In Part 1, I introduced the idea that housing supply and demand within a given metropolitan area are non-linear. Demand is bounded vertically at a minimum and maximum price. Supply is bounded horizontally at minimum and maximum quantities.
The price/income line that is fundamental to the EHT can give key clues about the relative location and shape of the demand curve at any given time in a metro area.
The way to interpret these charts is that supply notes the total number of units. Demand can be related to various factors that affect the use and price of housing: local economic conditions, changes in the average household size, local population growth, and migration.
Note that the y-axis I am using includes income. Rising incomes will raise local home values but in this framework, that might lead to no change in the average price/income level. Constructing it that way helps me conceptualize the situation in a city like LA, where rising incomes of current residents are mostly the result of outmigration because of housing costs. Empirically, there appears to be a maximum price/income households are capable of accepting, and so the demand curve has an upward bound using the price/income variable on the y-axis.
In general, I would think of net inmigration as moving the demand curve to the right - increasing the demand for housing. But, as home prices become elevated, outmigration increasingly becomes dictated by price. In that context, outmigration tends to be an expression of the demand curve - a decline in the quantity demanded. In other words, migration driven by aspirations or choice moves the demand curve left and right. Migration driven by housing costs is a change that happens on the demand curve in the host city - the curve flattens at the top.
When the migrating household arrives at its destination city, it triggers a rightward shift in the demand curve in the destination city, regardless of the motivation for their arrival. But, as I will outline below, there are important differences in how that demand is expressed.
At least that’s how I’m thinking of it. Please correct me if this seems wrong.
Motivated Migration
I have a speculative thought on this. I’m not entirely sure how to visualize this on the supply and demand charts, but I think it is important.
Think of all the Contagion city bubble events, like Phoenix 2006 in Figure 1, or Austin and Boise more recently. In every single case, the rush of demand was from Closed Access cities. New Yorkers and Bostoners to Florida. Angelinos and San Franciscans into various Western cities and tech centers.
I was a bit wrong, I think, in the way I was thinking about what is moving the demand curve when short term demand surges cause price appreciation like in Phoenix 2006 in Figure 1.
As you can see in the right panel of Figure 1 (Phoenix 2006), local demand booms dependably create a different type of change in home prices than limited supply does. Prices rise across the whole city when there is a demand surge. Since homes in the richer ZIP codes have a large luxury component, I associate their purchases with choices and with elastic demand. Families are willing to forgo luxuries if they are too expensive. So, it seems that there is some sort of froth temporarily motivating households to overpay, even when they have choices.
This relates to the categories of demand in the basket of services bundled as housing, which I have written about before:
Shelter
Price settles at cost, variable across metro
Neighborhood Amenities
Price determined by income, variable across metro
Metropolitan Area Scarcity
Price determined by income, uniform across metro
Endowments
Idiosyncratic, determines inter-metro migration
It also relates to the systematic intra- and inter-metropolitan area migration that is set off by inadequate housing supply. As I described it in “Price is the Medium Through Which Housing Filters Up or Down”, throughout a city deprived of housing, families have to make the choice between trading down to a worse neighborhood or moving out of the expensive city. This happens all across a city, not just in the poorest neighborhoods. And, we can infer from the price/income slope for Los Angeles in Figure 1 that families that remain in a housing-deprived city have a tolerance for giving up neighborhood amenities that scales with income.
We can infer from the price/income line that the poorest families in Los Angeles who still remain there are families that were willing to accept housing costs that are associated with a price/income level of about 14. Families with the average income in Los Angeles of a bit under $100,000 who still remain there are families that were willing to accept housing costs associated with price/income of 11.
At each income level, there is a range of families with different tolerances for high costs. The net result of this process is that there is a net outflow of poor residents every year, since they lose the bidding war from the richer residents that choose to remain in the city by trading down to a worse neighborhood. But there are some families at all income levels whose tolerance for compromising neighborhood amenities is lower and who make the decision to move away each year.
So, there is a sorting of families based on these tolerances. The families that remain in the expensive cities have self-selected as families willing to spend more to stay in their home city and/or willing to trade down to worse neighborhoods to stay.
So, a throng of families were moving from Los Angeles to Phoenix in 2006, and they (1) had self-selected as families who are unwilling to trade away neighborhood amenities and (2) were moving under duress. In other words, they have reached their financial limit. They have to get out. In terms of time, their demand for housing in another city is inelastic. They are moving now.
Their short-term demand inelasticity is motivated by their need to get out of the financial distress of the expensive city.
Aspirational movers don’t cause bubbles.
Here is a passage from “Housing was Undersupplied During the Great Housing Bubble”:
Many of those in-migrants were coming from California. They were moving to Phoenix largely to reduce their housing expenses. In fact, even though migration from California had continued to rise up through 2005, net migration into Phoenix had leveled off. That is because increasing numbers of households now began moving away from Phoenix, which had seen soaring home prices. From 2005 to 2008, migration into Phoenix declined each year while migration out of Phoenix continued to rise. By 2008, net in-migration into Phoenix was less than 10,000 households.
