T E X A S M U N I C I P A L R E T I R E M E N T S Y S T E M ACTUARIAL E X P E R I E N C E I N V E S T I G A T I O N S T U D Y AS OF D E C E M B E R 3

Transcription

1 T E X A S M U N I C I P A L R E T I R E M E N T S Y S T E M ACTUARIAL E X P E R I E N C E I N V E S T I G A T I O N S T U D Y AS OF D E C E M B E R 3 1,

2 May 20, 2011 Board of Trustees Texas Municipal Retirement System Austin, Texas Dear Members of the Board: Subject: Results of 2011 Experience Study We are pleased to present our report of the 2011 Experience Investigation Study for the Texas Municipal Retirement System (TMRS). Our report includes a discussion of the recent experience of the System, it presents our recommendations for new actuarial assumptions and methods, and it provides information about the actuarial impact of these recommendations on the liabilities and other key actuarial measures of TMRS. The proposed actuarial assumptions in this report are based upon the assumption that the System will be restructured effective as of December 31, If the current legislation (House Bill 997 / Senate Bill 350) to restructure does not pass in the 2011 Texas legislative session, some of the recommendations in this report may no longer be appropriate and we will reissue the recommendations later this summer with the appropriate changes. With the Board of Trustees' approval of the recommendations in this report, we believe the actuarial condition of the System will be more accurately measured and portrayed. This experience investigation study was conducted in accordance with generally accepted actuarial principles and practices, and in full compliance with the Actuarial Standards of Practice as issued by the Actuarial Standards Board. All of the undersigned are members of and meet the Qualification Standards of the American Academy of Actuaries. We wish to thank the TMRS staff for their assistance in this project. Sincerely, Joseph P. Newton, FSA, EA, MAAA Mark R. Randall, FCA, EA, MAAA Brad Stewart, ASA, EA, MAAA Linna Ye, ASA, MAAA J:\3052\2010\EXP\Report\ExperienceStudy.docx

3 Table of Contents Table of Contents PAGE COVER LETTER SECTION I 2 INTRODUCTION SECTION II 6 SUMMARY OF RECOMMENDATIONS SECTION III 11 ANALYSIS OF EXPERIENCE AND RECOMMENDATIONS SECTION IV 35 ACTUARIAL IMPACT OF RECOMMENDATIONS SECTION V 38 SUMMARY OF NEW ASSUMPTIONS SECTION VI 48 SUMMARY OF DATA AND EXPERIENCE APPENDIX I APPENDIX II TERMINATION EXPERIENCE BY CITY ACTUARIAL IMPACT BY CITY

4 SECTION I I N T R O D U C T ION

5 Section I Introduction Introduction In determining liabilities, contribution rates and funding periods for retirement plans, actuaries must make assumptions about the future. Among the assumptions that must be made are: Retirement rates Mortality rates Termination rates Disability rates Investment return rate Salary increase rates Inflation rate For some of these assumptions, such as the mortality rates, past experience provides important evidence about the future. For other assumptions, such as the investment return rate, the link between past and future results is much weaker. In either case, though, actuaries should review their assumptions periodically and determine whether these assumptions are consistent with actual past experience and with anticipated future experience. Historically, the Texas Municipal Retirement System (TMRS) generally has an experience study done every fifth year. However, this study and the prior study have been done on more frequent intervals to coincide with significant changes to the System. This study is generally based on experience during the four-year period of January 1, 2006 to December 31, 2009, with updates for data through 2010, as known. The last experience study was prepared in 2007 following completion of the December 31, 2006 actuarial valuation report. That report generally covered experience during the period of January 1, 2003 to December 31, In conducting experience studies, actuaries generally use data over a period of several years. This is necessary in order to gather enough data so that the results are statistically significant. In addition, if the study period is too short, the impact of the current economic conditions may lead to misleading results. It is known, for example, that the health of the general economy can impact salary increase rates and termination rates. Using results gathered during a short-term boom or bust will not be representative of the long-term trends in these assumptions. Also, the adoption of legislation, plan improvements or changes in salary schedules will sometimes cause a short-term distortion in the experience. For example, if an early retirement window was opened during the study period, we would usually see a short-term spike in the number of retirements followed by a dearth of retirements for the following two-to-four years. Using a longer period prevents giving too much weight to such short-term effects. On the other hand, using a much longer period increases the difficulty of identifying changes in behavior that may be occurring, such as mortality improvement or a change in the ages at which members retire. In our view, using a four to fiveyear period is reasonable. However, note that in our analysis of salary increases, we incorporated the results from the prior experience study. 2

6 Section I Introduction In an experience study, we first determine the number of deaths, retirements, etc. that occurred during the period. Then we determine the number expected to occur, based on the current actuarial assumptions. The number expected is determined by multiplying the probability of the occurrence at the given age, by the exposures at that same age. For example, let s look at a rate of retirement of 15% at age 55. The number of exposures can only be those members who are age 55 and eligible for retirement at that time. Thus they are considered exposed to that assumption. Finally we calculate the A/E ratio, where "A" is the actual number (of retirements, for example) and "E" is the expected number. If the current assumptions were "perfect", the A/E ratio would be 100%. When it varies much from this figure, it is a sign that a new assumption may be needed. (However, in some cases we prefer to set our assumptions to produce an A/E ratio a little above or below 100%, in order to introduce some conservatism.) Of course we not only look at the assumptions as a whole, but we also review how well they fit the actual results by gender, by age, and by service. Finally, if the data leads the actuary to conclude that new tables are needed, the actuary "graduates" or smoothes the results since the raw results can be quite uneven from age to age or from service year to service year. Please bear in mind that, while the recommended assumption set represents our best estimate, there are other reasonable assumption sets that could be supported. Some reasonable assumption sets would show higher or lower liabilities or costs. O R G A N I Z A T I O N O F R E P O R T Section II of this report summarizes our recommended changes. Section III contains our findings and a more detailed analysis of our recommendation for each actuarial assumption. The impact of adopting our recommendations on liabilities and contribution rates is shown in Section IV. Section V shows a summary of the recommended assumptions. Finally, Section VI presents detailed summaries of the data and comparisons of the A/E ratios. S E C T I O N V I E X H I B I T S The exhibits in Section VI should generally be self-explanatory. For example, on page 58, we show the exhibit analyzing the service-based termination rates. The second column shows the total number of members who terminated during the study period. This excludes members who died, became disabled or retired. Column (3) shows the total exposures. This is the number of members who could have terminated during any of the years. In this exhibit, the exposures exclude anyone eligible for retirement. A member is counted in each year they could have terminated, so the total shown is the total exposures for the study period. Column (4) shows the probability of termination based on the raw data. That is, it is the result of dividing the actual number of terminations (col. 2) by the number exposed (col. 3). Column (5) shows the current termination rate and column (6) shows the new recommended termination rate. Columns (7) and (8) show the expected numbers of 3

7 Section I Introduction terminations based on the current and proposed termination assumptions. Columns (9) and (10) show the Actual-to-Expected ratios under the current and proposed termination assumptions. 4

8 SECTION II S U M M A RY OF RECOMMEND AT I O N S

9 Section II Summary of Recommendations Summary of Recommendations Our recommended changes to the current actuarial assumptions may be summarized as follows: Economic Assumptions 1. Make no change to the 3.00% inflation assumption. Long-term inflation is likely to fall in the 2.5% to 3.5% range, and the current assumption is very close to long term historical averages. 2. Make no change to the 7.00% nominal investment return assumption. The expected arithmetic return based on the target asset allocation is 7.14% while the median expected geometric return over the next 20 years is 6.72%. 3. Based on a restructured System, as detailed in the current 2011 proposed legislation known as HB 997 and SB 350, the discount rate used to value the liabilities of the individual employers will decrease from the current 7.50% to the investment return assumption of 7.00%. The current 7.50% discount rate is based upon the existing separated fund structure and reflects the leverage on the Municipality Accumulation Fund (MAF) given the minimum asset return guarantees of the Current Service Annuity Reserve Fund (CSARF) and the Employees Saving Fund (ESF). After restructuring, the 7.00% discount rate will be applicable to the Benefit Accumulation Fund (BAF) of each employer. 4. Make no changes to the current salary scale assumption. The experience for salary increases in the public sector during the last decade was much higher than long term averages and we believe the next few years will likely be lower, ultimately returning to moderate increases. 5. Make no changes to the current payroll growth rate assumption. The payroll growth assumption does not impact the liabilities, only the development of the amortization of the unfunded actuarial accrued liability. Payroll growth over the prior five and ten year periods has averaged 3.75% and 4.01%, respectively, but is expected to moderate somewhat due to workforce demographics. Mortality Assumptions (Valuation Purposes Only No Impact on Annuity Purchase Rates) 6. Update the post-retirement mortality tables for non-disabled retirees to the RP-2000 combined mortality tables projected to the year 2003 by scale AA. The actual number of retiree deaths during the study period was less than expected based on the current mortality table, so rates were updated to bring them in line with actual experience plus build in a margin for future mortality improvement. Sample rates are shown on page Change the disabled post-retirement mortality assumption to the RP-2000 disabled life mortality rates for males and females, with the rates multiplied by an 80% factor for both males 6

