I O W A P U B L I C E M P L O Y E E S R E T I R E M E N T S Y S T E M

Transcription

1 I O W A P U B L I C E M P L O Y E E S R E T I R E M E N T S Y S T E M A C T U A R I A L A U D I T R E P O R T J U N E 3 0,

2 ABC Retirement System June 15, 2015 Investment Board Iowa Public Employees Retirement System 7401 Register Drive Des Moines, IA Dear Board Members: Gabriel, Roeder, Smith & Company (GRS) is pleased to present this DRAFT report of an Audit of the June 30, 2014 Actuarial Valuation of the Iowa Public Employees Retirement System (IPERS). We are grateful to IPERS Staff for their cooperation throughout the Audit process. In addition, we wish to thank Patrice Beckham and Brent Banister of Cavanaugh Macdonald for their assistance with this project. The Actuarial Audit has several related objectives: Review the appropriateness of the actuarial assumptions, Review assumptions and methods for compliance with Professional Standards, State Law, and Board Regulations, Verify the demographic data through independent collection and processing, and Express an actuarial opinion regarding the reasonableness and/or accuracy of valuation results based upon a test using a mathematical model of plan activity or sampling. In our opinion, the retained actuary s work provides a fair and reasonable assessment of the financial position of IPERS. We are pleased to report that we have found no substantial errors or omissions in the retained actuary s work. Throughout this report, the reader will note items where GRS sees things differently than the retained actuary. Indeed, our mission is to point out such items. In interpreting our comments, Trustees should be aware that while we are pointing out sources of difference, we are in agreement with Cavanaugh Macdonald on the vast majority of items reviewed. Respectfully submitted, Brian B. Murphy, FSA, EA, MAAA David T. Kausch, FSA, EA, MAAA Mita D. Drazilov, ASA, MAAA BBM:DTK:MDD:sc

3 TABLE OF CONTENTS Section I II III IV V VI VII VIII IX X XI XII Item Introduction Review of Data Elements Review of Experience Study and Assumptions Review of Annual Valuation Report Review of Statutes and Board Regulations Test of Valuation Results through Mathematical Modeling and Samples Average Compensation Case Studies Simulation Model Actuarial Funding Method Asset Data and Valuation Method Summary of Recommendations Formal Opinion and Concluding Remarks

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

5 INTRODUCTION The Iowa Public Employees Retirement System (IPERS) issued a Request For Proposal (RFP) for an audit of the June 30, 2014 Actuarial Valuation of IPERS performed by Cavanaugh Macdonald (the retained actuary). Gabriel, Roeder, Smith & Company (GRS, the Auditor) responded to the RFP and was awarded the work. The project commenced in March of An actuarial audit involves a review of the retained actuary s work by an independent actuarial firm. The purpose of the audit was to provide an evaluation sufficient to allow the Auditor to express an actuarial opinion regarding the reasonableness and/or accuracy of valuation results, actuarial assumptions, and actuarial methods in accordance with generally recognized and accepted actuarial principles and practices. This particular audit was not a full replication audit. It was what GRS refers to as a Level Two audit (i.e., a review of the actuarial reports of the plan and a test of the valuation results using a mathematical model of plan activity or sampling). For purposes of this audit: GRS received the same data from the System that the retained actuary received. GRS independently reviewed the data, and compared the original System data with the groomed data finally used by the retained actuary. The retained actuary supplied a detailed set of the actuarial assumptions that were used in the June 30, 2014 valuation. GRS loaded the actuarial assumptions into its computer system and modeled the benefit provisions of IPERS as of June 30, GRS and the retained actuary each use proprietary actuarial valuation programs that were independently developed over periods of years. As part of this audit, results of GRS software are compared with results of the retained actuary s software for a sample of various test lives, and differences are noted. Users of this report should bear in mind that an actuarial valuation involves a large number of intricate calculations and many individual judgments regarding rather arcane items along the way. Two independently written valuation programs will never agree. For actuarial audit purposes, we generally like to see principal valuation results within I-1

6 1% to 2% for retired and deferred vested members, and within 5% for active members. For active members, larger differences may be seen for some valuation results if the level of service for the active test case is relatively small. In the audit, we concentrate on those differences that we believe are important and do not pursue differences that we believe are the result of minor judgment items. In connection with this undertaking, GRS also received and reviewed the following items: Actuarial valuation funding report as of June 30, 2014, GASB Statement No. 67 accounting report as of June 30, 2014, The most recent experience study covering the period , A full set of census data for plan participants and beneficiaries as of June 30, 2014, A complete set of financial data as of June 30, 2014, and Screen shots and supplementary material detailing the calculations of the retained actuary s simulation model for IPERS. As part of the audit, GRS actuaries have commented on the assumptions and various actuarial methods. Once again, different individuals coming from different experience backgrounds often come to different conclusions from looking at the same data. This is no more surprising than two physicians recommending slightly different treatments to the same patient for the same condition. In total, we think that the IPERS is being well served by the retained actuary. Although we would perform certain aspects of the actuarial valuation differently than the retained actuary, and would probably arrive at slightly different assumptions and results, nothing in our report should be construed to mean that the retained actuary s work is at cross purposes with actuarial standards, good judgment, or the best interest of IPERS. I-2

7 SECTION II R E V I E W O F D ATA E L E M E N T S

8 REVIEW OF DATA ELEMENTS GRS received June 30, 2014 participant information from both IPERS Staff and the retained actuary. As part of the actuarial audit process, we performed a broad review of both participant data sets to ensure consistency in what IPERS reported to the retained actuary and what the retained actuary ultimately used for valuation purposes. The data received from the retained actuary included active/inactive and retired member information for the Regular Membership, Sheriffs and Deputies, and Protection Occupation groups. Some of the members included in the information received were excluded from the valuation. It is common for the retained actuary to make some small adjustments to the reported data for valuation purposes. The final data that is used by the retained actuary is sometimes referred to as the groomed data. In general, we found the active and retired data supplied by IPERS to be sufficient to perform the actuarial valuation and the adjustments made by the retained actuary to be reasonable. However, due to the recently enacted benefit changes in 2012, we are recommending that IPERS supply, and the retain actuary use, some additional data items for actuarial valuation purposes. In addition, we are recommending a change to the way the retained actuary is processing two small classes of retirees and beneficiaries for actuarial valuation purposes. Presented below are some observations from our data review. Active and Inactive Member Data This analysis included comparisons based on dates of birth, service, attained age, gender, pays, and status. In general, the data appears reasonable, and number counts match the June 30, 2014 valuation report. IPERS supplies a final average salary field on the active record. This final average salary field is the highest 3-year final average salary as of the valuation date. However, with the recently enacted benefit changes in 2012, it would appear that this field should now be based upon a 5-year final average salary for Regular members. In addition, we would recommend that another field be reported that contains the member s 3-year final average salary as of June 30, 2012 for Regular membership. II-1

9 We would also suggest that the member s control-year salary be reported on the member s record. This will allow the retained actuary to reflect the pay-spiking provision for these groups in the actuarial valuation process. Currently, a member s service is reported in total for each of the three valuation groups. The retained actuary then must make an assumption regarding whether the service was earned before or after June 30, We would recommend that service earned both before and after June 30, 2012 be reported for each of the actuarial valuation groups. Retired Member Data The data analysis consisted of many comparisons in data fields including dates of birth, pension amounts, benefit codes and post-retirement dividend amounts. There were no significant differences in the two data files. Total number counts match the valuation. For members that have adopted a 10-year Certain & Life option that have passed away as of the valuation date, it appears that the retained actuary is incorrectly determining whether a benefit should still be payable based upon a beneficiary s date of birth as opposed to the member s date of retirement. This is resulting in the inappropriate exclusion of approximately 34 records from the actuarial valuation. For members that pass away in the month of the valuation (i.e., June 2014), it appears that the retained actuary is excluding members that may have a continuing benefit to a beneficiary in the case where a joint and survivor benefit was elected. This appears to have affected 39 records in the June 30, 2014 valuation. We recommend that the retained actuary modify their valuation process to include these records in the actuarial valuation. II-2

