Tag Archives: Actuarial Accrued Liability

Banking on risky investments is no way to guarantee a public pension

Over the past several years I’ve spent a lot of time studying public pension systems. That’s involved diving into the economics and actuarial literature, reading through many individual plan reports, and analyzing the trends in those systems in the context of the principles of financial economics. Why do this? It isn’t just a public finance problem. Twenty million Americans participate in these plans. If research points to systematic structural weaknesses in public sector plans, that under the right conditions, can lead to plan failure, then it is an imperative to point it out and recommend solutions to ensure that retirees receive the pensions they’ve been promised without placing unnecessary burdens on taxpayers or forcing painful budget tradeoffs at the worst possible time: during a recession.

The only way to protect pensions is to accurately assess their true value and funded status and then contribute what is needed to pay out those benefits. Unfortunately, the story of US public sector pension is that they are built on investment risk and accounting illusions.

Pension finance is not without controversy. Misunderstandings can arise in part due to the very different approaches taken by financial economists and traditionally trained actuaries over how to most appropriately value pension liabilities and assets, as well as the nature of investment risk.

However, some of the conflict is due to the implications of the pension literature. Applying the economic approach to valuing pension fund liabilities reveals trillions more in obligations and far bigger funding gaps for states and cities. It shows how public sector plans have exposed themselves to an unwise amount of investment risk effectively linking guaranteed pension payments to market volatility and putting taxpayers on the hook for losses. Some state and local governments have responded to this debate either through small accounting reforms or policy changes meant to shore up pension systems. These reforms are not necessarily sufficient but it’s a tacit recognition that the math really matters.

There are some plans that continue to staunchly defend a “More investment risk = safe and guaranteed pension with no downsides” approach. And at least one system has gone on the offense against any suggestion that increasing investment risk in a government-guaranteed pension system amounts to gambling with employees’ pension benefits.

In May 2014 I authored a paper that made the case for economic accounting and better funding for Alabama’s three state-run pension plans.[1] My study was featured in The Advisor in July 2014, the newsletter the Retirement Systems of Alabama (RSA) provides to its members.[2] One article written by “RSA staff” purports to debunk my paper, but ends up missing the implications of both the literature and my analysis.

The RSA staff’s main complaint revolves around one sentence in which I cite a peer-reviewed 2010 study in the National Tax Journal by Joshua Rauh entitled, “Are State Public Pension Plans Sustainable?[3] Rauh finds that, without policy changes, Alabama might run out of assets to pay benefits by 2023, necessitating the move to a pay as you go system. To be sure, that is a sobering claim.

The RSA staff argues that the runout date calculated by Rauh is based on “bad data” from 2006, when Alabama offered a 3.5 percent ad hoc Cost of Living Adjustment (COLA). It further contends the runout date is based on the assumption of a risk-free discount rate and asset values from 2009, and this all unfairly inflates liabilities and cherry-picks a low-point for asset values. In addition, Rauh assumes that the plan only pays for normal costs going forward (not for past benefits), in keeping with the contribution behavior of most plans at the time of the study.

The first two claims by the RSA staff are incorrect. In the “run-out dates” paper, Rauh’s data is assembled from, “the individual plans and the Center for Retirement Research on a plan-by-plan basis.”[4] This dataset was originally developed for a previous peer-reviewed paper with Robert Novy-Marx entitled, “Public Pension Promises: How Big Are They and What Are They Worth?” which drew from the individual Comprehensive Annual Financial Reports (CAFRs) of 116 state-sponsored pension plans.[5] Nine data items were taken from the pension plan CAFRs that were available as of December 31, 2008. (The FY 2008 CAFR contains data for 2007 that the authors project to 2009). These CAFR-derived items are:

  • the plans’ stated liability
  • its state-chosen discount rate
  • the actuarial method (EAN or PUC)
  • a benefit factor
  • a Cost of Living Adjustment
  • an inflation assumption
  • the share of active workers in the plan;
  • the share of retired workers in the plan; and
  • the dollar amount of benefits paid in the most recent year.

