ACTUARIAL REVIEW REPORT FOR THE MISSOURI STATE EMPLOYEES RETIREMENT SYSTEM

Transcription

1 ACTUARIAL REVIEW REPORT FOR THE MISSOURI STATE EMPLOYEES RETIREMENT SYSTEM Prepared: March 10, 2017

2 March 10, 2017 Board of Trustees Missouri State Employees Retirement System 907 Wildwood Jefferson City, MO Dear Board Members: Cavanaugh Macdonald Consulting, LLC has performed an independent review of the June 30, 2016 actuarial valuations prepared for the Missouri State Employees Retirement System (MOSERS). As an independent reviewing, or auditing, actuary, we have been asked to express an opinion regarding the reasonableness and accuracy of the actuarial assumptions, actuarial cost methods, and valuation results for the Missouri State Employees Plan (MSEP) and the plan for judges (Judicial Plan). The retained actuary for the System is Gabriel Roeder Smith & Company (GRS). Our analysis of the actuarial assumptions and methods was based largely on the most recent experience studies prepared for the period from July 1, 2010 to June 30, Our opinion on the valuation results was based on a replication of the June 30, 2016 actuarial valuations for MSEP and a review of the June 30, 2016 Judicial Plan. We generally find the actuarial valuation results to be reasonable and accurate based on the assumptions and methods used. The valuation was performed by qualified actuaries and was performed in accordance with the principles and practices prescribed by the Actuarial Standards Board. This report documents the detailed results of our review. The undersigned are members of the American Academy of Actuaries and meet the Qualification Standards of the American Academy of Actuaries to render the actuarial opinion contained in this report. We are available to answer any questions on the material in this report or to provide explanations or further details as appropriate. Sincerely, Cavanaugh Macdonald C O N S U L T I N G, L L C The experience and dedication you deserve Joseph A. Nichols, ASA, EA, MAAA, MSPA, FCA Consulting Actuary Patrice A. Beckham, FSA, FCA, MAAA, EA Principal and Consulting Actuary Brent A. Banister, PhD, FSA, FCA, MAAA, EA Chief Pension Actuary 3906 Raynor Pkwy, Suite 106, Bellevue, NE Phone (402) Fax (402) Offices in Englewood, CO Off Kennesaw, GA Bellevue, NE

3 TABLE OF CONTENTS Page 1. Executive Summary Actuarial Assumptions... 4 Background on Actuarial Assumptions... 4 Economic Assumptions... 5 Demographic Assumptions Actuarial Methods Actuarial Cost Method Asset Valuation Method Amortization of Unfunded Actuarial Accrued Liability Method Data Review Actuarial Valuation Results Review Valuation Report Review GASB Reporting Review Actuarial Fee Review... 27

4 1. EXECUTIVE SUMMARY As an independent auditing actuary, Cavanaugh Macdonald Consulting, LLC (CMC) has been tasked to provide a general overview and express an opinion of the reasonableness and soundness of the work performed by Gabriel Roeder Smith & Company (GRS) for the MOSERS. The work to be reviewed includes the June 30, 2016 actuarial valuations for MSEP and the Judicial Plan, along with the most recent experience studies presented in March The specific items to be included in the actuarial audit, as outlined in the Request for Proposal, include verification and evaluation of: 1. Valuation results, including a review of the demographic and financial information used by the retained actuary in the valuation; 2. Actuarial assumptions for reasonableness and compliance with Actuarial Standards; 3. Experience study assumption recommendations; 4. Actuarial cost method and asset valuation method for reasonableness and appropriateness for MOSERS purposes, compliance with Actuarial Standards of Practice, and compliance with reporting and disclosures requirements set by the Governmental Accounting Standards Board. 5. Requirements of state law, i.e. compliance with applicable sections of Missouri Statutes; 6. Calculation of unfunded actuarial accrued liability, amortization period, and resulting contribution rates; 7. Actuary s reports for accuracy and completeness. Verification that the valuations were performed by qualified actuaries and in accordance with principles and practices prescribed by the Actuarial Standards Board. 8. Evaluation of fees paid to the retained actuary with respect to whether they are reasonable and proper based on the level of services received; 9. Any recommendations for alternative methods. We requested full member data and reports not available on the website from MOSERS. We also requested member data, as reconciled for the 2016 valuations, from GRS and complete descriptions of assumptions, methods and valuation procedures as well as sample life information. It is our belief that an audit should not focus on finding trivial differences between actuarial processes, procedures, philosophies, or styles utilized by two different actuaries, but rather to verify there are no material errors, and to identify potential improvements to the process and procedures utilized by the System s actuary. Because actuarial work draws on professional judgment, there is a subjective component that must be considered alongside the objective component of matching numerical results. In performing this audit, we attempt to limit discussions 1

