1. An application of behavioural
finance to household finance:
saving for retirement
2. In this lecture we will be seeing how nudge theory
has been applied to tackle a behavioural problem
“Libertarian Paternalism”
“choice architecture”
3. Examples of suboptimal financial decisions
• Payday loans with very high APRs when other cheaper forms of credit are available
(Agarwal et al. 2009, AER; Bertrand and Morse forthcoming, JoF))
• Failure to choose best credit card when facing multiple offers (Agarwal et al. 2006,
RFS R&R)
• Low participation in stock market (Cole & Shastry 2009, RFS R&R)
• Concentrated holdings of stocks (e.g. home bias)
• Excessive trading
• Best deal mortgages are achieved when zero up-front fee (Woodward 2008)
• Excessive use of credit cards (a problem of self-control and seduction of loyalty
rewards)
• Defined contribution retirement plans with generous employer contributions are
often not taken up
• Surveys show that most employees believe their savings are too low to prepare for
retirement
• Propensity for employees to invest in their employer’s stock (compounded by
company stock being exempt from diversification requirements of a 401(k) plan).
4. Readings
• Ackert and Deaves (2010) chapter 17
• Benartzi & Thaler (2007) `‘Heuristics and biases in retirement
savings behavior’, Journal of Economic Perspectives, 21, 81-
104.
• Benartzi & Thaler (2004) Save More Tomorrow: Using
behavioral economics to increase employee saving, Journal of
Political Economy, 112, 1(part 2), S164-S187
• Mitchell & Utkus (2004) Lessons from behavioral finance for
retirement plan design, in Mitchell & Utkus, Pension Design
and Structure: new lessons from behavioral finance, OUP.
• Thaler & Sunstein (2008) Nudge, Yale UP, chapter 6 & 7
• Statman (2013) Mandatory retirement savings, Financial
Analysts Journal, 63(3), 14-18.
• Also see references in slides
5. Trends in pension policy
Move from
• Defined Benefit pensions,
– pension income dependent on a formula based on
number of years in service
– Employer is generally responsible for setting aside a
pension fund to meet the needs of their employees.
To
• Defined Contribution pensions
– Income is dependent on amount invested
– Onus is on the employee to ensure sufficient invested
– DC pensions often also involve choice about what to
invest in
7. • Advantage of DB
– Low risk for employee
• Disadvantage of DB
– Risk is borne by employer, particular problem with
market risk
• Advantage of DC
– More easily transferred from job to job
– Predictable pension contribution for employer
• Disadvantage of DC
– Risk borne by employee
– Places a lot of responsibility on employee
8. Decisions are complicated!
• Example from Ackert & Deaves (2010, pp.298-300)
• 30 year old currently earning £50,000 and £40,000 already
saved for retirement. Planning to work till 60, when she
will buy a 30 year annuity to give her an annual income till
she is 90.
• She needs to calculate her deferral rate. Assume:
– Replacement rate = 70%
– Working years = 30
– Years of annuity = 30
– Wage growth = 0%
– Return on retirement savings = 5%
– Annuity interest rate = 2%
• Given the above, a deferral rate of 13% is implied
9. Example of DC: 401(k)
(similar to personal pension plan in UK)
• Permits employees to set aside tax-deferred
income, up to 25% of annual earnings to a
maximum of $17,500.
• Contributions often matched by employer
• Automatic enrolment is now allowed with opt-
out option
• The employer sets up the plan, but the
employee decides how much to save.
• Employees often choose what to invest in
10. Decisions to be made by employee
with DC pension
1) How much needs to be saved
- heuristics
- self-control
- Procrastination
- Inertia
2) What to invest in
- Poor knowledge of investment principles
- Diversification heuristic
11. Saving rate heuristics
• Multiples of 5
• Maximum employer matches
• Maximum permitted
• From research with Hewitt Associates,
Benartzi & Thaler (2007) illustrate saving rate
heuristics by looking at what happened when
the maximum savings rate was increased in
April 2002 from 16% to 100%
12. Prior to 2002 21% chose
the maximum allowable
rate
With removal of cap, many
switched to multiple of 5
heuristic
13. Do people save enough?
• Lots of survey data suggests employees are concerned that they are not saving
enough.
• An illustration of this from the US is provided by Bernheim et al (2001) "What
Accounts for the Variation in Retirement Wealth among U.S. Households?"
American Economic Review, 91(4), 832-857.
