Instantaneous Gratification:

1
Instantaneous Gratification Behavior, Models,
and Retirement Savings Policy
David Laibson Harvard University and NBER July
16, 2008
2
1. Motivating Experiments A Thought Experiment

• Would you like to have
• 15 minute massage now
• or
• B) 20 minute massage in an hour
• Would you like to have
• C) 15 minute massage in a week
• or
• D) 20 minute massage in a week and an hour

3
Read and van Leeuwen (1998)
Choosing Today
Eating Next Week
Time
If you were deciding today, would you
choose fruit or chocolate for next week?
4
Patient choices for the future
Choosing Today
Eating Next Week
Time
Today, subjects typically choose fruit for next
week.
74 choose fruit
5
Impatient choices for today
Choosing and Eating Simultaneously
Time
If you were deciding today, would you
choose fruit or chocolate for today?
6
Time Inconsistent Preferences
Choosing and Eating Simultaneously
Time
70 choose chocolate
7
Read, Loewenstein Kalyanaraman (1999)

• Choose among 24 movie videos
• Some are low brow Four Weddings and a Funeral
• Some are high brow Schindlers List
• Picking for tonight 66 of subjects choose low
  brow.
• Picking for next Wednesday 37 choose low brow.
• Picking for second Wednesday 29 choose low
  brow.
• Tonight I want to have fun
  next week I want things that are good for me.

8
In-class experiment with undergraduates

• Using incentive compatible revelation mechanism I
  measure
• 12 discount rate between day 0 and day 5
• 4 discount rate between day 20 and day 25
• Subjects choose relatively impatiently in the
  short-run.
• Strotz (1950s), Herrnstein (1960s), and Ainslie
  (1970s), were the first to understand this
  phenomenon.

9
Extremely thirsty subjectsMcClure, Ericson,
Laibson, Loewenstein and Cohen (2007)

• Choosing between, juice now
  or 2x juice in 5 minutes 60 of subjects
  choose first option.
• Choosing between juice in 20 minutes or
  2x juice in 25 minutes 30 of subjects choose
  first option.
• We estimate that the 5-minute discount rate is
  50 and the long-run discount rate is 0.
• Ramsey (1930s), Strotz (1950s), Herrnstein
  (1960s) were the first to understand that
  discount rates are higher in the short run than
  in the long run.

10
Conceptual Outline

• People are not internally consistent
  decision-makers
• Internal conflicts can be modeled and measured
• Scalable, inexpensive policies can transform
  behavior

11
Outline

• Motivating experimental evidence
• Theoretical framework
• Field evidence
• Neuroscience foundations
• Neuroimaging evidence
• Study 1 Amazon gift certificates
• Study 2 chips and juice
• Policy analysis

12
2. Theoretical Framework

• Classical functional form exponential functions.
• D(t) dt
• D(t) 1, d, d2, d3, ...
• Ut ut d ut1 d2 ut2 d3 ut3 ...
• But exponential function does not show instant
  gratification effect.
• Discount function declines at a constant rate.
• Discount function does not decline more quickly
  in the short-run than in the long-run.

13
Constant rate of decline
-D'(t)/D(t) rate of decline of a discount
function
14
Slow rate of decline in long run
Rapid rate of decline in short run
15
An exponential discounting paradox.

• Suppose people discount at least 1 between today
  and tomorrow.
• Suppose their discount functions were
  exponential.
• Then 100 utils in t years are worth
  100e(-0.01)365t utils today.
• What is 100 today worth today? 100.00
• What is 100 in a year worth today? 2.55
• What is 100 in two years worth today? 0.07
• What is 100 in three years worth today?
  0.00

16
An Alternative Functional Form

• Quasi-hyperbolic discounting
• (Phelps and Pollak 1968, Laibson 1997)
• D(t) 1, bd, bd2, bd3, ...
• Ut ut bdut1 bd2ut2 bd3ut3 ...
• Ut ut b dut1 d2ut2 d3ut3
  ...
• b uniformly discounts all future periods.
• d exponentially discounts all future periods.

17
Building intuition

• To build intuition, assume that b ½ and d 1.
• Discounted utility function becomes
• Ut ut ½ ut1 ut2 ut3 ...
• Discounted utility from the perspective of time
  t1.
• Ut1 ut1 ½
  ut2 ut3 ...
• Discount function reflects dynamic inconsistency
  preferences held at date t do not agree with
  preferences held at date t1.

18
Exercise

• Assume that b ½ and d 1.
• Suppose exercise (current effort 6) generates
  delayed benefits (health improvement 8).
• Will you exercise?
• Exercise Today -6 ½ 8 -2
• Exercise Tomorrow 0 ½ -6 8 1
• Agent would like to relax today and exercise
  tomorrow.
• Agent wont follow through without commitment.

19
3. Field EvidenceDella Vigna and Malmendier
(2004)

• Average cost of gym membership 75 per month
• Average number of visits 4
• Average cost per vist 19
• Cost of pay per visit 10

20
Choi, Laibson, Madrian, Metrick
(2002)(unreliable?) self-reports about
undersaving.

• Survey
• Mailed to 590 employees (random sample)
• 195 usable responses
• Matched to administrative data on actual savings
  behavior
• Consider a population of 100 respondents
• 68 report saving too little
• 24 of 68 plan to raise 401(k) contribution in
  next 2 months
• Only 3 of 24 actually do so in the next 4 months

21
Financial education Choi, Laibson, Madrian,
Metrick (2004)

• Seminars presented by professional financial
  advisors
• Curriculum Setting savings goals, asset
  allocation, managing credit and debt, insurance
  against financial risks
• Seminars offered throughout 2000
• Linked data on individual employees seminar
  attendance to administrative data on actual
  savings behavior before and after seminar

