I skimmed this paper. The paper compares the rent component of the CPI to two other rent indexes, Zillow’s ZORI and the Marginal Rent Index. The other indexes jumped up quickly and the CPI didn’t and that scared BLS.

The paper’s answer is that CPI measures the rent being paid now and the other indexes measure the rent on a newly signed contract. The contract, obviously, last 1 or 2 years so “the rent paid now” in the middle of the contract doesn’t match the rent on newly signed contracts. Thus, ZORI leads the CPI’s rent component.

Skimmed this paper.

It has how popular fixed and adjustable rate mortgages are by countries in the Euro area. In most countries, just one is popular. The authors try to identify features that make fixed or adjustable rate mortgages popular.

Estimated the value of land under American cities using 68,756 land transactions. The paper gets pretty mathy at points and I didn’t understand what they were doing. But, at its core, it is a hedonic regression. It has choices that I wouldn’t make. They do cross-validate their results. I wish they reported error bounds on their numbers — the price-per-acre in the center of the city is exponentially higher than average, so errors are magnified dramatically. They don’t look at the time series, to see if a change in any factor (like center-city prices) indicates a future change (like residential prices).

Separated samples by MSA, or primary MSA. Defined the “city center” to be City Hall or Mayor’s office. (That’s a problem in NYC! I would have chosen the tallest building.) Some MSAs have multiple city centers (like Minneapolis-St.Paul). Land is associated with the closest center. (That’s bad when a larger one dominates.) I don’t like that they don’t search for the center of the city and don’t try to discover more “centers”.

It uses hedonic regression. Log(land price) is a linear function based on distance from the city center and other factors. Distance is the natural log of 1 + the Euclidean distance in miles. The distance constant varies by city. The intercept varies by city and year. The linear function is also adjusted for lot size and proposed used, which can vary by time. Figure 1 shows a plot from Houston and the linear model seems like a good fit. When they allow multiple powers of their distance function in the linear regression, there is a higher peak downtown and a flattening of the rest of the graph.

They have little data and do some math to create an average city of a given area. I’m not sure how this works. They do a similar thing with changes over time.

To estimate the value of the whole city, they use their function to predict the price at the centroid of each census tract. Then they sum all the tracts in the city.

They find that larger cities have a higher intercept (central land value). Larger cities have a steeper slope. There is much variation here, but the trend is for large cities to have downtown values that are 7 times those 10 miles away. For smaller cities, the value is 1, meaning no change in land value from downtown to 10 miles away. The average land value for cities is pretty flat.

As for intended use, “Retail, apartment, mixed use, and medical proposed uses have substantially higher values, while commercial, industrial, and multifamily uses have lower values.” I’m not sure how “multifamily” is different from “apartment”. It doesn’t list single-family, but the appendix says it is near zero.

The highest central land values are NY, Chicago, D.C., San Francisco, and Los Angeles-Long Beach. The steepest price slope goes to Chicago and then Washington D.C.. The flattest expensive cities are San Jose and Orange County, CA. The higher average land values tend to be on the coast. (Why?)

Total land value is NYC at $2.5T, LA at $2.3T, then SF, D.C., and Chicago. They account for 48% of all urban land value. The total value for the USA in 2010 was $19.1T. An alternate way of measuring, “Flow of Funds”, put it at $6.2T. It only measures private land and covers the whole country; this paper’s method covers all of cities, including roads, parks, etc..

Over time, the land value drops 40% after the Financial Crisis. The Case-Shiller index drops 19%. Their values don’t line up with the stock market’s movements at the time.

This is a great book. It is easy to read. It is hard to summarize, given how much information is in it. It covers not just “exactions” but every aspect of housing and land development. The book isn’t the best organized, but it is full of great information.

The book’s central topic is “exactions”. Exactions are charges or costly requirements for building/property when connecting to city infrastructure, in order to pay for the infrastructure. This may be a charge for roads, water, or sewage. There are borderline may-be-exactions such as “inclusive zoning”, which requires builders of apartments to include some low-rent apartments.

The way cities pay for local infrastructure, like roads and sewage, has changed over time. Initially, developers just built houses and let the city put in after the homes were built. But developers didn’t take the cost of infrastructure into account and either, it was costly to cities or it didn’t get built.

Cities then mandated infrastructure as part of new developments. But that was expensive.

Then, came special assessment bonds. These bonds paid for the infrastructure, paid for by a property tax (or lien?) on the homes there. But, during an economic downtown, bonds defaulted and there were “widespread foreclosures against individual homeowners”.

It sounds like after that period, infrastructure was paid by property tax. But that is often not enough to pay for our expensive infrastructure. Especially after anti-tax “revolts”, like Prop 13 in California. Cities started charging fees or requiring land/buildings when new houses are connected to the infrastructure. Cities have expanded these include “social regulation”, such as including low-cost housing. But, in Berkeley, there are charges for child care, public art, local hiring, etc.. These non-monetary exactions are hard for economists to measure. (NOTE: Exactions have a long history; exaction of land for roads has existed since early American cities.)

The book goes into the political structure of cities. Often there is a “progrowth coalition” made of: landowners, developers, construction companies, law firms, real estate brokerages, other businesses, and organized labor. The businesses expect more customers and more profits. Organized labor expects more jobs.

In the 1970s, there started to be anti-growth coalitions. These were made of neighborhood associations, environmental groups, and anti-tax groups. Good government groups, like the League of Women Voters, have fought for transparency and community input, which have been used by anti-growth coalitions.

The authors comment that developers are actually very weak in city politics. They want permits and zoning changes and, therefore, cower to local politicians. They may organize at the state level.

Often, exactions are good city politics. They are charged only to new comers. The authors mention that state governments care about new comers and see their role as local oversight.

Cities will examine how much a developer can afford. This sounds like extortion to me. Some cities limit growth and start a bidding war of “voluntary” exactions from developers.

The book talks about the legality of exactions. Exactions are part of land use regulations, and regulatory authority is limited. Land use regulations fall under “the police power”, which allows states to regulate for public health, safety, and welfare. This means exactions are fees, which can only pay for the cost of the service. (They are not taxes, which can pay for anything.) Therefore, cities have built complex “linkages” where new business buildings pay for many things, including the low-cost housing their employees will require. (In Massachusetts, the state government made exactions into taxes, to avoid a legal challenge.)

State courts have ruled that subdivision of land is a privilege, not a right, so conditions can be required for it.

Developers encouraged court rulings to be put into state laws. This got rid of uncertainty. And, developers hoped that cities wouldn’t ask for more.

While early city laws set out the rules for building, the decisions have become more and more at the discretion of the city government. Courts accept this.(Even though we no longer have the rule of law!) Developers, in strong markets, are induced give “voluntary” exactions at multiple points in the approval process. Justice Scalia pointed out in Nollan that local governments could tighten regulations just to improve their negotiating position. Officially, “selling” permission is illegal, but many politicians think it is okay to squeeze developers for their constituents. Despite this power, corruption scandals have been rare.

Developers can always walk away. They can also appeal to the state government. And they can negotiate. Against smaller and medium-sized cities, the authors think that developers are often the better negotiators. Aurora, Colorado is an example of a harsh negotiating city. It has water rights in an arid region. The city uses special assessment districts to collect taxes for each annexation.

Chapter 5. The authors point out that infrastructure needs change over time, even without new arrivals. Economic growth means residents get richer and use more resources. Also, they demand better quality, forcing upgrades to infrastructure. E.g., driving more and demanding smoother roads. Also social changes, such as more working women and more divorce, increased demand for infrastructure. Highway demand grew 2 to 4 times faster than population from 1950 to 1990.

The authors point out that sprawl may have 3 different meanings: (1) development on the fringe, (2) ”ribbon” development along suburban highways, or (3) ”leapfrog” development that leaves undeveloped land inbetween it and the city core.

The book contains 2 studies that tried to measure actual costs of development. The numbers vary. It seems low-density single-family housing doesn’t cost that much more than higher density single-family housing. The infrastructure usually forms a tree — it grows slower than the area of land and has most costs in the nodes. Costs do drop by 1/3 to 1/2 when you get to townhomes and higher density. Those are new-built costs; infill development is expensive.

Unplanned communities had transportation costs that were 3% higher than planned communities. So, freedom doesn’t cost much more.

Chapter 6. Cities are bad at assigning costs to residents. Some financial restrictions, like bonds with short terms than the lifespan of the infrastructure, increase these.

It sounded like some cities specialized in housing for the rich (”bedroom communities”) and other cities specialized in commercial development. It separated housing from business. And no city wanted to specialize in housing for the poor.

Some cities use exactions to charge commercial development for the costs to workers’ housing. Others use exactions on residential development to charge workers for the cost of their housing.

The authors talk about different city expenses. Some are good for usage fees, like water. Others, like education, police, and helping the poor, we believe everyone should pay for. It is unclear exactly who should pay or how they should pay for the second type of services. Some development might be ”profitable” for a city to do on its own, but not when you include a given share of those additional costs.

(It was easy for me to see that a liberal city might spend a lot on education, helping the poor, etc.. And have a lot of those expenses might find it easiest to find the money by putting exactions on housing. But that might increase the need for subsidizing those services, leading to a spiral of increasing costs.)

Fast growth drives up costs for cities. But it’s hard to say if that’s really expensive or an investment. Fast growing cities and stagnant ones have the highest education investments — mediocre growing cities have the least. That’s strange.

Chapter 7. Developers try to pass on the costs of exactions to landowners and customers. It is usually true in the long run, but there are some exceptions. If there’s substitute land nearby (in cities with lower exactions), developers don’t. Another exception is during periods of high growth, where there are not enough developers to keep up with demand. Another is if local government restricts the supply of building permits. (Does this include having a large minimum lot size?)

The authors cite a few empirical studies. In Loveland, a fee increase led to a 3-times higher increase in home prices. Similarly in Dunedin. In multiple cities, land prices increased. So, rather than developers passing costs onto landowners, landowners expected higher rents in the future and demanded more money for their land!

Exactions effects on commercial real estate is tougher to measure. Costs get passed on to customers, employees (through wages) and suppliers.

An increase in exactions hurts landowners on the fringe of a city. Their land is much less likely to be bought for new development.

There are big questions of equity. When exactions increase, existing homeowners gain. Existing renters and all future residents lose. This also shifts money from poor to rich and from younger generations to older ones. Fixed-fee exactions are regressive, when compared to property tax, which falls heavier on the rich.

The poor rarely have to pay the full cost of police, education, and other services. But cities often charge them full price for water, sewage, garbage, and transportation.

Chapter 8. To achieve efficiency, economists recommend charging users a usage fee based on the marginal cost of supplying more infrastructure services (e.g., per gallon of water or per bus ride). They also recommend exaction fees based on the marginal cost of connecting the development to the infrastructure system. Sounds great. But that almost never happens.

Cities rarely take into account the costs to outside communities. (And courts encourage that.) The ”rational nexus standard” laws, which say cities can only charge fees for costs that can be specifically attributed to the new development/residents, encourages conservative measurement of costs to avoid court challenges. Cities also lower prices to remain competitive with nearby cities, especially for non-residental uses. Cities often provide a discount when a project can show it uses less infrastructure than average ... but then don’t charge more when it will use more than average.

But it goes much farther than that. Cities often charge based on the average user charge, not the marginal one. And they don’t change charges based on location, time of day, or season. For long-term costs, they charge resident based on the city’s current debt rather than the replacement costs of the infrastructure. Sometimes, they grandfather in older residents or older homes from true costs.

When marginal costs allow large profits, the authors suggest using ”lifeline” pricing, where a base amount is provided cheaply and each unit above it is priced at the marginal price.

There are difficulties to usage fees. It can be expensive to calculate the marginal cost, especially for every location and time-of-day. The authors consider if exactions can approximate the usage charges and say that they don’t, but they’re probably better than the other alternatives.

Where cities do implement usage fees, like metering water, usage drops sharply. Leaks are fixes. Lawns are watered less. The authors claim that, if prices are sufficiently high, residents will install low-flow toilets and shower heads. Usage fees would allow private business to compete. And might even let cities delay expansions indefinitely.

Chapter 9. The authors found exactions in many states in 1991 averaged $12,000 per unit. But costs to most cities were probably around $4,400 to $4,900.

Exactions are politically convenient. They are not visible on a tax bill. Developers hide the fee, as they try to pass the cost onto others. Developers don’t fight them; they care more about delays and need politician’s good will. And, unstated by the authors, it increases homeowner’s assets.

The authors point out that planning commissions have changed. They used to be independant and quasi-judicial. They resolved private conflicts. But, now, land use is about ”community objectives” and tied to environmental and fiscal policy. They commissions are political.

Likewise, regulation has changed. When it began, it prohibited anti-social behavior: conspiracy to drive up prices, adulterated products, and unsafe transportation. Next, it controlled behavior in some circumstances: it was okay to be loud and polute but only in this zone. Now, it is used for revenue generation. Where there is ambiguity in causes, there is a political fight to put costs on firms or people.

At federal and state level, population growth is a pure good. Income and sales tax increase. At the local level, population increase causes negative externalities. ”Linkage theory” connects negative externalities to new development and cities charge for it. (But I doubt that new development gets credit for the positive externalities!)

Different people have different values. Local advocates often want existing residents to be unharmed. (Is that ”liberal” or ”conservative” under Moral Foundations Theory?) Local homeowners hate property tax. It increases their monthly costs and immediately lowers the price of their home by a lot. ”Not surprisingly, therefore, the property tax is the central focus of the contemporary tax revolt.” (I wonder if that was even worse when prices were increasing by 10% to 15% per year in the 1970s, when Prop 13 was passed?)

After bitching about exactions for 100+ pages, the author wilt. They say they are ”highly imperfect, subject to abuse and requiring state oversight”. But often preferable to other anti-growth and anti-tax measures.

This book had a ton of good info. But it was ordered strangely. I would have expected the background about actual costs first, before setting up the fight over who pays for it. They also should have said usage fees + exactions are optimal earlier. Their conclusion is odd. If usage fees and exactions are optimal, why do they seem to complain about exactions for most of the book and sound resigned in their conclusion? I would have thought they’d spend more time on trying to discuss the right rules at the state/national level to push cities towards better usage fees and optimality. As much as I complain, I got a lot out of the book.

This paper creates rent indexes. The authors think that inflation is miscalculated because it isn’t handling rent correctly.

Salim Furth mentioned this paper on Twitter. He said that prices remain high because the population is elastic — if prices drop a little, more people move to the city, propping up demand.

Focuses on William A. Niskanen’s theory that bureaucrats maximize their agency’s discretionary budget. Presumes it can be measured by looking at City Managers who contract outside. When going outside, the price is visible to their “political sponsor”. So, if City Managers contract outside more than other top city officials, it validates the theory.

