By Ed Ring, February 18, 2013
SUMMARY: This study describes how actuaries calculate two key variables that govern pension solvency; the plan’s “accrued actuarial liability,” defined as the present value of all future obligations to pay pensions, and the plan’s “actuarial value of assets,” defined as the current value – adjusted upwards or downwards to account for market volatility – of the plan’s invested funds. The amount of the unfunded liability is the amount by which any pension plan’s liabilities exceed their assets. This study then calculates the impact of new credit evaluation standards, proposed by Moody’s Investor Services to take effect in 2014, on the calculation of a pension plan’s liability. Using the most recent data that consolidates all state and local pension plans in California, provided by the California State Controller’s Office, this study revalues the accrued actuarial liability according to Moody’s new criteria. The calculations reveal that the unfunded pension liability for all of California’s state and local government pension plans combined increases from the official estimate of $128.3 billion, to $328.6 billion using Moody’s new criteria. Included with this study are downloadable spreadsheets that allow the reader to conduct their own analysis using their own assumptions.
* * *
When assessing just how much debt California’s taxpayers owe their government, it is misleading to suggest that all future obligations represent current debt. After all, future tax revenues will be available to pay future obligations. But to the extent money should be set aside today to fund future obligations, and has not been, by any reasonable accounting standard, this funding shortfall is considered debt.
This principle is particularly applicable to retirement pensions, which must be funded with money set aside during each year a beneficiary is working. A retirement pension fund accumulates money that is set aside during the entire term of a beneficiary’s career, the money is invested and the returns are added to the fund. If all goes according to plan, by the time the beneficiary retires, sufficient funds (which continue to earn interest) are on hand to pay the expected pension for the duration of the beneficiary’s expected lifespan. To view a relatively simple financial model that illustrates these pension dynamics, download the CPPC produced spreadsheet “Flexible Pension Analysis.” A much more detailed explanation of how the model works can be found in the CPPC study “A Pension Analysis Tool for Everyone.”
When considering not an individual pension, but a pension fund, the principles just described all apply, but the contributions for all of the participating individuals are pooled into a single set of investments. This pooling results in considerably lower risk to participants for two reasons – financial and actuarial. The financial risk is lowered since a fund that pools the contributions of thousands (or millions) of participants can be diversified, professionally managed, and operated essentially in perpetuity (which mitigates the impact when the market has a few bad years). The actuarial risk is lowered because it is much easier to predict how long the participants will live on average, and fund their retirement accounts accordingly.
Despite the virtues of aggregated pension funds vs. individual retirement accounts, maintaining the solvency of a pension fund is a complex and potentially risky undertaking. Problems can arise if there is an unusually prolonged period of lower than expected investment performance. Problems can also arise if, during an unusually prolonged period of higher than expected investment performance, pension benefits are enhanced beyond what is actually sustainable.
This is exactly what happened with California’s state and local public employee pension funds. During the internet fueled stock bubble of the late 1990’s, followed by the real estate bubble during the early 2000’s, pension benefits were enhanced. Not only were these future obligations increased, but the new, more generous pension formulas were enhanced retroactively, so participants who were close to retirement saw their per year pension benefit accruals enhanced for their entire career. But what goes up, must come down.
The remainder of this report will not focus on whether or not it is practical to project long term investment returns at 7.5% per year, which is what pension funds currently use for their projections. Rather, this report will perform sensitivity analysis, using the latest consolidated data available for all of California’s state and local public employee pension funds, and determine what the unfunded liability would be based on lower long term rates of return. The assumptions we will use are taken from Moody’s Investment Services “Adjustments to US State and Local Government Reported Pension Data,” a proposal that was released for comments in July 2012. The methods we will employ mirror those used by John Dickerson, editor of YourPublicMoney.com, in a study he contributed to the California Policy Center in January 2013 entitled “The Impact of Moody’s Proposed Changes in Analyzing Government Pension Data.” But whereas Dickerson performed this analysis for the independent pension funds serving six Northern California counties, we will use data from the California State Controller’s office that consolidates the financial statements of every state and local public employee pension fund in California. This data is found in a document released on March 20, 2012 entitled “Public Retirement Systems Annual Report for the fiscal year ended June 30, 2010.”
