Saving for retirement and addressing inequality—an optimal plan
The best retirement savings plan offers more choice for some, not all
Published February 25, 2019
Few of us have the foresight or discipline to save sufficiently for old age. This isn’t a new phenomenon—a boomer defect or millennial flaw. Emperor Augustus of Rome established generous retirement benefits for his soldiers in 13 B.C. (if only to prevent insurrection by penniless vets), financed by estate and auction taxes. Germany adopted a social insurance plan in 1889 at the behest of Chancellor Otto von Bismarck, who urged his parliament, “Those who are disabled from work by age and invalidity have a well-grounded claim to care from the state.”
Today, Americans with a job are required to save for retirement through contributions to Social Security. We can also stash cash in 401(k)s, 403(b)s, 457s, SEPs, IRAs, Roth IRAs, and a range of other inscrutable initials. But few feel completely secure that we’ll have enough, and many doubt the federal plan’s long-term solvency.
Economists say that government can play a role here because individuals suffer “present bias”—we would rather consume now than later. Acting as our better selves, governments can implement policies that encourage and/or require us to save adequately for life beyond a paycheck, particularly for those who earn little. In the jargon of economics, such policies are “paternalistic,” not in a pejorative sense, but because they limit an individual’s freedom in her or his apparent best interest.
“Optimal Paternalistic Savings Policies,” a working paper from the Federal Reserve Bank of Minneapolis’ Opportunity & Inclusive Growth Institute, focuses on the question of choice under such policies. Assuming that a government has a dual mandate—to ensure that its citizens adequately fund their retirement and to address income inequality, how much latitude should individuals be given in saving for their own future?
The economists, Institute scholar Christian Moser of Columbia University and Pedro Olea de Souza e Silva of Uber Technologies, find that the optimal policy requires that low-income individuals save a large portion of their net income, but higher-income individuals choose among a range of subsidized retirement plans. The key lies in creating incentives for high-earning people to work and earn more so that they can generate higher tax revenue for redistribution.
“A trade-off exists between paternalism and redistribution,” write the economists. “The optimal policy enforces high savings rates at low incomes but offers a choice between various subsidized savings options at high incomes.” Their research suggests that the programs we now have—Social Security, IRAs, and employer-sponsored 401(k)s—already have features that resemble the savings system deemed optimal, but that those programs would need major changes to actually be optimal. By increasing mandatory savings and limiting savings choice, particularly for low-income individuals, overhauled savings plans could achieve the equivalent of 8.8 percent of disposable income for everyone according to the economists’ calculations—an enormous increase in economic welfare.
Vive la différence
The analysis takes into account two key observations about real-world savings behavior: large differences in how much people save voluntarily and continuously growing income differences. Put simply, people differ both in how much they care about their current versus future well-being and also in how much income they have when contemplating this trade-off. The economists capture these features in a model that allows for differences in both present bias and income ability.
The interplay of those ingredients generates theoretical and empirical results for how a government can best address income inequality and undersaving for retirement. Arriving at those results is more difficult than it might appear. The paper runs over 130 pages, and well over half is devoted to mathematical proofs and computer simulations. (The authors develop a “general, computationally stable, and efficient active-set algorithm,” if that helps.)
Ultimately, they find, to ensure adequacy of retirement savings and curb income inequality, government policies should require low-earning people to save as much as 46 percent of their income net of transfers for the future, which ensures fairly stable lifetime spending. The optimal savings rate would decline until an income level of around $95,000. Above that, individuals can choose their savings levels and use subsidized savings accounts, resulting in savings rates between 3 percent and 18 percent, varying by income level and present bias.
At first glance, this plan sounds punishing in that it forces the poor to sacrifice current consumption, but under the proposed plan, they benefit from more generous redistribution policies while young and old, thanks to better-funded tax-and-transfer programs. By giving richer people more choices, the plan ensures that the economy operates as productively as possible, thereby generating higher tax revenue to fund those programs.
“Intuitively, the planner offers savings choice to incentivize work and collect tax revenues, thereby facilitating redistribution,” explain Moser and de Souza e Silva. Or, as they later put it, bluntly: “High-ability types are treated as cash cows.”
Diverse needs and optimal policy
To ground their analysis, the economists look at U.S. data, enabling them to disentangle diverse savings motives: People have different spending needs when young and old. Savings motives vary widely by person and include unanticipated health conditions, family bequests, and uncertain longevity—the risk of outliving one’s savings. Based on myriad household characteristics revealed by the data, they find a substantial range of present bias and a positive correlation between current income and the degree to which people value the future.
Their analysis provides a framework for analyzing many real-world retirement savings plans. The optimal design for Social Security would provide benefits that are hump-shaped, rising from salary replacement rates during retirement starting at 68 percent for low-income individuals, rising to 145 percent at an annual income of $65,000, and declining thereafter.
Those earning more than $95,000 per year should be able to use retirement savings accounts such as 401(k)s or IRAs with progressive savings subsidies. And the optimal tax schedule, relative to the current U.S. personal income tax code, would feature increasing tax levels but decreasing marginal rates, again, to encourage work effort and tax revenue.
This research certainly isn’t the last word on optimal savings policies, but it provides a new framework to think about the paternalistic quest of taking care of an aging population. Moser and de Souza e Silva suggest, for example, that using similar methodology might help measure implied paternalism levels in retirement savings plans elsewhere.
Germany and Rome, perhaps.
For another Minneapolis Fed look at optimal tax design, see “Optimal income taxes,” on research by Jonathan Heathcote and Hitoshi Tsujiyama.