The third item – the actuarial method – was drawn from both the CAFR and information from the Center for Retirement Research at Boston College as of 2006.[6]

Novy-Marx and Rauh estimated a total of $42 billion projected liabilities as of June 2009 for all three of Alabama’s plans. [7] The authors’ estimate closely matches the reported value of $41.6 billion in September 30, 2009 in RSA’s FY 2010 CAFR. Novy-Marx and Rauh re-calculate the value of state promised pension liabilities when valued based on risk-free Treasury bonds. They find that Alabama’s total liabilities of $42 billion increase to $61.8 billion when discounted using the risk-free Treasury rate.

Their paper triggered a lot of attention. Clearly, the finding that GASB 25 was leading state plans to obscure the true size of their pension liabilities generates a lot of follow-up questions, such as, “When will they run out of money?”

In a subsequent paper Rauh (2010) tackles this very question. His assumptions are key to interpreting the run out date. Beginning with the data that he and Novy-Marx assembled, Rauh models the cash flows of these pension plans under the rate of return assumed by the plan itself, in the case of Alabama: 8 percent. A further assumption is made that future contributions to the plan will be equal in value to the benefits earned by employees in that year, “an assumption broadly in keeping with states’ recent contribution behavior.”[8] If the state fully funds benefits as they are accrued how long will the assets last under the assumption that the plans earn 8 percent each year?

Under an 8 percent discount rate with no COLA, and only funding the normal cost, Rauh projects that the RSA will run out of assets in 2023. The implication is that state contributions will have to increase, placing a greater demand on state budgets, necessitating increased taxes or cuts to spending. One thing going in Alabama’s favor is that they have a history of making the full contribution each year. However, this contribution amount is calculated under optimistic assumptions that I demonstrate in the paper are based on assuming a large amount of investment risk. And that is where the danger lies.

Contrary to the RSA staff’s claim:

  • There is no COLA assumption in Rauh’s 2010 run-out date study
  • The run out date of 2023 is based on a discount rate of 8 percent.

The RSA staff is correct to note that Rauh’s calculation is based on only paying the normal cost. Since Alabama has a history of making the full annual contribution this will help the system to forestall a run-out. The question is by how much, by how many years? As long as the RSA assumes an 8 percent discount rate and embraces a risky investment strategy they are operating under an accounting illusion that leads them to low-ball the annual contribution needed to fund the system.

If the market has a great run over the next decade with returns exceeding 8 percent per year and the RSA continues to to pay 100 percent of the ARC under these conditions it would stay solvent. The RSA points to the fact that between 2009 and today its assets have grown by 46 percent, or $35 billion. [9]

But there’s another problem. The RSA’s funded status continues its decade-long drop. Let’s look at Alabama’s assets, liabilities, and funded status of the plan between 2008 and 2013 (the most recent data available) taken from the plan CAFRs, with no adjustments to the data. The trend is clear. Liabilities are growing faster than the assets. Funding ratios are falling.

For Teachers’ Retirement System (TRS) over the period the total actuarial value of assets fell by six percent from $20.8 billion to $19.6 billion, while total liabilities grew from $26 billion to $29 billion (11 percent), leaving the system with a funded ratio of 66 percent.

Table 1. Teachers Retirement System Actuarial Accrued Liability and Actuarial Assets (2008-2013) Adjusted for Inflation

($ mil) 2008 2009 2010 2011 2012 2013 % change 2008-2014
TRS Liabilities $26,804 $27,537 $28,299 $28,776 $28,251 $29,665 11%
TRS Assets  $20,812 $20,582 $20,132 $19,430 $18,786 $19,629 -6%

Source: Comprehensive Annual Financial Report (CAFR) for Retirement System of Alabama (RSA) FY 2009-2014.

The same story can be told of the Employees Retirement System (ERS). Assets fell by 4 percent as liabilities grew by 11 percent over the period. The ERS is currently funded at 65 percent, down from 77 percent in 2009. Four years of increased returns have not reversed the decline.