It is unfortunate that the only data available that consolidates over 80 independent state and local public employee pension funds in California is nearly three years old. Much of the reason for this is unavoidable since the actuarial analysis itself takes several months. As a result, even if it were practical to attempt to duplicate the state controller’s efforts and consolidate these many pension fund financial statements independently, there still would be a considerable time lag. For example, in the annual reports for CalPERS and CalSTRS, the section that presents their solvency analysis, which depends on updated actuarial studies, lags a full year behind their financials. In reality, the state controller’s office works pretty fast to consolidate this data. The financial statements for the pension funds are released six months after their fiscal year ends; i.e., in December 2012, most pension funds released their annual reports for the fiscal year ended on June 30th, 2012. Those reports, in turn, contained solvency analysis for the year ended June 30th, 2011, since the actuaries can’t even begin their work until the financial statements are completed. Therefore, by March 2013, the state controller will have again consolidated all of this data in an updated annual report, allowing us access to consolidated pension fund solvency analysis for the fiscal year ended June 30, 2011. But for now, we will have to go with data for the fiscal year ended June 30, 2010.
Here then is the official unfunded pension liability for the fiscal year ended June 30, 2010. The source for this is page 15 (XV) of the introduction to the above referenced state controller’s report, and is based on a long term rate of return projection that, in general, is about 7.5% per year for the more than 80 funds consolidated here.
In order to explain how to revalue the estimated amount of underfunding using a lower projected rate of return, it is helpful to define the two variables that govern whether or not a pension fund is overfunded or underfunded.
The actuarial accrued liability represents the present value of all future retirement payments due all participants in the pension funds. This value must be continuously adjusted based on changes to the participant population. As new employees enter the funds with distinct benefit formulas, as current employees retire, and as retirees eventually die, the amount of this liability must be painstakingly recalculated.
The actuarial value of the assets represents how much money the pension funds have on hand to invest and to fund existing payouts to retirees. The value increases each year by the amount the funds earn through investment returns, as well as by the amount that is contributed to the funds. Their value decreases by the amount paid out by the funds to retirees, as well as by the amount it costs to administer the funds. Many funds don’t perform an actuarial revaluation of their assets because they use the market value of their funds when calculating solvency. CalPERS, for example, recently began using the market value of their assets instead of the actuarial value for their solvency analysis. But the state controller uses the actuarial value, as does CalSTRS and many other funds. The only difference between the actuarial value and the market value is that the actuarial value is based on an adjustment to market value that takes into account historical trends. This means, for example, if the market value of the funds have dropped faster than average over the past few years, the actuarial value of the funds will be higher than the market value, on the assumption the markets will recover. Conversely, if the market value of the funds have appreciated faster than average over the past few years, the actuarial value of the funds will be lower than the market value.
While there is room for debate as to whether or not the actuarial value of assets is accurate, or, for that matter, whether or not many of the invested assets have sufficiently transparent data regarding their value or are sufficiently liquid to have any reliable market value, it is the actuarial value of the liability to pay future pensions that generates the most controversy. This is because once the future value of these future financial obligations is calculated, in future dollars, the rate at which they are discounted to reduce them to a present value has an extreme impact on just how much these liabilities are worth. As of June 30, 2010, California’s consolidated state and local public employee pension funds had liabilities valued at $724 billion, and 82% of those liabilities were matched by assets, which were valued at $596 billion.
This discussion is more than academic, despite the fact that it is impossible to have a meaningful dialogue with anyone about pension solvency if they don’t have a clear understanding of the difference between present value and future value. Because if the present value of the future pension liabilities is not equal to the present value of the pension fund assets, the plan is underfunded. And to the extent it is underfunded, it will not generate sufficient investment income to maintain whatever level of funding it does enjoy. That is, when a pension fund is underfunded, unless contributions from participants are increased to cover the shortfall in investment returns, the fund will become even more underfunded in an accelerating cascade that can eventually result in a bankrupt fund. Put another way, if a fund is 50% funded instead of 100% funded, it cannot survive by hitting its return on investment target of 7.5%, because it is earning that 7.5% on half as much money as it needs. It has to earn 15.0%, just to stay at 50% funded. From this perspective, it is misleading to suggest that California’s pension funds are within adequate bounds if they were, as the data from FYE 6-30-2010 indicates, 82% funded.
