Route survival after downturns

Why do some airline routes reappear after every crisis while others never come back?

A shock hits aviation — a recession, a pandemic, a fuel spike — and an airline suspends two hundred routes. Traffic returns a few years later, sometimes fully. Yet the network that comes back is not the one that went away. Some city pairs fly again within months at higher frequency than before; others were suspended "temporarily" in 2020 and are still gone.

The tempting reading is that recovery was uneven, or that management was arbitrary. But look at which routes return and the pattern is far too consistent for that. The interesting question is not how much demand came back. It is what the crisis did to the network on its way through.

Start with an observation about the state of things before the shock, because that turns out to be where the answer lives. A mature airline network is not uniform. It is a population of routes with wildly different economics, and in good times you cannot easily tell them apart, because good times cover a great deal.

Ask what that variation consists of. One route earns most of its money from a handful of business travellers paying many times the leisure fare. One is thin but feeds forty connections a day into a hub, so the accounts understate it badly. One exists because a regional authority pays a subsidy. One runs on a fifty-seat regional jet nobody would order today. One survives only because it was already there.

Now ask what a downturn actually does. Not "reduces demand by thirty per cent" uniformly. It suspends nearly everything at once and then forces a sequence of decisions about what to restart, in an order set by scarce aircraft, scarce crew and scarce cash. And this is the move worth being careful about: at that moment, restarting a route is not the default with cancellation as the exception. Every route has to win its aircraft back, competing against every other route for the same airframe. Variation that was invisible while everything flew becomes the whole basis of the decision.

That is a selection process in the strict sense, and it has the three parts one needs. There is heritable variation — the routes differ, in stable ways, before the shock. There is a filter — the restart decision, which is far more discriminating than ordinary trading. And there is differential survival, with the survivors defining the network that carries on.

So what does the filter select for? Follow the airline's actual constraint. With fewer aircraft than routes, the question is not "is this route profitable?" but "does this route earn more per aircraft-day than the next candidate?" That reorders things sharply. A route feeding a hub earns its keep partly through fares booked on other flights, so it survives a filter that would kill an identical standalone route. A leisure trunk with dense demand refills quickly and gets an aircraft early. A thin point-to-point link with no connecting value, flown on an aircraft type the airline is now retiring, has no argument to make.

Then notice two things that make the pruning stick rather than reverse.

The first is that assets get released. A suspended route frees a slot pair at a congested airport, a gate lease, a based crew. Those do not sit idle waiting; slots at coordinated airports are reallocated under use-it-or-lose-it rules, which were waived during the 2020 shock and then progressively reimposed. Once another carrier is using the slot, coming back is not a restart but a fresh entry against an incumbent — a much higher bar than the one the route failed.

The second is fleet. Aircraft retirements accelerate hard in a downturn because a crisis is when it finally makes sense to write off the oldest types. The 747 and A380 exits during 2020, and the long attrition of fifty-seat regional jets before that, did not merely shrink capacity; they removed the specific gauge that certain routes needed. A market too small for a hundred-seat jet and no longer served by a fifty-seat one is not a route waiting to return. It has lost its instrument.

Is this the whole story? Not quite, and one caveat matters. Selection acts on what the airline can see under stress, which is short-run contribution — so a route with slowly-built long-run value, a new market still developing its traffic, is cut on evidence that would have looked different in three more years. The filter is real but not wise, and some of what it removes was worth keeping.

The main thing to correct is the word "recovery". It suggests a return to a previous state, and invites the expectation that if traffic reaches its old level, the map should too. But traffic returning tells you the demand exists; it says nothing about whether the aircraft, slot, crew base and connecting bank that made a particular city pair viable still exist. Those are the things selection removed, and they do not come back with the passengers.

It is also worth resisting the reverse error — treating the surviving network as optimal because it survived. What the filter selects for is contribution per scarce aircraft under crisis conditions, which is a narrower criterion than long-run value and is measured at the worst possible moment. The network afterwards is better adapted to the constraint that was binding during the shock, and that is not the same as better.

Finally, note how you would check all this. Routes with high connecting value, dense leisure demand or a protected position should reappear quickly; thin standalone routes, subsidy-dependent links and anything tied to a retired aircraft type should be persistently absent. Broadly, that is what post-shock network maps show.

• How slot use-it-or-lose-it rules turn a suspension into a permanent exit, and what waiving them changes.
• Why hub-feeding value is systematically understated by route-level profitability accounting.

Every term the collection defines is gathered in the glossary.