The Economist as Therapist: Behavioural economics and "light" paternalism

1
The Economist as Therapist Behavioural
economics and "light" paternalism
2
(No Transcript)
3
In fact, people often dont know whats best for
themselves (and when they do, often have trouble
getting themselves to do it)

• Health behaviors
• 1960-2000, obesity (BMIgt30) in U.S. adults
  increased from 13 to 31
• 1985 no states in U.S. with obesity rate gt 15
• 2005 no states with obesity rate lt 15
• Lifestyle diseases caused by, e.g., tobacco
  alcohol use account for 1/3 of all deaths in
  US.
• Potential benefit of many medicines e.g., to
  control blood pressure, cholesterol and avoid
  strokes -- stymied by poor adherence
• e.g., 1/2 of patients who have a heart attack
  stop taking cholesterol medication in first year
• Spending/saving/investing
• Average U.S. savings rate -1
• Median Net worth, excluding home equity (year
  2000)
• All U.S. households 13,473
• Households 65 23,369

4

• Traditional economics not well equipped to deal
  with these problems assumes that people..
• know whats best for themselves
• are able to act on that understanding
• ?little or no need for intervention (beyond
  problem of externalities)
• ?focus on prices and/or information as main tools
  for policy

5

• Behavioral economics allows for mistakes.
  People often...
• dont know whats best for themselves
• do know, but cant do it
• ?motivates intervention (much as one intervenes
  in diet of children)
• ?inspires new approach to policy light
  paternalism

6

• Paternalism
• policies intended to benefit individuals,
  premised on the idea that people cannot be relied
  upon to pursue self-interest commonly justified
  for children and others deemed unable to behave
  rationally
• distinct from regulations intended to deal with
  externalities i.e., to protect others
• Behavioral economics
• expands the possible application of paternalism
  by identifying systematic mistakes made by a
  broad cross-section of the population

7
However, widespread discomfort with
heavy-handed paternalism

• Fear that policy-makers wont make better
  decisions
• Fear of regulatory capture (e.g., cigarette
  companies knew warnings wouldnt help but would
  shield them from liability)
• People may have good reasons for behaving as they
  do (that policy-makers dont understand)
• Inherent value of autonomy/liberty

8
Proposed solution light paternalism

• Blanket term intended to encompass
• Libertarian paternalism (Thaler Sunstein,
  2003)
• Point out that paternalism is often unavoidable
  why not make decisions that are best for people?
• Asymmetric paternalism (Camerer, Issacharoff,
  Loewenstein, ODonoghue Rabin, 2003)
• Argue that its often possible to implement
  policies that improve welfare of those behaving
  suboptimally without limiting freedom of those
  behaving optimally

9

• Common theme using errors and biases that
  ordinarily hurt people to, instead, help them

10
An example defaults

• Substantial evidence that defaults matter e.g.,
  organ donations, savings behavior, investment
  allocations

11
Status quo bias and defaults in organ donation
(Johnson-Goldstein Sci 03)
12

• Defaults fit the definition of both libertarian
  and asymmetric paternalism
• No avoiding a policy on defaults
• Defaults (e.g., on amount of salary allocated to
  401k) are almost unavoidable (and lack of default
  itself involves a policy decision)
• Asymmetric in two respects
• Desirable defaults can improve the welfare of
  those who mindlessly adhere to the default
  without restricting the options of those who do
  not.
• Carefully selected defaults should be
  unobjectionable to those who dont believe there
  is a status quo bias but should be embraced by
  those who believe that there is such an effect.