5 1. EXECUTIVE SUMMARY concerning stylistic preferences and focus more on the significant philosophical approaches, the accuracy of calculations, the completeness and reliability of reporting, and the compliance with generally acceptable actuarial practices and standards of practice in all of the work reviewed. As described in our report, we have determined that the actuarial methods, assumptions, processes, and reports are consistent with the applicable Actuarial Standards of Practice (ASOPs) and our understanding of GASB. Throughout the report, we have noted a few issues where we believe there are opportunities for improvement. We would like to thank GRS for their cooperation and assistance in providing the required information for our review. We would also like to mention that many of the recommendations made in previous actuarial audits, especially those related to the valuation reports, have been incorporated by GRS. The prior recommendations that have not been adopted, but where we believe they are still appropriate, have specifically been noted. In Section 2 of our report, we analyze the set of actuarial assumptions used by GRS. The actuarial assumptions are a critical component of the valuation process and, thus, were reviewed as part of the audit. We believe the assumptions to be generally reasonable and appropriate for the valuations, but we have comments and recommendations, for consideration in the next experience study. (1) Recommendation 1: In the analysis for determining the real rate of return, specific data provided by Summit Strategies and the MOSERS investment staff should be presented. Also, in our opinion, the investment return data provided by Summit Strategies should be considered in determining the assumption (with greater weight than the other consultants) and not just used as a comparison to the overall analysis. (2) Recommendation 2: Additional explanation should be provided as to how the actual wage growth of 1.50%, which was used to isolate the actual merit-based pay in the study period, was determined. (3) Recommendation 3: As recommended in previous reviews, we feel that analysis should be conducted to present information on pure service-based withdrawal rates even if the approach is not ultimately recommended. (4) Recommendation 4: The experience study should include detailed analysis on the assumption regarding the selection of plan benefit provisions by MSEP members and the impact of the BackDROP on retirement experience. Our comments and suggestions reflect our ideas and perspectives. We acknowledge that setting assumptions is as much an art as it is a science so there is room for differences of opinion. In Section 3 of our report, we review the actuarial methods that are used to develop the actuarial contribution rate: the asset valuation method, the actuarial cost method, and the amortization policy. We believe the actuarial methods being used are reasonable. 2

6 1. EXECUTIVE SUMMARY In Section 4 of our report, we compare the data used by GRS with data provided by MOSERS. We find that the data is consistent and appropriate. In Section 5 of our report, the results of our independent calculations of the liabilities of MSEP are compared with the results prepared by GRS. We were able to match all liabilities, by decrement, within a reasonable range. We find the calculation results to be reasonable and appropriate for their intended purposes. In Section 6, we provide our comments on the valuation reports produced by GRS. We found the reports to be in compliance with the ASOPs, but offer some minor suggestions for improvement. Most of the suggestions involve the disclosure of some actuarial assumptions not currently included in the report. In Section 7, we discuss our review of the GASB reporting. Our review confirmed that the determination of the single equivalent interest rate for both plans was properly performed and that calculations were accurate. Because of the complexity of actuarial work, we would not expect to match GRS valuation results exactly, nor would we necessarily expect our opinions regarding the selection of assumptions and methods to be the same as those of GRS. While we offer some different viewpoints or ideas, we believe that GRS work provides an appropriate assessment of the health and funding requirements of both valuations we reviewed. The remainder of this report provides the basis for our findings for each of the requested tasks, including our recommendations. 3

7 2. ACTUARIAL ASSUMPTIONS BACKGROUND ON ACTUARIAL ASSUMPTIONS The actuarial assumptions form the basis of any actuarial valuation or cost study. Since it is not possible to know in advance how each member s career will evolve in terms of salary growth, future service, and cause of termination, the actuary must develop assumptions in an attempt to estimate future patterns. These assumptions enable the actuary to value the amount of benefits earned and to reasonably estimate when and how long these benefits will be paid. Similarly, the actuary must make an assumption about future investment earnings of the trust fund. In developing the assumptions, the actuary examines the past experience and considers future expectations to make the best estimate of the anticipated experience under the plan. There are two general types of actuarial assumptions: Economic assumptions these include the investment return assumption (expected return on plan assets), assumed rates of salary increase, price inflation, wage inflation, and increases in total payroll. The selection of economic assumptions should conform to ASOP No. 27 Selection of Economic Assumptions for Measuring Pension Obligations. Demographic assumptions these include the assumed rates of retirement, mortality, termination, and disability. The selection of demographic assumptions should conform to ASOP No. 35 Selection of Demographic and Other Noneconomic Assumptions for Measuring Pension Obligations. Different actuaries may have different philosophies when it comes to evaluating the experience study data and recommending changes to assumptions. When we see significant changes in experience, we consider them carefully and try to discern why the dramatic change occurred. If we believe the observed data is not an aberration and should be considered as credible, we typically recommend rates somewhere between the old rates and the new actual experience we rarely change the assumption to fully match the observed experience. If actual experience during the next experience study period shows the same result, we will recognize the trend at that point in time or at least move further in the direction of the observed experience. On the other hand, if experience returns closer to its prior level, we will not have overacted, possibly causing unnecessary volatility in the actuarial contribution rates and funded ratios. Many of GRS recommendations were closely based on the actual observed experience, indicating that heavy credibility was assigned to the study period. This difference in philosophy may drive some of our comments. In our audit work, we have seen a number of other actuarial firms present their experience study results solely in a presentation, i.e., no formal report. We commend GRS on preparing a formal report that is complete and provides some discussion and explanation of the recommended changes. Although a formal report is not strictly required by actuarial standards, we believe this is a best practice and allows for a more complete explanation and justification regarding decisions to keep or change a given assumption, as well as providing a permanent record. 4

