Texas Municipal Retirement System Retiree Mortality Study June 20, 2013 Joseph Newton Mark Randall Copyright 2012 GRS All rights reserved.
Today s Agenda Mortality Assumption Overview Actuarial Standards of Practice National and Local Trends Valuation Mortality Assumption Current Annuity Purchase Factors Proposed Annuity Purchase Factors GRS Recommendation 2
Mortality Assumption The Board had requested an analysis of the mortality assumptions used in all of the actuarial processes of TMRS Valuation Mortality Assumption Used in determining liabilities and employer contribution requirements Annuity Purchase Rates Used to convert member balances to annuities at retirement Include partial lump sum and survivor options There is no requirement that the same mortality assumption be used for both Valuation assumption dictated more by actuarial standards Annuity Purchase Rates dictated more by policies and legal restraints However, if they are not set equal, there should be a general understanding of why and the possible impact 3
History of Mortality Assumptions Valuation assumption: In 1981, UP 1984 mortality table (retiree - 2, beneficiary -7) In 1988, UP 1984 (retiree -1, beneficiary -8) In 2007, updated to RP- 2000 mortality table In 2011, updated to RP- 2000 mortality table, projected forward three years Annuity Purchase Rate: In 1981, annuity purchase rates set based on UP 1984 mortality table (retiree -2, beneficiary -8) 4
Life Expectancy for the General US Population - from Age 65 25.0 20.0 15.8 16.8 18.4 19.0 19.1 20.3 15.0 10.0 13.0 13.0 14.2 15.1 16.1 17.7 5.0-1960 1970 1980 1990 2000 2010 Females Males Since 2010, life expectancies continue to increase. The latest published rates (2011) are 20.4 years for females and 17.8 years for males, both from age 65. 5 Source: National Vital Statistics Reports
What does Life Expectancy in a given year mean? It is important to understand how to interpret the data in the statistics When females age 65 in 2010 had a life expectancy of 20.3, that does not actually mean people who are age 65 in 2010 will live 20.3 years. In fact, they will live longer Instead, the 20.3 years represents a hypothetical life expectancy for the population using the probability of death from 2010 at all ages above age 65. We will defined this as static This is similar to watching a race in which the runners will run 4 laps. If you are sitting at the finish line at the end of lap 3, you can: Tell who is ahead Tell how fast they ran the last lap, or last quarter lap, etc. Tell who appears to be running the fastest Project out the final outcome and times The snapshot of the race at the end of lap 3 is the static, the projection of the final outcome is projected The question becomes how to project future probabilities of death? There are standard, published actuarial techniques to do this 6
Actuarial Standards of Practice No. 35 Selection of Demographic and Other Noneconomic Assumptions for Measuring Pension Obligations From Background Section: As mortality rates have continued to decline over time, concern has increased about the impact of potential future mortality improvements on the magnitude of pension commitments. In the view of many actuaries, the guidance regarding mortality assumptions should more explicitly recognize estimated future mortality improvement as a fundamental and necessary assumption, and the actuary s provision for such improvement should be disclosed explicitly and transparently. From Section 3.5.3: Mortality and Mortality Improvement Assumptions The actuary should consider the effect of mortality improvement both prior to and subsequent to the measurement date. With regard to mortality improvement, the actuary should do the following: i. adjust mortality rates to reflect mortality improvement prior to the measurement date. ii. include an assumption as to expected mortality improvement after the measurement date. Note that the existence of uncertainty about the occurrence or magnitude of future mortality improvement does not by itself mean that an assumption of zero future improvement is a reasonable assumption. 7
