University of Connecticut, Janet & Mark L. Goldenson Center for Actuarial Research Annual Advisory Board Meeting, September 20, 2013 University of Connecticut Department of Mathematics College of Liberal Arts and Sciences Table of Content Children’s Museum Enterprise Risk Management Analysis Long-Term Care Predictive Modeling Simulation Study of Modified Credibility Weighting for Group LongTerm Disability Termination Rates Affordable Healthcare Act National Retirement Satisfaction Index 2 Children’s Museum ERM Analysis Children’s Museum ERM Analysis 4 AAMIR ADDONNA KATHLEEN SNAJDER SUZANNE XIE ADVISOR: JAY VADIVELOO UNIVERSITY OF CONNECTICUT Purpose of Report 5 Provide Museum with an outside perspective on their risks and help determine the steps they can take to handle the risks. Risks deemed most pressing are addressed. Help Museum realize their strategic plan, not change it. Types of Risks 6 Legal Employee litigation, risk of harm while on premises Operational Staff/personnel, loss of buildings and contents, IT security Strategic Reputation, failure to stay competitive in market Financial Support from donations and government, insufficient financial controls Suggestions for Legal Risks 7 Inevitable part of running a business Should be discussed with Museum’s legal department or lawyers One suggestion: reevaluate current insurance coverage for cost effectiveness and completeness Suggestions for Operational Risks 8 Need to constantly improve to “keep it fresh” Gauging and improving reputation Administer quick survey or have a suggestion box Offer incentives for visitors to provide feedback Attracting and maintaining quality staff Host staff appreciation events Provide some sort of incentive compensation plan Attract new talent by upholding reputation (discussed later) Suggestions for Strategic Risks 9 Need to stabilize revenues and manage reputation Offer additional programs to counteract seasonality Attendance plummets when school year begins Local partnership Create partnership with some local businesses, such as zoos, to offer combined discounts Mutually beneficial scholarship Host a competition where high school students propose new exhibit ideas for Museum Suggestions for Financial Risks 10 Need to build reserves to absorb revenue fluctuations In 2011-2012, 50% or more of each month’s revenue was obtained from outside support. Gifts, grants and donations comprise the largest percentage of revenue Build endowment over the next 3-5 years Endowment of $30 million will generate a perpetuity of $122,000 each month at an annual interest rate of 5% Amount covers current development efforts (on average each month) Museum can only focus on other endeavors when it becomes financially independent from donations Game Plan to Build Endowment 11 Create 3-5 year plan Sell Museum to corporations and the community, like in 1995 Show Museum’s impact on the community Distinguish from competitors Relationships are crucial Develop endowment campaign in partnership with local schools and government Full visibility given to donors whose endowment donation is geared towards specific programs Embrace technology Market smart and efficiently; promote “lifelong learning” Conclusion 12 Conflicting challenges Remain affordable to community while remaining fresh and dynamic Plan needs to be long-term, since short-term suggestions will not solve main problems Should gain financial independence Achieved through high level of commitment, planning, and support from all constituents Long-Term Care Predictive Modeling Long-Term Care Predictive Modeling Project Project Advisors: Jay Vadiveloo, Brian Hartman Research Team: Peter Camacho, Wanqing Ma, and Gao Niu 14 Initial Goals and Data Description Our Client came to us seeking a better understanding of the factors that drive their future economic risk. Mortality, Disability, and Lapse decrements for their Individual and Group long-term care (LTC) active policies (A total of 6 Models) Historical claims data on in-force policies (2000 – 2012). 32 unique independent variables; 3 decrements (dependent variables) Over 10M records were submitted Less than 17% of the raw data were removed for missing and inconsistent entries. Altogether, 8 predictive models developed Poisson GLM Modeling (Industry standard). Modeling done in SAS and TW proprietary software Emblem. 15 Basic Model Modelling based on a Poisson GLM with log monthly exposure as offset Predictor variables developed based on client input and use of Stepwise Logistic Regression techniques Monthly Rate of Claims = (Monthly Claims /Monthly Exposure) Monthly Rate of Claims≈EXP(β0+β1*X1+...