Table 2. Employees’ Retirement System Actuarial Accrued Liabilities and Actuarial Assets 2008-2013

($ mil) 2008 2009 2010 2011 2012 2013 % change 2008-2014
ERS Liabilities $13,078 $13,756 $14,248 $14,366 $13,884 $14,536 11%
ERS Assets $9,905 $9,928 $9,739 $9,456 $9,116 $9,546 -4%

Source: Comprehensive Annual Financial Report (CAFR) for Retirement System of Alabama (RSA) FY 2009-2014

The Judicial Retirement Fund (JRF) had the steepest increase in liabilities. Assets fell by 6 percent and liabilities grew by 28 percent. JRF is the most weakly funded at 58 percent.

Table 1. Judicial Retirement System Actuarial Accrued Liability and Actuarial Assets (2008-2013) Adjusted for Inflation

($ mil) 2008 2009 2010 2011 2012 2013 % change 2008-2014
JRF Liabilities $323 $340 $358 $393 $380 $414 28%
JRF Assets $259 $252 $246 $235 $234 $243 -6%

Source: Comprehensive Annual Financial Report (CAFR) for Retirement System of Alabama (RSA) FY 2009-2014

Looking back at the decade shows an even more dramatic trend. These systems began 2003 with funding levels of 90 percent. They have fallen every year since to their current levels of between 66 percent and 58 percent.

The RSA has stated in the past that 80 percent funding is good enough and that investing assets in a risky portfolio currently comprised of 70 percent equities will enable the system to comfortably meet its obligations. But as these funding trends show a volatile portfolio comes with a downside. The assets may be back to where they were five years ago, but in the meantime, liabilities continue their steady growth.

The next observation the RSA staff makes is that these numbers are too bleak since they are based on 2009 asset values. Since then the assets have grown by 11 percent on average over the period. To be sure, once you exclude 2008, things look better. But that’s a bit like excluding the F when you calculate your average grade for the semester. Ignoring the downturn doesn’t mean it didn’t happen or that it didn’t erode the assets. It takes exceptional and sustained performance to make up for it.

The five and 10-year period tell a less bullish story.

Annualized returns for the RSA for the Fiscal Year ended September 30, 2013. (p. 60)

Total Portfolio 1 year Last 3 Years Last 5 Years Last 10 Years
TRS 14.93% 11.45% 6.68% 6.29%
ERS 14.6% 11.4% 6.17% 5.97%
JFR 14.05% 10.89% 8.74% 7.06%

While investments have rebounded for the RSA, plan funding status is falling despite increased contributions. Since 2012 employers and most employees are making bigger contributions to these plans. Alabama now operates a Two-tiered pension system. Tier 1 TRS and ERS employees (those hired before January 1, 2013) saw their individual contributions rates increase from 5 percent of pay in 2011 to 7.5 percent of pay in 2013. JRF members, firefighters, police officers and correctional officers contribution rates increased from 6 percent in 2011 to 8.25 percent of pay in 2013. Tier II members (those hired after January 1, 2013) will have lower contribution rates and diminished benefits. Both tiers will give something up.

Employers are also contributing more. The state’s contributions have increased. For the TRS (Tier 1 employees), the state’s contribution has risen from 6.3 percent of payroll in 2000 to 11.7 percent in 2014. Employer contributions for the ERS (Tier 1) rose from 4 percent to 12 percent of payroll over the same period. JRF has the largest employer contribution “In 2000, the state contribution to the JRF was 21 percent of payroll. It reached 35% by 2014.”

Rauh’s 2010 study points to a trend worth monitoring. Funding levels are dropping. Assets are not growing fast enough to keep up with the growth in liabilities necessitating more revenues, higher contributions or some other action. Yet the RSA staff points to its recent returns of 11%, as if that is something the RSA can sustain. The stock market does reward risk-taking with high returns in bull markets, but at a cost of negative returns in recession years like 2008. Increasing the risk of RSA assets to chase high stock market returns is banking on something neither the RSA nor anyone else can guarantee.

