Last year on streets.mn, David Levinson wrote about a study which projected what accessibility in the Twin Cities might look like in 20 years under a variety of planning scenarios. I think accessibility is pretty great, and so I was very happy to get to work with David on another accessibility-focused project. “Access to Destinations: Annual Accessibility Measure for the Twin Cities Metropolitan Area” was recently published by MnDOT, and it presents a process that can be used to evaluate accessibility for the entire metropolitan area on an annual basis. I’ll provide a brief summary — but first, I want to talk about what accessibility is, and why I love it.
(If you’d like to learn more about accessibility evaluation, we will be presenting both of these studies at a seminar hosted by the Center for Transportation Studies)
What is Accessibility?
Accessibility measures the ease of reaching valued destinations.
Think about the last trip you made. You were in place A, you wanted to be in place B, and you got there by paying some sort of cost. There are several components at work here.
First, lets look at your reason for wanted to be in place B. That’s the whole reason the trip took place! You didn’t just want to move around for a bit, and then come back to A. (People do that sometimes, but we don’t consider it travel — we call it things like recreation, exercise, or NASCAR.) No, you had a reason for traveling to place B. That reason, from a transportation economics perspective, is the value that place B holds for you.
Now that we’ve established that place B has some value to you, it would be great if you could get there whenever you want, for free. But you can’t. Even if you frugally decide to walk from A to B instead of drive, fly, take the bus, or ride a horse, you cannot escape spending a precious resource: your time. The total amount of resources you spend in getting to B, whether they are time, money, oats (for your horse), or some combination, is the cost you pay to reach place B. To generalize just a little bit more, we can say that when the cost of getting to B is low, then B is easy to reach.
That, in a nutshell, is accessibility. It combines the value that we get from travel with the cost of making the trip. I love accessibility because it neatly encapsulates what I consider to be the fundamental purpose of transportation systems: providing ways for people to get to places they want to reach, at costs they are willing to pay.
The goal of the “Annual Accessibility Measure” project was to develop a method that MnDOT could use on an annual basis to measure accessibility in the Twin Cities. The idea is that it would be a useful performance metric to measure over time. If, in 2013, MnDOT is able to say that more people in the Twin Cities are able to reach more valuable destinations more easily than in 2012, that would be a pretty clear sign that our transportation system is serving us well. To do this, we first had to settle on a specific accessibility measure to use.
Turns out there are lots of ways to actually measure accessibility. Think about all the different ways we might talk about cost: you could measure it in dollars, in minutes, or some combination; you could evaluate the opportunity cost of things you passed up in order to travel; you could even measure how much time it feels like you are spending when you wait for the bus instead of driving.
Things are even more complicated when we talk about the value of destinations. What is the value of reaching a grocery store? Do all grocery stores have the same value? Is it more valuable to you to be able to reach a Cub Foods, a Trader Joe’s, or a Byerly’s? Different people assign different values to different destinations, and those values even vary from one day to the next.
People have built very successful academic careers around teasing apart these components of accessibility, and there exist measurements of accessibility that incorporate very detailed individual preferences and constraints. But they are hard to implement; they require very detailed data collection. They are also hard to explain and interpret, since they are prone to invoking terms such as the space-time prism.
To avoid these pitfalls, MnDOT needed a much more straightforward way to measure and talk about accessibility. We settled on cumulative opportunities accessibility, which measures how many destinations can be reached within a given cost threshold. We also decided that we would consider costs only in terms of time, since this would produce results that would be meaningful to the greatest number of people. Cumulative opportunities accessibility is a locational measures, which means that it is measured for a place, rather than a person. As an example, I might calculate that from my home, I can reach 750,000 jobs within 20 minutes by car during the AM peak period.
With these guidelines, we developed a process that can calculate cumulative opportunities accessibility to jobs, by car and by transit, throughout the Twin Cities — and we applied it to evaluate accessibility in 2010. Lets take a look at those results. (For more details about the evaluation process, you can read the full report.)
