Planning For The Next Generation Of American Roads
American are due for replacement. This is a tremendous opportunity to fix inefficiencies, if only someone paid attention.
I. A Once-In-A-Century Opportunity
Eisenhower signed the Interstate Highway Act in 1956, and in so doing created one of the most extensive infrastructure networks ever built. Interstates today carry tens of millions of people on hundreds of millions of trips each day, contributing a vital part of the economy. Like any major infrastructure project, it was late. It was originally intended to take ten years, for a completion in 1966. However, more challenging sections were deferred, and construction took longer, and the primary system was “completed” in 1992.
Yes, even in the good ol’ days we were not all that good at building infrastructure. Full statewide networks took decades to complete, with the first state being Nebraska. Its section of the network, mostly just interstate 80, was completed in 1972. While highway surfaces are only meant to last around 18-25 years, bridges are expected to last for around 50 years.
[I wasn’t able to track down the original source on that 50 year claim. Literally every site that discussed bridges quoted 50 years as the typical lifespan. I assume it came from a 1950s-70s standard. For further context, the FHWA targets a 100 year lifespan, as of 2006.]
If most interstate highways were built in the 1960s and 1970s, that sets their lifespans to end around now. That tracks with current data - remember the panic about how America’s bridges were collapsing in the early 2010s? The current average american bridge is around 42 years old, although some of those presumably have lifespans greater than 50 years.
The fact that the interstate highway network is reaching its intended lifespan and due for replacement can be viewed as an expensive problem. At the same time, America has changed since the middle of the century. The population centers have moved. Cars have gotten faster. Engines have gotten more powerful. Incomes have risen, and technology has made toll collection easier than ever.
We have the opportunity to rebuild the interstate highways in the shape of a more modern America, taking full advantage of new advancements in concrete, cars, trucks, and tolling technology. The Biden Administration also has this opportunity - in fact, part of the bipartisan infrastructure law will go to building american roads.
However, the Biden administration has not taken this historic opportunity to put forward a vision of positive change for American roads. To the extent that he has outlined a purpose for the investment at all, it’s “restoring communities”.
The idea is that the first generation of interstate highways committed Original Sin by dividing the neighborhoods they pass through. Democrats then add a racial component by arguing that most highway users were from white suburbs, but most concentrated impacts were in black neighborhoods.
This is all more or less factually true - walking under a highway sucks, and highways do usually leave two neighborhoods on either side. It’s also bringing the highway to downtown inflicted the most damage to the highest density neighborhoods on either side - which were primarily black.
However, as an electoral message, this feels… uninspired? Depressing? Over the past 50 years, is there really nothing else that democrats want to improve about the actual highway usage experience? Is there no desire for modernization? This feels tremendously wasteful on Democrats’ part. It also leaves a huge messaging opportunity for Republicans to come up with something more positive and inspiring.
So, in the rest of this post, I’m going to try to outline some changes that Republicans should be advocating for to fill that gap.
IIa. The Speed Safety Con
Before I write this section, I feel the need to correct the record on the road safety, because as per usual various advocacies have warped public understanding of statistics. Despite having the fastest average travel speeds, interstate highways are by far the safest type of road to travel on. Fatality rates per 100 million miles reach only 0.55, compared to 1.3 for all other types of roads.
Just as a matter of basic intuition, “speed = danger” should clash with the idea that the fastest roads are the safest. If not speed, what makes a road dangerous? The answer is high speed conflict points. For the uninitiated, a “conflict point” is a way in which two possible vehicle movements intersect. For example, a typical T junction has 9 conflict points, and a 4 way stop is considered to have 24.
I have to admit that my intuition doesn’t really align with the specifics here - a divergence counts as a “conflict?” - but the general idea is sound. For example, roundabouts have fewer conflict points, and they’re a lot safer.