10 SECTION III R E V I E W O F E X P E R I E N C E STUDY A N D A S S U M P T I O N S

11 EXPERIENCE STUDY AND ASSUMPTIONS In this section of the report we review the study of July 1, 2009 to June 30, 2013 IPERS experience prepared by Cavanaugh Macdonald Consulting and dated May 27, It is our opinion that, in total, the end results of the experience study reasonably represent prior and anticipated future experience under the System. We have identified one small logical inconsistency in the assumption setting process (please see page III-8 Active Members), and recommend that in future studies more attention be paid to the decrement rates by age or service groups than the present study does. Section 1: Executive Summary This section summarizes the main results of the study. It presents results on actuarial methods, economic assumptions, demographic assumptions, and optional form factors. It then presents a page summarizing the effects on actuarial liabilities and contribution rates. We reviewed this section for readability and understandability. In general, we found this section to be a useful summary of the remainder of the study. We have just three comments on this section: 1. On the top of page three, it says The lower price inflation assumption results in an increase in the productivity (real wage inflation) assumption from 0.75% to 1.00%. We would not say that lowering the price inflation assumption results in an increase. While the increase may be correct, we think that the productivity assumption is a separate assumption that should be independently studied. 2. We were not able to understand from page 3, why the 20bp allowance for investment expense is no longer needed. We note that this was clarified in later sections, but page 3 itself is unclear. 3. We would like to see some type of grand summary in the executive summary that gives the reader high level information regarding all of the changes that are being recommended. An example is shown on the following page. III-1

12 Assumption Recommendation Financial Impact Withdrawal rates Higher Rates Decrease Disability rates Lower Rates Decrease Merit pay increases Lower Rates Decrease Retirement rates Various Increase Pre and post-retirement mortality rates Interest rate Wage inflation Price inflation Amortization method Lower Rates No Change Lower Rate Lower Rate No Change Increase N/A Decrease Decrease N/A Total Various Increase Section 2: Introduction This section gives the layperson a good introduction to the principles governing an experience study, and to the actuarial standards that must be followed. We like the description of philosophy on page 9 and find the discussion of ASOPs 27 and 35 to be germane and well written. Our only comment on this section relates to the second sentence on page 7 which we find to slightly overstate the responsibility of the actuary. It says that Our responsibility is to consider the impact our work will have on members, employers, and taxpayers, both current and future. In fact, ASOP 4 limits the extent to which actuaries are required to consider taxpayers and employers by saying that This standard does not require the actuary to evaluate the ability of the plan sponsor or other contributing entity to make contributions to the plan when due. III-2

13 Section 3: Actuarial Methods Actuarial Cost Method. We concur with the decision to retain the Entry Age Actuarial Cost Method for use in IPERS. Later sections of this report will provide comments on the retained actuary s implementation of the method. Actuarial Value of Assets. In our judgement, the asset method meets the requirements of ASOP 44. We are pleased to see that our recommendation to include a market value corridor as a component of the method has been implemented. Amortization of UAL. We understand that while updating the funding policy, the Board reviewed the amortization method and moved to the layered base approach. The approach involves amortizing the June 30, 2014 liabilities over 30 years as a level percent of payroll and future experience gains and losses over separate closed 20-year periods. This method applies to the determination of the Actuarial Contribution rate ACR. The actual contribution is set equal to the ACR, within certain limits defined by statute. In particular, the actual contribution cannot increase more than 1% of payroll per year for Regular members. We think that the ACR as specified is a reasonable method and that, if followed, it will ultimately result in a 100% funded system. We have some concern over the statutory provisions that limit the year to year increase in contribution rates. We were pleased to note that the retained actuary has changed the technical implementation of the amortization method to reflect growth in unfunded liability between the valuation date and the beginning of the contribution rate period as recommended in our prior review. We think that some minor technical issues remain with the method they are using, which we have outlined elsewhere in this report. Section 4: Economic Assumptions Inflation. We find that the retained actuary provides a good discussion of the issues and data surrounding development of an inflation assumption and ultimately concur with the recommendation to lower the assumed rate of price inflation. However, we think that the retained actuary s reasonable range (2.5% to 4.0%) is too wide. To us, it is difficult to justify 4% price inflation as being a reasonable assumption. An inflation assumption that high would likely imply a return to the high inflation rates of the late 1970 s and early 1980 s for an extended period. III-3

14 When inflation runs at such levels, it has been difficult to achieve the real rates of return over inflation that pension funds typically assume, and wage increases have driven liabilities significantly upward. We also think that the data that the retained actuary supplied would have justified lowering the assumption more, however, the retained actuary s philosophy on page 9: Don t Overreact is a wise one and we do agree that the 3% figure is reasonable. Rate of Interest Crediting on Member Contribution Balances. This is a very minor assumption that many actuaries would not include in an experience study. However, since it is there, we comment that the discussion is unnecessarily brief. We suggest that it should include some comment about the spread between one-year CD rates and inflation, and a little history. It is possible that a more complete discussion would yield a different result, but the impact on the valuation results would be minimal in any case. Investment Return. Essentially, the retained actuary developed a forward-looking expectation based upon the actuary s view of inflation and Wilshire s capital market expectations. We concur that the end result of 7.5% is reasonable, given the data that the retained actuary had, but we have some suggestions regarding the analysis. First of all, the entire analysis is heavily based upon Wilshire s views. While Wilshire is the plan s consultant, and Wilshire s views are certainly important, there is considerable diversity of opinion regarding capital market assumptions among the universe of investment consultants. If a different firm served as the IPERS consultant, the retained actuary might have arrived at different results given the same portfolio. We suggest that the retained actuary at least present the reader with an understanding of that diversity, or a statement that the retained actuary believes that Wilshire s assumptions are mainstream and representative of a majority of consulting firms before arriving at a recommendation regarding an assumed rate of return. In presenting an analysis such as that found in the bottom chart on page 22, we would like to see a discussion regarding the time horizon implicit in Wilshire s long term capital market expectations. Wage Growth. First of all, as a very minor point, but one that we found confusing, we suggest that the term Real Wage Inflation be replaced by the term Real Wage Growth throughout the report. The term real inflation appears to us to be an oxymoron and is inconsistent with the term real as used elsewhere in the report. Having said that, we think that the end result of real wage growth of 1% is reasonable long term, and that the analysis the retained actuary provides III-4

15 generally supports it. However, in our personal judgement, we would have been reluctant to raise this assumption from the prior level of 0.75%. Payroll Increase Assumption. When unfunded liabilities are funded as a level percent of payroll, a separate assumption is needed regarding the growth of total payroll. In our judgement, a sound method requires assuming that the active population does not increase for purposes of determining this assumption. If a constant active population is assumed, it is logical to develop this assumption as the sum of price inflation and real wage growth assumptions that the retained actuary discussed previously. This leads to the conclusion that this assumption should be 4% as the retained actuary says. However, a brief review of the data suggests that payroll growth at this level has simply not been happening. If the payroll does not increase at the assumed rate, and unfunded liabilities are amortized as a level percent of payroll, the result can be underfunding even if the employer contributes at exactly the required rate. One way to deal with this matter is to add an overlay assumption that says that for purposes of determining the unfunded liability portion of the contribution rate (in other words the rate of growth of the dollar amount of UAL contribution), payroll cannot be assumed to grow faster than it has in the past 5 or 10 years. In IPERS, for example, that would have led to a 3.4% growth assumption for determining the UAL portion of the ACR in the 2014 valuation. This is somewhat conceptually similar to the idea of a corridor in the asset valuation method. Section 5: Introduction to Demographic Assumption Generally, we found this section difficult to follow. Throughout the body of the report, data is for the most part presented as grand totals with Actual to Expected Ratios, but with very little information, if any at all, by age or service groupings. In order to understand the conclusions and recommendations, we constantly had to flip back and forth to and from the unnumbered, unreferenced pages at the back of the report where detailed information is presented. Presenting the material in this manner leads the lay reader, and perhaps in some instances even an expert reader, to rely much more on the total A/E ratio for the purpose of determining assumptions than is desirable. The end result is that the initial decision regarding the incidence of assumptions at various ages tends to get frozen, and ignored in future experience studies, even if emerging experience indicates differences. What happens at individual age and service groupings can III-5

16 affect valuation results materially and, in our judgement, should be carefully monitored in each experience study. Consequently, we prefer to tabulate the data by age, service, sex, employment type, and any other relevant parameter, even including salary, in the main part of the report. We think this is achievable and would result in a much more readable and possibly more accurate, and perhaps even shorter, report. If the decrement in question is being studied on a liability weighted basis, the tabulations can be done in that manner. The following chart shows the general type of tabulation that we use for this purpose. Following the chart, the actuary can discuss why specific changes were made to assumptions. Withdrawal Experience of Example PERS Members EXPECTED RATES WITHDRAWALS Age Service Withdrawal Exposure Crude Old New Old New 0 4,645 31, ,976 4, ,240 22, ,644 2, ,494 19, ,307 1, , ,535 1, , , &UP , , , , , , TOTALS 13, , ,499 14,383 Actual/Expected Ratio 79% 96% Previous Investigation Results Crude New : : : This type of presentation has several advantages: 1. A reader can judge if the exposure is approximately correct. For a five-year study, for example, each person is exposed once each year, so the exposure at each age should be similar to about 4 or 5 times the number of people in the current valuation. III-6

