Thought Leaders
 Designing Pension Plans for Real People

Many pension plan designs don't adequately take into account the foibles of human decision making. These plans often seem to assume that workers are "are rational, autonomous, microcalculators who exercise independent and unbiased judgment," says Professor Olivia Mitchell, Executive Director of the Pension Research Council and Academic Director of Wharton's new Pension Strategy: Designing Resilient Retirement Systems program. "The evidence suggests that people do strive to maximize their self-interest, but for a variety of reasons they often fail to act in line with the expectations of rational economic and financial theory."

The result is that employees, who have an increasingly important role to play in their own retirement planning, often make sub-optimal decisions. Fewer than 40 percent of U.S. workers have calculated how much they need for retirement, and 30 percent have not saved anything for retirement. This is not always because of a lack of knowledge about the right course. One survey of 10,000 employees at a single company found, for example, that 68 percent felt their retirement savings rate was too low. They knew what levels they wanted to achieve, but were unable to do so.

In their forthcoming book, Pension Design and Structure: New Lessons from Behavioral Finance (Oxford University Press), Mitchell and colleague Stephen Utkus examine the implications of behavioral decision making for pension plan design. They consider how framing, inertia, procrastination, overconfidence, loss aversion, and other factors affect employee pension decisions and how pension plans can be designed to work around them — or even take advantage of them.

Harnessing Hyperbolic Discounting

For example, Mitchell and Utkus point to the Save More Tomorrow (SmarT) scheme developed by Richard Thaler and Schlomo Benzarti. This program is designed to take advantage of our human propensity for "hyperbolic discounting" and inertia. Hyperbolic discounting means that people apply much higher discount rates for short-term than long-term outcomes. For example, people will tend to take an apple today, rather than wait to receive two tomorrow, but they will be more likely to wait if the proposal is to receive one apple in 100 days and two in 101 days. They can be more detached from the future decision.

The SmarT plan takes advantage of this fact by encouraging employees to commit themselves today to future decisions. Under the program, participants sign up to automatically increase their pension saving at pre-specified dates (their anniversary) or to coincide with pay raises. As "hyperbolic discounters," they significantly underestimate the impact of such future commitments. Their inertia tends to keep them on the course they chose earlier, and, if savings increases are linked to pay increases, it also makes it look like the deductions are at little or no cost, taking advantage of the "money illusion." When the program was offered to employees at a 300-person firm, after 3 years the employees who signed up for it had boosted their saving rates from 3.5 percent to 11.6 percent.

Insights on Design

What are ways that an understanding of behavior can be used to improve pension plan design and spur retirement saving? Among the insights from Mitchell and Utkus:

  • Create opt-out defaults: Because of loss aversion and framing, people are less likely to opt out of a program that they are automatically enrolled in than they are to sign up for a program they have to opt into. With automatic enrollment, employees tend to save at the levels recommended by the employer. A study of one large U.S. firm found that plan participation rates jumped from 37 percent to 86 percent for new hires after automatic enrollment was introduced. Companies can also use default choices in shaping payout decisions at retirement.
  • Use automatic deductions to take advantage of inertia and procrastination: Once employees make a decision, they are less likely to change it later. Participation rates in 401k-type plans, where payroll deduction is the norm, are at least four times as high as for Individual Retirement Accounts, where structured payroll deductions are uncommon.

  • Take advantage of peer influences: Several striking studies have indicated that people with virtually identical demographic characteristics can have dramatically different saving rates, depending on whether their peers save for retirement or not.

  • Rethink education: While education is widely used to encourage employee retirement planning, behavioral evidence indicates that it often has limited impact. The programs tend to attract employees who are already motivated to learn about personal finances, so they tend to be preaching to the converted. Research shows that education may be more effective if workers are already practicing the right behaviors (through default participation, for example). Then education serves an ancillary role in explaining defaults and helping plan for future adjustments. "Creating defaults can be much more powerful than education in changing behavior," said Mitchell. "Instead of holding yet another seminar, it might be more effective to set a default saving rate of 6 or 10 percent of salary per year."
  • Limit choices: It is not always better to have more choices. Some 401k plans offer employees so many choices that they appear to be overwhelmed by them. One study found that reducing the choices offered to participants from five to two significantly increased participation.

  • Use new approaches to manage company stock risk: Even after Enron and other disasters, workers still significantly underestimate the risks of holding their own company stock. One option would be to allow employees to sign up for a mechanism that would automatically reduce this exposure with age (for example, limiting the percentage of retirement funds in company stock to 10 or 20 percent at age 65).

For copies of the chapter by Mitchell and Utkus (WP 2003-6), as well as other chapters in their forthcoming book, see the Pension Research Council Working Paper Library.

 

    

This month's articles: