In 1991 the City of Münster (Germany) aimed to raise support for the city’s spending on bicycling infrastructure. The press office of the city made a poster comprised of three panels. Each panel’s background displayed Münster’s Prinzipalmarkt, the cobblestoned main street in town; what differed in each panel was what filled the Prinzipalmarkt. The first depicted what the street looked like if 72 people were transported by car; the second, by bus; the third by bike. The main point was that bicycling is an extremely efficient way to use limited transport space in cities.
Twenty years later, Indian-based organization, Earthian, posted the figure to Facebook. It quickly went viral. The graphic has since been used prominently in presentations worldwide to advance the arguments that bicycling is a strategic means of transport; bicycling can help reduce traffic congestion. In theory, it makes sense. And, the “reduced congestion” argument is one of the more prevalent ones in the advocate’s arsenal. It carries water for places with exceptionally high cycling use or for specific corridors. The argument, however, is fatally flawed when applied to larger units of geography.
The source of the substitution
The implicit assumption in the reduced congestion argument is that there is a substitution effect between a cycling trip and a vehicle trip. If true, it is helpful to consider the source of the substitution and its magnitude. For example, using mode share figures based on the US, a starting point suggests bicycling would draw from other modes in proportion to their current mode shares. Bicyclists would therefore draw roughly 85 percent of their market from driving trips (solo or with others), 10 percent from walking trips, and the remaining from transit or other. If most converters would come from existing transit users, a reasonably safe assumption but a population whose travel comprises a meager two percent of travel (again, in the US), then these numbers are relatively small. Tapping into the mindset of converting auto users presents a greater challenge.
The magnitude of the substitution
Understanding the magnitude of the substitution effect is difficult mainly because it is based on a counterfactual condition (e.g., if Booth would not have killed Lincoln, then…). There are two ways to think about it. One can start with a fixed number of vehicle trips and assume a share of those trips that are replaced by bicycling. Alternatively, one can consider the number of bicycle trips and estimate the share of these trips that replace driving. The two analysis strategies are not comparable.
Our own work found a wide range using the later approach, ranging between 25 and 68 percent of bike trips claiming to substitute for car trips. This suggests that half of the existing bike trips, were they not by bicycle, would be by car. Such rates of substitution, I presume, vary wildly across the globe. Then think about places where cycling comprises more than half of all trips: Groningen (the Netherlands), Münster (Germany) or Copenhagen (Denmark). It is hard to conceive of the character of these iconic bike-friendly cities if most trips were by car. Most bike trips substitute and the congestion savings is remarkable. But outside of these select settings, it is hard to know. In the rest of the world, if the logic is applied to a particular corridor, the congestion savings argument might have a noticeable impact.
Vehicular travel is not fixed
For most cities, there is a there is an inherent fatal flaw with the foundations of the above logic. These flaws become apparent when congestion is considered on a larger scale or for places that are growing (the birth rates in the above mentioned towns are not exactly skyrocketing). The foundation assumes a fixed demand for travel, and vehicular travel in particular. Prevailing trends of population growth and auto ownership suggest otherwise. Coupled with what is generally known about travel behavior, and driver behavior more specifically, these factors paint a meager picture for cycling’s ability to address congestion.
Traffic congestion is a problem is that is both old and complex. Rome struggled with it, resulting in Caesar simply issuing a ban on carts and chariots back in the day. Things aren’t that easy in most cities and more recent thinking has focused on its source and solutions. A good portion of that thinking is framed around issues of capacity: building more roads or widening existing roads and what is referred to as the effects of induced demand. Any form of relief provided for drivers, the story goes, will quickly be gobbled up. Drivers will defect to the corridor where the relief is offered. Other drivers, previously sleeping in and starting their commute at 9:30 will join the 8:00 am rage. Still others who were previously carpooling or using transit, learning that their route is as bad as it used to be, will switch to driving. Add traffic from the inevitable new development down the street and any immediate gains become a wash.
A key outcome from most of this thinking has been coined the “Iron Law of Congestion,” suggesting that once congestion has reared its ugly head in a city, there is little the city can do get rid of it. Congestion is largely an inescapable condition in all large and growing metropolitan areas across the world.
