Erdmann Housing Tracker · Housing & Cities
TIER 5 Wed, 10 Jun 2026 19:38:16 +0000
Quantifying Agglomeration vs. Scarcity ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ ͏ | | ---|---|--- | | | Forwarded this email? Subscribe here for more --- --- # Facets of the Housing Crisis, A Series: Part 2 ### Quantifying Agglomeration vs. Scarcity | | Kevin Erdmann --- | Jun 10 --- | --- --- | | | --- | | --- | | --- | | --- | | READ IN APP --- Here I'm going to walk through some ways in which I think the scale of agglomeration economies gets accidentally twisted and misrepresented. The trick is that the way agglomeration economies are presented and discussed would be fine in a world without the shortage problem. The way it is discussed is the habit carried forward from the long history during which amenities, incomes, and aspirational migratory choices dominated the condition of the housing market, and so economists didn't have to be careful about describing agglomeration economies. The spatial selection of amenities and values, and the positive and negative amenities created by urbanization really did explain everything. In the before times, you didn't have to worry about overstating the case. Upgrade to paid Figure 1 visualizes the sources of value, costs, and conflict in urban housing. | | ---|---|--- Figure 1 The blue dots are meant to represent the economy if we had never developed the technology to have large cities. The orange dots represent the value of our 20th century cities, which were designed without some of the core features that made dense cities like New York City and Chicago valuable. It is true that if we were all still yeoman farmers scattered across the countryside, living near cities with a few thousand residents, we would be much poorer than we are today. Cities really are a big deal. But, that isn't the scale around which our current conflicts revolve. We have many regions with millions of residents. The housing crisis isn't turning us back into scattered yeoman farmers. That is a difference that is orders of magnitude larger than the difference between Phoenix and Los Angeles. I'm not sure if that is part of the mix of how people think about this or not. If I say to an economist that agglomeration economies and spatial equilibrium aren't important explanations for current trends in incomes and cost of living, do they think I am dismissing the difference between the dark blue dots and the orange dots? In a paper by Ed Glaeser & Matthew Resseger, titled "The Complementarity Between Cities and Skills", they estimate that in cities that attract a lot of workers with college educations, each doubling in size is associated with about 9% higher incomes. That's a big deal. In theory, incomes in Dallas are about 5% higher than in 2000 because the city is larger, and they would be another 4% higher if it becomes as large as Los Angeles. If there is a shortage of 15 or 20 million homes, and the pent up demand for living in the Closed Access cities means that every single unit will be built in the expensive coastal metro areas, and they all double in size, we could expect the average income associated with agglomeration economies to increase 9% in those cities. The effect of doubling, according to the Glaeser & Resseger numbers, would look like moving from the orange dots to the gray dots. Incomes would rise. Home prices would rise with them. Costs of building would rise. As a result, households would compromise on various margins, mostly living in smaller units, but also generally spending a bit more. In cities like New York City, especially families with lower incomes would spend more because where housing is dense enough, it starts to be bundled with other services, like public transportation. Keep in mind, price/square feet of either land or buildings might rise much more than price/income does. Much of the trade-off happens in size, quality, and location. That's why price/income didn't vary much in the 20th century when agglomeration was the source of rising incomes. So, it may be true that doubling New York's population will be associated with 9% higher incomes and 40% higher housing costs/square foot. That doesn't change Figure 1 very much. The new homes would trade off size for location and amenities. This isn't just a guess. In the past, there are many examples of cities doubling in value, costs per square foot rising substantially, and price/income levels remaining within the range shown in Figure 1. So, the gray dots compared to the red dots are a rough approximation of how the housing market in New York City looked as late as the turn of the century compared to, say, Atlanta, or Dallas, or Phoenix. And a similar shift would have happened again in New York City if it had doubled from 20 million residents to 40 million residents instead of capping its growth at 20 million. In the 1970s and 1980s, the average income in New York, Los Angeles, and San Francisco (the Closed Access cities) tended to run about 20% to 40% above the US average, because of some combination of size and local sources of productivity and innovation. Those 3 metro areas held about 14% of the US population in 1970. That has steadily declined to about 11% today. So, in terms of pure size, they have been losing pace compared to other cities, and explanations for rising incomes or home prices depend on those cities becoming more valuable in their current forms because of changing technological, economic, and cultural factors that leverage agglomeration economies only in these cities that had below-average growth trends, but not in cities that are growing. | | ---|---|--- Figure 2 There could be some truth to that. And, to the extent there is, they would have continued moving up and to the right along a similar trajectory, with higher incomes and home prices rising, roughly, in proportion. San Francisco's average income is now more than 80% above the US average. It has a strong claim to that sort of agglomeration leverage. Per