8 2. ACTUARIAL ASSUMPTIONS ECONOMIC ASSUMPTIONS Actuarial Standards of Practice (ASOPs) are issued by the Actuarial Standards Board to provide guidance to actuaries with respect to certain aspects of performing their work. As mentioned earlier, ASOP 27 is the standard that addresses the selection of, or recommendations regarding, economic assumptions for measuring pension obligations (liabilities) under defined benefit plans. The prior and recommended economic assumptions in the Experience Study report were: GRS Recommendation Prior (2015) Assumption Price inflation 2.50% 2.50% Real rate of return (net of expenses) 5.20% 5.50% Investment return 7.70%* 8.00% Total payroll growth 3.00%** 3.00% Price inflation 2.50% 2.50% Real wage growth (productivity) 0.50% 0.50% Total wage growth 3.00% 3.00% *Board ultimately adopted 7.65% **Not specifically stated in experience study. Each assumption is briefly discussed in the following narrative: Price Inflation: The price inflation assumption is a key assumption because it impacts both the assumption for the rates of salary increase (individual as well as total payroll) and the investment return assumption. The underlying price inflation component in all economic assumptions must be consistent, in accordance with the guidance provided in ASOP 27. Historical patterns of inflation show a long-term average of around 2.70% (last 30 years). Shorter timeframes of 5, 10, 20 years show inflation of 1.70% to 2.20%, below the 2.70% long-term average. Inflation has varied significantly over time, with some notably high periods in the 1970 s influencing the long-term average. Inflation has been consistently below the long-term average of 2.70% and the financial markets pricing of inflation (comparing Treasuries and TIPS) suggests that trend is expected to continue for the next 30 years. However, these results may be partially driven by the actions of the Federal Reserve Bank and, therefore, may not be indicative of the long-term estimation that actuaries need for their work. For a longer time frame, actuaries often consider the expected increase in the CPI used by the Office of the Chief Actuary for the Social Security Administration (SSA). As mentioned in the GRS report, the July 2015 SSA report estimated the ultimate projected annual increase in the CPI 5

9 2. ACTUARIAL ASSUMPTIONS over the next 75 years to be 2.7% under the intermediate cost assumptions. The lower cost assumption used a forecast of 2.0% and the high cost assumption was 3.4%. (For informational purposes, the 2016 Social Security report reflected an intermediate inflation assumption of 2.6%). While there can be arguments made for assuming inflation will remain low for a very long period of time, we note that inflation can be significantly affected by monetary and fiscal policy, and those policies may change dramatically and rapidly. Consequently, there are also some strong arguments for assuming that inflation could increase at some point in the future. In the experience study, GRS cites the following data to support their recommendation to keep the inflation assumption of 2.50%: current market pricing (bond market), actual inflation over the last 20 and 30 years, the inflation assumptions used by the Chief Actuary of the Social Security Administration, and the Federal Reserve s target for inflation. We believe that GRS provided ample data and explanation to support their recommendation to keep the inflation assumption at 2.50% and we believe it is reasonable. Investment Return Assumption: The investment return assumption (also called the valuation interest rate) should represent the long-term rate of return expected on the plan assets, considering the asset allocation, the real rate of return on each asset class, and the underlying inflation rate, net of investment expenses required to earn that return. The long-term relationship between price inflation and investment return has long been recognized by economists. The basic principle is that the investor demands a more or less level real return the excess of actual investment return over price inflation. If inflation rates are expected to be high, investment return rates are also expected to be high, while low inflation rates will result in lower expected investment returns, at least in the long run. The period considered for pension funding represents a very long time horizon. In reviewing this assumption, the actuary should consider asset allocation policy, historical returns, and expectations of future returns. Frequently, investment advisors focus on a period of the next 5 to 10 years since they are most concerned with how to invest the funds currently to maximize returns. The longer term is less relevant to them, but it is, of course, paramount to actuaries who are projecting benefits to be paid for the next 50 to 100 years. This difference in perspective can significantly influence how investment advisors and actuaries derive an investment return assumption and result in significant differences. GRS key analysis for the investment return assumption involved developing the expected return for the portfolio using the system s target asset allocation and forward-looking capital market assumptions. This is a common approach, which we believe is the best practice, and is standard practice for most public plan actuaries. For their analysis, GRS utilized the capital market assumptions of eight major investment consulting firms for their real rate of return analysis. Although not specified, it is assumed that the capital market data was for the 2016 calendar year. 6

