What Is a Pension? Moving Money From Working Life to the End of It

The fundamental distinction is about who bears the risk. A defined benefit scheme promises a specified income in retirement, usually calculated from salary and years of service, and the employer or the state carries the obligation regardless of investment performance or how long people live. A defined contribution scheme specifies only what goes in, invests it, and hands over whatever the pot has become, so the member carries the investment risk, the inflation risk and the risk of living longer than the money lasts. The shift from the first to the second across most of the private sector since the 1980s is the single largest change in retirement provision in modern history, and it happened because defined benefit promises proved far more expensive than sponsors had assumed once longevity rose, investment returns disappointed and accounting rules forced the liabilities onto company balance sheets.

A second distinction concerns where the money comes from at the moment it is paid:

• •Funded schemes accumulate assets in advance, so contributions are invested and the investments pay the pensions, which is how almost all private provision works
• •Pay as you go schemes pay current pensions out of current contributions or taxes, with no fund at all, which is how most state pensions work including the British and American systems
• •Pay as you go depends entirely on the ratio of workers to pensioners, which is why demographic change matters so much and why the systems are politically fragile
• •Funded schemes depend on investment returns and on there being something worth buying, which is why they are exposed to markets rather than to demography
• •Notional accounts, used in Sweden and several other countries, are pay as you go but record individual balances that grow with wage growth, making the arithmetic visible and adjusting automatically to demography
• •Multi-pillar systems combine a state safety net, a mandatory or occupational funded layer and voluntary saving, which is the arrangement most international bodies recommend

Three forces make pension promises harder to keep than they look. Longevity has risen substantially and, more importantly, was consistently underestimated, so schemes designed around a pension lasting fifteen years found themselves paying for twenty-five. Fertility has fallen across the developed world, reducing the ratio of workers to pensioners from around five to one in the mid-twentieth century towards two to one or lower in several countries by the 2050s. And interest rates fell for three decades, which sounds unrelated and is central: a scheme's liability is the amount of money that, invested at the prevailing safe rate, would fund the future payments, so a lower discount rate makes the same promises much more expensive today. Together these turned schemes that appeared comfortably funded in 1990 into schemes with enormous deficits by 2010, which is why so many closed.

The defined contribution model requires individuals to make several decisions that people reliably make badly: whether to join, how much to contribute, how to invest, and how to draw down a pot without exhausting it or dying with it untouched. The consistent findings are that people under-save relative to their own stated intentions, that they default to whatever the default is, and that present bias makes the trade-off feel worse than it is. The policy response, drawn directly from behavioural economics, has been to change the defaults rather than to exhort: automatic enrolment, in which employees are placed in a scheme and must actively opt out, raised participation dramatically wherever it has been introduced, most visibly in Britain after 2012 where participation among eligible private sector employees rose from around forty percent to close to ninety. Escalation schemes that raise contributions automatically with each pay rise address the second problem, and default investment funds that shift towards lower-risk assets as retirement approaches address the third. The drawdown phase remains the least solved, since converting a pot into an income that lasts an unknown lifetime is genuinely hard, and annuities, which do exactly that, are unpopular because they feel like losing the money.

State pension reform is the most reliably explosive subject in democratic politics, because the costs are visible and fall on identifiable groups while the benefits are diffuse and in the future. Raising the retirement age is the most effective single lever and has triggered large protests in France, Russia and elsewhere; it also has a distributional problem, since life expectancy differs by many years between occupational groups, so a uniform increase asks manual workers to give up a larger share of their retirement. Indexation rules matter enormously in aggregate and are barely noticed individually, since the difference between uprating pensions by prices, by earnings or by whichever is higher compounds into very large sums over decades. And the intergenerational question sits underneath all of it, since a pay as you go system is a promise made by one generation that the next will honour, backed by nothing except the expectation that they will in turn want the same deal.

Pensions move consumption from working life to retirement, and whatever the arrangement, retired people are supported by what workers produce at the time. Defined benefit schemes promise an income and put the risk on the sponsor, defined contribution schemes promise only the contributions and put it on the member, and the shift between them since the 1980s was driven by rising longevity, falling interest rates and accounting rules. Automatic enrolment fixed participation by changing the default, and converting a pot into a lifelong income remains the unsolved part.