Financial Literacy and Decision Making over the Lifespan Ye Li , - - PowerPoint PPT Presentation

Ye Li, Martine Baldassi, Eric J. Johnson, Elke U. Weber

Columbia University

Center for Decision Sciences

Financial Literacy and Decision Making

• ver the Lifespan

Seniors in charge

2

Is older actually wiser?

3

Why aren’t the older worse off? The older I grow, the more I distrust the familiar doctrine that age brings wisdom.

• H.L. Mencken (1922)

Decisions about Health and Wealth

• Decumulation of assets is a complex dynamic

programming problem. –Inputs include beliefs about longevity, market returns and inflation, future costs including health costs. –Annuitization involves similar complexities –Choice of health care insurance, both prescription drugs, and primary insurance

• A myriad of housing and health decisions.
• How can we structure these decisions to maximize

abilities?

4

A concern for financial services firms.

• Older American

(65+)s hold much in the way of assets: 34% of $53.1 trillion = $18.1 trillion (2007 Survey of Consumer Finances)

• But there is liability

5

Decisions relevant to Public Policy (Appelt et al. 2012)

• SSA retirement benefits are a primary source of income for
• ver 50% of older Americans (SSA, 2010)
• 40-50% of Americans claim benefits as soon as they are

eligible (Muldoon & Kopcke, 2008; Song & Manchester, 2007)

• 22% of consumers first think about the retirement decision
• nly a year before they retire, and another 22% first think

about it only six months before (EBRI 2008).

6

Changes with age…

Fluid intelligence declines with age

Salthouse, 2010

Fluid intelligence

Fluid intelligence (Gf) is the ability to generate and transform information on the fly

• Seems critical for decision making!

Decision- making performance

8

• Older adults are worse for some decisions…
• Framing (Finucane et al 2005; Kim et al 2005)
• Applying decision rules (Bruine de Bruin et al 2007)
• Overconfidence (Crawford & Stankov 1996)
• Risk Aversion (Dohmen et al 2011)

…but sometimes no different from young…

• Framing (Mayhorn et al 2002; Roennlund et al 2005)
• Iowa Gambling Task and endowment effect

(Kovalchik et al 2004)

…and sometimes older adults are better!

• Sunk-cost fallacy (Strough et al 2008)
• Attraction effect (Kim & Hasher 2005)

Changes with age…

Yet, mixed results on decision making

Fluid intelligence

Decision- making performance

9

Changes with age…

Perhaps experience is compensating

Fluid intelligence

Salthouse, 2010

Crystallized intelligence

Crystallized intelligence (Gc) is a stable depository

• f knowledge acquired through culture, education,

and life experience (Carroll, 1993; Cattell, 1971, 1987)

Decision performance

10

aGf × bGf < 0 Negative mediation (Age is a detriment) aGc × bGc > 0 Positive mediation (Age is a benefit) Compensating Cognitive Capabilities Hypothesis Age

Mediator Gc +aGc +bGc c

• aGf

+bGf Mediator Gf / c‘ (direct effect)

Indirect effect via Gc Indirect effect via Gf Observed effect

• Financial Literacy (Lusardi & Mitchell, 2007)
• Ability to understand financial information and decisions
• Debt Literacy (Lusardi & Tufano, 2009)
• Ability to understand debt contracts and interest rates
• Temporal Discounting
• Degree to which people discount future gains and losses
• Related to health (BMI, smoking) and wealth (savings, mortgage choice)
• Loss Aversion
• Degree to which valuations of losses outweigh gains of the same magnitude
• Related to mental accounting, numerous investing mistakes, default effects

11

Decision tasks with real-world economic consequences

• FL1. Imagine that the interest rate on your savings account was 1% per year

and inflation was 2% per year. After 1 year, would you be able to buy more than, exactly the same as, or less than today with the money in this account? – More than today – Exactly the same as today – Less than today – Do not know

• FL2. Do you think that buying a single company stock usually provides a

return that is more safe, equally safe, or less safe than the return on a stock mutual fund? – More safe return than a stock mutual fund – Equally safe return as a stock mutual fund – Less safe return than a stock mutual fund – Do not know

• FL3. Is using money in a bank savings account to pay off credit card debt

usually a good or a bad idea? – Good idea – Bad idea – Do not know

• DL1. Suppose you owe $1000 on your credit card and the interest rate

you are charged is 20% per year compounded annually. If you didn't pay anything off, at this interest rate, how many years would it take for the amount you owe to double? – 2 years – Less than 5 years – More than 5 but less than 10 years – More than 10 years – Do not know

• DL2. You owe $3,000 on your credit card. You pay a minimum payment
• f $30 each month. At an Annual Percentage Rate of 12% (or 1% per

month), how many years would it take to eliminate your credit card debt if you made no additional new charges? – Less than 5 years – Between 5 and 10 years – Between 10 and 15 years – Never, you will continue to be in debt – Do not know

• DL3. You purchase an appliance which costs $1,000. To pay for this

appliance, you are given the following two options: a) pay 12 monthly installments of $100 each, b) borrow at a 20% annual interest rate and pay back $1,200 a year from now. Which is the more advantageous

• ffer?

