Especially when you are banking on those assumptions for decades.
I’m going to look at some broad versus a very specific set of assumptions. One of the most influential assumptions set in trying to determine the cost of a pension, as well as the strength of the fund to back up pension promises, is that of the return on assets (or the discount rate on the liability).
Many public plans use something close to 8%. There was some recent fanfare when CalPERS lowered its discount rate to 7.5% – ten years prior it had been lowered from 8.25% to 7.75%, and this was the most recent change.
But how important is that rate of return assumption, really?
Let’s take a look:
The above chart comes from this page at the Cleveland Fed, involves a variety of assumptions, but the main assumption being that the rate of return on assets differ in the two situations. As far as I know, that’s the only difference in the assumptions. There is a dispute about some of the other underlying assumptions, but I’m not going to get into that here.
The point is that for these set of assumptions, over 35% of the plans fail within 15 years under the 6% scenario, as opposed to less than 20% in the 8% return case. The starkest disparity can be seen with those projected to survive past 40 years – about one third surviving in the 8% return case and only 10% surviving in the 6% case.
It’s not even clear to me that when they’re giving these percentages that this is weighted by amount of assets or benefits promised — it can be that this is just a pure count. While Rhode Island pensions failing is not a good thing, it’s not quite as bad as, say, the California pensions failing. There’s a different number of people, and different dollar amounts, involved.
Speaking of California, let’s talk about another set of assumptions related to pensions. Back in 2002, a retroactive pension sweetener was given to Sonoma County employees, and part of the reasoning behind making this decision was that it would be easily paid for via employee contributions, with a bit of county support for public safety workers. But the point was that it wasn’t going to cost the county that much more over base benefits.
So how has that worked out?
The first document estimating the total cost for the increased General employee benefits is a letter dated March 20, 2002 from Rick Roeder of GRS to Robert Nissen, Plan Administrator and Gary Bei, Assistant Plan Administrator of SCERA. The letter said the cost was based upon the year 2000 Annual Actuarial Valuation. It estimated the 3% at 60 benefit enhancement cost was $60,016,104 for a 5.43% contribution increase from CURRENT employees. At the end of the letter Rick Roeder states “If you like, we can perform additional analysis to reflect the fact that increased benefits may trigger earlier retirement.”
The second and only other document estimating the cost is a letter dated June 5, 2002 that updates the benefit enhancement to 3% at 60 based upon the 2001 valuation indicating the cost is $68,614,650 for a net contribution increase of 5.78%. At the end of the letter, Rick Roeder states that “It is IMPORTANT to consider that the combination of increased benefits and earlier ages at which multipliers hit a ceiling may trigger earlier retirements, especially for the 2.7% at 55 proposal. This in turn, would have some cost increase impact. Please let us know if you wish to do additional analysis in this regard (as we did for safety 3% at 50 analyses).”
The actuaries had done a calculation of the benefit with no assumption changes with respect to employee retirement age. They did offer to do additional analysis…. but that analysis was never done.
So what’s happened?
The Cost of Accelerated Retirements
It became clear to SCERA a year after the new benefits took effect that accelerated retirements (that were not included in the original cost analysis) had became a problem which dramatically increased the number of retirees and the average pension.
The minutes for the Employee Retirement Association Special Meeting May 4, 2005 state, “In 2003 there were 38 General members retiring with an average annual pension of $22,468. In 2004, due to increased benefits there were 217 general members retiring with an average annual pension of $37,715. A long-term structural question needs to be explored – to what extent are people retiring earlier due to increased benefits and consequently should the actuarial assumptions be adjusted? The Retirement Board may want to look at adjusting the assumptions after the next experience study is performed.”
Obviously, if people are retiring earlier, they’re not making contributions to the fund as longer….and thus not covering the cost as projected in the base assumptions.
You can go to the Union Watch post for some more cost projections, but you can see how sensitive pension costs can be to particular assumption sets.
Reality does usually unfold differently from the assumptions, but one hopes that good assumptions are picked such that deviations average out…. as opposed to keep being biased in the same direction.