10 Section II Summary of Recommendations and females. Sample rates are shown on page Update the pre-retirement mortality tables to the RP-2000 combined mortality table projected to the year 2003 by scale AA, with a five-year setback for both males and females. Current rates result in substantially more expected deaths than occurred during the study period; therefore, rates are being updated to better match experience. Sample rates are shown on page 43. Other Demographic Assumptions 9. Maintain the current retirement rates as a base table. Further analysis of the data shows that cities with more generous benefit provisions generally have higher rates of retirement. Therefore, we will adjust the retirement rates for each city based upon the plan provisions (employee contribution rate, employer match, and presence of a COLA) of that particular municipality. The current and proposed tables are shown on page Reduce the select period for termination from 20 years to 10 years with rates after the select period based on years from retirement. Apply different multipliers to the base rates based on job classification (police, fire, or other). Use individual multipliers of the base tables for each employer rather than the five tiers currently being utilized. Overall, these changes lower the expected number of terminations. Proposed termination rates are shown on pages 40 and Decrease the forfeiture rates for vested members not eligible for retirement. Also, base them on age rather than service and introduce adjustments based on the employer match. Proposed forfeiture rates are shown on page Make no change to the rates of disability, since the current rates produce a reasonable fit to the observed experience. Sample rates are shown on page Make no change to the current 40% Partial Lump Sum assumption. Actuarial Methods and Policies 14. Recommend no change to the use of a 10-year smoothing technique to determine the actuarial value of assets, used for determining the annual employer contribution rates. 15. Recommend no change to the use of a soft corridor around the market value of assets when determining the actuarial value of assets. The soft corridor utilizes three-year smoothing for deferred gains or losses outside of the predefined range. Based on the proposed restructured System, we recommend decreasing the threshold for the corridor from the current 25% of market value of assets to 15%. 7

11 Section II Summary of Recommendations 16. Based on the proposed restructured System, recommend substantially reducing, if not eliminating, the current 20% reserve policy. We believe the asset smoothing techniques along with a comparatively lower investment return assumption will provide an adequate reduction in contribution rate volatility. 17. Recommend continued use of the Projected Unit Credit Actuarial Cost Method. Even though not recommended in this experience study, in light of the proposed financial disclosure changes by the Governmental Accounting Standards Board (GASB), the Board could consider the Entry Age Normal Cost Method as a viable alternative. The Exposure draft of the proposed accounting rule changes is due to be released later this Summer. 18. Recommend continued use of closed amortization periods for experience gains and losses for underfunded plans. However, we recommend the Board consider adopting a policy to eventually decrease the current 30 year period allowable to TMRS employers. We believe an amortization period of years would be preferable over the longer term once the System has been restructured and all phase-in periods are complete. 19. Recommend use of a 25-year open amortization policy for overfunded plans. In addition, once a plan reaches overfunded status, all prior closed non ad hoc bases will be erased. This will provide for adequate surplus management and less contribution volatility. 20. Recommend changing the accrual period for members with service with other TMRS employers to be based on service only with the current employer. This change will only be applicable for future hires. Current members will continue to accrue over all TMRS service and therefore this change will have no impact on the current valuation. 21. Recommend continued use of a 15 year level dollar amortization policy for ad hoc benefit enhancements. Small City Policies 22. Recommend continued use of lower termination assumptions for smaller cities, with maximum multipliers of 75% for employers with less than 6 members, 85% for employers with between 6 and 10 members, and 100% for employers with between 11 and 15 members. 23. Recommend application of the assumption load on the post-retirement life expectancy for employers with less than 15 active members. The life expectancy will be loaded by decreasing the mortality rates by 1% for every active member less than 15. For example, an employer with 5 active members will have the baseline mortality tables multiplied by 90% (10 years times 1%). Lower multipliers create longer life expectancies. 24. Recommend application of shorter amortization periods for employers with less than 20 8

12 Section II Summary of Recommendations members. The maximum amortization period for amortizing gains and losses will be lowered from current levels by 1 year for each active member less than the 20 member threshold. For example, an employer with 8 active members and a current maximum amortization period of 25 will use (25-(20-8)) = a 13 year amortization period for the gain or loss in that year s valuation. Under this policy, the lowest amortization period will be 25-(20-1) = 6 years. 25. Based on the proposed restructured System, recommend discontinuing the policy of placing a floor under the normal cost for cities with less than 3 members. As described above, the shorter amortization period along with lower termination assumptions and longer life expectancies will provide adequate conservatism for these smaller groups. The proposed policy will significantly reduce the rate volatility currently seen in the contribution rates for this group. 9

13 SECTION III A N A LY S I S OF EXPERIEN CE A N D R E C O MMENDAT I O N S

14 Section III Analysis of Experience and Recommendations Analysis of Experience and Recommendations We will begin by discussing the economic assumptions: inflation, expenses, the investment return rate, the salary increase assumption, and the rate of payroll growth. Next are the demographic assumptions: mortality, disability, termination and retirement. Finally, we will discuss all of the actuarial methods used. I N F L A T I O N By inflation, we mean price inflation, as measured by annual increases in the Consumer Price Index (CPI). This inflation assumption underlies all of the other economic assumptions we employ. It not only impacts investment return, but also salary increase rate and the payroll growth assumption. Our current annual inflation assumption is 3.00%. Over the five-year period from December 2005 through December 2010, the CPI-U has increased at an average rate of 2.18%. However, the assumed inflation rate is only weakly tied to past results. The table below shows the average inflation over various periods, ending December 2010: Periods Ending December 2010 Average Annual Increase in CPI-U Last five (5) years 2.18% Last ten (10) years 2.34% Last fifteen (15) years 2.40% Last twenty (20) years 2.50% Last twenty-five (25) years 2.82% Last thirty (30) years 3.16% Last fifty (50) years 4.07% Since 1913 (first available year) 3.23% Source: Bureau of Labor Statistics, CPI-U, all items, not seasonally adjusted As you can see, while inflation has been lower than 3.00% over the last 20 years, if we look back over periods of more than 25 years, inflation has averaged around or above 3.00% per year. The chart on the following page shows the average annual inflation in each of the ten consecutive five-year periods over the last fifty years: 11

15 Section III Analysis of Experience and Recommendations Average Annual Inflation CPI-U, Five Fiscal Year Averages 10.00% 9.00% 9.23% 8.00% 7.00% 6.88% 6.00% 5.00% 4.00% 4.59% 4.84% 4.13% 3.00% 2.00% 1.31% 2.79% 2.54% 2.49% 2.18% 1.00% 0.00% yr Avg. Increase Most observers expect inflation to be low or non-existent for the next 1-3 years, as the US economy works out of the recent recession. In fact, over the year ending December 2010, the CPI-U increased only 1.50%. After that, some observers expect inflation to increase as a consequence of the large amount of government borrowing. One source of information about future inflation is the market for US Treasury bonds and TIPS (Treasury Inflation-Protected Securities). For example, the December 31, 2010 yield for a 20- year inflation indexed Treasury bond was 1.59% plus actual inflation. The yield for a 20-year non-indexed US Treasury bond was 4.13%. Simplistically, this means that on that day the bond market was predicting that inflation over the next twenty years would average 2.54% (4.13% 1.59%) per year. One year earlier, as of December 31, 2009, the spread between the 20-year inflation protected and constant maturity bonds was only marginally higher, with a difference of 2.55%, so there has been little change in this expectation. However, this analysis is known to be imperfect. It ignores the inflation risk premium that buyers of US Treasury bonds should ask for, and it ignores the differences in liquidity between US Treasury bonds and TIPS. For a number of years, the Cleveland Fed published on its website an adjusted inflation expectation, using formulas to adjust the raw results for these two factors. However, because of the unprecedented rush to safety and liquidity following the market meltdown, demand for US Treasury bonds soared, and the spreads between treasuries and TIPS 12