17 2. The crude rates in the chart on the previous page are simply the ratio of the actual number of withdrawals to the exposure. By viewing the crude rates, the present rates, and the proposed new rates, the reader can gain an insight into the actuary s smoothing techniques and into the judgments that were made. 3. The presentation allows verification that the expected figures in the experience study are derived from the assumption rates being used in the valuation. 4. It is easy to see if the proposed new rates match the final assumptions, and the assumptions used in the valuation reports. 5. It provides a link to prior studies and allows broad comparisons to be made with earlier results. Section 6: Mortality We are pleased to see that the recommendation that we made in the prior audit to use separate mortality tables for State vs. school retirees has been implemented. Mortality Experience: Regular Membership, Healthy Retirees. Given the same data, it is likely that we would have arrived at different results. The actual to expected ratios based on the recommended tables are 0.97 and 0.94, respectively. If we assume that the data is perfect and represents underlying trends, this would indicate that the proposed tables are already overstating expected deaths by 3% and 6%, respectively. In addition, the retained actuary is continuing the use of Projection Scale AA, which according to the Society of Actuaries 2012 white paper (Mortality Improvement Scale BB Report) is understating the rate of future mortality improvement. At a minimum, we would have used scale BB, and we would have adjusted the base rates so that the actual to expected ratios are closer to unity. Having said these things, though, we must also say that we do concur that the retained actuary s end result is within reasonable bounds, given the size of the group being studied. The retained actuary notes that at the time the experience study was done, the Society of Actuaries was close to releasing a complete new mortality table and projection scale. That has now happened (October 2014), and we encourage a careful review of the new RP 2014 mortality table and MP 2014 projection scale in the next experience study. III-7

18 Beneficiaries. We agree with the retained actuary s decision to use the table for retirees of the same gender for estimating beneficiary mortality. Disabled Members. We think the data would have supported a modest lowering of the mortality rates for disabled individuals. Left to our own devices, we probably would have averaged the two A/E ratios and tried to develop a table with ratios on the order of 90%. Active Members. This is a very minor part of the mortality assumption. However, we continue to believe that the following statement on page 31 of the study is incorrect: For active members, it is more conservative to have an A/E Ratio less than 100% because active member death benefits are generally less costly than retirement benefits. We think a correct statement would be: For active members, it is more conservative to have an A/E Ratio greater than 100% because active member death benefits are generally less costly than retirement benefits. It is easy to see this. If, for example, we assume that all active members die prior to retirement (so that the A/E ratio would be a number close to 0), costs would obviously go down because the cheaper death benefit would always be paid. If, on the other hand, we assume no active members die prior to retirement (so that the A/E ratio would be a number much larger than 100%), costs would obviously go up, because in all cases, the more expensive retirement benefit would be paid. It is therefore more conservative to have a small number of assumed active member deaths, which leads to a high A/E ratio, i.e. over 100%, the opposite of the retained actuary s statement. Unfortunately, therefore, we believe the retained actuary s judgement not to change the preretirement mortality table was based upon an incorrect premise. We would have reduced the assumed death-in-service rates, at least for females, in order to increase the A/E ratio toward III-8

19 100%. We think keeping the present active member mortality table may result in a very small understatement of IPERS liabilities. This is most likely an exceedingly minor issue. Special Service Classes. For Sheriffs, Deputies, and Protection Occupation, we would expect to use a post disability mortality table that reflects, at most, a minor impairment in life expectancy because typically, disability is granted for less significant health issues in these cases than for regular members. We did not see any discussion of this matter in the study. Section 7: Retirement Regular Membership. The reasoning in this section makes sense to us, and we probably would have arrived at a similar result in total. However, looking at results by age, for example on Exhibit F-9, we note that the shape of the current assumption and the shape of the proposed assumption are materially different. We think that the shape of the curve at individual ages should be considered. In the case of Exhibit F-9, we would have adjusted the retirement assumptions at individual ages, potentially ending up with an A/E ratio that is, in total, similar to that produced by the retained actuary, but with an effect on the actuarial valuation results that is not the same. We think that it would be a good addition to this section to include a statement saying what the proposed change in rates is, if any. We had to flip back to the unnumbered pages at the end of the report to figure that out. Regarding the retirement age for inactive vested members, we did not understand the basis for the retained actuary s recommendation. In the discussion on the bottom of page 38, the retained actuary points out that the average age at retirement for inactive vested members is 60.5, and says without providing a reason, that the age 62 assumption should be maintained. There is a statement saying that the average age at retirement has increased recently (from what?), but we do not see the connection between that statement and the decision to maintain the assumed age at 62. Of course, this is a minor issue. Section 8: Disability The retained actuary presents data showing that the expected disability rates are too high. We think some credibility should be assigned to the data, and that the rates should be lowered somewhat. Because of the volatility in this assumption that the retained actuary mentions, as III-9

20 well as the difficulties of measurement, we would probably not lower the rates by the amount the data indicates. We might split the difference, as the retained actuary suggests in their philosophy on page 9 ( Don t overreact ). Section 9: Termination of Employment In this section, we are pleased to see that termination information is presented by both duration and sex. We would still like to see the exposure and the assumed and experience termination rates displayed, as we discussed above. Generally, the expected rates (on a liability weighted basis) are well above the actual rates, and in most circumstances we would move in that direction. Since the retained actuary expects termination rates to increase in the future (upon which we present no opinion), we concur with maintaining the present rates. However, we would have constructed rates, at least initially, that differed between males and females, since the experience shown on the detail pages appears to be different. For example, the average weighted turnover rate for school females is 20% higher than for males, but the assumptions do not recognize that difference. (Data summaries H-6 and H-9). Section 10: Refund/Probability of Electing a Vested Benefit Although experience from the prior period is not shown, the retained actuary says that it is consistent with the experience in this period. We would have increased the probabilities to move the expected close to the actual. Most likely an A/E ratio less than 100% on a weighted basis will produce a small understatement of liabilities. Section 11: Merit Salary Scale We concur with the retained actuary s method of isolating the portion of the pay increase attributable to general wage growth and with adjustments for 27 vs 26 pays. We found the presentation of data in the data summaries J-1 etc. difficult to follow because the pages presented raw data that was not adjusted for the difference between assumed general wage growth and estimated actual general wage growth (2.0%) during the experience period. We suspect that if we saw data summaries adjusted in that manner, we would have made incremental changes to proposed pay increases at certain service years. III-10

21 One caution we present is that when a service based salary scale is used, a person s assumed future pattern of pay increases in valuation software is sometimes affected if he or she purchases prior service credit. The valuation software needs to have the ability to use a separate service field for the pay increase assumption in order to measure liabilities properly when the plan permits service purchases. However, this minor issue is almost always ignored. III-11

22 SECTION IV R E V I E W O F A N N U A L VA LUAT I O N R E P O RT

23 ACTUARIAL REPORT CONTENT Generally, we find the report to be complete, and readable. Further, the report satisfies actuarial standards, and the retained actuary s work product is a quality product. In this section, comments are provided on certain items in the report where we believe the presentation could be improved. These comments are judgment matters, and while we disagree slightly with the retained actuary on certain items, this disagreement does not detract from our overall opinion that the work is sound. Our comments are provided by reference to section and/or page number in the retained actuary s report. Retained Actuary s Cover Letter The retained actuary includes the following sentence in the cover letter: The Investment Board has the final decision regarding the appropriateness of the assumptions and adopted them as indicated in Appendix C. While we agree that the Investment Board has the final decision regarding adopting the actuarial assumptions (as is the case with most public employee retirement systems), this does not alleviate the retained actuary s responsibility as to determining whether the assumptions are reasonable or not. According to ASOP No. 41, Actuarial Communications, for any assumption that is not prescribed by law (e.g., private plan valuation discount rates and mortality assumptions), if the actuary is silent regarding the reasonableness of the assumption, then the actuary is indicating that the assumption is reasonable for actuarial valuation purposes. We recommend that the retained actuary modify this sentence to indicate the Board s and retained actuary s responsibilities regarding the actuarial assumptions. Retained Actuary s Section I: Executive Summary Generally, we find the executive summary to be a little long to be compatible with the usual idea of an Executive Summary. Perhaps it could be retitled. IV-1