Concludes: “City manager cities are slightly more likely to contract out. This general pattern of contracting out is more in line with the professionalism in government hypothesis than the greedy bureaucrat hypothesis.”

The author suggests metro-area GDP-linked bonds. He calls them “GDI” or Gross Domestic Income bonds. The author claims that renters would want to buy these, since they hedge the cost of housing. Likewise, homeowners would want to sell them, to hedge their ownership.

This is an innovative paper that showed a lot of statistics on cities follow a power-law distribution of the city’s population. An important statistic is GDP and, because the coefficient is greater than 1, it means that GDP grows super-linearly with the size of the city. Or, equivalently, that GDP-per-capita grows with the size of a city. But many other statistics follow the same pattern: new patents, R&D spending, bank deposits, crime, walking speed, gas stations, road surface, etc.. The coefficients are different for different statistics. For roads and gas stations, the coefficient is below 1, so the larger the city is, the fewer gas-stations-per-capita.

The authors go further and set up an differential equation for city growth based on a power law distribution. I don’t like this extrapolation. It leads to infinite growth in a finite amount of time. The authors get around this by saying that innovation must happen which reset the equation to initial conditions. I don’t like that fix. It would be easier to add additional terms to the differential equation. E.g., crime or disease grows with a different coefficient, restricting city size.

It’s an interesting paper. The collection of different statistics governed by power laws is very interesting.

Have not read. Brendan O’Flaherty mentioned it as a paper that studied homelessness and inequality.

I skimmed part of this paper.

It covers the design decisions for a house-price insurance plan for Syracuse, NY. Syracuse’s economy had shrunk and home prices were falling. No one wanted to buy a house there, because they feared they were bad investments.

I read it’s history section (#3).

Home equity insurance was legalized in California in 1935, but then made a felony in 1939. No explanation is known.

Oak Park in Illinois, a women’s group wanted to integrate the neighborhood. They feared “white flight” and dropping home values if Black families moved in. They created an insurance scheme For a one-time fee of $175, homes were covered for 80% of losses after 5 years. The program was paid for by a tax on all homeowners in the neighborhood.

99 houses bought insurance in the first 4 months. This number rose to 1500. But it had fallen to around 150 by the year 2000. Insurance was not popular because it insured only the neighborhood, not movement in city or nation-wide home prices. Also, doing a proper sale to claim the payout was onerous.

At the time, insurance inspired confidence that a “bank run” like exit would not happen. Later, it inspired fear, because its existence implied that insurance was needed! The program was ended.

The Illinois state government in 1988 allowed new taxing districts to be created for insurance purposes.

The author did not want to copy the Oak Park scheme for Syracuse. It didn’t cover Syracuse’s regional downturn. If that was changed, since it was backed by taxes, the higher taxes could precipitate an exodus and financial collapse of the city and housing market.

I didn’t read all the details, but they proposed insurance based on an index of the Syracuse area housing. The insured could redeem it when they sold their property. They evaluate it on historic data.

They look at the legal aspects. It is not insurance (in NY state law). If it is a mortgage, it is illegal, because it changes the value of the property! Regulators worked to make it something new.

An interesting feature is putting the mortgage lender first in line to receive payment. This makes it more like mortgage insurance. It encourages more loans.

It went on sale in 2002.

I skimmed this paper.

Used data from HUD’s Section 8 voucher renters. They matched results of renter surveys with Census data on tract. (Tracts have 1,200 to 8,000 people.) Calculated results for 331 metro areas and non-metro areas of 49 states.

“one-bedroom apartments rent for about 19 percent more than efficiencies, two-bedroom apartments for 35 percent more than efficiencies, three-bedroom apartments for 53 percent more, and each additional bedroom adds about 10 percent to rent. Living in a census tract where the mean travel time to work is 30 minutes longer reduces rent by about 10 percent. Households with an additional person per bedroom plus one pay about 14 percent more for an identical unit.”

Other price indices (ACCRA’s cost of living index, median rent, HUD’s Fair Market Rent, and indices derived from PUMS and AHS) all diverge from their price index.

Distance to city center was done as the average commute of the census tract. I’m pretty sure I disagree with that in general.

If the other price indices diverge from this one, how do they know which is “real”?

This is a tedious book that has a really good core idea and stretches it out to 220 pages. The core idea is that suburbs are expensive. There is a high cost for the infrastructure to attach a suburban house to the road network, water network, sewage network, etc.. This is because suburban houses are far apart. Cities often don’t notice that those high costs often exceed the property tax that is collected from the house. Thus, suburbs hurt city’s finances. The same applies to “big box” stores on the edge of a city.

It would have loved if this book had a good “aha” story, elaborated on the core idea, provided examples, and discussed solutions. But that’s not what this book does. It talks about “complicated systems” vs. “complex systems”. It messes up finances. (The author doesn’t count suburban infrastructure as “assets”, but calls them “liabilities”. What he should call them is “non-performing assets”. ) He talks about pre-building infrastructure vs. waiting for demand. He talks about “infinite baseball games” in one of the worst metaphors I’ve seen. He gives bad investment advice. And more. This book is a mess. The author is a professional engineer and doesn’t include a detailed example with numbers. WTF?!

The author does explain that the state and federal government helped cause the problem. By subsidizing new infrastructure, especially road construction, its costs do not show up in the city’s budget. In fact, building more infrastructure can help solve temporary budget problems! Only when growth has stopped and that infrastructure needs to be replaced do cities see the financial mess they are in.

Because this book is disorganized, it doesn’t say clearly how to measure the problem and how to fix it. I believe the best way to measure the problem is to assign infrastructure costs to each address and then compare that against the taxes paid. The solution, which the author stuck in corner, is for cities to prioritize the “profitable” parts of town, where revenues exceed costs. If there’s a surplus, the city should try to preserve the parts that don’t lose the city too much money. To avoid bankruptcy, cities will need to abandon the least profit parts of town. And those least profitable ones are the fringe of suburbs and big box stores.

I do not recommend this book. Its idea is very important. You should know the idea. But I think this summary explains the idea clearly and you will not gain any more insight about the idea by reading the 220 page book.

Skimmed this paper.

It is reportedly a landmark paper (first paper?) in measuring agglomeration effects. That is, that productivity is increased when we have a lot of people near each other.

It focuses on counties. It assumes productivity increases multiplicatively with density. I skipped a lot of math in its model. Workers have a preference for low-density locations, which prevents everyone from moving to a single city and being super-productive. Page 59 and 60 mention land prices, but I’m not sure it is in the model.

Data was labor input, that excluded the self-employed. It came from the Regional Economic Measurement Division of the US Bureau of Economic Analysis. They excluded data from states where mining was more than 15% of output: Alaska, Louisiana, West Virginia, and Wyoming.

They adjust density with education, state-wide effects, and population of county (without taking into account its area). Education has a big effect, but the others do not.

The result is that doubling density is correlated with an increasing labor productivity by 6%.

I’d need to read the paper in detail to figure out what this really means. That is, that the number does capture productivity. And figure out how land prices work into this. I don’t like that it is on a county basis — that’s such a large scale that it is surprising to see any effect. My data on land prices in Austin shows that prices change dramatically in a few blocks; Travis County is many many block-widths across. If density increases exponentially as we get to the center of Austin, the average density of Travis County doesn’t measure that peak. What I’m saying is that we need good find-detail data for measuring cities.

The authors felt that the press, public, and even policy makers had a mistaken view of the causes of homelessness. They wanted to present simple plots of data showing that many commonly discussed “causes” are not the cause. And that the price of housing was.

Page 13 and 14 have a great metaphor: musical chairs. A person with a cast doesn’t get a seat, but it isn’t because he cannot get to a chair. It is because there is 1 fewer chairs than people. Such is homelessness — we see a lot of individual factors (mental health, addiction, race, etc.) that affect who doesn’t gets a home, but we need to look at citywide factors that cause a home shortage.

The data is strange. They want to use the 35 largest cities by metro area. (Why 35??) They drop 6 because HUD’s Continuity of Care (CoC) region covers “too large a geographic area”. The dropped cities are Houston, Riverside (CA), Denver, Orlando, Pittsburgh, and Kansas City (MO). The rest are split into two groups, where the CoC matches the city (20 of them) or the county (10 of them). Chicago appears in both groups. They cover years 2007 to 2019, but admit that the PIT Count (their measure of homelessness) is iffy in some cities and especially in the early few years.

Over time, homelessness has dropped on average. But the variance has increased.

They talk about the PIT Count (”homeless census”). Most homelessness is temporary, but some is long-term. Because the PIT Count is a single day, it emphasizes the long-term. If it was people over the year, the temporary ones would show up more.

Homeless are more likely to be: male, unmarried, low-income, old, non-white, and queer. They are socially isolated: fewer family ties and far from family. They are more likely to be depressed, mentally ill, abused drugs, abuse alcohol, have a conviction and been incarcerated. 70% are individuals, usually male. 30% are families, which tend to undergo temporary homelessness.

Cost per homeless is $30,000 to $100,000 for government services.

The book has good data on poverty, which is negatively correlated with homelessness. Unemployment is negatively correlated with homelessness. It is odd that they plot data points for every year and don’t estimate fixed-effects for cities. (E.g., does an increase in poverty cause an increase in homelessness?)

Mental illness and drug use are measured on a statewide-basis. That seems to large a scale.

Race is negatively correlated with homelessness. Black people are more likely to be homeless, but cities with more Black residents have a lower rate of homelessness!

Temperature is handled badly. They use the average temperature for the month of January ... even though the PIT Count was on a specific day? The county data shows no correlation of warmer winters with homelessness, but the city data does. They fight that result by combining the data using “indexed homelessness” (described on pages 20 to 22) which is, let’s say, not a well-known technique. The anecdotal evidence — that cold NYC has more homeless than warm southern cities — is much stronger than their data.

They admit that temperature is correlated with unsheltered homeless. They explain that this may be because cold cities spend more on shelter than warmer cities, which may spend on outdoor services.

Generous social services is also done differently. It measures the state’s maximum TANF (”Welfare”) benefit against family homelessness rate. It uses families, because individuals rarely qualify for TANF. This does not seem like a good measure.

The show that low-income migration is negatively correlated with homelessness. They try to say this shows the cities with high homelessness are not “welfare magnets”. To me, that data means other things.

They handle political party anecdotally, citing that Chicago, Detroit, and Cleveland are highly Democratic cities with low homelessness. The authors inveigh against harsh police treatment of the homeless and say there is no evidence that it does more than hide the problem.

They do a strange interlude to look at cost-burden-ness as a cause of homelessness. They find when the median person spend more on housing, homelessness goes down in cities. But not in counties. And the result doesn’t hold when looking at the “first-quartile housing cost burden”.

What should be the finale is the comparison of median rent, vacancy, first-quartile rent, and “first-quartile vacancy” (vacancy in lowest 1/4 rent apartments). Rent and vacancy are inversely correlated and both predict homelessness. The strongest is median rent which has an R-squared of 55% for cities and first-quartile rent which has an R-squared of 28% for counties. (For vacancy, they regress against log(vacancy), which is a strange choice since both homeless rate and vacancy are on a scale of 0 to 100%. They don’t apply log to rent, where it would be more natural. And they don’t do log(vacancy) when showing it correlates with rent. These are inconsistent strange choices.)

Page 134 says that they did a full model with fixed-city effects ... but don’t show its results. Not even in an appendix. WTF.

The rest of the book doesn’t follow their thesis. They see if growth is correlated with homelessness and find it isn’t. They mention a study shows that inequality is causal ... but say it isn’t in their data (without any plots).

They eventually say that economics believe supply is limited by geography (oceans, mountains, etc.) and by regulation. They then group cities by elasticity (a measure of supply limitations) and growth. (So, now, they’re saying growth does matter.) And the worst cities have high growth and low elasticity. I expected the book to end here saying “we need to fix the regulation in high-growth cities”.

But, no. We get 38 pages of their vague plan to fix homelessness. After a data-driven book focused on proving one point, it is huge jumble of opinion at the finish. They dismiss filtering without a citation on page 134. They put together a mushy argument against markets on pages 171 to 173. They say private markets won’t create houses for people with no income, but ignore the idea of giving money to those people. After citing economists earlier, they say “housing must be de-commodified”. They mention that building houses has not come down in price like building iPhones, but don’t mention that most cities ban manufactured homes on most land. It goes on: public perception, funding, and more. Way too much for a single chapter.

The book’s data is passable — it is clear enough to make most of their thesis’s points — but I think it needs a much stronger analysis in an appendix. The book never really addressed how to handle correlations in data samples from the same city in different years. The writing began well but it totally flubbed the landing. The last chapter should have been cut and become the basis for the author’s next book. But it did well enough to make its point and that’s an accomplishment.

This paper tries to compute a land price for a large portion of America’s cities. It uses property appraisals, probably for mortgage applications, from the GSEs (Government Sponsored Enterprises, like Ginnie Mae). It puts those into a model to find land prices. It interpolates those land prices to all land using an averaging technique called “kriging”.

The model is pretty simple. It says that, for land plus a structure, the value of the structure decreases over time. The results of the model are two. For recently build houses, the land is probably valued at the price of the whole property minus the replacement construction cost for the building. For very old houses, the property is almost all land-value and not building-value. I don’t like this model — there is an option-value not accounted for: to tear down a building and rebuild whatever is most valuable then.

They limit their data to homes where the “effective age” is less than 15 years old. I guess these are well-maintained buildings. They drop appraisals that they think are generated from tax appraisals (which they don’t trust). They only allow reconstruction costs from a handful of well-known sources.

They adjust for the size of the lot, etc..

Given those land prices, they interpolate the data to all lots using kriging. That’s an average of nearby known data points, weighted by distance. They used the 20 closest points. It looks like weights were determined for the closest 6.9 miles in 15 bins. Its a good technique, but it does use averaging, so it doesn’t follow trends to predict peaks or cusps.

They generate prices for 960 counties, 7,742 ZIP codes, and 10,515 Census tracts.

They show prices as a distance from the CBD for various city sizes. They go down almost exponentially, as you get farther from the CBD. Larger cities are more expensive than smaller ones. Cities with more regulations have higher prices. Cities with more natural barriers have higher prices.

In larger cities, the land makes up more of a share of the property price. It is 40 to 45% downtown in cities with over 2 million housing units.

Figure 6(a) shows how counties with more expensive land have more square-footage of structures. But the slope is smaller and the variance much higher than I expected. Figure 6(b) shows that expensive land is usually associated with smaller lot sizes. That has a high variance too. Perhaps counties are too large and that’s causing the high variances?

