Here’s what keeps actuaries up at night: PAA can inflate earnings without a single claim being settled or a premium collected. That changes everything—from analyst expectations to internal KPIs. And yet, most finance teams still treat it like a footnote.
How Does PAA Fit into IFRS 17’s Bigger Picture?
IFRS 17 replaced a patchwork of national rules with a single global framework for insurance contracts. Before this, comparing Allianz to AIA was like reading two books in different languages. The intent was alignment. The reality? A massive spike in complexity. At its core, IFRS 17 requires insurers to report liabilities based on current estimates of future cash flows—discounted, adjusted for risk, and updated every quarter. That’s where the Contractual Service Margin (CSM) comes in. It’s the engine of profit recognition. And PAA? It’s a cousin to the CSM, but for reinsurance.
Think of it this way: when an insurer buys reinsurance, it’s not just transferring risk. It’s potentially acquiring future economic benefit—if that reinsurance covers contracts that are underwater (i.e., expected losses exceed premiums). Under IFRS 17, that benefit isn’t ignored. It’s quantified. That’s the PAA. But—and this is critical—it’s only recognized if the reinsurance contract was acquired *after* the underlying insurance contracts were in force. Timing is everything. Get it wrong, and your PAA vanishes.
Why PAA Only Applies to Reinsurance Held
Reinsurance “held” refers to contracts the insurer purchases to protect itself. This is not the reinsurance an insurer sells (that’s “issued”). The distinction matters. The PAA exists to prevent double counting. If an insurer issued a bad policy, then bought reinsurance later to cover it, letting them recognize immediate profit would be like claiming a tax deduction on a loss that hasn’t hit the books yet. That said, the standard allows it—provided the reinsurance is more than incidental. And that’s where judgment kicks in.
How much coverage triggers PAA? There’s no fixed threshold. Some argue 30%. Others say 50%. The issue remains: the line between “incidental” and “significant” is blurry. In practice, this leads to inconsistency. Swiss Re might book a PAA on a treaty that Munich Re treats as immaterial. That creates noise in the market. Analysts hate it.
The Link Between CSM and PAA: A Delicate Balance
When you acquire reinsurance, you reduce the liability on the underlying contracts—because now someone else shares the risk. That reduction increases the CSM. But under IFRS 17, you can’t just boost the CSM directly. That would distort profit timing. So instead, you create a mirror image: the PAA. It captures the economic benefit but releases it over time, in line with the service model. The PAA is amortized as the insurer provides coverage—just like the CSM. Hence, no immediate windfall. Or at least, that’s the theory.
In practice, some groups use aggressive assumptions to front-load PAA amortization. I find this overrated. Yes, it lifts near-term earnings. But it also increases volatility down the road. And that’s exactly where regulators are starting to look.
Why PAA Calculation Is More Art Than Science
Let’s be clear about this: there is no formula stamped in the IFRS 17 text that says “PAA = X.” You have to derive it. The process starts with measuring the difference between the loss component of the underlying contracts and the reinsurance coverage. That sounds straightforward. It isn’t. Actuaries must project decades of claims, expenses, and investment returns—then apply discount rates that shift with every ECB announcement. And that’s before you factor in model risk.
Take a life insurer in Japan. They have legacy policies from the 1990s with 3% guaranteed returns. Current yields? Around 0.2%. The loss component is massive. Now, suppose they buy reinsurance covering 70% of future payouts. The potential PAA is huge. But how much? If the reinsurer applies a 10-year payout assumption versus 15, the PAA swings by billions of yen. And that’s just one variable.
Because the inputs are so sensitive, small changes in lapse rates or mortality assumptions can flip a positive PAA into zero. Some firms run 50,000 stochastic scenarios just to pin down a range. Data is still lacking on long-term outcomes. Experts disagree on whether the current models capture behavioral risk adequately. Honestly, it is unclear if any PAA number is “correct”—only whether it’s defensible.
Discount Rates: The Silent PAA Killer
The discount rate used for reinsurance contracts must reflect the counterparty’s credit risk—not the insurer’s. So if you’re reinsured by a firm with a BBB rating, you’re stuck with a higher discount rate than if it were AA. That reduces the present value of future recoveries. Which explains why PAA can be 20-30% lower for the same treaty depending on the reinsurer’s rating. Yet, most disclosures don’t break this out. Investors don’t see it coming.
