Rideshare and Congestion: Mixed Messages and Conclusions

I was in NYC last week for a wedding and when I travel, I like to use ride-share to get around.  I talk to my drivers about being a driver, the city, and anything else they care to share all while observing the city through a passenger window.  All the rides I took on this trip were interesting, but the ride from my Airbnb in the East Village to Penn Station on my way out of town was truly eye opening.  It was just before 9:00 am when I got picked up and it took over 40 minutes to make the trip, a bit more than it would have taken to walk.  I don’t feel bad about this, as I had a suitcases with me.  It felt like we got caught at every light.  Bikes and pedestrians seemed to be the biggest holdup at each intersection.  I heard sirens approaching and as they got louder I turned around and saw that an ambulance was a little ways back from us.  Nobody moved to make space and it seemed to move at a crawling pace.  After it finally got by, I asked my driver about this but he just asked “move to where?”

He was right.  Street-side parking had our three lanes completely hemmed in.  The backup of cars at the intersections kept anyone from moving forward.  The ambulance got caught in gridlock.  How is this happening?  I know that my single occupancy use of a ride hail app isn’t helping, but I was hardly the tipping point that created this situation.  Isn’t the combination of ride-share, bike-share and public transit supposed to reduce traffic?   I watched congestion impede an ambulance!

Since ride-sharing came into the American consciousness, it has been touted as the future of automotive mobility and a solution to urban congestion.  And why wouldn’t it?  It makes sense one person giving a ride to others would decrease the need for those others to use their own car.  Even recent studies by MIT and Arizona State University have indicated that this is true.  The Arizona study concludes that ride hailing apps do decrease congestion in cities but does not reach a conclusion on why.  The MIT study concludes that a fleet of 3,000 four-person cars offers the capacity to meet 98% of the demand on New York City’s current fleet of 13,000 taxis, if single-passenger occupancy were replaced with actual ride-sharing.

Both of these studies are being used to promote ride-sharing, but I have a few problems with that.  One objection is that the use of the term “ride-share” has come to encompass different things.  “Ride sharing” is when a commuter can pull out an app on her way to work, see if anyone on the way needs a ride to the same general destination, pick him up, and then go.  “Shared ownership” is offered by ZipCar, where users pay monthly fees, can reserve the use of a car and then pay by mile until it is returned.  “Ride hailing” involves pulling out an app, requesting a point-to-point limo service, and getting picked up by a driver who is on the road for the purpose of picking up fares.  These are completely different use cases.  The motivations behind them, the relative convenience of one over the other, time commitment, first/last mile, ease of use, cost, and availability are all different.

The study published by ASU took traffic congestion data from 87 cities from 2004 to 2014 and analyzed monthly data to compare traffic density before and after the months in which Uber launched.  Since Uber didn’t launch in any US city until 2011, there is more before data than after for all cities analyzed.  The study repeatedly cites research that was published before 2011.  This is significant as Uber is a point-to-point ride hail service.  Before it’s launch, the prevailing “share” model in the US was the shared ownership model.  The fact that this paper specifically concludes that Uber reduces congestion while repeatedly citing studies on different types of shared use without making this distinction is misleading.  Further misleading is the fact that there is no discussion of Lyft or other services entering those same cities.  The discussion of Uber is treated as though it operates in a rideshare vacuum.

Policy factors can play a role in reducing congestion (a great deal of policy is designed to do this), but policy initiatives are not discussed in the ASU paper either.  Investments in road infrastructure, pedestrian infrastructure, bike share and public transit all influence congestion but these are done in phases and often have no obvious before/after date.  Any such projects that came to fruition between 2004 and the introduction of Uber could lead to decreased congestion over time.  Any such efforts by cities would have been undertaken for the express purpose of decreasing congestion and are expected have an inverse causal relationship to congestion.  Uber was introduced to make money.  The ASU paper did not discuss any those as contributing factors to decreased congestion.

The economics of car ownership and driving matter too.  Per capita vehicle miles traveled (VMT) is, well, the average miles traveled in a vehicle per 100 people in the US.  As people travel more in cars, there are more cars on the road at any given time.  As they travel less, VMT decreases and there are fewer cars on the road.  Per capita VMT has historically been linked to economic factors such as GDP and the price of gas.  It peaked in 2004, the year from which the authors began culling data, and decreased every single year through 2011 – when it essentially flatlined – before beginning to increase again in 2015 and 2016.  The drop was most significant during and immediately after the recession.  The ASU paper assesses congestion data from 2004 to 2014 and uses the month of Uber entry into any given city as the before/after cutoff for analysis.  I feel that the VMT data provided by the State Smart Transportation Initiative at the University of Wisconsin (pulled from data provided by the US Census Bureau and the Federal Highway Administration) indicates that congestion declined before Uber appeared.

I’m bothered by the ASU study over the omission of so many policy and economic factors that could have played a role in decreasing congestion in US cities.  I’m bothered by the fact that the authors suggest that policy makers specifically consider Uber for its potential benefits instead of advocating for the consideration of any or all shared mobility.   To me, this paper reads less like an academic paper than a paid advertisement.  Despite that, ASU’s W.P. Carey School of Business actively promoted the publication by Arizona researchers as the first to “prove” Uber’s effect on decreasing urban congestion.  It is important to note that their research has not yet  been published in an academic journal that requires a peer review process, but that didn’t stop sites like the Seattle PI from reporting it’s claims.