Figure 6(c) compares county land prices for housing and for agriculture. The authors say it “shows a strong positive relationship”, but the scatter-plot looks odd. And the scales are off completely. The median agriculture price for a county is $3,000 per acre, while the median single-family land price is around $100,000 per acre! I don’t get it.

Figure 9(a) is interesting. It breaks the counties into bins by the number of single-family units. The counties with the most houses saw the largest price increases from 2012-2019. Also the most increase in land value — especially in the top bin. That was enough to drag the national average to a 7% annual increase, even though smaller counties averaged a 2 or 3% increase.

I think this is a good paper, but it’s mostly because I see it as an engineering project. The model and some choices aren’t ideal, but they got something that works for the whole country. I’m interested to download the data and take a look at it.

This is a very good article. I dislike the author’s conclusion, but the article identifies a problem and marshals a lot of data to demonstrate it.

The author attacks the idea that government should protect housing as an investment.

Making money on housing is mostly about luck. “price appreciation accounts for roughly 86 percent of the wealth associated with owning a home.”

Government housing policy is “built on a contradiction”. Politicians announce higher house prices as a good thing, but it is obviously hurting people trying to find housing.

Investing in a single risky asset is a bad investment strategy. The lower your wealth, the more you are likely not to have other assets.

Inequality is inevitable. She mentions that housing has not been a good investment for Black families, which experience racism and depressed home prices.

The author’s solution is: “Policy makers should completely abandon trying to preserve or improve property values and instead make their focus a housing market abundant with cheap and diverse housing types able to satisfy the needs of people at every income level and stage of life.”

I don’t like the solution. It ignores all the good things about homeownership. It recommends government mistreat an investment good, and one that represents half of all investment!

This is a great paper which covers traffic congestion in cities. The show that cities with more commuters have more traffic congestion and that the congestion follows a power law. But then they look at individual cities and its different!

Some cities follow a power law, but with a different exponent. Other cities’ plots of population vs. delay have two regimes: a power law where delay rose quickly until a certain amount and, then, a flatter portion following another power law. The authors believe a capacity or behavioral limit was hit. E.g., in Birmingham, the distance driven per driver was increasing until 1998, at which it started to decrease and the delay-per-driver flattened.

The delay is on a log scale, so fast growing cities really hit a wall in congestion. So, Cincinnati started on the steep portion of the plot with 10 hours of delay per year. A decade later, it hit 25 or 30 hours of delay per year and then started to flat-line. Three decades later, the delay was still under 40 hours per year.

I’m fascinated by the paper’s Figure 2. It shows that the power-law coefficient for all cities dropped from 1980 to 2007, and then flattened. So, delay was much worse in large cities in 1980, but the difference between large and small cities got less over time. Figure 3 makes me believe that this is because delay got worse in smaller cities, not better in large ones. Maybe it is because more small cities grew into larger cities and hit their capacity limits?

Uses a public dataset from Texas A&M.

This paper has an interesting insight, but the idea could have been presented better.

The ”price puzzle” is that, when interest rates are raised, prices also seem to go up rather than down. The paper’s main finding is that interest rate increases shift people from buying to renting. This drives up rents, which are a major part of the CPI and PCE inflation measures. Prices of non-housing goods decrease as expected. This doesn’t solve the price puzzle but ”greatly ameliorates” it.

The authors accurately state: ”Although housing was not completely absent from the macroeconomics literature before the global financial crisis, it was seen as a minor component of the economy which did not deserve special attention.” I find that a valid damning critique of macroeconomics. Housing and land is 50% of stored value — half of all investments. And housing is the largest consumption good in almost every family’s spending. That macroeconomists didn’t consider it worth of attention is a stain on their work.

A 0.25% increase in interest rates leads to about a 0.1% increase in rents. ”Price Level”, which I believe is non-rent goods and services, has a 0.1% decrease. (Interest rates were measured using 3-month ahead Fed Fund futures.)

The authors believe the mechanism is the renter/buyer’s tenure decision. That is, how long to stay in a place.

The authors need to label their graphs better. They have ”Housing Rents” on page 6 and on page 8, with different graphs. I think the page 8 is houses (not apartments) and page 6 is all housing, but it is not clear.

If I’m interpreting this right, house prices take 12 months to see a 0.1% decrease. Also over 12 monhts, house rents increase by 0.05%. (NOTE: CPI delays some rent values, because they use the average rent and it takes time to contracts to expire and be re-signed at the higher rent.) Vacancy rates and homeownership rates drop almost immediately after a shock. (Vacancy depends on the supply, but there isn’t a supply graph.)

Including housing in CPI and PCE dampens the scale of price changes. Without it, CPI moves about twice as much in response to a credit contraction shock. PCE, which puts a smaller weight on housing, moves about 50% more.

I’d be interested to see what the authors found if they did this same study on cars. Cars are a large purchase often bought on credit. Cars are very different - traded more often, more mobile, less regulated, and operating costs are a larger portion of the spending on them. Still, the comparison might be enlightening.

This is a good paper. A good insight, but could have been presented a lot better. The authors need to label their graphs better. I wish they had done a better job of explaining the price puzzle, reproduced previous graphs about it, and then showing how the graphs change with this new insight.

An impressive paper. This paper tries to determine if building roads or widening lanes is better for a city. The title comes from a quote from Downs 1962, that the “fundamental law of highway congestion” is that, if you widen a congested highway, the traffic will increase until the congestion is just as bad as before construction. The authors say that they verify the statement: “increased provision of interstate highways and major urban roads is unlikely to relieve congestion of these roads”. They also find “no evidence that the provision of public transportation affects” vehicle-kilometers-traveled (VKT).

They measure how VKT changes as more lane-kilometers of road are added to a city.

They study 228 MSA (metro areas of cities), in 1983, 1993, and 2003. Roads have 6 different classifications. They ignore local roads. They study interstate highways separately. The 4 other classes of non-local roads get grouped into “major urban roads”. I didn’t find a description of “local roads” or how much of a city that makes up. (Are Manhattan’s streets “local roads”?) For interstates, they separate them into the “urban” and “non-urban” parts, but I don’t have a sense of how that division was made.

Under plain OLS, for interstates in urban areas, the coefficient between ln(roads) and ln(V KT) is very close to 1. R-squares are .95 to .96, too. The coefficients for “major urban roads” is .70 to .88. For non-urban interstates, they’re from 0.82 to 0.84.

The coefficients for ln(population) is “much less than 1 in all specification” ... but seems to be 1.01 in Table 3, regression #3. All others are in the range .5 to .3, which says population doesn’t affect driving as much as building roads.

They realize that congested cities are more likely to get roads added to them, so they use instrumental variables to separate that causality. But they only do this for all interstates. Not “major urban roads” and they don’t break interstates into “urban” and “non-urban” parts. The instruments are old maps of the railroads, explorations, and the plans for the interstates in 1947.

For the IV regressions, they get coefficients around 1. They say their preferred estimate is panel A, column 3, which has a value of 1.03.

They do another IV regression, but for how much buses (not trains) contribute to reducing VKT. Their instruments are the highway plan, railroad plan, and the 1972 election. They use an “ILML regression” rather than 2SLS “since our set of instruments is sometimes marginal weak”. The coefficient for buses is close to zero and, in some regressions, statistically indistinguishable from zero.

They do a regression that tries to show that there’s a “natural level of traffic”. I didn’t understand what they did here. They got negative coefficients and claimed there was mean reversion.

They try do determine where the additional interstate VKT are coming from. They look at commercial trucking, residents, population growth, and diversion from one set of roads to another. They find commercial trucking increases with interstate lane-km. As does residential driving, a little. Population too. But more interstate lane-km doesn’t divert very much from other roads.

For a traffic increase, they conclude: trucking accounts for 19 to 29%, population increase for 5 to 21%, residential driving for 9 to 39%, and diversion from other roads 0 to 10%. The true values would have to be at the upper end of those ranges to account for all the traffic.

The authors mention congestion, but the paper doesn’t actually mention delay. They show coefficients around 1 for urban interstates in almost every test. Still, it seems those highways carry more cars if population increases. Maybe not at rush-hour? I would have liked to see the breakdown by “roads that have stoplights” and “roads that do not”. As for the expanded capacity, it is used by more by local residents and diverts capacity (thru-truckers and migrants) from other cities. So it is productively being used, not just causing needless delays.

Maybe we need to expand interstates even more? I’d definitely prefer congestion pricing, though.

Did not read. It got suggested on a reading list and sounds interesting.

This was difficult to read. I had it recommended to me, so I read it. But I ended up putting a copy into an editor and massaging it until I got its arguments.

This blog post’s big point is that we’ve split the housing market in two: one for landlords and another for owner-occupied housing. The owner-occupiers get tax benefits, mortgage deduction benefits, etc. and pay less per-year for their house. That’s an economic rent. To exclude people from gaining it, there’s a bidding contest and the creditworthy and wealthy win it.

The author packed a lot of other points into the blog post. One is that a property is a risky asset. In a well-run market, that risk should be diversified by combining it with other risky assets. Under Market Portfolio Theory, only the (entire) market risk has a risk-adjusted excess profit. But we’ve structured the housing market so that houses are not diversified. Thus, homeownership comes with a risk premium. This adds to the benefits of homeownership. (If you’re wealthy enough to survive the risk!)

Politicians say they want everyone to be middle class and owning a home is part of that. But benefits of owning a home come from the economic rents, so they cannot be available to everyone.

The author says that the economic rents are reserved for the creditworthy. In the lead up to the Financial Crisis, credit was extended to a lot more people. This expanded the number of owner-occupiers bidding on houses and prices rose. When credit was withdrawn, it was left to landlords to bid on houses. Given their higher per-year costs, they couldn’t bid as high for the houses. Thus, the crash. (This paragraph has been what I believe the author meant; it wasn’t perfectly clear.)

This blog posts contains some interesting ideas. I wish it stated them clearer.

The author talks about the agency costs of renting housing.

If the law gives more rights to the tenant, the landlord raises rent to pay for the costs and risks.

For low-income housing, renters are more credit constrained and less able to be owners. They are also higher risk. Yields earned by landlords are especially high.

The author makes a very interesting comment that usury rates should be determined, not as fixed value, but relative to the rental payment. This would allow ownership anywhere the mortgage payment (even at a high interest rate) was less than rent. That would open up ownership to more people.

The author doesn’t say much about customization of the property. Nor further investment in the property.

The author enumerates some of the benefits provided by landlords: lower transactional costs (vs. buying and selling house), loan of capital, and diversification.

Skimmed. I find its graphs on page 7 to be very interesting. I had discounted ZIP code based analysis before, but this shows it is very useful.

Figure 1 on page 7 has scatter plots of rent-to-income ratios for zipcodes in Boston and Atlanta. Rent-to-income increases as people get poorer. For Boston, it curves upward sharply, with poorer zipcodes averaging 80% of the their income on rent! WTF! In Atlanta’s graph, rents peak at 45% of income. (The graphs use log income, which explains some of the sharpness, but still that’s how I would graph it.) Unless Boston has major housing assistance programs that don’t show up as “income” (e.g., rent stabilization), something is very wrong.

The same figure has house price-to-rent ratios. The price-to-rent ratios seem to be linear, with more variation in the richer neighborhoods. The lines are increasing in income. I’d expect an equal price-to-expected-return in each city. The increasing slope makes sense, because the fixed costs of renting cut into the returns of lower-income housing. But why are price-to-rent ratios in the 15 to 38 range in Boston, but 5 to 25 range in Atlanta? Does Atlanta have higher taxes or fewer amenities? Are construction costs cheaper in Atlanta, capping prices? S&P 500 had a price-to-earnings ratio of 21 to 24 at the time; to go higher than that, I have to believe Bostonians expect rents to increase in the future. Or that ownership is subsidized.

The same figure has price-to-income ratios. There is good data there (rents are scarce), but I don’t think it is meaningful. It is interesting that, in Atlanta’s data, price-to-income is flat around 3 for all income levels. Boston’s is 3 at the high end, but goes as high as 12 for low-income ZIP codes.

Figure 32 on page 69 has the slopes of price-to-income for various cities over time. Austin’s is pretty steady around -1 to -2. NYC and LA have extreme values that went as far as -5 in the lead-up to the Financial Crisis.

Figure 33 on page 71 has data on Austin. Lower income zipcodes have higher price-to-income ratios. Figure 34 on page 72 shows that Austin’s property taxes are almost twice Boston’s and Phoenix’s.

Figure 15 on page 44 shows that some cities have non-liner price-to-income ratio plots. Figure 30 on page 66 resolves this issue. After property taxes are accounted for, the plots become more linear. Property tax was gotten by looking at the average property tax claimed in each ZIP code in federal income taxes.

Interesting data. I may try to read it in full later. The scatterplots show ZIP codes as circles; I wish those circles were sized based on the population.

The paper claims that zoning took off in the 1910’s because of large gas-powered vehicles. The bus and truck allowed apartments and industry to operate on land in the suburbs. Previously, hauling goods was expensive and most people moved by train (street car).

With the large vehicle’s ability to go on any road, it was necessary by law to protect the neighborhoods. Developers used covenants for a time, but didn’t protect the edge of the neighborhood. Then they pushed for legislation.

The author says that other excuses for zoning were present at other times. But zoning took off all over the USA in the 1910s. It was enacted to preserve home prices in an era of technological change.

The author blames zoning for the fragmentation of American metropolitan areas into many cities. Previously, if there were two separate cities, the urban downtown of the larger could expand into the smaller one. The smaller town would merge with the larger, to get its services as business developed. But, with the new legal power, smaller cities could remain suburbs and stop incursions from commercial and industrial users.

Regional government entities were proposed, but killed. Local control stayed.

A great quote: “That the detached, owner-occupied home is at the top of the zoning pyramid is evident in nearly all zoning laws. ... The primacy of homeownership remains so widespread that we hardly think of it as something requiring explanation. Yet there is no theoretical reason why other uses of land should be regarded as less important. Apartment dwellers are as much citizens as home dwellers; owning has long commanded no special municipal privilege.” (pg. 17)

He says that homeowners invest in a single large un-diversifiable asset and form political entities to protect its value.

The author says that people only started to notice zoning’s affect on housing prices around 1970. He’s not sure why it showed up then. One theory is that “industrial parks” showed up and suburbs got some jobs. This drew poor and Blacks (freed by civil rights lawsuits) to move nearby. Suburbs forced high land prices to keep them out. The author believes that exclusion is far more income-based than racial.

To validate the new exclusion, suburbs used environmental and “growth management” language. The rules advocated for larger lots and open space zoning. (Open space zoning used large lots and banned development in large areas under the term “conservation”.)

Also in the 1970s era, Regional government entities were set up. This was called “The Quiet Revolution” by Bosselman and Callies (1971). The regional entities used “double veto”: things could be vetoed at either the local or regional level.