capita income in Los Angeles and New York City is still in the 15%-30% range above the US average. They don't have as strong a claim. In fact, I'm not sure they have a claim at all. I suspect that if they didn't have binding constraints on growth, they would have had average, or even below average, income and home price growth. A lot of highly productive workers and wealthy consumers do move to New York City. It's a big city and the economic center of the country. It was attracting more than its fair share of those workers and consumers when it used to be affordable. I think impressions about trends in agglomeration economies in New York sometimes are a simple level vs. rate of change issue. #### The Problem The problem is the arrows in Figure 1. The arrows are what most conflict in housing is about. It doesn't matter if a city is growing, stagnant, or shrinking. Cities evolve, breathe, and change over time. And, whenever they do, the locals feel betrayed. As noted by the spatial equilibrium models that explain housing expenditures in the before times when all cities could grow, families sort themselves spatially out across a city and across the country on many different margins involving what income they can earn, what amenities they value, what constraints they have, etc. The locals picked those locations for a reason, and changing the character of those locations is a threat. So, they block change where they can. And the motivation for that is basically _entirely_ to keep all those dots in the same order. They don't want their dot to move up. They don't want their dot to move down. So, it doesn't matter what some new apartments will be like. It doesn't matter if they will be cheap or expensive. They are change. Does more density make neighborhoods too expensive, or does it draw in the underclass? It can do both, in different contexts. It doesn't matter. It's change. The problem is that every neighborhood is engaged in this battle. The change can happen somewhere else. "Not here," says everyone. Now, at the end of the day, freezing the physical form doesn't stop change. It can't keep the dots in order. People move around between neighborhoods. Homes naturally depreciate, and so the condition of neighborhoods over time is mostly a product of how the families that live in them reinvest in them. We have little control over that. So, the battles over the arrows fail at their own goal. They have to. It's inevitable. But, additionally they fail because of their side effect. The blue dots in Figure 3 approximate the value of homes in New York, LA, San Francisco, and Boston today. | | ---|---|--- Figure 3 Compare the scale of the peculiar pattern that our current housing markets take to the pattern that the spatial equilibrium models explain. When housing becomes so constrained that it requires regional displacement from a significant number of existing local families every year, that displacement is very regressive. Poor families move away. And, in our current condition, where these cities have been displacing their poorest families for decades, the only families with low incomes that are left have chosen to remain because they are very resistant to displacement. They are the price setters. They are the reason high end prices might rise by 30% while low end prices double. This is a very different pattern than the 20th century pattern, and of a scale that dwarfs the shifts in housing expenditures that were driven by agglomeration economies. And, since so many poor families have been displaced, the average income of the cities in blue rises significantly. So, for instance, those recent papers put out by San Francisco Fed researchers that claimed demand and income growth are driving home prices higher instead of supply constraints? They use spatial equilibrium models and city averages. If all you have are city averages, and you think spatial equilibrium models still explain everything, you would think that the cities in blue are actually just swimming in agglomeration value, and they are just way further up and to the right in terms of creating value. But the average income rises because of who moved away and the average home price rises because those who stayed are paying a ransom. Both those changes are driven by behavior of poor residents. You might say, surely, San Francisco fits the description of a market moving up the agglomeration hill. And, yes, San Francisco (along with San Jose) is probably further up the agglomeration hill than any other market. But, at its peak, there was more than a $500,000 scarcity premium, and even after several bearish years in the San Francisco housing market, it is still $330,000. In other words, wherever San Francisco is on the agglomeration hill, take all the dots at a given income, and move them up $330,000. The blue dots in Figure 3 are set $380,000 higher than the baseline agglomeration dots. There probably are complicated mathematical and empirical ways to try to measure how much of San Francisco's very high average income is due to the missing blue dots and how much is due to how far it has moved up the agglomeration hill. As the San Francisco Fed researchers have made clear, they just don't know anything about this. The pattern is distinct, of a massive scale, and historically peculiar. They don't know about it. And, at the center of all of it is this strange mechanism, where the arrows fail at their goal but produce this massive negative side effect that is still a secret to the academic field that would have measured and studied it. Any paper that uses metropolitan averages to study these conditions is coming to conclusions about cities that don't exist. They are inferring cause and effect from a wonderland world. This is one facet where we all talk past one another. If I claim that building more homes in the blue city will help turn it back to a gray city, attract or retain households with low incomes, lower housing costs, and reduce the total value of the stock of housing, even