Aging and Financial Decision Making

Abstract

This study examines how cognitive changes associated with aging impact the financial decision making capability of older Americans. We find that a decrease in cognition is associated with a decrease in financial literacy. Decreases in episodic memory and visuospatial ability are associated with a decrease in numeracy, and a decrease in semantic memory is associated with a decrease in financial knowledge. A decrease in cognition also predicts a drop in self-confidence in general, but importantly, it is not associated with a drop in confidence in managing one's own finances. Participants experiencing decreases in cognition do show an increased likelihood of getting help with financial decisions; however, many participants experiencing significant drops in cognition still do not get help.

1. Introduction

There are concerns that financial decision making in the older population is compromised by the presence of cognitive decline in advanced age. These concerns combined with demographic changes resulting in an increasingly larger older population have sparked several recent studies on aging and financial decision making.¹ Recent studies find that older individuals are prone to worse financial decision making. For example, Korniotis and Kumar (2011) find that older investors exhibit less investment skill, and Agarwal et al. (2010) find that suboptimal credit behavior increases past age 53. Finke, Howe, and Huston (2011) suspect that financial decision making ability declines with age as financial literacy declines; indeed, they show that average financial literacy scores are lower by about 1% for each year after age 60. These existing studies indirectly examine the effects of cognitive aging on financial ability by comparing across individuals of different ages. Such comparisons confound the effect of cognitive decline with other differences, such as cohort effects. For example, Malmendier and Nagel (2011) find the cohort effect of early-life economic conditions on risk taking decades later. Direct measures of cognition collected repeatedly from the same individuals are needed in order to identify the effect of a decrease in cognition on financial ability.

We use longitudinal data from the Rush Memory and Aging Project, a large cohort study of aging, to identify, within individuals, the impact of decreases in cognition on financial literacy, financial confidence, and self-participation in financial decisions. In analyses restricted to persons without dementia based on a detailed clinical evaluation, we find that a decrease in cognition is a significant predictor of a decrease in financial literacy among older Americans. Drops in cognition are associated with decreases in each of the components of financial literacy we measure, both numeracy and financial knowledge. We use our rich cognitive assessment including measures of five domains of cognition to identify which components of cognition are driving these changes in financial literacy. We find that decreases in episodic memory and visuospatial ability are associated with a decrease in numeracy while a decrease in semantic memory is associated with a decrease in financial knowledge.

Although participants experiencing decreased cognition also show declines in their financial literacy, these participants may not recognize or may be reluctant to admit to this decline in their financial capability. We find that a decrease in cognition predicts a drop in self-confidence in general, but importantly, it does not predict a decrease in confidence in managing one's own finances nor a decrease in confidence in one's financial knowledge. Similarly, Holland and Rabbitt (1992) find that individuals in their 70s do not rate their sensory abilities as poor any more so than individuals in their 50s despite significant declines in their measured ability. Importantly, they find that those older individuals who recognize their decline in sensory ability adjust their road-use behavior and have fewer accidents. Perhaps there is an analogy between driving and financial choices, and older Americans who have a drop in cognition would be more likely to take precautions in their financial decision making if made aware of the connection.

The detrimental effects of cognitive aging on the financial choices of older Americans can potentially be mitigated with help for financial decisions provided within or outside of the household. We find that individuals who experience a decrease in cognition are more likely to stop managing their own finances and pass on this responsibility to their spouse, and they are more likely to get financial help from outside their household. However, there are still many participants who are experiencing cognitive decreases who are not getting help with their financial decisions. Even among the participants experiencing statistically significant decreases in cognition, about half are not getting help with their financial decisions. While these participants are likely to benefit from trustworthy, knowledgeable advice, knowing who to trust in financial matters can be problematic.

2. Data Description and Construction of Measures

Our data come from the Rush Memory and Aging Project (MAP), an ongoing longitudinal study of aging (Bennett et al. (2012)). Since beginning in 1997, MAP has enrolled older participants from throughout the Chicago metropolitan area. Participants undergo yearly interviews and detailed clinical evaluations, including medical history, neurological, and neuropsychological examinations. The MAP data include demographic information for each participant, such as age, sex, and education. In 2010, a decision making assessment was added to MAP. The Institutional Review Board of Rush University Medical Center approved MAP and the decision making substudy.