Here is a passage from “Build More Houses”:
In 2000, net migration into the major contagion cities consisted of about 28,000 households from the closed access cities and 21,000 households from other areas. By 2005, net migration from the closed access cities was up to 54,000 households, and population pressure from the closed access cities had actually turned contagion city migration (with the rest of the country) negative. By 2008, net migration from the closed access cities had dropped back to 9,000 households, and net migration into the major contagion cities from the rest of the country remained negative—about 10,000 net households moving away.
You see what was happening there? The traditional migrants that normally move into places like Phoenix and the other Contagion cities had elastic demand. When prices increased, they stopped coming!
The source of demand that was inelastic enough to raise housing prices in the Contagion cities was the demand from Closed Access cities from migrants under duress. They had to move, and so their demand was such that it could put pressure on supply that is inelastic in the short term but elastic in the long term.
I’m going to save my analysis of supply and demand in individual cities for part 3. I think this is something worth thinking about and chewing on as its own post.
I have long noted that bubbles only occurred before 2008 where the marginal new households were moving into less expensive cities. In other words, home prices in certain locations rose to levels unsustainable with long-term supply and demand conditions. But, the new individuals who were moving to those cities were moving their to lower their housing expenses.
But, now that I revisit this issue with a few more years of background, with the price/income slope framework, and with the thesis about how systemic migrations triggered by the housing shortage lead to these patterns, I think the case is even stronger than I imagined.
Bubbles are motivated by the outmigration from housing-deprived cities. I think it is a very reasonable assertion to say that the credit-bubble types of motivations that are a popular purported cause of quickly rising home prices are completely inadequate to create those bubble conditions.
Aspirational migration into the Contagion cities, on net, (the cities that had bubbles in Florida, Arizona, Nevada, and inland California) completely disappeared by 2005! Aspirational, hopeful migration was highly sensitive to rising costs.
The only local housing bubbles this country has experienced in the last 30 years have been motivated by financial distress. The credit bubble thesis is WEAK. There aren’t enough ways to emphasize a word to make the statement strongly enough.
Families that move to cities by choice, with hope, wait out temporary price spikes.
We now have had two episodes where migration from expensive cities led to cyclical spikes in home prices in the destination cities. One (2006) happened during a credit boom and one (2020) happened without a credit boom. Is there a single instance of a significant cyclical boom - one large enough to be called a bubble - that was motivated by a credit boom without an economizing migration element?
Maybe Washington, DC before 2008? I suppose the evidence suggests some interaction between credit and supply conditions in Los Angeles and New York City before 2008. Since there was a mass exodus of poorer families out of those cities, that was probably mostly borrowers with very high incomes getting mortgages with risky terms rather than the “anyone with a pulse can get a mortgage for a McMansion in Las Vegas” type that dominates the conceptions about the period.
There are two other mild caveats. Some of the demand coming out of the Closed Access cities is from homeowners who sold at high prices. When they move and buy a new home in a destination city, that isn’t demand under duress or demand that depends on generous lending markets, but it is still time-sensitive demand that isn’t particularly sensitive to price.
And, as Figure 1 in part 1 demonstrated, rents in the Closed Access cities generally moderated or fell because the motivation for moving away was either from the pandemic or from new work-from-home norms. So, again, the demand for homes in the destination cities wasn’t necessarily under financial duress, but it was still somewhat time-motivated and not credit dependent.
Conclusion
The main point here is that the “credit fueled bubble” is the dominant topic of discourse about the Great Recession and American housing. It’s the presumption which the Financial Crisis Inquiry Commission spent hundreds of pages arguing about - what form of excess credit was most responsible. There is a lot of evidence against that thesis that isn’t even on the conventional radar screen.
It is tempting to say “Why can’t it be both?”. Credit-fueled bubbles lead to rising prices where supply is tight and more building where supply is loose. The evidence for that interaction is weak for a number of reasons. One reason is that there wasn’t a building boom in most cities with loose supply conditions.
The gross migration flows suggest that the “why can’t it be both?” thesis is even weaker than it first appears. When the credit was flowing, aspirational migration from less expensive cities to the cities that were booming dried up. Literally turned negative, on net.
I wonder if the same is true of more recent migration flows into Austin and Boise. I suspect it is - that potential in-migrants into those cities took a break and that more locals moved away when the migration from California surged.
I still am wrestling with what the demand curve should look like in Figure 1, for a city like Phoenix in 2006. The elastic demand curve doesn’t quite seem right. But, I think what is tripping me up is that the elastic demand is from all the potential in-migrants that didn’t move to Phoenix when prices increased, or the residents that moved away. The elastic demand curve represents all that pent-up, price sensitive demand that put off their plans. It is the families that intended to move from Tulsa to Austin in 2022 but instead waited until 2024 when prices settled back down.
Consideration of supply and demand within individual cities over time will be the focus of part 3.
I actually think the 2001-2009 economy was a dysfunctional economy due to an underlying energy crisis. So the headline GDP growth numbers looked good but the underlying numbers were always suboptimal in the context of the first boomers turning 65 in 2011 AND we now know they were a healthy generation and the most productive members didn’t start retiring until long after 2011. Also the Bush Tax Cuts led to a record low revenue as a percentage of GDP and defense spending grew as a percentage of GDP and those two elements were the exact opposite of the Clinton economic boom. I could go on and on pointing out anomalous economic data and at some point it’s not nitpicking like Republicans are doing today and it’s all evidence for a very dysfunctional economy.
Ahh… thanks