Section 1
Framework: reading the shakeout with three models
The event to explain: a sub-sector crossed into loss at scale in one quarter, and now begins to restructure. Model 1: Comparative statics. The cost line that broke was per-head. Start by naming exactly what moved, because "costs went up" hides the mechanism. The decisive change was the National Insurance threshold dropping to 5,000 pounds. That converted employer NI from a tax that mostly bit on higher wages into something close to a fixed charge per employee. The arithmetic makes it concrete. A worker on 20,000 pounds took the employer's NI bill from 1,504 pounds under the old rules to 2,250 pounds under the new ones, a 50 percent rise for the same role (The Access Group). A part-timer on 6,000 pounds who previously generated no employer NI at all now generates a charge, because the threshold fell below where they earn. A business staffed by a wide rota of shift workers does not have one expensive problem. It has a hundred small ones, and they sum to a reshaped cost base. That is why this hit hospitality and not, say, a software firm with twenty well-paid engineers. Same policy, opposite exposure. The sector that tipped is the one whose cost structure was built from exactly the kind of employment the policy repriced. Comparative statics. The equilibrium-shift lens. • Assumes: you can isolate one moving variable and read its first-order effect. • Fits because: a single, nameable policy change hit one specific cost line, and we need to know which. • Breaks when: second-round effects dominate, once closures thin the field and survivors gain pricing power the simple picture shifts. • Counteracts: the vague "everything got more expensive" framing that hides who is actually exposed. • May reinforce: static thinking, treating a one-shot calculation as the end of the story. Failure mode in one line: it explains why the sector broke, not how the break propagates. Model 2: Threshold and cascade. Why the field thins in waves, not evenly. The jump from 15 to 23 percent in a single quarter is a tipping signal. A large group of operators was clustered just above breakeven, sharing the same shallow margin, so when the combined cost hit exceeded that shared buffer, they crossed the line together rather than one at a time. But the shakeout that follows is not a single event. It cascades. The first closures do two things to the survivors. They remove local competition, which hands the remaining operators a little pricing power and some transferred demand. And they spook landlords, lenders and suppliers, who tighten terms across the sector, which pushes the next-weakest cohort toward the edge. So closures come in waves. An early wave of the clearly unviable, then a slower grind as tightened credit and nervous suppliers squeeze the marginal-but-hanging-on. The sector does not settle at 23 percent loss-making and hold. It works through the distribution from the bottom up. This is the most important model for positioning, because it says your fate depends less on your absolute numbers and more on where you sit in the distribution relative to the operators around you. You do not have to be strong. You have to outlast the cohort below you long enough to inherit their demand. The distressed-asset window is the payoff at the end of the cascade, and it opens on a schedule you do not control. When the zombie cohort finally capitulates in a cluster, a lot of supply hits the market at once: sites with fit-outs already paid for, equipment at pennies on the pound, trained staff suddenly available, and landlords who have watched two tenants fail and now want a covenant that can actually pay rent. The operator holding cash at that moment buys growth cheaply that would have cost double in a stable market. The operator who spent their reserves keeping a marginal site alive through the same period misses it entirely. This is why balance-sheet strength is not a defensive virtue in a shakeout. It is the mechanism by which survivors compound their advantage while the field is still clearing. Threshold and cascade. The tipping-and-contagion lens. • Assumes: operators are clustered near a critical line, and each failure changes conditions for the rest. • Fits because: the loss share jumped discontinuously and closures visibly feed further closures through credit and supplier behaviour. • Breaks when: failures are independent rather than contagious. If closures do not affect anyone else's terms, you get steady attrition, not waves. • Counteracts: the assumption that a bad quarter is a one-time event that then stabilises. • May reinforce: doom loops in your own head. Not every closure nearby is a signal about you. Failure mode in one line: it maps the wave pattern, not the timing of any specific wave. Model 3: Behavioral. The zombie phase before the capitulation. Between "loss-making" and "closed" there is a long, dangerous middle, and it is governed by human psychology