24 Introduction. We think the introduction is well written, and really presents the main items that will be of interest to most people. We would like to see the funding horizon specified, rather than the amortization period. In cases where there is a time lag between the valuation date and the contribution beginning date, we have encountered confusion regarding when the 30-year period will be up. In this example would it be 2044, or 2045? Experience from Last Year. We would like to see this section begin with a discussion of how the population developed during the year as is currently shown in the Appendix on page A-4. Experience from Last Year (Assets). This section of the report presents asset experience on both a market and funding value basis. We suggest that it could be worthwhile to show market rates of return by year, perhaps accompanied by a 5 and 10-year trailing averages. This might help give the reader a sense of the reasonableness of the 7.5% assumption. Experience from Last Year (Liabilities). The presentation of liabilities on page 4 gives the reader a good description of the development of the unfunded liabilities this year, and of the reasons that the results changed from the prior year. Since the experience study is done on a liability weighted basis, we think it would be particularly useful to present a gain/loss analysis by source here. An example chart is shown below. Gain(/loss) in $ Millions % of Beginning Liabilities Risk Area FY 2014 FY 2013 Investment Return % 2.1% Pay Increases % 1.0% Normal Retirement (234.0) (0.8)% (0.9)% Early Retirement (67.0) (0.2)% (0.4)% Vesting % 0.2% Death-in-Service % 0.0% Disability % 0.0% Other Terminations % 0.0% Total Actuarial Gain % 2.0% A chart of this type gives the user an idea regarding whether or not the liability weighted assumptions are minimizing actuarial losses, and if not, which decrement or decrements might be the problem. In addition, it shows the user what the actuarial gain or loss actually was in total. We could not find the amount of the actuarial gain expressed in clear fashion anywhere in the IV-2

25 executive summary. We think it is $519 Million based on the chart on page 4. (This figure is actually confirmed in subsequent sections of the retained actuary s report.) Contribution Rate. The contribution rate section provides the reader with a good understanding of the contribution rates and their history. In this section it would be useful to present an estimate of the contribution rates in dollars. We find that Board members often inquire about the dollar effect of rates. Summary. We are pleased to see figures in this section based upon both market value of assets and actuarial value of assets. The chart on page 11 gives a good feel for what actually happened in the valuation. We would have put this chart on the first page of the executive summary. We would also suggest including the ratio of both assets and liabilities to payroll in the actuarial valuation report. Retained Actuary s Section II: System Assets We find this section to be well presented and we have no comments. Retained Actuary s Section III: System Liabilities We find this section to be well presented and we have no comments. Retained Actuary s Section IV: System Contributions We are unable to find the actual reported payroll anywhere in the retained actuary s report. By summing the salary data in the retained actuary s groomed data file, we find that the payroll reported in the valuation was $6,988,249,727. Upon review of the data supplied by IPERS, the payroll is actual pay paid during the year, which the retained actuary has annualized for people who were not paid in each of the 4 quarters of FY Thus, if the population is stable and pay increases happen according to assumption, the payroll for FY 2015 should be 4% higher than the $6,988.2 million or approximately $7,267.7 million dollars. But the three payroll figures in Exhibit 10 sum to $7,439.7 Million. We cannot explain this figure. Because of that, the UAL number that we would project in Exhibit 10 is a higher number than the number derived by the retained actuary. IV-3

26 Now, moving to Exhibits 11, 12, and 13, we would get a higher required dollar contribution for the UAL. The payroll that we would project for fiscal 2016 would be lower than what the retained actuary projects, so that our UAL contribution rate would be yet higher. The following chart summarizes the results that we would get. Item Formula Regular Sheriffs Protectives 1 FY 2015 Rate 14.88% 19.76% 16.90% 2 Normal Cost 10.23% 16.50% 16.04% 3 FY 2015 UAL Rate % 3.26% 0.86% 4 UAL 6/30/2014 $ 5,515,853,614 $ 29,124,989 $ (950,600) 5 Annualized FY 2014 Pay $ 6,575,693,979 $ 96,734,749 $ 315,820,999 6 Payroll Growth Factor 104% 104% 104% 7 Expected Pay FY x6 $ 6,838,721,738 $ 100,604,139 $ 328,453,839 8 Projected UAL Payment 3x7 $ 318,000,561 $ 3,279,695 $ 2,824,703 9 Interest Rate 7.50% 7.50% 7.50% 10 Half-year Interest Sqrt(1+9) % 3.68% 3.68% 11 Interest on UAL 9x4 $ 413,689,021 $ 2,184,374 $ (71,295) 12 Interest on payment 10x8 $ 11,709,438 $ 120,765 $ 104, UAL 6/30/ $ 5,599,832,636 $ 27,908,903 $ (3,950,609) 14 Amortization Factor Retained Actuary Amortization Payment 13/14 $ 300,274,484.5 $ 1,496,532.5 $ (211,839.8) 16 Expected Pay FY x7 $ 7,112,270,608 $ 104,628,305 $ 341,591, UAL Amortization Rate GRS 15/ % 1.43% -0.06% 18 UAL Amortization Rate in Valuation Report 4.12% 1.40% -0.06% This is obviously a very minor issue in the present circumstance; however, we believe the method we have illustrated is more mathematically correct than the method that the retained actuary uses. In some circumstances, the difference between the two methods can be much more meaningful than they appear to be in the illustration. For example, the differences will become more important when the amortization years are shorter than they are now. Retained Actuary s Section V: Plan Accounting Information This information is in line with what we would expect. The retained actuary has retained information that formerly was produced for GASB 25. We understand that there is a separate report covering the new accounting rules. A review of that report is included later in this section. IV-4

27 Retained Actuary s Appendix A: Summary Statistics on System Membership This section contains numerous summary schedules and charts that tabulate the current population in many different ways. We would like to see at least one of them tabulate the FY 2014 pay data that IPERS provided (after annualization). We are pleased to see that the retained actuary included the historical row schedules that we requested in connection with the prior audit. Retained Actuary s Appendix B: Summary of Plan Provisions We did not find a description of the plan provision for crediting interest to member accounts for purposes of refunds in this section. We also do not find a description of how deferred vested benefits are calculated if the members account balance exceeds the value of the defined benefit. Finally, we would like this section to include the assumption for determining actuarial equivalence in the case of option factors, service purchases, etc. Retained Actuary s Appendix C: Actuarial Assumptions and Methods In accordance with ASOP No. 41, Actuarial Communications, and ASOP No. 4, Measuring Pension Obligations and Determining Pension Plan Costs or Contributions, we recommend that the retained actuary include the statutory reference that gives the Investment Board the authority to adopt the actuarial assumptions for valuation purposes. We would like to see this section include some sample reserve values. Additionally, we would like to see the following outlined somewhere in this section: 1. What is assumed about the pays reported for the valuation? Are they actual earnings or earning rates? Is the pay for new hires annualized in any way? When are pay increases assumed to occur? 2. When are decrements assumed to occur during the year? 3. What is the basis for option factors? Are the factors actuarially equivalent? If not what loads or other methods are used in the valuation to adjust for the non-equivalence? 4. At what point in time is benefit eligibility tested? At what point in time is benefit service calculated? 5. When are contributions assumed to be received? IV-5

28 Retained Actuary s Appendix D: Contribution Rate Funding Policy In this section we would like to see the full policy for determination of the ACR spelled out, including the funding horizon, the 20-year funding of gains and losses, etc. We would like to see the policy modified to say explicitly what the amortization period would be if liabilities change due to benefit changes, early retirement windows, etc. REVIEW OF THE GASB STATEMENT NO. 67 REPORT We reviewed the June 30, 2014 GASB Statement No. 67 report for consistencies with the accounting standard as well as generally accepted actuarial practice. In general, we find the GASB No. 67 report to be clear, concise, and consistent with the accounting standard. The report does a very good job of disclosing which elements of GASB No. 67 are included in the report and which are excluded including references to specific paragraph numbers of the statement. One of the biggest potential changes required by GASB No. 67 is the use of a single discount rate that may be lower than the assumed rate of return. The report contains an explanation of the process for determining the single equivalent discount rate including a description of how the long term rate of return may be used until the Fiduciary Net Position (FNP) is expected to deplete and a municipal bond rate is used thereafter. The report includes a statement that Our calculations indicate that the FNP is not projected to be depleted, so the bond rate is not used Based on our understanding of the Contribution Rate Funding Policy and our review of the simulation model, we believe that this is an accurate statement. However, in our experience system auditors generally have requested a demonstration of the projection to be included in the GASB No. 67 report. GASB No. 67 requires recognition of ad hoc COLAs in the Total Pension Liability to the extent that they are substantively automatic. Consistent with the funding valuation, the Total Pension Liability reflects an assumed 3% COLA and no Favorable Experience Dividend (FED). Future IV-6