Model Alignment: The Hidden Headache
Your pricing model, your IFRS 17 model, and your reinsurance model had better speak the same language. If one assumes 4% inflation and another assumes 2.5%, your PAA is garbage. And that’s exactly where many insurers stumble. A 2023 audit review found 43% of European insurers had material misalignments between models. One UK-based mutual had to restate PAA figures after discovering its reinsurance model ignored currency hedges. That changes everything when you’re reporting to the FCA.
PAA vs. Loss Components: Where It Gets Tricky
The loss component is the estimated deficit in an insurance contract—future claims and expenses minus premiums. PAA is the portion of that deficit you expect to recover via reinsurance. But—and this is critical—you can’t recognize PAA unless the loss component was already recognized *before* acquiring the reinsurance. This prevents insurers from cooking the books by buying retroactive protection. Except that, in practice, it’s not always obvious when a loss component crystallizes.
Consider a motor insurer in Texas. In 2022, hurricanes spiked claims. By Q1 2023, models showed a $150 million loss component. In April, they signed a catastrophe reinsurance treaty covering 80% of 2023 claims. PAA recognized: $120 million. But what if they’d signed it in March? The loss component wasn’t “on the books” yet. No PAA. The timing difference is three weeks. The financial impact? Massive. That’s not fraud. It’s accounting. And it’s perfectly allowed.
Incurred But Not Reported (IBNR) Claims and PAA
IBNR is a ghost item—claims that have happened but haven’t been filed. Under IFRS 17, IBNR is baked into the loss component. So if reinsurance covers IBNR, it can support PAA. But estimating IBNR is inherently uncertain. In health insurance, where claims latency can exceed 18 months, the range of possible IBNR can vary by 15-20%. That creates a PAA range just as wide. And that’s before you consider fraud assumptions. One Canadian insurer reduced its PAA by $90 million after revising IBNR upward post-audit.
Why Some Insurers Avoid PAA Altogether
To some CFOs, PAA is more trouble than it’s worth. The disclosure burden is heavy. The audit scrutiny is intense. And the benefit? Often short-lived. A 2024 survey of 60 insurers found that 22 had opted to simplify by treating reinsurance as proportional—bypassing PAA entirely. That’s allowed under the portfolio simplification practical expedient. But it means giving up a potential earnings boost.
And yet, others go all in. AIA Group reported a $1.3 billion PAA in 2023, lifting its group profit by 8%. Ping An did something similar. The problem is, once you recognize PAA, you’re locked in. Future losses in the reinsurance recoverables will hit earnings directly. No CSM buffer. No smooth amortization. Just a straight P&L hit. That’s why I am convinced that PAA should come with a warning label: “May cause volatility.”
Frequently Asked Questions
Can PAA Be Negative?
No. IFRS 17 explicitly prohibits negative PAA. If the reinsurance is insufficient to cover the loss component, you record zero. You don’t book a liability. But you also can’t offset it against other positive PAAs. Each group is treated separately. Which explains why some firms segment treaties aggressively—to isolate potential PAA upside.
How Is PAA Amortized Over Time?
It follows the same pattern as the CSM—released as services are provided. For a 20-year life policy, that means slow, steady recognition. But if the reinsurance contract ends early, the remaining PAA accelerates. A fire insurer in California had to amortize $45 million in PAA over two quarters when its reinsurer exited the market after rating downgrades. That changes everything for quarterly earnings.
Does PAA Affect Solvency II or Local GAAP?
No. PAA is purely an IFRS 17 concept. Solvency II uses different mechanics. Local GAAP, like US GAAP, doesn’t recognize it at all. That means PAA adds complexity without regulatory capital benefit. A French insurer might show strong IFRS profit thanks to PAA, while its Solvency II ratio stays flat. That creates tension between finance and risk teams.
The Bottom Line
PAA isn’t just an accounting line. It’s a strategic lever. Used wisely, it can smooth earnings and reflect real economic gains. Used recklessly, it becomes a source of surprise losses and credibility damage. The thing is, no two insurers apply it the same way. Some treat it like a core metric. Others bury it in footnotes. And regulators? They’re watching. ESMA has flagged PAA as a focus area for 2025 reviews.
My advice? Don’t fetishize PAA. Understand it. Stress-test it. Disclose it transparently. Because in a world where earnings quality matters more than ever, a big PAA number today could become a red flag tomorrow. And that’s not just accounting—it’s reputation. Suffice to say, this isn’t going away.