more than accounting. Sunk-cost keeps operators trading a losing site because of the deposit, the fit-out and the years already spent. Loss aversion makes cutting staff you care about feel worse than absorbing losses that are, on paper, larger. Optimism bias reads every decent weekend as the turn, right up until the cash runs out. The result is a population of zombie sites: technically loss-making, still open, quietly consuming the owner's reserves and, in aggregate, propping up rents and supplier expectations at levels the market can no longer support. The behavioral model predicts the shape of the capitulation. It comes late and then suddenly, because people hang on past the rational exit point and then all run out of runway around the same time. For a survivor, this phase is both a threat and an opportunity. A threat because zombie competitors keep prices and rents artificially high while they burn down. An opportunity because their eventual, clustered failure is when distressed assets, sites, equipment and staff, come available cheap to whoever kept their powder dry. Behavioral. The human-reaction lens. • Assumes: operators deviate from the clean exit decision in predictable, documented ways. • Fits because: the gap between loss-making and closure is a psychological delay, not an accounting one. • Breaks when: operators are unsentimental and well-capitalised enough to exit on the numbers. Chains behave less emotionally than owner-operators. • Counteracts: the assumption that loss-making businesses close promptly and clear the market. • May reinforce: cynicism, mistaking every slow-to-close operator for a zombie when some are genuinely turning. Failure mode in one line: it tells you the capitulation is coming, not the month it arrives. The structure-break flag Every model here carries the same warning. The April 2025 changes did not raise the sector's costs. They rewrote the sector's unit economics. The shape of the labour cost function changed from marginal-on-wages to fixed-per-head. That means any benchmark, valuation multiple or "normal" margin from before 2025 is describing a different industry. When you value a distressed site, model comps, or plan a new opening, the pre-shock comparables are not conservative. They are wrong. Rebuild them on the post-2025 cost function or you will overpay and over-hire into a structure that no longer supports it.
Section 2
Solution: how to be a survivor, not the median
The three models describe the shakeout. Now position inside it. Rank the moves, schedule them by reversibility, then sanity-check against how shakeouts usually end. GEER: rank the survivor levers 1. Pick a defensible end of the format ladder and commit. The squeezed middle dies first in a shakeout. Wet-led community locals and genuinely premium, differentiated food both have defensible positions. The mid-market pub-restaurant that is neither cheap nor special is exactly the cohort the cascade takes. Choose an end and mean it. 2. Raise output per head, because per-head is the cost that broke. This is the single highest-leverage internal move. Multi-skilling, simpler menus that need fewer hands, booking and ordering systems that remove a role rather than reshuffle it. You are attacking the specific line the policy repriced. 3. Fix your balance sheet before the distressed-asset window opens. Survivors in every shakeout share balance-sheet strength. Runway is not just defence here. It is the ammunition to acquire good sites, staff and equipment cheaply when the zombie phase capitulates. 4. Widen the day-part and the revenue mix. A site that only works on Friday and Saturday nights is a fragile site. Trade earlier, add off-premise or retail lines, use the space across more of the week. Each added stream lowers your exposure to any single one being repriced by the next Budget. 5. Decide the weak-site question honestly. If you run multiple sites, the shakeout is your prompt to exit the marginal ones on your own terms rather than let them drag the group down. Holding the median site out of loyalty is how a survivable group becomes an unsurvivable one. 6. Flag the no-lever case. A single marginal site on a fixed long lease that cannot reach breakeven on the new cost function is not an optimization problem. The honest lever is a managed exit before a forced one. A planned exit preserves the operator to fight again. A forced one does not. RADAR: schedule the moves by reversibility Do now (reversible, right in every scenario): re-forecast every site on the post-2025 cost function this month. Claim the Employment Allowance if eligible, which rose to 10,500 pounds in 2025 with the eligibility cap removed and takes many smaller operators to zero employer NI (The Access Group). Cut the trading hours and menu lines that lose money on any honest reading. Preserve cash. Hedge (cheap tail insurance): build the flexible-staff pool and the second revenue stream now, while you can do it calmly. Negotiate