Accessibility in 2010
The map below shows the number of jobs reachable within 20 minutes of travel by car. People who live in red areas can drive to over a million job locations in 20 minutes, while people in blue or green areas can reach only a fraction of that. This is measured during the AM peak period (7:00 – 9:00), and it accounts for average speeds on roads and highways during that period. We can draw some conclusions and form some theories from the distribution of accessibility in the region:
- All roads lead to Minneapolis.
From a regional employment accessibility standpoint, Minneapolis is where its at. The highest regional job accessibilities peak in downtown and south Minneapolis.
- Highways matter…
We can see arms of accessibility reaching along the region’s radial highways — especially I-35W south of Minneapolis and I-394 to the west.
- …but so do employment centers.
High accessibility stretches farther along I-35W to the south of Minneapolis than it does to the north, and farther along I-394 than along I-94. That’s because high accessibility comes both from being able to get to far away jobs quickly, and from having a lot of jobs close by. What sets I-35W and I-394 apart is that they connect not only to downtown Minneapolis, but also to the corporate offices, malls, and supporting businesses in places like Richfield, Bloomington, and Golden Valley. The “Crosstown” Highway 62 is a particularly high-accessibility stretch, equipped with direct freeway access to the Golden Triangle, the Mall of America, the southern I-494 corridor downtown Minneapolis. Perhaps thats part of the reason that average home value in Edina is twice as much as for Hennepin County as a whole.
Taking a look at accessibility to jobs by transit, we see a strikingly different picture:
- A fraction of the accessibility.
At any given location in the Twin Cities, the transit network provides only a very small fraction of the jobs accessibility that can be achieved by driving.
- Islands of accessibility.
Due to the lower speeds of transit compared to driving, accessibility by transit is less continuous across the region, forming distinct clusters around major job centers. When speeds are low, proximity becomes a more important determinant of accessibility.
It’s All (Modally) Relative
Looking at the maps of car and transit accessibility side-by-side, it’s easy to just notice the extreme differences and leave it at that. But the pattern of the differences is important, and so we created maps that show the relative accessibility provided by driving and transit. Here, the values shown on the map are the ratios of transit accessibility to driving accessibility — so if a location has a value of 0.25, that indicates that the number of jobs reachable by transit is 25% of the number reachable by car.
(Note that for this map I am showing the accessibilities based on a 40-minute travel threshold because it does a better job of illustrating some spatial patterns.)
- Unexpected influences
There is a high-ratio cluster centered at Snelling & University – what’s going on there? You might not guess it from looking around, but this area is one of the places in the Twin Cities where transit accessibility is most comparable to auto accessibility. This is in part because of the local transit network, but also because of the road network.Snelling & University is an intersection of two high-frequency, high-ridership routes: the 84 on Snelling and the 16 (plus its limited-stop sibling the 50) on University. Just to the south, riders can transfer to the 94, which offers direct freeway access to both downtowns. That all adds up to pretty good transit accessibility to jobs.At the same time, accessibility to jobs by car is actually fairly low here — at least, lower than we might expect given the location. This is a result of both geographical factors and traffic factors: the area is hemmed in by railroads, the fairgrounds, and Como Park, and the only local highway access is via the notoriously congested I-94.This combination of high transit accessibility and low auto accessibility means that we might expect transit to be used more frequently in this area than in other parts of the region. But that is a topic I will leave for another day.
- Express transit service matters. Notice the scattered locations in the western suburbs with high transit/auto accessibility ratios? Those are places that offer express transit routes from park-and-ride stations direct to downtown. These services rarely stop and can take advantage of high-occupancy/toll (HOT) lanes on highways (known locally as MnPASS) to provide travel times that are more comparable to driving.
Challenges to Measuring Accessibility
We also identified some things that make annual evaluation of accessibility challenging, but which can be addressed through planning, practice, and research.
- How fast are people going?
We have an excellent system of loop detectors embedded in the pavement of our urban freeways, but actual measurements of speeds on arterials and lesser roads are rare (and very expensive to implement). To accurately measure accessibility by car, we need to have an accurate picture of roadway speeds. Installing loop detectors everywhere would cost a fortune (though… how many could we get for the price of a stadium?), but there are other ways to get this information. GPS is a promising technology for monitoring speeds on arterial roads.
