Network Expansion Limits

The special magic of the network effect: Adding node C to path AB not only makes journeys from B to C, but also A to C. So there is an inherent non-linear scaling to adding links. However, while each link adds more because of the network effect, but each link is less-good than the last, because you always build the best links first. So at some point, any network faces declining marginal returns on expansion. 

Further, in any infrastructure network, expansion is perilous, because infrastructure has non-linear maintenance, and those maintenance costs are end-loaded--cheap initially, but with a ruinously expensive (and unavoidable) balloon payment for reconstruction at the end of their useful life*.

*Rail miles, like road miles, are more discrete than things like bridges, so the necessary 'balloon payment' is much smaller, and it's possible to make a lot of low-quality 'patches' that neither work well nor are cost efficient. (If the patches were either, you'd have built the how system out of 'patches' from the beginning).

Railroads aren't quite State-Owned-Enterprises.

If you weren't previously aware, the US railroad network consists of a series of regional duopolies: (PDF here). Two of the railroads shown on the 2021 map merged in 2023, and two more on the edge of merging. 

Which begs the question: given the benefits of network effects (one combined network is more efficient than two networks of the same size) that  in transportation infrastructure, why do we have more than one railroad company? Why don't we just have the Interstate Railway System the way we have the Interstate Highway System? 

Partially, it's a legacy issue. Prior to World War 2, the Federal government simply didn't have the state capacity to achieve anything of that magnitude. (The U.S. had to develop it to facilitate war production, and then we used it to do things like win the space race). So railroads were developed by massive public corporations. Before highways, railroads were the only way to get anything anywhere overland, so a railroad could basically charge whatever they felt like. And as a local monopoly, they did. (Like charging $2000 for the equivalent of a plane ticket). So there is a lot of railroad hate in U.S. history.  

The only competitor to a railroad is another railroad, and since railroads were highly profitable, it attracted competitors. But railroading is a brutal business, because it requires huge amounts of capital to build. And until it's done, built and operating, it returns very little money. So there was a lot of BitCoin level speculative building, and a whole lot of partially build railroads that went bust. That other railroads then bought up, plugged into their network, and brought to completion. Competition being bad for business, railroads owners of course conspired to restrict competition and split things up 'fairly' in a way beneficial to both (but detrimental to purchasers of railroad services). 

Anyway, all the railways started out as profitable monopolies, then stabilized as profitable semi-monopolies. And the whole concept of splitting the operation (of vehicles) from the ownership and maintenance of the right of way didn't exist--the operations make money, the right-of-way doesn't. And it took the advent of the gas tax (1954?) to join those two together. 

There are of course, a limited number of exceptions, most notably turnpikes and tollways. But the US had a system of turnpikes (still present in name all over Maryland) which collapsed upon the advent of the railways. (The railways themselves collapse upon the advent of the interstate). 

So nowadays, there is no reason for us not to have an Interstate Rail Network. It would be remarkably simple per mile charge for any railroads using the track, and adding capacity where things are congested. But first we'd have to buy the railroad right of way from the railroad corporations. (Or just buy the corporations outright, and spin off the operating companies while keeping the track). 

But would it work? Amtrak owns very little of the railways over which it travels--basically just the NEC (North East Corridor). And Amtrak is an absolute punching bag. Part of the reason the Interstate Highway System worked is a whole series of political dodges. The gas tax is federal, but the Federal government returns all the gas tax money collected to the states (albeit with some conditions attached). And the way the Feds return that gas is by providing an 80-90% match on building highways. 

Quirks in American constitutional history: The Federal government is not allowed to build roads. They can fund them, plan them, etc.... but they can't own/operate them. Best guess is that it was a deliberate effort to protect the (private) turnpike trusts from (public) competition. 

Other issue is that railroads aren't exactly profitable these days. Competition from the publicly subsidized interstate highway system (combined with over-regulation), drove most of them into bankruptcy in the 1970's. So in some cases, the railroads can't even pay to maintain their own right of way. But when a railroad closes, so do railroad dependent industries. And so we have a variety of programs to subsidize them. Railroads bonds are tax free, railroads are eligible for special zero-interest loans from the federal government, etc. 