As for solutions, the author suggests taxing the imputed rent of owner-occupied houses. This would lower home ownership. Switzerland does this and has 30% home ownership. He admits that this is an “unlikely policy”.

The other solution is home equity insurance. The author cites many sources for a few programs on page 30. He wants to insure the price of a neighborhood, not the whole city. He realizes there is moral hazard. And it would be hard to calculate. And hard to get off the ground.

My opinion is that the author does a great job of demonstrating that the invention of buses and truck contributed, and may have been the primary cause, of zoning being adopted nation-wide. It seems like that zoning was to preserve home values. But it changed later, as a means to fight growth. I don’t think the author finds the meat there. I don’t think taxing the imputed rent is politically viable, at any level of government. I like the idea of insuring home values, but insurance is a one-sided option with a lot of down-side for the seller. I don’t think you could find investors with enough capital to insure every house in America’s cities.

This paper explores the motivations of NIMBY’s, who oppose new construction in their neighborhood. He says, “It’s the variance, dummy!”. Homeowners have sunk all their money into a single risky asset and they fear its crash. When evaluating any change to a neighborhood, they look at the potential downside.

He points out that you rarely find NIMBYs among either apartment owners or apartment dwellers. Occasionally, you find business owners, but they are usually trying to prevent competition. He believe apartment owners face the same effects as homeowners, but are more diversified.

He says that new homeowners are just as NIMBY as old ones.

He says homeowners are astute judges of their house’s value. Prices include costs such as property taxes, health hazards, congestion, school quality, crime rates, and air pollution. They also price-in events in the future, like potential industrial construction nearby.

He discusses house-price insurance, but says it is hard to write the contract that a homeowner wants and an insurer will accept. It is hard to price the effects on a particular house or neighborhood. If a homeowner signs, they might not maintain the house, in order to get a larger payout. Also, there is a risk of an insurer going bankrupt, since the one-sided risk is so large.

He mentions that house-price insurance is not a panacea. If homeowners are insured against losses, they don’t do what’s best for the neighborhood! We need some NIMBYism. So, he doesn’t advocate becoming a renter-majority society.

He says, “it is worth wrestling with the problems of homeowner insurance contracts as a solution to the NIMBY problem”. He says other solutions are possible. One is Caplin et al. 1991, where homeowners enter partnerships. Another is Case-Shiller home-value insurance using a city-wide index. Any diversification helps. “It may be that the NIMBY problem will fade as a result of more efficient financial markets rather than political reforms.”.

This is a good paper. It has an interesting concept and has lots of interesting ancedotes to go with it.

The abstract defines the term: ”Fiscal zoning is the practice of using local land-use regulation to preserve and possibly enhance the local property tax base.” It’s strange that it isn’t in the text of the paper. It would have been good if the paper kept repeating that theme. It wanders, but it wanders in intersting places.

The paper says land taxes are optimal, but hard to administer. It says zoning might be a crutch to try to make property taxes more efficient and closer to a land tax. He credits the idea to Bruce Hamilton.

Author says property taxes are better than sales taxes ... because sales taxes entice high-volume retail and local residents don’t like high-volume retail. I’m not sure I buy that. I think that, if one city did pure sales tax, the tax rate would be big and high-volume retail would go elsewhere, killing the scheme.

Author says that income taxes encourage rich residents to move to another city.

If cities instituted property tax without zoning, rich residents might convert mansions to apartment complexes and move outside of the city. With zoning, the mansion must remain single family. It must be sold to another rich person. The author says this avoids a deadweight loss, because the building’s usage is changed before its time. (I’m not sure I buy that.)

One quote is “Property taxation without zoning does not work very well.” But I think a more telling quote is from an old plannind document: “multi-family homes were excluded because they attracted ‘a class of tenants who add nothing to the revenues of the town, but who, on the contrary, become the cause of increased expense in all departments.’ ” If City Managers are trying to maximize income-per-existing-resident, zoning helps.

Another telling quote is from The Pittsburgh Committee on Taxation: “ ‘If Pittsburgh is to continue to raise practically all its revenues by taxing real estate values, steps must be taken to prevent the needless destruction of those values and to stabilize and promote their increase in every way possible.’ ” The committee’s report mentioned New York. M. Nolan Gray’s book “Arbitrary Lines” mentioned that NYC instituted zoning because an egotistical developer had built a huge building that collapsed commercial rents.

Zoning usually requires a neighbor complaint to start enforcement. The zoning board usually allows variances, if none of the neighbors objects. But zoning violators (especially outsiders?) face harsh penalities. A NYC skyscraper had to remove 12 stories after a mistake!

Zoning changes usually involve “side payments”, like park land and impact fees. A factory might help a whole area, with employees in many cities, but the city hosting the factory has “disamenities”. One funding option is a local, temporary property tax increase called “tax increment financing” (“TIF”) for local infrastructure.

“Regulation of housing development can bring complaints from housing advocates, state legislatures, and the courts, but exquisitely detailed regulation of nonresidential property seldom raises such issues.”

Cities are more heterogenious than the economic models would predict.

“Another argument against fiscal zoning is the simple observation that local officials allow structures likely to house children who could attend the local schools. By most calculations, a family with children is a fiscal drain on the community.” Also related to schools, quality of schools is a big factor in house price.

Homevoter hypothesis: Homeowners consume housing and own local assets (land and buildings) and are financially dependent on the city’s success. They vote for their financial improvement. Suburbs were built for homeowners. “Large city government officials are attentive to homeowners, but they also pay attention to development and employer interests as well as those of residents.”

“Most rural townships (in the East) and counties (in the South and West) now have zoning, and so they possess the devices that enable the local government to manage the future property tax base.” What?!?!

Page 20 and 21 have a discussion of semi-rural cities where a lot of land is undeveloped. It goes through an example of Celebration, Florida. This did not make sense to me. When it is easy to build, the option-value of land is low, so prices are low.

The example does show the motives/actions of a single developer over land. And, if there are multiple developers, they still have the same motives for stable prices and predictable income. Zoning is a way to enable that.

The author makes an interesting claim about Prop 13. He says it came in response to Serrano v. Priest, which said that school funding couldn’t vary by local property prices and taxes. The ruling “made the property tax into a statewide tax.” He says, with lower property taxes, the rich moved back into central cities causing exclusivity. (And suburbs didn’t change their existing exclusivity.) Zoning still matter for school districts, but possibly because of peer effects. It’s an interesting theory.

Overall, it’s a good paper. It would have been better to clearly state its theme and relate everything to it. I missed the theme when I first read it, because it was in the abstract not the body. I’m not sure the voters have a clear idea of what’s going on; it might be better to look at City Managers and informed players.

A very good paper. Simple, informative idea. Clearly expressed.

The idea is that rent burden can be broken into pieces. The rent burden for household i is rent_i∕income_i, but it is also the multiplication of four factors: The “rent gap” is rent_i divided by the median rent for a unit of that size. The “excess size cost” is the median rent for a unit of the size divided by the median rent of households of that size. The “demographic baseline” is the median rent of households of that size divided by the median income of households of that size. And, lastly, the “income gap” is the median income of households of that size divided by that particular household’s income.

Multiplying those 4 factors together gives you the rent burden. Now, Furth looks at each factor. He separates households where the rent burden is greater than 30% and where it is less than and compares the values of the factors.

3 of the factors are roughly the same between the two groups. The biggest difference is in the “income gap”. The poor are rent burdened.

He does a more localized breakdown for New England. In Southwest Connecticut and Massachusetts Bay, the “rent gap” has a larger effect there, almost equal to the “income gap”.

A very good paper. My hypothesis is that families at all income levels spend about 57% of their consumption on housing and transportation. Thus, a higher rent burden is where transportation is cheap. That is, public transit is common. That the “rent gap” is largest in the densest parts of the state, seems to support that idea.

This page goes into details about regional inflation. Basically, how inflation differs from one metro area to another.

Regional Price Parity (RPP) is the price of the same basket of goods in different metro areas. So, it’s rent for the same house. Or the same basket of goods. Or services. The average of all RPPs in a given year equals 100. (I looked up their definition and apparently they are intra-country PPPs!)

CPI measures inflation over time. For regional CPIs, they are indexed to 100 back in 1982. Regional inflation is also different because the baskets are different. For example, heating is a larger expense in colder climates.

About half of the regional differences is due to housing. It is persistent.

In tighter labor markets, employers pay higher wages and charge more for their products. This creates inflation. So another regional difference is, when unemployment goes down, prices go up. But it looks like housing markets are the cause of tighter labor markets! When you adjust for housing costs, the difference (mostly) disappears. That’s fascinating!

This is a piece of data I’ve been looking for! It says that rent increases in an MSA explain about half of inflation in the MSA. Prices depend on rent in a significant way.

The article starts by saying home prices are not a predictor of regional inflation. And that makes sense to me. If the price for a home is the sum of discounted future rents, it is dominated by the expectation of future rents and the discount factor (usually an interest rate).

The second half of the article focuses on rent as a predictor for regional inflation. That does work. Their linear prediction has an R-squared of 0.49. The trend is clear, but I’m not sure how precise that R-squared value is. In their plot, the changes don’t look normally distributed and the higher values seem to have larger movements than smaller ones.

The coefficient they get in their regression is much lower than I expected. Households spend about 32% of the their income on rent, so I expected if rent increased by X, prices should increase by at least .32 ⋅ X. And, given local purchases, probably closer to .50⋅X. It’s true that only 1 in 3 households rent at market prices, but the usual interpretation is that homeowners are landlords that rent to themselves. In fact, their rent metric includes owner-imputed rents! The conclusion says the author’s answer: “There still exists more variation in rents than in inflation, due in part to the fact that households will adjust how much housing they purchase when the price changes.”

The authors use “regional price parities”, which are an inflation measure for many metroareas. The CPI is only available in 27 MSAs.

I skimmed this book.

This is a classic book, famous for proposing to tax land. It was very popular in its day. The land tax is often called “Georgism” in the author’s honor. It is an emotional and pop-sci book — it doesn’t have equations or math. It is long and goes into detail.

The author focuses on the farm economy, but says his ideas also apply to cities. He says that production is caused by 3 things: labor, capital (tools and machines), and land. Income from them has their own names: wages, interest, and rent. A lot of income goes to renting land, because of monopoly and speculation. The remedy is: “We must make land common property.”. He says that taxes on wages, sales, and interest harm the economy because less labor is done and less capital put to use. Thus, he suggests to tax land, to its full value, and to abolish all other taxes.

For a simple idea, I expected a small treatise. This is a 400 page book. The author is a clear and impassioned writer. I could recite the book’s introduction at a housing conference today and I’m sure it would raise a cheer. It seems to be an entertaining book, but full of one man’s speculations and there have been more than 100 years of economics written since then. I’ll spend my own time on more recent works.

This book is a broad summary of Urban Economics research, packaged in a well-written mass-market book. It is about 300 pages and I would guess has 600 items in its bibliography. But it doesn’t read like an academic paper. It’s a flowing conversations about cities. It just happens that every other paragraph includes a number or a chosen morsel from a paper.

I normally try to summarize the important concepts in a book or paper, but this book is full of them. The other thing this book conveys is its mood. Cities are where things happen and are full of possibilities. They are beneficial to society and improving. We don’t know everything about cities, but what we do know should inform local, state, and federal policies.

I cannot recommend this book enough. If you want to get started in Urban Economics, this should be your first read. You’ll not only be informed about the field, you’ll be inspired about its potential.

A very readable book that argues that zoning is unnecessary. It describes what zoning does (restricts use, restricts size) and what it does not do (mandate construction, protect health and safety, protect the environment, plan for the future). Zoning doesn’t even “solve” the problem of externalities like noise, smell, shade, etc..

In NYC, zoning was instituted to restrict building size. The Equitable Building was so large that the price of office space plummeted. Zoning limited building volume so that didn’t happen again.

Berkeley implemented an early zoning plan for other reasons. “Residential areas” kept apartments of poor (including “negroes and Orientals”) out of the suburbs. Banning businesses kept Chinese-laundry (and the Chinese) out of the suburbs. Berkeley’s zoning was non-hierarchical — it kept residential out of industrial areas. This was used by business interests to force housing out of downtown areas and attract more businesses (with jobs and tax revenue).

Zoning drives up prices. Single-room occupancies (SROs) are banned. Parking required. Heights limited. Empty land mandated. The approval process delays housing. Etc. It makes cities worse and America poorer.

Zoning was enabled by and enabled segregation. By race and class.

Fixing the problems requires loosening zoning, but the author considers abolishing it all together. Zoning doesn’t accomplish its goals. It doesn’t address most externalities and, those it does, it does roughly. It isn’t the best mechanism for planning growth.

Houston is the example of a city without zoning. Most of its residents’ desire for zoning is fulfilled by deed restrictions, which are enforced by the city government. Some of zoning’s aspects show up elsewhere (don’t build in a floodplain, optional parking minimums, some business banned everywhere in the city). Houston is still segregated by use, but that is driven by the market. I wish the author had gone into more detail on this important aspect of the solution: “As Siegan observes, it’s not uncommon for homeowners’ associations to simply pay off owners of abutting properties to avoid more offensive uses, such as gas stations or car dealerships”. (Page 155)

The author says that if zoning is abolished, planning needs to change. Laws need to focus on externalities directly, like noise and smell. Planner’s duties can be replaced by mediators.

The author says that planning needs to fight the segregation that it helped create. Not just remove zoning, but use the federal Low-Income Housing Tax Credits (LIHTC) and Section 8 vouchers to put poorer families in high-opportunity areas. The author believe inclusionary zoning is bad. He supports community land trusts (CLT). (I find CLTs very very expensive for their benefit.)

This report looks at zoning regulation and the distribution of lot sizes in fast-growing suburban Texas cities: Round Rock (near Austin), Pfluggerville (near Austin), Frisco (near Dallas), and Pearland (near Houston).

The fascinating part of this paper is the graphs of the distribution of lot sizes. You can see spikes near the minimum lot sizes.

The Round Rock graph has wide distribution. It makes me think that the distribution is actually a log-normal one. It has a small spike and a number of lots smaller than the minimum lot size of 6,500 sqft.

The Pfluggerville distribution has huge spike at the minimum lot size of 9,000 sqft.

I love this paper, because it has data that isn’t available elsewhere. I really would have loved if they had plotted the lot sizes on a logarithmic axis and tried to fit a log-normal distribution to the tail above the minimum lot size. If that is the natural distribution for demand, we can use it to measure the short fall (not enough supply) and the distortion (people who buy a larger lot than they want).

The Director of Austin’s Planning and Development Review responded to City Council request to look at “micro units”, which have less than 500 square feet of space. It defines them and the restrictions in Austin’s zoning and building code.