We exclude data from the 71 participants who were diagnosed with dementia at the time of their first decision making assessment. For these participants even completing the decision making assessment is rare; only 19 of these participants provided answers to each of our outcomes of interest. Our conclusions are robust to including these participants, but we exclude them to avoid any selection bias due to the participants who could not complete the survey. Dementia is diagnosed in accordance with the standards set by the National Institute of Neurologic and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (Bennett et al. (2005)). At the time of these analyses, 575 participants without dementia at the initial decision making assessment had completed at least two decision making assessments. Two decision making assessments are required to observe increases and decreases in decision making measures over time.

Since its beginning, MAP has collected yearly cognitive test scores for each participant. Cognition is assessed with 19 tests, which are listed in the appendix by the 5 cognitive domains assessed in the battery: episodic memory, perceptual speed, semantic memory, visuospatial ability, and working memory. Episodic memory captures the memory of specific events whereas semantic memory captures the knowledge of concepts. Working memory captures the ability to store and process transitory information. Perceptual speed involves the ability to process information quickly and make mental comparisons. Visuospatial ability involves understanding visual representations and the spatial relationships among objects. The raw scores of each of the 19 cognitive tests are converted to z-scores using the baseline mean and standard deviation of the entire MAP cohort on that test. These 19 z-scores are averaged to compute the global cognitive function score. The z-scores within each domain are averaged to compute each cognitive domain score. We calculate each participant's change in cognition from the first decision making assessment to the most recent decision making assessment.

We connect each participant's change in cognition to the concurrent change in a variety of measures related to financial decision making capability. The exact wording of each decision making question used in this study is provided in the appendix. The decision making questionnaire includes 16 standard financial literacy questions including 9 to test numeracy and 7 to test financial knowledge.² We measure financial literacy, numeracy, and knowledge by adding the number of correct answers in each category of questions. Participants may respond that they do not know the answer, and they can refuse to answer any question. Participants are aware of these possibilities: 26% of participants refused or said do not know to at least one financial literacy question during the decision making study. Typically, these participants refused or said do not know only rarely. Among the participants who used these options at least once, the average number is 1.65 refusals or do-not-knows per survey. These responses are treated the same as incorrect answers in this analysis.

Each financial knowledge question includes a follow up question immediately after to assess the participant's confidence in her answer to the preceding knowledge question using a four-point scale from extremely confident to not at all confident. We score an extremely confident answer as a 3, fairly confident as a 2, a little confident as a 1, and not at all confident as a 0. We measure each participant’s confidence in her financial knowledge by summing the confidence scores to these 9 questions.

We also use two additional measures of confidence. We assess self-confidence using a single question that asks participants to report their general level of confidence on a ten-point scale with 1 indicating that they are not at all confident and 10 indicating that they are completely confident. We assess financial confidence with a single question that asks participants to report to what extent they agree with the statement: "I am good at managing day to day financial matters such as keeping up with checking accounts, credit cards, payments, and budgeting." Responses are reported on a seven-point scale from strongly agree indicating the highest level of financial confidence (6) to strongly disagree indicating the lowest level of financial confidence (0).

Participants are also asked who are primarily responsible for making their financial decisions. They are asked explicitly if they, their spouse, their child, or someone else is responsible, and they are asked to specify the relationship for a response that includes someone else. Thus, we can identify participants who make their own financial decisions, households who make their own financial decisions (participant or spouse), participants that get help with financial decisions (spouse or other person is specified, possibly in addition to self), and participants that get help from outside of the household (someone other than the participant or spouse is included as primarily responsible).

3. Cognition Change

3.1. Procedure for Cognition Change Sample

We use simple regressions to identify the effect of a change in cognition on these decision making variables. Each regression is of the following form:

$$\mathrm{\Delta }{y}_i=a*\mathrm{\Delta }{\mathit{\text{Cognition}}}_i+b+{\epsilon }_i$$

In each regression the dependent variable is the change in the decision making variable (y) from the participant i’s first decision making assessment to her most recent (Δy_i). When this dependent variable is binary, we use the logistic form for the regression.

