# Insurance premia: cat bonds and merger arbitrage Every diversifier reviewed so far wins, when it wins, against another market player. Trend rides the moves that someone else is stuck on the wrong side of ([[managed-futures]]). Long volatility buys the convexity that someone else is selling ([[long-volatility]]). Discretionary macro bets against the consensus ([[global-macro]]). One family gets far less attention, and in it the counterparty is not a speculator. It is a professional trying to get a risk off his books, because it is too heavy for him to carry alone. A Florida insurer cannot absorb the bill for a major hurricane by itself. A shareholder in a company being acquired would rather have a sure price today than a check in eighteen months that depends on the deal closing. In both cases somebody pays a premium to sleep at night, and that premium is your return. Two markets put this family within reach of a fund buyer: **cat bonds**, which securitize natural catastrophe risk, and **merger arbitrage**, which buys companies under acquisition to collect the gap between their price and the price promised. This article takes both mechanisms apart, prices what they have really delivered, measures what they do to a retirement portfolio, and ends on the only question that matters, which is how much to own. ::: cle You are on the other side of the counter The return on these strategies is an insurance premium, and their loss is a claim. The claim has a physical or a legal cause, a hurricane or a competition authority's veto, never an economic one. So the low correlation with stocks does not come out of a statistical window that could close: it comes from the nature of the risk itself. The corollary always disappoints, so here it is plainly. An uncorrelated asset protects you from nothing. It does not rise when stocks fall, it simply lives its own life elsewhere, which is useful but is not a hedge ([[defensive-assets]]). ::: ## The cat bond: a reinsurance contract cut into securities An insurer, more often a reinsurer, wants cover against a rare but ruinous event: the hurricane that crosses Miami, the earthquake under Tokyo. Instead of buying that protection from a peer, it issues a security. Investors pay in the principal, which goes straight into money-market instruments and sits there as collateral. In exchange they collect, every quarter, the short rate of the collateral currency plus a **spread**, and that spread is the price of the protection. If the catastrophe written into the contract happens, the collateral does its job: the principal is cut, in part or in full, and pays the claims. If it does not, the investor gets the money back at maturity, usually after three years. The loss, though, does not heal: a bond cut is cut for good, and what "recovers" after a bad season is not the burned paper, it is the wider spreads on the next vintage of issues. The contract names a **trigger**. It can be indemnity based (indexed to the issuing insurer's own losses), parametric (indexed to a physical measurement, wind speed or the magnitude of a quake) or industry loss based (indexed to the losses of the whole profession). The market has concentrated on a handful of well modeled perils: US hurricane first, then California and Japan earthquake, European windstorm, flood and wildfire. The market has been running since 1997, and it has just changed scale. Outstanding issuance ended 2025 a little above USD 61 billion, up 24% in a year. This is no longer a curiosity. It is a listed asset class, with its own indexes, its own specialist managers and, since recently, an ETF. **The track record, as published.** The reference index, the Swiss Re Global Cat Bond, shows an average return of about 6.7% a year over twenty-three years, with a single negative year, 2022, the year of Hurricane Ian. The last three are dizzying: 19.7% in 2023, 17.3% in 2024, 11.4% in 2025. Those numbers are accurate, and they are misleading, for two reasons the next section takes apart. They are dollar returns, inflated by a US money-market rate back up at 5%. And they date from after Ian, when spreads blew out and insurers paid top dollar for their cover. Nothing promises the next buyer that regime. ::: figure sinistres-calendrier Twelve years of real monthly returns in euros, cat bonds on top, world stocks below, on the same scale. The top panel holds two accidents: September 2017 (Irma and Maria) and September 2022 (Ian, which the inflation of 2022 deepens in a real series). September 2017 was a good month for stocks. Out of 146 months, only two show both panels losing together: March 2020, marked in red, when forced sellers took everything down at once, and September 2022, by pure coincidence, Ian on one side and the rate shock on the other. ::: ## What the premium really pays This is the passage the sales material leaves out, and it changes the verdict. ::: attention The collateral currency trap A cat bond's principal sleeps in money-market instruments, in one currency. The total return is therefore that currency's short rate plus the spread. Two things follow. The cash leg is a large share of any headline number, so an average struck across an era of high short rates says much less about the premium than it seems to. And a