Valuing a guaranteed pension based on the expected returns of risky and volatile assets increases the chance of a funding shortfall. It is likely that Alabama will find it will need more revenue to fund the RSA. Already inadequate funding levels are falling. The investment portfolio is heavily exposed to market risk. And contribution rates are rising.

The RSA staff’s response to my research is part of a more general problem. Many of those responsible for public sector pensions think that investment risk can be ignored or it can just be passed on to taxpayers. The point of this entire body of literature drives home one theme consistently: public sector pension accounting flaunts the established principles of finance by claiming that there is no price for assuming investment risk. Financial theory can be abstract. But recent history gives us a demonstration of these core principles. Many pensions systems, the RSA included, have ignored the lessons of the Great Recession and are exposing pensions to even more investment risk.

[1] Eileen Norcross, “Pension Reform in Alabama: A Case for Economic Accounting,” in Improving Lives in Alabama: A Vision for Economic Freedom and Prosperity, The Johnson Center at Troy University, May 2014 (https://nebula.wsimg.com/35b439dc51fd0dae2bd46e38024dadd2?AccessKeyId=F0B126F45D4E1A4094F7&disposition=0&alloworigin=1)

[2] “Troy University Report on RSA has Erroneous Assumptions,” by RSA Staff, The Advisor, July 2014 (http://www.rsa-al.gov/uploads/files/Advisor_July2014.pdf)

[3] Joshua Rauh, “Are State Public Pension Plans Sustainable? Why the Federal Government Should Worry about State Pension Liabilities,” National Tax Journal 63(3) p. 585-601, May 2010. (http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1596679)

[4] Ibid, p. 6 and p. 9.

[5] Robert Novy-Marx and Joshua Rauh, “Public Pension Promises: How Big Are They and What Are They Worth?” Journal of Finance 66 (4), 1211-1249, 2011 (http://www.jstor.org/stable/29789814?seq=1#page_scan_tab_contents)

[6] Ibid p. 1224, “The actuarial method (item 3) combines our own data collection with information from the state and local pension data made available by the Center for Retirement Research (2006.)

[7] Ibid, p. 1239

[8] Rauh, (2010) “Are Public Pensions Sustainable?” p. 2.


Waking Up Warwick, Rhode Island

Last week an article ran in the Warwick Beacon that was based on a chart I produced. I have since updated the chart to reflect the most recent FY 2011 numbers contained in the FY 2013 budget. (The first chart was based on the original FY 2011 budget.)

The chart shows Warwick, Rhode Island’s municipal budget (excluding the school budget) carved up according to current costs for funding the town’s pension benefits, Other Post Employment Benefits (OPEB), current employee healthcare costs and General Obligation bond payment. The figures come from official budget documents.

My value-added is that I estimate the additional amount needed to fully fund pensions based on the risk-free discount rate. It’s a ballpark estimate backed into based on the plans’ valuation reports. The actuaries, with access to all the plan data, can model the effect of applying the risk-free rate to plan costs more precisely.

It caused quite a stir.

I think the Beacon article shows the myriad problems with how public sector benefits have been valued, accounted for and funded. The piece underscores the misconceptions and cloudy thinking that surround pension finance.

Let’s go through it.

Here’s the question the chart addresses:  If Warwick, Rhode Island were to fully fund employee benefits while paying healthcare costs for current retirees and active employees and make its annual bond payment, how much would be left over to fund everything else in the city budget?

Warwick operates four public pension plans. They are the locally-funded: Firefighters/Police I Pension Plan, Fire II Pension Fund, and Police II Fund. The fourth plan, the Municipal Employees Retirement System (MERS) is jointly funded by the state and participating local governments. (correction: the MERS plan is also locally operated and funded and is distinct from the state-run MERS plan).

Using these plans’ assumed 7.5 percent discount rate to value the liability, only one plan appears to be in deep distress. The Police/Fire I plan is 22 percent funded and requires an annual contribution of around $14 million. The remaining plans seem to be relatively well-funded. Together they add a further $12 million in annual contributions. In total, according to the pension valuation reports for the town of Warwick, fully funding these four pension systems requires an annual contribution of $26.4 million from the city.

