The Pension Protection Act turned twenty this month. The anniversary is worth marking because what the law changed was not a set of plan features but the structure of retirement security itself, taking on two of the three risks that the death of pensions had dropped on American workers’ laps.

Defined benefit pensions spread across American industry in the 1940s and 1950s though the tax architecture that made them possible arrived earlier with the Revenue Act of 1921 and its extension to pension trusts five years later. Wartime wage controls did much of the rest. Employers barred from raising pay competed on benefits instead, and after the Inland Steel decision of 1948 made pensions a mandatory subject of bargaining, pensions became a structural feature of unionized manufacturing and trade.

Then, that foundation eroded from several directions at once. Private sector union membership fell from roughly a fifth of the workforce in the early 1980s into single digits and with it went the bargaining leverage that had won the pensions in the first place. Life expectancy kept climbing, so employers found themselves paying retirees longer than the original actuarial work had assumed. And accounting caught up. The Financial Accounting Standards Board began pulling pension obligations toward the balance sheet in 1985 and completed the job in 2006 when companies were required to report the funded status of their plans in full.¹ What had been a footnote became a line item and a volatile one. Sponsors froze and closed their plans.

Closing a plan does not extinguish the obligation, so employers that wanted out bought their way out. They purchased group annuity contracts from insurers, moving the future income payments and the administration that comes with them onto insurance companies’ balance sheets. It is worth noticing where the risk went. Even the companies most determined to be rid of the obligation did not hand it to their retirees. They transferred it to another institution capable of carrying it.

The replacement was already in the code. The Revenue Act of 1978 added Section 401(k), permitting employees to take part of their income as deferred compensation, and the first working plan appeared in 1980. Early adopters treated it as a perk, a way for employees to save something extra on top of the corporate pension. Employers then noticed what it made possible. A plan funded entirely by the company could give way to one funded by the employee with a modest match, as the cost of providing a pension was bearing down on them. Over the following two decades. the pension gave way to the account, and by March 2025, the Bureau of Labor Statistics found that 70% of private industry workers had access to a defined contribution plan while only 14% had access to a defined benefit plan.

What actually changed hands

The shift is usually described as a change in retirement vehicles. More precisely, it was a transfer of three distinct risks from the employer’s balance sheet to the individual worker. It dropped something less obvious as well. The pension had calculated all three for each worker by name, and nothing in the account that replaced it does that.

The first is funding risk. Under a defined benefit plan, the employer carries the obligation to set aside sufficient contributions to meet the promise. Actuaries calculate what the promise costs, the company writes the check, a shortfall is the company’s problem to solve, and the Pension Benefit Guaranty Corporation stands behind it in the last resort. Under a defined contribution plan, the worker decides how much to contribute which means the worker decides whether the plan will be adequately funded.

The second is investment risk. A pension trustee builds a diversified portfolio, rebalances it, monitors it, and absorbs the consequences when markets disappoint, because what the participant owns is a contractual retirement income promise rather than an account balance. Under a defined contribution plan, the worker selects the investments, designs the portfolio, and lives with the consequences of the retirement outcome.

The third is longevity risk. A defined benefit plan pays income to retirees until death, whenever that happens to be. A large employer pools thousands of lives and lets the averages do the work, which is the only way longevity risk can be managed. Many sponsors decline to carry the distribution obligation in perpetuity and transfer the liability to an insurance company by using group annuities that pay retirees their promised income for life. Either way, the longevity risk pooling happens. The employer carries it or an insurance carrier does, and in both cases, an institution is responsible. A defined contribution plan hands the worker a lump sum at retirement and asks them, in effect, to estimate how long they are going to live. A longer lifespan and they run out of money. A shorter lifespan and they spend their retirement needlessly poor.

The wrong party to hold them

The transition from defined benefit to defined contribution did not take into consideration the capacity of an average worker to manage these three risks.

Funding risk runs into present bias. A retirement contribution is a cost borne today for a benefit decades away. Human beings discount the future steeply. This served us well for most of our existence, but it serves us poorly in retirement savings. Money in hand is vivid. A retirement thirty years out is an abstraction, and it competes for attention against obligations that are neither distant nor abstract.

Investment risk runs into a competence problem. The assumptions sitting underneath the plan investment menus are that every participant has the time, knowledge, and acumen to design and maintain an investment portfolio. For most people, that has never been true.