29 transfers to the FED may occur once the plan is fully funded. We expect that it will be several years before the System as a whole is fully funded and it is not entirely clear whether the FED transfer statutes and administrative rules make future transfers to the FED substantively automatic once the plan does reach 100% funding. The assets reported as Fiduciary Net Position appear to include the assets in the Qualified Benefit Arrangement (QBA). While the amount in the QBA is extremely negligible ($378 out of $28 billion), it is not clear whether the QBA assets meet the definition of Paragraph 13 of GASB No. 67 which requires that all assets may legally be used to pay benefits of any of the plan members. Our understanding of the QBA is that those assets are strictly for paying benefits that exceed IRS limits under IRC Section 415(b) and that the QBA is maintained as a separate trust agreement. Ultimately, this is a legal and/or accounting question. There is the potential that if the QBA is treated as a separate plan, it will require a separate single equivalent discount rate that may be required to be the municipal bond rate. Recommendations We recommend that the actuary include a projection of the FNP to illustrate that the assumed rate of return may be used as the single equivalent discount rate. We recommend that the actuary disclose the assumption that there will be no future transfers to the FED and consequently no ad hoc COLAs are reflected in the TPL. In the event that future excess benefits occur, we recommend that the System research the treatment of the QBA in the reporting of the System s FNP for GASB No. 67 purposes. IV-7

30 SECTION V R E V I E W O F S TAT U T E S A N D B O A R D R EGULAT I O N S

31 REVIEW OF STATUTES AND BOARD REGULATIONS As part of our analysis, relevant sections of Chapter 97B of the Iowa Code, IPERS Purpose and Principles, and IPERS Funding Policy were reviewed as they apply to the Iowa Public Employees Retirement System actuarial valuation. This information was provided to us by IPERS Staff. Our review of the above sources was focused primarily on determining the accuracy of the main benefit provisions described in the retained actuary s Appendix B, Summary of Plan Provisions, on pages B-3 through B-9 of the Actuarial Valuation Report as of June 30, In addition, the documents were reviewed to ensure that all material benefit provisions were being disclosed in the actuarial valuation report. Summary of Review Based on our review, GRS believes that the significant majority of all material issues in the law and regulations are reflected in the actuarial valuation. In particular, presented below are some observations resulting from our review: We reviewed Chapters 1A, 4, 7, 8A, 9, 9A, 10, 11, 14, 15, 42 53, 65, 66, 68, 70, 80, 80C, and 82 of Chapter 97B of the Iowa Code. The description of the cost-of-living dividend in the Summary of Plan Provisions complies with 97B.49F(1). The statute describing Option 1 allows the retiree to change the beneficiary for any designated lump sum death benefit, but not the amount. We understand that the actuary assumes that the lump sum death benefit will be paid regardless of the original beneficiary s age in the determination of actuarial equivalence. We concur with this approach. The description of Option 4 (and therefore also indirectly Option 6) describes contingent benefits of 100%, 75%, 50%, or 25% of the retired member s benefit. The specific percentages listed are not stated in Chapter 97B.51 1.d.(1) nor are they in Chapter 11 of IPERS Administrative Code (495). However, the state percentages agree with those described in the member handbook and are described as the available percentages for QDROs under IPERS Administrative Code (2)k. V-1

32 Chapter 97B.1A 10A.a.(2) describes a limitation of pay spiking in the five-year average compensation that is not included in the valuation report. We did not attempt to measure the impact of this provision, but we do not anticipate that it has a material effect. The Favorable Experience Dividend (FED) reserve has been fully paid out as of June 30, However, Chapter 97B.49F 2.c. provides that future reserve transfers into the FED reserve may occur in future years if the System has favorable experience, is fully funded and is expected to remain fully funded after the transfer. IPERS Administrative Code Chapter (1) indicates that the System shall determine by rule the allocation of the System s favorable actuarial experience under these circumstances, but does not specify what that rule is. Consequently, even though the FED is depleted and the System is not fully funded, future favorable experience transfers appear to be allowable if not required under current statutes and administrative code. Chapter 97B.49H refers to active member supplemental accounts. There is no discussion of supplemental accounts in the actuarial valuation. Under Chapter 97B.49H 3., no transfers into active members supplemental accounts may occur unless the plan is 100% funded. However, if the plan is 100% funded, transfers may occur up to the difference between the combined employer and employee statutory contribution rate and the normal cost, if positive. The plan s Funding Policy delays reduction in the statutory employer and employee contribution rates after the plan reaches 100% funding. This may result in an unintended consequence of contributions to members supplemental accounts outside the objectives stated in the Funded Policy. Chapter 97D.5 2.a. requires the disclosure of the actuarially required contribution rate with a 30-year amortization of the unfunded actuarial accrued liability. The actuarial valuation report indicates that, beginning with the June 30, 2014 valuation, the amortization period is a closed 30-year period for the Actuarial Contribution Rate (ACR) and that future gains and losses will be amortized in separate bases over 20 years. Thus, the ACR in the 2014 report meets the requirements of this Chapter. Future valuations will likely result in an ACR with effective amortization periods other than 30 years and will therefore require a separate disclosure with a 30-year amortization. The Contribution Rate Funding Policy includes guidelines that are well articulated and principles that are widely used for public pension plan funding. The Policy requires automatic increases in the Required Contribution Rate to the ACR (limited to a 1% V-2

33 increase for Regular members, as required by statute) if the previous Required Contribution Rate is less than the ACR. If the Required Contribution Rate for a membership group is less than the ACR, the rate may be decreased, but not automatically. The rate may only be decreased in increments of 0.50% of payroll and only if the plan is 95% funded. Moreover, the rate may not be reduced for amortization of a negative unfunded actuarial accrued liability unless the plan has been at least 110% funded for three consecutive years. The asymmetry of the rate increases and decreases is consistent with Guideline 2 of a disciplined funding approach. However, there may come a time when prohibiting a rate decrease may be perceived as conflicting with Guidelines 3 and 4, interperiod equity and supporting an affordable, sustainable plan. Moreover, prohibiting rate decreases when the plan is well funded may result in unforeseen consequences such as contributions to members supplemental accounts (as mentioned above) and required transfers to the FED reserve. Many public employee retirement systems adopted written funding policies in conjunction with the new GASB No. 67 and 68 accounting standards. The Conference of Consulting Actuaries (CCA) published a white paper on Actuarial Funding Policies and Practices for Public Pension Plans in October of Since the publication of this White Paper was after the June 30, 2014 valuation date, we did not review the Contribution Rate Funding Policy as it relates to the CCA white paper. One observation we make from the CCA white paper and an emerging trend in general is that negative amortization of the unfunded actuarial accrued liability is undesirable. Negative amortization occurs when the unfunded actuarial accrued liability is expected to increase in nominal (not real) dollars if all actuarial assumptions are met. We expect this will be the case with IPERS 30-year amortization of the current unfunded and 20-year amortization of gains and losses in the short run, but perhaps less so as the outstanding period for the current unfunded decreases. The Contribution Rate Funding Policy provides transition to shorter closed amortization periods such as those outlined in the CCA white paper. V-3

34 Recommendations, Suggestions and Observations We recommend that the Actuarial Assumptions Section in Appendix C of the actuarial valuation report be expanded to include a description of the treatment of pay spiking in the three-year average and five-year average compensation in the valuation. We recommend that the Actuarial Assumptions and Methods in Appendix C of the actuarial valuation report be expanded to include a description of the assumed form of payment elected, and that no future contributions to member supplemental accounts or the FED reserve are assumed. Similarly, we recommend the report disclose that the assumed future FED dividend is zero. V-4

