Americans are living longer — and for many, that poses a problem. Retirement accounts have not kept pace as the cost of life after retirement has skyrocketed, says Damon Jones, an associate professor at Harris Public Policy. “You can’t support that longer retirement with the same number of working years.”

Add in the fact that a person’s final years can require costly medical interventions, such as assisted living or palliative treatments for chronic illness, and the takeaway is clear: Old age is a terrible time to run low on cash.

So it’s troubling that elderly Americans may be leaving huge quantities of money on the table.

In new studies, Jones and fellow researchers have found hard evidence that many senior citizens stop working prematurely, not because they necessarily want to quit working, but rather because they misunderstand a complex facet of Social Security policy: the “earnings test.”

The earnings test applies to everyone who receives Social Security benefits but has not yet reached full retirement age (currently 66). According to federal policy, if you continue to earn an income above a certain threshold while receiving benefits, some portion of your benefits will be withheld — but not forever. In most cases, that withheld money is incorporated into your monthly payments down the road, after you reach full retirement age.

From a policy perspective, the earnings test is meant to ensure that a person receives more of their Social Security entitlements later in life, when they’re most beneficial. “But in practice, some people seem to be responding to this like a tax,” Jones says.

He and his coauthors (Alexander Gelber of UC Berkeley, Daniel Sacks of Indiana University, and Jae Song of the Social Security Administration) have found that an individual whose income is close to the earnings test threshold is likely to cut back their hours, while higher income people might quit working altogether.

The researchers have good reason to think that this is the result of a pervasive misunderstanding — the incorrect belief that the money the government withholds is not paid back later in life, but forfeited forever.

Among the Social Security-eligible group that Jones and fellow researchers studied, the earnings test caused incomes to drop an average of at least 9.8 percent — that’s $579 per person.

The finding confirms the accusations of detractors who claim that the program senselessly diminishes earnings. So is this retirement policy ready for retirement?

Earnings test, past and present

If the earnings test seems draconian, Jones suggests considering it within in the broader history of retirement benefits.

Through the early 1900s, individuals who were too old to work would often find themselves unable to make ends meet once their regular incomes stopped. “A lot of people didn’t save, and they were old and impoverished,” Jones says. That was the impetus for including a government-run retirement income program in the Social Security Act, signed into law by President Franklin D. Roosevelt in summer of 1935.

In the original act, the earnings test was much harsher than it is today, Jones explains. “It used to be that you would just lose your benefits and you wouldn’t get them back.”

At the time, Social Security was framed as a program intended only for those unable to work — namely, the very elderly and the disabled. If you could do your job, the thinking went, then you didn’t need the government benefit at all.

But gradually, the policy began to look more and more like a savings scheme. In 1940, Congress set an earnings threshold, meaning elderly individuals earning $14.99 or less per week could keep their benefits. In 1950, retirees over age 75 were exempted from the policy. Since 1972, all retirees have been able to reclaim their postponed benefits once they hit full retirement age.

Jones sees this evolution as in line with the general philosophy of Social Security — rather than punishing people for choosing to work, the program would ensure that they received benefits when they most needed them, after they could no longer continue working. “Implicitly, we are being paternalistic and saying, ‘You may not be able to save well on your own,’” he says.

“Implicitly, we are being paternalistic and saying, ‘You may not be able to save well on your own.'"

- Damon Jones, associate professor, Harris Public Policy 

Lost labor and diminishing dividends

From an economist’s perspective, the earnings test should not necessarily deter a person from working, since — as long as their annual income exceeds a certain amount — their benefits are not actually forfeited, but simply delivered after they reach full retirement age.

But Jones is well aware that people don’t always act as classical economists expect.

Much of his research falls into the category of behavioral economics, a field that takes into account how psychological factors, particular contexts, and imperfect information lead people to deviate from fully rational decision-making.

In this case, there was reason to believe that people were misunderstanding the earnings test. In a 2011 survey, the AARP had found that more than half of senior citizens believed that the benefits lost by continuing to work would never be recouped.

It made Jones wonder: How many people were leaving the workforce early, or reducing their hours substantially, in order to remain exempt from the earnings test?

He and his colleagues were able to obtain a random sample of a restricted Social Security Administration database containing detailed information on every retirement benefit recipient from the last several decades. “That data was very rich,” says Jones. “We were lucky to be able to access it.”

Their final sample, further winnowed to meet their research needs, included 3.6 million data points. In a pair of studies, the authors used this data to form evidence-based conclusions about the impact of the earnings test.

They found that people whose pre-Social Security incomes fell within $3,000 of the earnings test threshold — that is, the group most likely to adjust their incomes to avoid being affected by the earnings test — saw their earnings fall nearly 10 percent in the years before they reached full retirement age. A full quarter of this decrease was from people who chose to stop working altogether.

They found that people whose pre-Social Security incomes fell within $3,000 of the earnings test threshold — that is, the group most likely to adjust their incomes to avoid being affected by the earnings test — saw their earnings fall nearly 10 percent in the years before they reached full retirement age. A full quarter of this decrease was from people who chose to stop working altogether.

And these were the conservative estimates. If people were indeed treating the earnings test exactly like a tax, as some economists have argued, then Jones and colleagues found that eliminating the earnings test could eventually increase seniors’ earnings even more — by as much as 23.4 percent. If the test disappeared, the number of Social Security-eligible people in the workforce could also increase by two to three percent.

The findings reinforced Jones’ suspicion that the earnings test is giving seniors the wrong incentives regarding work. But by better understanding those incentives, he and fellow researchers also got some critical clues about how to move forward.

“If we can find systematic types of mistakes, that’s very informative for policy,” he says.

Time to retire the earnings test?

Over the years, observers who suspected that the earnings test was keeping people from work have suggested drastic policy changes. “There’s definitely a camp that thinks we should cut this completely,” Jones says.

On the surface, Jones’ findings appear to lend support to this camp. But he cautions against axing the earnings test altogether. “The risk is that people will draw down their Social Security more rapidly, and be more likely to have less money at very old ages,” he says.

Jones and his colleagues say there are less dramatic alternatives. For example, the Social Security Administration could devise a new benefits schedule that backloads benefits towards the end of a person’s retirement, but no longer penalizes those who keep working. The idea would be to retain some benefits of the earnings test, without affecting the labor supply.

There are less dramatic alternatives. For example, the Social Security Administration could devise a new benefits schedule that backloads benefits towards the end of a person’s retirement, but no longer penalizes those who keep working. The idea would be to retain some benefits of the earnings test, without affecting the labor supply.

There’s also a more immediate remedy — to make sure people understand the policy as it currently works, so that they can respond more rationally. Jones hopes to partner with the Social Security Administration on new research exploring ways to reduce misinformation about the earnings test.

Change or no change, the stakes of this program are rising quickly. The full retirement age will soon increase to 67, meaning that the earnings test will suddenly apply to a larger segment of the population — and that thousands more seniors suddenly stand to lose out on work and earnings.

The implications could be severe for not only retirees, but for society at large, Jones says. “You probably want to think hard about any policy that’s discouraging people to work.”

Related assets

• Read the full paper