22
Effect of education is positive but small
23
100 bills on the sidewalkChoi, Laibson, Madrian
(2004)

• Employer match is an instantaneous, riskless
  return on investment
• Particularly appealing if you are over 59½ years
  old
• Have the most experience, so should be savvy
• Retirement is close, so should be thinking about
  saving
• Can withdraw money from 401(k) without penalty
• We study seven companies and find that on
  average, half of employees over 59½ years old are
  not fully exploiting their employer match
• Average loss is 1.6 of salary per year
• Educational intervention has no effect

24

• Financial education effects are small
• Seminar attendees have good intentions to change
  their 401(k) savings behavior, but most do not
  follow through
• Financial education alone will not dramatically
  improve the quality of 401(k) savings outcomes
• Choi et al (2005) study the effect of the Enron,
  Worldcom, and Global Crossing scandals on
  employer stock holding
• No net sales of employer stock in reaction to
  these news stories
• These scandals did not affect the asset
  allocation decisions of new hires.
• These hires did not affect the asset allocation
  decisions of new hires at other Houston firms.

25
Laibson, Repetto, and Tobacman (2007)

• Use MSM to estimate discounting parameters
• Substantial illiquid retirement wealth W/Y
  3.9.
• Extensive credit card borrowing
• 68 didnt pay their credit card in full last
  month
• Average credit card interest rate is 14
• Credit card debt averages 13 of annual income
• Consumption-income comovement
• Marginal Propensity to Consume 0.23
• (i.e. consumption tracks income)

26
LRT Simulation Model

• Stochastic Income
• Lifecycle variation in labor supply (e.g.
  retirement)
• Social Security system
• Life-cycle variation in household dependents
• Bequests
• Illiquid asset
• Liquid asset
• Credit card debt
• Numerical solution (backwards induction) of 90
  period lifecycle problem.

27
LRT Results

• Ut ut b dut1 d2ut2 d3ut3
  ...
• b 0.70 (s.e. 0.11)
• d 0.96 (s.e. 0.01)
• Null hypothesis of b 1 rejected (t-stat of 3).
• Specification test accepted.
• Moments

Empirical Simulated
(Hyperbolic) Visa 68 63 Visa/Y
13 17 MPC 23 31 f(W/Y)
2.6 2.7
28
4. Neuroscience Foundations

• What is the underlying mechanism?
• Why are our preferences inconsistent?
• Is it adaptive?
• How should it be modeled?
• Does it arise from a single time preference
  mechanism (e.g., Herrnsteins reward per unit
  time)?
• Or is it the resulting of multiple systems
  interacting (Shefrin and Thaler 1981, Bernheim
  and Rangel 2004, ODonoghue and Loewenstein 2004,
  Fudenberg and Levine 2004)?

29
Shiv and Fedorikhin (1999)

• Cognitive burden/load is manipulated by having
  subjects keep a 2-digit or 7-digit number in mind
  as they walk from one room to another
• On the way, subjects are given a choice between a
  piece of cake or a fruit-salad

30
Meso-limbic dopamine system vs. Fronto-Parietal
System
Frontal cortex
Parietal cortex
Mesolimbic dopamine system
31
Relationship to quasi-hyperbolic model

• Hypothesize that mesolimbic dopamine system is
  impatient.
• Hypothesize that the fronto-parietal system is
  patient
• Heres one implementation of this idea
• Ut ut b dut1
  d2ut2 d3ut3 ...
• (1/b)Ut (1/b)ut dut1
  d2ut2 d3ut3 ...
• (1/b)Ut (1/b-1)ut d0ut d1ut1 d2ut2
  d3ut3 ...
• limbic
  fronto-parietal cortex

32
Overview of candidate discount functions, D(t),
and value functions, V(t).
quasi-hyperbolic
affine transformation of quasi-hyperbolic
Limbic
Cortical
two system generalization
two system double exponential
33
Hypothesis
Limbic system discounts reward at a higher rate
than does the prefrontal cortex.
34
5. Neuroimaging EvidenceMcClure, Laibson,
Loewenstein, and Cohen Science (2004)

• Do agents think differently about immediate
  rewards and delayed rewards?
• Does immediacy have a special emotional
  drive/reward component?
• Does emotional (mesolimbic) brain discount
  delayed rewards more rapidly than the analytic
  (fronto-parietal cortex) brain?

35
Methods

• Subjects choose between two rewards
• 15 gift certificate at d or 20 gift certificate
  at dgtd
• Identify regions that show elevated activation
  only when immediacy is an option (i.e., d0 v.
  dgt0) b regions.
• Identify regions that show elevated activation
  when making any intertemporal decision relative
  to benchmark of resting state d regions.
• Hypothesize that b regions are limbic and
  para-limbic.
• Hypothesize that d regions are fronto-parietal.

36
Choices involving Amazon gift certificates
Time

• delay dgt0
  d
• Reward R R
• Hypothesis fronto-parietal cortex.
• delay d0 d
• Reward R R
  
• Hypothesis fronto-parietal cortex and limbic.