The same logic applies to gains from getting more people on bikes; it might provide temporary respite which would be gobbled. Suppose a community cycling initiative can leverage the previously explained phenomena to the fullest. Transit users and car users alike convert to bike travel for most trips. Getting more people on bikes might mean less people driving cars; it might mean temporary relief in the previously congested corridor. But using the Iron Law and considered across the region, the effects on congestion would be futile.
Congestion is less about capacity
But that’s not all. Driver behavior is also a culprit to congestion. Tom Vanderbilt’s book, Traffic, climbed its way to become a top ten best seller on the New York Times list in August 2008. It is the only book focused on transport that has claimed such a coveted spot. One of Vanderbilt’s principle claims is that traffic congestion has more to do with driver behavior than capacity issues. Drivers switch lanes, rubberneck, merge too early or overcompensate when braking—all relatively small flaws in driving behavior—but in the aggregate, they have a major impact on congestion.
Such driver errors and their consequences on congestion will diminish over time; automated vehicles will likely clean up such mishaps. However, this stream of thought suggests congestion has less to do with overall capacity in general. Even further evidence of the importance of the operations of a system in terms of congestion comes from New York City’ recent experience. Over the past years, the city has been aggressively reallocating select street space for cyclists and pedestrians, thereby decreasing overall vehicular capacity. Analyzing taxi cab logs, however, the Bloomberg administration contends that travel speeds have remained steady despite such decreased real estate—an outcome that is largely attributed to improvements to the city’s traffic signal system.
Cycling’s limits to reduce congestion is perhaps best understood when the argument is taken to an extreme. Scaling up to large and quickly growing cities, the latent demand ready to consume any relief to existing congestion is overpowering. Projections from the United Nations paint a picture of adding more than one million people to the earth every five days for the next dozen years. Admittedly, the majority of this growth will be absorbed in cities of developing countries and its impact being unequally distributed and perhaps even barely felt in Europe. But with rising incomes generally across the globe, the more than 60 million cars that are produced each year (more than 100 new cars every minute) will likely find willing drivers. The primary drivers to congestion (pun intended) are more powerful than anything that bicycling can realistically impact over the next decade or two, globally speaking.
There might be good reasons to spur bicycling and build more facilities to do it. The congestion relief might be felt locally. Aiming to reduce congestion on a regional level, however, is one of the least reliable rationales for doing so. Cyclists are better off making hay from other, more reliable arguments.
 Transport econometricians refer to this as the principle of Independence from Irrelevant Alternatives (IIA)
 Summary of Travel Trends: 2009 National Household Travel Survey, http://nhts.ornl.gov/2009/pub/stt.pdf
 Piatkowski, Dan, Kevin J. Krizek, and Susan Handy (2013). Accounting for the Short Term Substitution Effects of Walking and Cycling in Sustainable Transportation. Working paper available from the Active Communities / Transport (ACT) Research Group.
 Most of the original research is attributed to David Lewis [(1977), Estimating the influence of public policy on road traffic levels in greater London. J. Transport Econ. Policy, 11, pp. 155–168] and Martin Mogridge [(1990),Travel in towns: jam yesterday, jam today and jam tomorrow? Macmillan Press, London].
 Anthony Downs is largely credited with popularizing some of predominant thinking, starting with his 1992 book, Stuck in Traffic: Coping with Peak-Hour Traffic Congestion (1992), The Brookings Institution: Washington, DC. The concept is better explained in his updated work a decade later, Still Stuck in Traffic (2004), The Brookings Institution: Washington D.C.
 See, for example, http://www.nytimes.com/2013/09/05/nyregion/in-bloombergs-city-of-bike-lanes-data-show-cabs-gain-a-little-speed.html or other reports such as the “Green Light for Midtown Evaluation” (January 2010), from: http://www.nyc.gov/html/dot/downloads/pdf/broadway_report_final2010_web.pdf
 Population Division, United Nations, Department of Economic and Social Affairs. World Population Prospects: The 2012 Revision.
 Figures provided by the International Organization of Motor Vehicle Manufacturers, see: http://www.oica.net/wp-content/uploads/2013/09/total-production-2013.pdf