Share on Social

Meet our episode 33 guest Simon Berrebi

Across the US, bus ridership is floundering. But instead of looking at this problem regionally, transit researcher Simon Berrebi studies bus ridership at a hyperlocal level and created a scalable way for the whole community to make their local transit better.

Read our Rail~Volution blog post to find out what happened at this year’s conference.

Episode Transcript

Cohen: Josh Cohen
Berrebi: Simon Berrebi

Cohen: What happens when you unite a bunch of Atlanta transit nerds to help improve transit in their local community? My guest today, transit advocate and researcher Simon Berrebi, shares how he cofounded the advocacy group MARTA Army and applied those same skills of hyper locality to his research on bus ridership. This is the second of two episodes I recorded in Vancouver at Rail~Volution. Let’s go.

F: Mobility is an essential component to the cities of our future. To build this future, we need to do more than invest in technology; we need to invest in the people who will make the hard decisions necessary to create vibrant, equitable, and sustainable cities. Welcome to The Movement where we talk to the brave leaders who are effecting change in an effort to build a coalition of leaders who will make tomorrow real. Here is your host, TransLoc’s National Director of Policy, Josh Cohen.

Cohen: Welcome to The Movement podcast. I am joined by Dr. Simon Berrebi. He is a researcher at Georgia Tech. He also cofounded the MARTA Army, and I am sitting here with Simon at Rail~Volution. We’re in Vancouver, British Columbia. It’s beautiful, and we are just taking a little bit of a break from the conference to have this conversation, so welcome.

Berrebi: Hello, Josh. And thank you so much for having me.

Cohen: Well, good. Well, I want to get started here with your research. You have several different areas of research that I think would be interesting here for our audience. Obviously, ridership; you’ve done a lot of research into transit ridership and then also some bus bunching and so forth, some ways to kind of address some of the issues that kind of impact potentially ridership. So I’d love to maybe start there with transit ridership, because I think that’s a fundamental question. And it’s obviously one that has impacted a lot of transit agencies, most transit agencies around the country here in the U.S.

Berrebi: So as you may know, transit ridership and so by extension cities and the rest of the world are currently in a state of unprecedented crisis. Transit ridership over all has declined every year since 2015, and bus ridership in the United States has declined every year since 2012.

Cohen: Wow.

Berrebi: In 2018 we actually attained the lowest level ever recorded in history, which started in 1965 when the American Public Transit Association started keeping track. It’s actually likely that 2018 was the lowest bus ridership year since way before that, possibly even since the dismantlement of the streetcar system.

Cohen: Wow.

Berrebi: So when we look at this crisis, you know, there still lacks a consensus on what is causing this decline. It’s particularly troubling when you think of the fact that service levels have increased since 2012 throughout the United States. Urban population have never been so high, and they’ve increased every year as well. Meanwhile, you know, the economy is doing better; unemployment is lower in particular for transit or demographics that correspond to transit dependent population, so in those circumstances typically you would expect transit ridership to increase.

Cohen: When you look at all those reasons why transit ridership should be high and yet is not, where do we go with that? What can we really understand to really try to tease out what’s really going on?

Berrebi: So to answer that question, we’ve been looking at the problem on a hyperlocal level.

Cohen: Okay.

Berrebi: Basically all of the research that has been done so far trying to understand the ridership trends over time have looked at the problem at a regional level. “How is Vancouver different from Atlanta? How is Raleigh different from Toronto?” But that obscures all of the variation that is taking place within cities and within transit agencies. And so our hypothesis was that whatever is causing the ridership decline or whatever will enable us to understand it is happening on a hyperlocal level.

And so in order for us to evaluate that we’ve used stop-level ridership data in Portland, Miami, Minneapolis, and Atlanta which essentially come from automated passenger counters, which are little lasers mounted on transit vehicle doors that count the number of legs crossing in and out. And so through that source of data we know how many passengers got on and off at each stop. And we cross these data with other sources that are also at the neighborhood level providing us information about service levels, population, jobs, demographic, etcetera.

And so through that we asked, “What are the hyperlocal changes happening that can explain the changes in transit ridership or in bus ridership happening over time?” And so what we found was that bus ridership in all four cities declined across neighborhood characteristics but declined the most in neighborhoods that were white and in some cases in neighborhoods that were highly educated. And so when you think about these trends, what that corresponds to is in the last few years an uptake in telecommuting, which of course is only available to people who work in white-collar jobs, and ride hailing.

Cohen: Right. Sure.

