"We added a tiny amount of new highway capacity in comparison to the population growth and pent-up demand for it"

There is no reason we should be adding highway capacity in proportion to highway growth. Trying to build our way out of congestion is a dead paradigm [1]. Times have changed [2]. We know that GDP grows at a faster rate than population, and that GDP and VMT (vehicle miles traveled) are closely correlated. 

Keeping up
If we want to keep vehicle delay (roadway congestion) under control, we need to add lane miles at the same rate we add VMT. Seems plausible, given that GDP grows at the same rate as VMT. But only if we keep up--once congestion on a road reaches a certain level, people choose different routes, different times, and different modes. And then when we widen the road again, the people just change back, the 'triple convergence'. So once a road is congested, it will never be uncongested again.

Even if we double the size of the road?
Nope. People will just divert from other times, other routes, and other modes, to the new wider and faster route. And in the long-term, people will react to higher driving speeds by moving further away from things. When we first built the interstates, there was no congestion, and everyone moved out the central cities--until about 1970, when the 'Urban Transportation Problem' of roadway congestion happened, and people stopped moving, because there was too much congestion. And so we began to widen, and have never been able to stop.

Digging a Hole 
When you did a hole, the deeper the hole goes, the harder it gets, because each increment of depth you've already created is harder than the last--you did the easiest stuff first, when you started digging. Adding lane miles is like that. At first, we just paved the existing street. Then we eliminated on-street parking. Then we took the parkstrip. Then the sidewalk. Then the front-yards in front of houses. But at some point, to widen the road, you have to buy the whole house. And for the biggest roads, that's where we are at. TexDOT is proposing to take an entire subdivision (?!!) to widen it's central freeway. UDOT is going to take a row of houses along Foothill Boulevard, at $350k each.

Rising obligations
The second half of the problem is maintenance. Roads don't stay built--we have to maintain them. Every lane mile requires a certain amount of money to stay in good condition. The amount of money varies by the age of the road--costs are low at first, but rising, with a big lump sum every 45 years or so, when the road has to be 'rebuilt'.  The more roads we have, the more we have to maintain. [3].

So when we try to build our way out of congestion, we're taking on not only the cost of new roads, but the cost of maintaining all roads we've already built. GDP/capita is increasing as fast as VMT/capita, but not as fast as VMT/capita+maintenance per capita. [4]

The reality is actually worse, because the money from gas tax doesn't exactly track GDP. Gas tax revenue is a function of VMT and fuel efficiency. Fuel efficiency rose for decades, so the amount of gas tax collected per VMT fell over time, and we refused to raise it. (Why the Highway Trust Fund is broke, btw).  [5]

In conclusion, we can't build our way out of congestion, because the cost of keeping up with congestion grows faster than our ability to pay for it.

#----------------------------------------------------------------------------------------------------------------

[1] If you are non-metro area with slow population growth, you can do it! But you'll got broke paying for it, 30 years hence.  Check out Chuck Marohn of Strong Towns for more on that.

[2] Gas used to cost five cents a gallon. I suspect it was not the 'Freeway Revolts' that put an end to US roadbuilding, but rather the Arab Oil crisis (1973, 1978(?)). Check the spikes in 1973, and again in 1978. Oil prices jumped 20x in 10 years! Destroyed demand for driving as well as the major input into road construction.

[3] It's worse if we try to 'stretch' the road when the repairs come due. Water gets into the cracks of the roads, the freeze-thaw cycle levers the asphalt apart or washes out the dirt underneath the road.Then average annual cost of maintenance per lane mile increases.

[4] How do I know? It's a bit complicated, but I used Excel. I found some numbers on construction costs per lane mile, and cost of the annual maintenance costs of roads, on a per lane-mile basis (and co I had one column of GDP growing at a certain rate, one column of VMT growing at the same rate, one column of lane miles growing at the same rate. And a cost per lane mile, growing at the same rate as GDP and VMT. Then I added a cost for maintenance, based on the numbers I found, times the cumulative lane miles built. Then I summed the cost of new lane miles and the cost of lane mile maintenance. Then I graphed them, to see if GDP kept pace with the total cost of roads. It didn't. Over time, the cost of roads outstripped the increase in GDP. When this happened varied by what I used as my inputs, but the dynamics are inevitable: the combined growth in the cost of lane miles and the cost of lane-mile maintenance outgrew GDP.
  Mathematically, it would be something like X + .00025X > X

[5] Ah, the ever-reliable "Only because the Mass Transit Account!" argument. I'll address that elsewhere. Today, I'll simply note that:  I would love a per ticket charge on every transit ticket to fund mass transit capital spending on buses and guide-way. The Highway Trust Fund has been a reliable slush fund to promote road building for the past 100 years. I'd love transit to have the same thing.
I'd also like  mandatory mass transit parking with every new development, and exclusive 'transit-only' lanes, the way cars have. Fairs fair, right? Oh, and complete elimination from liability from manslaughter--that'd be cool. And heck, I'd take as many dedicated lane miles as proportional to the number of person-miles traveled. If mass transit is 3% of the total, 3% of the lane miles seems fir. 