From a recent Report by the Society of Actuaries (SOA) Retirement Plans Experience Committee (RPEC), published in September of 2012 A new projection Scale BB has recently been published, based on very recent data and trends. The following are excerpts from the Report which introduced Scale BB. As part of its periodic review of retirement plan mortality experience, RPEC initiated a Pension Mortality Study in 2010. This study, which is still in progress, includes a comprehensive review of recent mortality experience of uninsured retirement plans in the United States. The SOA anticipates publishing a new set of retirement plan mortality tables and mortality improvement rates in late 2013, or early 2014, that would be the successors to the RP-2000 tables and Scale AA. RPEC recognizes that there is a wide range of opinion with respect to future levels of mortality in the United States and that the assumptions underlying any set of mortality improvement rates must necessarily reflect some degree of subjectivity. RPEC characterizes the assumptions that underpin Scale BB (including a 1.0% long-term rate of mortality improvement and limited cohort effects) as middle-of the road, being neither overly optimistic nor too pessimistic with respect to future longevity improvements in the United States. In light of the nearly continuous pattern of increasing longevity in the United States over the past century, the Committee recommends that actuaries incorporate adequate provisions for future mortality improvement into their calculations. Taking into consideration the methodology used to develop Scale BB (Section 5.3) and RPEC s preference for generational projection of mortality over static approximations (Section 7.1), the committee encourages users of Scale BB to do so on a fully generational basis. 8
Valuation Mortality Assumption An Actual to Expected ratio (A/E) is used to measure the Actual experience to the assumption, or what was Expected For example, if the actuary expects 100 deaths and 80 deaths actually occur during the observation period, A/E ratio would be 80 / 100 which = 80%. An A/E of 100% would be a perfect match Traditionally for this assumption, setting a static assumption with 10% to 20% margin was preferred to allow for future increases in life expectancy An example from the 2011 TMRS Experience Study: The prior valuation mortality assumption expected 1,725 deaths during the observation period, actual deaths were 1,850, or an A/E ratio of 107% (1,850/1,725) GRS recommended updating this assumption to assume 1,632 deaths, or an A/E of 113% (1,850/1,632) Historically (pre 2000), this would have been similar to adding in one (1) decade of mortality improvement However, this has led to upward drift in contribution rates as the assumption is slightly modified every 4-5 years. Over time, this adds up An additional UAAL was created every time the assumption was updated This was a customary practice for two main reasons: (1) a general belief that there was a limit on the ultimate longevity, and (2) the added complexity of implementing a generational mortality type model and limitations in computational power 9
TMRS Experience Prior assumption was based upon the RP-2000 mortality table, projected to 2003 based on Scale AA At the time, this provided approximate 13% margin for improvements in future life expectancies (using 2007-2009 data) Also at the time, RP2000 was the most recently published table and Scale AA was the most recently published projection scale In 2 years, this margin has decreased to 8% (2009-2011 data) This is consistent with other sources of mortality data which shows a recent jump in the rate of longevity improvement 10
TMRS Experience (cont.) Therefore, GRS could recommend an update to the current mortality assumption to add back in 5-7% of additional margin to account for future improvements Projected four (4) more years would produce a 113% A/E ratio This trend has been consistent across most of our clients and in recent national surveys It is becoming more widely accepted that this trend of ever-increasing life expectancies is likely to continue 11
Other Factors to consider Texas ranks below average in life expectancy when compared to the other States In the current member data, 70% of TMRS actives (future retirees) are male. Twenty years ago, this was probably closer to 90%. This should increase life expectancy for the overall population compared to TMRS data 20 years ago 12
Mortality Rates by State Above data from National Vital Statistics 13