16 Section III Analysis of Experience and Recommendations shrank. As a result, the Cleveland Fed discontinued publication of its adjustments, believing their formulas would not work in the current economic climate. We also reviewed the inflation assumptions used by several investment consulting firms. In our sample of seven firms (including TMRS investment consultant, RV Kuhns & Associates), these ranged from 2.00% to 3.25%, with an average of 2.71%. However, the investment consulting firms typically set their assumptions based on a five or ten year outlook, while the calculation of the actuarial valuation must make a much longer view. In the Social Security Administration s 2010 Trustees Report, the Office of the Chief Actuary is projecting a long-term average annual inflation rate of 2.8% under the intermediate cost assumption. (The inflation assumption is 1.8% and 3.8% respectively in the low cost and high cost projection scenarios.) These inflation assumptions were unchanged from their prior year s report. The Philadelphia Federal Reserve conducts a quarterly survey of the Society of Professional Forecasters. Their most recent forecast (fourth quarter of 2010) was for inflation over the next ten years to average 2.20% which is down slightly from their third quarter estimate of 2.30%. Another source of information about this assumption is the Public Funds Survey that is prepared on behalf of the National Association of State Retirement Administrators (NASRA) and the National Council on Teacher Retirement (NCTR). This report surveys about 125 plans, including all of the largest public funds covering state employees or teachers. The current survey shows that the median inflation rate assumed for large public retirement systems in the U.S. is 3.50%. Our current 3.00% assumption is used by about 30% of the surveyed systems, with almost all of the rest using higher assumptions. Therefore, we believe a reasonable long-term inflation assumption will likely fall in the range of 2.50% to 3.50%. We believe that inflation may continue to be less than 3.00% annually over the next few years, but we believe a 3.00% rate of inflation is reasonable over the long term. This is in line with the average for the last 25 years, and a little below the long-term historical average. Therefore, we are recommending retaining the 3.00% inflation assumption. I N V E S T M E N T A N D A D M I N I S T R A T I V E E X P E N S E S Since the trust fund pays expenses in addition to member benefits and refunds, we must make some assumption about these. Almost all actuaries treat investment expenses as an offset to the investment return assumption. That is, the investment return assumption represents expected return after payment of investment expenses. On the other hand, there is a divergence of practice on the handling of administrative expenses. Some actuaries make an assumption that administrative expenses will be some fixed or increasing dollar amount, others assume that the administrative expenses will be some 13

17 Section III Analysis of Experience and Recommendations percentage of the plan s actuarial liabilities or normal cost, and others treat administrative expenses like investment expenses, as an offset to the investment return assumption. Our practice is to set the investment return assumption as the net return after payment of both investment and administrative expenses. This chart shows the administrative and investment expenses for the last five years expressed as a percentage of the assets each year: Annual Expenses Expressed as a Percentage of Assets Calendar Year Administrative Investment Total 2010 (est) 0.08% 0.04% 0.12% % 0.04% 0.11% % 0.03% 0.11% % 0.01% 0.09% % 0.01% 0.09% Average 0.08% 0.03% 0.10% As the Board is aware, the TMRS portfolio has undergone and continues to undergo significant changes. The number of asset classes is being expanded. As TMRS moves into these new asset classes, more internal or external managers must be hired to manage these investments. In addition, many of the new asset classes have higher expense ratios than the asset classes that were in prior use. Because of these changes, we don t believe past experience will be a good indicator of future administrative and investment expenses. Based on information provided by RV Kuhns & Associates and our own experience with other large state-wide retirement systems, we are recommending the assumption that investment and administrative expenses will consume 0.30% (30 basis points) of each year s investment return. This assumption is then used in setting the investment return assumption. I N V E S T M E N T R E T U R N Currently, TMRS assumes an investment return rate of 7.00%, net of investment and administrative expenses. This is the rate used in discounting future payments in calculating the actuarial present value of those payments. Even a small change to this assumption can produce significant changes to the liabilities and contribution rates. The 7.00% assumption is composed of a 3.00% assumed inflation rate plus a 4.30% assumed real return, for a gross expected return of 7.30%. This is offset by 0.30% for expected investment and administrative expenses. For this assumption, past performance, even averaged over a twenty-year period, is not a reliable indicator of future performance. This is particularly true for TMRS which is in the process of diversifying out of an all bond portfolio. The actual asset allocation of the trust fund will 14

18 Percentage of Plans Texas Municipal Retirement System Section III Analysis of Experience and Recommendations significantly impact the overall performance, so returns achieved under a different allocation are not meaningful. More importantly, the real rates of return for many asset classes, especially equities, vary so dramatically from year to year that even a twenty-year period is not long enough to provide reasonable guidance. There are strong reasons to believe the next twenty years will be different than the last twenty, in part because we are starting from higher price-earnings ratios on equities, and in part because the current bond returns are so low. The table below provides the distribution of the different investment return assumptions used by other large public retirement systems. TMRS 50% 45% 40% 35% 30% 25% 20% 15% 10% 5% 0% Public Fund Survey Investment Return Assumption 7.00% 7.25% 7.50% 7.75% 8.00% 8.25% 8.50% Source: Public Funds Survey (n=124) Median investment return assumption: 8.00% nominal return While we do not recommend the Board select an assumption based on this information, it is still informative to see TMRS assumption in relation to its peers. While the table shows that the 7.00% assumption is lower than most other systems, you should be aware that several large plans have recently reduced their assumption, and several others are in the middle of a review of this assumption. Also, the TMRS target asset allocation is generally more conservative than other systems. We view the investment return assumption as having three components: the assumed rate of (price) inflation, the real return net of inflation, and an offset for expected investment and administrative expenses. This building block approach is one explicitly permitted under ASOP 27. The inflation assumption and expense assumption have already been discussed, so we will proceed with the analysis of the real rate of return. To do this, we like to examine the results of applying a set of capital market assumptions to the plan s target asset allocation. Since we are not investment professionals, we looked at the results using information provided by RV Kuhns & Associates, the TMRS investment consultant, and 15

19 Section III Analysis of Experience and Recommendations we looked at results under the capital market assumptions for six other investment consulting firms: Callan JP Morgan NEPC PCA Strategic Investment Solutions Towers Watson Here is a table with the plan s long-term target asset allocation and the development of the plan s expected nominal investment returns using capital market assumptions provided by TMRS investment consultant, RV Kuhns & Associates. Long-Term Target Asset Allocation Expected Total Return Expected Portfolio Return (2) x (3) Asset Class (1) (2) (3) (4) Domestic Equities 20% 8.15% 1.63% International Equities 20% 10.50% 2.10% Fixed Income 35% 5.00% 1.75% Core Real Estate 10% 7.00% 0.70% Absolute Return 5% 7.75% 0.39% Real Return 5% 6.75% 0.34% Private Equity 5% 12.25% 0.61% Gross Return 7.52% Actuary s Plan Expense Assumption: (0.30%) Net Expected Nominal Investment Return 7.22% As you can see, RV Kuhns & Associates expected returns would support the current 7.00% return assumption. However, we also looked at the results using the capital market assumptions of the six other firms named above. These investment consulting firms periodically issue reports that describe their capital market assumptions, that is, their estimates of expected returns, volatility, and correlations between asset classes. While these assumptions are developed based upon historical analysis, many of these firms also incorporate forward looking adjustments to better reflect near-term expectations. The estimates for core investments (i.e. fixed income, equities, and real estate) are generally based on anticipated returns produced by passive index funds, and do not include an assumption for possible alpha generated returns. Given the plan s target asset allocation and the capital market assumptions made available by the investment consultants listed above, the development of the average nominal return, net of administrative and investment expenses, is provided in the table on the next page: 16

