Urban mobility is more than comparing congestion
Deconstructing traffic indices
“Real-time information is crucial for cities that want to take control of their smart city initiatives”
says Eran Shir, Cofounder and CEO of Nexar in this Forbes article.
Commercial traffic data suppliers have various strategies to promote the application of their products. It certainly is easy prey for the media to have a pseudo scientific top 100 list of “Most congested cities” conveniently aggregated at hand.
Basing such a complex comparison on just one dataset and a few indicators per city, is ignorant at best. Only listing the World’s Smartest Cities asks for even more dubious methodologies.
It’s no wonder every traffic index has a different city at the top of their leaderboard — they don’t even publish the exact methodology at work.
Marketing ploy or science?
The problem is that both, Inrix and TomTom’s traffic indices just compare average commute times based on rush hour traffic vs some weird average trips in free flow traffic. As Joe Cortright at City Observatory pointed out in 2016, this methodology doesn’t take into account that dense cities have shorter commute times and therefore MUST be more congested than sprawled urban areas.
For completeness, I’m adding insights on methodologies I actually found (but I may have missed some, so please share if you have more insights).
Inrix Traffic Scorecard actually updated its methodology in 2016:
an average congestion rate measures the impact of congestion on a typical driver’s trip; and the peak hour spent in congestion metric gives auto commuters insight on their drive to and from work.
TomTom’s About page similarly doesn’t tell us much:
The congestion level percentages represent the measured amount of extra travel time experienced by drivers across the entire year. This is in comparison to measured travel times during uncongested conditions. We calculate and report the overall congestion level (all day) and the morning and evening peak hour congestion levels for each city.
To illustrate, an overall congestion level of 36% means that an average trip made takes 36% longer than it would under uncongested conditions.
Ok, let’s look at some other congestion indices… are they actually different?
Australian Transport Metrics
Uber Movement published a lengthy report with Infrastructure Partnerships Australia (IPA) to introduce the new IPA Transport Metric, with the Travel Time Index at its core.
The Travel Time Index uses Uber journey times data to measure whether a city’s transport network is becoming more or less efficient. […]
The index is an aggregate figure representing the degree of variance in journey times on the city’s transport network; that is, comparing journey times experienced during peak periods to what would be expected under free flowing conditions. The closer journey times experienced by motorists are to journey times under free flowing conditions, the better the city will score in this measure.
This is probably the closest we get to detailed insights on how these indices are calculated. And it’s disappointing on so many levels because the City of Melbourne particularly is focused on getting people off the road onto public transport, with varying results as this report of overcrowded and delayed trains suggests.
The Social Cost of Congestion
The Australian Bureau of Infrastructure Transport and Regional Economies (short BITRE) released a study that looks at the ‘avoidable’ social cost of congestion which makes the topic relatable. Social costs are defined as
where the benefits to road users of some travel in congested conditions are less than the costs imposed on other road users and the wider community.
This is a smart move — and understandably is still driving media attention. The study estimates the ‘avoidable’ social cost of congestion in the 8 Australian capitals to be AUD $16.5 billion for the 2015 financial year.
However their methodology sounds all too familiar:
The study estimates the level of average trip delay (and other social impacts) of current road congestion levels and forecasts how these will likely vary over time, depending on future movements in population levels, travel patterns, infrastructure provision and traffic management.
There’s so much more to cities than traffic
The irony here is that only dense cities can be smart cities. Especially vertically dense cities allow for the intended sustainability benefits in mobility and general consumption that metropolitan areas promise (more on LSE Cities). By highlighting sprawled out metros, these indices precisely counteract all common knowledge on what makes a cities efficient and smart.
To be fair, TomTom’s Beat Congestion section on the website at least suggests alternative modes on 5th place. But beating congestion by outsmarting the others, by using all road space surrounding an incident equally ignoring local policies on through traffic in residential neighborhoods (like Waze offers) is not the goal.
The end goal should be to get a growing number of city dwellers into shared transportation and ideally off the roads entirely.
In addition, all this is sadly ignoring the efforts that many ‘smart’ cities take to future proof their transit networks, encourage alternative active transport and shared use of “high occupancy” vehicles while reducing single-occupancy trips (plus the similarly negative effects of ride hailing), as the US DOT’s 2016 Smart City Challenge showed.