– Option (A) – Option (B) – They are the same – Do not know

• Raven’s Progressive Matrices (Raven, 1936; Salthouse, 2005)
• Letter Series (adapted from Salthouse, 2005)
• Number Series (McArdle & Woodcock, 2009) (HRS)
• Numeracy (Lipkus et al., 2001)
• Cognitive Reflection Test [CRT] (Frederick et al., 2002)

Crystallized intelligence (Gc)

12

Fluid intelligence (Gf) Inhibitory control (IC)

• Stroop (Stroop, 1935) (Lumosity Labs)
• Flanker (Eriksen & Eriksen, 1974) (Lumosity Labs)
• Spatial 1-Back (Del Messier et al., 2010) (Lumosity Labs)
• Shipley’s Vocabulary (Shipley, 1986; adapted from CREATE)
• Antonym Vocabulary (Salthouse, 1993)
• WAIS – Information (Wechsler, 1997; adapted from CREATE)

Cognitive measures

13

• 336 participants
• Center for Decision Sciences Virtual Lab database
• No middle-age (30-59)
• Four-wave online study (last wave 1 year later)
• Same participants completed all four waves of the study
• Dropout rates do not differ between young and old
• No demographic, cognitive or decision-making variable predicts

dropout

Young (N = 173) Old (N = 163) 18 – 29 years 60 – 82 years mean = 24.8 years mean = 66.4 years 67% female 64% female

Sample characteristics

Time 1 Time 2 (12 months later)

*** p < .01 ** p < .05 * p < .10

FL Q1 FL Q2 FL Q3 Financial Literacy .81*** .76*** .51*** Debt Literacy DL Q1 DL Q2 DL Q3 .71*** .60*** .67*** Lambda $20 (short) Lambda $6a Lambda $20a Lambda $6a Lambda $20b Loss Aversion .63*** .90*** .62*** .89*** .87*** Discount $100, 12 months Discount $60, 4 months Discount $75, 3 months Discount $55, 3 months Discount $115, 3 months Temporal Discounting .48*** .64*** .55*** .77*** .72*** .11* .07 .39*** .25*** .39*** .75***

Decision-making traits

14

f

Letters Raven Numeracy CRT Numbers Gf .71*** .67*** .66*** .54*** .59*** Info Synonym Antonym Gc .79*** .80*** .67*** .73*** .34*** .17 .79*** .81*** .80*** Inhibitory Control Stroop Flanker Spatial 1-back .18*** .12**

• .003

Gender (male) Income

• .07

.05

• .02

Education .33*** .15**

• .02

*** p < .01 ** p < .05 * p < .10

Cognitive capabilities

15

f

Male Income Education Age Debt Literacy Gc

Inhib. control

Gf Loss Aversion Temporal Discounting

Financial Literacy

Structural Equation Model (SEM)

f

16

1 2 3

• 1.5
• 1
• 0.5

0.5 1 1.5 Young Old Z-Score

17

Fluid intelligence (Gf) Inhibitory control (IC) Crystallized intelligence (Gc)

Salthouse 2004, 2010

Cognitive capabilities by age

18

† p < .15; * p < .10; **p < .05; ***p < .01

Decision performance by age

*** ** *

†

• 1
• 0.75
• 0.5
• 0.25

0.25 0.5 0.75 1 Young Old Z-Score Financial Literacy Debt Literacy Discounting Loss Aversion

• Older are slightly more patient
• Older are slightly less loss averse
• Older are more financially literate
• Older are more debt literate