20 Section III Analysis of Experience and Recommendations Investment Consultant Investment Consultant Expected Nominal Return Investment Consultant Inflation Assumption Expected Real Return (2) (3) Actuary Inflation Assumption Expected Nominal Return (4)+(5) Plan Incurred Expense Assumption Expected Nominal Return Net of Expenses (6)-(7) (1) (2) (3) (4) (5) (6) (7) (8) % 2.75% 3.91% 3.00% 6.91% 0.30% 6.61% % 3.00% 3.91% 3.00% 6.91% 0.30% 6.61% % 3.25% 4.04% 3.00% 7.04% 0.30% 6.74% % 2.00% 4.58% 3.00% 7.58% 0.30% 7.28% % 2.50% 4.72% 3.00% 7.72% 0.30% 7.42% % 2.75% 4.95% 3.00% 7.95% 0.30% 7.65% % 2.50% 5.02% 3.00% 8.02% 0.30% 7.72% Average 7.12% 2.68% 4.44% 3.00% 7.44% 0.30% 7.14% 1 Expected returns are based on an arithmetic average. We have determined for each firm the expected nominal return rate, then subtracted that firm s expected inflation to arrive at their expected real return in col. (4). Then we have added back our 3.00% inflation assumption and subtracted 0.30% for expenses to obtain a net nominal return. As the table shows, the average one-year return of the seven firms is 7.14%, which is 0.14% more than the current assumption of 7.00%. Additionally, three of the seven other investment firms have an expected real return below the current assumption. The other four firms produced expected returns at least 0.28% above the current 7.00% assumption. In addition to examining the expected one-year return, it is important to review anticipated volatility of the investment portfolio and understand the range of long-term net return that could be expected to be produced by the investment portfolio. Therefore, the table on the next page provides the 25 th, 50 th, and 75 th percentiles of the 20-year geometric average of the expected nominal return, net of expenses, as well as the probability of exceeding the current 7.00% assumption. 17

21 Section III Analysis of Experience and Recommendations Investment Consultant Distribution of 20-Year Average Geometric Net Nominal Return Probability of exceeding 25th 50th 75th 7.00% * (1) (2) (3) (4) (5) % 6.18% 7.60% 34.8% % 6.18% 7.61% 34.9% % 6.36% 7.71% 37.4% % 6.96% 8.19% 49.0% % 7.08% 8.35% 51.8% % 7.12% 8.71% 52.1% % 7.17% 8.79% 52.8% Average 5.33% 6.72% 8.14% 44.7% *Plan's current return assumption net of expenses. As the analysis shows, there is a 50% likelihood that the 20-year average net real return will be between 5.33% and 8.14%. This becomes the best-estimate range under ASOP 27. However, the average results of the seven firms indicate there is slightly less than a 50% chance that the System will produce an average return that exceeds 7.00% over the next 20 years. We find the current investment assumption of 7.00% to be within the range for a reasonable assumption, as it is very close to and in-between the expected mean of 7.14% and the median expected return of 6.72%. Therefore, we recommend no change at this time. However, based on a restructured System, the discount rate used to value the liabilities of the individual employers will decrease from the current 7.50% to the investment return assumption of 7.00%. The current 7.50% discount rate is based on the separated fund structure and reflects the leverage of the MAF funds from the ESF and CSARF. After restructuring, the 7.00% discount rate/investment return assumption will be applicable to the Benefit Accumulation Fund (BAF) of each employer. 18

22 Section III Analysis of Experience and Recommendations S A L A R Y I N C R E A S E R A T E S In order to project future benefits, the actuary must project future salary increases. Salaries may increase for a variety of reasons: Step or service-related increases Increases for acquisition of advanced degrees or specialized training Promotions Merit increases, if available Bonuses, if available Our salary increase assumption is meant to reflect all of these types of increases, since all of these affect the salaries used in benefit calculations and upon which contributions are made. The actuary should not look at the overall increases in payroll in setting this assumption, because payroll can grow at a rate different from the average pay increase for individual members. There are two reasons for this. First, when older, longer-service members terminate, retire or die, they are generally replaced with new members being compensated with a lower salary. Because of this, in most populations that are not growing in size, the growth in total payroll will be smaller than the average pay increase for members. Second, payroll can change due to an increase or decrease in the size of the group. Therefore, to analyze salary increases, we examine the actual increases for individuals. We analyzed the salary increases based on the change in the member s reported pay from one year to the next. That is, we looked at each member who appeared as an active member in two consecutive valuations these are called continuing members and measured his/her salary increase. Salary increases can vary significantly from year to year. When the employer s tax revenues stall or increase slowly, salary increases can be small or nonexistent. During more economically favorable times, salary increases can be larger. Therefore, for this assumption in particular, we prefer to use data over a longer period in establishing our assumptions and have included the results from the prior actuary in our analysis. The salary assumption can be thought of as consisting of wage inflation (that part of the pay increase which is given to all employees) and an additional component to reflect step increases and other increases correlated with service. Most actuaries recommend salary increase assumptions that include an element that depends on the member s age or service, especially for large, public retirement systems. It is typical to assume larger pay increases for younger or shorter-service employees. Experience across retirement systems shows salaries are more closely correlated to service than age. 19

23 Section III Analysis of Experience and Recommendations Our current assumption follows this pattern. The current salary increase rates vary by service for the first 10 years. They range from 12.00% for a new member s first increase to 5.50% for members with 10 years of service. Age based rates are used for members with more than 10 years of service and vary from 5.25% for members under age 35 down to 3.50% for members age 65 and older. Members with 10 or more years of service received increases about 2.40% greater than actual inflation, while the current assumption is approximately 1.40% above inflation. However, we feel the last decade has provided public sector employees much higher salary increases than historically have been given and it is likely that the next decade could be closer to longer term averages. This is in line with national wage statistics, since over the last 20 years, wages have risen about 1.00% faster per year than inflation. Based on the fact that the last two valuations have produced no salary losses and the expectation of lower salary increases over the next few years, we are recommending no change to this assumption. Additional results of the analysis regarding this assumption are provided in Section VI on page 49. P A Y R O L L G R O W T H R ATE The salary increase rates discussed above are assumptions applied to individuals. They are used in projecting future benefits. We also use a separate payroll growth assumption, currently 3.00%, in determining the charge needed to amortize the unfunded actuarial accrued liability. The amortization payments are calculated to be a level percentage of payroll, so as payroll increases over time, these charges do too. The amortization percentage is dependent on the rate at which payroll is assumed to increase. Payroll has grown at 5.95% over the last five years and 6.20% over the last ten years. Part of this increase, though, comes from the growth in the number of active members, and GASB 25 prohibits systems from using anticipated membership growth in setting the payroll growth assumption. If we adjust to remove the effect of the increase in membership, payroll growth has averaged 3.75% over the last five years and 4.01% over the last ten years. However, inflation has been lower than expected during the 5 and 10 year periods ending December 31, Similar to the rate of salary increases, this can cause downward pressure in payroll growth rate. If we adjust the actual payroll growth rate experience for the difference between actual and assumed inflation, the normalized experience now becomes 4.19% and 4.48% respectively. 20

24 Section III Analysis of Experience and Recommendations However, we don t expect the rate of payroll growth to be as high in the future. In the near term, pressure on municipal budgets from the economic downturn should limit the growth rate. More importantly, the expected number of baby boomers retiring over the next 10 to 20 years should constrain growth as they are replaced by lower paid employees. Based on historical experience and the analysis shown above, we are recommending no change to the current 3.00% payroll growth rate. P O S T - R E T I R E M E N T M O R T A L I T Y R ATES (LI A B I L I T Y A N D C O S T C A L C U L A T I O N S ) TMRS actuarial liabilities depend in part on how long retirees live. If members live longer, benefits will be paid for a longer period of time, and the liability will be larger. It is important to point out that our analysis is focused on the mortality used in the annual actuarial valuations to determine the liabilities of the individual city funds and to determine the employer contribution requirements. The mortality assumptions used to determine the annuity purchase factors which ultimately calculate the monthly annuities for the individual members at retirement are based on different assumptions and are not part of this analysis or part of our recommendations. The mortality table currently being used for non-disabled retirees and for beneficiaries receiving benefits is the RP-2000 mortality table. This table has separate rates for males and females. The rates for females are then adjusted by using a one-year set-forward with no adjustment for males. (Set-backs and set-forwards are traditional actuarial techniques used to adjust a table to match the actual observed data. When a table is set forward one year, the actuary uses the table s rate for an age one year older than the person s actual age. For example, the mortality rate used for a 60-year old female retiree is the rate in the RP-2000 mortality table for females at age 61.) To analyze the data, we begin by determining the expected number of deaths in each year at each age for males and females. Then we compare the actual number to the expected number. The ratio of the actual deaths to the expected deaths the A/E ratio then tells us whether the assumptions are reasonable. For this assumption an A/E ratio of between 110% and 120% is generally desired for conservatism and to build in a margin for continued future improvements in mortality rates. There were 1,473 deaths among the male retirees and 377 deaths among female retirees during the study period. (These figures exclude deaths among beneficiaries and disabled retirees.) Based on the current mortality assumption, we expected 1,283 and 442 deaths, respectively. This produced A/E ratios of 115% for males, 85% for females, and 107% in total for both males and females. Because the A/E ratio of 107% for males and females combined was slightly less conservative than desired (and even lower for the core retiree ages of 60 to 79 at 103%), it is necessary to update the assumption to reflect past and anticipated future mortality improvements (longer life expectancies). We propose removing the set-forward for females and projecting the tables for both males and 21

25 Rate of Mortality Rate of Mortality Texas Municipal Retirement System Section III Analysis of Experience and Recommendations females to the year 2003 using Scale AA. The resulting A/E ratios for males and females increase to 119% and 96%, respectively. This change increases the combined A/E ratio to 113% overall and 110% for the core ages. A summary of the analysis and a table of median life expectancy at retirement under the current and proposed assumptions are shown in the following exhibits: 40% Male Post-Retirement Mortality 35% 30% 25% 20% 15% 10% 5% 0% Age Actual Experience Present Assumptions Proposed Assumptions 35% Female Post-Retirement Mortality 30% 25% 20% 15% 10% 5% 0% Age Actual Experience Present Assumptions Proposed Assumptions 22

26 Section III Analysis of Experience and Recommendations Average Life Expectancy at Retirement (Years) Males Females Current Proposed Current Assumptions Assumptions Assumptions Age at Retirement Proposed Assumptions Please refer to the exhibits on pages 50 and 51 for additional information regarding this assumption. D I S A B L E D M O R T A L I T Y R A T E S This is a minor assumption as there are relatively few disability occurrences and TMRS disability benefits are not subsidized. The current disabled mortality tables are the RP-2000 disabled-life tables for males and females, with a four-year setback on the male table and no adjustment on the female table. During the study period, there were 135 deaths for males and 36 for females compared to 144 and 26 expected deaths for males and females, respectively. This produced A/E ratios of 94% for males and 138% for females. We recommend removing the setback on the male table and taking only 80% of the unadjusted rates from the table. Because of the lack of credible data for disabled females, we recommend the same 80% adjustment to the female table. This new assumption will result in A/E ratios of 98% for males, 171% for females, and 108% in total. Please refer to the exhibits on pages 52 and 53 for additional information. A C T I V E M O R T A L I T Y R ATES This is another minor assumption with little impact on the employer contribution rates. However, it is very material to the supplemental death benefit plan of TMRS. Mortality across employee groups is generally lower than the mortality rates in the post-retirement mortality tables. The results of the analysis are shown on pages 54 and 55. As you can see, there were 303 actual deaths (239 males and 64 females), while there were 769 expected deaths (563 males and 206 females). This produced A/E ratios of only 42% and 31% for males and females, respectively. Although the number of preretirement deaths is not significant, we concluded that the assumption should be updated. We recommend utilizing the tables used for post-retirement mortality (RP-2000 combined mortality tables projected to year 2003 by Scale AA), with a fiveyear setback for both males and females. The updated assumption will increase A/E ratios to 71% for males and 54% for females. 23

27 Section III Analysis of Experience and Recommendations D I S A B I L I T Y R A T E S Disability is also a minor assumption. The results of the analysis are shown on pages 56 and 57. There were 261 new disabled retirees (183 males and 78 females) during the investigation period, while we expected 279 (206 males and 73 females). The A/E ratios were 88% for males, 107% for females, and 94% combined. These tables produce a reasonable fit to the observed experience, and we are recommending no changes at this time. T E R M I N A T I O N R ATES Termination rates reflect members who leave for any reason other than death, disability, or service retirement. They apply whether the termination is voluntary or involuntary, and whether the member takes a refund or keeps his/her account balance on deposit. The current termination rates reflect the member s gender and service. In addition, each employer has been placed in a rate category based on their past experience versus the group as a whole: Low, Mid-Low, Mid, Mid-High, and High. These categories are established for males and females separately. Historical practice has been for new cities to be placed into the Low category and then to limit the change between categories in a single experience study to a one category move, so an employer in the Mid category could only move to Mid-High or Mid-Low or stay at Mid. We are recommending significant changes to the structure of this assumption. While we feel a service only assumption is reasonable and appropriate for most employers, a more intricate structure can more accurately project the liabilities for the individual employers, especially for the smaller municipalities of TMRS. Select and Ultimate versus Service Only The first recommended change will be to move from the service only structure to a select and ultimate pattern. We are proposing a ten-year select period in the new assumption set. A select and ultimate structure splits the assumption set into two segments: a more robust structure for the first years of service (select) and then a more simplified structure once the member has reached the end of the select period. This type of structure recognizes the fact that the turnover rates for a new member in their 20 s are much different than the turnover rates for a new member in their 40 s. The following exhibit provides the actual rates of termination by years of service for males in the TMRS population for two cohorts based on entry age. Notice that the overall turnover rate for the member hired at 45 is lower than the younger member. The difference in these two patterns result 24

28 Section III Analysis of Experience and Recommendations in 30% more members who are employed at age 45 reaching the end of the select period than the younger members. Using a simple average among all members can overestimate the cost for the younger members and underestimate the cost for the older members. The larger the employer, the less difference there is between the old and the proposed structure since there will be a consistent number of members in each category. However, the smaller the employer gets, the more likely the population may not be as evenly spread out and therefore a mismatch can occur between the assumption and the actual experience. For the ultimate segment of this assumption, we will be using years from retirement eligibility as our grouping metric. Analysis has shown that the pattern of termination becomes highly correlated as members get closer to their retirement age. Traditionally, ultimate assumptions have been based on age, meaning that all members who are a given age and have more than 10 years of service will have the same termination rate. For some defined benefit retirement plans that are designed to base the retirement eligibility solely on age, this is typically a reasonable approach. However, for TMRS with the service only retirement eligibility, a 45 year old member with 18 years of service will likely have a lower probability of turnover than a 45 year old with 10 years of service as the former only has to work 2 more years to retire. The following exhibit provides the experience for males in the ultimate period measured by years from retirement. 25

29 Section III Analysis of Experience and Recommendations Member Category As displayed in the following graph, actual experience shows that there are distinct differences in the termination patterns for the diverse three member groups within TMRS, especially for firefighters. Therefore, our second proposed change to the structure is to introduce multipliers to the base tables based on the category of the individual member: Police, Fire, or Other. 26

30 Section III Analysis of Experience and Recommendations We believe this change will provide for a more objective distinction in the termination pattern between employers than the current structure which applies the same pattern to all individual members. For example, some employers have separate pension plans for their firefighters, so their TMRS population is only Police and Other employees. This employer will likely have greater turnover than an employer that includes firefighters. Under the current structure, this would be attained by placing the first group in a higher termination rate category. Under the new structure, these two employers may be in the same category because their non-firefighter populations may have similar experience, but the second employer would have their firefighters explicitly valued with the lower turnover assumption. Therefore, the following multipliers will be applied to the base tables to determine the termination pattern for each member in the actuarial valuation based upon the category of that member: Employer Multiplier Category Select Period Ultimate Period Police 92% 80% Fire 64% 54% Other 105% 109% Our final recommended change to this overall assumption structure will be to move away from the current five categories by gender to individual multipliers for each employer based on the experience of that employer in relation to the group as a whole. The following exhibit provides the distribution of A/E ratios for the employers with more than 500 exposures in the study period. We have also included the A/E ratio for the current five categories by grouping those employers together. Notice that the distribution has no natural breaks where employers should be separated between the categories. In addition, notice that the gap between the categories is 10% or larger. This means that to make a change to an employer s termination pattern requires a rather large change to the contribution rate. By changing to the proposed structure of individual multipliers, each employer will be given a multiplier between 75% and 125%. This means the changes that do occur could be much smaller in magnitude and will allow for minor changes within the assumption setting process. We also recommend that there no longer be separate categories by gender. Instead, we recommend that a base multiplier be applied to all members of a specific city. Therefore, the final assumption applied to a specific member will be the base table loaded by the employer multiplier and the member category. For example, if the member s age and service create a termination rate of 10% from the base table, the member is classified in the Police 27

31 Section III Analysis of Experience and Recommendations category (92% load), and the individual employer has a multiplier of 90%, then the termination decrement used in the valuation will be 10% * 92% * 90% = 8.3%. How employers were analyzed Generally, employers were ranked objectively based on a weighted A/E ratio. The ratio was determined by taking 1/3 of the A/E ratio during the select period and 2/3 of the A/E ratio during the ultimate period. The ultimate period was given a higher proportion because this segment of the termination assumption has a larger impact on the actuarial liability calculations, especially under the Projected Unit Credit cost method. We limited this process by not allowing a multiplier for an employer to change from the current category multiplier by more than 10%. If an employer ultimately needs to have their multiplier changed even further, the next experience study will allow for that transition to continue. In addition, we placed limits on the multiplier for smaller cities. Smaller cities experience higher turnover in general relative to larger cities. However, if an individual employer experiences very low turnover, the impact on the contribution rates will be substantially larger for smaller employers 28