10 2. ACTUARIAL ASSUMPTIONS One limitation of this approach is that the asset classes used in the target allocation may not perfectly align with the asset classes provided by the various investment advisors. The tables of results shown on pages C-7 and C-8 of the experience study report shows both the geometric and arithmetic returns. There are two additional components of the GRS analysis that are given weight in the analysis, but only by casual mention, and only as a general comparison. One component is information from Summit Strategies (MOSERS investment consultant) and the other is analysis made by MOSERS Investment Staff. GRS mentions that, in order to keep their analysis independent, they limited their discussion with MOSER investment staff to a confirmation level. We believe that providing detailed information on MOSERS expected return of approximately 7.70% would have been helpful to the reader. As for the information from Summit Strategies, there was no data provided specific to the MOSERS investments. We typically see information for the system s investment consultant separately identified with more weight applied in the evaluation process. This is due to the fact that the System s consultant has more credibility arising from its in-depth knowledge of the system s investments. Including the Summit Strategies returns, along with their inflation assumption would have provided greater insight into the appropriateness of the recommendation, and at least given some insight into the comparisons made by GRS. Based on the data and explanation included in the report, we find the 7.70% investment return assumption recommendation to be reasonable. We note that the recommended return of 7.70% maintains the use of a real rate of return of 5.20% (2.50% inflation % real return). We would note that the Board ultimately selected an investment return that differed from the recommendation, but this does not impact the validity of GRS analysis and recommendation. Given the analysis in the experience study report, the current assumption of 7.70% is reasonable as is the assumption ultimately adopted by the Board (7.65%). Salary Increase Assumption The individual salary increase assumption (salary scale) is used to project future salary increases for individual members. It consists of a wage inflation assumption that represents general wage increases in the overall economy plus a component for merit and promotion increases, typically in the earlier part of a member s career. These are each discussed below. Wage Inflation Assumption: GRS recommends keeping the wage inflation assumption at 3.00% (0.50% greater than the 2.50% price inflation). They use median wage and price inflation data during the period from 1990 to 2015 as the basis for their recommendation. The data includes information from the Social Security Administration (SSA) and directly from actual MOSERS experience. 7

11 2. ACTUARIAL ASSUMPTIONS The SSA data comes from the Social Security Administration Trustees Report, which contains many macro-economic indexes, including the Average National Wage. This index is commonly used for wage growth purposes. However, the Average National Wage is based on all wage earners in the US so it can be influenced by the mix of jobs (full-time vs part-time, manufacturing vs service, etc.) as well as by changes in certain segments of the workforce that are not seen in all segments (e.g. regional changes). In addition, if compensation is shifted between wages and benefits, the wage index would not accurately reflect the increases in total compensation. Based on the information presented, we believe that 3.00% is a reasonable assumption. Merit Salary Scale Assumption: The merit salary scale reflects the portion of salary increases provided at the individual member level, including promotion, increased skills, longevity pay, additional training, and other similar items. These increases typically occur earlier in a member s career, decreasing rather quickly and entirely disappearing over time. In order to determine the actual merit-based portion of the salary increases during the experience period, the general wage increase portion of the pay increases is subtracted from the total pay increases. Due to the fact that only total pay is reported, the wage increase portion during the study period must be estimated. One approach to determine the general wage increase portion is to look at the total salary increases for active members towards the end of their careers, generally assuming that there is no merit component for this group. GRS determined the actual merit-based salary increases by subtracting 1.50% from the actual pay increases. Although we agree with the subtracted amount based on actual wage increases, there is inadequate documentation of the process used to determine the 1.50% amount. It simply shows up in a footnote on an exhibit. In the Judges Plan, there is no mention of the assumed wage inflation that was used to determine actual merit increases over the experience period. We assume it is the same 1.50% amount used in MSEP, but that is not disclosed. Although we are not suggesting that 1.50% is not a proper estimate, an explanation of the reasoning behind its use is appropriate. It is very common for the merit salary scale in large, public plans to be developed as a servicebased assumption where the merit assumption varies solely with a member s years of service. In our 2012 actuarial audit, we suggested that GRS consider such a service-based merit salary assumption in the next experience study. We were pleased to see that GRS followed this advice, and ultimately recommended a service-based merit salary increase assumption. The recommended assumption fits the actual experience nicely and we believe it is reasonable. We would note that, when comparing the recommended rates to the actual rates, GRS also shows current rates. Since the 2015 valuation used age-based rates, there were no service-based current rates and GRS provides no discussion on how the displayed present rates were determined. We concur with the recommended service-based salary increases. 8

