A New Year, An Old Post on Pensions

One of my earliest research projects focused on the shift from defined benefit (DB) to defined contribution plans and its implications for retirement income. I started it in 1993, along with my colleague (then at the NBER, now at Dartmouth) Jonathan Skinner, and after a lot of updates and modifications in response to comments, it was eventually published in The American Economic Review in 2004.

The reason we wrote it, and the reason why I mention it today, is that many commentators on pensions, and the public policy toward pensions (the name of the conference at which we originally presented it), simply have no idea about the generosity of defined benefit pension plans as they existed before the shift to 401k plans began. What was noteworthy about our article is that it used data on defined benefit pension plans, collected through the Pension Provider Surveys of the Federal Reserve's Survey of Consumer Finances, (SCF) to simulate the distribution of future retirement benefits that would obtain under the vintage of defined benefit pension plans that existed as this trend emerged.

We had data on DB pension plans in 1983 and 1989, and we were able to compare the distribution of benefits under those plans to the analogous distributions of defined contribution or 401k plans as they emerged in 1989 through 2001, the latest year of data prior to our publication. So we had the true heterogeneity in the DB plan universe, and our simulation framework incorporated uncertainty in both future wages and asset market returns. Our universe of 401k plans, and the contribution and investment behavior of participants, was drawn from those SCF respondents who relied on the 401k as their sole pension plan, not those who had it as a supplemental plan.

The key point of the paper was that even for workers who remained with the same pension plan for an entire working career, the vintage of 401k plans that existed by 1995 was providing a comparable distribution of retirement income. Put simply, participants who relied on the 401k as their sole source of pension income were, along with their employers, contributing enough and holding enough in stocks (as opposed to bonds) to match what they would have received under the DB plans that were supplanted. In a nutshell, the DB plan formulas were not necessarily that generous in terms of their replacement rates, and their reliance in many cases on "final average pay" exposed participants to a lot of uncertainty.

The comparison is even worse when we consider workers who switch pension plans one or more times during their career. When the job changes happen before the worker reaches the DB plan's early retirement age, the worker is typically entitled only to "vested deferred" benefits, the value of which is eroded by inflation (if taken at the plan's normal retirement age) or full actuarial reductions. The balances in defined contribution plans like 401k plans are fully portable and do not suffer from this "backloading" of benefits under DB plans.

This study is what informs my view, which I expressed to Bloomberg View reporter Ramesh Ponnuru in his article that appeared today. I am quoted as saying:

Criticism of 401(k)s frequently idealizes the defined-benefit plans they have largely replaced. It’s true that 401(k) participants have more responsibility for their retirements than defined-benefit plans involved, and they are also exposed to market risk. Andrew Samwick, a professor of economics at Dartmouth College, pointed out in an interview that defined-benefit plans had their own risks. The company sponsoring those plans could skimp on pension contributions, or allocate its investments poorly, or go bankrupt and leave plan participants short.

And the shift to 401(k)s has coincided with a large increase in the number of people with retirement plans: Most workers didn’t have those defined-benefit pensions. “People have a very distorted notion of how good things were under defined-benefit plans or how good they would be today if that system of defined-benefit plans had continued,” Samwick said.

Our comparisons also did not consider the problem of DB plans that the security of the participants' retirement income depends on the solvency of, and funding decisions made by, the plan sponsor. It was not over 401k plans that Daniel Gross declared "the decade of cramdown."