The phases of retirement planning

Transcription

1 Lessons from Behavioral Research for Retirement System Design Olivia S. Mitchell The Wharton School The phases of retirement planning Accumulation phase: $ in Investment phase: $ grows Payout phase: $ out Rational economic men (and women) Forecast earnings, retirement & death date Compute needed saving deferred consumption Buy insurance When young: for premature death When old: for longevity protection Goal of retirement risk management: Reduce old-age economic insecurity Three-legged stool Private sources (work, saving, insurance, family) Employment-based pensions Government Social Security What do people really do? HRS Longitudinal survey (199-present):, households (age 51+) Demographic, health, wealth, income, transfers, family info (respondents and spouses) Links to admin data: Earnings and Projected Benefits File w/ labor market history, earnings SS wealth Pension Provider File Pension wealth. Medicare files (expenditures). 3 Uneven diversification among older populations (median wealth, on the verge of retirement) Total US $K (US) Net Housing Net Fin. Soc. Sec. Pension 1% % 1% % HRS retirement wealth distribution: % 15% 1% 5% % housing, pension, financial % Social Security NZ $5K (NZ) 9% 1% % % % -5% Ventile US has better diversification versus NZ: less in home and SS, better company pensions US: Mitchell/Moore () All HH age 51- (US$1) NZ: Scobie/Gibson (3) Couples age 5-55 (NZ$1) T15-5 th decile 5-1% Moore/Mitchell Net Housing Social Security Net Financial Pension 1

2 Estimating Saving Shortfalls: Select Retirement Age & Project Assets; Set Retirement Adequacy benchmark; Calculate Replacement and Prescribed Saving Choose Initial Replacement Rate Y ( 1 s) T + T Y RR = P P F YP F = YP Distribution of saving shortfalls: Earnings Decile 1. Nonmarried Earnings Decile Determine Need AF * RR * Y P Prescribed Savings Rate. 1 Determine Shortfall Need - Projected Assets. Solve for Prescribed Saving Rate T * AF [ YC ( 1+ wg) TP + TF] PROJ s = T YC [ Z+ AF( 1+ wg) ] Decile 1 Prescribed Savings Rate. Find New Replacement Rate * YP( 1 s ) TP + TF * RR = YP To retire at & smooth consumption: must save 15-% more (half at 5) Married Decile 1 What predicts saving shortfalls? To smooth consumption in retirement: Median US household in mid-5 s would have to save 1%/yr more Saving shortfalls Larger for higher earners Smaller if have pension (3-% pa) Smaller if own a home (7-% pa) Why don t people do better? Sources of financial information Group Relative Friend Personal Judgment Finan. Prof l Magaz./ News Age % % 3% 9% 1% Age Age Men Women College None Bernheim 1 So this motivates pensions. Why companies provide them? Mandatory (e.g. Chile, Australia) Tax-deferred compensation Attraction and retention: Attracts workers with long horizon Vesting employer contributions: e.g. 3 yrs Backloaded benefits keep workers to retirement age Retirement inducement DB plans often penalize delayed retirement Two main types of pensions: Defined Benefit (DB): Benefit formula promised e.g. B t =%*(Final Pay)*(Years Service) Defined Contribution (DC): Contribution amount specified: e.g. C t =%*Salary 1 13

3 The DC plan has taken over 7,, 5,, 3,, 1, E-1999 Active Participant Total DC Active Participants DB Active Participants K corp plans, 55M participants, ~$3Tr in assets PWBA 55 Bulletin, What s the problem? 1(k) plan participants have great intentions, poor execution Action Planned Change Actual Change Nonparticipants Enroll in 1(k) plan 1% 1% Participants Increase contribution rate % % Change fund selection 7% 15% Unless pension saving is mandatory: Procrastination widespread Inertia problematic Choi et al. 15 What to do? Design for selfcontrol But beware: defaults can cut accumulation Prior approach: opt-in New approach: Automatic enrollment Boosts participation dramatically Few opt out P a rtic ip a tio n R a te o f N e w H ire s 1% % % % % % 37% Before automatic enrollment % After automatic enrollment Automatic enrollment boosts participation: But some who would have enrolled voluntarily & saved at high rate, will now settle for default: Low saving rates (~3%) Conservative investments (GIC/MMF) SO overall saving might not rise What to do? Madrian and Shea 1 Save More Tomorrow (SMarT) Plan 1% 1% % % % % % Boost contribution rate automatically later! (when get a raise) Self-control commitment device Inertia they stay in the plan 11. Preadvice First Increase Second Increase Third Increase SMarT Participants Accepted Advice Declined Advice Investment Issues why fund a pension? Pros Cuts uncertainty re future benefits and contributions Allows diversified portfolio Cons Requires deferred consumption Requires supervision: investments and administration Benartzi / Thaler 19 3