Time
37
Methods
Subjects given a series of choices between
(R at d) and (R' at d') where RltR' and
dltd'.
d d'-d (R'-R)/R
? Today, 2 weeks, 1 month ? 2 weeks, 1
month ? 1, 3, 5, 10, 15, 25, 35, 50
38
Q Which regions only show elevated activation
when a subject considers an immediate reward? A
Limbic systems and para-limbic cortex.
ventral striatum
medial PFC
hippocampus
7
T13
ventral striatum
0
x -4mm
posterior cingulate cortex
medial OFC
medial OFC
39

• Areas respond only to immediate rewards

BOLD Signal
Seconds
40
McClure, Laibson, Loewenstein, and Cohen Science
(2004)
Emotional system responds only to immediate
rewards
7
T13
0
Neural activity
Seconds
41
b Analysis Summary of Significant Voxels
All voxels significant at p lt 0.001
42
d Areas respond equally to all rewards
43
Analytic brain responds equally to all rewards
VCtx
RPar
PMA
x 44mm
DLPFC
VLPFC
LOFC
x 0mm
15
0
T13
44
d Analysis Summary of Significant Voxels
All voxels significant at p lt 0.001
45
Effect of Difficulty
BOLD Signal
Sec
(R-R)/R
BOLD Signal
Response Time (sec)
Sec
1-3 35-50
5-25
(R-R)/R
46
Brain Activity in the Frontal System and
Emotional System Predict Behavior(Data for
choices with an immediate option.)
Frontalsystem
0.05
Brain Activity
0.0
Emotional System
-0.05
Choose Larger Delayed Reward
Choose Smaller Immediate Reward
47
Hypothesized brain regions predict choice (Trials
with immediate reward as an option.)
Z-score
48
Conclusions of Amazon study

• Time discounting results from the combined
  influence of two neural systems
• Mesolimbic dopamine system is impatient.
• Fronto-parietal system is patient.
• These two systems are separately implicated in
  emotional and analytic brain processes.
• When subjects select delayed rewards over
  immediately available alternatives, analytic
  cortical areas show enhanced changes in activity.

49
Overview

• Emotional brain doesnt see the future (myopic as
  in Shefrin Thaler, Bernheim Rangel,
  ODonoghue and Loewenstein, Fudenberg and Levine)
• Fronto-Parietal/Analytic brain treats the future
  the same way that it treats the present.

50
Open questions

• What is now and what is later?
• Our immediate option (Amazon gift certificate)
  did not generate immediate consumption.
• Also, we did not control the time of consumption.
• How does the limbic signal decay as rewards are
  delayed?
• Would our results replicate with a different
  reward domain?
• Would our results replicate over a different time
  horizon?

• New experiment on primary rewards Juice
• McClure, Ericson, Laibson, Loewenstein,
  Cohen
• (Journal of Neuroscience, 2007)

51
Subjects water deprived for 3hr prior to
experiment
(subject scheduled for 600)
52
A
15s
10s
5s
Time

i
ii
iii
iv. Juice/Water squirt (1s )
B
(i) Decision Period
(ii) Choice Made
(iii) Pause
(iv) Reward Delivery
Free (10s max.)
2s
Free (1.5s Max)
Variable Duration
15s
Figure 1
53
Experiment Design
d d'-d (R,R')
? This minute, 10 minutes, 20 minutes ? 1
minute, 5 minutes ? (1,2), (1,3), (2,3)
54
Behavioral evidence for non-exponential
discounting
0.8
0.6
P(choose early)
0.4
0.2
0
This minute
10 minutes
20 Minutes
Delay to early reward (d)
55
Behavioral evidence for non-exponential
discounting
d-d 1 min
0.8
0.8
d-d 5 min
0.6
0.6
P(choose early)
0.4
0.4
0.2
0.2
0
0
This minute
10 minutes
20 minutes
This minute
10 minutes
20 Minutes
Delay to early reward (d)
Delay to early reward (d)
56
Discount functions fit to behavioral data
ß 0.53 (se 0.041) d 0.98 (se 0.014)
Limbic
Cortical

• 0.47 (se 0.101)
• d 1.02 (se 0.018)

• Evidence for two-system model
• Can reject restriction to a single exponential
  t-stat gt 5
• Double exponential generalization fits data best

57
Juice and Water treated equally (both behavioral
and neurally)
SMA
Ins
VStr
y 12mm
Time (2 second increments)
58
Neuroimaging data
Areas that respond primarily to immediate rewards
ACC
PCu
NAcc
ACC
PCu
PCC
MOFC/SGC
NAcc
SGC
x -12mm
x -2mm
x -8mm
z -10mm
Areas that show little discounting
BA9/44
BA10
BA46
PCC
SMA/PMA
PPar
Vis Ctx
Ant Ins
x 0mm
x 40mm
x -48mm
Figure 4
59
Comparison with Amazon experiment
Impatient areas (plt0.001)
Impatient areas (plt0.01)
x 0mm
y 8mm
Patient areas (plt0.001)
Patient areas (plt0.01)
Figure 5
60
Measuring discount functions using neuroimaging
data

• Impatient voxels are in the emotional
  (mesolimbic) reward system
• Patient voxels are in the analytic (prefrontal
  and parietal) cortex
• Average (exponential) discount rate in the
  impatient regions is 4 per minute.
• Average (exponential) discount rate in the
  patient regions is 1 per minute.

61
Fitting discount functions using neuroimaging
data
ß systems
N(t) ßdR ßd'R' X(t)? e(t)
ß 0.963 (0.004) 0.96325 0.39
d systems
N(t) ddR dd'R' X(t)? e(t)
d0.990 (0.003) 0.99025 0.78
62
Fitting modular exponential discount functions
using neuroimaging data.
b systems
d systems
63
(No Transcript)
64
(No Transcript)
65
What determines immediacy?
Is mesolimbic reward activation associated with
relatively early (or earliest) options?
or Do juice and money have different discount
functions?
or Does thirst invoke more intense discounting?
66
Summary of neuroimaging evidence

• One system associated with midbrain dopamine
  neurons (mesolimbic dopamine system) discounts at
  a high rate.
• Second system associated with lateral prefrontal
  and posterior parietal cortex discounts at a low
  rate.
• Combined function of these two systems accounts
  for decision making consistently across choice
  domains, including non-exponential discounting
  regularities.

