The pension did not disappear because workers preferred 401(k)s. It disappeared because a 1978 tax provision written for executive deferred compensation turned out to be cheaper and less risky for employers than a defined benefit plan, and because the accounting and funding rules that followed made pensions expensive to keep. The shift moved investment risk, longevity risk and the contribution decision itself from the employer’s balance sheet to the worker’s. That transfer, not any change in investing technology, is the whole story.
What a pension actually promised
A defined benefit pension promised an outcome. The plan committed to pay a formula amount for life, usually based on years of service and final average pay. The employer had to fund that promise, invest the assets, and make up any shortfall if the investments underperformed or retirees lived longer than the actuaries assumed.
Three risks sat with the employer under that structure. Investment risk, because a bad decade in markets was the employer’s problem to close. Longevity risk, because the payment continued until death regardless of how long that took. And contribution risk, because the funding obligation existed whether or not the company had a good year.
A defined contribution account promises an input instead. The employer contributes a defined amount, or matches a defined share, and the obligation ends there. What the account is worth at 65 is whatever the market and the worker’s own contributions produced.
The legal sequence
Two pieces of federal law bracket the change.
The Employee Retirement Income Security Act of 1974 set minimum funding standards, vesting rules, fiduciary duties and reporting requirements for private pension plans, and created the Pension Benefit Guaranty Corporation to insure them. It was passed to protect workers after a series of plan failures left people with nothing. It worked, and it also made sponsoring a pension materially more expensive and more legally exposed.
The Revenue Act of 1978 added section 401(k) to the tax code, permitting employees to defer part of their salary into a qualified plan on a pre-tax basis. The provision was drafted with executive compensation in mind. Benefits consultants read it as a general-purpose savings vehicle, the Treasury issued proposed rules in the early 1980s allowing salary reduction contributions, and the plan type spread through large employers within a decade.
Later accounting changes compounded the pressure by making pension obligations and their funding gaps visible on corporate financial statements, where they moved with interest rates and market returns. An obligation that shows up in reported earnings is an obligation a public company works to shed.
Why employers preferred the new structure
Cost predictability explains most of it. A 401(k) match is a known percentage of payroll. A pension obligation is an actuarial estimate that moves with discount rates, asset returns and mortality assumptions, and it can move badly at exactly the moment a company can least afford it.
Portability helped the transition look like a gain for workers, and for some it was. Pensions rewarded long tenure and punished job changes, since benefit formulas weighted final salary and vesting schedules ran for years. An account balance travels. In a labor market where people change employers regularly, that is a real advantage.
What the portability argument leaves out is that the two features were traded together. Workers gained an account they could take with them and lost a benefit somebody else was required to fund.
What the transfer produced
The Federal Reserve’s 2022 Survey of Consumer Finances found that 54.3 percent of U.S. families held a retirement account of any kind, up from 50.5 percent in 2019, with a median holding of $86,900 among those who had one. Broken out by age, families aged 55 to 64 with a retirement account held a median of $185,000, and 57.0 percent of families in that bracket held any retirement account at all.
The Federal Reserve’s Report on the Economic Well-Being of U.S. Households in 2025 found that 35 percent of non-retirees thought their retirement saving was on track, a share unchanged from the prior year.
Those figures describe a system where participation is optional, contribution levels are a household decision, and the outcome depends on a surplus that many households do not have. A defined benefit plan produced a benefit whether or not the worker understood investing or had cash left at the end of the month. A defined contribution plan produces a benefit only if both conditions hold.
The part the design did not anticipate
Voluntary savings systems assume a household surplus to save out of. That assumption held better when the plan type spread than it does now.
The U.S. Census Bureau put median household income at roughly $80,000 in 2023. Against that, KFF put the total annual premium for employer family health coverage at about $25,000 in 2024, with workers paying more than $6,000 of it directly. Child Care Aware reports center-based childcare commonly costing $10,000 to $17,000 or more per child per year. Median home sale prices ran $400,000 to $420,000 in 2024 according to National Association of Realtors and Census data, roughly five times median household income against about three times in the 1980s.
A worker facing those costs and a voluntary savings plan does not make an investing mistake. The money is spoken for before the contribution election comes up. Automatic enrollment, which many plans adopted after Congress encouraged it in 2006, raised participation precisely because it stopped requiring the worker to find the surplus consciously. It did not create the surplus.
What the change did not do
It did not reduce the cost of retiring. A person leaving the workforce at 67 still needs roughly the same income stream regardless of which plan type funded it. The obligation was reassigned, not retired.
It did not transfer the expertise along with the risk. Pension assets are managed by institutional investors under fiduciary standards. Defined contribution assets are allocated by individuals, most of whom have no training and no access to the same instruments. Target date funds narrowed that gap. They did not close it.
And it did not make the underlying outcome more certain. It made it more variable. Some workers do better under a defined contribution plan than they would have under a pension, particularly those who change jobs often or who work for employers whose pensions were poorly funded. Others do far worse. Dispersion went up, which is what happens when you replace a promise with a market outcome.
Reading the shift honestly
Nostalgia for pensions overstates how widely they were held. Coverage was concentrated in large firms, unionized industries and the public sector, and many workers in the pension era retired on Social Security alone. The comparison worth making is not between today and a universal pension that never existed. It is between two ways of allocating the same risk, and the direction the allocation moved.
Groups working on household economics have started treating this as a question about costs rather than about investing. Fight For A Living Wage, a nonpartisan grassroots 501(c)(3), puts retirement saving in the same category as housing and healthcare, and makes the point that saving is the only one of the three a household can postpone. That is why it absorbs the pressure first. The Federal Reserve balance sheet data fits that reading.
The pension era ended for reasons that were rational from the employer side and legible in the tax code. What followed put the burden of a lifetime income stream on the household surplus at exactly the point in history when that surplus started disappearing. Those two facts are usually discussed separately. They belong together.