Berrebi: And so this research suggests that what is causing the ridership decline is not the behavior of people who depend on transit who may be encouraged to purchase cars now that gas prices are lower and that the interest rates have remained low over the last few years but rather choice riders who have access to a vehicle but who are either not making those trips any more or who are using ride-hailing companies for that.

Cohen: So do you think that’s a function of the ease of, say, TNCs, is that it’s the same whether you’re in San Francisco or Seattle? Right? You can go to the same Uber app or Lyft app, and it’s the same, whereas you can’t use your ORCA Card from Seattle in the Bay Area. Right? And so, you know, is it something about the ease in which people can choose that option? Is that part of what you’re seeing with that TNC? Or is that kind of, like, a step beyond kind of where your research would end?

Berrebi: I don’t have direct evidence on the net impact of TNCs on transit ridership, but from, say, a theoretical standpoint there’s two effects that are happening simultaneously. On one hand somebody who is a regular bus user might choose to ride an Uber or a Lyft on a day because they are late or it’s raining outside, and so in that type of situation the ride-hailing trip replaces a bus or a train trip. And there’s been some behavioral research about that through surveys, etcetera.

And then there’s another effect in which people like myself who ride transit every day and on some occasions might substitute a transit trip with a ride-hailing trip but also who justify not owning a vehicle because of the safety of being able to make that appointment or to get where they need to go if it comes down to it. And so, you know, I think that there is inherently both a complimentary and a competition between transit and ride hailing.

Cohen: Yeah.

Berrebi: You know, whether that effect is positive or negative, I think that there is some evidence at this point that overall ride hailing probably causes ridership decline from some of the research that we’ve seen from, for example, Greg Erhardt at University of Kentucky and others. But maybe in a different scenario it would be possible through a set of both regulation and incentivisation for ride hailing to be able to support public transit.

Cohen: Yeah, so let’s talk about that. So you’ve kind of given a couple of maybe kind of high-level ways that we could do something about this. Right? So do you have any—have you gotten into, like, that—obviously as a researcher maybe not but maybe in your background as an advocate you might kind of put that hat on, which is like, “So now that you kind of have done some of this research, what should we do about it?” Right? How do we actually bring about the change so that we can maybe start to address some of these challenges with ridership?

Berrebi: So I think that’s a fundamental question, because as folks are probably acutely aware climate is changing.

Cohen: Yeah.

Berrebi: And, you know, that is expected to generate 250 million refugees by 2050, create 250,000 premature deaths through diseases and natural disasters, and so as the greatest contributor to greenhouse gas emissions, we need to ask how can we get more people to travel together to make a better use of both space, resources, and at the same time minimize pollution while generating all of the positive societal outcomes of economic development and equitable access to opportunities.

The obvious answer to that is fixed-route transit service, whether it’s bus or rail. You know, getting people to travel together really is the most efficient from a societal perspective, however today if you look at the mode share of bus, rail, bicycle, TNCs, and taxis combined, they still represent significantly less trips than just pickup trucks alone in the United States.

Cohen: Wow.

Berrebi: So transit remains a niche market, and actually some of the research that we did in Portland, Miami, Minneapolis, and Atlanta helps us kind of understand what is the current business model of transit agencies and how that could be expanded. So one thing that we found in all four cities is that the most frequent routes are the most productive. So in your city, the route that runs at 10-, 15-minutes interval will typically be more crowded than a route that only comes every hour.

Cohen: But which comes first? Right? So in my mind I feel like—you know, the kind of heuristic I like to say is supply begets demand.

Berrebi: Mm-hmm.

Cohen: Right? But I guess you could also look at it the other way, which is, “Well, the reason we have 10-to-15-minute service is because there’s a lot of people on that corridor.” Right? So have you teased that out, that dynamic about which comes first there?

Berrebi: Absolutely. And in science we call that endogeneity, is when you’re trying to tease out a causal relationship but really both the explanatory factor and the response factor actually both affect each other. And that’s exactly the case because transit planners strive to maximize ridership, and that’s why they’ll allocate service in places where they believe the greatest latent demand lies.

Cohen: Right.

Berrebi: And so that’s why—you know, we looked at how ridership changes over time when the transit agency will add or remove service, and what we found is that bus ridership is generally inelastic to frequency. What that means is on almost any route every bus that you add to a route is going to generate less ridership than the average boardings per bus already on that route. And so that means there is negative returns.