share class hedged into another currency does exactly what it promises: it swaps the collateral's short rate for yours. You keep the spread and you give up the difference between the two rates. Before you read a track record, check which currency the collateral sits in and what the hedge costs. ::: Three other costs finish the bill. **Fees** first: about 1.3% a year on the vehicles you can actually buy, a quarter to a third of a normal spread. Over fifteen years, one defensive fund charging 1.75% delivered about 1.9% a year: in this asset class, picking the share class is not a detail, it is half the decision. **Model risk** next: a cat bond is priced off event probabilities computed by three specialist firms, and the climate is moving faster than their models. **Liquidity** last, decent in normal times and gone in March 2020, like everything else that trades over the counter. History adds a plumbing warning: in 2008, cat bond funds fell with everything else, with no hurricane in sight, because the collateral behind some bonds sat in structures signed by Lehman Brothers. The market fixed it, principal now sleeps in government paper or the like, but the episode is a reminder that causal decorrelation does not cover the pipework. Strip out the cash leg and the fees, and what is left of a 6.7% headline is a spread of a few points over cash, in a generous year for spreads. That is still worth owning. It is not what the brochure shows. The premium itself, though, rests on solid ground. It is one of the few on the alternative menu whose origin needs no econometrics: somebody has to shed this risk, there is not enough capital to carry it, so he pays. The question is never whether the premium exists. It is what it is worth today, and what survives the fees. ## Merger arbitrage: a return paid by the legal calendar The mechanism fits in three sentences. A company announces it is buying another at $50 a share; the next day the target trades at $48, not $50. The gap of $2 pays for two uncertainties, the wait until closing and the risk that the deal breaks. The arbitrageur buys the target, shorts the acquirer or an index to neutralize the market, and pockets the gap when the deal completes. The payoff is again the payoff of an insurance seller: small regular gains, set by a legal calendar rather than by the business cycle, and rare brutal losses when a deal falls apart. Two properties follow, and a retiree should care about both. The spread is quoted over the short rate, so the premium rises with rates. This is enhanced cash, not a substitute for stocks. And market sensitivity is low by construction, since the long leg is hedged. The other side shows up when deals break. Failures do not arrive at random, they arrive in waves: when credit shuts and financing falls through, or when a competition authority hardens its doctrine, as US arbitrageurs learned the hard way between 2021 and 2023. Correlation to stocks, low on average, therefore climbs in exactly the episodes that worry you. **In practice, fees decide everything.** A merger arbitrage fund commonly charges 1.8% of management plus a 20% performance fee. On a gross premium of 3 to 4 points over cash, not much is left, and what is left is still taxed. An index ETF that tracks announced deals costs a fifth as much, but it buys every deal without screening any of them, including the ones whose spread is wide because the market is right to doubt ([[building-it-with-us-etfs]]). ## What these blocks really do to a retirement portfolio This part is missing from almost everything published on the subject, and it is the part that decides. An uncorrelated block is never good on its own. It is good, or not, **instead of something else**, and the choice of that something else settles the verdict. The exercise starts from a retirement portfolio that is already diversified, of the kind [[all-weather-portfolios]] describes: stocks, duration, trend, gold, inflation-linked bonds. Add ten points of one of the two blocks, over a window common to every variant, and read the sustainable withdrawal rate at 5% failure over forty years. | Variant | Real return | Volatility | Sustainable rate | |---|---|---|---| | Starting portfolio | 6.4% | 7.2% | 4.94% | | +10 points of cat bonds, funded pro rata across every holding | 5.9% | 6.6% | 4.70% | | +10 points of arbitrage, funded pro rata across every holding | 6.0% | 6.8% | 4.74% | | +10 points of cat bonds, taken from the bond sleeve | 6.6% | 6.9% | 5.11% | | +10 points of arbitrage, taken from the bond sleeve | 6.7% | 7.1% | 5.16% | The reading is clean. Funding the block by shaving every holding, stocks included, **degrades** the plan: you trade return for carry. Funding the same block out of the bond sleeve improves it, by roughly two tenths of a point of withdrawal rate. The general rule sits in that contrast: enhanced cash is measured against cash, never against stocks. Three caveats travel with the table. The common window stops at twelve years, for lack of deeper cat bond history, so neither 2008 nor the 1970s is in it. And no public fund history starts before 2006: Katrina, the worst modern season, sits outside every backtest of the asset class, a structural optimistic bias. That window is also the worst of the