Future Recommendation to Valuation Mortality Assumption GRS recommendation would be to move to full generational mortality Instead of a static assumption with margin, the goal would be to find a good fit for today s experience and then project the rates using standard tables We are recommending the use of the RP-2000 table with blue collar adjustment, loaded by 109% for males and 103% for females This recognizes the lower than average life expectancy in the general population of Texas With no other adjustments, contribution rates for employers would increase on average by about ten percent (10%) For example, if the current employer contribution rate is 5.0% of payroll, this adjustment would increase the rate to 5.5%. If current rate is 15.0%, would increase to 16.5% Would add approximately $750 million to the System s Unfunded Actuarial Accrued Liability (UAAL), about a 25% increase. Current UAAL is $2.899 billion With this fully generational projection approach, a gradual and consistent improvement over time would be in the valuation process Future rates would not have to be reset every 4-5 years Keep future UAAL s from being systematically created With no other changes, this would be our recommendation in the next Experience Study, currently scheduled to be summer of 2015 14
Post Retirement Mortality Average Life Expectancy in Years from Current Age 65 35.0 30.0 25.0 20.0 15.0 10.0 5.0-22.0 21.3 21.3 23.1 24.3 25.5 24.5 24.6 24.6 26.4 27.4 28.4 Male Female Current Assumption Actual 2009-2011 Proposed Static Assumption (2010) Proposed Projected Assumption, Baseline 2010 Proposed Projected Assumption, Baseline 2020 Proposed Projected Assumption, Baseline 2030 RP-2000 with blue collar adjustment projected to 2010 by Scale BB, with 109% load for males and 103% load for females. Full projection after 2010 by Scale BB. 15
Annuity Purchase Factors The current factors are based upon the UP 1984 Mortality Table, setback two (2) years A setback is a technique used to recognize some improvement in longevity Setback 2 years means the mortality rate for a 65 year old member will be treated as if the member is 63 years old As discussed previously, the current factors were adopted in 1981 The current factors are not gender specific In 1981, approximately 90% of retiring members were male Today, approximately 70% of future retirees are male The current factors create an A/E ratio of 80% Underestimates life expectancies, based on the static 2010 data, by approximately 7.5%, or 1.6 years for a retiring 60 year old Underestimates life expectancy for a retiring 60 year old by approximately 13%, or 2.7 years when projected improvement in mortality is incorporated With a 5% discount rate, the current factors are underpriced by 10-11% The difference between the current factors and the experience will continue to widen by about 1 year, or a little over 3%, per decade 16
Annuity Purchase Factors (cont.) If it is desired to update the factors, we would recommend a projected, generational approach Factors will change slowly over time to reflect mortality improvements Projection scales would be put in place today so that members could plan their retirements The factors need to continue to not be gender specific Proposed assumption is that 70% of future retiring members will be male The factors would be based on the same mortality assumption as used in the valuation process However, it would be prudent to phase into the new factors since no member should be provided an incentive to retire to protect a current benefit Can create losses due to a high number of retirements Creates human resource difficulties 17
Proposed Assumption for Annuity Purchase Rates Average Life Expectancy in Years from Indicated Current Age 35.0 30.0 25.0 20.0 15.0 10.0 5.0 18-55 60 65 70 75 80 Current Factors 24.4 20.4 16.8 13.5 10.6 8.1 Proposed Projected Assumption (2010) 28.7 23.8 19.3 15.2 11.6 8.6 RP-2000 with blue collar adjustment projected to 2010 by Scale BB, with 107.5% load, weighted 70%/30% male/female
Phase In If the factors changed overnight, members currently eligible to retire could see their annuity decrease by as much as 9% to 11%. This could take at least a year to make up the difference, and many people may choose to retire immediately Based on the current factors, once eligible to retire, the average member will realize an increase in their annuity at about 10% per year due to new contributions, a 5% interest credit on their prior balance, and a smaller annuity factor 19