A typical rural highway has a gazillion conflict points. Every possible turn or driveway adds at least 9 conflict points, and way more on a multilane highway. If you drive down a typical state highway, you can feel this intuitively - the driveways are frightening. This picture of a typical state highway, for example, contains several hundred conflict points:
The worst possible situation for road safety is when two of these roads meet. Taking a look at the most fatal intersections in America reveals the same type of intersection over and over again. A big state/US highway meets another one at a traffic light, in a cluster of about a hundred conflict points. Here’s an archetypical example:
By contrast, interstate highways are safe because they do not have conflict points. By definition, interstates don’t have traffic lights, driveways, or stop signs. Almost all of the things that can cause crashes, even when drivers are behaving normally, stop existing. The framework of “speed is dangerous” is absolutely correct where roads have conflict points, but it has very little validity for interstate highways.
As a demonstration of this theory, we can look to Germany. Famously, its highways have no speed limits. If high speeds on highways were the primary driver of highway fatalities, we would predict high fatality rates. However, Germany also has none of the 3x3 arterial intersections which are especially common in the sun belt. In reality, Germany has a crash rate of 0.26, lower even than US interstate highways.
For a comparison closer to home, we can discuss Texas State Highway 130. I talked about this in the P6 post last week, but tl;dr the road is a toll highway parallel to I-35 between San Antonio and Austin. The highway has the highest speed limits in the country, at 85mph, and a crash rate of 35 per 100 million VMT. This is lower than the Texas average for comparable (rural interstate) highway average in the state, which is 53 per 100 million VMT.
IIb. Correcting Highway Speed Limits
The common understanding of speed limits is that they represent a tradeoff. High speed limits mean people get to places quickly, but the high speeds cause fatal accidents and kill people. On the other hand, a universal 10mph speed limit would result in zero deaths, but people would hate it. That’s because a 10mph speed limit wastes people’s time. Speed limits are best understood as a tradeoff between those two extremes.
However, as we’ve just established, that tradeoff doesn’t hold for interstate highways. The natural conclusion, therefore, is to simply not have highway speed limits. They are a waste of time for road users. However, mandatory interstate speed abolition would probably not go over well with voters.
If Americans don’t want to eliminate speed limits, the very least republicans can strive for is to implement speed limits that people actually want to drive. Highway design guidance already allows for this - sort of. In some cases, default speed limits are an egregious imposition and drivers disobey them even at the median. In those cases, best practice recommends raising the speed limit to the 85th percentile speed.
The flaw in 85th percentile speed limits is that they are reactive, not predictive. They don’t reflect the actual desires of drivers, because the speed limit influences how fast people drive even when they do not obey it. Future generations of american highways should aim to set speed limits at the speeds Americans actually want to drive, before considering the speed limit.
How fast is that? To answer, we can look to Germany again. Be prepared to shift your Overton Window, because it is *fast*. A study in 2007 determined that the average speed on the autobahn’s unrestricted sections was 88mph. That’s faster than every speed limit in America, but it might still be too slow.
Cars have gotten faster since 2007, and a second analysis was performed in 2019. That analysis found that the average speed was around 75mph, but the highest segment average speed was 100mph.
That was just the middle of the bell curve. 15th percentile speeds weren’t reported, but 10th percentile speeds were, and those averaged 115mph, and the maximum was 127. Electorally, a 125mph speed limit is likely a terrible idea, but it shows that the US DOT has significant scope to save drivers time by recommending less extreme raises.
IIc. Speed Limit Research
Part of the problem with recommending higher speed limits is that the federal government does no research on higher travel speeds. For example, speed limits are set so that drivers are able to stop after seeing an obstacle. That distance has been studied by the American Association of State Highway and Transportation officials (AASHTO).
They publish their recommendations in the AASHTO green book, which is updated about once every five years. However, stopping distance research in the most recent 7th edition (published 2018) is based on research done in 1984 and 1997. Acceleration data cuts off at 70mph.
Obviously, cars have improved in the past 25 years. Cars produced in 1984 struggled to even reach 90 miles per hour, so stopping distance research from that speed didn’t make much sense. Cars now have approximately twice the horsepower, and automatic braking systems. It’s safe to say that for any safety threshold, a “safe speed” in 1997 is a lot slower than a “safe speed” today.
