Unlike many workers, I get a statement every quarter from my defined contribution provider that includes a lifetime income projection. This number estimates how much monthly income I can buy from my retirement savings. And it is not a very big number.
That’s the point of the lifetime income illustration. Social Security and defined benefit plans are always presented in terms of income. How much are we going to get paid each month? You just run a quick formula and out pops a number. The number makes sense because we can relate it to our current expenses to see whether we have enough to retire comfortably. The defined contribution era means we have a bigger number that we need to turn into a smaller number. That requires math skills that most workers don’t have.
This inability to make sense of the number in their account statements is one of the reasons that the Department of Labor is moving forward with proposed rulemaking that would give plan participants a simple monthly income projection. Research suggests that these income projections will give workers much more realistic expectations about when they should retire and how much more they need to save. It also suggests that information isn’t a panacea and that assumptions may be overly optimistic.
Pensions are expensive. In a low-yield environment, it takes a very big number to replace an income using lump sum savings. As Stanford economist John Shoven has noted: “You can’t finance 30-year retirements with 40-year careers without saving behavior that is distinctly un-American.” Americans hoping for a financially secure retirement will likely need to save more and retire later. And seeing a big number on an account statement can lull a worker into a false sense of retirement readiness.
Many Americans don’t realize how much it costs to build a lifetime income in retirement. In a 2009 article, UC-Davis professor Victor Stango and Dartmouth professor Jonathan Zinman show how investors who are bad at comparing dollar amounts across time periods (they call it exponential growth bias) are far more likely to make incorrect financial decisions. If one could simply correct this bias by bypassing the financial calculation in the first place, some of these mistakes could be avoided.
That’s the reason the DOL is moving to add a lifetime income number to the lump sum amount reported in the traditional ERISA defined contribution disclosure statement. A DOL official noted that “adding this information to quarterly account statements lets workers know whether their savings are on track. For those saving too little, it provides a wake-up call to step up their savings.” You can’t dodge the reality that a gaudy six figure account balance provides a three figure monthly payment.
Understanding how little income you can buy with current savings is, of course, depressing. But the DOL wants workers to know that there’s a way out. The proposed rulemaking will also show workers how much higher their lifetime income could be if they saved a little more each month. So beat them down with a dose of reality and lift them back up with a new and improved savings plan that might just allow them to reach their retirement goals.
Will It Work?
I’m a big fan of information policy. Showing calories at restaurants is a great example. Knowing that fettuccine alfredo has more calories than a bacon burger helps me make better lunch decisions. Knowing how much income I’ll get from my 401(k) balance helps me make better savings decisions. Better information tools help me choose the diet and the savings amount that makes me the happiest.
A recent study by Stanford University economist Gopi Shah Goda and her co-authors looked at how providing lifetime income projections influenced savings behavior among 17,000 employees at the University of Minnesota. The authors split the sample into four groups—no information (a control), a group that just received general information on saving for retirement including how to sign up for the DC plan, a group that was shown how increased savings rates would allow assets to grow over time into a lump sum at retirement, and a group that received a lifetime income projection. This way the authors could see what type of information policy gave workers the greatest motivation to save.
Their finding showed that information policy works in general, but it doesn’t work for everyone and its aggregate impact isn’t huge. The income projection was most effective at changing savings rates and increasing the amount saved compared to general retirement information and information on projected aggregate savings balance. So giving workers a specific lifetime income estimate was more effective than policies like workplace retirement education. The increased likelihood of making a change was 29% when given a lifetime income projection and the amount increased by $1,150 per year. The not-so-great news was that, even after the intervention, only about 5% of employees participated in the optional DC plan, so the average increase among employees was about $85.
To me, the most interesting finding is that the lifetime income projection worked better for workers who were most motivated to save. Getting the lifetime income projection was actually more effective for those who already considered themselves better informed about retirement. Perhaps unsurprisingly, the lifetime income projection was also most effective at increasing savings for those who placed a greater value on the future. It turns out that information doesn’t necessarily help less informed workers make better retirement decisions. It helps the ones who are more informed and motivated to save even more for retirement.
Many of the provisions of the 2006 Pension Protection Act have had a big impact on improving retirement outcomes. The most successful changes didn’t necessarily help workers make better choices—they helped them make no choice at all. Employer incentives to auto-enroll new workers into retirement plans had an immediate impact on employee participation. Automatic escalation increased how much retirees saved. Defaulting into target-date funds improved portfolio quality and equity exposure.
These features worked because they used a policy known as soft paternalism. Since financial decisions are often difficult and we have a limited amount of time to develop expertise, a benevolent policymaker can identify pretty good choices and make them the default option. Studies of defaults, whether locating desserts in a less convenient part of a cafeteria or making drivers opt out of organ donation, show the power of so-called choice architecture. Laziness trumps carrots and sticks.
Income illustrations are great for someone like me. I’m already saving for retirement, I want to make sure I’m saving enough, and the illustration substitutes a simple tool for a more complicated calculation. That improves my ability to save the right amount of money. And you’re probably like me. But a lot of workers aren’t.
So is a mandatory lifetime income illustration a good idea? They’re a great idea for those who are going to pay attention to it. These, however, are probably the workers who already save more than their co-workers. So we might get a policy that’s a good idea but increases the dispersion of retirement savings outcomes. The rich literally will likely get richer, but better information may do little to improve the prospects for those workers who are most vulnerable to running out of assets in old age.
When the DOL asked for comments on lifetime income projection assumptions, my first thought was that the best answer was probably the easiest. The only foolproof calculation is to use the current inflation-adjusted returns on safe investments (inflation-protected Treasuries, or TIPS) and then assume the retiree buys an inflation-protected annuity at retirement.
There are some problems with this extremely conservative assumption. First, you end up with a really depressing outcome—TIPS rates are tiny and you’d need over a million dollars to generate $50,000 of inflation-protected income through an annuity at age 65. Second, most workers with a long-run time horizon would hold some share of retirement savings in risky assets and they would also be able to generate a higher nominal return from corporate bonds in a tax sheltered account.
In their advanced notice of proposed rulemaking, the DOL suggested safe harbor projections of a 3.9% real return on pre-retirement portfolios and a withdrawal rate that assumes a 10-year T-bill rate and mortality credits (basically a fixed immediate annuity). The DOL notes that this is a somewhat conservative after-fee estimate since about 60% of 401(k) assets are invested in equities and a 60/40 portfolio had a historical U.S. real return of 5.6%. If someone told me right now that I could invest in an instrument that provided a certain 4% real return on my portfolio, you can bet that I’d shift every penny into it.
Comment letters to the DOL reflect this general disagreement about what assumptions are best for retirees. If you assume the same rate for all retirees, you ignore the wide range of asset allocation within retirees’ portfolios. You might also create expectations that the retirement income projection is the amount they will actually be able to buy at retirement, which makes many wonder how employees will respond when they aren’t close.
Then there’s the assumption that employees will buy something that looks like an annuity at retirement when many are wary of the potential liability risks of including annuitization options as a plan fiduciary. And if you give plan providers too much latitude in projection assumptions, you open the door for the possibility that more rosy assumptions will attract assets or greater investment risk. It’s complicated.
Even though it’s complicated, it’s hard to see anyone opposing a policy that’s proven to increase 401(k) assets. Employees will likely get a lifetime income projection and they probably won’t like what they see. That’s a good thing if it causes them to take action. Just don’t expect that it’s going to solve the retirement crisis on its own.