+βn * Xn ) Client Call Internal Meeting Modeling Work 16 EVOLUTION OF RESEARCH Additional Variables. Special interest in Alzheimer’s related Claims. Creation of conditional Alzheimer’s model gave the client a powerful new way to anticipate these high liability claims. Challenging to model due to low frequency of Alzheimer's incidence relative to high total months of exposure. As such, we needed an alternative to classic modeling procedures 17 CONDITIONAL MODEL Conditional Monthly Incidence of Alzheimer’s Of all those who went on claim, what proportion when on claim due to Alzheimer’s? Model 3: Overall Monthly Rate of Alzheimer’s Claims Monthly Rate of Alzheimer’s Related Claims ≈ (Monthly Claims )* (Monthly Conditional Rate of Alzheimer’s) Model 1: Original Model Model 2: Conditional Model 18 SECTION II: RESULTS 19 CONTEXT EXAMPLE : GROUP LONG TERM CARE MORTALITY MODEL Note: Have beta estimates to 2 decimal places Model Interpretation E[Monthly Mortality Rate] = EXPONETIAL(β0 + β1* X1 + β2 * X2 + β3 * X3) X1= AttAge X2=1 if Gender=Female, -1 if Gender=Male X3= Duration Example: For a 40 years old female whose contract is in duration 10, X1 = 40, X2 = 1, X3 = 10 E[Monthly Mortality Rate per 1000] = 0.01 20 POWER OF OUR MODELS 21 22 23 24 25 26 27 DATA TRANSFER AND FINAL DELIVERABLES Written report of findings and PowerPoint presentation to client SAS codes for all 8 predictive models 8 Predictive models 3x ILTC ( Mortality, Lapse, Incidence) 3x GLTC ( Mortality, Lapse, Incidence) 2x Conditional Alzheimer's Incidence (ILCT, GLTC) Rate Calculator Single user interface collection of all of all 8 models. Manual variable input. Audit check 28 RATE CALCULATOR : DEMO Note: [Here I think we should rap things up with Demo of the rate calculator that Gao was able to put together. This allows us to leave the audience with a tangible final product of the work we did]** 29 Simulation Study of Modified Credibility Weighting for Group Long-Term Disability Termination Rates Simulation Study of Modified Credibility Weighting for Group Long-Term Disability Termination Rates 31 HUITZE RUAN SUZANNE XIE ADVISOR: JAY VADIVELOO UNIVERSITY OF CONNECTICUT Abstract 32 Work Group’s model Limited Fluctuation Credibility(LFC) Theory Selected Variance Factor (A) Our model Positive and negative dependence of termination rates Monte Carlo simulation process Results Reduction in credibility weights is significantly smaller than the reduction in credibility weights proposed by Work Group LFC Model 33 Suggested formula for blended termination rates: Z ´ q +(1- Z)´ q A E qA: Company’s actual termination rates with 85% margin qE: Experience terminations rates based on 2012 GLTD Valuation Tables Z: Credibility weighting factor using the LFC model LFC Model 34 Independence Model Use confidence level of 85% and error tolerance of 5% 2 K= æ 1.44 ö ç ÷ è 0.05 ø Z= min( N ,100%) K K=number of expected terminations needed for full credibility N=number of expected terminations for the same period using company’s actual exposure Dependence Model 35 Work group suggests the use of a Selected Variance Factor (A), an arbitrarily defined number, to reflect the dependence characteristic Our model suggests the use of a Selected Variance Factor (A*) is defined as: æ std(q̂ D ) ö A* = ç I ÷ è std(q̂ ) ø 2 Where std(q̂ I )and std (qˆ D ) are standard deviation of actual termination rates based on the independence and dependence model, respectively Dependence Model 36 Work Group’s model æ 1.44 ö ç ÷ ´A è 0.05 ø 2 K= Z= min( N ,100%) K Our model æ 1.44 ö æ std(q̂ D ) ö K*= ç ÷ ´ç I ÷ è 0.05 ø è std(q̂ ) ø 2 Z= min( 2 N ,100%) * K Our Selected Variance Factor (A*) std(q̂ D ) 2 ) =( I std(q̂ ) Methodology (Independent Model) 37 Monte Carlo simulation Use 1000 lives and 1000 simulations per set of 1000 lives Use termination rates from a 2008 LTD table for a male aged 42 with an elimination period (EP) of 90 days Methodology(Independent Model) 38 Choose two stage termination rates by different durations as suggested in the Guideline Group Months after EP 1 <=3 2 >3 and <=24 3 >=24 and <=60 4 >=60 and <=120 5 >120 For each simulation, capture initial exposure counts at the beginning of stage 2 (B) and actual claim terminations in stage 2 (A) Average claim termination rate in stage 2 is A/B Methodology (Dependent Model) 39 Same methodology as independent model for first stage q1A=(actual terminations in stage 1)/1000 The new expected termination rate