Airport access road widening
A Socratic walk-through of airport access road widening — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why does a wider road to the airport stop saving anyone time within a year?
The airport approach is jammed every morning, so a lane is added. For a few weeks it is wonderful: the run from the motorway takes eleven minutes instead of twenty-six, and everybody says so. A year later the same drive takes twenty-five, and nobody can point to what went wrong.
The natural explanation is that traffic simply grew. But that relabels the puzzle — growth was happening before, and the scheme was sized for it. Something is restoring the delay rather than merely eroding the gain.
Reasoning it through
REASONING #Start with the assumption buried in the scheme. It treats the number of vehicles wanting the road as a fixed quantity, to be served by more or less capacity. Is that fixed?
Consider what a traveller heading to an airport is actually deciding. Not merely which road, but which mode — car, taxi, rail link, hotel shuttle, coach; whether to park at the terminal or off-site; and above all when to leave. That last one deserves attention, because a trip to a flight is unlike an ordinary commute. Being late is catastrophic and being early is merely dull, so people do not aim at the average journey time. They leave early enough that they will still make it on a bad day.
Now follow what a faster road does to that choice. It reduces the typical journey and, at first, its variability. So the traveller who used to leave at half past five, judging that the worst case demanded it, discovers she can leave at quarter past six and still be safe. She has not saved fifteen minutes of travel; she has spent them on an extra quarter-hour at home. Multiply that by everyone making the same recalculation, and the departures that were spread across a wide, defensive window pull inward toward the hours people actually prefer. Traffic re-concentrates on the peak — and the peak is precisely where the congestion was.
That is the first refill channel, and it is the one specific to airports. Two more follow behind it. Drivers who had been using back lanes to dodge the approach return to the main road. And travellers who chose the train or the bus because the road was unreliable now find driving competitive again — an important channel here, because airport ground access competes across modes far more than an ordinary commute does, and because every rideshare or drop-off trip that switches puts two vehicle journeys on the road for one passenger, not one.
Then the slow channel, which makes it permanent. A faster approach makes new off-site car parks viable, makes an airport hotel or a freight shed worth building further out, and makes the airport itself a better place to add flights and therefore passengers. Those decisions run on the timescale of planning permission and are not reversed when the road fills again.
So where does it stop? Ask what each traveller is weighing. She keeps driving, at that hour, as long as the journey is tolerable relative to her alternatives — the train, an earlier start, a night in an airport hotel. As the road refills, the drive gets worse, and one by one people at the margin switch away or leave earlier again. The road stops filling exactly when the delay has climbed back to the point where the marginal traveller is indifferent between driving and her next-best option. That is an equilibrium, and it is the key move: the delay is not set by the road's capacity at all. It is set by how bad a journey travellers will tolerate before doing something else. Capacity determines how many people are enduring that delay, not how large it is.
Which yields a testable implication. If the equilibrium delay is pinned by the alternatives, improving the alternatives should lower it and adding capacity should not. Broadly that is what is observed: airports with a fast, frequent rail link tend to hold lower road delays, because the indifference point sits at a smaller number. The general road result is the one Gilles Duranton and Matthew Turner found for American cities, where vehicle travel rose roughly in proportion to lane capacity.
The analogy
THE ANALOGY #A busy surgery that takes no appointments makes patients pay in waiting. Add a second doctor and the wait halves — briefly. Then those who used to arrive at eight to beat the queue come at ten, those who gave up and went to the pharmacy come back, and the wait settles at about the length patients were willing to sit through all along. The surgery treats many more people, a real gain, but nobody in the waiting room can see it.
a patient's wait ends when she is seen, whereas the airport traveller's cost is dominated by uncertainty rather than delay — she is buying a margin against missing a flight, so a road that becomes reliable changes her behaviour even if it never becomes fast.
Clarifying the model
THE MODEL #The first correction is to the word "saving". A widening does deliver a time gain; what it does not do is let anyone keep it. The gain is real and gets spent — on a later alarm, on a trip that was not worth making before, on a flight schedule the airport can now support. So the honest verdict is not that the scheme achieved nothing, but that it bought access and volume rather than speed, and a business case built on time savings was built on the one benefit that gets competed away.
The second is about where the queue actually goes. An airport is not an open network; it is a funnel into a small number of fixed points — kerbside drop-off, the car park entrance, the terminal doors. Widening the approach does not widen those. Very often the visible effect of a wider road is that the queue relocates rather than dissolves: the approach flows and the kerb backs up. When the binding constraint is downstream, upstream capacity moves congestion without reducing it.
Two caveats belong on the finding. Short-run relief is genuine and can last years where little suppressed demand exists. And a genuinely new link is doing something other than adding lanes to a saturated corridor. What does hold delay down is anything acting on the price of the peak directly — charging for kerbside access, or making rail fast enough to move the indifference point.
A picture of it
THE PICTURE #How to readLeft to right is one year after opening, with the two things happening in each window. The channels are ordered by speed, not size — route switching is nearly instant, departure timing takes months, mode switching longer, land use longer still. Nothing here is anyone behaving badly.
What became clearer
WHAT CLEARED #The delay on an airport approach is not a property of the road. It is the level of misery at which the marginal traveller stops driving, and capacity cannot move it — only the alternatives can. A widening therefore converts a time saving into extra trips, later departures and eventually new buildings, and hands the queue on to the kerb, which was never widened at all. That is not a failure of execution. It is what happens whenever a scarce thing is given away at a price of zero and rationed by delay instead.
Where to go next
ONWARD #- Why travellers optimise against the variance of a journey rather than its mean, and what that implies for reliability schemes.
- Kerbside pricing and drop-off charges as the direct instrument, and who they fall on.
Key terms
TERMS #| Term | What it means |
|---|---|
| Induced demand | additional travel called into existence by a reduction in the cost of travelling, rather than by underlying growth. |
| Latent demand | trips that were worth making only below some cost, suppressed by the old conditions and released by the new ones. |
| Departure buffer | the extra margin a traveller allows to protect against missing a flight, sized by the worst plausible journey rather than the typical one. |
Every term the collection defines is gathered in the glossary.