Defined-Benefit and Defined-Contribution Plans of the Future

Abstract

Defined-benefit pension plans are in decline. What went wrong? The lessons the investment community learns from this decline will help us create better plans in the future—plans that incorporate desirable features of both defined-benefit and defined-contribution plans.

Two main causes explain the decline of defined benefit (DB) pension plans. One cause starts with the confusion between what benefits are worth (calculated by discounting benefits at bond yields) and what constitutes the best estimate of the long-term funding target (where discounting includes an equity risk premium). This confusion led to overly generous benefit promises. And with funding based solely on the best estimate, without an additional reserve for investment risk, benefits became underfunded when the risk premium did not arrive. The second cause is legislation that, on the old-fashioned assumption that most corporations last forever, permitted underfunding to continue.

The next generation of DB plans will look different from yesterday’s plans. The benefit will probably start as a career average (possibly enhanced periodically by a catch-up provision). It will be valued and funded on the basis of bond yields, with a full funding requirement. There will be no benefit confiscation on an employee’s early death or termination; the benefit will transparently be the reserve held for the employee and will be communicated to the employee each year. Investment risk will require an additional reserve.

Future defined contribution (DC) plans will have “autopilot” features—default options designed to mimic desirable DB features, such as extended coverage, contributions increasing with age and pay, and a sensible investment policy that is professionally executed. Other arrangements will be able to convert the employee’s account into postretirement income less expensively than through individual annuity purchases.

When we have transparency of DB and DC plans, we will be able to see that retirement income guarantees are expensive. We will have to make individual decisions about bridging the gap between what we wish for and what pension plans can help us achieve. That situation is no different from the past, but perhaps we will confront it more explicitly in the future.

Notes

¹ The Dutch FTK regulations require DB pension funds to calculate their liabilities by discounting at market-based rates taken from the euro-swap curve (0.05 percent above the government bond rate) as the discount factor rather than a fixed actuarial interest rate of 4 percent. Assets must exceed liabilities by a margin that depends on the extent of investment risks taken but is never less than 5 percent of liabilities. FTK is scheduled to become effective 1 January 2007.

² I do not discuss the public sector here, but I will say that the public sector world goes on pretty much as previously, contrary to the principles I espouse.

³ Portability and quick vesting, together, mean that employers will no longer gain from an employee’s early termination of service. So, in principle, employers will need to make a lower benefit promise or an adjustment to some other aspect of total compensation.

⁴ Steve Kandarian, former executive director of the PBGC, in a presentation to the Annual General Meeting of the Society of Actuaries, New York City (November 2005).

⁵ Some genuine hybrids do exist. An example is Canada’s multi-employer plans or “collective DC” plans, as they are termed in the Netherlands. These plans are DC plans as far as employers are concerned. Then, an actuary estimates the DB formula that the contributions can support, and the DB formula is used as the payout. If the actuary’s assumptions are wrong, however, it is the benefit, not the contribution, that gives way. A different example is the Mercer Human Resource Consulting “Retirement Shares Plan” (October 2005), which is basically a career-average DB plan meant to be invested in matching bonds, but it has the creative twist that a participant can elect to take investment risk with the present value of his or her benefit and receive a benefit reflecting actual investment experience—in effect, the participant can make it individually a DC plan.

⁶ The passage of the Pension Protection Act in August of 2006 effectively prevents funding with 70-cent dollars.

⁷ See www.retirementsecurityproject.com. This project is supported by The Pew Charitable Trusts, Georgetown University’s Public Policy Institute, and the Brookings Institution.

⁸ The brainchild of Richard Thaler and Shlomo Benartzi (2004).

⁹ It is often said that DC plans are transparent whereas DB plans are opaque, which is cited as a reason employees, particularly younger employees, prefer the DC model. But the saying is not true. The value of defined contributions is transparent in the accumulation phase, but the retirement income it will generate in the decumulation phase is opaque. The DB model is the opposite: It is opaque as to value in the accumulation phase but transparent as to retirement income. Experience indicates that younger employees tend to prefer transparency in the accumulation phase; older employees prefer transparency in the decumulation phase. Why not make both models transparent in both phases?

¹⁰ Such a plan would give identical people identical outcomes if they made identical investment choices—for example, if they defaulted to the same investment option. But presumably, the election of a nondefault option is made consciously in order to increase the employee’s utility, so “identical outcomes” can be replaced in the socially desirable list by “opportunity to increase utility.”

¹¹ Indeed, the DC approach is typically more costly than the DB approach because it requires more detailed record keeping and is associated with higher investment management charges and more expensive rates to purchase annuities.