32 Section III Analysis of Experience and Recommendations because there is less payroll over which to spread the losses. Moreover, to provide conservatism for small employers, we have not allowed the multiplier to be set higher than 100% for employers with between 11 and 15 active members, 85% for employers with between 6 and 10 active members, and 75% for employers with less than 5 active members. Final Results Detailed analysis results are shown in Section VI on pages 58 through 65. In the aggregate, the current assumptions produce an A/E ratio of 88% in total and 77% during the ultimate period. For this assumption, A/E ratios over 100% are conservative. Based on our proposed recommended assumptions, the overall A/E is 99% and 102% during the ultimate period. Changes in this assumption will increase the liabilities and contribution requirements for most employers. R E T I R E M E N T R ATES We currently use rates of retirement that vary by age, gender, and entry age range. Experience showed that 7,496 active members retired during the study period compared to 8,237 who were expected to retire resulting in an A/E ratio of 91%. (Note that these numbers exclude previously terminated members who retired during the period.) Although somewhat lower than ideal, an A/E ratio of 91% indicates a relatively good fit overall especially given the economic uncertainty in years 2008 and In most retirement systems with more typical defined benefit structures, an A/E ratio less than 100% is actually desired for conservatism since it is generally more valuable to the member to commence earlier rather than delay retirement to gain additional accruals. However, the cash balance design of TMRS makes the liabilities less sensitive to retirement patterns. A thorough analysis of the retirement rates is still needed to more accurately calculate a municipality s liability and project future cash flows. The current retirement rates at a given age are generally higher for earlier entry ages indicating that members who have been in TMRS longer and had the opportunity to accumulate larger account balances are more likely to retire. However, this doesn t take into account the impact that plan provisions may have on a member s behavior. The member contribution rate and employer match both directly impact the income replacement ratio that a member will receive immediately upon retirement, while the expectation of receiving a Cost of Living Adjustment (COLA) after retirement allows for some security that inflation won t erode the purchasing power of the calculated retirement benefit. Based on an analysis of these factors, we determined that a change in the structure of the assumption was necessary for members who are under age 62. We maintained the current retirement tables as base rates and made two adjustments based on a city s plan provisions: 1) employee contribution rate plus the employer match and 2) a recurring COLA. For example, 100% of the base rates are used for cities with a 7% employee contribution rate and a 2 to 1 match (total 21% of pay contribution), but only 75% of the base retirement rates 29

33 Section III Analysis of Experience and Recommendations are used for cities with a 5% employee contribution rate and a 1 to 1 match (total 10% of pay contribution). The percent of the base rates applied for cities with other provisions is determined through interpolation using the total contribution as a percent of pay. In addition, only 90% of the rate calculated based on the employee contribution rate and employer match is used if the city does not have a recurring COLA. Neither adjustment is made to the base rate for members who are age 62 or older. The following chart shows the A/E ratio by age band and total contribution as a rate of pay (the 15% total contribution rate is comprised of cities with a 5% employee rate & 2 to 1 match and a 6% employee rate & 1 ½ to 1 match). In total, the A/E ratios vary from 64% for cities with a 5% employee rate and 1 to 1 match to 79% for cities with a 15% total contribution rate and 92% for cities with a 7% employee rate and 2 to 1 match. 120% A/E Ratios by Employee Contribution Rate & Employer Match 100% 80% 60% 40% 20% 0% & over Age Band A/E With 5% EEC & 1-1 Match (10% total contribution) A/E With 5% EEC & 2-1 Match or 6% EEC & Match (15% total contribution) A/E With 7% EEC & 2-1 Match (21% total contribution) The next chart shows the A/E ratio by age band for cities with no COLA compared with cities having a recurring COLA. In total the A/E ratio for cities without a recurring COLA was only 74% compared to 90% for cities with a recurring COLA. It is worth noting that the presence of a recurring COLA rather than the percentage of CPI granted appears to be the key determinant. 30

34 Section III Analysis of Experience and Recommendations 120% A/E Ratio by Presence of Recurring COLA 100% 80% 60% 40% 20% 0% & over A/E with no COLA A/E with Recurring COLA After changing the retirement rates based upon the plan design, the A/E ratio increases from 89% to 92%. This change to the structure of the retirement rates will also allow for more accurate cost analysis of proposed alternate benefit designs that are available to TMRS cities. Additional details are shown on pages 66 through 68. O T H E R A S S U M P T I O N S There are other assumptions made in the course of an actuarial valuation, such as those listed below, and GRS believes that these are generally realistic, accurate and somewhat conservative. Therefore, it is recommending only one minor change at this time for members electing refunds at termination, as described below. 1. Percent Married: 100% of employees are assumed to be married. 2. Members are assumed to take a Partial Lump Sum Distribution equal to 40% of their Employee Savings Fund balance at retirement. 3. Age difference: Male members are assumed to be three years older than their spouses, and female members are assumed to be three years younger than their spouses. 4. Percent of vested members electing refunds at termination (forfeiture rates): Decrease the forfeiture rates for vested members not eligible for retirement. Also, base the rates on age rather than service and introduce adjustments based on the employer match. 31

35 Section III Analysis of Experience and Recommendations 5. Percent electing annuity on death (when eligible): All of the spouses of married participants who die after becoming eligible for a retirement benefit are assumed to elect an annuity that commences immediately, in lieu of a deferred annuity or a refund. 6. Assumed age for commencement of deferred benefits: Members electing to receive a deferred benefit are assumed to commence receipt at the maximum of age 60 or the age first eligible for retirement. 7. Administrative expenses: The assumed investment return rate represents the anticipated net return after payment of all investment and administrative expenses. A C T U A R I A L C O S T M E T H O D The actuarial funding cost method currently being utilized by TMRS is known as the Projected Unit Credit Actuarial Cost Method. The Projected Unit Credit Actuarial Cost Method develops the annual cost of the Plan in two parts: that attributable to benefits accruing in the current year, known as the normal cost, and that due to service earned prior to the current year, known as the amortization of the unfunded actuarial accrued liability. The normal cost and the actuarial accrued liability are calculated individually for each member. The normal cost is the present value of the portion of projected benefits that is attributable to service accrued in the current year. T he unfunded actuarial liability reflects the difference between the portion of projected benefits attributable to service credited prior to the valuation date and assets already accumulated. The unfunded actuarial accrued liability is paid off in accordance with a specified amortization procedure. For cities with three or more employees, the amortization payment as of the valuation date is based on a level percentage of payroll over a closed period of either 25 or 30 years. Even though a change is not recommended in this experience study, based on the published preliminary views of the Governmental Accounting Standards Board (GASB) detailing future changes to financial disclosures for TMRS municipalities, and TMRS system-wide change to a structure typical of most defined benefit plans, the Entry Age Normal Actuarial Cost Method could be give future consideration by the Board as a viable alternative funding method. It is the most common actuarial cost method among statewide pension funds. By design, its primary attribute is that it keeps normal costs level as a percentage of payroll. A C T U A R I A L A S S E T V A L U A T I O N M E T H O D Actuaries generally recommend using a smoothed actuarial value of assets (AVA), rather than market value (MVA), in order to dampen the fluctuations in measurements such as the funded status and the GASB Annual Required Contribution (ARC). The actuarial value of assets is based on the market value of assets with ten-year smoothing applied. This is accomplished by recognizing each year 10% of the difference between the market 32

36 Section III Analysis of Experience and Recommendations value of assets and the expected actuarial value of assets, based upon the assumed valuation rate of return. We continue to believe this method is appropriate. It does not distinguish between types of return (interest, dividends, realized gains/losses, and unrealized gains/losses), like some other methods. It treats different asset classes and different investment styles the same. We do not believe the method has a bias relative to market. In other words, we expect the ratio of the AVA to MVA to average about 100% over the very long term. We believe this method does a good job of smoothing asset gains and losses, and reduces fluctuations in the funding periods. Therefore, we are not recommending a change to this method. The actuarial value of assets is further adjusted by 33% of any difference between the initial value and a 25% corridor around the market value of assets, if necessary. Based on a restructured TMRS Fund, we believe this corridor should be decreased as the current leverage on the MAFs will be removed. We are recommending a 15% corridor for the restructured TMRS Fund. I N T E R E S T R E S E R V E Based on the proposed restructured System, we recommend substantially reducing, if not eliminating, the current 20% reserve policy. We believe the asset smoothing techniques, described above, along with a comparatively lower investment return assumption will provide an adequate reduction in contribution rate volatility. 33

