From a private number to a public one
You arrive at this guide carrying a whole toolkit. You can discount a stream of future claims to today, set a policy reserve, run a profit test, and price in a margin for uncertainty. So far almost all of that work has produced *private* numbers — figures an actuary computes to manage a book of business. This guide is about the moment those numbers go *public*: when they must appear on an audited financial statement that an investor in London, a regulator in Singapore, and an analyst in Toronto can all read and compare. That demand for comparability is the entire reason IFRS 17 exists.
Before 2023, comparing two insurers across borders was like comparing recipes written in different languages and units — almost every jurisdiction accounted for insurance contracts its own way, so the word *profit* meant something subtly different everywhere. The previous guide drew the line between statutory and GAAP accounting: statutory rules ask "can this company pay claims if things go badly?", while general-purpose reporting asks "how is this company truly performing?". IFRS 17 is a general-purpose answer to that second question, written to be the same answer the world over. It does not change how much an insurer ultimately earns; it changes *when* and *how clearly* that earning is shown.
Building the liability one block at a time
How do you honestly put a value on a promise to pay uncertain amounts, at uncertain times, possibly decades from now? IFRS 17's answer for long-duration contracts — life cover, annuities — is the building-block approach, also called the general measurement model. Rather than guess one opaque lump sum, you assemble the liability out of clearly separated, separately disclosable pieces, so a reader can see exactly what drives a change rather than facing a single mysterious figure. This is just the discounting and margin-setting you already know, dressed in the discipline of public reporting.
There are three blocks plus a fourth element. Block one is the present value of the *best-estimate* future cash flows: premiums coming in, claims and expenses going out, discounted at current rates — exactly the discounting of liabilities you met earlier, with no hidden conservatism baked into the central estimate. Block two is the risk adjustment for non-financial risk, a deliberate margin for the uncertainty in those cash flows. Together blocks one and two are called the *fulfilment cash flows* — your honest best guess of what it will cost to fulfil the promise, plus a price for the uncertainty. The fourth element, layered on top, is the contractual service margin.
IFRS 17 building blocks (issue of a new annuity book)
Block 1 PV of best-estimate cash flows ....... 1,000
Block 2 Risk adjustment (non-financial risk) . 70
----------------------------------------------
Fulfilment cash flows ................ 1,070
+ CSM Contractual service margin ........... 130
----------------------------------------------
Insurance contract liability at issue 1,200
Day-one profit recognised ..................... 0
(the CSM absorbs the expected profit so none is booked up front)The risk adjustment deserves a careful word, because it is where your judgement about genuine riskiness becomes a hard balance-sheet number. It covers *only non-financial* risk — mortality, lapse, expense — never interest-rate risk, which is already handled inside the discounting. Think of it as the extra you would pay, above the pure expected cost, to be rid of the uncertainty itself. IFRS 17 does not mandate one method (a cost-of-capital calculation or a value-at-risk confidence level are both common), but it does demand you disclose the confidence level your number corresponds to — so a reader can tell that one insurer is reserving to a stronger percentile than another. As risk runs off over time, the risk adjustment is released into profit.
The contractual service margin: profit kept on a leash
Here is the idea that makes IFRS 17 feel different from anything you have priced so far. When a gym sells a year-long membership upfront, it would be misleading to declare the whole year's profit on day one — the member has not received eleven of the twelve months yet. Insurance is the same, only longer and far more uncertain. The contractual service margin (CSM) is IFRS 17's device for *storing* the expected profit on a group of contracts and dripping it into earnings only as coverage is actually provided. At inception the CSM is set to exactly cancel the day-one profit — so a profitable contract reports *zero* gain when written. The profit you computed in your profit signature is real, but it is not allowed to land all at once.