I’m not critical of the MIT study.  Back in November, a team from the Computer Science and Artificial Intelligence Laboratory at MIT published the results of a simulation study in the Proceedings of the National Academy of the Sciences (which is peer reviewed and super prestigious!).  If, all ride-sharing eliminated single passenger point-to-point ride hailing and replaced it with actual carpooling, then a fleet of 3,000 four-passenger cars could “serve 98 percent of taxi demand” in NYC.  This is important research in terms of capacity and policy.  It shows that densely populated cities have room to implement policies that increase existing capacity utilization while decreasing congestion and emissions.  “Demand” however, is a relative term.  Engineers and marketers use the term very differently.  The demands on capacity could be met with a smaller fleet but the preference for a private ride over carpooling indicates that the proposed scenario would fail to meet consumer demand.  Consumers are willing to pay a premium for the convenience of unshared, point-to-point service, as evidenced by the fact that UberPool is only used by 20% of Uber users.  In theory, it is possible to get more people around on that many fewer vehicles, but consumers are unlikely to make that transition willingly as consumers seem to actively dislike it, especially in the US.

The problem with the MIT study is not in the publication itself, but in the interpretation thereof.  The crux of the study is to quantify the extreme limit of efficiency that could be reached and therefore demonstrate how much wiggle room there actually is for places like NYC to find the happy medium between consumer demand and greater societal needs.  This study involved creating a navigation algorithm that makes that outcome a possibility.  In a discussion of the study on MIT’s website, author Daniela Rus even discusses “…the trade-off between fleet size, capacity, waiting time, travel delay and operational costs…”.   Her’s is a discussion of finding a happy medium.  Publications like The Mercury News however, touted the potential for Uber and Lyft to reduce congestion by 75%.  Going from 13,000 taxi cabs to 3,000 cars is about a 75% reduction in the size of that of cabs fleet that operates in NYC, but that is very different from reducing the number of all vehicles by that much.  There are researchers actively tying research like this to advocating better policy, but even important research like this is contributing to an echo chamber of misinformation.

The companies selling on-demand ride service apps are selling the idea that they will help decrease urban congestion.  There are studies that seem to validate that assertion.  There are studies that don’t but are quoted as doing so anyway.  Are there studies that contradict those claims?  It turns out that there is one.  Alejandro Henao, a post-doc studying transportation systems at the University of Colorado in Boulder took the initiative to drive for Uber and quiz his passengers about their use cases.  34.1% of his passengers indicated they would have walked, biked or taken public transit, indicating a cannibalization of mobility modes meant to decrease congestion.  Another 12.2% reported that they would not have made the trip at all.  The ease of use and low cost of hailing a ride service that conveniently adds first and last mile service to the ride drives use.  His survey makes a strong case that congestion is significantly increased by apps like Uber and Lyft, and do not, on their own, reduce congestion.

Back in my Uber, watching an ambulance struggle to get through congestion that I’m actively a part of, my driver tells me that this has been getting worse.  He told me that access to emergency services was a growing problem.  And it is!  Hospital closures like that of Beth Isreal and St. Vincent as the population of Manhattan increased by 5.0% from 2010 to 2016 have put a strain on the city’s emergency services.  Fire officials reported a 17% increase in life-threatening incidents in 2015 over the previous year.  Mayor de Blasio actively campaigned on increasing access to emergency services, but budget increases and the addition of more response vehicles in 2014 still resulted in increased emergency response times.  Only the response times for life-threatening situations are reported, but in the time between the call and the arrival on scene, situations can escalate.

The traffic on either side of us is as dense as it can get.  No bike or pedestrian could fit between the rows of cars on either side of us.  The light turns green and cars up and down the row start honking while pedestrians are still in the crosswalk.  When the pedestrians are gone there are cars still stuck in the intersection trying to navigate across our path.  The honking dies down but another siren is blaring.  Coming from behind us a firetruck is at a standstill.  The van ahead of us isn’t moving but cars are inching forward to our left.  Right as the firetruck was about to pull alongside, my driver cut in front of it.  I was shocked, but he rationalized it by claiming that he could make more space by getting to the intersection.  I didn’t believe him, but the observation was still important.  Drivers in NYC are getting so jaded that some have stopped caring about giving right of way to emergency vehicles.  He has somewhere to be and the fire department can’t give him a ticket.  I understand intuitively that his behavior is a symptom.  It compounds an existing problem but is not the cause.  He even starts offering up solutions.  Delivery to stores should be done at night or early morning.  Personal vehicles from outside the city should have limited or restricted access.  I notice that he doesn’t suggest requiring his drivers to pool.  As we pulled up to Penn Station and I walked to my train, I came to a realization:  of all the economic, environmental, and health issues tied to congestion and the changing mobility landscape, access to emergency services in growing cities may be the first to become an acute problem.




One thought on “Rideshare and Congestion: Mixed Messages and Conclusions

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s