Longevity risk needs no behavioral explanation at all. It is not a matter of bias or effort. An individual cannot pool their own mortality, and no amount of diligence changes that.

Three milestones

The federal response has come in three acts, roughly a generation apart, and each has taken on one part of the problem.

The Employee Retirement Income Security Act of 1974, known as ERISA, came first, and it did something more fundamental than any provision that followed. It established that someone must be accountable. The statute installed the fiduciary, a named person or committee bound by duties of loyalty and prudence, obligated to act for the exclusive benefit of participants and beneficiaries. Everything since has been built on that foundation because a system in which nobody bears responsibility for outcomes cannot be trusted.

ERISA was written to protect, not to encourage. Its concern was benefits already promised and money already set aside, and its duties were framed accordingly. That was the right design for a pension world, where the fiduciary, held to the highest standard, managed and safeguarded the assets and the participant bore none of the risk.

By 2005, the evidence was not in dispute. The work of Brigitte Madrian and Dennis Shea on automatic enrollment,² and of Richard Thaler and Shlomo Benartzi on committing future raises to savings³ showed that defaults and precommitment work and work at scale, but employers largely could not use them. Two obstacles stood in the way. State wage payment statutes, written decades earlier to stop coercive payroll deductions, barred an employer from moving money out of a paycheck without a signed instruction, and a company operating across twenty states faced a different answer in each of them. Plus, a sponsor who selected an investment for a participant who had never made an election was making a fiduciary decision and was fully exposed under the prudence standard, with no relief available.

The Pension Protection Act of 2006, the PPA, cleared both. A new preemption provision superseded state law to the extent it would prohibit or restrict an automatic contribution arrangement, conditioned on the employer giving participants advance notice of the deduction and of their right to stop it.⁴ This was not permission to garnish wages. The money never leaves the worker. It moves into an account the worker owns and can shut off at any time. What the provision removed was a requirement that a signature preceded the transfer, and what it substituted was advance notice and an unconditional right to decline.

What the Act built came down to three important provisions, and each recognizes something the transition had left unresolved.

First, automatic enrollment lets an employer include every eligible worker in the retirement plan, at no less than 3% of pay. That is the law recognizing that the hardest part of retirement saving is beginning the habit of regular saving.

Second, the qualified default investment alternative, known as the QDIA, gives the fiduciary protection for putting those contributions to work in a professionally managed multi asset portfolio rather than leaving them idle.⁵ That is the law recognizing that an average worker most likely does not have the skill to be a professional investment manager, and that behavioral biases are likely to affect portfolio performance negatively.

Third, automatic escalation raises the contribution over time, to as much as 15% under current law.⁶ That is the law recognizing that starting to save is not the same as saving sufficiently.

What the QDIA actually did

The Department’s final regulation, effective in December 2007, established the QDIA and did something unusual for a safe harbor. Rather than name products, it described a portfolio. A diversified mix of equity and fixed income, managed by a professional, without a guarantee of principal. Three forms qualified: a risk-based fund, such as a balanced fund, which holds an allocation appropriate to the participant population as a whole; a target date fund, known as a TDF, which allocates by reference to the age of the participant or an expected retirement date; and a managed account, in which a professional builds and maintains an allocation for the individual.

All three are multi asset portfolios, and all three do the same essential thing. They take the portfolio construction decision away from the participant and give it to a professional manager. The TDF became the dominant choice, holding $5.3 trillion as of the middle of 2026 according to Sway Research.

Whichever form a sponsor selects, the professional managing it applies the same discipline that pension trustees had been applying for decades. Diversify across asset classes, seek the most return available for each unit of risk taken, and rebalance as markets and the participant’s horizon move. That is modern portfolio theory, orthodoxy among institutional investors since the 1970s and almost entirely unavailable to individual participants before 2007. The QDIA did not invent it. It installed it as the resting state of the defined contribution account.

None of this is perfect, and none of it was meant to be. But taken together, these provisions addressed two of the three risks squarely. Automatic enrollment and escalation put money into the plan at a rate no education campaign had ever achieved, and the QDIA put that money into a portfolio built by someone qualified to build it.