He says micro units are already legal to build, as small as 250 sqft., but other parts of the code interfere with building them. “Site area requirements specify the minimum amount of land required per dwelling unit. In Austin’s zoning code, these requirements differ depending on the zoning district. ... Under Austin’s current code, the smallest site area requirement is 800 square feet, for efficiency units located in MF-5 zoning.”

Also, there is a parking requirement. He says “Austin’s multifamily use requires at least one space per dwelling, unless it meets one of the above conditions for reduced parking. That means that aside from CBD and DMU zoning districts, a multifamily development will have close to one parking spot per dwelling (at the 20% central core reduction) or more.”

Have not read yet. Looks like a redo of Glaeser & Gyourko 2003, but with much better data. It doesn’t study Austin. It also doesn’t study Houston, which would be interesting.

I skimmed this paper.

Looks at data from Ohio and North Carolina to see if homeownership affects voting in local elections. They do differences-in-differences.

The effect varies from 3% to 8%, depending on the regression. Richer are more likely to vote. When zoning is on the ballot, it is higher.

This is an executive summary of inclusive zoning. It describes it and its effects. It has a summary of 6 different studies of inclusionary zoning. “... four find that inclusionary zoning increases prices.” The author points out that all these studies assume the cities’ restrictive zoning policy is “a given”. In the conclusion she states: “No studies of its effects indicate that it increases housing supply or contributes to broadly lower prices. It benefits a small portion of low and moderate-income households rather than targeting aid at the households that need it most.”

I skimmed this paper.

This paper tries to measure the effect of Houston’s 2013 zoning change that reduced the minimum lot size outside the I-610 loop (ring road). The core of the paper is a regression on the logarithm of appraised land prices from 2005 to 2021, except 2013. The treatment area is within 2, 1, or 0.5 miles of the I-610 loop. The treatment is being in the affected time and area: after 2013 and outside the I-610 loop. Fixed affects include the year, ”demographics at the Census Tract level” and ZIP code ”linear time trend”.

The land appraisals are geographically clustered and not independant. The author attempts to correct for this, but I didn’t understand what they did.

The results say appraised land prices fell outside the I-610 loop, relative to inside. But most of the results are not significant. A few regressions, where the distance was set at 0.5 miles and using many fixed effects, reach a p ¡ 0.1 significance.

I only skimmed this paper, so I can’t draw too many conclusions. I didn’t expect a large effect, because outside I-610 is far from downtown, where land values are highest. (I wish we had data for the 1998 minimum-lot size change that did affect downtown!) Minimum-lot sizes have their biggest effect on the bottom of the market, so I don’t think this style of regression is the best tool to identify their affect. A cut-off model might be better. Or a regression of log of price-per-acre that includes the size of the lot as an input. That said, it seems to be a respectable study of the effects in the only large city to change its minimum lot size, and that’s an asset to the research.

(A side note, I mostly skimmed the paper because it was a drudge to read. It felt like a ”wall of text” in places. I’m not sure if that’s a factor of the Mercatus Center’s format or the author’s long sentences and paragraphs, or a combination of the two.)

This research evaluates walkability’s effect on home prices on a zip code basis. I don’t trust this paper’s results at all.

The paper’s literature review says that Walk Score sometimes didn’t line up with real walkability on small scales. Hence, it would be odd to see its effect at larger scales, like zip code.

The Walk Score for a zip code was taken at its centroid. The other variables were average rooms, average housing age, average commute time, average distance to CBD, income, vacancy, and density. They controlled for county-level fixed effects. (The hope this adjusts for crime and public amenities.) They regress log(price-per-sqft) with linear effects. It isn’t clear if that is a sqft of land or interior space. They use linear average income, not log(average income), which is strange. In the results, they say “The coefficients on distance from the CBD, average commute... are insignificant”. That’s a bad sign.

They attempt to validate the model with rental rates, which I would think would be the primary way to measure the effects.

The authors find price increases with Walk Score. But zip codes are not randomly drawn — they were drawn for delivery of postal mail. Nor is the centroid of them random. The Walk Score varies on a small scale and they should be measuring it on that scale. I don’t think they take enough confounding variables into account. Even with the ones they do, they find a negative coefficient for both commute distance and distance to CBD. I don’t trust these results at all.

A touching and enlightening article. The author explains how she dealt with poverty by occasionally living in her car.

It covers a lot of the practical issues of being homeless. Low pay and expensive housing cause homelessness. Landlords have requirements (deposit, proof of income, etc.) and discriminate against homeless (let alone incarcerated, etc.). Roommates are problematic: propositions for sex, mental health, lack of privacy, etc.. When you live in your car, you need to find a place to park at night.

The most important section is the first: “Homelessness is expensive”. It opened my eyes to how costs increase as you go down the ladder. Without a stove, you must buy pre-cooked food and cannot buy in bulk. Without storage, you must buy a new winter coat each year. Without a shower, you need to buy a gym membership to get clean. With all those higher costs, trying to “climb up the ladder” is harder the lower you are.

I skimmed this paper. This paper looks at land prices in Detroit. They’re still high downtown but fall off sharply. They pick up a bit in the suburbs. The image in Figure 5 is sad.

This is the kindest, wholehearted, literate attack I’ve ever read. It is dense with ideas and feelings and examples.

The first sentence is “This book is an attack on current city planning and rebuilding.” The victims of the attack are architects and planners who think that the best cities have tall buildings in park-like environments. (These go by brand names “City Beautiful”, “Garden City” and “Radiant City”.) She also attacks governments who subsidize suburbs and who essentially “clear the slums and rebuild”.

Her first chapter is on Safety. She defines a city as a place with a lot of strangers and the residents of a city block (people living there and business owners) need to defend themselves from strangers. There are 2 ways to do this. One way is to demark “Turfs” which are patrolled by gangs (in poor areas) or private security guards (in rich ones). The alternative is an organic defense: the dangerous person is detected, the alarm communicated to fellow residents, and enough residents mass in the street to fend off the danger. This organic defense only comes through visibility of danger, stability of residents, communication, building of trust through voluntary interactions, and fast response.

The organic defense is encouraged by fostering interaction of residents on sidewalks using mixed-use, short blocks, high traffic at all hours, recreation in public rather than private, etc. You also need big windows looking over street, no blind spots, etc.. Parks are bad — plenty of blind spots, far from windows, usually low-traffic. It’s where gangs fight, where boys sneak off to do bad things, and where the homeless gather. A few parks, which are small and have traffic at all hours with various crowds at different times, are successes. Activity-based amenities, like swimming pools (and libraries?), can be good. In addition to parks, other borders between streets, like highways, railroads and rivers, also cut the organic fabric.

Chapter 6’s title says “neighborhoods” but means “governing”. The author doesn’t see a boundary for a neighborhood — the city is a continuous piece of overlapping relationships: friends, customers of business, schools, etc.. In her mind, there is your block and your city and nothing in between. She says there is a political role to be played by districts of size about 100,000. Basically, no citywide politician cares about your block. You need a district size to be where someone on your block can gather enough voices from nearby blocks to get the attention of an assemblyman at the city level and that the assemblyman has enough power to get stuff done with the police, the traffic planners, etc.. She comments that many cities are not very democratic because of their structure and hoops that residents must try to run through.

Quote from pg. 181, “Here is a seeming paradox: To maintain in a neighborhood sufficient people who stay put, a city must have the very fluidity and mobility of use ...”

I have not read Chapters 7 through 15 yet.

Chapter 15 is about money. There is real damage done to neighborhoods that could not get loans because banks blacklisted them after planners designated them as “slums” or bad. Existing buildings are not improved; bad buildings are not torn down and replaced. Her early example of Boston’s North End is brought up. It was only sustained and improved by people paying cash and by construction people living there. When money does flood into a slum, it often displaces residents and destroys the social fabric.

Tax law also drove people to be slumlords. Owners did not invest so that profits came as capital gains rather than income.

The author mentions that eminent domain hurts because it pays only for the land/building and not everything the owner had invested betting on a stable future. Some cities have “quick-take” laws that grab the land immediately and settle the price later, so that profiteers cannot buy property that will be later eminently domained.

Chapter 16 is about subsidized housing. The author has sharp comments about lack of housing is about lack of money. She recommends subsidizing rent and guaranteed loans to building owners who provide it. She would let owners choose their residents.

I did not read Chapters 17 through the end.

I didn’t have time to read all of this book but I want to! It is an engaging and eye-opening. The author has new ideas with lots of examples. The author never uses the words “organic fabric”, but that’s what she means. It’s a difficult concept for planning — improving may require a local “gardener” rather than a distance mathematical bureaucrat. Even the first question that I’d want to ask, “how do you measure it?”, is a quandary. Page 174-179 talks about the social network in a neighborhood and mentions the number of connections between any two people. It’s a kind of “six-degrees” but within the neighborhood. Perhaps that’s a way to measure it. But the next question, “what do planners do to make a neighborhood better connected?” gets even more complex. Even if there’s complexity, the author has redirected everyone’s gaze at it.

This is a big-thought book about the economic development of humanity, like “Guns, Germs, and Steel”, by Jane Jacobs. She wrote “The Death and Life of Great American Cities” which is a classic book that upset the world of urban planning. This book is fascinating, dense with ideas backed by anecdotes from cities. It tries to show that cities are vital to economics and a particular process in cities drives economic growth. My major problem with it is that it doesn’t use math. While this book is fertile for ideas, its main argument should have use the precision and rigor of mathematics to prove its point.

Her big idea is that economic growth is driven by “import substitution” in cities. In her model of a city, it has exports, imports, export-producing industry, and (self-)consumption-producing industry. The exports pay for the city’s imports. Growth happens rapidly when a city starts producing goods that it used to import. E.g., Tokyo used to import bicycles, but now makes them locally. This new industry increases the local economy but also has another effect. The city’s exports pay for the imports and now the city can buy even more of other imports. These can help the city produce more goods that it used to import, causing growth and the cycle to happen again. It is that cyclic feedback that Jacobs believes drives rapid growth in cities (and economies).

The book is about more than that idea. She believes economics is about cities. (She points out that industry is not evenly spread across the land, but concentrated in cities.) That cities pre-date agriculture. (She speculates about a city in Turkey trading obsidian before agriculture.) That the conventional story if “agriculture blooms and then cities develop” is wrong. She says cities produce tools that help farmers be more productive.

She has an organic view of cities. She says cities that focus on mass production in a single industry do not grow and develop. She thinks the best cities have an mix of small companies which grow and innovate. The book contains many stories about new products being developed, such as 3M changing from being a mining company to making a wide varies adhesives. She mentioned how Japan learned to make bicycles slowly: first repairing bikes, then making their own parts to repair bikes, and then, eventually, making the whole bike. She tells the story of a factory being set up in a small town and it failing, but it succeeded quickly when relocated into a city. She tells many stories of employees “breaking away” to start a new business. She has a great metaphor about organic development on pg. 129: biologists once thought a fertilized egg was a tiny baby, only needing to grow larger. We now know the egg is a single cell and the fetus develops by cells splitting and specializing.

She has a great set of examples about how geography is not destiny. There were other settlements near New York and London with better harbors. Those settlements should have had the advantage in growth. But they didn’t. Something else drove growth and development.

I will say that two of her strongest stories did not move me. They were about the growth in Japan and Los Angeles after WWII. Both were far away from where they imported goods from (USA, East Coast). So, the standard economic motivations to move production locally make sense — transportation costs were high (at that time). It would be interesting to compare where expensive transportation drives local growth and where it does not. In the 1600 to 1900’s, colonies were far away but only some grew. Jacob’s view is that colonies were either focused on single-product mass production or the colonizers prevented locals from making imports locally.

She has disdain for Adam Smith’s view that specialization leads to more production. Smith’s classic example is the 10 specialists doing different jobs will make 4,800 pins per day per person, and 1 generalist struggles to make a handful. Jacobs says “Division of labor is a device for achieving operating efficiency, nothing more.” (pg. 82). She considers the idea of creating the pin in the first place to be the important part. Innovation eventually lead to a pin-making machine that put all the specialists out of business! She does value the division of labor because a 10-step process has 10 (or even 45) places where innovation can occur, if allowed.

For a book from 1969, it feels very predictive of the last 50 years. She mentions recycling. Investing in startups. That services will differentiate and grow faster than manufacturing.

This is an amazing book because it has lots of stories and will get you thinking in new directions. I don’t accept its story of “import substitution”, because it is too reductionist. It may explain a city’s growth but not the economy’s. Doesn’t import substitution mean that manufacturing has moved from the exporting city to the importing one, not new manufacturing has been created? It might grow the economy if local resources are used, but then it is those resources that are causing growth, not the movement of the manufacturing.

I do like her view of cities as organic. If we view a city as a network of processes, where each takes input goods and produces output goods, then the network can only change by little steps. A process can grow. A process can mutate to produce a slightly different product. A process can move from outside the city to inside it. A process can split or a new process can replicate an existing process. Innovation can create a brand new process, but it still has to connect it into the network. It helps where there are lots of places to connect it to. That may be why cities succeed: not just lots of products for customers but lots of places in a network where innovation can find a place.

This is a weird book that contains an interesting idea. The author, Jane Jacobs, is a seasoned observer of cities and wrote the book in Urban Planning, “The Death and Life of Great American Cities”. In this book, she tries to explain the agglomerations effects of cities by exploring the metaphor of ecology. That is, that businesses succeed in a vibrant city in similar ways to species succeeding in a verdant environment. It is an interesting idea. But she doesn’t explain the idea directly. The book written as a sequence of conversations by a handful of characters. That’s weird.

Explaining science using dialog has a long history. The Socratic dialogs date to around 400 B.C.. Galileo’s “A Dialog Concerning the Two Chief World Systems”, which convinced people that the Earth goes around the Sun, was published in 1632. It can be an interesting informal format, where questions and answers go back an forth between an expert and a skeptic. Eventually, the skeptic is won over and so is the reader. But, sadly, this book is not in that format. It’s a handful of not-very-well-defined characters talking. It’s just weird.

But the idea deserves our consideration. Silicon Valley has seen amazing productivity and growth since 1957. Startups have thrived there and changed the world. But there’s nothing special about the land there. An Internet-based startup could be started anywhere. But that place seems to be different. Jane Jacobs thinks it is the ecology of businesses there.

Jane Jacobs believes that growth comes through innovation. Focusing on efficiency can help growth, but a single-company city or a single-industry city is like a mono-crop. A mono-crop doesn’t change and A single predator or disease can wipe it out. A single-company or single-industry city can collapse. Rochester, NY was tied to Kodak’s rise and fall. Detroit to America’s car industry.

