing costs we don’t dig into too much. I was showing you an aggregated number there of how much people are spending on housing over time and that’s going up. We do get to see how much they’re spending on insurance for homeowners. I can’t quite see that for renters, but we have other data to see that a little bit. We know, especially over the past few years, property insurance rates have been rising a fair amount. You asked about the regulatory part. My understanding is most of that’s done at the state level. I know that there’s some federal overlay on top of that, but mostly it seems like an area that I don’t know a lot about, so I can’t tell you too much about how much the regulatory environment in particular is affecting insurance costs.

But yeah, I know that over the past few years, property insurance in particular has been going up a fair amount. But yeah, I would just put it as part of what I’m interested in looking at in the future is how things like regulations impact housing costs and quality. There’s perhaps going to be a little bit of a trade-off in there at times of, “Well, if you only allow the best quality housing to be built, that’s going to be really expensive and that’s going to be a burden on people.” But I think at the low end, you always want to be careful about making sure that people are using materials and building [construction] techniques that aren’t going to result in a condo falling down, for instance, or something along those lines.

There’s some push and pull on where the optimal level of regulation is, but that’s tough to include in this type of work since there’s just so many things that go into regulation, so I don’t have a simple answer there. So yeah, I don’t know if that’s overly helpful for what you’re looking for there. It is true that there are these other components of housing that we can’t measure, so something like property insurance or how property taxes are changing over time and what you’re getting for that, like local amenities, I’m not able to observe that very well in this paper as well. So there are other aspects of housing that we can’t really observe over time well in our paper that are worth thinking about.

Colleen Dushkin

Thanks, Erik.

Lauren Lambie-Hanson

Thanks for that question. Next, we have Kevin. Kevin, welcome. If you could introduce yourself.

Kevin Chu

Hi, there. I’m Kevin Chu, executive director of the Vermont Futures Project. I serve on the Boston Fed CDAC. My question is around demographics and how that factors into your research. This is for both of you. And I’m thinking about demographics and median household size. I can imagine, for example, 20 years ago, a family of four, two adults, two kids, and now the kids have grown up today, they’ve moved out.

So, Erik, for your research on the quantity side, per person, square footage just went up just because two people left, or number of bedrooms. And then connected to John’s research, those two kids, now they’re looking for housing, and if they’re looking in their community, is that part of the demand curve? So how does demographics and median household size factor into your research? And I ask this because Vermont is a very old state.

Erik Hembre

Okay. I can start off and then maybe hand it over to you, John. So demographics, absolutely interesting to analyze alongside this, and you’re kind of spot on with your guesses there, so I’ll repeat your point there a little bit. It is true that number of people per house has been going down over time. So when you’re looking at amount consumed, in many ways, we’re living in similar sized houses, but with less people. And that’s how you can at least rationalize some of those gains. It’s also true that in general, newer houses that are built, things that we’re adding to the stock and replacing are larger than what’s on average there.

So we are seeing a growth in average home size, although new multifamily units, at least from what I could tell, aren’t especially larger than existing multifamily units in terms of space. So most of that house size growth that we do observe is more on the single family side of things. I’d note your example of two kids moving out, it’s not clear that that’s happening as much as it used to anymore. It is true that we are definitely seeing plenty of adult children living with their parents longer than they used to. I’m going to make just a very small plug that this is something that I’m working on at the moment and we’re going to be releasing.

So if you want to think about the share of people that own their home as opposed to the number of homes that are owner-occupied, those can be very different numbers. So if you look at the US homeownership rate right now, it’s about 65%. That’s going to tell you that 65% of units have the owner living in them. But if you look at the share of adults that own their home, it’s only about 53%, and that’s because a lot of these adults are now co-residing with their parents. So not quite what you asked, but it’s on my mind at the moment. So just wanted to share that and turn it over to, John.

John Mondragon

Yeah. And similarly, your intuition is exactly right. Demographic, it’s very interesting, but when you look over time and in the cross section of what does explain growth in housing units, a very first-order thing that pops out of the data is that growth and incomes actually doesn’t. And it doesn’t matter if you’re more or less constrained, more or less regulated, growth and incomes will show up in growth in prices, but it doesn’t necessarily show up in growth in quantities. That is, if I know you got richer, I know your prices are likely to have gone up, but I actually have no idea if you had additional housing units get added. And that’s because the thing that really drives housing unit demand is demographics, is population growth and changes in household size.