Each period a slice of the CSM is released to profit, in proportion to the *coverage units* — a measure of how much service was provided that period. A CSM of 120 on a four-year contract with level coverage releases about 30 of profit a year; spread a CSM of 240 evenly over six years and you book 40 a year. The CSM also acts as a *shock absorber*: when estimates of future cash flows change for reasons relating to *future* service, you adjust the CSM up or down instead of jolting profit. That smoothing has a crucial, honest limit, though — it is not symmetric.
On a life insurer's IFRS 17 balance sheet the CSM is often the single largest, most-scrutinized line, and tracking it through every assumption change demands serious actuarial machinery. There are lighter routes for special cases — a simplified *premium allocation approach* for short-duration contracts (much like the unearned-premium accounting you saw in property-casualty), and a *variable fee approach* for participating business, where policyholders share in investment returns. But the building-block model with its CSM is the conceptual heart, and the place where your craft as an actuary becomes the most visible number a public company reports.
Embedded value: the worth a balance sheet won't show
Even a flawless IFRS 17 balance sheet stays oddly silent about a life insurer's biggest asset: the long stream of future profit locked inside policies *already* on the books. The statement shows today's assets and liabilities, but barely hints at the decades of earnings those in-force policies will quietly throw off — earnings IFRS 17 deliberately holds back inside the CSM. Embedded value is a separate, management-and-investor measure built to surface exactly that hidden worth: an estimate of what an insurer's existing business is really worth to its shareholders today.
Broadly, embedded value is the sum of two parts: the *net worth* (the shareholders' share of net assets today) plus the *value of in-force business* — the present value of future profits expected to emerge from existing policies — less a deduction for the cost of holding required capital. As a sketch, an insurer might report a net worth of 400 plus a value of in-force of 600, for an embedded value of 1,000. Its movement matters even more than its level: the *value of new business* written each year is a leading signal that this year's sales created economic value, often well before accounting profit catches up.
The hard part is valuing *uncertain* future profits, and that is where the variants come in — and where the word "market-consistent" finally earns its place in this guide's title. *European Embedded Value* (EEV) brought a consistent, disclosed methodology and an explicit allowance for the cost of the options and guarantees buried in policies. *Market-Consistent Embedded Value* (MCEV) goes further: instead of leaning on a single, possibly optimistic real-world investment return, it values the cash flows — and especially the financial options and guarantees — using market-based, risk-neutral techniques, so the answer agrees with what financial markets themselves would price. A guarantee that markets say is expensive is booked as expensive, not wished away by an upbeat return assumption.
How to read these numbers honestly
Hold all of this at arm's length, the way the previous rung taught you to hold a capital figure. The most common misconception about IFRS 17 is that it changes how much an insurer ultimately makes — it does not. The total economic result of a contract is fixed by the cash flows; IFRS 17 only governs the *timing and presentation* of that result, spreading it across the years of coverage instead of booking it up front. If you ever hear that a company "made more money under IFRS 17," reach for the cash flows, not the income statement.
And every one of these figures is a *model* output, dominated by assumptions you cannot fully verify. The discount curve, the risk adjustment's confidence level, the lapse and mortality assumptions, the choice of coverage units — each is a judgement, and small changes can move the headline number materially. Embedded value is the most assumption-driven of all: it usually *excludes* future new business, it is *not* an audited liability, and a shift in the investment-return or lapse assumption can swing it substantially. This is precisely why MCEV reports, like IFRS 17 disclosures, come wrapped in *sensitivity tables* — the figure means little without the band of uncertainty drawn around it.
Step back and the shape of this rung comes into focus. Reporting is the discipline that takes your private calculations and makes them *answerable* — comparable across firms, disclosed with their assumptions, defensible to an auditor and a regulator. The building blocks force you to separate honest expectation from margin from unearned profit; the CSM forces profit to follow the service that earns it; embedded value and MCEV force the future into a number consistent with how markets value risk. None of these is reality — each is a transparent, defensible story about reality. That is the whole point: in reporting, the actuary's job is not to produce the truth, but to produce a number you can *show your work* on, and stand behind.