Jane Jacobs sees innovations happening when a process mutates into a new process, just as a species mutates into a new species. An entrepreneur might first learn to bake bread as an employee at a bread shop. Then, spends his savings to rent his own bakery. The process of the new baker at the new bakery will be slightly different: new products, new ways to deliver products, etc. If the new bakery is successful, it will grow and hire its own employees, who might go on to start their own bakeries.

But starting a business requires a lot more things than know-how and starting capital. A new business needs to find suppliers and customers. It needs banks to give mortgages or landlords who rent buildings. It needs lawyers to draw up founding documents and contracts. This means that new processes don’t arrive in a city, sprung from the forehead of an entrepreneur. They usually come by small changes to the existing processes in the city. It is far easier for that new baker to get started if his previous employer is an ally, introducing him to suppliers, advertisers, lawyers, etc.. The new bakery would look separate, but it has a lot in common with the old bakery.

Some of the small changes are: Replacing a supplier. Adding a new customer. A slightly different product, tunes to specific customers. Growing the process. Making the process more efficient. Outsourcing part of the process. Opening a new location or moving to a more advantageous location. Turning waste into a product. (E.g., kitchen oil became “bio diesel”, waste rags from making clothes became stuffing for teddy bears, manure became fuel or fertilizer.) A new delivery mechanism.

A common story of innovation is adding some new customers, discovering that they’re using the product in a new way, making a slightly different product for that group of customers, which draws in even more customers who want to use the product that way, etc.. After a little while, there’s a completely different product available with a completely separate group of customers. Jacobs calls this splitting a “bifurcation”.

But part of the process is brand new companies. Every new business needs startup capital. In an ecology, we can think of this as the energy stored in eggs and seeds.

The book has an interesting discussion of energy. Life on earth is powered by sunlight. A desert has sunlight. So does a tropical forest. But only the forest has lots of life. Why? It seems the forest has lots of small steps between sunlight being captured and it is used. Plants grow, become food for herbivores, which become food for carnivores, which die and become food for mushrooms. It wasn’t clear to me what “energy” was in the metaphor of business. Maybe it is “demand”?

The book talks for some time about feedback cycles. An interesting point was that subsidies can cause unintended feedback cycles. Cod was over-fished, partially because Canada subsidized fisheries. When fewer fish should have meant fewer profits and fewer fisherman, the subsidy kept them fishing.

Again, this is a weird book. It was hard to summarize the content, because it’s written as dialog. And it is hard to read — I never did understand who the characters were. But I do like the idea that businesses are processes and cities grow by lots of little changes to those processes. It might explain why innovation is so localized.

An amazing paper that tries to fulfill its title’s claim: the rate of return on everything. It measures historical values of equities, bonds, and housing.

For housing, an interesting table is Table VIII on Page 39. It breaks down the mean return and standard deviation for rent and capital gains. The same is done for equity dividends and capital gains. Housing has a better Sharpe ratio!

This paper uses AI to decide which homeless people to assign to which treatment, such as sending them to the homeless shelter or rapid rehousing.

I skimmed the paper.

It looks to use the finite resources to minimize the reentry of people into homelessness. It identifies the most important factors, to the AI, to diagnose and prevent reentry. There are ethical considerations and the authors looked at a system that, for any individual, only increased their chance of becoming homeless again by a fixed amount.

The author looks at the effect of Minneapolis’s ”triplexes everywhere” policy passed in 2018. He does this using a differences-in-differences approach, comparing sales of properties inside the city with those outside the city.

He needs to compare different houses, so he uses a hedonic regression on log price. There is an additional variable for being inside the city, being after a chosen date, and the product of those two. There are a lot of details.

His default date for the before/after is Dec. 2018, when Minneapolis city council passed the plan. Figure 2 plots data over time and shows Minneapolis’s prices improving relative to control cities, which is worrying. There is seasonality, but he goes back and forward 1 year from the date, so is sure to include all seasons equally.

For inputs, he looked at transactions from Zillow. He doesn’t say which transactions were in/excluded. He mentions ”R1” zones and ”buildings zoned for 2- or 3- families”.

He added variables for zipcode and month of sale fixed effects, to control for closeness and prices changes over time. He looked at more detailed location (census tract) but results were similar. He does not list all the variables used in the regression and the paper does not have the equations for the regression. (There is a Technical Appendix, but it is only 3 pages. It has an equation but no list of variables.)

He does a number of regression. For properties within 1km, 2km, or 3km of the border, he sees price increases inside Minneapolis after 2019 of 5.7% to 2.9%. He looks at neighborhoods with below or above median income and the results are not significant. (But he quotes that result in his abstract?!) He also does a regression based on a house’s percentile rank of square footage (of interior space, presumably) compared to other houses within 200m. That was strange.

Overall, I think this paper is not good. I think difference-in-differences is a fine approach, although I would have preferred a discontinuity test if possible. I find it a big mistake to not list all the variables used in the regression, even more so because there was a technical appendix so page space was not an issue. I think the author reasonably controlled for distance/location.

I am most worried about time. I would have preferred a regression on years clearly before and after. Prices increase when information hits the market and that could have been at the election in Nov. 2017. The trend seen in Figure 2 is that prices in Minneapolis are increasing faster than those in control cities for the 2 years before Dec. 2018. Visually, Minneapolis is increasing .1 log dollars per 12 months vs .05 for control cities. So, if that trend continued, I’d expect the regression to get .05. Which is about what it got. (The author on page 389 says ”the general trend in price growth between the two groups is similar”, but I don’t think that’s so.) So, my view is this paper concludes ”prices increased 3% to 5% from 2018 to 2019, but it isn’t clear that that is attributable to an event around Dec. 2018.

This blog post covers a topic that I intended to write about. That being poor is expensive: buying pre-cooked food, not being able to store winter clothes, fewer transportation options, etc.. The author’s writing is clunky, but they cover a lot of topics.

The gist is that working your way out of poverty is hard. The walls get steeper the farther you go down.

The authors try to predict how rising rents will affect future inflation. The covid pandemic caused rents and housing prices to rise. Rents and owner’s equivalent rent (OER) are part of inflation measures, so the authors wanted to predict their effects.

They use Zillow’s ZORI rent index, which is current asking price in the market and a leading indicator of CPI’s rent. They also use Zillow’s Home Value Index, which is a leading indicator of rents. They also take into account vacancy, unemployment, and ”rent inflation trends”. I’m not sure what that last one is.

They study 15 MSA and predict the effect will be 0.5% on the PCE inflation measure. Most of the effect in 2023 will be from rents that rose in ZORI that carry over to the PCE (and CPI) as existing leases expire.

This is a good paper. It estimates the price of land around Phoenix, Arizona, using transactions on vacant land. It compares those prices to a previous estimate of land under buildings. The author’s goal is to evaluate the use of these estimates for land-value taxes.

The authors start with the prices of transactions on vacant land. They use a technique called “kriging” to estimate the price of land that was not sold. In kriging, the estimate of price-per-acre is a weighted average of the known price-per-acre for land transactions. The weights are a function of distance to the land in the transactions. The author tries many ways to calculate weights. They find that weighting nearby land more and averaging many pieces of land work better. Many techniques for weighting performed about the same.

There are adjustments for the size of the lot, zoning, and nearby amenities, like roads, lakes, golf courses, railroads, etc.

The authors compare their results to land-values assigned by the local property assessors. They find “Both of these values track each other closely in the growth rate, but are different in the level due to differences in both observable and unobservable characteristics”. Another validation finds similar results.

The authors compare these estimates, calculated using vacant land transactions, to estimates based on a previous paper[16] by the authors (and some additional co-authors). That paper used property prices minus the cost to replace the building to estimate land values. “Between 2012 and 2018, vacant land transactions generate land values that are, on average, 14% higher than land underneath structures.” Vacant land differs substantially from land under structures. It is larger. I spent 30 minutes trying to understanding the graphs and the author’s descriptions here. I didn’t find them clear. And I fear making any summary, since the author seem to give up at the end. They say “the large level differences are not due to differences in the option value of redevelopment nor teardown costs” “unobservable differences are not everywhere in the city” and “we infer that there are large, positive, unobserved factors that are positively correlated with the size of residential lots.”. It would have been nice to see where in the city those were.

The authors discuss how these results would affect a land tax. They say, “Because land is typically the more volatile component of a property, a tax exclusively on the value of land would therefore be more volatile in terms of revenue generated.” They show big swings where the revenue doubles or halves within 6 years. And those are in the only 18 years of 1 city covered by this dataset.

I find this paper very interesting, because land values are important and the prices of vacant lots are handy. (I’ve estimated price in Austin, TX using the prices of vacant lots.) I wish the comparison between the two techniques was better written. The plots were impossible for me to comprehend quickly and the description was not enough to help. (The paper is short; I wonder if page limits hurt the comprehension.)

I like the technique of kriging. It’s simple and fits well in the economics toolbox. It is an averaging and it doesn’t capture trends, such as prices increasing as you get closer to downtown, so there’s room for better techniques.

I’m not surprised that vacant land is more valuable that under structures. Vacant land is a very flexible option, because it can be subdivided and there’s a large variety of buildings that can be built on it (with time delays). The authors’ other technique for valuing land was based on the price of the whole property minus the cost of reconstructing the same building. That approach (minus the rent for the time to reconstruct the building) is a lower-bound for the price of land, because it is the profits from a single choice of a very flexible option. And exercising that option (having residents move and destroying the building) had high friction costs. Given the “land is an option” approach, its easy to see that swings in future demand cause big swings in land prices, because the option can go from “in the money” to “out of the money”. (Urban Economics needs a good framework for talking about future demand.)

It’s a good paper. Worth reading if you care about land prices and most Urban Economists do.

Skimmed.

Finds that 3 years after change in property taxes, rents increase to account for the tax change. It is a rather dramatic step up in the 3rd year. And error bounds are wide. It’s also strange that the averages in the past are also positive. (Perhaps tax increases are passed in the worse year?)

The 3 year lag is strange. Where it is significant, I assume it is a quantity effect. It takes a while to take unprofitable housing off the market or to build new profitable housing.

Weird. Assessed values for property never change. So reassessment cannot be the cause of increases 3 years later.

The paper contains dramatic welfare graphs, but these come from a simulation. The model shows that a tax increase of 1% causes more than a 1% drop in consumption utility for households in the bottom 15% or so. Households with higher consumption were less affected.

A pretty short review of hedonic regression for housing. The paper covers some history of the technique. Not much theoretical explanation of its foundations at all. The author recommends Follain and Jimenez (1985) and Sheppard (1999). He mentions that hedonic regression can be done on price and rent. He doesn’t talk about which studies did which (or that rent is better because it doesn’t involve interest rates). He lists good variables to include: rooms, rooms by type (bedrooms, bathrooms, floor area, single-family vs apartment, number of floors in apartments, age, type of heating/cooling, structural features (basements, fireplaces, garages), structural materials, quality of finish, neighborhood characteristics (including schools), and distance to CBD (and schools, groceries, other employment centers). Other important factors: tenure in building, are utilities included, race/ethnicity, and date of data collection. Models can be linear, semi-log (log price and linear characteristics), or log (log price and log characteristics). Another is ”trans-log”, which covers interactions of pairs of characteristics. If you assume houses follow a log-price hedonic model, repeated-sales models compute the exact price increase over time. I’d need some long-term studies to see if that’s correct in practice. It sounds like Case-Shiller’s work on repeated-sales indices is worth reading. The author talks about ”two-stage models”. This is not two-stage least squares, but two different regressions. One seems to use log-price and the other linear-price. I’m not sure why this is done. Maybe to calculate a specific dollar price for features, rather than the percent-increase of log-price models. The author talks about instrumental variables, but doesn’t name them. It is an important issue, because if you’re using the sale price, sale of a house doesn’t occur randomly. The author mentions hedonic regressions have happened for all sorts of housing, but rarely commercial property. This is a less-than-great summary paper. It didn’t explain basics and had uneven coverage. It’s got some references, but no summarized data.

The origin of Marchetti’s constant. https://en.wikipedia.org/wiki/Marchetti%27s˙constant

Says that Zahavi reached an empirical conclusion from data around the world: man travels a mean of about 1 hour per day. They believe its origination is biological, governing the territory controlled by primitive man.

In ancient Greece, villages have a mean territory of 22 square km. The radius is about 2.5 km and, at a walking speed of 5 km/hr, travel is about 1 hour round trip. The author claims that cities got larger due to changes in the speed of transportation technology, with similar-timed commutes.

Car travel averages about .9 hours per day, at most income levels. Very low income people travel about .7 hours per day. The miles driven per year in the USA is stable at 9,400. (NOTE: This seems too stable and the value now is 13,400, so it is not stable.) Travel consumption is stable around 13%. Death rates per year is also freakishly stable around 22,000 per year. (NOTE: This doesn’t match data on Wikipedia. The number there has much more variability, but is stable around 35,000 per year.)

The paper goes on to look at communication vs. travel, with technology. The number of messages or calls has increased exponentially, as have trips. The number of messages/calls follows a gravitational law, proportional to the populations of two cities divided by the distance apart.

City population in 1920 followed a Zipf distribution, linear in log(population) and log(rank by size). The author claims that, if you use the 30 minute commute and include planes and high-speed trains, it still holds because of areas like the Boston-to-DC corridor. The final graph shows mobility in France. Technology changes mean speed increases about 3% per year.

I skimmed this paper.

This paper examines the effects of the minimum lot size. Not just on the price of land, but the effect on households, who not only buy less land but may buy a different sized house too.

It studies Houston’s change to the minimum lot size in 1998/1999. It has a model where households can move between Houston with the new minimum lot size and a counterfactual Houston without the change.

Renting households, especially poor ones with few family members, gain upto $28,000 in their lifetime. The average effect for renters is $18,000. The author looks at homeowners, but I’m not sure how that is calculated. It seems poorer homeowners of small houses lose upto $10,000. I think the author says they lose on the house, but are able to gain by moving to a new ”right sized” house. I’m guessing owners are forced to stay in Houston for their whole lives?

I don’t think I’ve read enough of this paper to get a good feel of it. Salim Furth reviewed this paper here: https://marketurbanism.com/2022/10/20/is-affordability-just-you-get-what-you-pay-for/

Authors compare the PIT Count to the US Census, American Community Survey (ACS), and Homeless Management Information System (HMIS). They also compare HMIS shelter-use data from LA and Houston to the Census, to evaluate the usefulness of this daily data source to study this population.

I only skimmed this paper.

Each data source uses different definitions and they are applied inconsistently. Double counting happened for 10 to 15% of people in shelters in the 2010 Census in LA and Houston. The population is very mobile.

The authors conclude that the data is consistent with each other, assuming difference in definitions. They assume the data sources are accurate.

Read abstract and skimmed introduction.