after growing the stock of housing, someone who believes in the wonderland world city thinks I'm crazy. They think agglomeration value is on steroids, and if we build more houses, it's going to sling itself to the moon up that agglomeration hill. Look at all the agglomeration value those cities seem to have accumulated even with what little growth they have had! On this issue, it's actually surprising that New York City and Los Angeles don't have relative average incomes that are particularly higher than they were in earlier decades. They have both definitely displaced millions of their poorest residents. They sit about where I have placed the blue dots in Figure 3. I think this suggests that they have actually slid down the agglomeration hill a bit. Which makes sense. They haven't been growing. I think our initial expectation of Los Angeles and New York City should be that they have lost agglomeration value since the 1990s (relative to growing cities), but outmigration has countered that by compositionally pressing the average income higher. Now, if they did build the houses that would reverse that migration, they would start to accumulate agglomeration economies again. But, how many homes that would be and how much it would push them up the hill are very different going from blue to gray versus moving even higher up the hill from black. When others claim that building more homes in Los Angeles and New York City will only increase the value of the real estate in those cities, they are talking, empirically, about a different city than I am. I think they are talking about a pretend city. If they are using metro area averages to make their case, they are definitely in wonderland. If they don't know that these cities have a peculiar uniform unit premium, then they aren't talking about the same cities I am talking about. In the model I use to arrive at the scale of the scarcity premiums I cited above, there were no scarcity premiums as late as the turn of the century. The dots only started to move up from the gray dots to the blue dots after 2001, and when they did, the poorest residents were flooding out of those cities and flooding into Florida and Arizona. Every other city lived on the agglomeration hill until the mid-2010s, when, after their own rates of new home construction fell to the same pitiful level as the Closed Access cities, they also started moving up toward the blue dots. In their cases, because they have no density amenities, they rose from the red dots up toward the blue dots. Outside of a few laggards like St. Louis, every major city in every region of the country now is moving toward the blue dots. This is another area that I think creates some of the hallucinated conclusions in the San Francisco Fed paper. Under current conditions, the more demand there is for growth in a metropolitan area, the more constrained construction capacity is being steered to that market. But, under current conditions, it's not enough to keep the scarcity premium from pushing the dots up. And, this also creates a positive correlation between demand, growth, and home price appreciation. So, they also think those cities are climbing the agglomeration curve. They made a peculiar choice in measuring demand in cities. Instead of measuring average income growth, they measured total income growth - the growth in the number of new residents and the growth of their average income. There has been some discussion about the problem with doing that because total growth is dependent on building more homes. They are measuring supply conditions, to a certain extent. I think the reason that metric was so satisfying to them is that measuring it that way makes both cities with high in-migration and high housing induced out-migration look like they are climbing the agglomeration hill. Since they think spatial equilibrium still drives home values, this makes their findings seem robust. I will discuss the other cities more later in the series. Upgrade to paid Coincidentally, a new report was just posted today that seems to be a good example of these issues. It's titled "New York Families Are Doing Better Than You're Told". They compare the distribution of incomes in New York City today to the distribution of incomes in 1970, based on a fixed set of real income bins. And, there are 3 important issues here. 1. They use the period from 1970 to 2026. There wasn't a housing shortage for the first half of that period. At least not enough of one to create huge scarcity premiums in housing costs. I've complained about this before. It's uncanny how analysts who are skeptical of the problem of high housing costs can't manage to confine their analysis to the period where there is a cost of living problem. From 1970 to 1995, New York City was somewhere on the agglomeration hill. 2. They didn't account for migration. Something like 4 million residents (net of domestic newcomers) have moved out of New York City over the last 30 years, generally because of cost of living issues. The missing blue dots. New Yorkers today have higher incomes than New Yorkers did in 1995 because of who left. 3. They apply a uniform inflation adjustment. The housing shortage creates regressive rent inflation in order to motivate poor residents to leave and make housing available for residents with more income. Figure 4 shows excess rent inflation across New York City from January 2015 to today. In New York City, the trend was probably similar to this in the previous 2 decades, also, but Zillow only tracks rents from 2015. | | ---|---|--- Figure 4 So, those three choices make New York City look like its healthily climbing the agglomeration hill. That's a wonderland city. The real New York is the blue dots. You're currently a free subscriber to Erdmann Housing Tracker. For the full experience, upgrade your subscription. Upgrade to paid --- | | | Like --- | | Comment --- | | Restack --- (C) 2026 Kevin Erdmann 548 Market Street PMB 72296, San Francisco, CA 94104 Unsubscribe