in stage 2 is adjusted under two extreme situations: Positive relationship: q2*E = (q1A/ q1E)*( q2E) Negative relationship:q2*E = (q1E/ q1A)*( q2E) Calculate A/B, the average claim termination rate in stage 2 Methodology (Cont.) 40 Repeat simulation process 1000 times. Calculate mean (µ) and standard deviation (σ) of the actual termination rates in stage 2 (A/B) Take ratio of both models: Ratio1 = µdependence/µindependence ≈ 1 std(q̂ D ) Ratio2 = σdependence/σindependence = std(q̂ I ) Methodology (Cont.) 41 Test termination rates at beginning, middle, and end of each duration group Adjust q2*E to account for less direct dependency between termination rates in later durations Group 1(<4 months) : Z = 1; disregarded Group 2 (4 to 24 months) : no adjustment to q2*E Group 3 (25 to 60 months) : 75% of original q2*E Group 4 (61-120 months) : 50% of original q2*E Group 5 (> 120 months) : 25% of original q2*E Results (Cont.) 42 Positive Dependency Relationship Duration Group (Months) Selected Variance Factors (A) Selected Variance Factors (A*) Guideline Simulation 4 to 24 4.0 1.8 25 to 60 3.0 1.2 61 to120 2.5 1 > 120 2.0 1 Floor of 1 set for A* Results 43 Positive Dependency Relationship Duration Group (Months) 100% Credibility Value assuming independence 100% Credibility Value (K) Guideline Selected Values for 100% Credibility (K*) Simulation 4 to 24 830 3,300 1,500 25 to 60 830 2,500 1,000 61 to 120 830 2,100 830 > 120 830 1,700 830 Approximations made to raw calculated values to determine K* Results (Cont.) 44 Negative Dependency Relationship Duration Group (Months) Selected Variance Factors (A) Selected Variance Factors (A*) Guideline Simulation 4 to 24 4.0 1.8 25 to 60 3.0 1.6 61 to120 2.5 1.2 > 120 2.0 1 Floor of 1 set for A* Results (Cont.) 45 Negative Dependency Relationship Duration Group (Months) 100% Credibility Value assuming Indepedence 100% Credibility Value (K) Guideline Selected Values for 100% Credibility (K*) Simulation 4 to 24 830 3,300 1,500 25 to 60 830 2,500 1,300 61 to 120 830 2,100 1000 > 120 830 1,700 830 Approximations made to raw calculated values to determine K* Consolidated Table 46 • By taking an average of the 2 extreme dependency relationships, we obtain the following results: Duration Group (Months) 100% Credibility Value (K) Guideline Selected 100% Variance Credibility Factors (A) Factors (K*) Guideline Simulation Selected Variance Factors (A*) Simulation 4 to 24 3,300 4.0 1,500 1.8 25 to 60 2,500 3.0 1,150 1.4 61 to 120 2,100 2.5 915 1.1 > 120 1,700 2.0 830 1 Conclusion 47 Shown two extreme dependency relationships Positive dependence Higher than expected terminations in stage 1 imply higher than expected terminations in stage Negative dependence Lower than expected terminations in stage 1 imply lower than expected terminations in stage 2 Conclusion 48 The adjusted credibility factors to recognized dependency are slightly different for both dependence relationships The consolidated dependency factors can be used to adjust the credibility weighted factors for GLTD terminations For either type of dependency, reduction in credibility weights is significantly smaller than the reduction in credibility weights proposed by the Actuarial Guideline Affordable Healthcare Act Pocketbook guide to the Affordable Care Act (“ACA”) An ongoing collaborative project with UCONN Goldenson Center for Actuarial Research • Jay Vadiveloo • Pete Camacho • Nehal Sapre • Mark Spong American Institute for Economic Research (AIER) • Stephen Adams • Jules Clark • Natalia Smirnova Understanding ACA Creating a Plan Long Term Projections • Legal requirements • Demographic context • Political climate • Compare apples to apples • Identify key groups • Project costs • Determine implications 50 What is the AIER/Goldenson Center Study? First-level analysis aimed at helping individuals anticipate the personal financial impact of the ACA next year and in five years • Narrow – direct costs to individuals – premiums and out-of-pocket expenses • Does not include: • • • • • Tax impact (state and federal fiscal impact) Potential changes in employer-based health insurance Potential changes in the delivery system that affect costs Follow-on work will address these important issues 51 Why the AIER/Goldenson Center Study? ACA is arguably the most significant public personal finance policy act since Social Security was established in 1935 • Also a deeply politicized issue • • The public is getting misleading and conflicting information from all directions 52 Why: People Confused About ACA 53 