Why a Good-Weather Number Isn't Good Enough:

The Case for Probability-Based Pension Communication

In an earlier post in this series, Why PPIs Can't Afford to Leave Participant Understanding to Employers, we argued that PPIs face a structural risk in the Wtp transition: the traditional model of routing pension communication through employers is breaking down, and the providers who own participant understanding directly will be better positioned commercially and regulatorily than those who don't.

In today’s blog, we explore how even when PPIs communicate directly with participants, the wrong methodology produces the wrong outcomes. The industry has defaulted to a communication model built around optimistic projections, and regulators, academics, and increasingly participants themselves are starting to notice. The question isn't just whether to communicate more but whether the numbers being communicated make sense.

The number that looked too good

Imagine a 32-year-old participant receives her transition statement this year. She's moved from a collective DB scheme to a flexible DC arrangement under the Wtp. On the statement, three numbers are shown: her expected pension, her bad-weather pension, and her good-weather pension.

The good-weather number is substantially larger than her previous DB projection: in some cases, according to the AFM, a few times larger for young participants in premium schemes. There is no contextual explanation, so she focuses on the largest figure and forms an expectation.

The AFM flagged this problem explicitly in its Sector in Beeld Pensioenen 2025 - the first time it published data on the spread between good-weather and bad-weather scenario amounts across the Dutch pension market. Its finding: for younger participants in premium schemes (DC arrangements), the gap between the two can be very large. And without contextual explanation, this creates unrealistic expectations.

This is a feature of how DC projections work, and AFM is not criticising the methodology but the communication. There's a critical difference between those two things, and it is where most Dutch pension providers are currently getting it wrong.

What a good-weather number actually is

Under the Uniforme Rekenmethodiek (URM), the standardised calculation methodology that all Dutch pension administrators must use, scenario amounts are derived from a quarterly scenario set published by DNB. The set currently consists of 10,000 stochastic economic scenarios, modelling future interest rates, inflation, and equity returns across a range of possible economic futures.

From these 10,000 simulations, three figures are extracted for participant communication: the 5th percentile (bad-weather), the 50th percentile (expected), and the 95th percentile (good-weather). These three numbers are what participants see on their UPO and on Mijnpensioenoverzicht.nl.

The problem is not the model but the presentation. Running 10,000 simulations and then showing only the 5th, 50th, and 95th percentile outputs to participants discards almost all the probabilistic information that was generated. The participant receives three isolated data points: not a distribution, not a range, not any indication of the relative likelihood of each scenario.

PGGM, one of the Netherlands' largest pension administrators, has published their own analysis of this phenomenon. Their finding: the bad-weather scenario for a typical DC participant already deviates rapidly from the expected outcome, and for certain types of young participants, the deviation can reach 50%. The spread, they note, is large enough that the exact precision of any single figure matters far less than communicating the honest shape of the uncertainty.

This is the core of the good-weather problem: a participant who anchors to the 95th percentile outcome, the largest number, prominently displayed, is forming a pension expectation based on a scenario that, by definition, only materialises in the best 5% of all modelled economic futures.

The median isn't the solution eit Drafther

Showing the median outcome is better than showing only the good-weather figure. It is more honest. But it is still not enough, and understanding why is important for any PPI that is serious about participant communication.

A median figure tells a participant: "there is a 50% probability that your pension will be at least this amount." That is a meaningful statement. But it still gives no indication of what happens below the median, and for retirement planning, the shape of downside risk matters enormously.

The asymmetry of retirement risk is fundamental. A participant who retires into a bad-weather scenario does not get a second chance to rebuild their pension capital. Unlike an accumulating investor who can hold through a downturn and wait for recovery, a retiree drawing down in a poor market faces sequence-of-returns risk that can permanently impair their income. The bad-weather scenario is not a theoretical curiosity: it is the scenario that, if it materialises, produces the outcomes participants most need to understand and prepare for.

Netspar research on visualising pension uncertainty confirms that participants who receive the standard three-scenario "navigation metaphor" focus primarily on the expected amount and treat the range almost as a formality. The good-weather and bad-weather figures are acknowledged, but their probabilistic significance, the fact that one is a 95th percentile outcome and the other is a 5th percentile outcome, is not readily understood from presentation alone.

The navigation metaphor was a genuine improvement on what existed before. But it was designed for a DB world where the dominant communication task was showing that future pension income was somewhat uncertain - not for a DC world where participants bear investment risk personally and need to understand the full probability landscape of their outcomes.

What probability-based forecasting looks like

The answer is not more numbers. It is a fundamentally different way of presenting what the underlying stochastic model already knows.

Rather than extracting three-point estimates from 10,000 simulations and discarding the rest, probability-based forecasting presents the full distribution of outcomes in a form participants can understand and act on. The underlying methodology, stochastic Monte Carlo simulation, is already mandated by DNB and used by every administrator in the Netherlands. The question is what to do with the results.

