Home » Environment » Why More Roads Will Not Reduce Congestion

Why More Roads Will Not Reduce Congestion

Sign up to the Barefoot Economist’s Newsletter

© Barefoot Economic Services

All content on this website (unless otherwise specified) is copyright of Barefoot Economic Services. Please acknowledge any content used in your work appropriately. Thank you for your consideration.



I remembered in the good old days when it was a breeze getting around Melbourne, assuming you were driven (by my parents, not a chauffeur). Now, that I often travel by car for meetings, I notice how some of my favourite short cuts are no longer the fail-safe ways of cutting travel time. Even driving in the middle of the day could test the patience of a saint. We are lead to believe that more roads will reduce congestion. But the facts will tell you otherwise.

Congestion is Getting Worse

According the Australian Government’s Bureau of Transport and Regional Economics, the ‘social costs’ of congestion will increase from $A9.4 billion in 2005 to $A20.4 billion in 2020. This is made up of time delay costs for private and commercial use of $A7.4 billion and $A9 billion respectively, extra vehicle operating costs of $A2.4 billion and $A1.5 billion for additional air pollution costs. In the US, congestion costs are already estimated at $US200 billion a year.

Clearly, the issue of congestion costs is not insignificant here and around the world. The delay externality imposed by commuters is time that could have been spent on working or leisure, things that people value. But why wouldn’t more roads reduce congestion for commuters?

Why Roads Will Not Reduce Congestion

Economists are surprisingly in agreement on why roads will not reduce congestion. More roads, whether tolled or free actually increase congestion eventually for two reasons: the rebound effect and the displacement effect.

Rebound Effect or Induced Demand

More roads means it is less costly for a commuter to drive, at least in the short-term. However, if everyone thinks like that than more people are going to start driving to work. This is called the Braess Paradox where people do not take in account the responses of other users to the same problem. So, new road capacity effectively ‘induces’ demand for roads by reducing the cost of using it. While additional road capacity may temporarily reduce congestion, demand ‘rebounds’ as people observe the less congested roads.

Importantly, this Rebound Effect exists for toll roads as well as new free roads. The reason for that is that tolls are often set a rate sufficient to recover costs or generate a return. But tolls are often not set to discourage road users. If the road is privately owned, this wouldn’t make very good business sense.

The Rebound Effect also explains why population growth only partly explains road congestion. For example, the opening of the Sydney Harbour Tunnel increased car traffic across the harbour by 38% from 1992 to 1995 while population grew by only 4% during this period. Similarly, in the US, the introduction of ‘High Occupancy Toll’ (HOT) roads saw an increase in single person car travel, despite charges applying to such vehicles.

Despite the wealth of economic research on the Rebound Effect or induced demand, this is often ignored when governments propose a new infrastructure road project. The East West link is a notorious Australian example of a government providing an over-optimistic assessment of a project’s ‘congestion-busting‘ credentials. Obviously, ignoring the Rebound Effect means you can claim the reduction in ‘social costs’ from reducing congestion ad infinitum. However, research here and abroad would seem to suggest you could claim it for three years at best.


Tolls may be successful in reducing congestion on tolled roads, but this may encourage motorists to search for un-tolled alternatives to get to their destination. Unfortunately, if one motorist thinks like that, you can assume that many others will think like that. The effect is congestion on unpriced roads and potentially a decline of road safety, especially if drivers are choosing to cut through residential streets or school zones.

This displacement effect could be moderated by anticipating that motorists will change their behaviour in such ways and provide options to accommodate this such as widening unpriced roads. However, the toll operator is unlikely to accept such actions if they aren’t compensated for the fall in demand for their business.

How Can Governments Solve Congestion?

With all this economic research into roads and congestion, surely economists have thought of a solution? Yes they have and you may or may not be pleased to learn that it is a price-based solution. Economists have typically argued for a ‘congestion price’ charged to motorists that explicitly discourages people from driving (see here and here). It differs from the basic toll because it is set higher than what is required to earn a reasonable return, and it is actually designed to reduce demand. There are variations such as ‘dynamic’ congestion charges where the price changes depending on how congested traffic is. Or simply charged at peak hours to encourage motorists to drive in off-peak hours. London and Singapore are probably the best known examples of congestion pricing.

But congestion pricing isn’t sufficient to reduce congestion. Governments need to provide alternatives to other forms of transport such as public transport or cycling. Otherwise, commuters will have no option but to pay the congestion price with little scope to change their behaviour to reduce the cost. This means the government would need to develop long-term and intensive congestion-reduction policies in order to encourage significant private investment by infrastructure providers but also commuters to switch transport modes. Unfortunately, for governments, this will men there will be fewer photo opportunities at ribbon-cutting ceremonies but hard, thankless policy slog over there time in government. But maybe, they will create an urban environment that will work well where people aren’t spending unproductive hours in traffic. I know which future I would prefer.