By NMA Board Member Christopher DiPrima
In previous newsletters (A Primer on Induced Demand Part 1 and Part 2), I outlined the complicated relationship between new highway lane capacity and induced demand – a much-misunderstood concept, which is often summarized incorrectly as, “you can’t build your way out of congestion.”
In those newsletters, I described some of the key mechanisms which cause new trips to be created after new capacity is added – most significantly, that people change their behaviors (they travel to places that they would not otherwise have traveled) and that land uses around new highway miles are often intensified (a new outlet mall is built near a highway interchange).
Following those newsletters, an NMA member asked a more fundamental question – a vital one which sounds so obvious that it doesn’t even seem like it needs to be proved: Do increases in population lead to traffic congestion?
The short answer is “yes,” but to explain why it is useful to understand the way in which planners model transportation systems.
The Four-Step Model
Since the 1950s, most transportation planners use a standardized four-step model to forecast traffic patterns:
Step 1: Trip Generation and Attraction: Predicting the number of trips originating or destined for a given traffic analysis zone (TAZ).
The first step in the process is trip generation and attraction. For a morning peak model, the trip generators will be mostly homes, while the trip attractors will be workplaces, shopping malls, and other places where people want to go. In the evening, this will be reversed.
Typical inputs for a trip generation model include population, income, land use, and employment. The single most significant variable in the trip generation model is population. Put simply, you can’t have trips if you don’t have population.
Step 2: Trip Distribution: Matching origins and destinations
In this step, trip generators and trip attractors are paired into a matrix of how many trips will be taken between each origin and destination TAZ.
Step 3: Modal Split: Assigning each trip to a given travel mode
The third step is to assign the trips to various travel modes. This is typically done using a statistical model called a nested logit model, which estimates the utility of each mode for a given trip. After accessibility (does the mode serve the trip?), the most significant inputs are typically the cost of a trip in both time and dollars – and those values will usually be different based on a person’s income and trip purpose (work versus leisure).
Step 4: Trip Assignment: Assigning each trip per mode to a given route
In the final step, the trips – which have now been assigned to a travel model – are assigned to a specific route in the network.
All of the later steps flow from the initial trip generation and attraction model. This means that holding all other variables equal, more population will lead directly to more trips. With a fixed, inadequate system capacity, more trips will lead to more congestion.
In practice, there are equilibrating factors, which temper congestion:
- Trip distribution – Under congestion conditions, people change their trip choices in both time and space. All else being equal, the more congested the transportation network, the fewer trips are made in the peak hour as a proportion of the day’s trips. Conversely, when new capacity is added, some of the trips made in the off-peak move to the peak. This is a source of latent demand, which is often mistaken for, or equivocated with, induced demand.
- Modal split – As congestion increases, the utility of a given mode decreases. Given other options, a larger proportion of the total population will choose another mode. However, this assumes that another mode is available and can provide comparable utility – if it doesn’t, congestion will simply continue to increase.
- Trip Assignment – As congestion increases, people’s choice of route will change. All else being equal and lacking desirable alternative modes, congestion should increase vehicle miles traveled as people choose less-direct routes to optimize their travel times.
Limitations of the Four-Step Model
The four-step model has been the standard in transportation planning since the dawn of the computer era when computers first became powerful enough and cheap enough to run massive simulations. However, it does have limitations. The four-step model is an aggregate model at its core since it works at a zonal level. Newer techniques, called activity-based models, attempt to solve some of the deficiencies of the four-step model by operating at a disaggregate level – for example, an individual household.
One limitation of any traditional transportation model is that it does not account for external changes in trip generation habits. The best example of this is the pandemic lockdowns: the population and number of jobs did not change, but trip generation was decimated because people stopped going to work. Therefore, decreasing the likelihood to travel can overcome the effects of increasing population.
An Urbanist Rebuttal and Counterpoint
The people most likely to dispatch the induced demand complaint are urbanists. They argue that population growth is at worst inevitable and is in most cases an unqualified good. A 2014 article in the journal Scientific Reports (link below) illustrates this line of thinking refreshingly directly. Earlier in the paper, the authors Rémi Louf and Marc Barthelemy write,
“The superlinear increase of congestion delay with population, and thereby of gasoline consumption and CO2 emissions, has terrible consequences on the economy, the environment, health, and well-being. The outlook is nothing short of grim in our ever-urbanizing world. As the proportion of human beings living in cities dramatically increases – the UN expects the world population to be 67 percent urban in 2050 – wages are likely to increase but not enough to compensate for the negative effects of congestion.” [emphasis added]
Later, they opine,
“It might be time to cut down considerably the use of individual vehicles, or to consider the possibility of living in smaller or medium-sized cities: the infrastructure costs (LN) may be larger, but the impact on the environment (CO2 emissions) and on the well-being of people (delays in congestion) would be beneficial.”
Urbanists suggest that only the first option is a good outcome and that we should put all of our efforts into making “superstar cities” bigger and more hostile to motorists. This has enormous geographic equity implications that are simply ignored because urbanists believe, almost by definition, that the only correct way to layout a city is in the pattern of old, large, coastal cities with extensive mass transit. They would also suggest that small and medium cities should attract more people by becoming more hostile to motorists and thus more urban. I call this an aesthetic preference toward urbanism – urbanists like urban cities, so they willfully exclude other viable options and claim that their preferred option is the only option.
I would counter that it is an equally effective approach, and in many ways preferable, to encourage growth in regions, which want and need the growth. Why should Amazon choose to locate its second headquarters in the congested DC area? What if our incentive systems encouraged development in depressed regions rather than further enriching the wealthiest parts of the country? However, this approach requires a degree of centralized economic planning that has fallen out of favor since the 1970s and may come with its own negative consequences.
If interested, here is additional information on transportation modeling and the effects of population growth on congestion?
- “Advanced Practices in Travel Forecasting” (National Cooperative Highway Research Program, 2010)
- “Why is traffic congestion getting worse?” (Case Studies on Transport Policy, 2020)
- “Our results show that road and transit network changes [in San Francisco] over this period have only a small effect on congestion, population, and employment growth each contribute about a quarter of the congestion increase and that TNCs [Uber, Lyft, etc.] are the biggest contributor to growing congestion over this period, contributing about half of the increase in vehicle hours of delay, and adding to worsening travel time reliability.” [emphasis added]
- “Scaling relationship between traffic congestion versus population size of 164 global cities” (Gachon Center of Convergence Research Working Paper, 2018)
- “However, the degree of congestion penalty increases rapidly, as the population size of cities in subgroups increase.” [emphasis added]
- “How congestion shapes cities: from mobility patterns to scaling” (Scientific Reports, 2014)
