The Brookings Institution, Washington, D'C'www'brookings'edu

1
Strengthening Retirement Security

• Peter R. Orszag
• Joseph A. Pechman Senior Fellow Economic Studies
• November 2005

2
Tiers of retirement income
3
Adequacy of retirement saving

• Significant share of low- and middle-income
  families appear to be undersaving for retirement
• Median value of 401(k)sIRAs for households on
  verge of retirement only about 10,000

4
Two core impediments to retirement saving

• Complicated and difficult
• Inertia and freezing in response to complexity
  translate into non-saving
• Upside-down incentives
• Weak financial incentives for moderate and
  lower-income household to save

5
Problem 1 Complexity/decision making

• As pension system shifts to 401(k)s, workers
  must
• Decide whether to participate
• Decide how much to contribute
• Decide how to invest
• Decide when and how to take the money out

6
Making It Easier Automating the 401(k)

• Where we are now
• Employee must
• Actively elect to join
• Decide how much to contribute
• Decide how to invest
• Decide what to do at a job change

• Where we should be
• Defaults at each step
• Automatic enrollment
• Automatic escalation
• Default investment in well-designed fund
• Default rollover at job change

7
Impact of 401(k) Auto Enrollment
Default is participation
Default is nonparticipation
Under 20k in earnings
Actual results for employees with between 3 and
15 months tenure. Study by Professor Brigitte
Madrian, University of Pennsylvanias Wharton
School, and Dennis Shea, United Health Group.
8
Automatic Escalation Save More Tomorrow
Source Thaler and Benartzi
9
How We Get There

• Clarify federal ERISA preemption of state law to
  permit auto enrollment
• Provide measure of fiduciary liability protection
  for well-designed default investments
• Allow short grace period for employees to opt out
  and receive penalty-free refund of deferrals
• Explore reforms to non-discrimination rules
• Incorporate key features within Thrift Savings
  Plan

10
Problem 2 Existing tax preferences upside-down

• Tax preferences provide larger incentives to
  high-income households (in 35 percent bracket)
  than lower-income households
• To raise private saving, must not simply cause
  shifts of assets but generate additional
  contributions
• Empirical studies confirm that upper-income
  households disproportionately tend to shift
  existing savings, not save more

11
Alternative form of financial incentive Matching
contributions

• Match for contributions Put 1 into retirement
  account, get a matching contribution (e.g., 50
  cents)
• Matches can be independent of marginal tax rate
• Previous research literature finds mixed effects
  from match rates

12
Research project with HR Block

• First large-scale experiment examining effect of
  match rates on contributions to retirement
  accounts
• Randomized assignment of match rates, overcoming
  major shortcoming of previous studies
• Research team
• Esther Duflo, MIT
• William Gale, RSP and Brookings Institution
• Jeffrey Liebman, Harvard University
• Peter Orszag, RSP and Brookings Institution
• Emmanuel Saez, UC Berkeley
• Research undertaken in conjunction with HR
  Block, nations leading tax preparation firm

13
Research design

• Experiment run in 60 HR tax preparation offices
  in St. Louis area, from March 5th to April 5th
• 15,000 clients
• Each client was randomly assigned a match rate
  for IRA contributions of up to 1,000
• 0 match (control group)
• 20 percent match
• 50 percent match
• Contributions were easy and convenient to make
• Effectively allowed refund to be split into
  retirement contribution and other uses
• Easy IRA signup

14
Matches boost participation
15
and contributions for participants
16
both of which boost average contribution
(including participants and non-participants)
17
Effects stronger for high-income married couples,
but still significant for low-income couples
18
Bottom line

• Significant increase in retirement contributions
  possible, even among low- and middle-income
  households, with
• clear and understandable match for saving
• easily accessible savings vehicles
• opportunity to use part of income tax refund
• professional assistance
• Possible policy reforms
• Extend matching incentive to low- and
  moderate-income households
• Deposit matching money into account, rather than
  providing as tax credit

19
Other ways to boost incentives Attenuate steep
implicit tax on saving

• Asset tests in means-tested benefit programs
  (Food Stamps, SSI, Medicaid, TANF) often impose
  implicit tax on retirement saving
• Rules are also complicated and seemingly
  arbitrary
• Food Stamp Program exempts 401ks but not IRAs, so
  workers rolling over can be disqualified

20
Conclusions

• Two tiers of retirement income
• First tier Social Security
• Second tier Additional saving
• Exciting new evidence on ways to boost saving on
  top of Social Security for low- and
  moderate-income households
• For more information www.retirementsecurityprojec
  t.org