That’s particularly true on high-frequency routes except in Minneapolis. That’s a bit different for other reasons. In Atlanta, Miami, and Portland, low-frequency routes are actually the most sensitive. They’re the least productive to begin with, but they’re the most sensitive to changes in frequency.

Cohen: I mean, that makes sense up to a certain point. Right? Like, because, again if you—so that makes sense that you would see it’d be less sensitive on the more frequent routes and more sensitive on the less frequent routes because you have the highest upside—right—or the highest potential upside from the ones that have the—I guess depending on demand.

Berrebi: Absolutely.

Cohen: Okay.

Berrebi: I mean, you know, if you double the frequency on a route where the bus comes every hour, you know, all of a sudden your bus comes every 30 minutes, so potentially you have to wait 30 minutes less. If you double the frequency on a route where buses come every 10 minutes, you know, the difference between waiting 10 minutes or five minutes may not be as great.

Cohen: Right. And then also I’m sure there’s a—this is probably not the right word, but there’s a—is it a temporal? Like, I guess if you roll out a change like that, people’s habits aren’t going to change over night. Right? Development patterns may emerge if you kind of move, you know, create something. So it’s maybe a lagging indicator. It may take six months or a year for that to start to work through the system perhaps or at least get down to the data where you’re capturing at the stop-level, passenger counter data?

Berrebi: Absolutely. And this research is over the scope of five to six years.

Cohen: Wow.

Berrebi: So I think we’re able to capture that lagging effect, but there’s a much more important lagging effect which is development, you know, where people live.

Cohen: Right.

Berrebi: And today the vast majority of Americans, even urban Americans, live in places that were built and designed for the automobile. And so I think one of the implications of this research is that transit, you know, fixed-route transit in these four cities has already attainted the point of diminishing returns.

Cohen: Hmm.

Berrebi: Right? So all the service that you’re going to add is not going to generate the same efficiencies of scales as the service that you already have. And the reason for that is because transit, the transit market is very niche. You know, it primarily targets people who live in dense places and in particular people who depend on transit as their main mode of transportation because they don’t have access to a private car. So I think that if—I mean, I know this sounds crazy, but hear me out. If we were serious about curbing climate change, if we were serious about significantly increasing transit mode share, just increasing fixed-route service would be a necessary but insufficient condition.

Cohen: Mm-hmm.

Berrebi: Another thing that we would definitely need to do is invest in transit-oriented communities.

Cohen: Of course, yeah.

Berrebi: And while that would take several decades to come into place, it’s something that would definitely need to happen to increase transit mode share in the long run. Another thing that would be necessary is to price single occupancy driving according to its real cost—

Cohen: Totally.

Berrebi: —not just on the space that it takes from society and on the inefficient use of that space that it makes but also on the global scale of destruction that it engenders.

Cohen: Yeah.

Berrebi: And finally if we want to be able to reach beyond this niche and to some extend insular market segment that transit is currently targeting, if we want to attain the critical mass to have transit ridership on the scale of this country and society as a whole, then we need to be able to connect people who live in places that were built for the automobile. And I think that’s where ride-hailing companies can transition from being the very real threat that they currently are into becoming an opportunity. You know, demand-responsive can be more efficient than fixed-route service in very low densities.

Cohen: Yeah.

Berrebi: And one point that has been discussed and that I think is correct is that at the density at which demand-responsive is more efficient than fixed-route, neither of them are efficient. You know, if you think about providing service in low-density suburbs and even in rural areas, demand-responsive might be more efficient than fixed-route, but neither of them are really viable.

Cohen: Right.

Berrebi: But the point that I think we need to consider here is that if you provide a door-to-door service that gives people a seamless experience from a ride that will pick them up at their porch within minutes and bring them to a high-capacity, fixed-route BRT or rail station where they’ll be able to tap their phone and just get on their ride to their final destination, then that can generate a lot more demand. One piece of evidence for that is Innisfil, Ontario, which is not first and last small but where the providence of subsidized ride-hailing service has reached beyond its budget and had to be curtailed by the City of Innisfil.

So at that point you’re not comparing low-density, fixed-route with low-density demand-responsive. At that point you can actually attain greater density not of origins and destinations or not of residential areas, residential homes and workplaces, but density of demand, so density of origins and destinations.

Cohen: Yeah.

Berrebi: And at that point when more people are willing to engage in this demand-responsive, first-and-last mile service, then that can become more effective than providing fixed-route all the way through.

Cohen: Yeah.