In a well-designed probability-based communication experience, a participant can understand:

• The range of plausible pension outcomes, expressed as a probability distribution rather than three isolated points

• Their probability of meeting a target income - for example, "there is a 72% chance of achieving at least €1,800 per month in today's money"

• How that probability changes when they adjust their contribution rate, change their investment profile, or modify their planned retirement date

• The gap between their current trajectory and a personally meaningful retirement goal - not just a regulatory benchmark

This is the difference between a statement that reassures and a tool that informs. The former shows three numbers and hopes the participant draws the right conclusions. The latter puts the participant in a position to make a genuinely informed decision about their own financial future.

The weather forecasting analogy is instructive here. A meteorologist does not say "it will rain 12mm on Thursday." They say: "there is a 70% chance of rain above 5mm." The forecast is probabilistic because the underlying model is probabilistic, and presenting it otherwise would be misleading. Pension forecasting should work the same way: the model already knows the distribution and the communication should reflect it.

The regulatory tailwind is clear.

The AFM's direction of travel on this question could not be clearer. In its October 2024 research report covering 62 pension funds, the AFM found that 37 of them had shortcomings in balanced communication: the most common failure mode was presenting positive effects prominently in first information layers while burying risks and negative consequences in deeper layers. Information about negative outcomes was reaching participants far less reliably than information about opportunities.

In its November 2025 review of transition statements, the AFM found that contextual explanations of the difference between projected pensions in old and new schemes were "often still missing or too generic" - a finding the regulator explicitly linked to the risk of unrealistic participant expectations.

And in a July 2025 Transitiebulletin, the AFM specifically recommended that administrators address communication scenarios involving pension declines, noting that setting realistic expectations about possible future reductions is an important part of adequate communication planning.

This is consistent with Netspar's academic guidance on the same question. Their research on the explainability of the new pension system concludes directly: "Do not promise golden mountains for the new system; this will only lead to unrealistic expectations and therefore disappointment." Communication must set realistic expectations, not optimistic ones.

The direction of regulatory travel points clearly away from good-weather anchoring and toward communication that honestly represents the probability landscape of DC outcomes. PPIs that get ahead of this, that build probability-based communication into their participant experience now, will be better prepared for AFM scrutiny, more credible in employer tender processes, and more trusted by participants when markets disappoint.

What good looks like

The goal is not a PDF with better-labelled numbers. It is a participant experience that treats the underlying stochastic model as an asset rather than an engine for generating three compliance figures.

In practice, this means a participant portal that does not simply display a balance and a contribution rate but functions as a genuine planning environment. One where a participant can see their probability distribution of outcomes, explore what-if scenarios with meaningful visualisation of how their choices shift that distribution, and understand in plain terms the gap between their current trajectory and a retirement income that meets their needs.

This is precisely the context in which Kidbrooke's simulation and forecasting infrastructure is designed to operate. KidbrookeONE's stochastic engine runs the kind of probability-based analysis described above, generating participant-level outcome distributions that can be presented interactively, integrated into existing PPI portals, and used to support the kind of personalised, probability-literate communication the AFM's four communication principles point toward. For PPIs exploring this architecture, our documentation provides a detailed overview of the relevant forecasting, gap analysis, and scenario simulation capabilities.

The good-weather number made a kind of sense in a world where pension administrators wanted to reassure participants that the system would provide for them. In a DC world - where participants bear investment risk directly, where outcomes vary enormously across individuals and economic scenarios, and where regulators are actively monitoring the quality of uncertainty communication - that reassurance without honesty is not just inadequate, it may become a liability.

SOURCES:

All sources cited are primary Dutch regulatory, practitioner, or academic sources.

→ AFM — Sector in Beeld Pensioenen 2025

→ AFM — Scenariobedragen (supervisory guidance page)

→ AFM — Transitiecommunicatie: zorg voor maatwerk voor jouw deelnemers (April 2024)

→ AFM — Deelnemers hebben recht op het volledige verhaal (October 2024)

→ AFM — Prognose-transitieoverzichten: verbetering zichtbaar, maar aandachtspunten blijven (November 2025)

→ AFM — Nieuwe aandachtspunten communicatieplan (July 2025)

→ AFM — Vijf vragen over scenariobedragen (November 2023)

→ DNB — Definitieve scenariosets bij Wet toekomst pensioenen

→ DNB — Rekenmethodieken voor weergave van ouderdomspensioen in scenario's

→ PGGM — Tienduizend scenario's voor UPO-berekening onnodig

→ PGGM — Een betere toekomst voor scenariobedragen

→ Netspar — Hoe kunnen we de onzekerheid rond de verwachte pensioenuitkering visualiseren? (2025)

→ Netspar — Wijzigingen in de uniforme scenarioset (November 2025)

→ Netspar — Uitlegbaarheid van het nieuwe pensioenstelsel (April 2024)

→ Netspar — Pensioencommunicatie dossier

→ TKP Pensioen — Bedragen in transitieoverzicht lopen sterk uiteen: zo legt u het uit

→ Financial Investigator — Peter van der Nat: Vreemde rentesprongen in scenariosets DNB (May 2024)

→ Tweede Kamer — Initiatiefnota 36 775, vergaderjaar 2024-2025