Plenty of states today set their speed limits at 80mph, the maximum value in federal research. To rectify this, the federal government should include faster speeds in their research. These speeds should be set way higher than policymakers think is necessary - if international comparison is any reference this margin will be used quickly. Research should be done on speeds up to 110 mph at least, although I would consider something like 140mph to be a more future-proof number.
On a similar note, research is clearly not aware of improving vehicles. It certainly seems like the attitude at the NHTSA is that the one study from the 1980s will suffice in perpetuity. That’s reasonable for things like driver reaction time, which don’t improve with technology (at least, not until autonomous vehicles).
However, the same attitude obviously should not apply to things like braking distance, or maximum sideways force. Research in areas values depend on car technology should be updated at roughly the frequency which auto manufacturers replace models - about every 5 years.
IIIa. Truck Size Limits: Weight
Speaking of thresholds that eventually became a problem, the other notable highway threshold is maximum weights. Interstate highway weight limits were originally set at a maximum of 73,000 lbs, later raised to 80,000 lbs in 1974. Maximum weights exist to limit road wear, which appears to vary with the fourth power of axle weight.
To protect the pavement from wear, most states impose further axle-by-axle limits on how heavy truck can be. That includes a federal limit on single axles (20,000 lbs) and axle pairs (34,000 lbs). Some states, mostly in the midwest, allow larger truck combinations with a second trailer. Those trucks are allowed to weigh more, but the 34,000 lb limitation on axle groups still applies.
Importantly, however, this is a federal issue. The lowest possible maximum weight a state can impose is 80,000 lbs, and that governs the fleet that most trucking companies operate. The smallest possible longest truck, the 53’ tractor trailer, is also set by the federal government. These standards exist to make interstate fleet commonality possible, and are enforced with the typical federal funding threat.
Reviews of literature show that while per-axle weight matters but total truck weight doesn’t really change the wear rate. Furthermore, wear rate is also heavily dependent on pavement - roads designed to carry heavier trucks don’t suffer significant wear when used by heavy trucks, but roads designed for lighter trucks do.
The state with the most permissive rules on truck weights is Michigan. Instead of limiting total weight at 80,000 lbs, the state limits total weight at 164,000 lbs, but per-axle limits decrease as total weight increases. The theoretical maximum requires 11 evenly-spaced axles, as opposed to the typical 5. This saves roads in two ways - first, by incentivizing truckers to decrease maximum per-axle loads, and by enabling a single truck to do what otherwise would have required two trucks. By using one truck instead of two, one “tractor” is eliminated, which reduces road wear.
At the moment, no other states are even allowed to enact these laws. Michigan’s standard predates the federal standard, so it was grandfathered in. New states, however, can’t adopt such a standard, and Michigan couldn’t go back if it changed theirs.
This situation is, frankly, absurd. States should be adopting Michigan’s standards, which both preserve the roads and increase truck efficiency. Instead of limiting efficiency, the federal government should encourage other states to adopt Michigan standards, which would raise productivity and wages for trucking companies nationwide.
Because most goods have to travel by truck in the US, reforms like this would not only benefit truckers. Because these reforms would reduce per-weight shipping rates, truck deregulation is an easy way to combat inflation. In fact, the previous round of truck deregulation, the Motor Carrier Act of 1980, was passed at the peak of an inflationary spike in the 1970s-80s. Correlation isn’t causation, but it’s not nothing.
IIIb. Truck Size Limits: Length
Axle counts, however, are not the only way in which the US government restricts truck size. In addition to weights, most states cap truck size at a standard tractor plus a 53 foot trailer. This is up from 45 feet in the 1970s, following two rule changes to get it that far. Alternatively, trucks can tow two separate 28 foot trailers, which adds to nearly the same capacity.
However, some states allow longer trucks. Most Great Plains and Rocky Mountain states allow the “Rocky Mountain double”, a truck which tows both a 48 foot and 28 foot trailer, instead of two 28 foot trailers. Oregon, Kansas, Indiana, and Ohio allow three 28’ trailers. New York and Massachusetts allow two 48 foot trailers, but only on Interstates 90 and 87 (because those two are “turnpikes”). Surprise, surprise, Michigan from earlier permits the same type of double trailers on its roads.