- Comparing auto and transit accessibility.
This can be tricky. The issue is that when we drive, we can leave whenever we want — so it makes sense to use average speeds to calculate travel times. But when we use transit, we can only leave when trips are scheduled to depart. If we use the travel times provided by a single trip (as we did in this implementation), we are effectively assuming that everyone is perfectly willing to adjust their own schedules to match the transit schedule. That’s not very realistic, since schedule constraints are an oft-cited reason why people don’t use transit! We have some ideas for how auto and transit accessibility can be compared more meaningfully, but for now it is important to keep this issue in mind when comparing the two.
I Hope You Love Accessibility, Too
Accessibility can be a tricky thing to talk about; it is a fairly straightforward concept, but detailed discussions can quickly turn into geek-fests of metrics, distributions, and space-time prisms. But I think that when applied thoughtfully and with a consideration for simplicity and ease of understanding, it can be a very powerful way to discuss the performance of transportation systems. It incorporates the two fundamental motivators of travel — the value of destinations, and the cost of reaching them — into a single metric that is easy to compare across space. What do you think about accessibility as a tool for understanding transportation, particularly here in the Twin Cities?
“But I think that when applied thoughtfully and with a consideration for simplicity and ease of understanding, it can be a very powerful way to discuss the performance of transportation systems.”
It’s also a way to discuss the performance of our land use systems, since “opportunity accessibility” is a product of destination density AND transportation system performance. I hope you present this not just to MNDOT (if you only have a hammer…) but to the Met Council, MPCA, and local city council’s as they have land use authority. If the goal is to improve accessibility, is it cheaper to do it through more asphalt, or through higher densities of destinations?
That’s a great point, and I didn’t talk much about the land use side of things in this post.
I feel like the popular conception often sees land use as more static than transportation; it is the background canvas upon which the dynamic lines of transportation are drawn. Discussion of expensive, exciting transportation systems tends to magnify this effect.
But in reality land use is far more dynamic! Transportation systems change in big fits and starts, but land use is constantly evolving in countless small ways. Even if we completely halted new investments in transportation, we could still make huge changes in accessibility through the policies we choose to guide land use development.
…now that I think of it, transportation systems also change in Small Starts 🙂
Thanks for the great writeup!
One of the concerns I have with studies like this is that because they are data-driven and the data are necessarily summarized, it can lead to incorrect conclusions.
Here’s one example. Your point about Snelling and University is a good one. A layman might see that and ask, “why are we building Central Corridor if transit is so good there already?” We need to take into account the local situations of the neighborhoods where we are considering changes.
Different communities have very different accessibility needs. For example, the Harrison neighborhood did a job skills inventory and found that its residents have skills that match up very well with the job center developments going on in the SW suburbs. An averages-based study like this will miss this very important information.
These studies are very useful, but they also must be used in the proper context with lots of additional information, particularly from *people*.
David, you’re exactly right about the limitations to this type of evaluation. By measuring accessibility like this, we are sort of measuring potential trips. But as you point out, there’s no guarantee that the trips the system can provide are trips that people actually want, and there is no (simple) way to detect trips that people want but that the system is failing to provide.
isn’t that why it is better to use Utility-based accessibility measures?. I mean if you can collect data of destination choices and mode choices, and fit a Random Utility Model (discrete choice) then you are better able to predict the trips by mode that people want to take. You can use the Expected Maximum Utility (also known as inclusive utility, accessibility, logsum) to know the attractiveness of destination and mode pairs. The problem is that the sample should be representative, but if you use data from Household travel survey it should work, and also some planning data about travel times and travel costs (network skims).
Also another point we discussed last year when I was at MN, shouldn’t we calculate accessibilities based on land use?. There are important questions about reaching grocery stores, and reaching clothing stores, and so on, and what about coffee shops? :). Couldn’t we calculate accessibilities by certain land uses?