Berrebi: And of course the objective being that every mile between the first and the last be traveled in the company of perfect strangers.

Cohen: Hmm. I like that. You know, I think when you talked about the demand-responsive and the fixed route both being kind of inefficient in that sense, you know, what that inefficiency is a price of is the land-use decisions that we’ve made for the last 50 years. Right? And so here in Vancouver where just the sheer amount of density along their heavy rail lines or light rail—it’s their automated sky train—is just amazing. It’s just there is 30-story skyscrapers five miles, 10 miles outside of the city where they just all cluster around this in a way that I hadn’t—I don’t often see here in North America, which is just this like density that extends so far beyond and consistently beyond the downtown core, which I just think is so great. So I think that penalty that we’re paying is those land-use decisions that we made 40 or 50 years ago.

I agree with you though when you talk about some of the changes that need to be made, because I often think about this, that we’re all going to do what is the easiest for us unless we’re forced to do something different. Right? And so if the incentives are such that to be able to take a single-occupancy vehicle if you have one and drive somewhere and don’t have the pay the full cost, and because of the way the roads are allocated, you know, the single-occupancy vehicles get a couple lanes and there’s no bus lanes, dedicated bus lanes, or dedicated heavy rail or whatever.

People are going to kind of do what is the easiest for them, and so that’s why I really do think that the role of government and regulation and structure to say, “Yeah, we’re going to prioritize, and we’re going to make a bus lane. And guess what. For that other lane, if it’s a two-lane road, that other lane is going to be really challenging for those single-occupancy vehicles, so you’re going to have to think about whether you actually want to do that. And I’ll really force you to make that choice.” Right? And I think that’s where the role of that regulation really comes into play.

Berrebi: So you just said “the easiest for us,” and an important dimension in travel behavior is not just what’s easiest for me right now but what’s easiest for me or I guess what’s hardest for me.

Cohen: Okay.

Berrebi: So think about any transportation service. The way that you define quality in many cases is not by the typical experience; it’s by the worst experience. Because if even in five percent of cases you’re going to be, say, 20 or 30 minutes late because your bus departed just one minute early, that means you’re going to show up 20 or 30 minutes late to your appointment, your meeting, your work location. And for some people and for some instances that can have some very grave consequences, and so that’s why reliability is so important.

And I think that’s where public transit often fails the test in comparison to single-occupancy, private cars. And so the point that you were making, I think, is very important. Dedicated right-of-way is the only way for transit to be competitive because no one will use—you know, no one who has any kind of choice—and fundamentally everyone eventually has some kind of choice—will ride a service that slugs through traffic at the same speed as they would be able to in their own car.

Cohen: Yeah.

Berrebi: And so one thing that I found very impressive is efforts from King County Metro and other transit agencies to have a dedicated speed-and-reliability group that will work with local jurisdictions to cut a curb here, do a transit signal priority there.

Cohen: Oh, cool. Yeah.

Berrebi: And while they may only be saving 15 seconds at every intersection, these improvements, they scale up, and they end up making a real impact on not just travel speed but on reliability too.

Cohen: But that takes an agency to say, “This is important to do. This is important to invest in,” this particular group as you described, which, you know, they’ve got lots of different things they’re trying to do. And I really like that though, if you say, like, “Look; this is something we want to invest in and that we can have kind of excess returns, if you will, from this.”

Berrebi: Mm-hmm.

Cohen: I really like that.
Berrebi: So it’s not just one agency; it’s actually multiple agencies because the right-of-way typically belongs to the municipalities. So that means it takes one agency to work with the local municipalities, provide them with the guidance and the tools to make those decisions, and that’s a lot of people to have in the room when no ribbon is going to get cut.

Cohen: Right. [LAUGHS]

Berrebi: You know?

Cohen: Yeah.

Berrebi: Like, nobody is every going to notice this. Right?

Cohen: Yeah.

Berrebi: In the long term though, you know, that is going to affect people’s travel decisions.

Cohen: Yeah, totally. I love it. That may be a good way to maybe transition a little bit to the MARTA Army. I’d love to hear a little bit of the origin story there and what you really hoped to accomplish with the MARTA Army.

Berrebi: So it started at TransportationCamp South 2013. Lyle Harris, who actually still works at MARTA and at the time was the chief spokesperson—and he was actually at Rail~Volution—asked a bunch of transit nerds, “What would it take to make MARTA better?” Right? “What should we do in order to fundamentally improve service?” And the overwhelming reaction from the crowd was, “Leave it to us.”

