A Federal Retirement Plan Sounds Simple. The Design Choices Are Anything But

The idea of a federal auto-enrollment retirement plan has broad, almost reflexive, appeal. Tens of millions of American workers lack access to an employer-sponsored retirement plan. Give them one and problem solved. The rhetorical simplicity is part of why proposals like the Retirement Savings for Americans Act and the Automatic IRA Act keep surfacing in Congress, and why the concept draws support from across the political spectrum.

But simplicity at the headline level masks serious complexity underneath. 

A federal retirement plan is not a single policy. It is a bundle of design choices (e.g., enrollment structure, default contribution rates, escalation provisions, withdrawal rules, matching incentives, portability mechanisms), each of which can materially alter the program's impact on the people it is supposed to help. Get the design wrong and you risk creating the appearance of progress without delivering the substance of it.

This is why rigorous, independent analysis matters, and why the Morningstar Center for Retirement & Policy Studies undertook the research published this month by Spencer Look and Jack VanDerhei, titled “Access, Auto-Enrollment, and Accumulation: A Simulation of Universal Retirement Plan Coverage.” 

The study uses the Morningstar Model of US Retirement Outcomes, a nationally representative, household-level simulation framework, to test what actually happens to retirement wealth under different federal plan configurations. 

Not in theory. Not under idealized assumptions. Under conditions that reflect how people actually behave: they opt out, they change jobs, they withdraw funds early, and they cash out balances when they leave employers.

The findings are both encouraging and cautionary, and they carry direct implications for policymakers weighing how to structure any future federal coverage initiative.

The most unambiguous result in the research is also the least surprising: auto-enrollment dramatically outperforms voluntary enrollment. At a 3% default savings rate, auto-enrollment produces average wealth gains of 28% for affected workers, more than double the 13% observed under voluntary enrollment. At a 6% default, the gap widens further (49% versus 13%).

This is not a marginal difference. It is the difference between a program that meaningfully shifts retirement trajectories and one that serves primarily as a political gesture. 

Behavioral economics literature has been telling us this for more than two decades, since Madrian and Shea’s foundational work in 2001. The contribution of the Center’s research is to quantify the effect at population scale, across generational cohorts, income levels, and demographic groups, using a model that accounts for the messy realities of American working life.

Anyone designing a federal plan who treats auto-enrollment as negotiable is, in effect, choosing to leave most of the program’s potential impact on the table.

Conventional wisdom holds that higher default contribution rates produce proportionally better outcomes, and the research bears this out: auto-enrollment at 6% does produce meaningfully better outcomes than auto-enrollment at 3%. But the more useful finding for policymakers may be that the initial default rate matters less than you'd expect, depending on what other design features are in place.

Auto-enrollment at 3% with automatic escalation to 6% produces results that are remarkably close to the flat 6% default (e.g., average wealth gains of 44% versus 49%). The similarity is driven largely by real-world adoption patterns. Higher default rates create higher opt-out risk, especially among the lower-income workers that this plan is designed to support. And the political economy of default rates is real; opt-outs can be a bigger threat to program effectiveness than modestly lower contribution rates among those who stay. An escalation design offers policymakers a practical middle ground. It allows workers to start at a rate low enough to minimize opt-outs, then let the automatic ramp-up do the work over time.

This finding matters because the political economy of default rates is real. Higher defaults create higher opt-out risk, and opt-outs are a bigger threat to program effectiveness than modestly lower contribution rates among those who stay. An escalation design offers policymakers a practical middle ground: capture most of the savings benefit while reducing the behavioral friction that could undermine participation.

The study also tests an enhanced version of the Saver’s Match program, which takes effect in 2027. The modeled enhancement restricts access to match funds until age 62, expands income eligibility, and doubles the match rate to 100%.

The distributional implications are striking. Under the expanded Saver’s Match scenario, the lowest-income workers see average wealth gains of 108% to 136%, depending on enrollment design. Hispanic and non-Hispanic Black workers reach gains of 76% to 99%. Gen Z workers (i.e., those with the longest time horizon) see gains of 108% to 133%. Single women reach 100% to 128%.

For lower-income workers, women, and communities of color, these results reflect populations the retirement system has most conspicuously failed. For younger workers, they reflect the power of compounding when access starts early. In both cases, the structure and generosity of the matching incentive matter most. Auto-enrollment gets people through the door. The Saver’s Match determines whether the system works hardest for those who need it most.

This suggests that access policies and incentive policies should be viewed as complementary, not competing line items in a budget negotiation. A federal plan without a meaningful match is an access program. A federal plan with a well-designed match is a wealth-building program.

It is worth being direct about what this study does not claim. Larger retirement balances do not automatically translate into retirement-income security. A worker whose projected wealth increases by 50% may still face a shortfall between projected income and expenses in retirement. The authors are explicit about this, and a follow-up study examining retirement-income adequacy is planned.

The study also does not estimate the fiscal cost of the proposed Saver’s Match enhancements, which is an important consideration for any legislative effort. And it does not model potential crowd-out effects on the private employer-sponsored retirement system, which is a legitimate concern that deserves analytical attention.

What the research does provide is something that too often goes missing from the policy conversation: a rigorous, empirically grounded framework for comparing alternative program designs under realistic behavioral assumptions. It tells policymakers not just that a federal plan would help, but which design choices produce the largest gains, for whom, and why. 

That kind of analysis is what separates evidence-based policymaking from policy by anecdote.

Approximately 32.3 million workers would enter the retirement savings system under the auto-enrollment scenarios modeled in this study, even after accounting for opt-outs. Over a 10-year horizon, the proposals would add between $635 billion and $1.35 trillion in wealth to the system, depending on design. 

These are not small numbers. They represent a meaningful shift in the retirement security trajectory of a generation of American workers who have been structurally excluded from the employer-sponsored system.

But those numbers are projections, not guarantees. They depend on the choices that policymakers make about enrollment structure, default rates, escalation, portability, leakage prevention, and matching incentives. Every one of those choices has a quantifiable effect on outcomes, and every one of them deserves the kind of careful, independent analysis that the Morningstar Center for Retirement and Policy Studies has provided here.

The worst outcome would be to enact a federal plan that checks the political box of “expanded access” but fails to deliver meaningful retirement wealth accumulation because of design choices that were never subjected to serious scrutiny. We have the research tools and the data to do better than that. Policymakers should use them.