Financial Literacy

Age Financial Literacy Gc

ctotal = .41*** cdirect = .65*** agc = .41*** bgc = .59***

Inhib. control

aIC = -.83*** bIC = .31* abgc = .24***

Gf

agf = -.40*** bgf = .57*** abIC = -.26* abgf = -.23***

Male

mδ = .02

Education Income

Eδ = .01 Iδ = .12* *** p < .01 ** p < .05 * p < .10

19

Debt Literacy

Age Debt literacy Gc

ctotal = .19*** cdirect = .53*** agc = .41*** bgc = .58***

Inhib. control

aIC = -.83*** bIC = .32* abgc = .24***

Gf

agf = -.40*** bgf = .79** abIC = -.26* abgf = -.32***

Male

mδ = .21**

Education Income

Eδ = -.13 Iδ = .03 *** p < .01 ** p < .05 * p < .10

20

Age Temporal discounting (more patient) Gc

ctotal = .08 cdirect = .11 agc = .41*** bgc = .20***

Inhib. control

aIC = -.83*** bIC = .04 abgc = .08***

Gf

agf = -.40*** bgf = .20** abIC = -.03 abgf = -.08** *** p < .01 ** p < .05 * p < .10

Male

mδ = .04

Education Income

Eδ = .06 Iδ = .13*

Temporal Discounting

21

Age (less) Loss aversion

ctotal = .06 cdirect = -.12 bgc = -.08 bIC = -.25* abgc = -.03

Gc

Inhib. control

Gf

agc = .41*** aIC = -.83*** agf = -.40*** bgf = -.02 abIC = .20* abgf = .01

Male

mλ = .25***

Education Income

Eλ = .09 Iλ = .09 *** p < .01 ** p < .05 * p < .10

Loss Aversion

22

Decision performance

23

Summary Age

Mediator Gc +aGc +bGc c

• aGf

+bGf Mediator Gf / c‘ (direct effect)

*** ** *

†

• 1
• 0.75
• 0.5
• 0.25

0.25 0.5 0.75 1 Young Old Z-Score Financial Literacy Debt Literacy Discounting Loss Aversion

• 1.5
• 1
• 0.5

0.5 1 1.5 Young Old Z-Score

Fluid intelligence (Gf) Inhibitory control (IC) Crystallized intelligence (Gc)

• Our older group was 66 on average...

What happens for even older?

• We didn’t collected 30-59…

Are middle age even better?

24

What happens over the full life course?

Salthouse, 2010

• 2
• 1.5
• 1
• 0.5

0.5 1 20 30 40 50 60 70 80 Patience Financial Literacy Debt Literacy

New Results (Note….Preliminary)

• Anonymous credit reports from a major credit reporting firm

– 338 total data fields – Two pulls: February 2009 & January 2012 (r = .80) – Mean FICO score increased by 13pts (p < .001)

• Matches for 456/632 (72%) original participants!

– 245/332 full data points (74%)

• 231/332 in both time periods

– Higher match rate for old (136/16383%) than young (100/17358%)

Credit Report Fields

Age FICO

cobserved = .33*** cresidual = .38***

Crystallized Intelligence

bGc = .18* abGc = .07*

Fluid Intelligence

bGf = .31*** abGf = -.12***

FICO Mediation

27

Male

• .10

Education Income

.06 .13** aGc = .41*** aGf = -.40***

Total Indirect Effect = - .05

Predicting how long you will live (Payne et al, 2012)

• Fundamental Assumption of Economics Models.
• Can be asked two ways:

– What is the probability that you will live to be age 85 or

• lder?

– What is the probability that you will die by age 85 or younger?

• These should be the related p(alive at 85) = 1 – p(die by 85)
• Are they influenced by how we ask?

28

29

Implications

• How to accommodate elders?

– Financial Education can help Crystalized Intelligence. – But Not Fluid Intelligence. This is a challenge

• Better Choice Architecture.
• Lessons from Medicare Part D

– Should there be ‘dozens of options?” – Can People Opt-In to a supported environment

• Next Steps: Actual Behavior…Credit Reports of

participants.

30

Other Conversations

• Choice Architecture in general, no-action defaults,

specifically

• Applications

– Decumulation

• Annuitization

– What are the behavioral barriers? – How do people estimate longevity?

• Claiming Age for Social Security

– Design of Health Care Exchanges.

31

32

Thanks!

• National Institute of Aging for funding
• Society of Judgment and Decision Making meeting participants
• Columbia’s Cognitive Neuroscience Seminar participants
• Preferences-as-Memories Lab members:

Kirstin Appelt, Dan Bartels, Isaac Dinner, Bernd Figner, Dave Hardisty, Maria Konnikova, Jing Qian, Eric Schoenberg, Katherine Thompson, Liza Zaval

• Special thanks to Jon Westfall for technical assistance
• Mark Heitman for SEM assistance
• Tim Salthouse and Yaakov Stern for feedback

Acknowledgements

33

Appendix

34

• Gc predicts better financial and debt literacy,

greater patience in temporal discounting

• Does general Gc stand in for more specific

knowledge or experience?