Those dynamics are incorporated in the way we run our tests. That is, if there’s additional demographic growth in this city because families were large and those people are moving out and they want to have houses there, that will show up in additional price pressure. And if it’s inelastic, it should show up in even higher price growth than in places that are relatively elastic. So that thing is there and it’s a real pressure. And I think it’s clear that it’s causing interesting changes in when people move out and all that sort of stuff, but it seems to be primarily a feature of distributions, I would guess, of incomes and prices and that sort of thing, as opposed to the metro-level regulatory environment or geographic environment or something like that.

Lauren Lambie-Hanson

And, John, you didn’t really have time to go into this, but in your paper, you do look at different measures of housing quantity, not just number of units, but also rooms per person. Could you just briefly tell us a little bit about that as people are thinking about any last questions they might have before we close today?

John Mondragon

Yeah. So nothing as detailed as what Erik is using in his paper, but we do have rooms per person and we also see that that has no correlation with any of these things. But what’s interesting there is you do see as places grow, they tend to have relatively fewer rooms per person. And I think that’s not saying that housing is getting worse. And it reminded me when Erik just said, it’s not clear you see a lot of growth in the size of recent multifamily units. I think that’s likely because you’re seeing growth in urban areas. That is, we’ve seen this pretty long secular trend to living in cities over the last 20 years, relative to the previous 20 years when cities were being emptied out in a lot of ways, a lot of cities at least.

So people are kind of trading off, “Well, if I want to have a lot of space, I need to live further out,” versus, “If I live in a smaller space, I can walk to the cool restaurants and take the train,” and all that sort of stuff. So I think those are very interesting things that are shaping our cities and what gets built where and how big it is. And it really speaks to what Erik was saying is, well, part of what happens when you buy housing is you have a certain amount of space that you can exclude other people from, but you also are putting that in a specific area that gives you access to certain amenities and those have value and you’ll pay for those. And it’s very hard to measure exactly what’s happening on that margin.

Lauren Lambie-Hanson

Right. Well, I don’t see any more questions that have come in, so I’ll just ask one last quick question of Erik. Erik, just as we’re ending and as we’re thinking about policy implications, we want to be clear, are your results suggesting that improving housing conditions for LMI households is no longer a concern? Is it mission completed, mission accomplished? What would you say to that?

Erik Hembre

I would not hang the mission-accomplished banner quite yet. Again, the point of the study is to try and understand how things have changed over time. And I think it’s worthwhile to take away from that, that things have improved, and pretty considerably for this group. So I think that’s worth noting and celebrating. Alongside that, it’s very important to note that while the absolute level of quality has changed, the relative quality has not necessarily changed. I showed that picture of the distribution of housing and where that average house was to our group, and that hasn’t changed over time, the distance between the average house and the lower-quality house.

So part of that speaks to how this process of filtering we talk about happens. What’s happened is we built a lot of housing over the last 35 years, so higher-income households have moved into the newer, better housing, lower-income households, in general, have moved up into this more average quality housing. So there’s always more room to improve because people value housing a lot. And the more that we can allow more housing to be built over time into the future, we might expect even more quality gains to happen after that as well.

Lauren Lambie-Hanson

Thanks, Erik. Well, thank you all for joining us today. We appreciate your time, your questions, and your insights that you shared with us. Information from today’s program will be sent up in a follow-up email next week with links to the video and audio recordings on fedcommunities.org. We also want to know what you think, so you’ll soon receive a survey, as Courtney mentioned, about today’s programming. Please share what you liked and what we could do better. As you may know, the Federal Reserve System is hosting the 2026 National Community Investment Conference, or NCIC, on March 23rd to 26th in Phoenix.

This year’s program, Innovations in Public-Private Partnerships will highlight strategies and effective models for advancing economic opportunity through access to credit, investment, and financial services. In addition, this is pretty cool, members of the Fed’s community advisory councils have been extended a special invitation by the Board of Governors to have lunch with Governor Michael Barr on Tuesday, March 24th at the conference. This lunch is an opportunity to share your perspectives with the governor and network with your System counterparts.

If you’re planning to attend the conference, please be sure to register by March 3rd, and please also email your advisory council liaison if you’d like to join the lunch with Governor Barr. Finally, for more information and resources on the Fed’s work in communities across the nation, visit fedcommunities.org. Thanks, everyone, and have a great afternoon.