4 Funded NO Individual Accounts NO US, Russia, much of Europe YES Notional accounts : Latvia, Sweden Global Retirement Assets Growing (% of GDP) 1 1 YES Provident Funds: Asia, some in Africa Personal Accounts: Chile, Australia Colombia Peru Argentina Korea India France Germany Phillipines Greece Sri Lanka Denmark Sweden Finland Japan Canada Chile Ireland Malaysia Singapore Australia US UK Netherlands Switzerland Geanakoplos, Mitchell, Zeldes (1999) Bateman/Kingston/Piggott 1999 and Iglesias/Palacios Risks differ by plan type Of course in a DC plan Defined benefit: Employer promises future benefits & invests to meet promise (firm bears capital mkt risk at least at first!). Rational agents should Balance expected returns and risk. Hold diversified portfolios: 3 stocks. Be independent of framing and co-workers. Defined contribution: Only contribution, not benefit, assured (employee bears capital mkt risk). Evidence? Investors see patterns in small sequences of randomly drawn numbers (Tversky/Kahneman). Workers hold lots of company stock and think it s safe! (Mitchell/Utkus) Workers have weak portfolio preferences: 79% select median portfolio over their own; pick the funds at the top of the list (Benartzi/Thaler) Investors plagued by choice overload (Iyengar et al.) Overlook fees and costs Exposure to company stock Few attend to pension expenses: What s the gold standard? Participants (Millions) million with access, 11 million concentrated (>%) 5 million highly conc. (>%) % 1-% 1-% 1-% 1-% Over % Allocation to Co Stock Account Basis Pts (1) Fund Size ($B, 5) CREF Stock $1.1 CREF Global Equity. CREF Growth 1.9 CREF Equity Index 3.9 CREF Social Choice CREF Bond Market CREF Infl-Linked Bond CREF Money Market 3.9 TIAA Real Estate 5 7. Source: Holden and VanDerhei (1b), VanDerhei (b); authors calculations 5

5 Administrative Costs Erode Pension Buildup Assuming ROR 1% % Saving decreased by: 1 basis point expense 33 % 5 % 5 basis point expense 15 % 1 % (Assumes 5% wage growth from age 1 to 5.) Payout Phase Issues To protect against longevity risk, people should annuitize But investors often take the lump sum! DOD: 11K officers, 55K enlisted offered separation package of annuity vs. lump sum IRR on annuity: 1-% (!) Extensive education program (and some tax advantage for annuity) % taking lump sum Forecast Actual Officers None 51% Enlisted personnel Half % Saved govt (cost soldiers) $1.7B Schieber Are annuities a good investment? compute EPDV of $1 premium EPDV = y (1 λ* τ )* A* Pj + j Πk=1 (1+(1 τ )*ik ) 1*T j=1 j=1*t +1 (1 τ )* A*P j j Πk=1 (1+(1 τ )*ik ) Pop vs annuitant mortality tables Taxes: inclusion ratio λ (principal untaxed at federal/state tax rate τ until attain life expectancy) Alternative discount rates (corporate yield curve usually) Adverse Selection in Annuities (US males). Those wanting annuities in private market likely to live. LONGER asymmetric information. Insurers price accordingly How important in annuity mkts?? Population Annuitants Age Asymmetric information: Annuity buyers live longer! 9 Annuities attractive: EPV of $1 premium (men age 5) So why don t people annuitize? Pop Ann UK.9.97 US.1.93 Australia Italy NA.9 Bequest motives Liquidity preferences Nursing Home costs loom large Framing effects - lump sum seems substantial Overconfidence? In investment projections In estimating mortality risk Mitchell (; Treasury Yield Curve) 3 5

6 What about Self-Annuitization? Goal: Finance periodic income stream for oldage consumption until (uncertain) date of death Starting Point: Retiree of a certain age/sex Retirement wealth at that point Compare alternative phased withdrawal patterns to benchmark real lifetime annuity Product Alternatives Life annuity: insurance companies Phased withdrawal plan: banks, mutual funds Why of interest? Importance: 1 st pillar DB pensions in decline, more DC plans Retirees responsible for decumulation phase UK, Germany require mandatory annuitization (75/5) Issues: Uncertainty about capital markets Uncertainty about investment horizon Multiple goals: Benefits, shortfalls, and bequests Incorporate asset allocation and withdrawal rules 33 3 Phased Withdrawal Plan Advantages compared to Life Annuity High flexibility, liquidity Bequest potential Higher benefits Risks of Phased Withdrawal Plans Lower benefits than Life Annuity Consumption Shortfall Longevity risk (No risk pooling) Betting on Death Capital market risk Betting on Capital Markets Our Approach Use risk value models: Incorporate both Prob and Size of loss Compare fixed with variable withdrawal rules Optimize asset allocation & parameters of variable payment schedule Study mixed portfolios of withdrawal plans and annuities Compare phased withdrawal paths to life annuity benchmark 35 Dus, Maurer,Mitchell 3 Findings: Phased withdrawal strategy Exposes one to risk of outliving assets; Appeals due to flexibility, bequests, & possibly higher consumption than annuity; Requires attention to asset allocation and withdrawal rules. To minimize the risk of falling below real annuity benchmark, must invest more in fixed income than equities. For a fixed withdrawal rule compared to no annuity: Mandatory deferred annuitization can enhance expected payouts & cut expected shortfall risk But at cost of reduced bequests. For a variable withdrawal plan, later annuitization may not reduce risk: It requires optimization of benefit to wealth ratio. Conclusions: People are imperfect computers! Confused by many choices Are overconfident. Lack self-control. Prone to inertia & procrastination. Do what others do. Are influenced by framing and default effects. 37

7 Implications So mortals could use help to do that which they know they should Save more Invest better Protect against outliving assets Designing retirement programs requires recognizing imperfections in people s concentration & processing abilities. Financial education useful but insufficient. Thank you - questions? For more information: The Pension Research Council: Books and working papers: Olivia S. Mitchell & Stephen P. Utkus, eds. Money Matters: Shaping Retirement Decisonmaking. Pension Research Council. Oxford, UK: Oxford University Press,. Executive Education:Pension Strategy Course September 5 Olivia S. Mitchell 7