13

• Changes introduced into benefits package of large
  corporation..
• NEW Cohort defaults (all could be changed easily
  by employee)
• Automatic enrollment
• 3 of salary allocated to 401k
• 100 of allocation to money market
• WINDOW cohort
• Immediately eligible for 401k
• No automatic enrollment or allocation default

14
Enrollment
15


16
Light paternalism has diverse ramifications

• Need for new welfare criterion
• Encourage a focus on process
• Need for expanded field research
• Implementation issues channeling existing
  economic interests

17
1. Need for alternative welfare criterion

• Standard preference-based welfare criterion
  assumes that people naturally choose whats best
  for themselves welfare measured by the degree to
  which individual preferences are satisfied
• But premise of light paternalism is that people
  cant be relied upon to choose whats best for
  themselves
• ? need for alternative welfare criterion to
  evaluate success of paternalistic interventions

18
Possible welfare criteria

• Experience utility (back to Bentham!)
• Problem adaptation
• Limiting utility to valid choices (Bernheim
  Rangel, forthcoming)
• Informed decision utility
• Problems
• Informing often difficult
• Information often not the issue
• Preponderance of preferences

19
Big problem with experience utility adaptation

• Wortman and Silver (1987) quadriplegics reported
  no greater frequency of negative affect than
  control respondents!
• Tyc (1992) no difference in quality of life or
  psychiatric symptomatology in young patients who
  had lost limbs to cancer compared with those who
  had not.
• Experience utility would place no negative value
  on quadriplegia or limb-loss!
• Also, our research shows that people with
  disabilities who report being perfectly happy
  also say that they would give up substantial
  money or life-expectancy to re-attain good
  health.

20
Possible welfare criteria

• Experience utility (back to Bentham!)
• Problem adaptation
• Limiting utility to valid choices (Bernheim
  Rangel, forthcoming)
• Informed decision utility
• Problem Informing often difficult if not
  impossible
• Preponderance of preferences

21
2. Importance of process

• Decision biases often contribute to suboptimal
  behavior e.g.,
• Hyperbolic time discounting ? inadequate saving,
  overeating, procrastination, poor medication
  adherence
• Status quo bias ? inadequate retirement saving
• Loss aversion ? poor investment decisions
• Overweighting of small probabilities ? burning
  money by playing the lottery
• Best light paternalistic programs use same biases
  to promote well-being

22
Example Save More (AR) Tomorrow plan(Benartzi
Thaler)

• Inadequate saving due in part to
• hyperbolic time discounting (which leads to
  overweighting of present gratifications)
• underweighting of opportunity costs relative to
  out of pocket costs (because putting money aside
  is seen as an out of pocket cost)
• Status quo bias (because default on most 401k
  plans is zero)
• SMarT Plan
• Employees agree to increase 401k put-aside next
  year
• Financed out of wage increase (which is reduced)
• Uses all three biases to promote saving
• Hyperbolic time discounting ? willingness to save
  tomorrow
• Underweighting of opportunity costs ? willingness
  to save out of salary raises
• Status-quo effect ? continued adherence once
  implemented
• ? Highly effective!

23
First implementation of SMarT plan
24
3. Need for expanded field research

• Paternalistic policies often have unintended
  consequences
• Different biases come into play
• People may have had good reasons for behaving as
  they do
• Social/economic interactions produce unexpected
  consequences
• ? Need for field experiments

25


26
Asset allocation
27
4. Channeling existing economic interests

• Many economic interests stand to gain from
  individual mistakes
• credit-card companies overspending
• Fast-food companies over-eating
• Diet industry obesity
• Banks overdrafts, etc.
• Realtors/mortgage companies overextension
• Hospitals pharmaceutical companies poor
  preventive care
• States lottery play
• Casinos, Pornographers, etc. etc.