break clauses into anything you sign. Keep dry powder specifically earmarked for the distressed-asset window, because that window opens on someone else's timetable, not yours. Defer with a trigger (irreversible, wait and pre-commit the signal): do not expand into the shakeout on optimism. Write the trigger. If a target site or acquisition clears breakeven on the new cost function and you hold the runway to absorb a bad first year, act. If the promised reliefs fail to materialise at the next Budget or a further wage step is announced, execute the contraction plan you wrote in advance rather than deciding under pressure. The discipline is deciding now, calmly, what future signal changes your mind, so the frightened version of you does not get a vote later. CHAIN: check against how shakeouts actually end Match this to the right reference class, on structure not surface. Not "restaurants," but "thin-margin service sectors that absorbed a structural cost reset and then consolidated." The UK casual-dining reckoning of 2018 and 2019, when chains like Jamie's Italian, Byron and Prezzo went through closures and restructurings. Pub closures grinding through the years after 2008. Casual dining shakeouts elsewhere. The base-rate pattern is stable across them. The sector does not shrink and revert. It consolidates. Format polarises toward value and premium while the middle hollows out. Closures cluster rather than spread evenly. And the survivor cohort is defined less by clever operations than by two structural traits: a variable rather than fixed cost base, and a balance sheet strong enough to still be standing when the distressed assets come cheap. Operational brilliance on a fragile balance sheet loses to average operations on a strong one, most of the time. The present tilt: this shock is concentrated on headcount cost specifically, which raises the payoff to the labour-productivity lever and the penalty for a large fixed rota beyond what the historical analogs would suggest. It also changes who the natural consolidators are. In a headcount-driven cost shock, the operators best placed to buy up distressed sites are the ones who have already solved output-per-head, because they can run an acquired site on a leaner rota than the seller ever could. That is a different survivor profile from the 2008 pub closures, where cheap acquisition favoured whoever had the strongest drinks-supply deal. Read the analog for its shape, then adjust for the fact that this time the edge is operational leverage over labour, not purchasing scale over stock. One more caution on the reference class. Every analog above happened without a favourable policy reversal arriving mid-shakeout. If a relief lands here, the cascade can stall halfway, which strands anyone who acquired aggressively on the assumption that the field would keep clearing. That is not a reason to sit out. It is a reason to size acquisitions so that a stalled shakeout is survivable, not fatal. And the matrix-break flag. If a hospitality VAT cut arrives, the kind campaigners like Tom Kerridge have pushed and that Ireland has moved toward, the unit economics rewrite again and the whole base rate resets. A plan that only works if policy stays frozen is not a plan. Build for the reset, because in this sector another one is coming.
Section 3
The blind spot
Here is what none of these models can see. They map how a sector restructures under a cost shock. They cannot price the politics that caused it or the politics that might reverse it. Whether reliefs get extended, whether the sector gets singled out again, whether a VAT cut actually lands, those are decisions made in a system that does not read your P&L. The entire framework is built to make you robust to a decision you cannot forecast, not to help you forecast it. If you catch yourself betting the business on a specific political outcome, in either direction, you have left the framework and started gambling.
Section 4
The fitness test
Run your operation, or the sites you own, through two branches. Branch one: the shakeout accelerates. Credit tightens, two more waves of closures roll through, and the reliefs you were hoping for do not come. Are you positioned to outlast the cohort below you and inherit their demand and their assets, or are you in that cohort? If you cannot answer that cleanly, you do not yet know where you sit in the distribution, and in a cascade that is the only thing that matters. Branch two: the sector gets relief. A VAT cut or a rates reprieve arrives and the pressure eases. Are you built to seize a recovering market, with the balance sheet and optionality to expand into it, or have you cut so deep and locked in so hard that you cannot move when the moment comes? Surviving the downside and missing the upside is only winning half the game. The operators still standing in three years will not be the ones who called the policy right. They will be the ones who built a business that did not need to.