$785 billion is bidding against itself at the top of the tower

The reinsurance industry raised a record amount of capital this cycle, pointed most of it at risks that were already funded, and now stands at $785 billion, per PWD's records, while the cat bond market had a record year of its own. Over the same stretch of business, 84% of US earthquake exposure carried no coverage at all, and Moody's sized the protection gap at $700 billion. The money did what money does at the top of a tower: it made the top of the tower cheap.

Cheap is worth something. A cedent buying top-of-tower property catastrophe cover is buying tail protection it could not otherwise hold, and a soft market at that layer is real money to the buyer, which is why January's conversations will open on price even when they close on terms. What cheaper top-of-tower cover cannot do is reach a risk nobody has modelled, priced, or sold: the uninsured earthquake exposure, the perils outside the footprints the market underwrites to, the losses that take two years to measure and therefore two years to pay.

Fitch's renewal survey is the cleanest statement of what the industry expects next: 86 per cent on terms against 60 on price, and the same work has reinsurers expecting to sell property cat for less in January while attaching less protection to it. Those two expectations are consistent with each other, and the second should concern a cedent more than the first, because a discount bought with a narrower insuring clause gets paid for out of the buyer's own balance sheet.

That is why the protection gap has proved so resistant to cheap capital: rate on line can fall for years without a single uncovered building getting covered, because most of the gap sits outside the perimeter the reinsurance market prices at all. An uninsured building is not uncovered because the price is too high; for most of them, no price exists. Data, take-up and measurement bind long before rate does.

The record cat bond year is the clearest evidence that this cycle's new capital is being spent on risks that were already bankable: cat bonds are sponsored by insurers and reinsurers laying off peak-zone property risk, so record issuance deepens the market at attachment points that were already clearing. Nothing about that depth creates a first-time buyer in a county with no loss history and no schedule of insured values, and deepening a market is a different project from widening one.

Two additions to capacity now sit in front of cedents, neither pointed at the exposure that has no coverage, and AM Best has assigned Charp Re a Good rating, which puts a new reinsurance name in front of buyers without obliging anyone to buy from it. What matters is what Charp Re writes at, because a new book in a soft market gets built by being the cheapest pen at the table, and a Good rating is a licence to quote rather than a franchise. Whatever the talk of capital formation, the agencies decide which of it reaches cedents: no rating, no cedent, no business.

Arcadian is the more revealing of the two. Multiyear capacity has been committed behind a casualty line that is still being established, which is how sponsor-backed money is entering insurance in this cycle: fund the team first, and the paper it will write second. That commits capital against a hiring pattern rather than a loss history, says more about the route sponsor capital is taking than the hire does, and puts the sponsor's clock well ahead of the book that will have to justify it.

Both, though, are entering lines whose pricing other people's capital has already cleared: Charp Re will sell cover into a reinsurance market where that cover is getting cheaper, and Arcadian is building a casualty book whose economics will be settled as much in the terms as in the rate. Neither is underwriting the earthquake exposure without coverage, and neither moves the 84%.

Earthquake is the sharpest illustration, because it is the peril the market has the least excuse to ignore: it is modelled, securitised, and most of the US exposure still carries no policy. What the gap is missing is a submitter—schedules, ownership, and a buyer who thinks of the exposure as insurable in the first place. If the best-covered peril in the industry's toolkit leaves that much risk bare, the perils with thinner models should not expect a record year of capital to find them on its own.

Scale was never the obstacle. The capital-markets pool runs 280 times the property-casualty pool, which is one way of saying the search for insurance risk has never been limited by the supply of dollars. What is scarce is a trigger a cedent will hold and a regulator will accept: a parametric structure that pays on a measurement, a model the market trusts in a geography where loss data is thin, a casualty line that can be underwritten on its own terms. A lower rate on line produces none of them.

The cat bond market's own limit makes the point: Swiss Re's $300 billion Florida tail figure is larger than the entire cat bond market, so even in the best-modelled and most liquid peril in the world the capital markets absorb a layer of the eventual loss and reinsurance balance sheets carry the rest. Bermuda's $220 billion of disaster exposure sits in the same place, concentrated in the perils that have cleared for a decade. The resilience report's own data caveat points to parametric cover as the practical route into the exposures the models handle badly, which is also the route the market has been slowest to take.