Median Std dev Min Max
Income replacement ratio 0.60 0.32 0.02 2.87
Financial Wealth-income ratio 0.88 5.80 -6.24 63.51
Total Wealth-income ratio 2.36 6.45 -6.10 65.86
• How would you expect the pre- and post-retirement consumption
behaviour of someone with a low income replacement ratio to
someone with a high income replacement ratio?
15. Limited self control
• Evidence of not saving enough and knowing it
• Self control is a common problem, especially
when the pay-off is in the future
16. Exponential or hyperbolic discounting
• When you’re keen to save for the future but not to
save now for the future you’re said to have hyperbolic
discounting. Note, we are considering subjective
discount rates rather than objective discount rates
• Hyperbolic functions capture the procrastination
problem relating to saving and is associated with
deferred gratification.
• Note how after the first period, the hyperbolic function
has the appearance of an exponential function. This
means we can commit to saving in the future but as
the time approaches if we can reverse that decision we
will!
18. Procrastination
• Follows from hyperbolic discounting
• Similar in effect to self control problem, but
different. It involves putting off a decision.
• Procrastination is particularly associated with big
decisions. Decisions that involve a lot of time and
(cognitive) effort. Like setting up a pension!
• One solution – automatic enrolment (e.g. UK
Pension Act 2008, automatic enrolment started in
2012)
19. Inertia
• Sunstein and Thaler (2008) refer to examples
of DB plans in the UK which don’t require any
employee contributions (free money!) but
they do need the employee to act to take up
the plan. Only approximately 50% of
employees take these up!
20. Asset Allocation
Harry Markowitz was asked how he allocated his
retirement account. He answered: “I should
have computed the historic covariances of the
asset classes and drawn an efficient frontier.
Instead … I split my contributions fifty-fifty
between bonds and equities” (quoted in
Sunstein and Thaler 2008, p. 123)
21. Asset allocation
• Diversification heuristics such as 1/n
• Availability and framing biases
– Benartzi & Thaler (2001, AER) use experiments and
aggregate investment data to show proportion invested
in equity depends on % of equity funds offered.
– However, Huberman & Jiang (2006) using actual 401(k)
fund data show that the 1/n heuristic is used over the
chosen funds rather than the range of choices.
In a study of university pensions, the median
number of changes to asset allocation was zero
(Samuelson & Zeckhauser 1988)
22. Source: Sethi-Iyengar et al. (2004) How much choice is too much? Contributions to 401(k) in Mitchell and
Utkus (eds.) (2004)
Participation and cognitive overload
23. Participation, heuristics and cognitive
overload
• Is greater choice a virtue? Greater choice
implies greater complexity
– Increases chance of non-involvement (Huberman
et al. (2007, J Fin Services Res) shows that every 10
funds added to a 401(k) plan is associated with a
drop in participation of 1.5-2%)
– Cognitive overload is likely to increases likelihood
of use of heuristics (Morrin et al. (2012, J
Marketing Res) shows increasing number of funds
available reduces likelihood of investing in all funds
but increases the likelihood of using 1/n heuristic)
25. An example of importance of defaults
• In 1998 a large US company changed from an
opt-in pension plan to automatic enrolment
which requires action on part of employee to
withdraw (Mardian et al. 2001, QJE)
• Prior to 1998, take-up was low for first years
of employment, despite matching employer
contributions up to 6% of income
• The plan increased take-up for new employees
to 86%
26. Madrian et al. (2001) ‘The power of suggestion: inertia in 401(k) participation and
savings behavior’, Quarterly Journal of Economics, 116, 1149-1187
27. Issues raised by example
• Intervention of company requires them to set
other defaults
– Default saving rate was only 3%,
• Not enough to save sufficient for pension
• Employer only matches 3%
– Default investment was in money market fund
• It could be argued that those who wouldn’t have
joined are better off. But what of those who did
join?
• Does this example show possibility of nudging
some in the wrong direction?
28. Two approaches to tackling pension
‘crisis’
• UK: 2008 Pension Act
• US: Save More Tomorrow (SMarT)
29. Save More Tomorrow™: SMarT
Benartzi & Thaler (2004)
• Introduces the idea of a Scheduled Deferral Increase
Programme (SDIP).
• SMarT has four features:
1. Employees asked to commit in future – we don’t mind
deferring good behaviour. A long gap is desirable.