35 SECTION VI T E S T O F VA LUAT I O N R E S U LT S T H R O UGH M AT H E M AT I C A L M O D E L I N G A N D S A MPLES

36 CASE STUDIES To evaluate the reasonableness and/or accuracy of valuation results in accordance with generally recognized and accepted actuarial principles, we requested from the retained actuary valuation results for individual test cases. Valuation results were requested for 20 active test cases (12 Regular Membership, 4 Sheriffs and Deputies and 4 Protection Occupation), 26 retired test cases (19 Regular Membership, 4 Sheriffs and Deputies and 3 Protection Occupation) and 9 inactive test cases (7 Regular Membership, 1 Sheriffs and Deputies and 1 Protection Occupation). Results between GRS and the retained actuary were generally within expected boundaries, especially for the retired test cases and active test cases with higher amounts of service credit. A small number of active test cases had larger differences than anticipated. Different valuation systems will round service and age amounts differently, and that type of rounding probably explains the differences in test cases with small amounts of service. For retiree test cases 17 and 18, we were able to match the retained actuary s results if we used a beneficiary gender of male. We recommend that the retained actuary review their valuation process for these types of cases. Page VI-2 shows a summary of the results for the individual test cases by valuation group. Based upon the results shown on this page, we have validated the accuracy of the valuation results to a reasonable degree as presented in the June 30, 2014 actuarial valuation report. VI-1

37 SUMMARY OF INDIVIDUAL TEST CASE RESULTS A. ACTIVE MEMBERS CavMac GRS % Diff CavMac GRS % Diff CavMac GRS % Diff Regular Membership $ 2,344,455 $ 2,276, % $ 1,766,848 $ 1,727, % $ 87,020 $ 85, % Sheriffs and Deputies 678, , % 237, ,663 (1.9)% 32,896 33,719 (2.4)% Protection Occupation 1,141,084 1,120, % 643, ,579 (0.5)% 56,719 56, % Total Active Test Cases $ 4,163,767 $ 4,055, % $ 2,646,972 $ 2,615, % $ 176,635 $ 174, % B. RETIRED MEMBERS C. INACTIVE NON-RETIRED MEMBERS PVFB EAAL EANC PVFB CavMac GRS % Diff Regular Membership $ 3,624,936 $ 3,709,815 (2.3)% Sheriffs and Deputies 1,354,489 1,363,074 (0.6)% Protection Occupation 707, ,428 (1.4)% Total Retiree Test Cases $ 5,687,080 $ 5,790,317 (1.8)% PVFB CavMac GRS % Diff Regular Membership $ 258,345 $ 250, % Sheriffs and Deputies 53,168 50, % Protection Occupation 276, ,308 (8.4)% Total Inctive Test Cases $ 587,596 $ 603,040 (2.6)% VI-2

38 ACTIVE TEST CASES REGULAR MEMBERSHIP Annualized Valuation Age Service Salary Gender Result CavMac GRS % Diff Test Case $ 52,790 F PVFB $ 307,906 $ 308, % EAAL 292, ,700 (0.3)% EANC 4,515 4, % PVFS 161, ,065 (0.2)% Test Case $ 98,094 M PVFB $ 327,526 $ 319, % EAAL 194, ,432 (1.3)% EANC 11,037 11, % PVFS 1,189,923 1,131, % Test Case $ 13,834 F PVFB $ 30,260 $ 28, % EAAL 12,536 12,571 (0.3)% EANC 1,449 1, % PVFS 169, , % Test Case $153,072 M PVFB $ 854,361 $ 852, % EAAL 827, ,221 (0.1)% EANC 10,814 10, % PVFS 319, , % Test Case $ 68,990 M PVFB $ 156,149 $ 152, % EAAL 85,387 79, % EANC 4,734 5,111 (7.4)% PVFS 1,028,290 1,021, % Test Case $ 54,255 F PVFB $ 247,397 $ 242, % EAAL 216, , % EANC 4,550 4,601 (1.1)% PVFS 369, , % VI-3

39 ACTIVE TEST CASES REGULAR MEMBERSHIP (CONTINUED) Annualized Valuation Age Service Salary Gender Result CavMac GRS % Diff Test Case $ 28,045 M PVFB $ 43,095 $ 37, % EAAL 12,805 12, % EANC 4,276 3, % PVFS 203, , % Test Case $269,770 F PVFB $ 230,712 $ 201, % EAAL 75,683 41, % EANC 34,371 33, % PVFS 1,099,019 1,186,155 (7.3)% Test Case $ 5,788 F PVFB $ 12,362 $ 12,405 (0.3)% EAAL 5,897 6,544 (9.9)% EANC (0.7)% PVFS 52,150 48, % Test Case $ 49,959 M PVFB $ 80,868 $ 75, % EAAL 15,807 15, % EANC 5,053 4, % PVFS 650, , % Test Case $ 2,664 M PVFB $ 1,765 $ 2,046 (13.7)% EAAL NA EANC (10.1)% PVFS 24,946 29,878 (16.5)% Test Case $ 33,730 F PVFB $ 52,054 $ 44, % EAAL 27,031 24, % EANC 5,344 4, % PVFS 157, , % VI-4

40 ACTIVE TEST CASES SHERIFFS AND DEPUTIES Annualized Valuation Age Service Salary Gender Result CavMac GRS % Diff Test Case $ 4,993 M PVFB $ 10,640 $ 10,700 (0.6)% EAAL 7,748 8,520 (9.1)% EANC 1,344 1, % PVFS 9,345 8, % Test Case $ 65,811 F PVFB $ 263,909 $ 271,806 (2.9)% EAAL 154, ,083 (3.3)% EANC 9,555 10,467 (8.7)% PVFS 749, , % Test Case $ 43,068 M PVFB $ 120,689 $ 109, % EAAL NA EANC 6,784 6,920 (2.0)% PVFS 839, , % Test Case $ 78,660 F PVFB $ 282,990 $ 265, % EAAL 74,566 72, % EANC 15,213 15, % PVFS 1,094,451 1,047, % VI-5

41 ACTIVE TEST CASES PROTECTION OCCUPATION Annualized Valuation Age Service Salary Gender Result CavMac GRS % Diff Test Case $113,796 M PVFB $ 435,599 $ 452,383 (3.7)% EAAL 221, ,667 (16.0)% EANC 18,870 18, % PVFS 1,291,890 1,238, % Test Case $ 79,464 F PVFB $ 457,098 $ 430, % EAAL 394, , % EANC 16,519 16, % PVFS 288, ,991 (2.3)% Test Case $ 45,656 F PVFB $ 100,662 $ 106,269 (5.3)% EAAL 12, NA EANC 13,463 14,288 (5.8)% PVFS 332, ,960 (15.1)% Test Case $ 45,421 M PVFB $ 147,725 $ 132, % EAAL 14,496 13, % EANC 7,867 7, % PVFS 850, , % VI-6

42 RETIRED TEST CASES REGULAR MEMBERSHIP Member Member Beneficiary Beneficiary Option Elected Valuation Age Gender Age Gender at Retirement Result CavMac GRS % Diff Test Case NA NA M Active Member Death Benefit PVFB $ 606,586 $ 620,145 (2.2)% Test Case 2 Option 5: 120 Month Term Certain F NA NA Annuity PVFB 289, ,060 (0.2)% Test Case 3 Option 2: Annuity with Variable F NA NA Decreasing Lump Sum PVFB 367, ,274 (0.2)% Test Case 4 Option 2: Annuity with Variable F NA NA Decreasing Lump Sum PVFB 142, ,458 (0.7)% Test Case 5 Option 2: Annuity with Variable F NA NA Decreasing Lump Sum PVFB 12,897 12, % Test Case 6 Option 1: Annuity with Fixed M NA NA Lump-Sum Benefit PVFB 20,950 20, % Test Case 7 Option 1: Annuity with Fixed F NA NA Lump-Sum Benefit PVFB 600, ,683 (3.3)% Test Case 8 Option 1: Annuity with Fixed M NA NA Lump-Sum Benefit PVFB 81,130 81, % Test Case F NA NA Option 3: Single Life Annuity PVFB 388, ,088 (0.5)% Test Case 10 Option 4: 100% Joint & Survivor M F Annuity PVFB 170, ,030 (2.9)% VI-7