12 2. ACTUARIAL ASSUMPTIONS Payroll Growth Assumption The unfunded actuarial accrued liability (UAAL) is amortized as a level percentage of payroll over the amortization period. As a result, a payroll growth assumption is necessary to develop the UAAL contribution rate. Given a stable active population, the payroll growth assumption is typically set equal to the wage inflation assumption. Although it is common to use the wage inflation assumption as the payroll growth assumption, some plan use a lower payroll growth assumption. The experience study did not include a specific recommendation for the payroll growth assumption and we would suggest that be added in the next study. DEMOGRAPHIC ASSUMPTIONS The major demographic assumptions are the assumed rates of retirement, termination of employment (with or without a vested benefit), disability, and mortality (death before or after retirement). In the following paragraphs, we make specific comments on each demographic assumptions. Please note that GRS went through the same demographic analysis for the Judicial Plan as they did for MSEP. We generally agree with GRS analysis and recommendations for demographic assumptions for the Judicial Plan, and due to the very small number of members, our discussion below is limited to the analysis in MSEP only. We would note that in our last actuarial audit, we recommended that GRS consider a mortality assumption for Judges that reflects longer life expectancy, given the strong correlation between education/income and mortality. GRS followed that suggestion and recommended an assumption for Judges that reflects a longer life expectancy, when compared to the MSEP participants. Rates of Mortality: One of the most important demographic assumptions in pension valuations is post-retirement mortality because it projects how long benefit payments are expected to be made. The longer retirees live and receive benefits, the larger the liability of the system, thus increasing the contributions required to fund the system. In addition, if members live longer than expected based on the assumption, the true cost of future benefit obligations will be understated and contributions will increase over time as the unfavorable experience unfolds. Because of potential differences in mortality, healthy retirees, disabled participants, and active members are usually studied separately. The mortality assumption applies to members both before and after retirement, although the post-retirement mortality assumption has a far greater impact on valuation results of the retirement system. Most often, gender distinct rates are used since studies consistently show that females live longer than males (although that gap has been shrinking according to recent mortality studies). It is commonly recognized that rates of mortality have been declining through time, which means people, in general, are living longer. ASOP 35, Selection of Demographic and Other Noneconomic Assumptions for Measuring Pension Obligations, requires the actuary to include 9

13 2. ACTUARIAL ASSUMPTIONS an assumption as to expected mortality improvement (even if the improvement is assumed to be none) after the measurement date. It further requires the actuary to disclose what, if any, future mortality improvements are assumed and how the improvements are reflected in the mortality assumption. There are two approaches to anticipating future improvements in mortality: (1) Setting the mortality assumption so that it includes a margin, and (2) Using the generational mortality improvements. The first approach intentionally selects a mortality table with lower mortality rates than are currently being observed, thus providing room for mortality improvements in the future. Under this approach, the Actual/Expected (A/E) ratio is well over 100% initially because the actual deaths are higher than expected by the mortality table (creating margin, cushion for mortality to improve). Under the second approach (generational mortality), the starting mortality rates are set close to the observed experience (with resulting A/E ratios around 100%), and then future improvements are directly reflected by applying a mortality improvement scale to the mortality rates in each future year to reduce the probability of death. Under the generational approach, the greatest change in life expectancy is reflected for younger members who have more years of future mortality improvement. In the experience study report, GRS has a very detailed explanation of the adjustments that are made to the tables, including how the projection scale is applied and how the load factor is determined. Retiree Mortality Generally, our preferred approach for reflecting mortality improvements is to use generational mortality. Therefore, we concur with GRS recommendation in the experience study to change to generational mortality. However, we recognize that there are definite differences of opinion on how future mortality improvements will unfold over time even among experts. GRS recommendation was to use the most recent mortality table published by the Society of Actuaries, the RP-2014 Health Annuitant Mortality Table for males and females, with future mortality improvements anticipated using Scale MP The resulting A/E ratios were generally near 100% which is appropriate when using generational mortality. After further discussions, the Board ultimately adopted a static mortality table, created by projecting mortality improvements from 2006 to 2026 with Scale MP-2015 instead of moving to a fully generational approach. The resulting A/E ratio was about 116% which does provide room for future mortality improvements. Based on our conversation with GRS, the choice to remain with a static table was due to skepticism regarding continuing mortality improvement in the long term. The static mortality table was chosen to provide for some improvements with the idea that the assumption would continue to be reviewed in future experience studies and adjusted at that 10