So why don't we have an Interstate Railway Network? It's because it's entirely a legacy issue--existing legacy providers protected by existing legislation, as once the Constitution protected the turnpike trusts.

We may yet see. Pennsylvania made a massive investment in the "Main Line" of public works, building railroads and canals. So while it can't be done at a Federal level, there is no reason a state couldn't do it. But because of the scale (railroads work best at abut 500 mile distances), the network effects, and the sheer cost, it would take a big state to do it: NY, CA, TX, FL, OH. 

Cities exist for the agglomeration economies--to enjoy jointly what we cannot enjoy independently. In a modern context, that means highly specialized services. (Between Amazon and Walmart, most specialty goods can be enjoyed largely independent of location). For most Americans, their intersection with specialty services is the medical field (if you are old) or college (if you are young). So if you are looking for something to center a town (or small city) around, it's the same thing existing towns have already centered themselves around--a hospital or a college.

Every other place is grasping to find a place for itself in the 'space of flows', where it adds value to the things being produced. Anyplace that can't manage that is just left with transportation and warehousing, being a link in the larger network. Traditional extractive industries (mining, fishing, forestry, farming) are subsistence at best, vulnerable to global fluctuations to commodity prices (in inputs and outputs), and threatened by every greater mechanization and industry concentration.

Cities revived by leveraging human capital--people adding flows by connecting to people. And that's true of finance or technology. There are a lot of places hoping that 'information' is a new form of capital. (It is, but not the local economic developers want it to be.) A data center isn't a factory--the inputs and outputs of the data center aren't consumed or produced locally. Local power and water are consumed, but the outputs are immediately transported elsewhere for consumption. So there is no chance for a local value add. A data center will never catalyze a local supplier to scale up, nor induce a new local consumer of its outputs.

NYC Transit Fares

At some point, you have to ask if benefit from fares is really exceeding the cost of fares. Time has value. Fifty people on a bus waiting fifteen seconds while someone fumbles for a quarter is a grotesque waste of time. Simple Benefit cost analysis says that even if every person on that bus is only making $10/hour, that's over two bucks wasted, to obtain a quarter. It's a a calculation every transit agency should be making, but for NYC it's especially pertinent.

You may ask "But it is worth forgoing the revenue from the 50 fifty people?". Say it's $2.50/persons, so that's $125 in revenue, and we can assume not everyone is taking 15 seconds. Say everyone is taking five seconds to pay when boarding (on average--most people take 2 seconds, Spiders George takes 30), so every rider is imposing about 66 cents in lost time on every other rider. Then add in the cost of the fareboxes. And the cost of the man who comes to empty the fareboxes. And the armored truck that comes to collect the coins. And the person who sorts the coins and takes them to the bank. And the machine that smooths out all the bills.

Then add the opportunity costs of slowing down your bus. A bus is a $600,000 vehicle, and a bus driver runs about $100k a year (with benefits). Quick google says it costs $215/hour to operate a bus in NYC. Now, as a though experiment, lets say I can double the speed of my bus if I stop collecting fares. Suddenly, I need half as many buses and driver to provide the same service, and my cost per hour drops to $110/hour.

And you start to wonder "How much money are we actually making by collecting this money, once we subtract out our costs?". And then consider the benefits to riders--people where they are going, twice as far or twice as fast. A 20 minute bus trip becomes a 10 minute bus trip, for 50 people. That's 83 dollars of benefits, right there.

Actual math for non-fare collecting might only be ten percent faster, but the principle applies: 66 cents less in time cost per rider ($33), $21 dollars less in operations costs, a couple dollars less in fixed costs, $8.30 in benefits provided to riders, and the seemingly impressive $125 you've collected in fares looks a lot less impressive.

Now consider the imposed congestion costs. Everyone not in a bus is in a car, and every car is adding to congestion, imposing delay on all the other drivers, all of whom are very likely making much more than $10/hour.

A final note to the status quo folks, who claim "but we already have the fare collection equipment": overhead walks on two legs. Everything bought needs maintained, needs cleaned, needs repaired, and (eventually) needs replaced. And those replacement costs all need financed, all of it with debt. And all of that needs done by people who need paid, and who need paid enough to afford NYC rents.