Why: Few Trusted Sources 54 Why Goldenson Center & AIER? • The mission of the Goldenson Center for Actuarial Research is to provide high-quality applied actuarial research services to serve the needs of the insurance and financial services industry in the region • The mission of the American Institute for Economic Research is to provide factual, unbiased research and critical analysis of the economic issues affecting the everyday life of people in America 55 Summary of the ACA • Individual and employer mandate • Creation of state-based health benefit exchanges - Four benefit categories of plans along with a catastrophic plan - Provision of premium credits and cost sharing subsidies • Expansion of Medicaid to 133% of Federal Poverty Line • Guaranteed availability of coverage; i.e., restrictions on excluding individuals with pre-existing conditions • Single risk pool for individuals and small groups per state 56 Change of Coverage Landscape Health insurance coverage of the total population preACA employer individual 16% 1% 13% 16% medicaid 49% 5% medicare other public ? uninsured 57 Apples to Apples Comparison Issues • Typical annual increase in healthcare premiums regardless of ACA • Level of coverage discrepancies • What does being uncovered pre-ACA actually mean? 58 Cost Drives Premiums • Initial premiums will be available in October 2013 • Estimate longer term premiums by estimating costs and assuming prior loss ratios are a reasonable proxy • Included in long term cost estimates • Wellness Programs • Morbidity estimates • Cost controlling by insurers • Pent-up demand 59 Next Steps Decomposition and estimation of healthcare costs over the next five years considering parameters such as severity, frequency, changes in morbidity, pent up demand etc. Calculate total premiums collected using historic loss ratio and the estimated costs • Compare expected costs for a large group with the expected costs of a “pooled” individual to see the effect of the pooling principle • Survey of insurance carriers and large employers 60 National Retirement Satisfaction Index National Retirement Satisfaction Index 2013 Review A presentation to the 2013 Goldenson Center Board Meeting by Gary Rohrig and Gao Niu September 20, 2013 Agenda Project Overview Project Background Index Definition Conceptual Model Financial Projections Non-Financial Improvements Analysis Scenarios Next Steps Acknowledgements 63 Project Overview Problem Philosophy Due to limitations in the current retirement preparedness indices, there is a need to reliably account for satisfied living in retirement In order to give a full and positive picture of someone's level of satisfied living after leaving the work force, we will research, develop and implement a national index to accurately account for both financial and non-financial drivers related to retirement Scope & Value Determine partnerships – Idea was initiated by two members of the Advisory Board – Current project is supported by the Goldenson Center, UConn’s Roper Center and Towers Watson – UConn project team consists of five faculty members and a number of graduate students under the direction of Jay Vadiveloo Develop a National Retirement Satisfaction Index (NRSI) which will be updated annually The findings of the NRSI will be publicly available and actively marketed 64 Project Background Existing national indices National Retirement Risk Index (Boston College) International Retirement Security Survey (AARP) Retirement Confidence Survey (Employee Benefit Research Institute) Others focused on specific areas Across Generations Retirement Income Survey (New York Life) Retirement Preparedness Survey (Merrill Lynch) Fidelity Retirement Index (Fidelity) 65 Project Background Comparison with other Indices Other Indices NRSI Do not rely on national U.S. data Relies on U.S. Census data and other nationally recognized data sources Capture only purely economic data Captures both economic and noneconomic factors Infrequent updates Will be updated annually Use arbitrary constraints that put people in an “at risk” category Takes a more holistic view of retirement preparedness and uses a financial metric that evaluates retirement satisfaction on a scale of 0 to 100 66 Index Definition Benchmark Retirement Satisfaction Equity (RSE): A measure of retirement satisfaction. The actuarial PV of equity in financial and non-financial factors at retirement RSE = Accumulated Assets + Social Security + Potential Income – [Health & Living Expenses] For working population, RSE is measured at an