The right-hand side includes participant i’s change in cognitive function score (ΔCognition_i) and a constant. The coefficient of the first term (a) captures the effect of a one unit change in cognitive score. The error term is ε_i. The coefficient a captures the effect of both increases and decreases in cognition collectively. Since the focus of this study is on understanding the impact of decreases in cognition on financial decision making, we also run the previous regression using only the subset of participants who experience a decrease in cognition. For this subset the coefficient a only captures the association of decreases in cognition with the dependent variable.

3.2. Summary Statistics for Cognition Change Sample

Table 1 presents summary statistics for the 575 participants in the cognitive change sample. They are mostly female, well-educated, older Americans. The average age is 82.23 years, and only 23% are male. The participants average 15.11 years of education. About two-thirds (377 participants) of the sample experience a decrease in their global cognition z-score from their first decision making assessment to the most recent. The average decrease in measured cognition among this group is −0.29. Many participants (34%) increase their cognition score. The same questions are repeated each year, and participants benefit from the effect of practice. The average increase is smaller in size at 0.19.

Table 1.

Summary Statistics for Cognition Change Sample

	All Participants		Decrease in Cognition
Participants	575		377
Male	23%		22%
Age	82.23 (7.36)		83.17 (7.19)
Education	15.11 (2.86)		15.21 (2.93)
	Initial Level	Change	Initial Level	Change
Cognition	0.22 (0.54)	−0.13 (0.35)	0.22 (0.55)	−0.29 (0.30)
Financial Literacy	11.20 (2.30)	−0.15 (2.10)	11.08 (2.36)	−0.34 (2.21)
Numeracy	6.32 (1.35)	−0.11 (1.54)	6.26 (1.37)	−0.24 (1.58)
Financial Knowledge	4.88 (1.47)	−0.04 (1.32)	4.82 (1.49)	−0.10 (1.40)
Self-Confidence	7.17 (1.83)	0.08 (1.94)	7.17 (1.88)	0.06 (2.04)
Confidence in Managing Finances	4.98 (1.38)	−0.03 (1.33)	4.88 (1.45)	−0.12 (1.52)
Confidence in Financial Knowledge	14.77 (4.33)	0.02 (3.56)	14.52 (4.41)	−0.19 (3.60)
Participant Makes Financial Decisions	88%	−13%	87%	−16%
Household Makes Financial Decisions	91%	−10%	91%	−12%
Gets Help with Financial Decisions	41%	12%	45%	12%
Gets Help Outside of Household	25%	11%	29%	12%

This table presents summary statistics for the whole sample and the subsample of participants who experienced a decrease in cognition score during the Decision Making assessment. Age and Education are stated in years. Cognition is a z-score scaled to all participants in the Memory and Aging Project at baseline. Values are reported as means (standard deviation) or percentages.

Studies around the world find low levels of financial literacy (Lusardi and Mitchell (2011a)). Participants in this study perform similarly. Participants answer on average 11.20 of the 16 financial literacy questions correctly in their initial decision making assessment. They correctly answer the same percentage of numeracy questions and financial literacy questions on average (70%). This percentage does not change by much overall from first assessment to the most recent. Lusardi and Mitchell (2011b) analyze a three-question financial literacy module included in the 2004 Health and Retirement study. Two of their questions, one about inflation and one about compound interest, match questions used in our measure of literacy. They find that only 50% of respondents answered both the questions about inflation and compound interest correctly. In contrast, 65% of our respondents answered the same questions about inflation and compound interest both correctly.³

Participants display a high level of self-confidence: their self-confidence averages 7.17 on the 10 point scale with a 10 meaning completely confident. Confidence in managing finances is similarly high on average (4.98 out of 6), meaning that most participants agree with the statement that they are good at managing their day to day financial matters. Confidence in financial knowledge averages 14.77 out of a maximum of 21, which is a little higher than the score for a participant who indicates they are fairly confident for each question (14).