Measure VMT, not LOS

With the passage of Senate Bill 743 and official adoption of the measures into the 2018 California Environmental Quality Act (CEQA) Guidelines, transportation impacts will no longer be determined using level of service (LOS), a measure of traffic congestion. Rather, projects will be required to determine transportation impacts based on vehicle miles traveled (VMT). This standard goes into effect July 1, 2020, though some jurisdictions may adopt VMT analysis sooner.

-https://dudek.com/measuring-transportation-impacts-using-vehicle-miles-traveled/

There is a saying: As goes California, so goes the nation. And in this case, I do urgently hope it to be tree. VMT is an excellent measure: of energy consumption, to pollution produced, to exposure to automobile accident. Combined with a road capacity metric, it is also a good way of measuring congestion. 

Level of Service is a garbage measure. Contrary to the school grading system it implies, LOS C indicates a road/intersection is actually at maximum efficiency: carrying all the capacity it is designed to carry. LOS A means an under-used road. And LOS F almost never happens: when LOS E hits, people divert to other routes, other times, and other modes. (My friend leaves his house at 515a to avoid traffic). The 'peak hour' is no longer an hour long. Reaching LOS A would mean doubling the width of every road out there, at a cost in billions, to ensure free-flowing traffic, everywhere, at once. And thanks to induced demand, it would last about a year. As the free-flowing traffic induced developers to build in ever more distant locations, where land is cheaper.

"It is difficult to get a man to understand something, when his salary depends upon his not understanding it!" - Upton Sinclair.

I think it's time we recognized that most traffic engineering is deeply unethical. And that there is a whole structure built up, bureaucratically, to facilitate the construction of roads, and make efficient the movement of cars, regardless of the costs. Automobile travel both kills and sickens people. Why is we want more of it?

Soapbox a bit on VMT

To soapbox a bit on VMT: VMT is not a measure of transportation demand. It's a measure of transportation consumption, like gasoline. Vehicle miles traveled means mileage on your car, time lost in congestion, and wear and tear on the road.

It's been used for a long time because it's an easy metric to interpret. You invest in places where that investment is likely to be used.  But while that make sense when you are the Bureau of Public Roads in 1956, trying to decide which highway to pave next, it's less useful today.

Because most of the VMT today isn't on rural highways. It's on urban roads. And urban roads are different than rural highways--they aren't 'corridors' but networks (pfah to 'corridor studies' everywhere). All the roads interconnect, and there are multiple pathways between two different points. And in a network, ideally, you want people to take the shortest point between two routes. Because every mile on the network is road capacity consumed, and pollution generated. But that never happens, because we have freeways, which are much much faster than surface streets. And so there is always the temptation to detour to the freeway, drive along the freeway, and then drive back to where you wanted to go. And which you pick all depends on how fast the freeway is compared to the surface streets.

There is an entire field of modeling, known as transportation demand modeling, designed to overcome the limitations of corridor studies. Millions of dollars every year go into developing, maintaining and using these models, designed to answer the question of WHICH corridor traffic will use to get between two points. (Or combination of corridors, or combination of streets, or series of network segments, etc). All in order to best predict which route is going to be most crowded in 10 years, so they can get a jump on things.

There is a fundamental shift there--from providing connectivity between places (rural highways) to make places more accessible to one another, to ensuring mobility between places (urban highways).

Here is the fundamental of urban economics: There is an equilibrium between land costs and transportation costs. Anything you do to change the cost of land or the cost of transportation shifts that equilibrium. But nothing can change the fact that there is an equilibrium. Lower transport costs, and you lower land costs. But as transport costs rise (traffic, tolls, etc), land costs rise as well.

But here is the rub in urban transportation: The cost of each increment of speed improvement isn't linear. Every increment costs more. The second lane costs more the first. The 8th lane costs more than the 4th. Because the first increment is always built using the easiest/cheapest way. So that's used up, and the second increment has to make do with the second best bit of right of way: The part that requires demolishing a whole house, digging deeper into the edge of the hill.

But I've strayed from the topic at hand: Soapboxing about VMT. VMT is a cost of travel. For corridors, it's a reasonable way to allocate funds. For urban networks, it's silly: The most VMT doesn't necessarily indicate the best (most direct) route between places--just the fastest. And so we keep dumping money into congested routes, hoping they will get better. But they never do. The fastest route is still the fastest route, and if you make it a little bit better, more people use it. The scientific name for it is 'induced demand', as people divert from other corridors, other times, and other modes of travel.