67
Outline

• Experimental evidence for dynamic inconsistency.
• Theoretical framework quasi-hyperbolic
  discounting.
• Field evidence dynamic decisions.
• Neuroscience
• Mesolimbic Dopamine System (emotional, impatient)
• Fronto-Parietal Cortex (analytic, patient)
• Neuroimaging evidence
• Study 1 Amazon gift certificates
• Study 2 juice squirts
• Policy
• The Age of Reason

68
6. PolicyDefaults in the savings domain

• Welcome to the company
• If you dont do anything
• You are automatically enrolled in the 401(k)
• You save 2 of your pay
• Your contributions go into a default fund
• Call this phone number to opt out of enrollment
  or change your investment allocations

69
Madrian and Shea (2001)Choi, Laibson, Madrian,
Metrick (2004)
Automatic enrollment
Standard enrollment
70
Employees enrolled under automatic enrollment
cluster at default contribution rate.
Fraction of Participants at different
contribution rates
Default contribution rate under
automatic enrollment
71
Participants stay at the automatic enrollment
defaults for a long time.
Fraction of Participants Hired Under Automatic
Enrollment who are still at both Default
Contribution Rate and Asset Allocation
Fraction of Participants
Tenure at Company (Months)
72
Survey given to workers who were subject to
automatic enrollment You are glad your
company offers automatic enrollment. Agree?
Disagree?
Do people like a little paternalism?

• Enrolled employees 98 agree
• Non-enrolled employees 79 agree
• All employees 97 agree

Source Harris Interactive Inc.
73
The power of deadlines Active decisions
Carroll, Choi, Laibson, Madrian, Metrick (2004)

• Active decision mechanisms require employees to
  make an active choice about 401(k) participation.
  
• Welcome to the company
• You are required to submit this form within 30
  days of hire, regardless of your 401(k)
  participation choice
• If you dont want to participate, indicate that
  decision
• If you want to participate, indicate your
  contribution rate and asset allocation
• Being passive is not an option

74
Active Decision Cohort
Standard enrollment cohort
75
Simplified enrollment raises participation Beshear
s, Choi, Laibson, Madrian (2006)
2005
2004
2003
76
Consumer policy

• Savings interventions have several key features
• Effective (large behavior changes)
• Scalable and inexpensive
• Well-received by public
• Can we find similar types of policies for
  consumer goods?
• Problems with consumer policy
• Employer cant play such a central role
• Consumer preferences are highly heterogeneous
• People dont like delegating most choices

77
Consumer Policy Two reasons for optimism

• Most people say that they want and plan to
  live healthfully (nutrition, diet, sleep,
  smoking, etc)
• Firms and workers have aligned incentives
  vis-à-vis health outcomes, so firms may be
  interested in greasing the rails of health
  behavior change

78
Possible directions

• Change defaults
• Make default portions smaller and more healthful
  (e.g. dont bundle sandwich with potato chips)
• Make good behaviors easier
• Put healthy snacks in vending machines (dried
  fruit/nuts replace candy diet vs. sugar soda)
• Design inviting (sunlit) stairways in office
  buildings
• Choose in advance (long-run preference is
  patient)
• Employees place electronic lunch orders before 10
  AM
• Gyms encourage appointments or pre-commitments
• Early decision regulations (e.g. tobacco)

79
Early Decision RegulationsBeshears, Choi,
Laibson, Madrian (2006)

• Regulations can tip the behavioral balance in
  favor of the (patient) long-run perspective and
  away from the (impatient) short-run perspective.
• If people followed their own long-run plans they
  would
• eat more fruit
• smoke fewer cigarettes
• drink less alcohol
• make fewer wagers
• How would such regulations work?
• What are the costs and benefits?

80
General implications

• In quasi-hyperbolic model, decisions made today
  for tomorrow implement patient preferences
• ß does not play a role, since ß multiplies all
  future periods.
• Ut ut b dut1 d2ut2 d3ut3
  ...
• If society wants people to be patient, then it is
  optimal to allow consumers to choose future
  consumption (partially) in advance.
• We call this an early decision.
• If consumers have different but predictable
  prefer-ences, early decision regulation may
  complement sin taxes.
• Can regulation implement early decisions?

81
A regulatory example

• To buy cigarettes you must have a free tobacco
  sticker, available to adults who apply for one.
• Smokers who want to quit can destroy their
  sticker (it takes 4 wks to get a new one).
• Smokers who want to quit can fail to apply for a
  new sticker when their old one expires.
• System creates limited ability to pre-commit.
• If you dont have a sticker its hard to smoke on
  impulse (need to find an enabling friend).

82
Conceptual linkages

• Request sticker today so you can smoke in the
  future.
• Happy smokers will get sticker.
• Some ambivalent smokers wont.
• System enables happy smokers to keep on smoking,
  and helps ambivalent smokers quit.
• Differential impact.
• Early Decision regulations primarily affect
  consumers whose long-run goals contradict their
  short-run impulses.

83
Other Early Decision regulations

• Location-based regulations
• Restrict density of vendors
• Time-based regulations
• Restrict hours/days of availability
• Delay-based regulations
• Eliminate immediacy (e.g. mail-order)
• Self-regulation
• Consumers can constrain themselves (e.g.
  cigarette sticker or web-based self-rationing)

84
All work on the same principle

• Restricting continuous availability forces
  consumers to decide for the future.
• So early perspective is given an advantage.
• If consumers want to smoke they can.
• But, if early preference is to quit, that
  preference can be (partly) enforced.

85
The path to hell...Unintended costs of early
decisions.

• Encourages black market
• But sin taxes may be an even greater inducement
  for black markets
• Reduces consumer autonomy/agency
• Really a transfer of agency to early perspective
  from immediate perspective
• May induce perverse demand effects
• Creates inefficient nuisance costs for some
  consumers

86
Early decision regulations

• Long-run preferences tend to be patient.
• Short-run preferences tend to be impatient.
• Early Decision regulations give an edge to our
  long-run preferences.
• Early Decision regulations help consumers follow
  through on their long-run goals, by making it
  harder to impulsively reverse long-run plans.
• Need small scale testing.
• The proofs in the field experiment.
• The UK government has recently developed a plan
  that matches our proposal.

