In 1997, as a result of state legislation, the pension plan for the Michigan State Employees’ Retirement System underwent a significant change. State employees who qualified for MSERS and who were hired on or after March 31, 1997, were placed in a “defined-contribution” retirement plan. Under this system, they were provided with individual retirement savings accounts to which the state government makes mandatory contributions and the employees make voluntary contributions.

This retirement savings plan, which defines the state’s deposits to the retirement account but not the level of future retirement benefits, stands in contrast to MSERS’ ongoing “defined-benefit” pension plan for state employees who were hired before March 31, 1997. Under that traditional plan, state government promises an employee a defined annual retirement income. To finance these future pension benefits, state government sets aside money and invests it annually, using the assets accrued over time to pay employees’ retirement benefits as they come due. Under this traditional plan, the investment risk lies with the state — ultimately, with the taxpayers.

In this Policy Brief, the author analyzes state pension data to determine whether state taxpayers have saved money because of the decision to close the MSERS defined-benefit plan to new members and to place them in the MSERS defined-contribution plan instead. The author reviews three areas of potential cost-savings: annual “normal costs”; unfunded liability; and “political incentives.”

The “normal cost” of a defined-benefit plan is the annual cost to state government of prefunding the future retirement benefits that working members earned in that particular year. The average normal cost of the MSERS defined-benefit plan from fiscal 1997 through fiscal 2010 — i.e., from the first year of the MSERS transition through the most recent year for which complete data is available — was 8.1 percent of the previous year’s payroll (the previous year’s payroll is typically employed by the state when measuring this cost).

The state’s annual cost of benefits earned under the MSERS defined-contribution plan cannot exceed 7 percent of the current year’s payroll, due to the plan’s design. Using data from the Michigan Office of Retirement Services, the Michigan Senate Fiscal Agency and the MSERS defined-benefit plan’s comprehensive annual financial reports, the author estimates that from fiscal 1997 through fiscal 2010, state government saved a total of $167 million in MSERS defined-benefit plan normal costs by switching new employees to the defined-contribution plan. This estimate includes an adjustment for the increased normal costs that can result from the closing of a defined-benefit plan.

A second potential area of savings involves the defined-benefit plan’s unfunded liability. This liability occurs whenever contributing the normal costs proves insufficient to ensure that a defined-benefit plan remains on track to meet its future pension obligations. As of September 30, 2010, the MSERS defined-benefit plan had an unfunded liability of approximately $4.1 billion. This shortfall has developed for several reasons, including the fact that the plan’s assets have not been growing at the actuarially assumed rate of 8 percent annually and the fact that the Legislature did not make the annual required contributions needed to finance the unfunded liability once it arose. If new employees had continued to enter the MSERS defined-benefit plan, the plan’s unfunded liability would almost certainly have been higher — an estimated $2.3 billion to $4.3 billion higher, given a proration based on state data.

Some argue that any savings from switching new employees from a defined-benefit to a defined-contribution plan is mitigated by the fact that closing the defined-benefit plan requires future amortization payments to be made on a “level-dollar basis,” which is initially more expensive than the level-percent-of-payroll basis used for an open plan. This “transition-cost” argument is dubious, however. The switch to a level-dollar amortization pattern does not alter the benefits ultimately paid, and in MSERS’ case, the state has generally failed to make the level-dollar amortization payments.

A final area of cost analysis involves the change in political incentives that occurs with the creation of a defined-contribution plan. A defined-benefit plan can carry considerable unfunded liabilities, while retroactive benefit increases can be enacted and necessary funding significantly deferred. Indeed, since proper funding of a defined-benefit plan requires taxing current voters to provide pension benefits that may not be paid out for years, sound funding policy can be unappealing to legislators seeking re-election and hoping to provide visible benefits now.

In contrast, a defined-contribution plan cannot be legally underfunded, and any increase in the plan’s benefits must essentially be paid for when the change is made. A defined-contribution plan thus reduces the political opportunities to defer funding of pension benefits to a future generation of taxpayers and avoids placing a questionable burden on taxpayers who may have been too young to vote when benefits were granted and funding was postponed. While it is difficult to quantify the savings from improved political incentives — the author offers no estimate — this category may be the single largest area of savings over time.

Thus, from fiscal 1997 through fiscal 2010, the MSERS defined-benefit plan is estimated to have saved state taxpayers $167 million in pension normal costs, $2.3 billion to $4.3 billion in lower unfunded liabilities, and important but unquantifiable sums by improving the political incentives of pension funding. These considerable savings and the fact that the plan is predictable, affordable and current in its obligations make it a model for reform of other state government pension plans.