2. Commitment kicks in on next pay rise: side steps loss
aversion and exploits money illusion
3. Starts modestly, but builds in future increases to
coincide with pay rises: exploits inertia
4. Plan members can change their mind at anytime –
minimises deterrence to take up the scheme
30. Evidence on SMarT/SDIPs
• Benartzi & Thaler (2004) report evidence from a
pilot that took place in 1998 in which an
investment consultant visited a firm to advise low
income employees on required savings rate to
meet pension needs. Possible outcomes:
32. Two nudge policy prescriptions
1. Automatic enrolment
2. Accelerated (SMarT) contributions
• SMarT is now used widely in the US and is
increasingly seen in the UK
• In the US, the 2006 Pension Protection Act
require employees who offer a 401(k) with
matching to build in a SMarT component (see
Sunstein & Thaler 2008, chapter 6, fn 15)
33. In the UK: Pensions Act 2008
This Act introduces two key requirements for employers:
• “To automatically enrol eligible jobholders who are not in
a qualifying pension scheme into an automatic enrolment
scheme. Jobholders will be enrolled by their employers from
the first day they become eligible but retain the right to opt
out of the scheme from this point. Employers may choose
the pension scheme they provide, which must meet certain
criteria. The Act gives the Secretary of State powers to
establish a pension scheme in which employers can choose
to participate. This scheme will be run by an independent
trustee body.
• To maintain the jobholder’s membership of a qualifying
scheme, including making relevant contributions, so long as
the jobholder chooses to be part of it.”
http://www.legislation.gov.uk/ukpga/2008/30/notes/division/3
34. 2008 Pension Act
National Employment Saving Trust (NEST)
Target Group
• Aged under 35, no savings, low to modest income
around £18k
• Tends to be risk averse and highly loss averse, so
much so they’re liable to withdraw if face losses
• Not financially literate
In response, some features of NEST include
• Low risk with primary objective to beat inflation
• Risk management through asset allocation, not
derivatives
35. National Employment Savings Trust (NEST)
• Auto-enrolment (can opt out, but auto-enrolled every
3 years)
– Employers required to contribute minimum 2% of earnings
– Employees required to contribute minimum 1%
– Mandatory contributions scheduled to reach total of 8% by
2018 (3% from employers, 4% from employees, 1% from
government)
– Default offerings are 47 target-date funds, at 1 year
intervals the mix of stocks, bonds and cash becomes more
conservative as participant ages
– Centralised investment management, fees of 30bp (Dept of
work & pensions estimate a reduction in management fees
from 150bp to 75bp would save up to £100k for lifetime
saver, 46 years saving, initial annual contribution of £1200
increasing 4% per year, investment growth of 7%)
36. The Building blocks of NEST target date funds
Global Equity
Fund
Passive FTSE All World
Developed Index
UBS
UK Gilts Fund Passive FTSE Actuaries All
Stocks Index
State Street
(SSgA)
Index-Linked Gilt
Fund
Passive FTSE Actuaries Index
Linked Gilts Over 5-
Years
State Street
(SSgA)
Low-Risk
Liquidity Fund
Active 7-Day LIBID BlackRock
Diversified Beta
Fund
Mostly
Passive
UK Risk Free Rate +2%
to +4%
BlackRock
Global Ethical
Equity Fund
Active MSCI World F&C
Sharia Compliant
Global Equity
Fund
Passive Dow Jones Islamic
Titans 100
HSBC
Sterling Bonds Active iBoxx Sterling Non-Gilt
All Maturities Index
Royal London
Asset Mgt
Source: Source: “Developing and delivering NEST’s investment approach,” NEST Corp. 2012.
www.nestpensions.org.uk
37. Some criticisms of libertarian
paternalism
From the left
• Doesn’t address underlying problems
• About rolling back the state
From the right
• Too paternalistic
Doubts relating to effectiveness
• Unproven: mainly small trials; it is easier to nudge people
downhill than to nudge them uphill
• Can nudge people into either too cautious or too ambitious
saving plans
• Some doubt whether it works for people who really need it.
Statman (2013) proposes mandatory DC (acknowledging
death of DB)
38. Statman (2013) proposal: Mandatory
defined-contribution retirement plan
1. Mandatory contributions from both employers and
employees, combined to 12% of earnings (15% is
preferable)
2. Central agency to administer plans for employees
whose employees don’t provide retirement plans
3. Default offerings of well-diversified funds with one-
year intervals (as with NEST)
4. Fees not exceeding 30bp
5. No borrowing from retirement fund or cashing out
before retirement.
6. Enhanced financial literacy through education