Now, when valuing these four systems using the risk-free discount rate, the picture changes. The risk-free rate adds a further $29 million to the annual required contribution. Valuing these plans on a market basis doubles the annual contribution to $55.7 million. That’s 48 percent of the town’s municipal budget in FY 2011.

Employees also receive Other Post Employment Benefits (OPEB) – largely comprised of healthcare – in their retirement. According to the OPEB valuation Warwick spends $7.2 million to pay for current retirees who are now receiving these benefits. (Note, that the budget gives a slightly different total than the OPEB valuation. In the budget,  OPEB ran about $6.68 million in FY 2011)  If the town were to fully fund OPEB benefits for their current workers they would need to contribute a further $14 million. That represents the amortization of OPEB taken from the valuation report.

On top of this Warwick is contributing $12.5  $11.8 million to pay for current employee healthcare benefits in FY 2011 (see, Annual Budget, FY 2013). I added in General Obligation debt (principle and interest payments) of $8.5 million, since I assume this is a non-negotiable expense for the municipality. That leaves Warwick with about 18 percent of its FY 2011 budget remaining.

(Are my numbers wrong? have a look at their reports and let me know if I have made a mistake.)

On the Beacon article, I will underscore three points.

1) The risk-free rate and why it matters. To value and fund a pension plan requires figuring out how much is needed to be set aside today to fund a promised benefit to be paid in future. One must equate the value of a pension to be paid in the future with its value today (the time value of money). That means one must assume a discount rate (i.e., rate of interest) to convert the future value into the present. That enables one to figure out the amount needed to be set aside today to fund tomorrow’s payments.

As most followers of the pension story know, in choosing that discount rate, public sector pension plans look to what rate of return they expect the plan’s assets will return when invested. Most public plans have assumed a rate of return of 8 percent on their assets.

This approach, embedded in GASB 25 and ASOP 27 has been strongly criticized by financial economists as violating several established precepts of economics. Firstly, assets and liabilities should be kept separate for the purpose of valuation, otherwise known as the Modigliani-Miller theorem.

Public pensions represent a secure, government-guaranteed benefit and are not likely to be defaulted upon. Public pensions should be valued like a government bond. The rate to use is the return on Treasury bonds, currently 2.3 percent.

But what policymakers are worked up over is not the economic principles behind discount rate selection. It’s the practical effect that many politicians and plan sponsors protest, as The New York Times story of yesterday highlights.  Lowering the discount rate increases the liability and the amount needed to fund the plan. That has a real impact on the budget, as the Warwick chart shows.

The best and most lucid explanation for market valuation of pensions, I think, can be found in Pension Finance, by M. Barton Waring. (whose intent, I might add, is to save the defined benefit plan.) Other excellent explanations are provided by Douglas Elliot of the Brookings Institute, economists Joshua Rauh and Robert-Novy Marx, Andrew Biggs, and Jeffrey Brown.

David W. Wilcox, the director of research and statistics for the Federal Reserve Board has stated:

 These [public pension benefits] happen to be really simple cash flows to value. They’re free of credit risk. There’s only one conceptually right answer to how you discount those cash flows. You use discount rates that are free of credit risk. This is one of those things where it just really is that simple.*

Now for a few common objections to the risk-free rate. These are perennial and have been very elegantly addressed elsewhere by economists.

2) “But, the private sector uses…”

Private sector defined benefit plans suffer from their own set of accounting and moral hazard problems; and, they use a variety of discount rates for a variety of reasons.

Pension plans governed by the Taft-Hartley Act are collectively bargained-for plans. These plans use the return on assets (7.5%) to value the actuarial liability. According to a March 2012 analysis by Credit Suisse, such discount rate “hocus pocus” means Taft-Hartley plans are now “crawling out of the shadows” with an unfunded liability of about $369 billion when using the corporate bond rate.