28
Economist as therapist needs to identify
commercial/government interests aligned with
those of individuals

• Health insurers, Veterans administration
  preventive health
• Drug companies medical adherence
• Investment companies adequate saving

29
Some projects..

• Warfarin adherence
• Weight loss
• Promoting saving

30
Warfarin Adherence(project with Kevin Volpp
Stephen Kimmel)

• Illustrates..
• Importance of process
• Value of field experiments
• Alignment of commercial and individual interests

31
Warfarin (Coumadin)Anti-stroke medication

• Benefits are unambiguous (welfare evaluation is
  trivial)
• Chance of stroke / past stroke 21
• Warfarin (if taken correctly) ? 3
• However, only 66 adhere, even in best conditions
  (warfarin clinic)
• For every 10 increase in missed doses, 14
  increase in odds of under-coagulation
• Patients with more erratic doses have higher risk
  of both under- and over-coagulation
• Our goal increase adherence through incentives

32
Importance of process..
Insight 1 need to deliver frequent feedback and
rewards ideally at the daily level Insight 2
lotteries give more bang for the buck (in part
because people overweight small
probabilities) Insight 3 ideal lottery gives
frequent positive feedback plus the hope of big
payout (because people are both forward- and
backward-looking) Insight 4 play on regret...
33
Our solution..

• Patient gets 2 digit number (e.g., 27)
• Every evening we draw a two digit number
• If first two digits match (e.g., 25) or second
  two digits match (e.g., 57), they get 10
• If both digits match (27), they get 100
• But.....
• ONLY IF THEY TOOK THEIR WARFARIN
• Message transmitted to subject whether they won
  or would have won (if didnt take medication)

34
How do we know if they took their warfarin?
35
Planned study

• 4 conditions
• 2 (reminder/no reminder) x 2 (lottery
  incentive/no lottery incentive) factorial design
• Outcome meaures
• Primary proportion of time out of therapeutic
  INR range
• Secondary rates of thromboembolic events,
  bleeding, adherence, costs
• Who can fund? Aetna (health insurance company)

36
But, first need to prove it works..

• Pilot Project 1
• 10 subjects for one month
• reminder incentive condition only (controls
  same patients before intervention, and other
  patients)
• got the incentives wrong (5 EV/day)
• Pilot project 2 (3 EV/day)

37
Rates of non-adherence
38
Weight Loss Study

• All Participants
• Attempt to lose 1 pound per week for 16 weeks
• Asked to return to lab each month for weigh-in
• 3 conditions
• Control
• Experimental (2 conditions)
• Both conditions
• Daily weight-monitoring Ps asked to phone in
  weight every day
• Feedback Sent daily text message
• E1 Lottery incentives
• E2 Deposit contracts

39
Incentive conditions

• Lottery
• Same lottery scheme as for warfarin
• And, similarly, participants only receive lottery
  payment if.. a) called in weight that day and b)
  are at or below daily weight loss goal
• Message transmitted to Ps via pager whether they
  won or would have won
• Deposit Contract
• Ps can put money down toward weight loss at
  beginning of month
• All deposits matched 11
• If P is at or below weight loss goal at end of
  month, he gets the deposit contract back
• If not, he forfeits the

40
First month
1-way ANOVA F(2,38) 19.96, p lt 0.0005
41
First month
?2 (2) 25.61, p lt 0.0005
42
Results to date(4 months into study)
43
Results to date(4 months into study)
?2 (2) 29.07, p lt 0.0005
44
Saving(in progress need partner!)

• The problem (welfare criterion)
• Out of 122 million working Americans, 42 Million
  Save through Defined contribution plans
• Median family owns zero stocks, even in
  retirement plans
• Average U.S. savings rate -1
• Median Net worth, excluding home equity (year
  2000)
• All U.S. households 13,473
• Households 65 23,369

45

• Usual Assumption People dont care about
  retirement
• ? Try to make retirement more vivid (either
  positively or negatively)

46
Alternative approach lottery-based rewards

• Participants set a yearly saving goal (e.g.,
  1,000)
• Make weekly deposits (e.g., 20)
• Every week they are on or above target,
  participate in a lottery (at the time of making
  their deposit)
• Plays on hyperbolic time discounting, goal
  gradients, peanuts effect

47
Conclusion

• Public policy is currently the most important
  application of behavioral economics
• Possible to design light paternalistic policies
  that help people without reducing their autonomy
• Many of the most successful interventions use the
  same biases that typically hurt people to,
  instead, help them