Of course, this is a deeply NYC-centric argument--most places can't manage to put 50 people on a bus. But it's still a good thought experiment on the trade-offs that come from collecting fares, and especially for onboard fare collection. It was painful to ride Pittsburgh's light rail--in an era BRT has definitively proved the benefits of off-board fare collection, a light rail was still stopping to collect fares. 

Transportation Mad Libs

Hat tip to Russel King's post on LinkedIn

Everyone praises The U.S.'s freeway network. It is actually a financial, environmental and transport disaster.

One of The U.S.'s top urban planners has reviewed his country's freeway network.

The findings are sobering.

What he discovered should worry every transport planner on Earth.

The U.S. has built many fast highways.

But they have built far too many.

In the wrong places.

For routes that don't need them.

The claims are shocking:

• 40,000 km of routes didn't make sense
• Some lanes carry almost no passengers
• Interchanges are built far from city centres

Government officials love cutting ribbons on big transport projects. 🚄

But they forgot to ask simple questions first.

Do people actually want to travel here?

Is there a better way?

The U.S.'s regular roads are profitable.

They carry both people and goods.

And there are not enough of them.

The U.S. has spent a fortune on freeways but has neglected its intra-city transport.

That means more cars, more congestion and more pollution.

This matters for every country looking at freeways.

Freeways is brilliant when it connects large cities.

But The U.S. keeps building freeways to smaller places.

They want the fastest speeds even when conventional roads would be better.

The U.S. shows us what happens when you chase the shiny object. ⚠️

Instead of proper options analysis.

My blog post this Thursday will dive deeper into freeways.

And what other countries can learn from The U.S.'s and others' expensive mistakes.

Does your country have a sensible freeways strategy, or is it chasing the shiny object?

Accessibility and Nodes

Urban theory pretty clear that you have the highest land prices in the most central place. (Actual distances largely irrelevant to to what is most central, as access depends on transportation). So 'crossroads' inevitably have the highest local access anywhere you go. Empirical research pretty clear that trip frequency is inversely proportional--if something is half as far, you make 4x as many trips; if something is twice as far, you make 1/4 as many trips. How 'far' that trip is is a function of the 'cost' of at trip (in terms of cost + time). If something lowers transportation costs less (in time or money), you don't make more trips, but you tend to make longer trips. Lower transportation costs means you can do things like skip over the local market to go to the supermarket (which has better selection) or go to Walmart and get both groceries and household supplies. Lower transport costs also mean you are willing to have your house be further away from things. Further away means lower rent or--for the same money, more space (property or square footage). 

Urban land prices tend to decline as you get further from a central point. For any distance (d), the he coefficient (c) in the 1/d^c equation is set by transportation costs. The bigger that value gets, the easier it is to travel. Now, the equation suggests a smooth (if non-linear) decline from the central place. Which any realtor can tell you is crank--it's being next to a freeway that matters. Which brings up an important point--access points. A network defined by node access points (freeway exits, stations) doesn't spread access equally. It declines as you get further away, just as from the main central point. So in any urban environment, land values have a central 'peak' that declines, interrupted by smaller peaks around sub-centers, located near the point access points of the network. 

In urban land use, different uses are willing to pay more for access. For roads, it can generally be thought of as office, retail, housing, industrial, farms. Office uses will snaffle the locations near the most accessible freeway exits, retail will take the next tier down. Industrial needs big lots, so they'll typically snaffle the next tier before residential (subdivisions) have a chance. 

Mode choices & transit pricing

I make a regular trip to a yoga studio. It's about a mile and change from my house. I can drive, bike or take the train--all will get me there in 17-23 minutes. I drive more often than I'd like, because the parking AT THE METRO STATION PNR is free (unenforced), while the Metro ride will cost me $5.  This is despite the fact that I know (intellectually) that driving is costing me 62 cents a mile--but I also know that most of that is a marginal cost--my car will depreciate regardless, and I'll still need to pay insurance (regardless of how many miles I drive). Most people don't even consider the marginal cost of driving, especially beyond gasoline. (Everyone I've ever talked to 'budgets' by curtailing trips by pay-period.) 

We want more people to choose transit. If you want people to do more of an activity, you can make it cheaper. But you can also make it seem cheaper by hiding the cost from them: 1) reduce the marginal cost of a journey to zero, 2) put a lump-sum payment on auto-renewal (with unlimited use), 3) make someone else pay for it.