What Moody’s has suggested is to calculate the present value of pension fund obligations at the high-grade long-term corporate bond rate of 5.5%, which they deem to be less risky than 7.5%. Since the rate at which pension funds project their annual earnings is the same rate at which they discount their future liabilities, the adjustment calculations are relatively easy. Along with assuming a 5.5% discount rate, Moody’s has to make an assumption regarding what point in the future the liability must be discounted from. In reality, the actuaries calculate the future obligations for every year in the future, starting with next year, and projecting out to the point at which they estimate the participants who were hired this year (or their survivors if they receive survivor benefits) will have all died. They calculate the total amount that they estimate participant population will collect in every year between now and, say, 2072, taking into account mortality statistics, projected earnings growth (which affects the final pension calculation), cost-of-living adjustments, and any other variables that would affect the estimates. They then discount each year’s estimated obligation back to the present, and add them all together. In order to revalue these liabilities without having access to every actuarial calculation from every fund, what Moody’s proposes is to simply estimate the midpoint of the future payments stream. They select 13 years into the future, which seems a bit conservative. Here is their rationale:
Moody’s Adjustments to US State and Local Government Reported Pension Data, request for comment, July 2nd, 2012, page 6:
“To implement the discount rate adjustment, we propose using a common 13-year duration estimate for all plans. This is a measure of the time-weighted average life of benefit payments. Each plan’s reported actuarial accrued liability (“AAL”) is projected forward for 13 years at the plan’s reported discount rate, and then discounted back at 5.5%. This calculation results in an increase in AAL of roughly 13% for each one percentage point difference between 5.5% and the plan’s discount rate. For example, a plan with a $10 billion reported AAL based on a discount rate of 8% would have an adjusted AAL of $13.56 billion, or 35.6% greater than reported.
We recognize this duration estimate may be higher than warranted for some plans and lower than warranted for others. Each pension plan has a unique benefit structure and demographic profile that affects the time-weighted profile (duration) of future benefit payment liabilities. However, plan durations are not reported, and calculating duration individually for each plan is not feasible. Our proposed 13-year duration is the median calculated from a sample of pension plans whose durations ranged from about 10 to 17 years. Plans with shorter durations usually are closed or have a preponderance of older or retired members.”
Here is the formula that governs this readjustment:
Adj PV = [ PV x ( 1 + official %i ) ^ years ] / ( 1 + adjusted %i ) ^ years
Here is the formula with the actual variables provided by the California state controller (in billions):
Adj PV = [ 724.4 x ( 1 + 7.5% ) ^ 13 ] / ( 1 + 5.5% ) ^ 13
Here then is the adjusted unfunded pension liability for the fiscal year ended June 30, 2010, using a discount rate of 5.5% instead of 7.5%:
As can be seen, if the discount rate is lowered to 5.5%, and the actuarial accrued liability is revalued according to Moody’s proposed criteria scheduled for adoption in 2014, it results in the estimated funding status of California’s consolidated state and local government pension plans lowering from 82% funded to 64% funded.
The next table is also available as a downloadable spreadsheet entitled “Impact of Discount Rate on Pension Liability.” The reader is invited to download this spreadsheet and conduct their own analysis. In the table presented below, seven scenarios are considered, each using distinct, but credible assumptions regarding the discount rate and the duration in years between the present value and the future value of the pension liability. Scenarios 1 and 3 (columns 1 and 3) have already been covered; the first is the official amount of California’s combined unfunded pension liability for all state and local government pensions according to the California state controller, the third is the restated liability and consequent amount of underfunding using Moody’s proposed new credit evaluation criteria.