Phase In an example Let s assume someone age 50 is employed at a 7% and 2-to-1 match TMRS city This individual is currently earning $45,000 annually and has 20 years of service with the TMRS city This member would have approximately a $994 per month annuity if they retire If this member puts off retirement for one more year, their monthly annuity would increase to about $1,101, or about a 10.8% increase, assuming a 3% annual salary increase This year-over-year increase will lower to about 10.0% per year by age 55 20
Phase In an example (cont.) If the new factors were put in place immediately, the age 50 annuity would decrease from $994 to $909, a decrease of 8.6% Our preferred method of phase in would extend the full implementation by 13 years Thirteen years was chosen because the average member would still receive a 9% increase in their annuity each year (1% less than the current growth rate) and a vast majority of cities would not have an increase in rate With the phase-in, no member would ever have a decrease month over month in their annuity 21
Proposed Factors (Life Only) - with a 13-year phase in 18.0 16.0 14.0 12.0 10.0 8.0 6.0 4.0 2.0 Annuity Factor for a retiring 60 year old 22-2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 Current Factors 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 13 Year Phase In 12.1 12.2 12.4 12.5 12.7 12.8 13.0 13.1 13.2 13.4 13.5 13.6 13.7 13.8 Proposed Unadjusted Factors 13.4 13.4 13.5 13.5 13.5 13.6 13.6 13.6 13.7 13.7 13.7 13.8 13.8 13.8 RP-2000 with blue collar adjustment projected to 2010 by Scale BB, with 107.5% load, weighted 70%/30% male/female
Projected Annuity for Example Member A Member reaching age 50 with 20 years of service, $45,000 salary, 7% and 2-to-1 City $6,000 $5,000 $4,000 $3,000 $2,000 $1,000 $0 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 Current Factors 1,392 1,541 1,705 1,883 2,078 2,290 2,523 2,778 3,058 3,363 3,699 4,068 4,474 4,920 13 Year Phase In 1,392 1,525 1,667 1,821 1,988 2,166 2,359 2,568 2,793 3,038 3,303 3,591 3,906 4,248 Proposed Unadjusted Factors 1,273 1,404 1,547 1,702 1,870 2,055 2,254 2,471 2,708 2,965 3,245 3,550 3,885 4,248 Approximately 30% of members choose a pure life only, with another 20% choosing a certain and life 23
Projected Annuity for Example Member A Member reaching age 50 with 20 years of service, $45,000 salary, 7% and 2-to-1 City $600,000 $500,000 $400,000 $300,000 $200,000 $100,000 $0 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 Additonal Subsidy From Current Factors - 2,224 4,886 8,020 11,661 15,839 20,604 26,043 32,180 39,031 46,664 55,125 64,478 74,767 Subsidy From Phase-in 16,330 16,327 16,147 15,746 15,126 14,273 13,199 11,927 10,437 8,724 6,807 4,701 2,425 - Account Balance 173,508 189,197 205,916 223,559 242,349 262,323 283,493 305,999 329,961 355,185 382,049 410,599 440,829 472,726 24
Projected 100% Joint & Survivor Annuity for Example Member A Member reaching age 50 with 20 years of service, $45,000 salary, 7% and 2-to-1 City $4,500 $4,000 $3,500 $3,000 $2,500 $2,000 $1,500 $1,000 $500 $0 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 Current Factors 1,194 1,314 1,445 1,585 1,738 1,903 2,082 2,277 2,488 2,717 2,966 3,237 3,533 3,853 13 Year Phase In 1,194 1,308 1,431 1,563 1,706 1,859 2,024 2,202 2,395 2,603 2,828 3,072 3,338 3,625 Proposed Unadjusted Factors 1,154 1,268 1,391 1,524 1,667 1,823 1,990 2,171 2,368 2,581 2,811 3,060 3,332 3,625 Annuity is increasing by about 0.5% less per year in the phase in scenario versus the current factors Approximately 50% of members choose a survivor option 25
GRS recommendation We see the change in factors as a reaction to an anticipated future increase in the contribution rates, not as a reaction to current rates If the annuity purchase factors are changed to reflect current expectations, GRS would recommend not letting the contribution rates change, or at least minimizing the change, due to a change in the annuity factors The current valuation process takes the difference between the annuity purchase factors and the actual experience into account. The 13 year phase in was determined by balancing the speed of increase in member annuities with keeping the employer contributions mostly unchanged If a different phase in approach is used, there will be a positive or negative impact to contribution rates The longer the phase in, there will be some increases in contribution rates A shorter phase in could create a decrease in contribution rates 26
Next steps August Board Meeting Estimated effect on City contribution rates and other options Timeline for implementation Communication efforts 27