This is ridiculous. The same literature review from earlier has found no impact on safety stemming from these truck size differences. At the end of the day, trucks carry most interstate commerce, and the variety in regulations helps nobody. States with decline to legalize trucks that improve productivity with no safety cost do so either because they are incompetent, or due to willful ignorance in the name of safety.
The federal government needs to intervene here. The distribution of funds to rebuild interstate highways presents the ideal opportunity to standardize weight and length limits at new, higher, and more uniform levels. This should be the guiding policy of any Republican-controlled FMCSA or NHTSA. As an added benefit, truckers are typically Republican, so this delivers concrete benefits for a large and important constituency.
IV. Interchange Frequency
Current federal guidance suggests that highways should aim for roughly an exit per mile in urban areas and an exit per 2-3 miles in rural areas. However, as American cities (and suburban areas) grow, demand for long regional trips through urban areas has also risen. These trips are long, and thus require high speeds, but are also urban, and therefore have to pick their way through traffic traveling far shorter distances.
The relatively novel occurrence of long trips through suburban areas raises some questions about how to deal with them, and what length of trip highways should cater to. Placing highway interchanges every mile means highways that serve both very short and very long trips.
I’ll use a highway near where I live - Interstate 95 between Greenwich and Stamford in Connecticut. A short ten-minute drive is ideally something suited to local roads. However, because of frequent exits, it makes more sense to use Interstate 95, which is the same road that must be used to get to New York City from the same region.
If cars doing the first type of trip are using interstate 95 even though they don’t need it, they cause congestion. That slows down longer regional trips, which can afford the slowdown less. As suburban regions expand, solving this problem becomes increasingly urgent. Large cities have increasingly turned to express lanes in some form.
These lanes reduce congestion by reducing vehicle eligibility, either through HOV restrictions, bus or truck only restrictions, or tolls. However, the most substantial examples use separated lanes, which have another feature: restricted physical access. Most express lanes only have a few entrances, around one per 3-5 miles.
By reducing the number of entrances, express lanes become unhelpful for shorter trips, even when wealthy suburbs could buy out the highway capacity for short trips. However, there are some roads (like I95 in Connecticut) which are important for regional travel and yet are too expensive to add express lanes to.
For roads like this, which are primarily in the Northeast, a program of exit removal needs to be seriously considered. Furthermore, future highway plans across the country should be classified as either “local” freeways (exits every 1 mile) or “regional”/“express” freeways (exits every 3-5 miles).
Roads which currently fit the “express” standard include major high-speed toll roads like the New Jersey Turnpike, the Massachusetts Turnpike, and the Indiana Toll Road. These roads provide high-speed travel through relatively suburban environments, and enable cities to access more land. Furthermore, the high exit distance enables higher traffic speeds, which fits well with the speed limit policy above.
V. Tolling, and Public Private Partnerships
I’ve already discussed public-private partnerships on this blog, namely all of the ways they can go wrong. That being said, I’m concerned about implementation details, not the idea as a whole. Getting the private sector involved in American roads will enable more construction and unlock more land for development.
Republicans should lean towards more private involvement, rather than less. A republican policy should not only solicit private operation of government-planned roads, but also making government eminent domain available for privately-planned toll roads. This would allow roads to be built where the market demands them, rather than where centralized state planners would like them to be.
Regardless of who leads the route selection process, roads will also increasingly require a paid component. Highway construction CPI has risen 50% since the pandemic, and infrastructure projects across all sectors have been getting more expensive.
Toll roads, or roads with a tolled component, are not only an effective way to pay for increased road construction. Increasingly, tolls are an essential component in finding enough project financing at all. Without toll roads, regions such as New York continue to choke on their own traffic. Everyone, no matter their urgency, must sit in the same traffic, while no mechanisms can pay for new roads.
Republicans need to fix this. With some thought and effort, they can fix this. They can make travel faster for everyone, increase truck efficiency, lower consumer prices, and fix traffic. Especially since Democrats have declined to put forward a substantial agenda, this is effectively a free lunch. The only thing Republicans need to do here is notice, and think.