This paper studies places for stores in NYC. Landlords sometimes keep the places vacant for over a year.

The abstract says “We find that tenant heterogeneity and move-in costs jointly explain long-run vacancy by generating dispersion in match surplus and therefore option value for landlords. In a counterfactual exercise, eliminating either feature results in vacancy rates of close to zero. Search frictions and aggregate uncertainty play much smaller roles.”

They use their model to evaluate a vacancy tax. “Proponents of the tax argue that landlords who hold storefronts vacant impose unnecessary costs on their neighborhoods by reducing local economic activity through underutilization of retail space, as well as posing a threat to neighborhood safety via a reduction in “eyes on the street” (Jacobs, 1961). They view the tax as a Pigouvian measure which would cause landlords to internalize the impacts of their vacancies on urban vibrancy.”

“We find that the proposed commercial vacancy tax would indeed encourage landlords to fill vacant spaces more quickly, reducing vacancy rates and retail rents. However, the tax would also distort the set of stores present, with lower-earnings stores arriving at opportune moments crowding out higher-earnings stores that might have arrived later. These lower-earnings stores are more likely to exit, increasing retail churn and reducing welfare.”

Slides on a proposed amendment to Austin’s zoning to allow “units” of 200 sqft. The minimum at the time was 400 sqft.

The amendment would still require 400 sqft per unit on the site. And, require 0.6 parking spots per unit. And, require 10% affordable units! AND, it would only be allowed on a “CTC”, which I think is a “core transit corridor”, but not districts that already have affordability requirements, live VMU! WTF!!!

Applies to zoning districts MF-1 through MF-5, and MU combining districts. Requires 10% be affordable (at 60% MFI for rental units or 80% MFI for owner-occupied). Requires 0.6 parking spaces per unit. They are high restricted by Austin’s “compatibility” rules.

Did not address land requirements in zoning rules. MF-5, the highest density, allows 54 units per acre, or 1 unit for every 807 sqft of land.

I found a chart on this page very valuable. Regional Price Parities are like Purchasing Price Parties, but for regions of the US. They compare the cost of the same goods in different states or cities. They’re good for measuring regional price differences across space. You need an inflation measure, like the PCE, to measure price differences across time. Chart 1 on this page compares how the RPP index (”all items”) changes based on a change in one of its components: rent, goods, or non-rent services. We see that, even though rent is a small part of the index (I think 16% or 22%), a change in rent causes a larger change in the index, like 34%. This indicates that the prices of goods and other services are also affected by rent.

Nordlinger bought a house in California and was paying 400% more property tax than their neighbor, due to Prop 13. They challenged the law under the Equal Protection Clause of the US Constitution.

The Supreme Court used a very loose “scrutiny” and found that there was a rational basis for the law. The California state legislature was rational to incentivize people to stay in the neighborhood. Another rational basis was that, under Prop 13, homeowners could predict how much taxes they would owe in the future. Using this level of scrutiny, the Court didn’t evaluate how well this law achieve those goals, nor its other effects. (The Court assumed the legislature took those into consideration and the plaintiff could ask the legislature (rather than the Court) to change the law.)

The Court also held that there was a rational basis to keep the lower tax if the residents moved within California and if children inherited the property. As Nordlinger already lived in California as a renter before buying the home, they were not allowed to argue that these terms were unfair to out-of-state buyers. Another plaintiff would need to bring that case.

The decision was 8-1. Stevens dissented. He thought the rational basis for the law was nonexistent. The law broke taxpayers into classes, but it wasn’t by any rational scheme but by the purchase price of the property, irrelevant of time. He also thought the “rational basis” had to serve all taxpayers and the rational that this law served newer homeowners (and people wanting to buy homes) was close to absurd. He also thought that, by allowing inheriting children to keep the lower tax, it was a tax-break based on heritage.

The author builds economic models to explain homelessness.

The basic model has 3 levels of housing: homeless, low quality, and high quality. Agents have an income and a standard utility model, increasing in money and home quality. The model demonstrates that poor people, in the right situtations, will choose homelessness.

The author extends this to a model with a continuum of quality ranges, from 0 to positive infinity. And then adds construction of homes, maintenance, filtering, etc.. Models are similar to filtering models of Sweeney 1974 and Arnott 1987.

SKIMMED.

On page 49, the author looks at “comparative dynamics”. That is, how changes in income distribution, interest rates, and other factors affect the housing market. I didn’t read all the details.

Other aspects of housing are considered: rent collection costs, public housing, shelters, building codes, and tastes.

I think the model is more complicated than necessary to explain the homelessness dynamics. There isn’t any data connecting it to the real world. Still, the author seems pretty happy with it.

The book starts pretty boring. Chapter 2 is a discussion of the various definitions of “homeless”. For example, are people in a shelter “homeless”? Chapter 3 is an undramatic discussion of why homelessness is bad for other people. Strangely, for an economist, it doesn’t try to put it in financial terms. Chapter 4 is a history of homelessness and data for it. Which is pretty awful, up to 1996.

Chapter 5 is interesting: it compares the homeless, who sleep outdoors or in shelters, to what most people call “homeless”, the people that beg, collect cans, and wash windshields. He refers to those as “colloquially homeless” or “daytime streetpeople” or simply “streetpeople”. There is overlap. The author’s study in New York said about 3/4 of streetpeople are homeless, but 1/4 live with someone or have their own place. Most have slept in a shelter but do not like them. The streetpeople earn a variable income that is below minimum wage. Those that reported hours worked around 7 hours a day. 2/3 work 7 days a week; the rest are evenly split between 1 and 6 days a week. 3/4 say they would take a low-wage job. About 1/2 get public assistance, which might go away with a job. If given extra money, about 1/3 said they would spend it on alcohol or drugs. 2/3 said they didn’t have a way to save money!

Chapter 6 describes an economic model of housing and homelessness, without any equations. But it uses phrases like “price-quality schedule”, so good luck understanding the chapter without an advanced econ degree. I have one and, for me, it might have been easier with some equations. The model has a high-end market, where houses are built and maintained, a low-end market, where houses are not maintained and deteriorate in quality, and, finally, homelessness. The interesting part is the low-end market, where changes in costs ripple down determine who is homeless. It is a model of a market at equilibrium and the author walks through various scenarios that will change the equilibrium. They’re summarized in Table 6-1 at the end of the chapter. It is a well-constructed model but I’m not sure how well it represents reality. Buildings take decades to age, so are cities actually near equilibrium? Also, the

Chapter 7 looks at the effect of income inequality. The author uses an unusual metric: the slope of a line fitted to the lower-half of an income histogram. Yeah. Very unusual. He leaves out a lot of details. He claims that it explains homelessness very well, “allowing for lags” (pg. 132). But what is an appropriate amount of “lag”? Increases in the price of low-rent housing predicts homelessness. On page 142 he starts a history of very low-rent housing. Every place has a different name and regulation for them: flop houses, SROs, cubicle hotels, etc. Many closed due to profitability in the early 80’s. Some had high vacancies. It’s not clear why that explains anything. He uses fire reports and demolition records but I don’t find the data and/or story compelling.

Chapter 8 looks at financial influences on housing. Interest rates and inflation are too small to affect operating costs in his model. The opportunity cost of land (which he calls “gentrification”). This didn’t explain homelessness in the 80’s, but it might now in Austin and San Francisco.

Chapter 9 examines other studies that compare multiple cities against each other. It includes Tucker 1987, Tucker 1987b, Quigley 1990, Elliot and Krivo 1991, Burt 1992, and Honig and Filer 1993. He doesn’t like their data! There are different definition of metro-area: Census Bureau and Rand-McNally. Pg. 167 has some insights: “wealthier cites should provide more shelter, since charity is a normal good, and so should cities with more valuable real estate, since removing homeless people from such real estate yields a higher return”.

Chapter 10 covers government. “New York has not prevented the construction of shanties, but has confined their construction to the most primitive technology.” And, now, Austin has confined people to tents. He mentioned that health regulations may include window size, ventilation, and minimum floor area to limit the spread of TB. He says that regulations may be needed to keep buildings maintained: the owner going bankrupt will not profit from maintenance (the bank will) and may not maintain the building. He notes a distinction between “grandfathered” regulation, where the existing uncompliant are legalized, and “retrofit” regulations, which requires owners to invest to meet the standard. Owners of grandfathered buildings have “something valuable — an exemption from regulation”. NYC banned SROs, but the author doesn’t think it affected homelessness. One argument is high vacancy rates. One cost is parking spaces, even if the law only requires 1 for every 6 rooms. Low-cost landlords suffer from capricious enforcement of rules. Rent control means owners do not invest in maintenance and that owners discriminate on residents (rather than price), which may mean the unhoused are “lemons” that the landlords don’t want. Making evictions harder hurts low-cost landlords, because more tenants are disruptive and fewer can pay (or have seizable assets) to make up for the eviction costs. “A former SRO owner in Newark, for instances, gave as his main reason for leaving the business the hassle and liability that had become part of it; he thought that the only persons who would buy a cheap hotel now was the ’thickest-skinned SOB in the world’.”. O’Flaherty thinks tenants rights contributed high vacancy and, a small amount, to homelessness. He believes “public housing for the middle class increases homelessness; ... for the very poor, reduces it.”. 1/3 of homeless women in a shelter said they wouldn’t live in Chicago public housing due to the lack of safety.

He says Section 8 has a certificate, which pays the difference between actual rent and 30% of income, and a voucher, which pays the difference between “fair market rent” and 30% of income. Those with voucher still need their residence checked by the housing authority! Why?! Food stamp funds increase with rent, so it’s a rental assistance program! WTF?!

Rental vacancy was studied by Rosen and Smith 1983 and Arnott 1989. Vacancy is complicated. Pairing a landlord with a tenant is a two-sided process. Landlords decide the rent and also how much to advertise and any renter requirements. Tenants decide their price limits, where to search, how much to search, and how much effort to meet landlords’ requirements. Vacancy around 7% or 9% is “natural”. O’Flaherty doesn’t think vacancy explain homelessness.

Chapter 11 is about welfare (”income maintenance”) in wealthy countries. New York’s General Assistance encouraged living on the street: Unsheltered receive $305; those in a shelter get $149. Bureaucracy in the USA makes it hard to sign up and hard to stay on the list. “one chance in eight of opening a case and maintaining it for a year.” System is complex and time-consuming. Making and keeping appointments is difficult for homeless. Many homeless seek independence and avoid hassle. Mental health and disability determinations are hard and bureaucratic. (They’re medical decisions in Canada.) Programs incentivize black-market work, like begging and cleaning windshields, because official job income reduces their income from SSI or other programs. The “tax” is 100% for New York’s general assistance.

Under sharing housing, the author says social security (SSI) and foot stamps have “strong disincentives for sharing housing”. SSI reduces by 1/3. Foodstamps give more for 2 people living apart than living together. This programs were growing rapidly in the 1970s, before homelessness took off in the 1980s!

Chapter 12 is about mental health. The chapter looks at the release of people from mental hospitals in the 1960s and says that was unlikely to be a cause of homelessness later. Some women make themselves dirty and act crazy to scare off attackers. For low-cost housing where people shared rooms, a few people with mental health problems has a large effect. Also, repair costs and liability go up. Many have “innkeeper liability”, which means the law blames them for bad guests causing problems to other guests! There may be a tipping point, where enough mentally ill poison all low-cost housing, but the author finds little evidence for it.

Chapter 13 is about drugs. Taking drugs is a choice and the author sweetly says, “Volition means economics.” He says crack became popular because it is cheap. (Powdered cocaine was sold by gram; crack by 1/20th of a gram.) Different drugs are substitutes and crack displaced powdered cocaine and amphetamines. Maybe alcohol too. But crack is not a substitute for housing! Cheaper highs mean that more money is left for housing. Housing demand should have increased!

But that analysis skips a lot, such as the lost productive hours of someone who gets high and the instability of highs and lows (which causes people to lose jobs). And maybe other changes? Drugs and alcohol are common among shelter residents. Half took cocaine in one study. Cocaine use did spike in the 1980s. Cirrhosis deaths spiked in the 1970s ... why? The timing of homelessness doesn’t really line up with emergency room visits. The author concludes that “at least as good an argument can be made that crack reduced homelessness as that it increased homelessness.”.

Chapter 14 is about cops and prisons. Many homeless have been in jail (31%) or prison (18%). Homelessness doesn’t cause people to go to prison, but releases of prisoners increases homelessness. They have a hard time getting work, may have a harder time finding a place to rent, and have fewer social connections to help. Prisoner releases didn’t cause the problems in the 1980’s, but maybe contribute today? Common laws used against homeless are: loitering (unconstitutional), vagrancy (unconstitutional), obstructing sidewalks, begging, public drinking or drunkenness, and commitment to mental institutions. Police are usually sympathetic to the homeless: they have to deal with them regularly, there’s little benefit in conflict, and the police like support from other institutions who might react negatively if they abuse the homeless. Police just displace the homeless, not fix things. Homeless might prefer jail, with warmth and food. We don’t have much data on Black homelessness during the 1950’s and 1960’s, but it increased in 1970’s. Why? The author mentions less racism, but doesn’t mention migration, which happened then.

Chapter 15 is the author’s diagnosis and prescription. He thinks homelessness in NYC in the 1980’s was caused by the “middle class shrinking” (income distribution changes) in the 1970’s. His model predicts that less low-cost housing was provided. Also contributing were zoning regulations, strict housing-maintenance laws, difficulty evicting tenants, more ex-prisoners, and more generous shelters. Low priced crack may have helped end homelessness. Newark’s story is similar. Chicago didn’t see a shift in the income distribution and didn’t see homelessness. He waves his hands with Toronto. He dodges with London and Hamburg. In fact, he goes so far as to say “perhaps maintenance is cheap relative to construction so that the models of Chapter 6 don’t apply.”!!! This is my biggest problem with his model: how does he know it works at all?

His policy recommendations are: Do not discriminate against or in favor of the homeless. E.g., different welfare payments, access to services, etc.. The police should focus on crime, not anti-social behavior. This includes prosecuting crimes against the homeless and crimes by the homeless. He says police should do “aggressive but non-punitive referral” but I’m not sure what that means. Right after, he talks about the Port Authority bus terminal and Grand Central terminal being clear of homeless. He says that the homeless should not be allowed to build shanties, because they’re unsafe and violate the law.

Cities should not run shelters; they should hand out an allowance and let private companies run shelters. Shelters will then compete to be the preferred place and improve services to the homeless. (Cities may run shelters for natural disasters, etc.) Private shelters should be allowed to offer better rooms for cash on top of the allowance, creating a working market for low-cost housing. (NOTE: In Austin, there is probably a huge discontinuity between a free bed in a shelter and $700 for an apartment!) The author prefers entitlements, because they provide predictability for everyone. For disruptive homeless people, which private shelters will not house for the allowance, he recommends selectively increasing the allowance. “This scheme won’t work perfectly ... but I can think of no scheme that would work better.” (NOTE: Let charities deal with them?) Later, on page 289, he says “A commitment to zero homelessness is unworkable”.