anticipated retirement age For the current retiree population, RSE is measured as of today Overall RSE is a weighted combination of both RSE measures RSE is an extension of the accounting principal of Equity = Assets – Liabilities NRSI calibrated to range between 0 and 100 When RSE is negative, NRSI is set to 0 Maximum NRSI of 100 calculates RSE assuming individuals do not retire and continue to work indefinitely – Ex: If Maximum RSE = $100,000 and Projected RSE = $50,000, then NRSI = 50 67 Index Definition Financial Factors Assets from savings, social welfare, investments Liabilities from living, out-of-pocket health costs and other expenses – Our liabilities will represent the minimum needs for survival in retirement Non-Financial Factors Health Status: As health status declines due to aging and disease incidence, an increase in out-of-pocket medical expenses can result, increasing liabilities – We have measured improvements in health status that cause a decrease in health liabilities Adaptability: Captures post retirement part-time income that could be earned by all retirees, thus increasing future part-time retirement income – For the working population, the effect of improvement in educational levels also increases the potential for part-time income Financial Planning: Applied only to the working population, improvements in the level of financial planning helps to increase asset growth rates and thus increase future assets for retirement Job Satisfaction: Applied only to the working population, the delaying of retirement is measured in terms of national job ranks and captures the impact of accumulating more assets before retirement and decreasing lifetime liabilities after retirement 68 Financial Projections • Data sources Economic Factors Retirees Workers Accumulated Assets HRS U.S. Census Projected Income & Expense U.S. Census U.S. Census Social Security Income HRS HRS Interest Rates U.S. Federal Reserve US Federal Reserve Non-Economic Factors Retirees Workers Future Health Improvements HRS HRS Adaptability HRS HRS Financial Planning N/A HRS Job Satisfaction N/A U..S Census, U.S. Bureau of Labor Statistics, Wall Street Journal, HRS 69 Actuarial Projections • Working population Assets at retirement accumulated by interest only • • Future liabilities at retirement (living expenses and health care expenses) discounted by interest and survivorship Future social security benefits and potential post-retirement part-time income discounted by interest and survivorship • • • Retiree population All components of RSE discounted by interest and survivorship • • Projected up to retirement age 63 Inflation rates used in projections for wages, living expenses, and health expenses Actuarial Present Value of Expenditure for Retirees Total Actuarial Present Value of Expenditures for Workers $2,500,000 $1,500,000 $2,000,000 $1,000,000 $1,500,000 $500,000 $1,000,000 $500,000 $0 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84 $0 Total Maximum Expenditures Total Actual Expenditures Total Maximum Expenditures 70 Total Actual Expenditures 70 Health Expense Projections • We use multi-variable regression for annual health expenses HRS Data contains: • Health Score (State of Health) • • Attained Age Out-of-Pocket (“OOP”) Health Expense • • OOP increases by attained age and higher health score • • Best health status: score = 1 Retiree OOP Expenses are modeled using average HRS health scores Annual Out-Of-Pocket Medical Expenses for Health Conditions $10,000 $8,000 $6,000 $4,000 $2,000 $0 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91 93 95 97 99 101 Attained Age Medical Expenses: Health Score 1 Medical Expenses: Health Score 2 Medical Expenses: Health Score 4 Medical Expenses: Health Score 5 Medical Expenses: Health Score 3 71 Health Expense Projections • Annual out-of-pocket expense projection Predict improvements in health score over time • Using a five-year “calendar-year effect” on the 2012 HRS average health score by age Cumulative improvements over multiple HRS waves in health score improvement factor by age • • Project health expenses • Baseline expenses • • Expense by age using average reported HRS health score: 3.027 Improved expenses • • • Expense by age using improved health score Improved health score apply the improvement factor to the overall average HRS reported health score Average Projected Health Score for Workers 2.7 2.7 2.6 2.6 2.5 2.5 2.4 2.4 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 Improved Health Score 72 Health Expense Projections • We