The capital markets framing cuts through so much noise around housing debates. Fellman's point about density being a symptom of high prices rather than a cure flips the YIMBY supply obsession on its head. The Japan example is particularly telling since permissive zoning coexisted with 10x price increases during the asset bubble, proving that credit conditions and investor demand trump zoning reform every time. His insight about GSEs as redistributive instruments rather than just mortgage facilitators reframes the entire privatization debate in ways that should terrify anyone who thinks housing abundnce is just a construction problem.
I have a lot of problems with Fellman’s points, and I am only about halfway through the podcast. I probably will have even more problems once I am finished.
Fellman says that Austin will get an expensive as New York City and San Francisco once it reaches the density of New York City and San Francisco. This makes no sense for many reasons. Let us simply compare Brooklyn and San Francisco. Brooklyn has about twice the density of San Francisco. In other words, if San Francisco had the density of Brooklyn, then another ~800,000-900,000 people would live in San Francisco. Per a calculation made by ChatGPT 5 Thinking, the cost per square foot to rent the average apartment in San Francisco is very close to the cost of the average apartment in Brooklyn (~$4.4–~$5.0/sq ft/month). Thus doubling built density does not inherently make a city much pricier per square foot. If San Francisco simply matched Brooklyn’s density, the primary effect would be a much larger housing supply (hundreds of thousands more residents), not a sharp jump in price per square foot. Per-sq-ft rents are set by demand and supply, and land-use regulations (broadly defined) determine supply. In short, similar prices at very different densities undercut the claim that “more density means much higher prices”.
Fellman makes a big deal about how cost increases with the density of construction. This point is absolutely true. I believe even that the cost of building higher increases non-linearly; in other words, it costs more and more to add each additional floor.
If someone is arguing that housing per square foot should cost the same in an exurb as the downtown, then someone is wrong. But no one argues this; it is a strawman as far as I know.
The basic idea here is that housing—in a city where supply is not deeply constrained via land-use regulations—should roughly reflect the cost of constructing new housing. If the new housing costs a lot to build (because, say, the new housing is a glass and steel skyscraper), then the housing will cost more. But closed-access cities (i.e., cities with a thick bundle of binding land-use regulations) do not have housing costs which roughly approximate the cost of constructing new housing. To provide an extreme example (and maybe imperfect—these trailers also have private beach access), trailers in Paradise Cove Mobile Home Park in Malibu regularly sell for ~$1-5 million. Manufactured housing, meanwhile, does not cost that much to build out, to put it mildly. Given the valuable floor space here, why does not the land owner build something like the Steinway Tower and make a huge bundle of money? Well, we know why: it is very illegal.
To reiterate the point above in a different way, if the high cost of construction caused the high prices of the Bay Area, then you would not see low-cost construction (i.e., a ranch house) sell for boo koo bucks. But we do see low-cost construction sell for boo koo bucks in the Bay Area.
Fellman also states some misleading facts in his discussion of Japan. First off, Japan’s population is shrinking. However, Tokyo’s population has increased: the core 23 wards grew from 7.97 million in 1995 to 9.88 million in 2024. (The broader Tokyo Metropolis (the whole prefecture) likewise grew from ~11.77M (1995) to ~14.05M (2020).) I do not think it makes sense to talk about Tokyo’s cost during the late ‘70s and ‘80s, because Japan’s growth rate back then was 3-5% per year. A lot of different factors pushed up housing costs, not least of which would have been surging construction costs due to the Baumol-like effects. Japan also does have binding supply constraints in some respects—especially height—so even Tokyo ain’t perfect. Tokyo does show, in my opinion, the simple power of less binding land-use regulation: comparatively moderate housing cost growth (albeit with recent increases) in a very dense city while adding millions of new residents.
Fellman’s points and emphases to me appear as someone saying, “Hey, the demand side still matters.” Insofar, as supply-sider YIMBYs ignore the demand, side, it is a good point to make. Both sides matter. Higher interest rates, ceteris paterbis, mean people have less money to spend on housing. I suspect Fellman invests in markets (e.g., exurbs) where demand matters more.
In 2022, equities fell 19% and shelter inflation hit multi-decade highs (BLS: +7.9% y/y by January 2023). Fellman seems to think this combination is unlikely, but it is not uncommon.
"Additionally, if you object to any of Mike’s claims or think he (or Boyd) is misreading the current housing landscape, we’d love to hear your perspective"
You can't say this, and then get grouchy when someone responds. I mean you can, but it makes you look petty. Especially since you didn't at all engage with his points, you just expressed irritation that he made them.
