Brokers placing property programmes for clients with catastrophe exposure are working from numbers that are reliable on one side of the ledger and, on the other, often little better than arithmetic.
That is the assessment of Patrick Hunter (pictured), Pacific head of risk consulting at Lockton in Auckland, New Zealand. He's spent his career on both sides of the loss - building pre-loss models and then standing on sites after the event to work out what actually happened.
"If I was to focus on the key area where that divergence can occur, it'd be around the business interruption (BI) side of things," Hunter said.
The distinction matters for brokers because the two exercises ask different questions and draw on different evidence. "If we consider cat modelling, for example, that's probabilistic because it's looking at probabilities of things happening," he said. "Whereas when you're looking at post-loss, the thing has actually happened, it's in reality."
Hunter reached for an unusual authority to make the point. "I think it's a Mike Tyson quote - everyone's got a plan until they get punched in the nose," he said. "Well, you're dealing with the company that's been punched in the nose."
The business continuity plans that a company relies on are rarely a match for the event itself. "Major losses often test business continuity plans in ways that are difficult to fully replicate before an event occurs," Hunter said. "Practical post-loss experience provides valuable insight into how organisations actually respond under pressure and can significantly strengthen the development and testing of BCPs."
Hunter's argument is not that catastrophe models fail. On the property side, he rates them highly.
"For a seismic study, a huge amount of data goes into the property loss side of things – looking at return periods, the seismology of the area, so on and so forth," he said. "They're generally really quite good at coming up with those probable property loss numbers."
The problem starts at the next line item.
"But then you roll into the BI - and often the business interruption side of things is really quite high level," said Hunter.
Hunter's example: a building is destroyed and might take 24 months to rebuild. The calculation that follows is often no more than arithmetic.
"The BI often is no more sophisticated than saying, okay, revenue divided by 12 times 24, that's the interruption," he said.
That approach fails in a specific and expensive direction. "The reality is that's never the case and the reality is quite often a business interruption loss for a major loss will be as big or bigger, sometimes multiples bigger, than a property loss," Hunter said. "So when you are using high-level assumptions to determine what a business interruption exposure is, that's where a key divergence occurs post-loss, because the reality is something very different."
For a broker, the practical consequence is that the least rigorous figure in the submission is attached to the largest potential exposure.
Asked what is routinely missing, Hunter pointed to the gap between what a model assumes about a recovery timeline and what a recovery actually involves.
"You don't start reinstating a building day one of a loss," Hunter said.
What happens instead, he said, rarely makes it into the pre-loss assumption. Sites have to be cleared. Where there has been loss of life, that carries its own process. Brownfield reinstatement is a different exercise to building on clear ground. Meanwhile customers make other arrangements and supply chains reroute – so the interruption propagates well beyond the site that was modelled.
His direction to intermediaries was explicit. "You've got to think far, far broader than that," he said, urging brokers and risk managers to allow for workforce disruption and for a regulatory environment that may shift after a loss, and to build buffers for both.
He added: "It's just far more complicated than divide revenue by 12 times whatever."
The depth of any of this analysis, Hunter said, comes down to who is doing it. "So much of it comes down to the experience of the consultant doing the work," he said – knowing how far down the rabbit hole to go, depending on the scenario.
For brokers, that reframes a due diligence question. When a client presents a BI declaration supported by a modelling exercise, the question is not only what the model produced but whether anyone involved in producing it has stood on a site after an event and watched a recovery run late.