century for long bonds, which mechanically flatters any idea built on replacing them. And take two points of annual return off the block, to stand in for a stingier spread regime, and the edge falls to zero, without turning negative. The most useful result is elsewhere, in what the household lives. Replayed under a flexible spending rule, the same comparison moves the bottom quartile of the spending delivered by less than 0.4%. These blocks do not change a retiree's life. They shift the nature of the risk and file a little off the volatility. It is an engineer's adjustment, not a transformation ([[flexibility-in-practice]] says the same thing about withdrawal rules). ::: exemple Five points of cat bonds, taken from the right place A household holds 13 points of inflation-linked bonds and 4 points of long government debt. It moves 5 points into a cat bond fund. What it buys: an expected return of 3 to 5 points over cash while spreads are wide, with no interest-rate risk, and hurricane risk in its place. What it gives up: the convexity of long duration on the day of a deflationary crash, because cat bonds do not rise in a crash, they ignore it. What it costs: 1.3% of fees on the holding, and a taxable brokerage account, since these funds are not on the menu of ordinary retirement wrappers. And the sentence to write down before buying, the one that has to hold on the day of the bad season: "this holding can lose 10% in a month with no market falling at all, and I will not sell it for that". ::: ## Sizing, and the conditions to check first Four conditions, in order. - **Your base defense is already in place.** These premia replace neither duration, nor trend, nor gold, each of which answers one precise regime ([[defensive-assets]]). They come after, or not at all. - **The funding comes from the short sleeve.** Cash, short bonds, maybe part of the inflation-linked bonds. Never stocks, never trend, or the arithmetic turns against you. - **The price of risk is good today.** For cat bonds, read it off the spread on offer and the short rate of the collateral currency, both public. The cycle works for the patient buyer: spreads widen after the bad seasons, never before, and the best vintages followed 2017 and 2022. For arbitrage, off the average spread on live deals. Selling insurance at a discount is a bad deal, uncorrelated or not. - **The vehicle is clean.** Total fees under 1.5%, assets large enough to keep the fund from closing, and a share class in your own currency unless you want the exchange rate to supply three quarters of the variance. The first cat bond ETF, launched at the end of 2025, is worth watching, but it is still small and has no track record. The deep share classes trade at net asset value through fund platforms, rarely on an ordinary broker's menu: in this family, access is a constraint to check before anything else. Reasonable sizing runs from 0 to 10 points, taken from the short sleeve. Zero is a perfectly defensible answer for a simple portfolio, exactly as it is for long volatility. Ten points is a ceiling: a bad catastrophe season can cost the holding 10% to 20%, and you have to be able to ride that out without touching the plan ([[the-psychology-of-spending]]). The market's concentration argues for that ceiling too: about two thirds of the total risk sits in two perils, US hurricane and US earthquake, and a single major landfall on Miami, a perfectly plausible event, can take out on the order of a fifth of an entire fund. ## The essentials - Cat bonds and merger arbitrage get paid to carry someone else's risk: the hurricane, or the deal that may not close. Their low correlation with stocks is causal, not statistical. But an uncorrelated asset protects you from nothing. - The cat bond market is worth USD 61 billion and its index shows 6.7% a year over twenty-three years, with a single negative year. Most of that headline is the cash leg. The premium itself is the spread, and fees of about 1.3% take a quarter to a third of it. - Merger arbitrage is enhanced cash: the premium rises with rates, and the risk, which is regulatory, arrives in waves. Fund fees often absorb half of it. - In a retirement portfolio that is already diversified, the funding decides everything. Taken from the bond sleeve, ten points add about 0.2 point of sustainable withdrawal rate; taken pro rata across every holding, they take as much away. - The effect on the standard of living is marginal, under 0.4% on the bottom quartile of spending. Size it at 0 to 10 points, funded from the short sleeve, with a vehicle under 1.5% of fees and a share class in your own currency. --- ## Going further - Artemis (artemis.bm): the daily chronicle of the cat bond market, with outstanding issuance, new deals and losses, free to read. - Swiss Re, Global Cat Bond Index: the reference series since 2002, and its methodology. - Morningstar, "Catastrophe Bonds as Portfolio Diversifiers": the balanced review, upside and limits, for a reader new to the subject. - In this book: [[defensive-assets]] (the spec a defensive block has to meet), [[global-macro]] (the catalog of the other alternative premia), [[enhanced-cash]] (the other way to put the short sleeve to work), [[designing-a-portfolio]] (where these points come from).