Cohen: Oh, interesting.

Berrebi: You know, “You’re big and clunky; we’re young and innovative, but that’s not really what it’s about. It’s about the fact that we can impact the most important component of service of the transit experience, which is the local component.” You know, we can help improve transit in our own neighborhoods. And these efforts can then scale regionally to improve the transit service as a whole and to generate a constituency for better public transit in a fragmented metropolitan area, which is the Atlanta region. And so Lyle worked internally, and that was definitely an uphill battle to get a transit agency to allow a bunch of transit nerds to place signage at its bus stops and crowd fund trashcans and do a lot of things that definitely raised the hair of the legal department—

Cohen: Yeah.

Berrebi: —but that eventually created a group of transit nerds from every stride who did their part because they cared about their local neighborhoods and who together helped sink public transit deeper in the fabric of their own community.

Cohen: Well, what I think is so interesting about that is that as riders and so-called transit nerds y’all were closer to the problem, if you will. Right? So you’re the ones who are day to day dealing with some of those challenges. And whether it’s a lack of trashcans or a lack of wayfinding, whatever the case may be, you’re closest to that problem. And so what I’m hearing there, I think, is kind of around some design-thinking kind of principles around like, “You’re closer to the problem. Let’s let you kind of direct that a little bit,” as opposed to it being this top-down from the clunky MARTA. Right? I think that’s an element of that that I’m hearing and I think is really interesting that—I don’t know if it was intentional or not, but I think it definitely comes out.

Berrebi: It was intentional for two reasons. The first one is we quickly realized that if we were just to advocate for better transit we’d get the same group of people who would show up at any midtown or downtown location who were very homogenous. But for people who are transit nerds but who don’t necessarily work in public transit, they may not understand what the stakes are in terms of policy, and they may not feel connected to it.

When you give people an opportunity to improve transit in their own back yard, then the people who will show up truly represent the diversity of the united transit nerds of the Atlanta region, which correspond to the people who ride and love transit. And so when you get these people to work locally, what you create through that is a grassroots network. And that, we found, was very important not just for improving the transit ridership experience on the ground but also for providing information to MARTA whereas to the needs and desires of its riders.

And so, you know, for a transit agency engaging in a public engagement process following the successful referendum to expand transit in the City of Atlanta and also for other projects such as TOD, etcetera, the opportunity to connect with transit riders from every neighborhood and to target public engagement to get the opinion of people in such neighborhood or such neighborhood was something that was very valuable.

Cohen: Hmm.

Berrebi: And so we were very careful from the very beginning to create a scalable model of engagement where our role was to create programs, develop software, create a scalable structure that would enable anyone to participate. I never went out and placed signage at bus stops here and there because, you know, there’s 10,000 bus stops, and just driving from one end of the region to another can take an hour too. So putting it in the hands of the people who care about transit and giving them an opportunity to take a stand, we found, was the best way to maximize our impact and also to create this network of engagement which could help MARTA better respond to the wants of its riding population.

Cohen: That’s fantastic. I love it. So where can our audience learn more about either your research or The MARTA Army? What would be the best way for them to do that?

Berrebi: People can find out about MARTA Army’s current activities at They are currently doing a lot of software development to improve wayfinding and real-time information. They’re also working with young students to teach them how to ride public transportation, and so that’s some work that I’m proud of.

As for me, my research will come out sometime later this year or earlier next year. It’s part of a very large research project where we’ve been waiting to get the full picture before starting to put it out. In the nearer future we’ve actually written a TCRP study on transit ridership trends which should come out this fall and which I’m very excited about, which look at how bus and rail ridership have changed over the last few years throughout the United States and doing some case studies on several transit agencies which have seen particularly interesting trends.

Cohen: Excellent. Well, thank you so much for joining me, Simon. I really appreciate you taking us into the mind of the original transit nerd there that was kind of spearheading a lot of these kind of hard questions there and thinking a lot about that and then continuing to delve deep into those questions. And I appreciate the service that you did in founding the MARTA Army.

Berrebi: It was my pleasure, and I love the work that you do, and I can’t wait to hear more of your podcast.

Cohen: Awesome. Thank you.

Berrebi: Cheers.

F: Thanks for listening. If you like what you hear, head to Apple Podcasts and subscribe, rate, and review this podcast. You can find out more at or follow Josh Cohen on Twitter at @CohenJP. Be sure to join us next week for another episode of The Movement.


Read our Rail~Volution blog post to find out what happened at this year’s conference.