37 SECTION IV A C T U A R I A L IMPA C T O F RECOMMENDAT I O N S

38 Section IV Actuarial Impact of Recommendations Actuarial Impact of Recommendations (Based on the December 31, 2010 valuation) Calculation of Contribution Requirements ($ Millions) TMRS in Total Current Proposed Difference (1) (2) (2) - (1) 1. a. Actives & Inactives $ 13,330 $ 12,738 $ (592) b. Annuitants 2,319 7,598 5, Total actuarial accrued liability (a + b) $ 15,649 $ 20,336 $ 4, Actuarial value of assets 10,761 16,750 5, UAAL (2-3) $ 4,888 $ 3,586 $ (1,302) 5. Funded ratio (3 / 2) 68.8% 82.4% 13.6% 6. UAAL/Payroll 104.1% 76.4% -27.7% Contribution Rate for TMRS Plan Year: 7. Full retirement rate a. Normal cost 9.81% 9.07% -0.74% b. Prior service 6.26% 4.37% -1.89% c. Full retirement rate 16.07% 13.44% -2.63% 8. Estimated Contributions $ 823 $ 688 $ (135) 35

39 Section IV Actuarial Impact of Recommendations 120 New Assumptions compared to Current Rates, PUC No Reserve, Additional Distribution by Interest Credit Cities With More Than 5 Contributing Members

40 SECTION V S U M M A RY OF NEW A S S UM P T I O N S

41 Section V Summary of New Assumptions Summary of New Assumptions I. Economic Assumptions A. General Inflation General Inflation is assumed to be 3.00% per year. B. Discount/Crediting Rates 1. System-wide Investment Return Assumption: 7.00% per year, compounded annually, composed of an assumed 3.00% inflation rate and a 4.00% net real rate of return. This rate represents the assumed return, net of all investment and administrative expenses. This is the discount rate used to value the liabilities of the individual employers. 2. For the Supplemental Death Benefits Fund, the rate is 4.25% per year, compounded annually, and derived as a blend of 5.00% for the portion of the benefits financed by advance funding contributions and a short-term interest rate for the portion of the benefits financed by current contributions. 3. Assumed discount/crediting rate for Supplemental Disability Benefits Fund and individual employee accounts: an annual rate of 5.00% for (1) accumulating prior service credit and updated service credit after the valuation date, (2) accumulating the employee current service balances, (3) determining the amount of the monthly benefit at future dates of retirement or disability, and (4) calculating the actuarial liability of the system-wide Supplemental Disability Benefits Fund. 38

42 Section V Summary of New Assumptions C. Overall Payroll Growth 3.00% per year, which is used to calculate the contribution rates for the retirement plan of each participating city as a level percentage of payroll. This represents the expected increase in total payroll. This increase rate is solely due to the effect of wage inflation on salaries, with no allowance for future membership growth. D. Individual Salary Increases Salary increases are assumed to occur once a year, on January 1. Therefore, the pay used for the period year following the valuation date is equal to the reported pay for the prior year, increased by the salary increase assumption. Age Rate (%) & over 3.50 The above age-related rates are assumed for service with more than 10 years of service. For participants with 10 years of service or less, salaries are assumed to increase by the following graduated scale. Years of Service Rate (%) E. Annuity Increase The Consumer Price Index (CPI) is assumed to be 3.00% per year prospectively. Annuity Increases, when applicable, are 30%, 50%, or 70% of CPI, according to the provisions adopted by each city. 39

43 Section V Summary of New Assumptions II. Demographic Assumptions A. Termination Rates 1. For the first 10 years of service, the base table rates vary by gender, entry age, and length of service. For each city the base table is then multiplied by 75% to 125%. A further multiplier is applied depending on an employee s classification: 1) Fire 64%, 2) Police 92%, or 3) Other 105%. A sample of the base rates follows: Males Service Age Females Service Age

44 Section V Summary of New Assumptions 2. After 10 years of service, base termination rates vary by gender and by the number of years remaining until first retirement eligibility. For each city the base table is then multiplied by 75% to 125%. A further multiplier is applied depending on an employee s classification: 1) Fire 54%, 2) Police 80%, or 3) Other 109%. A sample of the base rates follows: Years from Retirement Male Female % 2.19% % 3.07% % 3.74% % 4.31% % 4.80% % 5.25% % 5.66% % 6.04% % 6.40% % 6.74% % 7.06% % 7.37% % 7.66% % 7.94% % 8.22% Termination rates end at first eligibility for retirement B. Forfeiture Rates (Withdrawal of Member Deposits from TMRS) for vested members vary by age and employer match, and they are expressed as a percentage of the termination rates shown in (A). The withdrawal rates for cities with a 2-to-1 match are shown below. 2% is added to the rates for 1-1½-to-1 cities, and 4% is added for 1-to-1 cities. Percent of Terminating Employees Choosing to Age Take a Refund % % % % % % % Forfeiture rates end at first eligibility for retirement. 41

45 Section V Summary of New Assumptions C. Service Retirees and Beneficiary Mortality Rates 1. For calculating the actuarial liability and the retirement contribution rates, the Gender-distinct RP2000 Combined Healthy Mortality Table projected to the year 2003 by Scale AA. Age Males Females For determining the amount of the monthly retirement benefit at the time of retirement, the UP-1984 Table with an age setback of two years for retirees and an age setback of eight years for beneficiaries. D. Disabled Annuitant Mortality Rates 1. For calculating the actuarial liability and the retirement contribution rates, the gender-distinct RP2000 Disabled Retiree Mortality Table with both male and female rates multiplied by 80%. Age Males Females For determining the amount of monthly retirement benefit at the time of retirement, the UP-1984 Table with an age setback of two years for retirees and an age setback of eight years for Beneficiaries. 42

46 Section V Summary of New Assumptions E. Pre-Retirement Mortality Rates-Gender-distinct RP2000 Combined Healthy Mortality Table projected to the year 2003 by Scale AA, with a 5 year setback for both males and females F. Disability Rates Age Males Females Age Males Females

47 Section V Summary of New Assumptions G. Service Retirement Rates, applied to both Active and Inactive Members The base table rates vary by gender, entry age group, and age. These rates are adjusted then multiplied by 2 factors based on 1) employee contribution rate and employer match and 2) if the city has a recurring COLA. Males Females Entry Age Groups Entry Age Groups Ages 32 Ages Ages 48 Ages 32 Ages Ages 48 Age & Under & Over & Under & Over and over Note: For cities without a 20-year/any age retirement provision, the rates for entry ages 32 and under are loaded by 20% for ages below 60. Plan Design Factors Applied to Base Retirement Rates Employee Contribution Rate Employer Match 5% 6% 7% Recurring COLA: 100% No Recurring COLA: 90% 44

48 Section V Summary of New Assumptions III. Methods and Assumptions A. Valuation of Assets The actuarial value of assets is based on the market value of assets with ten-year smoothing applied. This is accomplished by recognizing each year 10% of the difference between the market value of assets and the expected actuarial value of assets, based upon the assumed valuation rate of return. The actuarial value of assets is further adjusted by 33% of any difference between the initial value and a 15% corridor around the market value of assets, if necessary. B. Small City Methodology For cities with fewer than twenty employees, more conservative methods and assumptions are used. First, lower termination rates are used for smaller cities, with maximum multipliers of 75% for employers with less than 6 members, 85% for employers with 6 to 10 members, and 100% for employers with 11 to 15 members. There is also load on the life expectancy for employers with less than 15 active members. The life expectancy will be loaded by decreasing the mortality rates by 1% for every active member less than 15. For example, an employer with 5 active members will have the baseline mortality tables multiplied by 90% (10 years times 1%). The maximum amortization period for amortizing gains and losses is decreased from current levels by 1 year for each active member less than the 20 member threshold. For example, an employer with 8 active members and a current maximum amortization period of 25 will use (25-(20-8)) = 13 year amortization period for the gain or loss in that year s valuation. Under this policy, the lowest amortization period will be 25-(20-1) = 6 years. C. Actuarial Cost Method: The actuarial cost method being used is known as the Projected Unit Credit Actuarial Cost Method. The Projected Unit Credit Actuarial Cost Method develops the annual cost of the Plan in two parts: that attributable to benefits accruing in the current year, known as the normal cost, and that due to service earned prior to the current year, known as the amortization of the unfunded actuarial accrued liability. The normal cost and the actuarial accrued liability are calculated individually for each member. The normal cost is the present value of the portion of projected benefits that is attributable to service accrued in the current year. The unfunded actuarial liability reflects the difference between the portion of projected benefits attributable to service credited prior to the valuation date and assets already accumulated. The unfunded actuarial accrued liability is paid off in accordance with a specified amortization procedure. For cities with twenty or more employees, the amortization as of the valuation date is a level percentage of payroll over a closed period of either 25 or 30 years. 45