14 2. ACTUARIAL ASSUMPTIONS time, if appropriate. This is an acceptable approach and the assumption is reasonable. We also note that the use of a static mortality table is still commonly used by public plans. For analysis of mortality, GRS determined the actual to expected ratios on a benefit-weighted basis, as recommended in our last audit; i.e., a greater weight is given to participants with higher benefits. Because there tends to be a correlation between benefit size and longevity, weighting the mortality analysis by benefit amount helps to ensure that the assumption is a good fit for measuring the retiree liability, and not simply providing an estimate of the number of retirees dying. For example, if two individuals are the same age, the one receiving a $1,500 monthly benefit has twice the influence on the liabilities as does an individual with a $750 monthly benefit. In addition, the Society of Actuaries tables (such as the RP-2014 Table) are developed in this benefit-weighted manner so it is appropriate to analyze the actual experience on this basis. We have seen varying results when benefit-weighted analysis is performed, confirming the importance of this approach. We agree with GRS benefit-weighted method for mortality analysis. Active Mortality Rates The active mortality tables presented in the experience study are the employee version of RP-2014 tables with a 95% load for males and a 90% load for females. GRS noted that there was insufficient data for active mortality, but then used the data to make adjustments to the basic table. GRS gave very little indication as to how the load factors were determined. These assumptions resulted in an A/E ratio of 94% for males and 90% for females. The probability of active death is low so the actual number of deaths tends to be small and variability often occurs. Overall, the recommended assumptions fit both the male and female experience fairly well. Overall, the assumption is reasonable. Disabled Mortality Rates The disabled mortality tables recommended in the experience study are the disabled versions of RP-2014 tables with a 95% load for males and a 90% load for females. GRS probably chose the RP-2014 disability mortality table as a companion to the employee and retire tables. However, the experience study includes no data, and very little discussion, of how the load factors were chosen. That being said, it is very likely that with the appropriateness of using RP-2014 tables for actives and retirees, the use of the RP-2014 disability mortality table for the disabled mortality assumption is reasonable. Disability Rates The disability tables recommended in the experience study resulted in an A/E ratio of 95% for males and 96% for females, indicating actual experience was assigned full credibility. The prior disability table assumed a higher level of disabilities in the aggregate than occurred. As a result, all the rates were shifted downward, with the exception of the age groups where disabilities were assumed too low using the prior table (ages for males and ages for females). 11

15 2. ACTUARIAL ASSUMPTIONS A/E Ratio Current Recommended Male Select 83% 95% Female Select 78% 96% Although we might not have lowered rates as far as GRS, we believe the assumption is reasonable based on the available data. Rates of Termination Termination rates reflect members who leave for any reason other than death, disability or service retirement, whether voluntary or not. The MOSERS plans use a select and ultimate assumption. This means that one assumption is used during the select period (generally the first five years of employment for MOSERS) and a different assumption regarding termination applies after the expiration of the select period (referred to as the ultimate period). For MOSERS, GRS uses agebased rates during the ultimate period. In our last actuarial audit, we recommended that GRS at least review pure service-based withdrawal rates. We repeat that recommendation as we continue to see evidence that points to a much higher correlation to service, versus age, when looking at the probability of terminating service. The termination rates during the select period vary for males and females. Based on the actual experience, GRS recommended certain changes to the termination rates. A summary of the A/E ratios on the current and recommended assumptions are summarized below: A/E Ratio Current Recommended Male Select 113% 109% Male Ultimate 108% 104% Female Select 113% 108% Female - Ultimate 114% 109% Overall, the recommended changes are reasonable, but we repeat our recommendation that the next experience study include analysis using pure service-based termination rates. 12

16 2. ACTUARIAL ASSUMPTIONS Rates of Retirement: Separate analysis is performed for four different eligibility stages of retirement early retirement, first year eligible for normal (unreduced) retirement, second year eligible and third year eligible and beyond. The A/E for the recommended assumption for each stage is as follows: A/E Ratio Current Recommended Year 1 Eligibility 91% 97% Year 2 Eligibility 89% 93% Year 3 + Eligibility 97% 97% Early Retirement 93% 98% The recommended assumptions closely follow the actual experience, as indicated by A/E ratios near 100%. The changes to the retirement rates are reasonable based on the data. There are two assumptions regarding retirement that are absent from the experience study. One of these is the election of the MSEP or MSEP 2000 benefit structure for those eligible to make the election. This assumption could have significant impact on the calculation of liabilities. In the valuation report, we notice that there is a heavily emphasized assumption regarding the selection of benefits for MSEP members. It is as follows: The liabilities for active members hired on or after January 1, 2011 were based on MSEP 2011 benefits. The liabilities for active members hired on or after July 1, 2000 (April 26, 2005 for Administrative Law Judges) were based on MSEP 2000 benefits. The liabilities for active members hired before July 1, 2000 for Elected Officials, General Assembly, and Uniformed Water Patrol were based on MSEP benefits. The liabilities for all other active members hired before July 1, 2000 were based on the assumption that members would elect MSEP 2000 prior to age 62 and MSEP on or after age 62. In the experience study, GRS points out that they reviewed this assumption by comparing the liabilities for MSEP to those of MSEP 2000 for each individual and that the difference was 0.20%. What is not clear is if the total expected MSEP liabilities were compared to the total MSEP 2000 liabilities (by individual), or if the comparisons were made at the individual decrement age level. We feel that since this is an important assumption in valuing the benefits, data should be presented detailing the comparison and that, in addition to the liability comparison, experience regarding the actual elections of MSEP and MSEP 2000 benefits should be reviewed and presented. The second is an assumption regarding the election of the BackDROP, including the duration of the election period. There are many factors that may lead to a retiring member choosing the 13