accumulate annual expenses to get the total actuarial present value • Current retiree health expenses • • Baseline - using average HRS reported health score Actual - using actual current HRS reported health score Working population health expenses • • • Baseline - using average HRS reported health score with medical inflation Actual - using improved HRS reported health score with medical inflation 73 Adaptability Improvement • We use multi-variable regression for annual part-time income by level of education HRS data contains: • Age of retirement Level of attained education (years) Part-time income earned • • • Higher education level higher part-time income Higher attained age lower part-time income • • Part-Time Income Part-Time Income by Attained Level of Education, All HRS respondents $3,500 40% $3,000 35% $2,500 30% 25% $2,000 20% $1,500 15% $1,000 10% $500 5% $0 0% Less than High School Some High School HS Graduate HRS Population Percentage Some College 4-year degree Graduate Degree Average Part Time Income 74 Adaptability Improvement • Part-time income increases income in retirement Retirees • • • Potential part-time income calculated for each retiree using a regression equation based on attained age and level of education Workers • HRS data selected for regression – people over 55 reporting any part-time income amount (about 1:16) • • Baseline income amount using US census average education level Improved income amount using projected gain from educational improvement Improvements in level of education • • Forecasted using a 10-year “calendar year effect” on US Census data The additional annual part-time income is accumulated • • Inflation and discount factors are applied the same way as in the baseline financial projection Average Annual Part Time Retirement Income $60,000 $50,000 $40,000 $30,000 $20,000 $10,000 $0 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 Average HRS Part-Time Income, positive responses only Projected Part-Time Income using Logarithmic Regression of Average HRS Part-Time Income, positive responses only 75 Financial Planning Improvement • We use HRS reported assets by financial planning level to model gains in projected assets • HRS Data contains: • • • • Attained age Level of financial planning (score of 1 to 5, 5 being the most planning) Total assets Higher level of financial planning higher accumulated assets Average HRS Assets by Financial Planning Score $1,000,000 $900,000 $800,000 $700,000 $600,000 $500,000 $400,000 $300,000 $200,000 $100,000 $Avg. Total Assets in 2000 Avg. Total Assets in 2002 Avg. Total Assets in 2004 1 2 3 4 Avg. Total Assets in 2006 Avg. Total Assets in 2008 5 76 Financial Planning Improvement • Gains from financial planning are forecasted for current workers • HRS data baseline of level of financial planning for retirees • Average current levels of financial planning are determined • Improvements in financial planning are computed • 5-year “calendar year effect” using HRS average financial planning score Accumulated assets are recalculated • • • • Improvement factor = ratio of improved financial planning assets to baseline financial assets The improvement factor is multiplied by the investment return rate of the baseline calculation of financial assets financial-planning adjusted return rate Accumulated assets are recalculated using the financial-planning adjusted return rate Average Total Assets - All HRS Waves - Weighted v. Financial Planning Score $800,000 $700,000 $600,000 $500,000 $400,000 $300,000 $200,000 $100,000 $- 0 1 2 Avg. Total Assets - All Waves - Weighted 3 4 5 6 Linear (Avg. Total Assets - All Waves - Weighted) 77 Job Satisfaction Improvement • We determine the number of years beyond retirement workers will be willing to work Delaying retirement affects RSE • • • • • • More satisfied working environment more years in employment (delay retirement) Use Wall Street Journal job ranks to model level of job satisfaction Top 200 jobs ranked – lowest rank = best job (2013, actuary = 1) Jobs matched to corresponding HRS (Retirees) and US Bureau of Labor Statistics (Workers) job groups Retirement Age & Job Rank Retirement Age • Increase accumulated assets before retirement – more accumulation years Decrease liabilities after retirement – less expenditures in retirement years 64.5 78.0 64.0 77.5 63.5 77.0 Retirement Age 63.0 76.5 62.5 76.0 Job Rank (Population Shifted Method) 62.0 75.5 Retirees Workers 78 Job Satisfaction Improvement • Overall job rank level of job satisfaction age of retirement From HRS