The line the law did not cross

Deciding what happens to someone’s paycheck when they say nothing is a paternalistic act, and calling it choice architecture does not make the imposition disappear. The answer is that the imposition is real and the exit is unconditional. A default that cannot be declined is a mandate, and Congress pointedly declined to write one. Every worker keeps the right to contribute nothing, to contribute more, to choose different funds, or to leave entirely.

That distinction carries more weight than it is usually given, and it is worth being fair to the people the defaults are designed around. The behavioral literature is often read as a claim that human beings are irrational. It is not. Attention is scarce, and a parent who declines to spend a Saturday comparing bond funds is allocating limited attention sensibly. The default is not a correction applied to defective people. It is a courtesy extended to busy ones, and the fact that it can be overridden at will is what keeps it a courtesy. There are limits to what the architecture can carry, and workers enforce them. Push a default too high and people decline, cut back, or take the money out through loans and hardship withdrawals, which is the system reporting accurately that a deduction exceeded what a household could bear.

The scoreboard

The results are visible in the data, and they are not marginal. How America Saves, the annual study Vanguard published in its 25th edition in June, reports plan participation at a record 86% of eligible employees, up from 65% when the series began. The figures cover the plans Vanguard administers, which skew toward larger employers. Sixty one percent of those plans now use automatic enrollment, roughly double the share in 2013, and 69% of participants sit in a professionally managed allocation, the vast majority in a single TDF, against 9% at the end of 2005.

One figure deserves particular attention from anyone who followed the original debate. Critics warned that automatically enrolled workers would anchor at the low default rate and save less than volunteers. In 2013, that gap was nearly two percentage points, and the criticism had force. However, in 2025, the automatically enrolled deferred 7.7% and the voluntarily enrolled deferred 7.5%. The gap has closed and inverted because sponsors raised their defaults and layered on escalation. Including the employer match, which has itself risen to a record 4.7%, total savings rates reached 12.1%.

The asset numbers follow. Sway Research counts 169 TDF series with assets at the middle of 2026. Packaged mutual fund and collective trust series alone grew 21% in 2025 to $4.8 trillion, with custom strategies built for individual large plans adding roughly another $371 billion, a segment most published tallies leave out. Defined contribution plans held $13.8 trillion at the end of the first quarter, of which $9.9 trillion sat in 401(k) plans.

A worker hired in 2007 who never opened an enrollment packet, never selected a fund, and never rebalanced has been enrolled automatically, escalated automatically, diversified automatically, and moved toward safety automatically as retirement approached. That worker made no decisions. That was the point.

The third risk

Longevity was left standing, and Congress came back for it in 2019. The SECURE Act created a fiduciary safe harbor for selecting an insurer to provide guaranteed lifetime income inside a plan, giving sponsors a defined path to satisfying prudence when they choose an annuity provider.⁷ The diagnosis was the same one made in 2006. Employers were not declining to offer lifetime income because they doubted its value. They were declining because selecting an insurer whose obligations run forty years forward is an open-ended liability with no map. SECURE 2.0 followed by easing the constraints on qualified longevity annuity contracts, and the market has moved as well. Sway now tracks TDF series carrying embedded or optional income guarantees as a distinct category, and the largest of them finished its first year on the market with more than $16 billion in assets.

The circle closes

ERISA established that someone must answer for retirement outcomes, which was the precondition for everything after it. The actual management of the transferred risks began in 2006, when the PPA gave that person the legal cover to address funding and investment risk, and continued in 2019, when the SECURE Act extended the same instrument to longevity. Twenty years and two Acts later, three risks were brought back under the control of the employer fiduciary.

What has not been restored is personalization, and the names of the two plan types explain why. A defined benefit plan defines the benefit, and it defines it for each individual. The formula ran on that worker’s own years of service and that worker’s own pay, so a career employee retiring at 65 might receive 60% of pre-retirement pay for life. Sixty percent of his pay, not an average of everyone’s. Added to Social Security and its cost-of-living adjustment, that retiree was well provided for and knew the amount in advance.

A defined contribution plan defines only the contribution. Everything downstream of the deposit, how much accumulates, how it is invested, and how long it lasts, remains an uncertainty the worker carries. The machinery built since 2006 is powerful but uniform. A default of 3% or 6% applies to everyone. A glide path keyed to a birth year treats two workers of the same age identically, regardless of their individual savings, pay, other resources, or what they will actually need.

