Research. Evaluation of Retirement Strategies. 1. Retirement Strategies Key variables Key questions

Transcription

1 Evaluation of Retirement Strategies 1. Retirement Strategies Key variables Key questions 2. Evaluation of Retirement Strategies Approaches proposed Pros, cons, and evidence 3. Concluding Thoughts Research 1

2 The Basics Exogenous (given) variables Size of the retirement portfolio Usually a round amount easy to scale up or down Endogenous (decision) variables Asset allocation Usually, but not exclusively, stocks and bonds Withdrawal rate The initial withdrawal rate plays a key role Main uncertainties Length of the retirement period Portfolio returns * Both have exogenous and endogenous components Asset Allocation (AA) Basic trade-off Aggressive AA High growth, high standard of living, high uncertainty Conservative AA Low growth, low standard of living, low uncertainty Key questions Should the AA be static or dynamic? If static, aggressive or conservative? * A lot could be said in favor of very aggressive AAs If dynamic, predetermined or tactical? If predetermined, DEG or REG? If tactical, depending on what variables? 2

3 Withdrawal Rate Basic trade-off: Spending... too much and running out of money early too little and lowering the standard of living (And leaving an unintended bequest) Key questions Should periodic withdrawals be fixed or variable? If fixed, amount or percentage? (Why not percentage) If amount, nominal or real? (Why not nominal) If real, at what level? * Quirk: Initial WR + Inflation-adjusted withdrawals If variable, depending on what variables? Withdrawal Rate The beginning Bengen (JFP, 1994) The 4% Rule: An initial withdrawal rate (IWR) of 4%, with subsequent annual withdrawals adjusted by inflation, never depleted a portfolio before 30 years Lower IWRs unnecessarily lower a retiree s standard of living (At 3% portfolios last over 50 years) Higher IWRs are risky (At 5% portfolios often fail to last at least 30 years) Therefore, 4% is a safe IWR Tricky issues: Results depend on... country considered (Bengen: USA) asset allocation considered (Bengen: 50-50) assets considered (Bengen: Stocks and bonds) the definition of an acceptable failure rate 3

4 Evaluation Key question What is an optimal AA and WR? This is where most of my research comes in * This is also an issue during the accumulation period Proposals Failure rate Bequest Success-to-variability ratio Maximum withdrawal rate Shortfall years (S Y ) Risk adjusted success (RAS) Downside risk adjusted success (D RAS) Utility based approach The Failure Rate The basics Definition Fraction of retirement periods considered in which a strategy failed (Frequency) Calculation Based on historical or simulated retirement periods (Usually 30-year retirement periods) Motivation Proxy for the probability of failure Shortcomings Two strategies may have the same failure rate but may have left very different bequests Measures how often a strategy failed but not by how much it failed 4

5 The Failure Rate Evidence 4% IWR The Failure Rate Evidence

6 The Failure Rate Shortcoming 1 Two strategies may have the same failure rate but may have left very different bequests The Failure Rate Shortcoming 2 Measures how often a strategy failed but not by how much it failed

7 Shortfall Years (S Y ) Measures the average number of years a strategy fell short, across all the periods in which it failed Risk Adjusted Success (RAS) (100 retirement periods / No bequest) Evaluation S1 vs S3 S1 (Same F but lower S Y ) S1 vs S2 S1 (Same S Y but lower F) * Note that neither variable is useful in isolation But what about S3 vs S4? F and S Y point in opposite directions 7

8 Risk Adjusted Success (RAS) RAS is similar to a Sharpe ratio but applied to the evaluation of retirement strategies Risk Adjusted Success (RAS) (100 retirement periods / No bequest) S3 vs S4 S4 Same E(Y S ) Lower SD(Y S ) Higher RAS 8

9 Two More Ideas The comparison between S3 and S4 suggested two additional insights What if the risk of Y S were measured not with its standard deviation but with its semideviation? If different individuals make different choices, then the utility they get from S3 and S4 may be different These insights were developed into two more analytical frameworks to evaluate retirement strategies Downside risk-adjusted success (D-RAS) A utility-based approach Downside RAS (D RAS) D-RAS is similar to a Sortino ratio but applied to the evaluation of retirement strategies It basically aims to avoid penalizing strategies that leave large bequests 9

10 Downside RAS (D RAS) D-RAS does not penalize aggressive strategies that leave large bequests For this reason, it typically selects more aggressive strategies than RAS A Utility Based Approach This approach involves defining a new variable, the coverage ratio (C), and then calculating C for each retirement period considered the utility of each C the expected (mean) utility of the strategy Define Y t : Number of years of inflation-adjusted withdrawals sustained by a strategy C t = Y t /L : Coverage ratio in retirement period t Note that, if a strategy fails, then C < 1 if a strategy leaves a bequest, then C > 1 C contains all the information in F, S Y, and Y S 10

11 A Utility Based Approach This approach involves a kinked utility function A Utility Based Approach 11

12 Evaluation The Big Picture Final Thoughts It is critically important for retirees to implement a good retirement strategy But what is a good (or better, or optimal) strategy? This is where evaluation approaches come in Many different approaches have been proposed to evaluate retirement strategies They all have pros and cons But some are more comprehensive than others My (probably biased) take on this line of research The failure rate is incomplete but here to stay I like the coverage ratio (My co-author s idea) It can be used to build risk-adjusted success measures It can be used within a utility-based approach 12

13 Evaluation of Retirement Strategies Thank you 13