data • Age of retirement Occupation group of retiree • • From US BLS and WSJ • Job rankings Occupations and job groups • • Determine the revised retirement age for working population • • 𝐴𝑔𝑒𝐽𝑜𝑏 = 𝐴𝑔𝑒𝑅𝑒𝑡 + 𝐴𝑔𝑒𝑅𝑒𝑡 ∗ 𝑅𝑎𝑛𝑘𝑊𝑜𝑟𝑘𝑖𝑛𝑔 −𝑅𝑎𝑛𝑘𝑅𝑒𝑡𝑖𝑟𝑒𝑒 𝑅𝑎𝑛𝑘𝑅𝑒𝑡𝑖𝑟𝑒𝑒 Job-Satisfied Retirement Age, by Job Ranking 65.5 35% 65.0 Retirement Age 40% 30% 64.5 25% 64.0 20% 63.5 15% 63.0 10% 62.5 5% 62.0 0% 24 35 37 58 63 64 76 78 Job Ranking 81 115 127 131 HRS Population Percentage HRS Average Retirement Age Regressed Retirement Age, by job rank 144 79 Analysis • Preliminary Results Increase to Baseline Factor Retiree Population Working Population Overall Baseline 54 48 49 Health Care 1 6 5 Adaptability 10 9 9 Financial Planning 0 5 3 Job Satisfaction 0 11 8 TOTAL 65 79 74 Note: These values are preliminary and subject to change 80 Next Steps • Put together final report after models are validated by UConn and TW and report is reviewed by select -industry representatives • Models should be set up so that the NRSI can be updated annually • Disseminate and publicize report • Explore with industry leaders how NRSI findings can be used to better design and market retirement products and services • Explore how an Individual Retirement Satisfaction Index that is company specific can be designed to be used in parallel with the NRSI 81 Thank you, your questions are welcome 82 Appendix 83 Scenarios Using the model we generate multiple scenarios • • • Econometric factors changed shift of interest rates Non-Financial factors changed shift of cumulative improvement factors Scenarios create a guide for improvements workers and retirees • • Interventions can be taken to improve one’s satisfaction in retirement Summary of Analysis Results by Scenario By Group Scenario Description Changes Investment Returns : 5.53% Discount Rate : 3.66% Medical Inflation : 5.81% Expense Inflation : 3.06% Wage Growth : 3.12% 1 Baseline US Population, All Ages 2 Mild Economic Stress US Population, All Ages Investment Returns Decrease by .74% Expense Inflation Increase by .13% 3 Moderate Loss of Assets US Population, All Ages Investment Returns Decrease by 1.49% Expense Inflation Increase by .27% Savings Ratio Decrease by 25% 4 Mild Economic & Medical Hardship US Population, All Ages Investment Returns Decrease by .74% Medical Inflation Increase by .28% Expense Inflation Increase by .13% Wage Growth Decrease by .17% Savings Ratio Decrease by 10% 5 Extreme Decrease of Savings US Population, All Ages Savings Ratio Decrease by 50% 6 Moderate Educational Improvement US Population, All Ages Educational Improvement Increase by 4x 7 Moderate Financial Planning Improvement US Population, All Ages Financial Planning Improvement Increase by 2x 8 Moderate Health Improvement US Population, All Ages Health Improvement Increase by 4x 9 Multiple Facet Improvement US Population, All Ages Health Improvement Increase by 4x Financial Planning Improvement Increase by 2x Educational Improvement Increase by 4x 84 Scenarios With economic stressors / non-financial interventions, the NRSI shifts • Summary of Analysis Results by Scenario By Group NRSI Index by Grouping Retirees Scenario Description Base Health Care Adaptability Working Population Financial Planning Job Satisfaction Base Financial Job Health Care Adaptability Planning Satisfaction 1 Baseline US Population, All Ages 54 55 65 65 65 48 54 63 68 79 2 Mild Economic Stress US Population, All Ages 54 55 65 65 65 39 46 58 63 74 3 Moderate Loss of Assets US Population, All Ages 54 55 65 65 65 0 10 25 31 41 4 Mild Economic & Medical Hardship US Population, All Ages 54 55 65 65 65 18 29 44 51 64 5 Extreme Decrease of Savings US Population, All Ages 54 55 65 65 65 7 13 22 27 35 54 55 65 65 65 48 54 64 68 79 54 55 65 65 65 48 54 63 78 90 6 7 Moderate Educational Improvement US Population, All Ages Moderate Financial Planning Improvement US Population, All Ages 8 Moderate Health Improvement US Population, All Ages 54 58 67 67 67 48 71 81 83 94 9 Multiple Facet Improvement US Population, All Ages 54 58 67 67 67 48 71 81 89 100 85 Summary NRSI Summary for Working Population Base Health Care Adaptability Financial Planning Job Satisfaction 90 79 11 5 9 6 48 79 74 11 5 12 7 39 64 41 10 6 15 10 0 13 7 15 11 18 35 8 5 9 6 7 12 Total 94 100 11 2 10 11 8 10 11 4 10 6 15 9 6 23 23 48 48 48 48 86 Summary NRSI Summary for Retirees Base Health Care Adaptability Financial Planning Job Satisfaction Total 65 0 10 1 65 0 10 1 65 0 10 1 65 0 10 1 65 0 10 1 65 0 10 1 65 0 10 1 67 0 9 4 67 0 9 4 54 54 54 54 54 54 54 54 54 87