Now, let's talk about urban area VMT? Is SLC the best, if we have the most VMT? Or the most VMT per capita? Or the most VMT/GDP? NO. Exactly, the opposite, actually. VMT/capita is a measure of how much the average person HAS to drive (HAS to pay) to get done the things they want to do. And VMT/GDP is a a measure of how much transportation (as an economic input) that has to be consumed per unit of GDP. It's like hours of labor: The amount of resource that must be consumed per unit of output. Economists have a term for that: "Labor productivity".  How much labor is required to make a dollar worth of value? HINT: a high number is not a good thing--it's the sort of thing that characterizes sub-Saharan Africa, or under-industrialized countries. They lack all the things that makes labor efficient: Tools, machines, training, education. So high VMT/GDP is a bad thing. It means the regional economy (the economy of the urbanized area) is inefficient.

So when the DOT sees VMT on a road going up, they feel like they should dump some money there. But when you add up all the VMT on all the roads, that's a cost. So when the DOT dumps money where there is VMT (to create more VMT), it's not actually a good thing. (Nor does it actually make your commuter any shorter--remember that the best/fastest route is still the best/fastest route, and if you make it faster, more people will take it, until it gets worse, and is no longer the best route).











 simple metric: What's a good place to put transportation dollars? Where people

"Shared vehicle, doesn't park, doesn't occupy curb space"

"Shared vehicle, doesn't park, doesn't occupy curb space"

This was the claim I heard made about shared vehicles. First off, the vehicle is rarely 'shared'. If the driver is in the vehicle, that's not a 'shared' ride. You are being chauffeured. The driver is not a passenger. (If I drive my kid to school, it's not a 'shared ride'--I'm driving him to a place, being a 'taxi-mama'.) It's only a shared ride when there are 2+ passengers in the car. Which is rare. How rare? In my experience, perhaps 1% of all the rides I've ever taken, did I share a ride with someone. (And that was in a major metropolitan city).

"Shared vehicle, doesn't park, doesn't occupy curb space"
A TNC is just a dial-a-ride taxi with a better fare structure. Taxis occupy curb space. They aren't in motion all the time. (Or perhaps even most of the time). They _do_ require parking space. That said, taxis are probably in motion more than TNCs, because they spend much of their time 'cruising' for fares, rather than parked and waiting for fares. So Taxi's probably take up more road space  than a TNC would (a moving car taking up about 3x as much space as a parked car, due to safety-required separation between two moving cars.)

I have seen Lyft Vehicles, pulled over on the side of the road, checking their phones, as they wait for Google maps to locate their next fare in the queue. (Peak times, not so much, as trips get added to the queue faster). 

Do they take up space in the middle of the day? Ie, do they take up all-day parking space, the way that a car driven to work does? No, clearly not. But cars still need to be stored someplace. (Likely in whatever suburban wonderland TNC drivers live). Which brings up a VMT issue: If a TNC driver drives someone to work, then drives home. Then, in the afternoon, drives someone from work to the grocery store, and then someone else from the store to home. Then they drive themselves home. Point being that the driver made two trips home (one morning, one evening) to drive one person to and from work. So if we assume that everyone will take a TNC to work, VMT is going to explode. Rule of thumb (from a PhD student who drove UBER was he drove 1.69 miles for every passenger mile). So 69% more VMT, to achieve the same amount of mobility? The roadway network will never withstand that kind of increase.

 Of course, it might. If the demands on the roadway network were spread over a wider period of time. Presumably, fewer drivers want to drive during congested conditions, because it's slower. And while they get paid both per mile and per minute, I believe the incentive structure also pays them by trip. So there is a strong incentive to maximize the number of trips you make. And even if not, it's better to be earning for both miles and minutes than sitting in traffic. 

So you have fewer drivers during congested times, and people have to wait longer to take a TNC during congested times. (So the reliability declines, and more drivers are inclined to say "Screw it, I'll just drive myself"). At which point, you've basically enacted congestion pricing. Ie, when things are congested, it costs more to drive.  

Peak VMT?

The Melbourne Urbanist has a pretty great graphic:

The post discusses 'peak travel', the idea that the industrialized world has hit a peak, and gross Vehicle Miles Traveled  (VMT) will never be as high as they were in mid-2007. Gas prices are frequently mooted as the cause, which seems plausible.



But what struck me was less then decline in per capita miles traveled then the decline in Per Capita GPD. Historically, there has been a strong correlation between vehicle miles driven and economic growth, but the exact causal relationship has always been fuzzy--do richer countries drive more, or is more driving required for greater economic activity? (The latter is a favorite argument of the road lobby).

But gas prices are high, additional credit constrained, and American consumers are broke. I theorize that the price of gas has disrupted the American economic engine in unanticipated ways. It's not just businesses that are suffering from high gas prices, but employees. The Journey to Work is typically the farthest Americans travel in a week, and thus the one that uses the most gas. When gas was cheap, it made sense to live far from your job, where housing was cheap. The time-cost was high, but time is cheaper than gas.

With the housing market frozen, moving is much more difficult, so 'home' is fixed, and workers have to choose from jobs that are close enough that they can afford to drive there. And thus we have lingering long-term employment.