87
Implementation?

• Sweden already has a (weak) form of Early
  Decision regulation in the alcohol market.
• Extensions to other markets (cigarettes)?
• Try on small scale first.
• E.g. a few small, isolated towns
• Need experimentation to determine which Early
  Decision regulations work (i.e. benefits outweigh
  costs).

88
7. The Age of ReasonAgarwal, Driscoll, Gabaix,
Laibson (2008)
89
(1,2) Home Equity Loans and Home Equity Credit
Lines

• Proprietary data from large financial
  institutions
• 75,000 contracts for home equity loans and lines
  of credit, from March-December 2002 (all prime
  borrowers)
• We observe
• Contract terms APR and loan amount
• Borrower demographic information age,
  employment status, years on the job, home tenure,
  home state location
• Borrower financial information income,
  debt-to-income ratio
• Borrower risk characteristics FICO (credit)
  score, loan-to-value (LTV) ratio

90
Home Equity Regressions

• We regress APRs for home equity loans and credit
  lines on
• Risk controls FICO score and Loan to Value
  (LTV)
• Financial controls Income and debt-to-income
  ratio
• Demographic controls state dummies, home tenure,
  employment status
• Age spline piecewise linear function of borrower
  age with knots at age 30, 40, 50, 60 and 70.
• Next slide plots fitted values on age splines

91
(No Transcript)
92
(No Transcript)
93
What is the Channel for the Age Effect?

• Banks offer different APRs when the loan-to-value
  (LTV) ratio is
• less than 80 percent
• between 80 and 90 percent
• over 90 percent
• Borrowers estimate their LTV by estimating their
  house value
• Banks form their own LTV estimates
• Rate-Changing Mistake when borrower and bank
  LTVs straddle two of these categories
• E.g., borrower LTV lt 80, bank LTV gt 80.

94

• Rate Changing Mistakes generate two sources of
  disadvantage for the customer
• If I underestimate my LTV (Loan-to-Value ratio),
  the bank can penalize me by deviating from its
  normal offer sheet.
• If I overestimate my LTV (i.e., underestimate the
  value of my house), the bank will penalize me by
  not correcting my mistake and allowing me to
  borrow at too high a rate.

95

• A Rate-Changing Mistake costs 125 to 150 basis
  points.
• Next slides plot
• Rate-Changing Mistakes by age
• APRs for borrowers who do NOT make a
  Rate-Changing Mistake

96
(No Transcript)
97
(No Transcript)
98
(No Transcript)
99

• For consumers who dont make a Rate-Changing
  Mistake, age effect is small
• All the action is due to consumers who make a
  Rate-Changing Mistake
• That is, consumers who over- or under-estimate
  their house values (relative to bank model)
• The propensity to make the mistake is U-shaped
  with age
• Hence, the final APR is U-shaped with age

100
Two channels by which RCM raise interest payments

• Direct channel old and young borrowers may have
  a higher ex-ante likelihood of making a RCM
• Indirect channel old and young borrowers may
  have a higher ex-poste likelihood of accepting
  the high interest rates they receive after they
  make a RCM (instead of shopping around)

101
(3) Eureka Learning to Avoid Interest Charges
on Balance Transfer Offers

• Balance transfer offers borrowers pay lower
  APRs on balances transferred from other cards for
  a six-to-nine-month period
• New purchases on card have higher APRs
• Payments go towards balance transferred first,
  then towards new purchases
• Optimal strategy make no new purchases on card
  to which balance has been transferred

102
Eureka Predictions

• Borrowers may not initially understand / be
  informed about card terms
• Borrowers may learn about terms by observing
  interest charges on purchases, or talking to
  friends
• We should see eureka moments new purchases on
  balance-transfer cards should drop to zero (in
  the month after borrowers figure out the card
  terms)
• Study 14,798 accounts which accepted such offers
  over the period January 2000 to December 2002

103
(No Transcript)
104
(No Transcript)
105
Seven other examples

• Three kinds of credit card fees
• Late payment
• Over limit
• Cash advance
• Credit card APRs
• Mortgage APRs
• Auto loan APRs
• Small business credit card APRs

106
(No Transcript)
107
(No Transcript)
108
(No Transcript)
109
(No Transcript)
110
(No Transcript)
111
U-shape for prices paid in 10 examples

• Home equity loans
• Home equity lines of credit
• Eureka moments for balance transfers
• Late payment fees
• Over credit limit fees
• Cash advance fees
• Auto loans
• Credit cards
• Small business credit cards
• Mortgages

112
Salthouse Studies Memory and Analytic Tasks
Source Salthouse (forth.)
113
DementiaFerri et al 2005

• Prevalence of dementia
• 60-64 0.8
• 65-69 1.7
• 70-74 3.3
• 75-79 6.5
• 80-84 12.8
• 85 30.1

114
Outline

• Motivating experimental evidence
• Theoretical framework
• Field evidence
• Neuroscience foundations
• Neuroimaging evidence
• Policy discussion
• Defaults
• Deadlines
• Simplicity (make it easy)
• A copy of these slides will soon be available on
  my Harvard website.

