One fork in the road: a promise, or a pot
In the previous guide you met the annuity — the only instrument that turns a lump of savings into income that lasts exactly as long as you do — and the longevity risk that makes it hard to price. A pension plan is simply that machinery built at scale, by an employer, for a whole workforce. But before any actuary can value one, the plan's designers must make a single decision that shapes everything afterwards: does the employer promise *how much income you will receive*, or only *how much money it will put in on your behalf*? That fork has two names — defined benefit and defined contribution — and almost every retirement system on earth is one, the other, or a blend of the two.
The cleanest way to hold the difference is a single dial: *who carries the risk that the future turns out worse than assumed?* In a defined-benefit plan the employer turns that dial fully toward itself — it owes you a stated income no matter how markets behave or how long you live. In a defined-contribution plan it turns the dial the other way — it owes only the agreed deposits, and from there the outcome is yours. This is the same risk-bearing dial you met in life products, where whole life keeps the risk with the insurer and variable life hands it to the policyholder. Pensions turn the very same dial, on the two risks that matter most for retirement: investment risk and longevity risk.
Defined benefit: the employer promises an income
A defined-benefit (DB) plan promises a pension defined by a formula, not by an account. The classic shape multiplies three things: an accrual rate, your years of service, and a measure of your pay. A typical final-salary formula reads *1.5% × years of service × final average salary*. Work thirty years, finish on a salary of 60,000, and you have earned an annual pension of 1.5% × 30 × 60,000 = 27,000 — paid every year for the rest of your life, and often a reduced amount to your spouse after you, through a joint-and-survivor form. Notice what is *not* in that formula: nothing about how the investments performed. The number is a promise.
DB benefit formula (final-salary type):
annual pension = accrual rate x years of service x final avg salary
= 1.5% x 30 x 60,000
= 27,000 per year, for life
Benefit accrual (how the promise is earned, one year at a time):
after 1 yr : 1.5% x 1 x salary
after 10 yrs: 1.5% x 10 x salary <- earned 'so far'
after 30 yrs: 1.5% x 30 x salary <- full promise at retirementThat promise is earned gradually, and the technical word for the build-up is benefit accrual. Each year of service adds another slice — another 1.5% of pay, in our example — to the pension you have locked in. The slice you build *this* year is the seed of what an actuary will later call the normal cost, and the value of all the slices earned *so far* is the accrued liability the plan must already hold money against. You will value those properly in the next guides; for now hold the picture: a DB pension is a stack of yearly slices that grows until you retire, then pays out for as long as you live.
Because the benefit is fixed, the *risk* lands squarely on the employer (the plan sponsor). If the plan's investments underperform, the sponsor must top up the fund from its own coffers. If retirees as a group live longer than the mortality assumptions predicted, the sponsor pays the extra years. The worker is shielded from both: market crashes and rising life expectancy are the sponsor's problem, not theirs. This is the quiet generosity — and the quiet danger — of DB. A promise made when a firm is young and growing can become a crushing weight decades later, which is exactly how so many corporate and public plans drifted into being underfunded.
Defined contribution: the employer fills a pot
A defined-contribution (DC) plan flips the promise. Here what is *defined* is the input, not the output: the employer agrees to pay a stated contribution — say 5% of your salary each year, often matched by your own 5% — into an individual account with your name on it. That money is invested, usually in funds you choose, and it grows (or shrinks) with the markets. At retirement you do not receive a promised income; you receive whatever the account happens to be worth. The familiar 401(k) plan in the United States and most modern workplace pensions worldwide are DC. The employer's obligation ends the moment the contribution is deposited.
Now run the risk dial the other way. The market's behaviour is *yours*: a strong decade swells your pot, a crash just before retirement can gut it. And here is the subtler trap — even after you reach retirement with a healthy balance, you still face longevity risk personally, because a pot of money does not know how long you will live. Spend it as though you will die at 80 and live to 95, and you outlive your savings; hoard it for fear of that, and you live needlessly poor and die rich. A DC account hands you, the individual, the very problem an annuity was invented to solve — and most people never buy the annuity that would solve it.
Vesting: when the promise truly becomes yours
There is a catch that applies to both designs: an employer's contribution or accrual is not automatically yours to keep if you leave early. Vesting is the rule that decides when the employer-funded portion becomes a non-forfeitable right. Until you are vested, walking out the door can mean walking away from the employer's money. Two common patterns exist. *Cliff vesting* gives you nothing of the employer's share until a threshold — say three years of service — at which point the whole amount becomes yours at once. *Graded vesting* hands it over in slices, perhaps 20% per year, so that you are fully vested after five or six years.
It is worth keeping vesting and accrual distinct, because beginners blur them. *Accrual* is how much benefit you have *earned*; *vesting* is how much of what you have earned you get to *keep* if you leave. You can have accrued a sizeable benefit yet be entitled to none of the employer's part because you are not yet vested. Your own contributions to a DC account, by contrast, are almost always immediately and fully yours — it is the employer's match and the DB promise that carry vesting strings. For the actuary, vesting matters because *withdrawal rates* — the probability a worker leaves before vesting, forfeiting benefits — directly reduce the plan's expected cost.
The real goal: the replacement ratio
Step back and ask what a pension is actually *for*. Not to accumulate the biggest pile — to replace your working income smoothly enough that retirement does not feel like a cliff. The yardstick for this is the **replacement ratio**: your retirement income as a fraction of your pre-retirement pay. If you earned 60,000 and your pension pays 27,000 a year, your replacement ratio from the plan is 45%. Designers usually aim for a *total* replacement ratio — counting state pension, workplace plan, and personal savings together — somewhere around 60–80%, because retirees no longer save, no longer commute, and often pay less tax, so they need less than 100% to keep the same standard of living.
Here the two designs diverge in a revealing way. A DB plan *targets the replacement ratio directly* — the formula is engineered so that, say, 1.5% per year over a full career lands you near 45% from the plan, a number you can read off before you ever retire. A DC plan, by contrast, delivers a *balance*, and the replacement ratio falls out only at the end, after you convert that balance into income — and that conversion depends on the annuity price on the day you retire, which in turn depends on interest rates and on how long insurers expect you to live. Two people with identical DC balances who retire two years apart can end up with materially different incomes, purely because rates moved.
This is why the DB-versus-DC choice is the deepest structural decision in pensions, and why the world has drifted heavily from DB toward DC over the last forty years. Employers, weary of carrying open-ended market and longevity risk on their balance sheets, handed it to individuals. That shift is not free: it has quietly transferred a vast amount of risk onto households who are, on the whole, poorly equipped to manage it. Everything you will study next — the accrued liability, the funding methods, the discount-rate assumption that swings the whole valuation — exists chiefly because *DB promises must be valued and funded today for cash flows decades away*. A DC plan needs almost none of that machinery, because it makes no promise to value.