White paper · November 2026The War Premium UnwindsMuch of shipping’s 2026 strength rests on the status of the Gulf and the Red Sea: spot rates on one main lane rose 305% after February, and Gulf tanker earnings are at records. The same state also drives costs, so the war is a windfall for some lines and a cost shock for others. Suez container transits are above half of pre-war levels, with a full return expected by early 2027, while the fleet is set to grow 10% in 2027 on an orderbook equal to 40% of the fleet. We consider three scenarios: A, contested normalisation (central baseline): earnings fall but stay positive. B, an energy supply shock (tail): costs and rates rise first, volumes fall later. C, the unwinding (the what-if; tail): rates revert as in 2023 while fuel, financing and ordered ships stay. For a stylised line the damage is not proportional: once the market tips into surplus, each further step of returning capacity costs several times more profit than the first. Thin-margin container carriers, tanker owners, lessors and older tonnage are most exposed, and no branch is benign for a large line.Read the paperPDFWhite paper · October 2026Firms in uncertain, dynamic networksFirms operate inside networks of suppliers and customers, and in many economies a small “rich club” of large, densely interconnected firms forms the core of that network. Strategy, performance and risk management at firm level are nonetheless built on single-shock sensitivities set against a central scenario. This paper asks when that is enough. It is enough as long as three things hold: a shock is small relative to the network’s distance to criticality, so that first-order perturbation theory is valid; the state of the network can be treated as known, although a firm in fact only holds a belief about it; and the network’s couplings can be treated as settled, although in fact they are continuously reshaped by the actions of goal-directed firms. In that calm regime, sensitivities are sufficient statistics, and variance bridges, strategic sensitivity tables, value-at-risk and hedge ratios all work. In most circumstances, substitutes are scarce, shocks can push networks into a dynamic or even a critical regime, most of the time there is significant uncertainty about the state of the network and network couplings are dynamic. Then all three assumptions fail: the response becomes non-perturbative, uncertainty about the state dominates the tail, and the network’s own dynamics take over. Superposition, attribution, sensitivity-based value-at-risk, risk contributions and plans built on fixed sensitivities then break down.Read the paperPDFWhite paper · September 2026European ResilienceEurope's strategy for resilience has to be realised in a world that remains highly interconnected: globalisation stood at a record-high 25% of its theoretical maximum in 2025, and trade grew faster than the global economy. This paper argues that connectivity and resilience are in tension, depending on the regime. In stable conditions, dense connections can diversify risk. In uncertain conditions, the same connections, where they carry concentrated dependence, transmit and amplify shocks non-linearly, in a way that is well described by the physics of systems close to a critical point. The evidence for the EU in 2025 shows a two-part structure: internal ties that are broad and dense (intra-EU goods exports of €4,025 billion against €2,584 billion to the rest of the world), and external ties in energy and digital infrastructure that are concentrated in a few hubs (the three US hyperscalers hold about 70% of the European cloud market, and the US supplied 56.0% of EU LNG imports). We conclude that Europe needs more autonomy as a whole and more interdependence within Europe, that companies should map and stress-test dependencies rather than connections, and that technology both accelerates the mechanism and enables preparation. The future cannot be predicted; it can be prepared for.Read the paperPDF