17 2. ACTUARIAL ASSUMPTIONS BackDROP age, most recent salary, and inflation experience, among others. We recommend that an analysis of this data be included in the next experience study. Miscellaneous Assumptions: In the valuation process, there are a number of other assumptions that are required for programming purposes that are fairly minor in significance and often difficult to measure. In these cases, it is reasonable to use some rough analysis or even simply professional judgment in setting these assumptions. GRS s experience study report identifies several of these assumptions related to marriage rates, spousal age differences and election of refund. Overall, we find their recommended assumptions to be reasonable. 14

18 3. ACTUARIAL METHODS ACTUARIAL COST METHOD For all pension plans, whether defined benefit or defined contribution, the basic retirement funding equation is: Where: C = employer and member contributions I = investment income B = benefits paid E = expenses paid from the fund, if any. C + I = B + E As can be seen from the formula, for a given level of benefits and expenses the greater I is, the smaller C is. This is the underlying reason for advance funding a pension plan, and, historically, investment income pays for 60% to 70% of the benefit dollars received by plan members. In other words, for every dollar paid to a member only 30 to 40 cents comes from contributions. Of course, the problem with the formula is that in order to figure out exactly how much to contribute, the plan would have to be closed to new members and allowed to operate until all retirees were deceased. At that point, the benefits and expenses actually paid out, and the investment income actually earned, would be known, and, using the equation above, the true cost could be determined. Since the vast majority of plans are ongoing and have no intention of closing, and since even with a closed plan it takes a very long time before all benefits are finally paid out, plan sponsors hire actuaries to estimate the cost of their plans and to create a budget for systematic contributions to meet that cost. In order to determine the contributions needed, the actuary s first step is to estimate on a given date (the valuation date) the value of all benefits (and expenses) that will be paid to the existing active and retired membership over their remaining lifetimes based on the plan s current benefit structure. This estimation requires the use of assumptions regarding both future events (termination, disability, retirement, death, etc.) and future economic conditions (return on assets, inflation, salary growth, etc.). The MOSERS assumptions were covered in the previous section. By combining the assumptions for future events and the salary growth assumption, the actuary generates an expected benefit payment stream. In other words, a string of annual payments expected to be made to the current active and retired members from the valuation date until all members are no longer living. Then the actuary applies the investment return assumption to discount each year s payments to the valuation date, creating the present value of all future benefits or the total liability of the plan. 15

19 3. ACTUARIAL METHODS The difference between the total liability and the current assets of the plan represents the present value of future contributions (PVFC) that have to be made by either members or the employers. Usually the members and employers cannot contribute the entire difference in one year, but rather desire a relatively smooth contribution pattern over time that also meets any external constraints. In order to budget for the PVFC, the actuary applies an actuarial cost method. There are several acceptable cost methods, but it s important to recognize that they are nothing more than budgeting tools. Different actuarial cost methods can provide for faster funding earlier in a plan s existence, more level funding over time, or more flexibility in funding. The choice of an actuarial cost method will determine the pattern or pace of the funding and, therefore, should be linked to the long-term financing objectives of the system and benefit security considerations. The actuarial cost method used by GRS for both MOSERS defined benefit plans is the entry age normal method. This cost method determines the normal cost as a level percentage of pay which, if paid from entry into the plan to the last assumed retirement age, will accumulate to an amount sufficient to pay the expected benefit. Entry age normal tends to result in reasonably stable contribution rates, a feature that is desirable for many public plans. An additional cost is determined by amortizing the unfunded actuarial accrued liability (discussed later in this section). The entry age normal cost method is also the method specified by GASB for financial reporting under GASB Statements 67 and 68. Entry age normal is the most common cost method used by public plans and we completely agree with its use for MOSERS. Entry age is also the method specified in Missouri Statutes as the required method for both MOSERS plans. ASSET VALUATION METHOD Since the purpose of actuarial funding is to build up an asset pool (remember the importance of I in C + I = B + E ) actuaries need to value the current asset pool on each valuation date. The market value could be used, but it would tend to create too much volatility from valuation date to valuation date, and a single day s measurement is not necessarily indicative of the true underlying value of the investments held by the plan. Thus, most actuaries use an asset valuation method which smoothes out these fluctuations in pursuit of achieving more stable funding measures and (when relevant) developing more level contributions. A good asset valuation method places values on a plan s assets which are related to current market value, but which will also produce a smooth pattern of costs. This is a question of balancing fit (measured against market value) and smoothness. Neither book nor market value of assets is generally felt to be appropriate in determining the actuarial contribution rate for an ongoing pension plan. Book value produces smooth, predictable employer contributions, but it ignores sizeable appreciation and is not a good measure of the fund s true value (i.e., a poor fit to market value). On the other hand, market value is a realistic current measure of the fund but, on a long-term basis, one day s market value may not be a very meaningful figure for a pension fund. Furthermore, sharp short-term swings in market value can 16

