Restaurant districts

Why do rival restaurants crowd onto one street instead of spreading across the city?

No city planner drew the restaurant quarter. There is no committee that assigns kitchens to streets, no rule that says the Vietnamese places go here and the wine bars there. Every restaurant is an independent business, opened by someone who chose a lease for private reasons, and most of them would prefer their competitors were somewhere else.

Yet the quarter exists, has a name, appears on maps, and often outlives every restaurant that was in it when it formed. Where does a structure like that come from, if nobody built it and nobody wants the part of it that hurts them?

Begin with what a single owner is actually choosing between, because the whole pattern has to be built out of decisions of that kind. She has a budget, a cuisine, and two candidate leases: one on a quiet residential road with no restaurants, one on a street that already has eleven.

The quiet road looks better on the obvious measure — no rival within half a mile, every diner in the neighbourhood is hers. So why does she so often take the other lease?

Ask what she needs, one item at a time, and the answer starts to accumulate. She needs a chef who can leave a bad employer on Tuesday and start on Wednesday; that labour pool only exists where kitchens are dense. She needs fish delivered daily in small quantities, which a supplier will do cheaply on a street with twelve stops and expensively for one drop-off. She needs a landlord who understands extraction ducts, a gas fitter who has done this before, an environmental health officer who is not surprised by her. None of that is competition. All of it is cheaper by the dozen, and it is not available on the quiet road at any price.

Then ask about the diners, which matters more. On the quiet road, someone must decide to eat at her restaurant before leaving home. On the busy street, people decide to go to the street — and choose when they arrive. That is a different and much larger flow, because a group of six can agree to go somewhere with options far more easily than they can agree on one menu. Her eleven rivals are also, collectively, the reason anyone walks past her window at all.

So notice the shape of what we have found: the street offers her things that no restaurant on it provides deliberately, and that none of them could provide alone. That is the sense in which the district is more than its members.

Now let it run. Each new arrival makes the street slightly better supplied, slightly better known, slightly more worth a journey — which makes the next lease slightly more attractive than the quiet road, which brings another arrival. The loop is self-reinforcing, and it is why these places grow rather than settle. It also explains their odd origins: the seed is usually an accident. A cheap building, a wave of migration housed nearby, a market that used to be there, a tram stop long since removed. The initial reason is often gone and irrelevant; once the loop is turning, it no longer needs its cause. What sustains the district is not the same thing that started it.

Is the competition simply not real, then? It is real, and it bites — but on a different margin than intuition suggests. The rivalry is over share of the diners who came to the street; the clustering is over whether they came. When a street is thin, the second effect dominates and openings help everyone. When it is saturated, the first takes over, and this is where the loop turns on itself: footfall raises what a landlord can charge, rents rise until only the operators who can pay them survive, and the survivors converge on whatever reliably fills tables. The quarter that emerged from variety can end up homogenised by its own success, which is not something any restaurant chose.

And there is a test to keep this honest, because a story this tidy should be checkable. If the pull were mainly agglomeration of the kind described, it should be strongest where the meal is a discretionary outing chosen after arriving, and weakest where it is not. That is what we see: takeaways serving their own block, motorway service food, and hospital canteens do not cluster, because nobody travels to a district to compare them. Restaurants that are an evening out do.

It is tempting to fold this into the familiar story about competitors edging toward each other to steal the customers in between. That story is about splitting a fixed set of diners, and it is not what is doing the work here. The forces above are almost all about enlarging the flow — cheaper inputs, a deeper labour pool, a bigger reason to make the trip at all — which is why the arrival of a good rival next door can genuinely make a restaurant better off, something the splitting story cannot produce.

Second, emergence here does not mean anything mysterious. Nobody has to intend the district, but nobody has to be ignorant either: owners know perfectly well why the street is good. The point is that its properties — the name, the reputation, the pull on a Saturday — belong to the street and to no restaurant, and cannot be bought, moved, or defended by one of them. That is why districts are fragile in a way individual businesses are not: a road closure or a change in where the buses stop can undo one, and no single owner can answer it.

• Whether zoning or rent control can protect a quarter, or only freeze it.
• The same loop in other trades — fabric wholesalers, luthiers, garages — and why some cluster tightly and others barely at all.

Every term the collection defines is gathered in the glossary.