The capital markets framing cuts through so much noise around housing debates. Fellman's point about density being a symptom of high prices rather than a cure flips the YIMBY supply obsession on its head. The Japan example is particularly telling since permissive zoning coexisted with 10x price increases during the asset bubble, proving that credit conditions and investor demand trump zoning reform every time. His insight about GSEs as redistributive instruments rather than just mortgage facilitators reframes the entire privatization debate in ways that should terrify anyone who thinks housing abundnce is just a construction problem.
I have a lot of problems with Fellman’s points, and I am only about halfway through the podcast. I probably will have even more problems once I am finished.
Fellman says that Austin will get an expensive as New York City and San Francisco once it reaches the density of New York City and San Francisco. This makes no sense for many reasons. Let us simply compare Brooklyn and San Francisco. Brooklyn has about twice the density of San Francisco. In other words, if San Francisco had the density of Brooklyn, then another ~800,000-900,000 people would live in San Francisco. Per a calculation made by ChatGPT 5 Thinking, the cost per square foot to rent the average apartment in San Francisco is very close to the cost of the average apartment in Brooklyn (~$4.4–~$5.0/sq ft/month). Thus doubling built density does not inherently make a city much pricier per square foot. If San Francisco simply matched Brooklyn’s density, the primary effect would be a much larger housing supply (hundreds of thousands more residents), not a sharp jump in price per square foot. Per-sq-ft rents are set by demand and supply, and land-use regulations (broadly defined) determine supply. In short, similar prices at very different densities undercut the claim that “more density means much higher prices”.
Fellman makes a big deal about how cost increases with the density of construction. This point is absolutely true. I believe even that the cost of building higher increases non-linearly; in other words, it costs more and more to add each additional floor.
If someone is arguing that housing per square foot should cost the same in an exurb as the downtown, then someone is wrong. But no one argues this; it is a strawman as far as I know.
The basic idea here is that housing—in a city where supply is not deeply constrained via land-use regulations—should roughly reflect the cost of constructing new housing. If the new housing costs a lot to build (because, say, the new housing is a glass and steel skyscraper), then the housing will cost more. But closed-access cities (i.e., cities with a thick bundle of binding land-use regulations) do not have housing costs which roughly approximate the cost of constructing new housing. To provide an extreme example (and maybe imperfect—these trailers also have private beach access), trailers in Paradise Cove Mobile Home Park in Malibu regularly sell for ~$1-5 million. Manufactured housing, meanwhile, does not cost that much to build out, to put it mildly. Given the valuable floor space here, why does not the land owner build something like the Steinway Tower and make a huge bundle of money? Well, we know why: it is very illegal.
To reiterate the point above in a different way, if the high cost of construction caused the high prices of the Bay Area, then you would not see low-cost construction (i.e., a ranch house) sell for boo koo bucks. But we do see low-cost construction sell for boo koo bucks in the Bay Area.
Fellman also states some misleading facts in his discussion of Japan. First off, Japan’s population is shrinking. However, Tokyo’s population has increased: the core 23 wards grew from 7.97 million in 1995 to 9.88 million in 2024. (The broader Tokyo Metropolis (the whole prefecture) likewise grew from ~11.77M (1995) to ~14.05M (2020).) I do not think it makes sense to talk about Tokyo’s cost during the late ‘70s and ‘80s, because Japan’s growth rate back then was 3-5% per year. A lot of different factors pushed up housing costs, not least of which would have been surging construction costs due to the Baumol-like effects. Japan also does have binding supply constraints in some respects—especially height—so even Tokyo ain’t perfect. Tokyo does show, in my opinion, the simple power of less binding land-use regulation: comparatively moderate housing cost growth (albeit with recent increases) in a very dense city while adding millions of new residents.
Fellman’s points and emphases to me appear as someone saying, “Hey, the demand side still matters.” Insofar, as supply-sider YIMBYs ignore the demand, side, it is a good point to make. Both sides matter. Higher interest rates, ceteris paterbis, mean people have less money to spend on housing. I suspect Fellman invests in markets (e.g., exurbs) where demand matters more.
In 2022, equities fell 19% and shelter inflation hit multi-decade highs (BLS: +7.9% y/y by January 2023). Fellman seems to think this combination is unlikely, but it is not uncommon.
I have a lot of problems with John putting words in my mouth. But alas…..
"Additionally, if you object to any of Mike’s claims or think he (or Boyd) is misreading the current housing landscape, we’d love to hear your perspective"
You can't say this, and then get grouchy when someone responds. I mean you can, but it makes you look petty. Especially since you didn't at all engage with his points, you just expressed irritation that he made them.
Happy we agree then, and I just misunderstood!