115
End
116
Should Defaults Influence Economic Outcomes?

• Standard neoclassical theory If
  transactions costs are small and stakes are
  large, defaults should not influence rational
  consumers.
• In practice, defaults make an enormous
  difference
• Organ donation
• Car insurance
• Car purchase options
• Consent to receive e-mail marketing
• Savings
• Asset allocation

117
Overview of defaults

• Defaults affect all saving and asset allocation
  outcomes
• Four psychological factors jointly contribute to
  the default effect
• How can we identify optimal defaults
• Alternative interventions (education) is much
  less effective

118
1. Defaults Affect Saving and Asset Allocation

• i. Participation
• ii. Contribution rates
• iii. Asset allocation
• iv. Pre-retirement distributions
• v. Decumulation / annuitization

119
Participation, Contribution rates, and Asset
Allocation

• Automatic Enrollment in a US 401(k) plan
• Welcome to the company
• If you dont do anything
• You are automatically enrolled in the 401(k)
• You save 2 of your pay
• Your contributions go into a money market fund
• Call this phone number to opt out of enrollment
  or change your investment allocations

120
Madrian and Shea (2001)Choi, Laibson, Madrian,
Metrick (2004)
121
Employees enrolled under auto-enrollment cluster
at the default contribution rate.
Default contribution rate under
automatic enrollment
122
Participants stay at the automatic enrollment
defaults for a long time.
123
Automatic enrollment

• Participants hired under automatic enrollment
  tend to stay at the automatic enrollment defaults
  (about 75)
• Default saving rates
• Default asset allocation
• Automatic enrollment results suggest that
  employees are passive

124
Automatic enrollment Conclusions

• Automatic enrollment dramatically increases
  401(k) participation
• Participants hired under automatic enrollment
  tend to stay at the automatic enrollment defaults
• Similar default effects are observed for
• cash distributions at separation
• saving rates at match thresholds

125
Additional evidence on Asset Allocation

• Private account component of Swedish Social
  Security system (Cronqvist and Thaler, 2004)
• At inception, one-third of assets are invested in
  the default fund
• Subsequent enrollees invest 90 of assets in the
  default fund
• Company match in employer stock (Choi, Laibson
  and Madrian, 2005b, 2007)

126
The Flypaper Effect in Individual Investor Asset
Allocation (Choi, Laibson, Madrian 2007)

• Studied a firm that used several different match
  systems in their 401(k) plan.
• Ill discuss two of those regimes today
• Match allocated to employer stock and workers can
  reallocate
• Call this default case (default is employer
  stock)
• Match allocated to an asset actively chosen by
  workers workers required to
  make an active designation.
• Call this no default case (workers must choose)

Economically, these two systems are
identical. They both allow workers to do whatever
the worker wants.
127
Consequences of the two regimes
128
Cash Distributions

• What happens to savings plan balances when
  employees leave their jobs?
• Employees can request a cash distribution or roll
  balances over into another account
• Balances gt5000 default leaves balances with
  former employer
• Balances lt5000 default distributes balances as
  cash transfer
• Vast majority of employees accept default (Choi
  et al. 2002, 2004a and 2004b)
• When employees receive small cash distributions,
  balances typically consumed (Poterba, Venti and
  Wise 1998)

129
Post-Retirement Distributions

• Social Security
• Joint and survivor annuity (reduced benefits)
• Defined benefit pension
• Annuity
• Lump sum payout if offered
• Defined contribution savings plan
• Lump sum payout
• Annuity if offered

130
Defined Benefit Pension Annuitization

• Annuity income and economic welfare of the
  elderly
• Social Security replacement rate relatively low
  on average
• 17 of women fall into poverty after the death of
  their spouse (Holden and Zick 2000)
• For married individuals, three distinct
  annuitization regimes
• Pre-1974 no regulation
• ERISA I (1974) default joint-and-survivor
  annuity with option to opt-out
• ERISA II (1984 amendment) default
  joint-and-survivor annuity, opting out required
  notarized permission of spouse

131
Defined Benefit Pension Annuitization

• Effect of joint-and-survivor default on
  annuitization
• Pre-1974 Less than half of married men have
  joint-and-survivor annuity
• Post-ERISA (I II) joint-and-survivor
  annuitization increases 25 percentage points
  (Holden and Nicholson 1998)
• Post-1984 amendments joint-and-survivor
  annuitization increases 5 to 10 percentage points
  (Saku 2001)

132
Four psychological factors contribute to the
default effect

• Financial illiteracy
• Endorsement
• Complexity
• Present-bias

133
i. Financial illiteracy

• John Hancock Financial Services Defined
  Contribution Plan Survey (2002)
• 38 of respondents report that they have little
  or no financial knowledge
• 40 of respondents believe that a money market
  fund contains stocks
• Two-thirds of respondents dont know that it is
  possible to lose money in government bonds
• Respondents on average believe that employer
  stock is less risky than a stock mutual fund
• Two-thirds report that they would be better off
  working with an investment advisor than managing
  investments solo

134
Financial illiteracy among Wharton MBAsChoi,
Laibson, Madrian (2006)

• Subjects allocate 10,000 among four funds
• Randomly choose two subjects to receive any
  positive portfolio return during the subsequent
  year
• Eliminate variation in pre-fee returns
• Choose among SP 500 index funds
• Unbundle services from returns
• Experimenters pay out portfolio returns, so no
  access to investment company services

135
One year of index fund fees on a 10,000
investment
136
Experimental conditions

• Control
• Subjects receive only four prospectuses
• Prospectuses are often the only information
  investors receive from companies
• Fees transparency treatment
• Eliminate search costs by also distributing fee
  summary sheet (repeats information in prospectus)
• Returns treatment
• Highlight extraneous information by distributing
  summary of funds annualized returns since
  inception (repeats information in prospectus)

137
Fees paid by control groups (prospectus only)
Maximum Possible Fee
443 average fee with random fund allocation
Minimum Possible Fee
N 83
N 30
0 of College Controls put all funds in
minimum-fee fund
6 of MBA Controls put all funds in minimum-fee
fund
t-test p0.5086
138
Ranking of factor importance

• MBA controls
• Fees
• 1-year performance
• Performance since inception
• Investment objectives
• Desire to diversify among funds
• Brand recognition
• Performance over different horizon
• Past experience with fund companies
• Quality of prospectus
• Customer service of fund
• Minimum opening balance