20 3. ACTUARIAL METHODS result in large fluctuations in the computed employer contributions required to fund the plan (i.e., not very smooth). The goal of the actuarial asset valuation method is to smooth or reduce investment market fluctuations. This is particularly important during periods of volatile capital markets in which abrupt changes in asset values, when factored into the funding valuation, produce sudden unnecessary changes in contribution levels. In this case, unnecessary implies that the change in asset values is not necessarily a true revaluing of the assets involved, but rather a fluctuation reflecting a current economic climate or a short-term reaction to specific news. In our opinion, desirable characteristics of an actuarial asset valuation method include the following: The method should be simple to operate. It should be readily calculable from financial statements. The method should be easy to explain to all interested parties. The theoretical underpinnings should be solid and not produce a long-term lag to the fair value of assets. The value produced should account for market values. The method should smooth the effect of market fluctuations. Investment decisions should not be affected by the actuarial asset valuation method, and vice versa. The value produced should be realistic; the price tag placed on assets should be sensible and should not cause other variables to be adjusted to account for unrealistic asset values. MOSERS Asset Valuation Method: The asset valuation method used by GRS in the valuation finds the dollar amount of difference between the expected actuarial value of assets and the actual market value of assets. The resulting gain or loss is then recognized 20% per year over five years. We found that the description of the asset smoothing method in the valuation reports accurately portrays the method used in the calculation. This method of spreading the difference between the expected and actual investment earnings is a feature typically found in smoothing methods. The difference between the actuarial value of assets and market value of assets should be equal to the existing unrecognized gains and losses. As a matter of difference, we typically determine the investment gain or loss to be spread as the difference between the expected investment earnings on market value minus the actual expected earnings on market value. The MOSERS method starts with the expected investment earnings on actuarial value of assets. This is a small difference, but one that should be explained with a 17

21 3. ACTUARIAL METHODS comment that states, in order for the actuarial and market value of assets to converge over a short term, the actuarial value of assets must earn the assumed investment return. Compliance with ASOP 44 Actuarial Standard of Practice Number 44, Selection and Use of Asset Valuation Methods for Pension Valuations, provides guidance to the actuary when selecting an asset valuation method for purposes of a defined benefit pension plan actuarial valuation. When considering the use of an asset valuation method other than market value, ASOP 44 states that the actuary should select an asset valuation method that is designed to produce actuarial values of assets that bear a reasonable relationship to the corresponding market values. Further guidance states that the asset valuation method must satisfy both of the following criteria: (a) The asset values fall within a reasonable range around the corresponding market value, AND (b) Any differences between the actuarial value of assets and the market value of assets are recognized within a reasonable period of time. In lieu of satisfying both (a) and (b) above, an asset valuation method meets ASOP 44 requirements if, in the actuary s professional judgment, the asset valuation method either: (i) Produces values within a sufficiently narrow range around market value, OR (ii) Recognizes differences from market value in a sufficiently short period. Several of the terms in the criteria of ASOP 44 such as reasonable and sufficiently narrow are not well defined. As a result, actuaries can differ in their opinion on these matters. As we consider the current asset valuation method used by MOSERS in light of ASOP 44, we believe it satisfies these requirements. We find the asset valuation method to be reasonable and in accordance with actuarial standards. AMORTIZATION OF UNFUNDED ACTUARIAL ACCRUED LIABILITY METHOD The UAAL amortization method is comprised of three main components: the number of years, a closed period and a payment determined as a level percentage of pay. The following discussion applies to plans that are funded by contributing the actuarial contribution rate, the result of the application of the various actuarial methods. A closed period decreases each year until the UAAL is fully amortized (similar to a standard home mortgage). If the UAAL is amortized as a level percentage of pay, the payment increases each year at the same rate as the assumed payroll growth. Level dollar amounts remain the same dollar 18

22 3. ACTUARIAL METHODS amount over the entire amortization period. In general, closed periods result in the plan reaching 100% funding as scheduled (assuming no future gains or losses). A recent white paper, prepared by the Conference of Consulting Actuaries, recommends amortization policies that set a date for full amortization of the UAAL (closed periods) and new layers for annual changes over periods of less than 30 years (15 to 20 years recommended). In the GFOA Core Elements of a Funding Policy, closed amortization periods of no longer than 25 years are recommended. In recent years, we have observed a trend in the industry of preference of closed amortization periods over open periods. By setting required contributions based on closed, shorter amortization periods, there is more volatility but a higher probability of achieving and maintaining the 100% funded status policy. The funding policy for both MOSERS plans is to finance the plans using actuarial recommended contributions that remain level as a percentage of payroll and from one generation of citizens to the next generation. Beginning with the 2014 valuation, both MSEP and the Judicial Plan changed from amortizing the UAALs from a 30-year open amortization period to a 30-year closed amortization period. As a result, in the June 30, 2016 valuations, the UAALs were amortized as a level percent to pay over a 28-year period. A 2014 amendment requires the plans to fund the percentage of pay levels not less than the rates determined in the June 30, 2013 valuations, until the plans are at least 80% funded. This should help accelerate funding ahead of the 28-year schedule. We find that the amortization of the unfunded actuarial accrued liability as presented by GRS in the valuation reports reasonable, and in compliance with the MOSERS funding policy. 19