The float

You are running the insurer now.

The underwriting result has been negative for nine consecutive years. Every year the company has paid out more in claims and costs than it took in premiums. Every year. The board has just approved a dividend.

Before you read on, commit to an answer.

Predict

An insurer pays out 104 in claims and expenses for every 100 of premium it takes, every year, for a decade. What happens to it?

Choose one answer

It runs down its reserves and fails within a few yearsIt survives by borrowing, which gets more expensive each yearIt can be one of the most profitable companies in the countryIt stays roughly at breakeven, with investment income offsetting the loss

Lock in answer

The money in between

A premium arrives in January. The claim it eventually pays might be settled in March, or in four years, or — for an injury claim being argued over — in eleven.

Between those two dates, the insurer is holding money that is not its own. It will have to be paid out. In the meantime, the insurer decides where it sits.

Turn the investment side on

This is the same model as the last lesson with one more control released. The underwriting is now set to lose money: a combined ratio of 104, a four percent loss on every premium, every year, for ten years.

The return on the float starts at zero. Move it.

Insurance float

≈ Hypothetical model, not a forecast

Premiums in, claims and expenses out, and the money held in between put to work.

It assumes

• The underwriting loses four percent of premiums every year for ten years.
• Claims take three years to settle, so the float is three times annual premiums.
• The investment return is earned on the whole float, every year, without a loss.
• Nothing catastrophic happens in any of the ten years.

It ignores

• Reserve development, where yesterday's claims turn out to cost more
• Reinsurance, which changes both the claims and the float
• Any year in which the investments themselves lose money
• Capital requirements, and the cost of holding that capital
• Catastrophe years, which is when insurers actually fail

Find the return on the float at which ten years of unbroken underwriting losses become a profit overall. Read the number carefully before you decide whether to be surprised.

Return on the float0%

Combined ratio

10,400%

Premiums in the final year

1,304,773

Underwriting result, final year

-52,191

Float held, final year

3,914,320

Investment income, final year

0

Total result, final year

-52,191

Underwriting result over the period

-458,555

Investment income over the period

0

Total result over the period

-458,555

What the float costs, as a rate

133.3%

Return the float must earn to break even

133.3%

• Underwriting result
• Investment income
• Cumulative total

◷ Not there yet — keep moving the controls.

One point four percent.

That is the whole thing. A company losing money on every policy it writes, in every one of ten consecutive years, turns a profit if the money it is holding in the meantime earns more than one and a half percent.

Look at the figure the model labels what the float costs. It is 1.33 percent — the four-point underwriting loss divided by a float three times the size of annual premiums. That is this insurer's cost of borrowing. It is available in unlimited quantity, it has no maturity date, and no committee has to approve the renewal.

Check

What is the float, on the insurer's own balance sheet?

Choose one answer

An asset — it is money the company is holdingA liability — it is money owed to policyholders who have not claimed yetNeither — it is a timing difference and nets to zeroEquity, since the shareholders control what it is invested in

Check answer

The operator's chair

Now the interesting part, and the reason this lesson is written from your side of the desk.

If the float is worth more than the underwriting costs, the underwriting is not the point. It is a means of gathering float. And two levers change how much float a given amount of premium produces.

The first is how much you are willing to lose on the insurance itself. Price more cheaply, accept a worse loss ratio, write more business.

The second is subtler and much more powerful: write the kind of insurance where claims take a long time to settle. Motor damage settles in weeks. A liability claim argued through courts settles in years. The same premium held for five years instead of one is five times the float.

The return below is fixed at five percent. You control the loss ratio and how long the claims take.

Insurance float

≈ Hypothetical model, not a forecast

Premiums in, claims and expenses out, and the money held in between put to work.

It assumes

• The float earns five percent every year, in every year, with no losing year.
• The float multiple is a choice about what business to write, not a dial to turn.
• Claims and expenses stay steady shares of premiums however the pricing changes.
• Writing more business at a worse price is assumed to be possible, which it is.

It ignores

• Reserve development, where yesterday's claims turn out to cost more
• Reinsurance, which changes both the claims and the float
• Any year in which the investments themselves lose money
• Capital requirements, and the cost of holding that capital
• Catastrophe years, which is when insurers actually fail

Build the position: the underwriting must lose more than 100,000 in the final year, and the company must still be profitable over the decade. Then look at which of the two controls did the work.

Claims, as a share of premiums7,400%

Float held, in years of premiums0.5

How long claims take to settle decides how much money is held in the meantime.

Combined ratio

10,400%

Premiums in the final year

1,304,773

Underwriting result, final year

-52,191

Float held, final year

652,387

Investment income, final year

32,619

Total result, final year

-19,572

Underwriting result over the period

-458,555

Investment income over the period

286,597

Total result over the period

-171,958

What the float costs, as a rate

800%

Return the float must earn to break even

800%

• Underwriting result
• Investment income
• Total result

◷ Not there yet — keep moving the controls.

The loss ratio makes the hole. The float multiple decides whether the hole matters. At half a year of float, a four-point underwriting loss is fatal within a decade. At three years of float, a twelve-point loss is comfortably profitable — and the second insurer looks far worse than the first on every figure a customer or a journalist would ever see.

Now turn it around

Float is not an insurance invention. It is what happens whenever a business is paid before it delivers, and it is sitting in your own spending.

Check

Which of these is NOT running on float?

Choose one answer

A gift card scheme, where cards are bought in December and spent over the following two yearsA software company offering two months free if you pay for the year up frontA builder taking a forty percent deposit before starting workA restaurant taking payment when the meal ends

Check answer

The part that keeps it honest

Three things stop this being the free money it can sound like.

The float is owed. A bad catastrophe year calls it in early, and it is called in at exactly the moment investment markets are usually down, because the things that damage economies damage both sides at once.

Leverage works downwards. Holding three times your premium income in investments means a ten percent investment loss costs you thirty percent of a year's revenue, on top of whatever the underwriting did.

Competition gives most of it away. Insurers know the float is valuable, so they compete for it by cutting premiums, which is precisely how the industry arrives at a combined ratio above 100 in the first place. The reason your cover costs less than the expected claims plus expenses is that somebody wants to hold your money in the meantime.

That last one is worth sitting with. The mechanism this lesson has spent thirteen minutes on is also the reason insurance is cheaper than it would otherwise be.

Who pays, who benefits, who carries the risk

Who pays. Policyholders, in the gap between paying today and being paid later — a gap they receive no interest on, and which is priced into the premium only through competition.

Who benefits. The insurer, on the investment return over the cost of the float. And the policyholder, through a premium lower than the claims and expenses alone would justify.

Who carries the risk. Split, and this is the part that matters. The insurer carries the investment risk. The policyholder carries the risk that the insurer loses on both sides at once and cannot pay — which is what capital requirements and guarantee schemes exist to cover, and why an insurer's balance sheet is regulated far more tightly than its prices.

Worth remembering

• Float is the money an insurer holds between collecting a premium and paying the claim, and it is a liability, not profit.
• An underwriting loss is the cost of that float: divide the loss as a share of premiums by how many years of premiums are held.
• In the worked example, a four percent underwriting loss against three years of float costs about 1.3 percent a year, so any return above roughly 1.4 percent makes the whole operation profitable.
• Writing insurance where claims settle slowly produces far more float from the same premium, which is why long-tail insurers can run above a combined ratio of 100 indefinitely.
• Any business paid before it delivers runs on float — gift cards, annual subscriptions, deposits, season tickets.