• College controls
• 1-year performance
• Performance since inception
• Desire to diversify among funds
• Investment objectives
• Quality of prospectus
• Performance over different horizon
• Brand recognition
• Fees
• Customer service of fund
• Minimum opening balance
• Past experience with fund companies

139
Effect of fee treatment (prospectus plus 1-page
sheet highlighting fees)

N 83
N 30
N 29
N 85
19 of MBA treatment put all funds in minimum-fee
fund
t-tests MBA p0.0000 College p0.1451
10 of College treatment put all funds in
minimum-fee fund
140
Ranking of factor importance

• MBA fee treatment
• Fees
• 1-year performance
• Performance since inception
• MBA controls
• Fees
• 1-year performance
• Performance since inception

• College fee treatment
• Fees
• 1-year performance
• Performance since inception
• College controls
• 1-year performance
• Performance since inception
• Desire to diversify among funds

141
Returns treatment effect onaverage returns since
inception


N 83
N 30
N 28
N 84
t-tests MBA p0.0055 College p0.0000
142
Returns treatment effect on fees

N 83
N 30
N 28
N 84
t-tests MBA p0.0813 College p0.0008
143
Ranking of factor importance

• MBA return treatment
• 1-year performance
• Performance since inception
• Fees
• MBA controls
• Fees
• 1-year performance
• Performance since inception

• College return treatment
• Performance since inception
• 1-year performance
• Desire to diversify among funds
• College controls
• 1-year performance
• Performance since inception
• Desire to diversify among funds

144
Lack of confidence and fees

N 64
N 46
N 36
N 136
N 50
N 5
t-tests MBA 1 vs. 2, p0.2013 MBA 1 vs. 3,
p0.0479 College 1 vs. 2, p0.2864 College 1
vs. 3, p0.3335
145
We conducted a similar experiment with Harvard
staff as subjects

• In this new version we have 400 subjects
  (administrators, faculty assistants, technical
  personal, but not faculty)
• We give every one of our subjects 10,000 and
  rewarded them with any gains on their investment
• 4,000,000 short position in stock market

146
Data from Harvard Staff
Control Treatment
Fee Treatment
518
Fees from random allocation 431
494
3 of Harvard staff in Control Treatment put all
in low-cost fund
9 of Harvard staff in Fee Treatment put all
in low-cost fund
147
ii. Endorsement

• A non-zero default is perceived as advice
• Evidence
• Elective employer stock allocation in firms that
  do and do not match in employer stock (Benartzi
  2001, Holden and Vanderhei 2001, and Brown, Liang
  and Weisbenner 2006)
• Asset allocation of employees hired before
  automatic enrollment (Choi, Laibson, Madrian 2006)

148
Choi, Laibson, and Madrian (2007)
149
Choi, Laibson, and Madrian (2007)
150
(No Transcript)
151
(No Transcript)
152
iii. Complexity

• Complexity ? delay
• Psychology literature (Tversky and Shafir 1992,
  Shafir, Simonson and Tversky 1993, Dhar and
  Knowlis 1999, Iyengar and Lepper 2000 )
• Savings literature each additional 10 funds
  produces a 1.5 to 2.0 percentage point decline in
  participation (Iyengar, Huberman and Jiang 2004)
• Also results on complexity generating more
  conservative asset allocation (Iyengar and
  Kamenica 2007).
• Quick enrollment experiments

153
Complexity and Quick Enrollment

• Conceptual Idea
• Simplify the savings plan enrollment decision by
  giving employees an easy way to elect a
  pre-selected contribution rate and asset
  allocation bundle
• Implementation at Company D
• New hires at employee orientation 2
  contribution rate invested 50 money market / 50
  stable value
• Implementation at Company E
• Existing non-participants 3 contribution rate
  invested 100 in money market fund

154
Complexity and Quick Enrollment

• Conceptual Idea
• Simplify the savings plan enrollment decision by
  giving employees an easy way to elect a
  pre-selected contribution rate and asset
  allocation bundle
• Implementation at Company D
• New hires at employee orientation 2
  contribution rate invested 50 money market / 50
  stable value
• Existing non-participants employee selects
  contribution rate invested 50 money market / 50
  stable value
• Implementation at Company E
• Existing non-participants 3 contribution rate
  invested 100 in money market fund

155
Complexity and Quick Enrollment

• Conceptual Idea
• Simplify the savings plan enrollment decision by
  giving employees an easy way to elect a
  pre-selected contribution rate and asset
  allocation bundle
• Implementation at Company D
• New hires at employee orientation 2
  contribution rate invested 50 money market / 50
  stable value
• Existing non-participants employee selects
  contribution rate invested 50 money market / 50
  stable value
• Implementation at Company E
• Existing non-participants 3 contribution rate
  invested 100 in money market fund

156
iv. Present-Biased Preferences

• Self control and savings outcomes Why do today
  what you can put off until tomorrow? (Laibson
  1997 ODonoghue and Rabin 1999 Diamond and
  Koszegi 2003)
• Framework exponential discounting with an
  additional factor, ßlt1, that uniformly
  down-weights the future.
• Ut ut b dut1 d2ut2
  d3ut3 ...

157
Laibson, Repetto, and Tobacman (2004)

• Use MSM to estimate discounting parameters
• Substantial illiquid retirement wealth W/Y
  3.9.
• Extensive credit card borrowing
• 68 didnt pay their credit card in full last
  month
• Average credit card interest rate is 14
• Credit card debt averages 13 of annual income
• Consumption-income comovement
• Marginal Propensity to Consume 0.25
• (i.e. consumption tracks income)

158
LRT Simulation Model

• Stochastic Income
• Lifecycle variation in labor supply (e.g.
  retirement)
• Social Security system
• Life-cycle variation in household dependents
• Bequests
• Illiquid asset
• Liquid asset
• Credit card debt
• Numerical solution (backwards induction) of 90
  period lifecycle problem.

