Tuesday, June 16, 2009

New paper: “Answer Quickly: How Much Do You Think You'll Get from Social Security?”

I have a new paper in AEI's Retirement Policy Outlook series that examines what I call the "predictability risk" of Social Security benefits – put simply, the risk that your retirement benefit will turn out to be different from what you thought it would be. This is more important than you'd think, since Social Security's benefit formula is so complex that the typical person could never figure out their benefits on their own. Unlike simpler corporate DB plans, it's pretty much impossible to guestimate your benefits "on the fly."

To examine this issue, we used the Health and Retirement Study, which includes both questions on how much you think you'll receive from Social Security and how much you are actually receiving. Using people who responded to more than one wave of the study, we could compare what near-retirees thought they'd receive one to two years prior to claiming to what they actually got when they retired.

About one-quarter of HRS respondents couldn't even hazard a guess as to their benefits. Of those who did make an estimate, their guesses were very close on average – but there was a great deal of disparity. Many people overestimated their future benefits, while many people underestimated them.

Both guessing low and guessing high have a cost, but overestimates obviously are more damaging for retirement income security. One-third of near-retirees overestimated their benefits by at least 10 percent, while one quarter overestimated them by more than 28 percent. One in ten retirees received a benefit less than half as much as they expected. This predictability risk is every bit as damaging as having your 401(k) account decline on the verge of retirement.

A number of policies could help. First, better publicity for the Social Security Statement, which appeared to improve people's knowledge when it was first rolled out in 1995 but doesn't seem to do as well today. (I have some ongoing research on this issue, so I'm not yet ready to say how well the Statement has really done.) Second, new sources of benefit estimates, such as the SSA's new online benefit estimator. And third, and probably most importantly, simplifying the benefit formula as part of a larger Social Security reform.


WilliamLarsen said...

“since Social Security's benefit formula is so complex that the typical person could never figure out their benefits on their own. Unlike simpler corporate DB plans, it's pretty much impossible to guestimate your benefits "on the fly."

If you think the Social Security Benefit is complicated, then you are in the wrong business. It is one of the most forward formulas there is. First the bend points are indexed based on the change in the US average wage. Therefore, whatever you guess the growth to be you can calculate these bend points. Does it matter if you are wrong, no? The reason is that rate of growth in the bend points mimic buying power in any given year. If you assume low, your benefit will be low, but you will still have the same buying power as if you had assumed a high growth rate.

Second, the two bend points have three fixed replacement rates, 90%, 34% and 15%. Very straightforward.

Third, your wages subjected to SS-OASI taxes determine your average monthly indexed wage. Is this complicated, no? Your W-2 form from your employer lists this. Most people know what they make by looking at the pay stub. All you need to do is keep track of your wages on a yearly basis. This is not difficult at all.

Fourth, using a spreadsheet one can easily identify their yearly wages, import the bend points, assume their own wage growth and one for the US economy if they think they will be different and project future bend points. Is this hard, no? Again if you simply assume your wages grow at the US average Wage, you will be very close. In fact if you want to be able to determine a cohort’s average monthly benefit, use the US Average wage in the year they retire.

The key is buying power consistency and not absolute dollars.

Andrew G. Biggs said...

There are a couple answers this, from the simple to the more complex:

First, the sheer fact that many people even on the verge of retirement have no idea what their benefit is going to be hints that the calculation may be difficult.

Second, your example is for a relatively simple stylized worker; in the real world benefits can vary a lot based on your marital status, length of marriage if divorced, number of years of earnings, etc.

For instance, take two people who earn the same amount over their lifetimes, except one has 9 years in the workforce and the other has 10. The latter person will qualify for benefits, the former won't. Likewise, take two identical lifetime earners, except one crams his earnings into 35 years while the other works at lower annual wages, but for 45 years. Different benefits. One more: take two households with identical lifetime earnings, except one is single earner and the other has two earners. Different benefits.

Finally, I think your explanation itself would be well beyond most people's capacity to follow. Many won't even be able to follow the concepts (e.g., what's 'wage indexing' in the AIME? You need to convert the wages on your W-2s.); others won't be able to find the necessary data; still others can't use a spreadsheet -- I know many of them!