Transit agencies should also get rid of 'monthly' passes, and start selling '31 day' passes, so someone who can't buy the whole loaf on the first of each month isn't screwed, and so there is never a reason anyone should delay buying a bundle. Which also enables households to budget to all the main expenses (rent+ & transportation) aren't falling due on the same day an in the same paycheck. 

And for every transit agency with space capacity (and that's every agency, some time of the day), there should be dynamic pricing. (Once you're running a vehicle-trip, there's almost no marginal cost to letting more people on the bus). An economist will tell you to charge a premium for the peak, but an advertiser will tell you to offer a discount for the off-peak--when prices change unexpectedly, people should get a pleasant surprise.  

Polycentricity

The term 'polycentricity' is widely abused to suggest a uniform repeating tessellation, which ignores the hierarchy of centers. People accept the hierarchy of C-store, strip mall, mall, but get uncomfortable with the fact that the hierarchy also implies the emergence of a regional uber-mall. Likewise, downtowns get uncomfortable that the fact that any sufficiently large downtown will see the emergence of regional office sub-centers. Place hierarchy is an emergent property of the trade-offs between transportation costs and rents.

Has remote work permanently damaged cities?

Remote work has loosened the tie between home and work. But the COVID premium people were able to enjoy by shifting central-city salaries to peripheral locations won't last. Prime age working folks shifted to low density because they needed more space for home offices. The 'gray wave' of retirees is also location independent. Remote work has been hard on offices, but it takes a while for companies to recognize the costs of peripheral locations and negotiate RTO terms. I'd also like to point out that even people who work remote still live in Manhattan, thanks to an unbeatable quality of urban amenities.

Birth Rate

Let's play a facts game: Half the birth rate decline is the elimination of teen pregnancy. The other half is the decline in women under 30 having kids. Generally agreed, the first is a good thing, the second a bad thing. Increasing age of first natality is tied to housing. Adding a kid requires adding a bedroom. If you can't afford the bedroom, you don't add the kid. So until houses with 2+ bedrooms are affordable to single-income households ages 20-30, the birth rate will continue to decline. As evidence, I adduce the fact that Provo, Utah (crazy natality) is also stacked with two bedroom apartments. It's the densest part of Utah, because people are totally clear that married studies have kids, and that requires appropriate housing.  Other interesting sub-populations with high birth rates (military families, the church-going) enjoy subsidies--either directly for housing, and indirectly, through care-taking.  

Population Decline

As any left-behind community or shrinking city can tell you, population decline is terrible. People leave, population levels fall below the limit to support essential services (doctors offices, grocery stores), quality of life falls for those who remain, inducing more to leave. 

 But people often miss the financial half of this--when people leave, the debt doesn't. (And every municipality is loaded with debt, because city infrastructure is a massive capital investment that can't be funded through equity).  So when people leave, that debt burdens fewer people. So tax rates go up, so more people leave. And in desperation, the city start preying on those who remain for revenue--fees and fines. 

Stage three is abandonment--when you've got more houses than households, and more shops than businesses. That's when's truly go to bits, because that kind of blight attracts crime. Squatters and arson at best, and at worst, a malignant criminal underworld that exists as a society apart, sheltered among the downtrodden and wretched.   

On gentrification

 Gentrification exists anywhere where it becomes viable for one household to buy out and combine two units. It's a normal process of urban regeneration. But it's hugely controversial because we've supressed the other half of the cycle--the conversion of depreciated single family residential into multi-family.

On Vancouver

Vancouver costs aren't too high--they are just too high for missing middle. (Sen̓áḵw suggests the 'market density' is 600 units per acre). As one of the excluded, I'm actively hostile to the interests of long-term residents--it's always in the interest of the incumbent to exclude both competitors (for amenities or parking). I think the examples of both Japan and California clearly show that major cities can't be trusted with the power of zoning--the potential for abuse is simply too great, and that state/provincial pre-emption is going to be the rule going forward. On an equity basis, I am actively hostile to allowing the limited supply of buildable land in proximity to quality urbanism to be squatted by an inheritor/oligarch/elite worker class.