43 RETIRED TEST CASES REGULAR MEMBERSHIP (CONTINUED) Member Member Beneficiary Beneficiary Option Elected Valuation Age Gender Age Gender at Retirement Result CavMac GRS % Diff Test Case 11 Option 4: 25% Joint & Survivor NA NA F Annuity PVFB 34,833 35,133 (0.9)% Test Case 12 Option 4: 75% Joint & Survivor M NA NA Annuity PVFB 231, ,577 (5.6)% Test Case 13 Option 6: 25% Joint & Survivor NA NA F Annuity with Pop-Up PVFB 58,682 60,797 (3.5)% Test Case 14 Option 6: 50% Joint & Survivor F M Annuity with Pop-Up PVFB 425, ,812 (0.4)% Test Case 15 Option 6: 75% Joint & survivor M NA NA Annuity with Pop-Up PVFB 153, ,076 (0.9)% Test Case 16 Old Option 1: Annuity with F NA NA Variable Dec Lump Sum PVFB 8,199 8,300 (1.2)% Test Case 17 Option C: Pre Ret Death NA NA F Month Certain Annuity PVFB 16,255 20,265 (19.8)% Test Case 18 Option B: Pre Retirement Death NA NA F Single Life Annuity PVFB 15,219 21,893 (30.5)% Test Case 19 Option 5: 120 Month Term Certain NA NA M Annuity PVFB - 11,313 NA For retiree test cases 17 and 18, we were able to match the retained actuary s results if we used a beneficiary gender of male. We recommend that the retained actuary review their valuation process for these types of cases. VI-8

44 RETIRED TEST CASES SHERIFFS AND DEPUTIES Member Member Beneficiary Beneficiary Option Valuation Age Gender Age Gender Elected Result CavMac GRS % Diff Test Case 1 Option 2: Annuity with Variable M NA NA Decreasing Lump Sum PVFB $ 417,777 $ 420,716 (0.7)% Test Case 2 Option 6: 100% Joint & Survivor M F Annuity with Pop-Up PVFB 348, ,573 (0.5)% Test Case 3 Option 7: Level Payment Option M NA NA Annuity PVFB 277, ,926 (0.1)% Test Case 4 Option 7: Level Payment Option M NA NA Annuity PVFB 310, ,859 (1.1)% VI-9

45 RETIRED TEST CASES PROTECTION OCCUPATION Member Member Beneficiary Beneficiary Option Valuation Age Gender Age Gender Elected Result CavMac GRS % Diff Test Case 1 Option 1: Annuity with Fixed M NA NA Lump-Sum Benefit PVFB $ 42,021 $ 42,481 (1.1)% Test Case 2 Option 7: Level Payment Option M NA NA Annuity PVFB 323, ,604 (1.7)% Test Case 3 Option 7: Level Payment Option M F 75% J&S Annuity PVFB 341, ,343 (1.1)% VI-10

46 INACTIVE NON-RETIRED TEST CASES REGULAR MEMBERSHIP Valuation Age Gender Result CavMac GRS % Diff Test Case F PVFB $ 24,848 $ 24,973 (0.5)% Test Case M PVFB 1,536 1,538 (0.1)% Test Case M PVFB % Test Case F PVFB 21,266 21,824 (2.6)% Test Case F PVFB 2,474 2, % Test Case F PVFB 170, , % Test Case F PVFB 37,725 31, % INACTIVE NON-RETIRED TEST CASES SHERIFFS AND DEPUTIES Valuation Age Gender Result CavMac GRS % Diff Test Case M PVFB $ 53,168 $ 50, % INACTIVE NON-RETIRED TEST CASES PROTECTION OCCUPATION Valuation Age Gender Result CavMac GRS % Diff Test Case M PVFB $ 276,083 $ 301,308 (8.4)% VI-11

47 SECTION VII AV E R A G E C OMPENSAT I O N C A S E S T U D IES

48 INTRODUCTION A common source of unexpected liabilities (actuarial loss) is associated with the Final Average Earnings and service credit of new retirees being higher than would be projected from the annual pay and service credit reported for the actuarial valuation. In this section, we analyze sample data for eight people who retired shortly after the June 30, 2014 valuation was completed. Five were regular retirees and three were from either the Sheriffs and Deputies or Protection Occupation Groups. One of the regular retirees was a deferred vested person as of June 30, The remaining regular retirees were active on June 30, We compared data and certain calculations for these people from four sources: 1. The Retained Actuary s groomed data 2. A data file that IPERS provided directly to us 3. Benefit calculation pages that IPERS provided 4. Output of our valuation program using IPERS data as input In performing this analysis, we identified certain issues with respect to both the retained actuary s groomed data, and the data that IPERS supplies. These issues are summarized below. First of all, the Final Average Earnings FAE in IPERS plan provisions is based upon the average of the highest five years of covered wages, but cannot be less than the 3-year average as of June 30, In the long run, the June 30, 2012 floor will not matter, but for the immediate future the floor affects the valuation results. IPERS provided a 3-year FAE in the data that was sent to us. We presumed that was the same data file that was sent to the retained actuary. Upon review, the 3-year FAE appeared to be a current value. Indeed, according to the file layout, it is Average Annual Wages for 12 most recent quarters. A review of the test cases supports that assertion. That particular data item is of diminishing value in performing the valuation. VII-1

49 In most of the cases, the Investment (member contributions) shown in the benefit calculation is slightly different from the amount shown in the data files. This is most likely due to crediting of some additional contributions and/or interest after the valuation date, and is not an issue. With respect to the benefit calculation pages, we notice that the Reserve Values that GRS calculates for females tend to be quite different from those that IPERS calculates, even after adjusting for differences in the calculated benefits. We recommend that IPERS work with the retained actuary to ensure that the reserve factors used are consistent with the retained actuary s assumptions. With respect to the retained actuary s groomed data, the groomed data did not contain at least the 3-year FAE as of June 30, 2014, which had been in the data that IPERS supplied. Since the retained actuary s groomed data contains no version of an FAE, the actuary may be estimating the FAE for valuation purposes directly from the actuarial salary scale (effectively assuming that pay increases of approximately 4% or so happened in the recent past). Such a calculation is likely to lead to an understatement of liabilities in the present economic environment. The thought process here may have been that the current FAE that IPERS supplied is no longer relevant to the calculation, which is true. But, the 2012 FAE could be relevant for those near retirement for the next few years. It may have been possible to match the 2012 data with the 2014 data and to move the 2012 FAE into the 2014 data in place of the FAE field that IPERS supplied. If that had been done, the retained actuary could have included the 2012 FAE in the groomed data and tested the 5-year FAE calculated in the valuation program against the actual 2012 FAE in the data and chosen the greater value, which is in accordance with plan provisions. In our judgement that would have enhanced the accuracy of the valuation. We do not have data to estimate the effect that it would have, but in our judgement, that would be the correct way to do the valuation. If, for some reason, that was impossible, we think that the actuary should then use the 2014 FAE that IPERS supplied, perhaps adjusted slightly downward, as a proxy for the true year FAE. We recommend that the retained actuary augment the groomed data as described above and develop the FAE in its valuation program in accordance with plan provisions. VII-2

50 In reviewing the cases, we also noticed that the retained actuary does not store the birth month. The birth month could matter for certain special studies, but generally not for a valuation. We think this is a very minor issue. Finally, we note that the retained actuary s groomed data matches the IPERS data with respect to all of the items that we studied, other than FAE. We ran our proprietary software on the sample data, assuming that the FAE shown in the data was the June, 2012 FAE (which of course it is not) just to assist the reader in gaining an understanding of the potential effect of applying a floor to the FAE calculation. This allows the user of this report to see how the calculations would have been done if we had access to the proper FAE in the data file. The next pages present data on each of the sample cases and comment on the surrounding issues. GRS shows a different number of quarters worked than is reported in the data, because GRS assumes that retirement occurs two quarters after the valuation date. VII-3

51 Data Source Case 1-Reg Retained Actuary IPERS Benefit Calc GRS Valuation Comment Status AV AV Retired 1/1/2015 AV OK Date of Birth /7/1947 8/7/1947 8/7/1947 OK Gender F F F F OK Quarters Worked OK Investment $16, $16, $16, $16, OK Current Earnings $8, $8, NA $8, OK Final Average Earnings $7, $14, $7, Exception Formula Benefit $ $ $ Exception Reserve $37, $73, $42, Exception In this case, we notice that the final average earnings in the benefit calculation are much higher than the final average earnings in the IPERS data. In order to perform the valuation, the GRS software would project current earnings ($8,012) backward into the past assuming that prior pay increases had occurred in accordance with the salary scale. If for example, the actuarial salary scale provided for an average increase of 4.0% to 4.5%, the 5-year FAE of a person who earned $8,012 in the last year of employment would be approximately 94% of $8,012 or $7,531. Since this value is less than the FAE supplied by IPERS, GRS software used $7,716 as the FAE. Of course, based upon the benefit calculation, the GRS software should have used $14, as the FAE, but there was no way for that to happen. Given the data supplied, and the fact that the retained actuary s report does not indicate anything to the contrary, we presume the retained actuary would have used a figure on the order of $7,531 for the final average earnings in this instance. In this case, both GRS and the retained actuary would have understated the benefit due to lack of data, although by a lesser amount for GRS. VII-4