159
LRT Results

• Ut ut b dut1 d2ut2 d3ut3
  ...
• b 0.70 (s.e. 0.11)
• d 0.96 (s.e. 0.01)
• Null hypothesis of b 1 rejected (t-stat of 3).
• Specification test accepted.
• Moments

Empirical Simulated
(Hyperbolic) Visa 68 63 Visa/Y
13 17 MPC 23 31 f(W/Y)
2.6 2.7
160
Procrastination (assume b ½ , d 1).

• Suppose you can join the plan today (effort cost
  50) to gain delayed benefits 20,000 (e.g. value
  of match)
• Every period you delay, total benefits fall by
  10.
• What are the discounted costs of joining at
  different periods?
• Join Today -50 ½ 0 -50
• Join t1 0 ½ -50 - 10 -30
• Join t2 0 ½ -50 - 20
  -35
• Join t3 0 ½ -50 - 30
  -40

161
Interaction with financial illiteracy

• Consider someone with a high level of financial
  literacy, so effort cost is only 10 (not 50)
• As before, every period of delay, total benefits
  fall by 10.
• What are the discounted costs of joining at
  different periods?
• Join Today -10 ½ 0 -10
• Join t1 0 ½ -10 - 10 -10
• Join t2 0 ½ -10 - 20
  -15
• Join t3 0 ½ -10 - 30
  -20

162
Interaction with endorsement and complexity

• Consider a plan with a simple form, or an
  endorsed form, so the effort cost is again only
  10 (not 50)
• As before, every period of delay, total benefits
  fall by 10.
• What are the discounted costs of joining at
  different periods?
• Join Today -10 ½ 0 -10
• Join t1 0 ½ -10 - 10 -10
• Join t2 0 ½ -10 - 20
  -15
• Join t3 0 ½ -10 - 30
  -20

163
Procrastination in retirement savingsChoi,
Laibson, Madrian, Metrick (2002)

• Survey
• Mailed to 590 employees (random sample)
• 195 usable responses
• Matched to administrative data on actual savings
  behavior
• Consider a population of 100 employees
• 68 report saving too little
• 24 of 68 plan to raise 401(k) contribution in
  next 2 months
• Only 3 of 24 actually do so in the next 4 months

164
3. Optimal Defaults public policy

• Mechanism design problem in which policy makers
  set a default for agents with present bias
  (Carrol, Choi, Laibson, Madrian and Metrick 2007)
  
• Defaults are sticky for three reasons
• Cost of opting-out of the default
• Cost varies over time ? option value of waiting
• Present-biased preferences ? procrastination

165
Basic set-up of problem

• Specify behavioral model of households
• Flow cost of staying at the default
• Effort cost of opting-out of the default
• Effort cost varies over time ? option value of
  waiting to leave the default
• Present-biased preferences ? procrastination
• Specify (dynamically consistent) social welfare
  function of planner (e.g., set ß1)
• Planner picks default to optimize social welfare
  function

166
Optimal Defaults

• Two classes of optimal defaults
• Automatic enrollment
• Optimal when employees have relatively
  homogeneous savings preferences (e.g. match
  threshold) and relatively little propensity to
  procrastinate
• Active Decision require individuals to make a
  decision (eliminate the option to passively
  accept a default)
• Optimal when employees have relatively
  heterogeneous savings preferences and relatively
  strong tendency to procrastinate
• Key point sometimes the best default is no
  default.

167
30
Low Heterogeneity High Heterogeneity
Offset Default
Active Decision
Center Default
0

0
1
Beta
168
Lessons from theoretical analysis of defaults

• Defaults should be set to maximize average
  well-being, which is not the same as saying that
  the default should be equal to the average
  preference.
• Endogenous opting out should be taken into
  account when calculating the optimal default.
• The default has two roles
• causing some people to opt out of the default
  (which generates costs and benefits)
• implicitly setting savings policies for everyone
  who sticks with the default

169
The power of deadlines Active decisions Choi,
Laibson, Madrian, Metrick (2007)

• Active decision mechanisms require employees to
  make an active choice about 401(k) participation.
  
• Welcome to the company
• You are required to submit this form within 30
  days of hire, regardless of your 401(k)
  participation choice
• If you dont want to participate, indicate that
  decision
• If you want to participate, indicate your
  contribution rate and asset allocation
• Being passive is not an option

170
(No Transcript)
171
Active decisions conclusions

• Active decision raises 401(k) participation.
• Active decision raises average savings rate by
  50 percent.
• Active decision doesnt induce choice
  clustering.
• Under active decision, employees choose savings
  rates that they otherwise would have taken three
  years to achieve. (Average level as well as the
  entire multivariate covariance structure.)

172
New directions for defaults

• Defaults for savings rate escalation
• Defaults with high savings rates
• Defaults for lifecycle rebalancing
• Defaults for annual rebalancing
• Defaults for employer stock
• Defaults at separation
• Defaults for annuitization
• Individualized defaults (savings rate and asset
  allocation)
• Defaults for employees not covered by DB/DC plans
• Defaults for investment of tax refunds

173
Public Policy and DefaultsAnnuitization

• Interesting aspects of the joint-and-survivor
  Social Security annuity default discussed earlier
• Differentiated default singles vs. marrieds
• Annuity election irrevocable
• Implicit deadlinemust either accept or opt-out
  of the default before receiving pension payments
• Note
• Largely homogenous preferences
• Similarities to active decision approach
• Reduced scope for procrastination
• Those who do opt-out of joint-and-survivor
  annuity appear to have economically sound reasons
  for doing so (Johnson, Uccello and Goldwyn 2003)

174
Public Policy and DefaultsAnnuitization

• Thinking more generally about retirement income
  annuitization and defaults in a defined
  contribution world
• Understanding annuitization optio