Other corporate pension plans are covered by the Pension Protection Act of 2006 and governed by FASB guidance 158. They use a composite return on corporate bonds to value their pensions, currently in the 5 percent range, which is lower than the rate used by public plans. The corporate bond is closer to a low-risk (though not a guaranteed) rate. Public plans carry a stronger guarantee, as they are backed by government, and therefore should be discounted using lower rates than used by the private sector – not a higher one – as is current practice.

 These different guidances explain the plethora of discount rates cited by public plan officials as justification for their current assumptions. And that leads to a great question.

So, why do public and private plans get to value their pension liabilities differently?” (Quick answer: exactly!)

If the Law of One Price is correct (which holds that in an efficient market there must be only one price for similar assets, otherwise opportunities for arbitrage exist) then then salad bar approach to selecting the discount rate is absurd.

The Long Answer:

Actuarial practice has not incorporated the lessons of modern portfolio theory into pension accounting. In the 1960s and into the 1970s the harm was not visible. Pensions were more heavily invested in bonds. The ticking time bomb that ‘high risk’ discount rates  presented to defined benefit plans was not really revealed until the behavior it encouraged began to manifest. These behaviors included the shifting of pension asset portfolios into more risky investments, enhancing benefits, and skipping payments during the 1980s and 1990s. The result was growing funding gaps that accompanied market downturns in the late 1990s, the early 2000’s, and lastly in 2008. Each of these episodes is a demonstration of the problem of valuing liabilities based on risky asset returns.

For some insight into how actuarial science remained largely frozen in time, Jeremy Gold and Lawrence Bader discuss the gap between corporate finance and actuarial practice.

3) “We’ll get the expected 7.5 percent”.

This is another recurring defense of the current public sector accounting. But, an investor doesn’t “get” the expected return. The investor realizes a random and uncertain draw from an increasingly wide distribution of possible realized returns (Waring, 2012).

An oft-expressed rejoinder is,  “…but the market returned an average of 8 percent over the past 20 years.”

This statement alone should be cause for alarm. There is always a chance you will either do better, or worse, than expected. Yet, by virtue of ASOP 27 and GASB 25 the risk of not achieving 8 percent annually, is simply ignored. (Or more accurately it is borne by future taxpayers and younger retirees.) Discounting benefits at a risk-adjusted interest rate captures the cost to taxpayers of having to supplement pensions should projected returns not be realized and the plan’s assets fall short.

The coming years will satisfy a proposition of Waring’s that I think is worth stating again:

Measures of the pension plan based on conventional accounting methods will always follow measures based on economic accounting, even with a lag. The accounting will follow the economics, sooner or later.

The economics on this issue is non-controversial. One can review the work of Nobel-Prize economists Bill Sharpe (one of the developers of the Capital Asset Pricing Model), Robert Merton, (expansion of Black-Schoeles option pricing model), as well as the contributions of finance professors Roger Ibbotson (Yale) and Olivia Mitchell (Wharton, UPenn) for further reading.

The policy message the economics points to is unsettling. Defined benefit plans are in trouble and they will require more funding and difficult budget and policy decisions starting now.

And, who really wants to hear that?

So, the best I can do to drive home the importance of market valuation is to re-state the analogy. You don’t calculate the employer’s annual payment to the pension system based on how the plan’s assets are expected to perform.  Just as you don’t value your home mortgage based on what you think your 401K might do. This video developed by Nobel-Prize winning economist Bill Sharpe makes the case perhaps better than I can do in this blog.


*Wilcox, David. Testimony before the Public Interest Committee Forum sponsored by the American Academy of Actuaries, September 4, 2008. Novy-Marx and Rauh present a similar argument; see Novy-Marx, Robert, and Joshua Rauh, “The Liabilities and Risks of State-Sponsored Pension Plans,” Journal of Economic Perspectives, vol. 23, no. 4 (Fall 2009), pp. 191–210. In analyzing federal employee pensions, the CBO used a discount rate 1 percentage point above the Treasury rate. However, the CBO explicitly noted that this was because federal pensions lack the legal protections that state pension plans like the WRS are entitled to.