52 Data Source Case 2-Reg Retained Actuary IPERS Benefit Calc GRS Valuation Comment Status IV IV Retired 12/1/2014 IV OK Date of Birth /31/1949 8/31/1949 8/31/1949 OK Gender F F F F OK Quarters Worked OK Investment $17, $17, $17, $17, OK Current Earnings $0.00 $0.00 NA $0.00 OK Final Average Earnings $27, $27, $27, $27, OK Formula Benefit $ $ $ $ OK Reserve $65, $65, $69, OK This is a deferred vested case. In this instance all of the data agrees. However, we still notice that our reserve value is higher than the value that IPERS calculates. Data Source Case 3-Reg Retained Actuary IPERS Benefit Calc GRS Valuation Comment Status AV AV Retired 10/1/2014 AV OK Date of Birth /1/ /1/ /1/1952 OK Gender M M M M OK Quarters Worked OK Investment $44, $44, $44, $44, OK Current Earnings $65, $65, NA $65, OK Final Average Earnings $57, $53, $61, Exception Formula Benefit $2, $1, $2, Exception Reserve $252, $236, $283, Exception In this case, GRS software estimates the current 5-year FAE as $61,569, based upon current year earnings and the actuarial salary scale. Since this figure is higher than the FAE in the data, the GRS software uses it. The retained actuary would most likely have done about the same thing. Unfortunately, this value overstates the actual final average earnings and therefore the benefit, and therefore the reserve value. It could be that the 2014 earnings of $65,320 were an anomaly. This is why we think that the current 5-year FAE should be shown in the data. In this case, if the reserve value is adjusted for the difference in benefits, GRS agrees approximately with the reserve value in the benefit calculation. VII-5

53 Data Source Case 4-Reg Retained Actuary IPERS Benefit Calc GRS Valuation Comment Status AV AV Retired 10/1/2014 AV OK Date of Birth /30/1948 6/30/1948 6/30/1948 OK Gender M M M M OK Quarters Worked OK Investment $6, $6, $6, $6, OK Current Earnings $7, $7, NA $7, OK Final Average Earnings $10, $16, $10, Exception Formula Benefit $ $ $ Exception Reserve $25, $38, $26, Exception We presume that the $16, Final Average Earnings in the benefit calculation is the 3-year FAE as of June 30, In this case, both the current year earnings and the 3-year average FAE in the data lead to an understatement of liabilities. In this example, both GRS and the retained actuary would understate liabilities, but GRS would have done so by a lesser amount because GRS applied the FAE floor of $10,884 to the calculation. If the reserve values are adjusted for the difference in benefits, the GRS value is very close to the value given in the Benefit Calculation. Data Source Case 5-Reg Retained Actuary IPERS Benefit Calc GRS Valuation Comment Status AV AV Retired 2/1/2015 AV OK Date of Birth /13/1986 9/13/1986 9/13/1986 OK Gender F F F F OK Quarters Worked OK Investment $8, $8, $8, $8, OK Current Earnings $25, $25, NA $25, OK Final Average Earnings $25, $22, $25, Exception Formula Benefit $ $ $ Exception Reserve $57, $50, $59, Exception In this case, the GRS software applied the floor of $25,082 to the FAE calculation (recall that for this illustration, GRS is assuming the FAE reported in the data is actually the June 30, 2012 figure when it is not). This actually leads to an overstatement of the true liabilities because the floor is below that. In this case the retained actuary s software would have produced a more accurate result by ignoring the floor. The GRS reserve value, when adjusted for the difference in benefits, is fairly close to the value in the benefit calculation. This case differs from all other cases because it is a disability case. VII-6

54 Data Source Case 1-SSG Retained Actuary IPERS Benefit Calc GRS Valuation Comment Status AV AV Retired 11/1/2014 AV OK Date of Birth /22/1958 9/22/1958 9/22/1958 OK Gender M M M M OK Quarters Worked OK Investment $181, $181, $184, $181, OK Current Earnings $87, $87, NA $87, OK Final Average Earnings $86, $87, $86, OK Formula Benefit $5, $5, $5, OK Reserve $707, $715, $705, OK This case worked rather well. It appears, though that the 3-year FAE at June 2012 was actually higher that the 3-year FAE in the IPERS data. The retained actuary would most likely project backward with the salary scale and arrive at a yet lesser figure. The GRS reserve value is very close to the value in the benefit calculation after adjustment for the difference in projected benefit. Data Source Case 2-SSG Retained Actuary IPERS Benefit Calc GRS Valuation Comment Status AV AV Retired 1/1/2015 AV OK Date of Birth /10/1959 4/10/1959 4/10/1959 OK Gender M M M M OK Quarters Worked OK Investment $87, $87, $90, $87, OK Current Earnings $75, $75, NA $75, OK Final Average Earnings $73, $73, $73, OK Formula Benefit $3, $3, $3, OK Reserve $487, $499, $498, OK This case is largely unremarkable. But once again, the retained actuary has no real way to calculate the final average earnings, and so would most likely arrive at benefit and reserve values less than shown. VII-7

55 Data Source Case 3-SSG Retained Actuary IPERS Benefit Calc GRS Valuation Comment Status AV AV Retired 12/2/2014 AV OK Date of Birth /3/1953 2/3/1953 2/3/1953 OK Gender M M M M OK Quarters Worked OK Investment $141, $141, $145, $141, OK Current Earnings $74, $74, NA $74, OK Final Average Earnings $73, $75, $73, OK Formula Benefit $4, $4, $4, OK Reserve $549, $568, $546, OK In this case, the actual final average earnings in the benefit calculation exceeds the current earnings. Both GRS and the retained actuary would understate liabilities in this situation. Presumably GRS would understate by a lesser amount because its software applied the $73,109 as the floor. VII-8

56 SECTION VIII S I M U L AT I O N M O D E L

57 ANALYSIS OF THE SIMULATION MODEL IPERS provided screen shots of the valuation simulation model under different scenarios for each group. The valuation simulation model allows the user to model numerous scenarios regarding different discount rates, future asset returns, contribution levels, and asset valuation methods to obtain projected valuation results. The simulation model does not appear to allow for modeling the FED. The results shown in the model include but are not limited to, funded ratios, actuarially determined contributions, assets and liabilities. In general, with regard to the scenarios we tested, the results of the tool were consistent with our expectations. Furthermore, we found the output to be presented in an effective manner. The baseline scenario for each group showed a substantial gain in the first year of the projection. This appeared to be caused by the actuarial value of assets being reset to the market value one year after the valuation date. The model has a feature to reset the actuarial value to market value, but there is no indication on the output whether or not this feature has been selected. This is clearly visible in the second year of the projected Funded Ratio for the Regular Membership in the baseline scenario shown below: The total payroll show in the simulation model does not agree with the payroll disclosed in the valuation report for valuation year Moreover, the projected payroll appears to grow at a different rate in the baseline scenario than the payroll growth assumption disclosed in the valuation. To the extent that the assumptions in the simulation model and valuation differ, distortions will occur at later years in the projections. If there is a rationale behind using VIII-1

58 different assumptions, we suggest that the rationale be disclosed on the output pages of the simulation model. The output shown below from the baseline scenario of the Regular Membership shows total payroll growing from 7,000 million to 19,025 million over 30 years for a compound average growth rate of 3.39% versus the valuation s assumed 4.00% payroll growth rate. Short term fluctuations in payroll growth are not uncommon in projections, but the long run growth should be close to assumed. During the last 10 years of the projection, the payroll is expected to increase from 13,325 million to 19,025 million for a compound growth rate of 3.63%. The last year of the projection, payroll is projected to grow from 18,327 to 19,025, an increase of 3.81%. We recommend that the retained actuary verify that the payroll growth rate becomes 4% soon after the end of the 30-year projection period. The simulation model does not appear to show an implied amortization period for a given Required Contribution Rate. The model does effectively illustrate how the current method for determining the Required Contribution Rate is expected to result in a funded ratio in excess of 100% for each group in the not-too-distant future (19 years for Regular, 1 year for Sheriffs and Deputies, and 0 years for Protection Occupations). This means the System as a whole is projected to exceed 100% funded in 19 years. IPERS Administrative Code (97B) states that IPERS shall in no event credit amounts attributable to favorable actuarial experience to the FED account, unless IPERS is fully funded and will remain fully funded after such VIII-2