The second scenario (column 2) uses a discount rate of 6.2%, which is referred to as the average long-term investment returns of a “Blended 20th Century Fund.” Here is how Stanford University professor Joe Nation describes that rate of return in his December 2011 paper entitled “Pension Math: How California’s Retirement Spending is Squeezing The State Budget:”
“This 6.0 to 6.5 percent figure is based on the performance of a hypothetical fund containing 80 percent equity and 20 percent income instruments between 1900 and 1999. It assumes an equity rate based on the 20th-century Dow Jones industrial annual average of 5.3 percent, plus 2 percent in dividends, less 0.5 percent in fees. Combined with income instruments with a net rate of return of 4.5 percent, this hypothetical fund would have earned an average annual rate of 6.2 percent.”
As can be seen, if the discount rate is only lowered to 6.2%, it still results in the estimated funding status of California’s pensions lowering from 82% funded to 70% funded.
The fourth scenario (column 4) uses what is called the “Low Risk or Treasury Rate” of 4.5%. As noted in Nation’s study, even this rate is not considered 100% risk free, and is by no means a worst case. Using a 4.5% projected rate of return means that the unfunded pension liability for California’s state and local public employees is not $128 billion, but $329 billion. And this still may be optimistic.
The next three scenarios, columns 5, 6, and 7, show what happens if the “mid-point” of the future obligations is not 13 years in the future, but 17 years in the future. And why wouldn’t it be? As Moody’s themselves state, “Our proposed 13-year duration is the median calculated from a sample of pension plans whose durations ranged from about 10 to 17 years. Plans with shorter durations usually are closed or have a preponderance of older or retired members.” One must take into account the fact that the center of gravity of pension plan obligations, or the “time weighted profile of future benefit payments” is not merely a function of the age distribution, but also of the generosity of the benefits bestowed. From this perspective, the older participants in California’s public sector pension plans, those who retired prior to the benefit enhancements of the late 1990’s and early 2000’s, are impacting the plan less than their numbers, because their benefits are significantly lower, per capita, than the benefits of the people who have just retired or are about to retire. This means the 13 year duration may be too short.
If you increase the duration of the pension plan discounting to what is probably a more representative 17 years, then, as the chart indicates, the unfunded pension liability at 6.2% is $295 billion (67% funded), at 5.5% it rises to $401 billion (60% funded), and at 4.5% it rises to $576 billion (51% funded). Scenario 7, which assumes a 17 year horizon for discounting and a discount rate of 4.7%, is not at all unrealistic.
The data used in this analysis, while the most current available, is nonetheless quite dated. It shall be interesting to see what the June 30, 2011 data will indicate. That data should be released by the State Controller’s office in a month or two. To try to get an idea, however, we looked at the June 30th 2010 solvency tests for CalPERS and CalSTRS as disclosed on their most recent annual reports (CalPERS, page 128, CalSTRS, page 113). Because the tables show solvency test results for multiple years, we were able to see the June 30, 2010 data, as well as the June 30, 2011 data, on their annual reports for the fiscal year ended June 30, 2012.
As it turns out, as of June 30, 2010, the actuarial value of the assets managed by CalPERS and CalSTRS combined accounted for 67% of the total assets disclosed by the state controller for all of California’s pension funds, and they accounted for 70% of the total liabilities disclosed for all of California’s pension funds. With this in mind, the performance of these two very large funds over the past few years could provide some indication as to whether or not the official calculation of the consolidated pension plans underfunding will have improved or worsened over the past two years. In the two years ended June 30, 2012, CalPERS earned 21.7% and 0.01%, respectively. CalSTRS earned 23.0% and 1.84% for the same two years. This suggests that the official level of underfunding for California’s state and local government employee pension funds has improved marginally. But these returns also evince the unsettling volatility of investment returns, even with very large, professionally managed funds. And as we have demonstrated here, what average rate of return is ultimately delivered over the next 10-20 years by these funds has an extreme impact on whether or not these funds can remain solvent.
This report was prepared by CPPC Research Director Ed Ring, with assistance from John Dickerson, Marcia Fritz, and Joe Nation.