July 18, 2012
[On Wednesday, July 18, Senators Jansen and Pavlov presented an alternative to the House “H-3” version of SB 1040. H-3 was ultimately voted down in the Senate 16-22, so SB 1040 will next be discussed in a conference committee with the hopes of a resolution of the issue at the one-day August 15th session.]
1. Stranded Costs Left Unaddressed. I will have a hard time supporting any proposal that does not make some effort to deal with stranded costs. To do otherwise is to kick the can down the road again, leaving it for a future legislature to again “fix the problem past legislators failed to do”. This “compromise” proposal fails in that regard.
2. Cost of Conversion to Defined Contribution ("DC") Plan. The major reasons our House Work Group recommended against closing the current hybrid plan for new employees in favor of substituting the a defined contribution plan equivalent to that available to state employees were (1) our belief that the DC plan normal cost was less than the cost of the proposed DC plan (with our desire to control costs, not raise them, the DC plan did not seem to make sense) and (2) with the inability to prefund the health care benefits AND fund the faster amortization indicated by the GASB guidelines, we were uncertain how negative the bond rating agencies would be if the state were unable to fund the calculated Annual Required Contribution. However, the DC plan has a number of very attractive features, and we did not wish to reject the idea without further study. So, we asked that a study be completed to assess the risk that the assumptions used to calculate the normal cost would not be realized, raising the risk of future unfunded liabilities for the hybrid plan that in total would cause the DC plan to actually be less costly, rather than more costly as our initial information indicated. The study was also to explore the variance from GASB issue.
Note that the total cost of a defined benefit plan is the "normal cost" (what the actuaries estimate the future cost to be based on a set of assumptions) plus or minus the costs or savings achieved by the fund actually experiencing less or more favorable results. If investment returns are less favorable than the assumed 7% of the hybrid plan, then additional costs in the form of "unfunded liabilities" would occur. The study is intended to evaluate how likely the assumptions would not be realized and what the impacts would be if various departures from the assumptions were to occur.
Investment Rate of Return Assumptions. Since our H-3 House version of SB 1040 passed the House in June, we have recently received more information. In summary, the DC plan is highly likely to cost more than the current hybrid plan, even according to the latest Arnold Foundation memo publicized by the Mackinac Center. The long term investment rate of return would need to average below 6% for the hybrid plan to be of equal cost to the state employees’ DC plan. Kathryn Summers’ estimates put the breakeven investment rate of return at about 5.5% (or even lower if the .55% cost of the disability and life insurance elements of the plan are included to get the total of 6.75% cost of the DC plan).
The difference between the Arnold Foundation results and the ORS paper dated 6/25/2012 (from which one would conclude that the normal cost is not sensitive to a change in the investment rate assumption) is a difference in assumption regarding the size of the fund investment portfolio. ORS looked at it from its current stage with a very small asset balance, while the Arnold Foundation study looked at it when the hybrid plan would be mature and fully funded. With a larger portfolio, the investment rate of return gets proportionately more important to the final, actual cost of the system. However, even by the Arnold Foundation numbers, it is 78% probable that the 6% rate of return will be achieved, and 84% likely that the 5.5% “breakeven” rate of return be achieved. Now, I admit, that leaves a 16% chance the hybrid plan might be more expensive, but that appears to be a small risk compared with the chance to avoid the higher cost of the DC plan.
The historic MPSERS Investment returns have averaged 9.78% over the 30 years through 2010, 8.93% over 25 years, and 8.15% over 20 years. Of course, past results are no guarantee of future performance, so what we will achieve in the future is cloudy in my crystal ball. Fitch believes the 8% used in the other MPSERS fund is optimistic and is evaluating risk based on 8%, 7% and 6%. The current hybrid plan uses a statutorily established 7%. The Arnold Foundation used information from the MPSERS CAFR to calculate the probabilities of various rates of return, but the reasonableness of any assumption depends on how optimistic one is concerning the future of our country and the world's economy.
Those two "studies" are not the end of the investigation needed, however. The Arnold Foundation is hardly an unbiased organization of the type we, as policy makers, should be engaging to get objective, third-party, unbiased information on which to make these serious policy decisions. Which view of the sensitivity of the investment rate is the better view: ORS's or Arnold Foundation's? Further, the investment rate of return is just one of the many assumptions used by actuaries to calculate the normal costs. We do not know what the assumption of 3.5% payroll increase does to the calculated normal cost and now much the actual total costs may vary if payroll actually grows at a slower or faster rate. Similarly, the retiree mortality rate has varied significantly from the actuaries' assumption in the past. Has a modified assumption been substituted? If not, how sensitive is the calculated normal cost to a different assumption? We don't know, but will know if the study called for in H-3 is completed.
When we get the results of the study, we might very well conclude that the attractive features of the DC plan (most importantly being the certainty funding a DC plan achieves) outweigh the risks that the hybrid plan might ultimately be more costly. I am just reluctant to make that decision on incomplete and biased information.
GASB Variance. It has become clear that the GASB guidelines are intended for reporting purposes, and are not funding mandates, despite the unfortunately named "annual required contribution". However, even if not actually required, the question remains what the bond rating agencies response would be if the ARC were not fully funded. My research indicates that Fitch does not see a variance from ARC as an automatic negative, if the variance is small. Their ratings are dependent upon a wide range of factors, however, so I would not expect that a negative reaction would occur if the reforms enacted, in total, reduced costs and more fully funded the liabilities than before, as our H-3 does. So, I believe that moving to a DC plan, with the result of closing the DB plan and needing to amortize the unfunded liabilities on a faster basis than our cashflow would allow, would not result in a negative bond rating agency reaction. However, getting some direct responses from bond agencies would make the answer to that question more certain, which the proposed study contemplates.