49 Section V Summary of New Assumptions Under the Projected Unit Credit Actuarial Cost Method, if actual plan experience is close to assumptions, the normal cost will increase each year for each employee as he or she approaches retirement age. However, if the age/service/gender characteristics of the active group remain constant, the total normal cost can be expected to remain somewhat level as a percentage of payroll. The total contribution is made up of the sum of the individual normal costs and the amortization payment on the unfunded actuarial accrued liability. 46

50 SECTION VI S U M M A RY OF DATA A N D E X P E R I E N C E

51 Section VI Summary of Data and Experience List of Tables SALARY EXPERIENCE POST-RETIREMENT MORTALITY EXPERIENCE FOR NON-DISABLED MALE RETIREES POST-RETIREMENT MORTALITY EXPERIENCE FOR NON-DISABLED FEMALE RETIREES POST-RETIREMENT MORTALITY EXPERIENCE FOR DISABLED MALE RETIREES POST-RETIREMENT MORTALITY EXPERIENCE FOR DISABLED FEMALE RETIREES PRE-RETIREMENT MORTALITY EXPERIENCE FOR ACTIVE MALES PRE-RETIREMENT MORTALITY EXPERIENCE FOR ACTIVE FEMALES DISABILITY EXPERIENCE FOR ACTIVE MALES DISABILITY EXPERIENCE FOR ACTIVE FEMALES SERVICE BASED TERMINATION EXPERIENCE ALL EMPLOYEES SERVICE BASED TERMINATION EXPERIENCE GENERAL EMPLOYEES SERVICE BASED TERMINATION EXPERIENCE POLICE SERVICE BASED TERMINATION EXPERIENCE FIREFIGHTERS ULTIMATE TERMINATION RATES ALL EMPLOYEES ULTIMATE TERMINATION RATES GENERAL EMPLOYEES ULTIMATE TERMINATION RATES POLICE ULTIMATE TERMINATION RATES FIREFIGHTERS MALE RETIREMENT EXPERIENCE FEMALE RETIREMENT EXPERIENCE RETIREMENT EXPERIENCE BY PLAN DESIGN ELEMENT

52 Section VI Summary of Data and Experience SALARY EXPERIENCE Current Salary Scale 03/09Actual Experience Proposed Salary Scale Years of Step Rate/ Above Step Rate/ Step Rate/ Service Total Promotional Total Inflation Promotional Total Promotional (1) (2) (3) (4) (5) (6) (7) (8) % 7.60% 20.97% 18.41% 15.97% 12.00% 7.60% % 4.60% 11.16% 8.60% 6.16% 9.00% 4.60% % 2.60% 7.99% 5.43% 2.99% 7.00% 2.60% % 2.60% 7.47% 4.91% 2.48% 7.00% 2.60% % 1.60% 6.96% 4.40% 1.96% 6.00% 1.60% % 1.60% 6.89% 4.33% 1.89% 6.00% 1.60% % 1.10% 6.35% 3.79% 1.36% 5.50% 1.10% % 1.10% 6.14% 3.58% 1.14% 5.50% 1.10% % 1.10% 5.97% 3.41% 0.97% 5.50% 1.10% % 1.10% 5.86% 3.30% 0.86% 5.50% 1.10% % 0.00% 5.00% 2.44% 0.00% 4.40% 0.00% Current Inflation Assumption 3.00% Proposed Inflation Assumption 3.00% Current Productivity Component 1.40% Proposed Productivity Component 1.40% Actual CPI-U Inflation for Dec/02 - Dec/ % Apparent Productivity Component 2.44% 49

53 Section VI Summary of Data and Experience NON-DISABLED RETIREES POST-RETIREMENT MORTALITY - MALES Assumed Rate Expected Deaths Actual / Expected Actual Total Actual Current Proposed Age Deaths Count Rate Current Proposed Current Proposed (2) / (7) (2) / (8) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) , % 220% , % 159% , % 136% , % 108% , % 113% , % 118% , % 109% , % 115% % 135% % 121% N/A N/A Totals 1,473 53,279 1,283 1, % 119% Male + Females 1,850 72,877 1,725 1, % 113% *Columns may not add due to rounding. 50

54 Section VI Summary of Data and Experience NON-DISABLED RETIREES POST-RETIREMENT MORTALITY - FEMALES Assumed Rate Expected Deaths Actual / Expected Actual Total Actual Current Proposed Age Deaths Count Rate Current Proposed Current Proposed (2) / (7) (2) / (8) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) , % 300% , % 275% , % 96% , % 92% , % 73% , % 91% , % 101% % 90% % 74% % 150% % 100% Totals , % 96% *Columns may not add due to rounding. 51

55 Section VI Summary of Data and Experience DISABLED RETIREES POST-RETIREMENT MORTALITY - MALES Assumed Rate Expected Deaths Actual / Expected Actual Total Actual Current Proposed Age Deaths Count Rate Current Proposed Current Proposed (2) / (7) (2) / (8) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) % 107% % 119% % 78% % 100% % 110% % 82% % 100% % 100% % 100% % 0% N/A N/A N/A Totals 135 3, % 98% Male + Females 171 4, % 108% *Columns may not add due to rounding. 52

56 Section VI Summary of Data and Experience DISABLED RETIREES POST-RETIREMENT MORTALITY - FEMALES Assumed Rate Expected Deaths Actual / Expected Actual Total Actual Current Proposed Age Deaths Count Rate Current Proposed Current Proposed (2) / (7) (2) / (8) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) % 350% % 400% % 80% % 100% % 50% % 100% % 200% N/A N/A % 100% N/A N/A N/A N/A N/A N/A Totals % 171% *Columns may not add due to rounding. 53

57 Section VI Summary of Data and Experience MALE PRE-RETIREMENT MORTALITY Assumed Rate Expected Deaths Actual / Expected Actual Total Actual Current Proposed Age Deaths Count Rate Current Proposed Current Proposed (2) / (7) (2) / (8) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Under % 0% , % 350% , % 100% , % 60% , % 83% , % 88% , % 84% , % 82% , % 59% , % 48% , % 42% % 33% 75 and over % 14% Totals , % 71% Male + Females , % 67% *Columns may not add due to rounding. 54

58 Section VI Summary of Data and Experience FEMALE PRE-RETIREMENT MORTALITY Assumed Rate Expected Deaths Actual / Expected Actual Total Actual Current Proposed Age Deaths Count Rate Current Proposed Current Proposed (2) / (7) (2) / (8) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Under N/A N/A , % 100% , % 50% , % 200% , % 140% , % 22% , % 80% , % 35% , % 50% , % 40% , % 55% % 60% 75 and over % 0% Totals , % 54% *Columns may not add due to rounding. 55

59 Section VI Summary of Data and Experience MALE DISABILITY EXPERIENCE Assumed Rate Expected Disabilities Actual / Expected Actual Total Actual Current Proposed Age Disabilities Count Rate Current Proposed Current Proposed (2) / (7) (2) / (8) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Under N/A N/A , % 0% , % 100% , % 160% , % 56% , % 84% , % 73% , % 98% , % 104% , % 63% % 100% % 200% 75 and over N/A N/A Totals , % 89% *Columns may not add due to rounding. 56

60 Section VI Summary of Data and Experience FEMALE DISABILITY EXPERIENCE Assumed Rate Expected Disabilities Actual / Expected Actual Total Actual Current Proposed Age Disabilities Count Rate Current Proposed Current Proposed (2) / (7) (2) / (8) (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) Under % 0% , % 0% , % 0% , % 0% , % 245% , % 147% , % 85% , % 139% , % 76% , % 57% % 0% % 0% 75 and over % 880% Totals 78 98, % 107% *Columns may not add due to rounding. 57

61 Section VI Summary of Data and Experience *Columns may not add due to rounding. 58

62 Section VI Summary of Data and Experience *Columns may not add due to rounding. 59

63 Section VI Summary of Data and Experience *Columns may not add due to rounding. 60

64 Section VI Summary of Data and Experience *Columns may not add due to rounding. 61

65 Section VI Summary of Data and Experience *Columns may not add due to rounding. 62

66 Section VI Summary of Data and Experience *Columns may not add due to rounding. 63