That is the honest measure of what twenty years accomplished. Automatic enrollment makes adequate funding far likelier without guaranteeing it. The QDIA makes the portfolio competent without making it specific. The annuity safe harbor makes lifetime income available without making it automatic. Three real answers, none of them a promise, and none of them yet aimed at a particular person.

This is where the next work lies, and personalization is the thread that connects all three. A savings rate is only adequate against a target, so the funding question becomes what rate this worker needs to reach a replacement rate that fits this worker’s circumstances. Portfolio risk follows from that same target, since the right level of risk is the level required to get there, not the level implied by a birth year. And the amount of guaranteed income needed in retirement is whatever remains after Social Security and the accumulated balance are counted, which is a different number for every household.

Personalize the outcome for each worker and the three risks stop being separate problems. They become one calculation performed for one person, which is what the pension did all along, and what the defined contribution system has never yet done.

None of this requires a fourth Act of Congress. The authority has been in place since 2007 because the managed account was one of the three qualifying forms from the beginning. What was missing was the ability to deliver it. Personalization was the most expensive option on the menu, and it repaid the cost only for participants who engaged with it, supplying their outside accounts, a spouse’s accounts’ balances, an intended retirement date, and a real sense of their own capacity for loss. A default that works only for people who are paying attention is not much of a default, which is why the managed account never became the resting state of anything.

Data abundance, better technology, and falling costs have changed that. Payroll and recordkeeping systems already know enough about a participant to calibrate a savings rate, a level of portfolio risk, and eventually an income need, without asking that participant for anything at all. It is the same bargain the QDIA struck in 2007, offered now at the level of the individual rather than the birth year.

ERISA established who is accountable. The PPA and the SECURE Act gave that person the tools. What remains is to point them at individuals rather than at populations, and for the first time, that is a practical thing to do.

Notes

1. Financial Accounting Standards Board Statement 87, issued in 1985, required employers to recognize a minimum pension liability and to measure pension cost on an accrual basis, though it smoothed asset returns and actuarial gains and losses over time. Statement 158, issued in 2006, removed that smoothing from the balance sheet and required a plan’s funded status to be reported in full.
2. Brigitte C. Madrian and Dennis F. Shea, The Power of Suggestion, Inertia in 401(k) Participation and Savings Behavior, Quarterly Journal of Economics, volume 116, issue 4, November 2001, pages 1149 to 1187. Available at https://doi.org/10.1162/003355301753265543. Circulated earlier as National Bureau of Economic Research Working Paper 7682, https://www.nber.org/papers/w7682.
3. Richard H. Thaler and Shlomo Benartzi, Save More Tomorrow, Using Behavioral Economics to Increase Employee Saving, Journal of Political Economy, volume 112, supplement 1, February 2004, pages S164 to S187. Available at https://www.journals.uchicago.edu/doi/10.1086/380085.
4. Section 902 of the Pension Protection Act of 2006 added Section 514(e) to ERISA, which supersedes any state law that would directly or indirectly prohibit or restrict an automatic contribution arrangement in a qualified plan. The relief is conditioned on advance notice to the participant of the deduction and of the right to stop it.
5. Section 624 of the Act added Section 404(c)(5) to ERISA. It deems a participant who gives no investment direction to have exercised control over the assets, provided they are invested in a QDIA. This differs from the participant direction relief of Section 404(c), which requires the participant to actually exercise control, and a plan need not satisfy Section 404(c) to rely on it. Selecting and monitoring the default remain fiduciary acts under either provision. The Labor Department’s implementing regulation was finalized in October 2007 and took effect that December.
6. Section 902 also created the qualified automatic contribution arrangement at Sections 401(k)(13) and 401(m)(12) of the Internal Revenue Code. A plan qualifies by defaulting participants at no less than 3% of pay, rising by one point a year to at least 6% and no more than 10%, and by providing either a matching contribution of 100% on the first 1% of pay and 50% on the next 5%, or a nonelective contribution of 3% of pay for every eligible participant. Employer contributions must vest within two years. The maximum default rate was later raised to 15% by the SECURE 2.0 Act of 2022.
7. Section 204 of the SECURE Act of 2019 added Section 404(e) to ERISA, establishing conditions under which a fiduciary is deemed to have acted prudently in selecting an insurer for a guaranteed lifetime income contract, principally a review of the insurer’s ability to make payments and written representations from the insurer regarding its licensing and financial condition.