He describes public housing rules as giving “authorities a choice between running empty buildings and running buildings full of fairly well-off people”. A study says that for every $2 spent on public housing, residents would pay only $1 for it. His non-discrimination policy says that homeless should not move to the front of the queue for public housing. He also says that there shouldn’t be a waiting list for public housing — it should be a random draw of all applicants! (Waiting longer doesn’t mean “more deserving”.)

As for regulations, he doesn’t like “apex zoning” (”hierarchical zoning”, I believe). It should be easy to evict for any reason (not just “good cause”). He points out that eviction takes time and money and, for poor residents, putting a lien on their possessions won’t pay for it. (NOTE: Does this mean that it would be a good policy to pay landlords for evictions, as a form of insurance?) He says we should remove “innkeeper liability” for lodginghouses and SROs; they should have the same liability as apartments. (NOTE: I’m not sure I agree with this; I think allowing eviction for any reason would help fix this.)

He says working homeless are good, because busy people cause fewer problems than welfare clients. He suggests finding productive work, paid in cash, in small amounts quickly. He suggest paying for each small piece of work. He suggests getting rid of the disincentives in assistance plans. He suggests hiring should be informal and without any eligibility criteria. His example of a good program is “Street News” newspaper in Chicago and recycling cans in NYC. He suggests discouraging begging by allowing charities, like Salvation Army, to solicit on the streets!

He suggests getting rid of the disincentives for doubling up in SSI and foodstamps. He suggests reducing effective marginal tax rates. For drug abusers, he says we should make it easier for them to get money like SSI. He suggests disability should be more than yes-no proposition, but have degrees. Lastly, US programs need to be as friendly and accessible as Canadian ones.

This paper covers research into homelessness by economists and practitioners.

It has an interesting discussion of the strengths of studies that focus on individual vs. averages. Practitioners want individual studies, since it informs how they act. But they don’t provide information on solving homelessness.

It has some interesting insights on the “general equilibrium”, by which the author means how everyone in the housing market reacts to a policy. A shelter is more pleasant than sleeping rough, so more shelters means that fewer will sleep rough and, also, that some people will stay longer. So, Housing First treats the homeless, but there aren’t studies on what it does to the system.

One simple model has a rate e at which the homed become homeless and a rate x at which the homeless become homed. The steady-state homeless rate is e∕x.

Discusses treatments, where money seems to be best. Mentions that rent is factor. Some non-housing interventions help, but they’re relatively expensive.

Journeys Home is a longitudinal study that covers Australians beyond homelessness. It produced a lot of good data. For men, trauma was seen on entering and exiting homelessness. Women experienced trauma while homeless. Violence might cause male homelessness, but not female. Addiction, in general, was not a cause of homelessness. Although heavy alcohol use (a binge) might be causal.

An increase in EITC payments lead to fewer families doubling up. But it didn’t decrease homelessness. The author asks “Why might this be?” and guesses either it has to do with averages or the near-homeless didn’t apply for EITC.

Evictions often precede homelessness. Giving legal representation during evictions help prevent them. But no one has studied if landlords change in response.

He casts a lot of doubt on HUD’s Annual Homeless Assessment Report. This includes the Point-in-Time Count (”homeless census”, “PIT Count”).

His own paper (2003) looked at shelter quality and the quality of living outside (rain, worse spots to camp) and computed an equilibrium for how many homeless would choose each. It says how the number of beds affects numbers too — meaning 100 more beds reduces the PIT Count by much fewer than 100! Permanent supportive housing is very high quality housing and has a predictable effect.

When prisoners are released, shelter numbers do not increase. But that may be because recently released criminals scare people away?

If housing is given to those homeless longest, it might incentivize people to stay homeless longer to increase their odds of “winning a lottery”.

Corinth 2017 has formula for predicting homelessness.

NYC ran a service “Homebase” which prioritized treated people in their neighborhoods. I’m not sure all the details, but it kept social connections.

Prevention is tricky. Rolston et al. found 14.5% of their control group entered shelters and 8.0% of the treatment group. Since only 6.5% were affected, the program could be expensive for the effect.

People have private information on how likely they are to be homeless. Still, practitioners may find it hard to get honest answers when the treatment depends on the answer. The Homebase program saw evidence with long walks to services. The NYC housing program does it by requiring long stays at a shelter before getting housing support.

In NYC, homelessness increased from 2011 to 2017, but the housing market was mostly unchanged. Vacancy in low-rent apartments fell by 1/3, but it was not significant. Migration doesn’t explain shifts in numbers, even in the NY/NJ area. LA saw a similar rise in the city but not surrounding areas.

Under Mayor De Blasio, NYC raised $-per-family-day from $103 to $171. (NYC sometimes rents hotel rooms and that’s close to the price of one.)

On page 62, does a “thought experiment” of how many Housing Choice Vouchers would be necessary to replace NYC’s shelters and other programs. He believes the money currently spent ($1.8 billion) might be able to do it. Pg. 66 says LA’s survey is different from NYC in major ways. E.g., visual not interviews. No transit. Non-random. Crappy in every way.

LA spends far less on homeless services: 1/15th that of NYC! That rose to about 1/3 in 2017. “prevention in Los Angeles might also require $30,000 to reduce the PIT Count by 1”. Wow. He compares that to rent subsidies, which in NYC, cost $40,000 to $80,000 to reduce the PIT Count by 1.

Economists have not studied the unsheltered homeless. Is the difference between cities a matter of enforcement?

Veterans saw a big decrease in homelessness from 2010 to 2016. But the age range for high homelessness (18 to 64) also fell sharply. There was a large voucher program (HUD-VASH) but it may have had little effect.

Homeless families fell by 2/3s between 2010 and 2017. It could be because NY state takes kids from homeless families and other states copied that. There are other possibilities.

Black people are more likely to be homeless, but states with more Black residents tend to have fewer homeless. Drug rates, alcohol abuse, and mental illness are the same rate in Blacks as in other groups. The proportion of Black incarcerated is similar to those homeless. The author says another candidate explanation is segregation: a shelter full of Black people may be off-putting to White people. (But wouldn’t that push up the number of unsheltered White homeless?)

The conclusion says: “We don’t know how to end homelessness.” “We can’t explain why homelessness in the two biggest American cities exploded in this decade, or why it fell in the rest of the country.” “On an aggregate level, weather, labor market conditions, and housing prices affect homelessness, and maybe policies too, but modest changes in unemployment or rents cannot be expected to trigger huge changes in homelessness.” “...; for the most effective policies a handful or two of people have to be treated to reduce aggregate homelessness by one.” “It’s a good time for more economists to start working on homelessness.”

These are slides from a class at Sciences Po in Paris, France. It works through the equations of the AMM model, one-by-one in a very clear manner. I like this a lot.

Skimmed.

The author builds a simple stochastic-calculus model of an individual’s wealth and, when their wealth falls to zero, the individual goes homeless. By “simple”, I mean it is still stochastic calculus and heavy on the math.

The author’s model leads to an obvious policy: a “Pigouvian homelessness prevention program”. That is, assuming homelessness has negative externalities (like smoking causes second-hand lung cancer), the government pays a subsidy if you are housed (like subsidizing people not to smoke). Usually the Pigouvian response is a tax, like a heavy cigarette tax, but taxing the homeless would be cruel. He says “he optimal shelter quality (and street enforcement regime) maximizes the instantaneous difference between benefits and costs”.

Another policy is income smoothing. He concludes that “any actuarily fair insurance that reduces volatility is welfare-improving.”.

I quickly skimmed this paper.

It tries to find the contributing factors to homelessness.

It first looks at the theory that the mentally ill were no longer in institutions. It finds the prison population increased proportionally. The remaining 34,000 were too few to account for the 100,000+ increase in homeless over the period.

They interpret O’Flaherty (1996) as saying income inequality can lead to a class of older housing where rent is too low to pay for the maintenance. Thus, residents must choose between substandard housing or homelessness. (Note to self: the oldest housing in Austin is near downtown. Growth has caused the least-valuable buildings to be on very valuable land. Has this wrecked filtering?)

Did not finish.

Brendan O’Flaherty mentioned this as one paper connecting homelessness to inequality.

Building near Plaza Saltillo claims to have “micro-units”. Average unit size is 355 square feet. However, the building has 34,364 sqft and, for 60 units, that means 572 sqft per unit.

It has a lot of amenities: coworking spaces, gym, communal kitchen, rooftop patio, etc.

It has 19 parking spaces.

The land is 0.26 acres. Density of 240 units per acre.

It was built by Watershed Development Group. It was architected by Mark Odom Studio.

Did not read.

Abstract claims full pass through, especially near downtown.

Page 5 says “When focusing specifically on buildings outside a 5km radius from the Boston central business district (CBD), the pass-through decreases to a range of $0.52 to $0.67. For buildings within a 5km of the CBD, rent increases by $1.06 to $1.39 indicating this market may be imperfectly competitive with capital owners exercising market power over tenants.”

This paper covers 2 techniques that might be useful in derivatives markets.

One section covers the math for combining hedonic regression with repeated-sales data.

The other section covers “perpetual futures”. This is a financial instrument that regularly pays the different between two indexes. One is the index to measure; the other is a reference index, like the CPI or risk-free short-term interest rate. The instrument lasts forever, making its price equal to the discounted future flows from the index.

I skimmed this paper.

I’ve been looking for an empirical paper like this, which looks at the effect of the minimum lot size. The author expected (as I did) that land prices would increase closer to downtown and fall in the suburbs. He also thought that rents/prices for housing downtown would drop relative to the suburbs.

The data is from Houston. Used tax appraisal values. Only single-family homes (not rental market!). Only the 2013 change in minimum lot size was studied, because there wasn’t data from before the 1998 change.

Model is simple. Fixed effects for year and parcel. Treatment effect is linear in price, not log-price, which is strange. In my skimming, I didn’t see a discussion of new buildings on the property, nor hedonic regression on the buildings. There isn’t a factor for distance from the center of the city — that’s important because the treatment area is basically the area around downtown.

The authors found that land prices increased dramatically with the smaller minimum lot size, as compared to outside it. The buildings dropped slightly in value inside the area, as compared to outside. So, after all, prices increased inside relative to outside. The effect on land prices was still widening 5 years after the law changed.

I don’t think this is a good paper, because it doesn’t include the distance or drive-time from downtown as factor in the regression. Anything that affected prices based on distance — like traffic congestion — would have a similar effect. The study would have been better if it regressed on the effect at the boundary.

I read this paper quickly. It’s a simple concept and a well-written paper. An easy read.

This paper looks at the hard problem of making sure there are lots of many sizes available for construction in a city. It is easy to divide a large lot into many pieces. But, once done, it is hard to aggregate lots of small lots into a large lot again. How does a city keep a set of large lots available for development? The author’s answer is to give the financial incentive of more density for larger lots.

The concept is pretty simple. Write the zoning laws so that larger lots can build taller and denser. This incentivizes owners of small lots to collaborate to create larger lots.

The author has an example from Simi Valley, California. The incentives allowed 18 single-family lots to be voluntarily grouped together and redeveloped as a master-planned community. One owner refused to sell; 3 others only sold part of their land. The sellers had a single realtor. According to the realtor, “the developer paid the land price expected for development at seven units per acre. The original owners thus captured most of the capital gains from the higher density”.

The author has some interesting comments. It is voluntary, so avoids the problems associated with eminent domain. After passing graduated density, neighborhoods remain the same; neighborhoods only change when lots are assembled. Owners may sell their land for fear-of-missing-out.

I do worry about the balance. The goal is to have many lots of many sizes. This law may tend towards the largest possible. E.g., imagine 4 lots on a block and two agree to merge. If those 2 are in the middle, will the others almost have to join now? If those 2 are on the end, will the other 2 have an incentive to join to make 2 pair or join the already merging lots to make it 3 or 4 lots together?

I don’t have time now, but would like to think more about the market solutions. If there’s a shortage, won’t larger lots command a higher price? That should provide incentives to sell. If we need many land owners to sell together, is there a way to form a legal entity that they participate in? E.g., create an adhoc corporation with certain bidding rules and a way to overrule holdouts? I feel like I need to be convinced that this is a real problem and that this is the best solution.

This is a short, easy read. It is a good subject and a solution worth talking about and considering.

Did not read yet. Recommended by Caitlin Gorback.

Skimmed incompletely.

The author studies the minimum lot size on a nationwide dataset and predicts the effects of changes to the law.

The author uses CoreLogic’s national dataset. It includes data from Multiple Listing Service for many cities. They don’t have zoning or minimum lot sizes for areas, so they synthesize them. The zoning districts are the US Census block group. (These are 600 to 3,000 people, so 240 to 1200 houses.) The minimum lot size was detected in the data by finding break points in the distribution of lot sizes.

Given a distribution of lot sizes, a break points cut the distribution into two pieces. The chosen break point minimized errors when each piece was fitted to a 7th order polynomial function. I find this an odd choice. I’m not sure the grandfathered lots (below the break point) should follow a particular distribution. And, for the ones above the minimum lot size, my own work indicates that these follow a Pareto (Power Law) Distribution.

They validated the results against known data. I wasn’t very happy with the description of how the validation was evaluated.

White people tend to live in neighborhoods with larger minimum lot sizes. Especially when the minimum lot size is 1 or 2 acres. Black, Asian, and Hispanic are more common in neighborhoods where the min lot size is 8,000 sqft or smaller. No surprise, larger minimum lot sizes are associated with higher incomes.

To measure effects, they do boundary analysis. They include houses within 0.5km of the border. The regression doesn’t seem to include distance from the border, which is odd. They find that larger minimum lot sizes are associated with higher prices, higher rents, wealthier residents, and whiter residents. Even when city and school districts are accounted for. There is a confounding effect that wealthier residents buy larger homes and people will pay to live near wealthier neighbors and white people will pay to have white neighbors. The raw coefficient said that the increase was 13.6% for prices and 5.7% for rents when lot size doubled. That is reduced to about 2% and 1% when the confounding factors are accounted for. I find it amazing that the effect gets magnified that much.

I feel like the minimum lot size shouldn’t bind in many areas. There didn’t seem to be a check for that. On second thought, if the constraint doesn’t bind, there isn’t a break point in the lot-size distribution and it isn’t detected. So maybe it’s not an issue?

The paper goes on to create a demand model for housing and a supply model for housing. They estimate the effects of changing the minimum lot size. I didn’t read this part.