Pensions Are a Promise, Not a Number

Every year or two, a report gets published estimating the "unfunded pension liability" of some national system, usually expressed as a number large enough to be genuinely alarming out of context several times the country's annual GDP, occasionally more. The number circulates, gets treated as evidence of an impending crisis, and then quietly disappears from the conversation until the next report revives it. I've come to think this entire genre of reporting, while not exactly wrong on the arithmetic, is deeply misleading about what a pension system actually is, and the misunderstanding has real costs.

What that big scary number actually represents

An unfunded pension liability figure is, at its core, the present value of every future pension payment the system has promised, minus the present value of the assets currently set aside to pay for it. For a pay-as-you-go system, the kind most public pension systems actually are, that gap is by design enormous, because the system was never built as a giant investment account meant to accumulate assets equal to every future promise. It was built as an intergenerational transfer mechanism, where today's workers' contributions pay today's retirees' pensions, with the expectation that tomorrow's workers will do the same for the retirees of that era.

Judging a pay-as-you-go pension system by comparing its assets to its total future liabilities is a bit like judging a functioning insurance company by asking whether it holds enough cash today to pay every claim its policyholders might ever file over the rest of their lives. No insurance company operates that way, and no one would want them to it would mean an enormous, wastefully idle pile of capital sitting around instead of being productively deployed, in exchange for a guarantee nobody actually needs, because the insurer is confident premiums will keep arriving to cover claims as they come due. A pension system's real question isn't whether it holds full funding today. It's whether the flow of contributions is likely to keep pace with the flow of benefit payments, now and over a reasonable planning horizon.

The demographic question that actually matters

That reframing points toward the question that genuinely deserves attention, which is demographic, not actuarial: what's happening to the ratio of contributing workers to receiving retirees, and how is that ratio likely to evolve. This ratio, not the headline unfunded liability figure, is the number that tells you whether a pay-as-you-go system is under real strain.

Norway, like most of Europe, has watched this ratio decline steadily as birth rates have fallen and life expectancy has risen fewer workers supporting each retiree, and each retiree drawing benefits for longer than the system's original design assumed. This is a genuine structural challenge, and I don't want to minimize it by focusing too much on debunking the scarier framing. But it's a fundamentally different kind of challenge than the "unfunded liability" framing suggests, because it's a problem with several plausible, gradual, well-understood policy responses adjusting the retirement age in line with rising life expectancy, adjusting contribution rates gradually rather than through crisis-driven emergency reform, adjusting benefit formulas at the margin rather than a looming cliff edge requiring some dramatic, singular fix.

Why the framing itself causes damage

Here's what concerns me about the recurring "unfunded liability crisis" framing beyond its technical inaccuracy: it actively undermines the thing pension systems most need in order to function well, which is confidence. A pension promise only works as intended if workers currently paying into the system believe, with reasonable confidence, that the same system will still be there in a form recognizably similar to what they were promised, decades from now when they retire.

Every alarming headline about trillion-krone unfunded liabilities chips away at that confidence, and the behavioral consequences of that erosion are real, even if they're harder to quantify than the liability figure itself. Workers who don't trust the state pension system save less through it and more through private, often less efficient, individually managed alternatives alternatives that carry higher fees, less effective risk pooling across a population, and considerably worse outcomes on average for anyone who isn't unusually financially sophisticated or unusually lucky with market timing. Younger workers, hearing repeatedly that the system they're funding may not be there for them, become more receptive to political arguments for dismantling collective, risk-pooled pension provision altogether in favor of individual accounts not because the economic case for that shift is strong, but because trust in the collective alternative has been worn down by a metric that was never designed to be read the way it typically gets reported.

What a more honest pension conversation would look like

I'd like to see reporting on pension sustainability lead with the dependency ratio and its projected trajectory, rather than the unfunded liability headline, because the dependency ratio is the number that actually predicts strain, and because it comes with a much clearer, more actionable menu of gradual policy responses attached to it. I'd like to see more explicit public communication about the difference between a pay-as-you-go system and a funded one, because I don't think most people currently paying into a state pension system fully understand which kind they're actually part of, and that gap in understanding is precisely what makes the scarier, less accurate framing land so effectively.

None of this is an argument that pension systems face no real challenges they do, and Norway's own system has required, and will continue to require, real adjustment as the demographic picture shifts. But a promise under manageable strain, with well-understood levers available to address that strain, is a fundamentally different situation from a promise that's broken, and conflating the two doesn't make the underlying policy problem any easier to solve. It just makes the public conversation around it considerably harder to have honestly, at precisely the moment honest conversation is what the problem actually needs.