• Or is it related to some other age-dependent

variable?

• Socioemotional Selectivity Theory (Carstensen, 2006)

How does crystallized intelligence compensate?

Loss Aversion titrator

35

Back

36

More information on our sample, and drop out (completion rates), screening procedures

Internet Availability in the US

37

• > 80% of Americans have a computer at home, almost

92% of those have internet access

Nielsen Claritas Convergence Audit, 2008

88% 87% 81% 76% 60% 78% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 18-34 35-44 45-54 55-64 65+ Total

Dial up/High Speed Internet

61% 75% 82% 87% 77% 78% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Dial up/High Speed Internet

Back

Internet Availability (Cont.)

38

Item Have Access to Internet at Home, Work, or Elsewhere Have Used the Internet in the last 30 days % of Total Population 85% 71%

• Male

48% of those… 48% of those… Age 18 – 34 33% 36% 35 – 54 41% 43% 55+ 26% 21% Income < $50,000 36% 31% $50,000 - $74,999 21% 21% $75,000 - $149,999 31% 34% $150,000+ 12% 13%

MediaMark Research & Intelligence, 2008

Back

Comparison of Sample Characteristics with US population

39

Comparison of Sample Characteristics with US population

Social Demographic

GRAD young

(18-30)

US Comparison young GRAD old

(60-82)

US Comparison old

Gender a Male 33% 51% 36% 44% Female 67% 49% 64% 56% Education b HS Degree or Less 33% 47% 26% 55% Some College/AA 12% 44% 15% 24% Bachelors Degree + 55% 9% 59% 23% Race c Caucasian 65% 61% 93% 78% African American 8% 14% 4% 8% Asian 15% 4% 0% 4% Hispanic 2% 18% 1% 8% American Indian/Alaskan 1% 1% 0% 1% Other 9% 2% 2% 1%

a b d Data Set: 2009 American Community Survey 1-Year Estimates

(a Young: 20-34, b Young: 18-24, d Young: 20-34, a b d Old: 60+)

c US Census Monthly Estimates by Age, Sex, and Race—July 1, 2009; young = 18-30, old = 60-85

Back

Completion rates: 632 participants (Nyoung=332, Nold=300) completed the 1st wave 562 participants (Nyoung=296, Nold=266) completed the 2nd wave 516 participants (Nyoung=274, Nold=242) completed the 3rd wave 336 participants (Nyoung=173, Nold=163) completed the 4th wave Overall dropout rate was 46.84% (Nyoung=159 [47.9%], Nold=137 [45.7%] )

• Low considering the length of time between waves (Musch & Reips, 2000)

IMPORTANT

• Dropout rates do not differ between young and old.
• No demographic, cognitive or decision-making variable predicts dropout

Details on completion and drop out rates

40

Back

41

More information on the measurement model for the cognitive variables

• Measurement Models:

– Cognitive Competencies – Decision-making Performance

• Overall SEM (structural equation model) combining cognitive and

decision-making factors

42

Measure of fit Cognitive Factors Decision- making Factors Overall SEM Overall SEM w/ Demographic Controls χ2 Test of Model Fit 117.2 141.7 566.8 601.0 Degrees of Freedom 40 95 347 410 χ2 /df 2.93 1.49 1.63 1.47 RMSEA (Root Mean Square

Error Of Approximation)

0.08 0.04 0.04 0.04 CFI (Comparative Fit Index) 0.94 0.95 0.90 0.91 TLI (Tucker Lewis Index) 0.92 0.93 0.88 0.89

Model fits

Back

Measurement model (Cognitive measures)

43

Info Synonym Antonym Letters Raven Numeracy CRT Numbers Gf Gc .790*** .560*** .535*** .594*** .618*** .780*** .799*** .625*** .795*** .792*** .830*** .703*** .336*** .106 Inhibitory Control Response Speed .437*** .278*** .159** .978*** .630*** Choice RT Simple RT Stroop Flanker Spatial 1-back

Anchoring (bad factor structure) Age Anchoring z-score Gc

ctotal = .454* cdirect = .567 agc = .465*** bgc = -.080

Inhib. control

aIC = -.829*** bIC = .175 abgc = -.033

Gf

agf = -.367*** bgf = .096 abIC = .145 abgf = -.038

Male

man = .003

Education Income

Ean = .238*** Ian = .105 *** p < .01 ** p < .05 * p < .10