Consistent with their high confidence in their ability to manage finances and their high confidence in their financial knowledge, the vast majority of participants (88%) are primarily or jointly responsible for their financial decisions at the time of their first decision making assessment. About 41% get help with financial decisions, including from a spouse, child, or outside advisor. Just 25% get help with financial decisions from someone other than a spouse. Over time fewer participants make their own financial decisions and more get help. At time of the most recent decision making assessment, the percentage of participants making their own financial decisions dropped by 13%, and 11% more got help with financial decisions.

3.3. Cognition Changes and Literacy

In this subsection we examine the impact of decreases in cognition on financial literacy and its components (numeracy and financial knowledge). Table 2 presents results of six regressions following the form specified in the procedures section of this paper. Changes in cognition are associated with changes in financial literacy and its components. A one unit change in cognition is associated with a literacy change of 1.084, which comes from a 0.648 change in numeracy and a 0.437 change in financial knowledge. Each association is statistically significant at the 1% level.

Table 2.

Decrease in Cognition and Literacy

	Literacy Change			Numeracy Change			Knowledge Change
	Estimate	SE	p-value	Estimate	SE	p-value	Estimate	SE	p-value
Cognition Change	1.084***	0.248	0.00	0.648***	0.183	0.00	0.437***	0.158	0.01
Intercept	−0.012	0.092	0.90	−0.031	0.068	0.64	0.020	0.058	0.74
Cognition Decrease	1.237***	0.375	0.00	0.765***	0.269	0.00	0.473**	0.240	0.05
Intercept	0.021	0.157	0.89	−0.017	0.113	0.88	0.038	0.101	0.71

This table presents regression results of the effect of cognitive changes on financial literacy, and separately on its components, numeracy and knowledge. Cognition Change provides the association between participants' cognition changes and their changes in the specified outcome variable. Cognition Decrease provides the previous association only among the subset of participants who experience a decrease in cognition.

*, **, and *** indicate statistically significant at the 10%, 5%, and 1% level respectively

We rule out the possibility that the positive association between cognition changes and literacy changes in this regression could be driven by those participants with improvements in their cognition score improving their financial literacy scores as well and not from those with decreases in their cognition score getting worse on literacy. Since the impact of decreases in cognition is the focus of this study, we rule out the previous possibility by running the same regression only for the subset of participants whose global cognition score dropped. Decreases in cognition are associated with decreases in financial literacy and its components. A one unit decrease in cognition is associated with a financial literacy decrease of 1.237, which comes from a 0.765 decrease in numeracy and a 0.473 decrease in financial knowledge. The associations with literacy overall and numeracy are also statistically significant at the 1% level while the association with financial knowledge is statistically significant at the 5% level.

The size of these effects of cognitive changes on financial literacy are modest, but it is important to consider that the changes in cognition we are measuring during the decision making assessment period occurs over just two to three years. Individuals experiencing cognitive decreases are likely to experience further decreases over time. Thus, the impact of decreases in cognition on financial literacy is expected to accumulate over time.

3.4. Breakdown by Cognitive Domain

The association of drops in cognition with drops in financial literacy and its components can be separated into the five domains of cognition tested. Table 3 presents summary statistics for these five domain-specific cognitive measures. As with the global cognition score, participants’ average score has dropped over time. The changes in these domain-specific cognitive measures are positively correlated; however, there is a lot of independent variation in these measures. The correlations range from a low of 6% between visuospatial ability and working memory to a high of 34% between episodic memory and semantic memory.

Table 3.

The Components of Global Cognition

	Decrease in Cognition
	Initial Level	Change
Episodic Memory	0.32 (0.74)	−0.30 (0.42)
Perceptual Speed	0.09 (0.78)	−0.31 (0.46)
Semantic Memory	0.22 (0.62)	−0.21 (0.39)
Visuospatial Ability	0.24 (0.74)	−0.26 (0.61)
Working Memory	0.15 (0.74)	−0.22 (0.49)

	Numeracy Change			Knowledge Change
	Estimate	SE	p-value	Estimate	SE	p-value
Episodic Memory Decrease	0.725***	0.242	0.00	0.294	0.213	0.17
Intercept	0.173	0.130	0.18	0.024	0.114	0.83
Perceptual Speed Decrease	0.369*	0.219	0.09	0.212	0.191	0.27
Intercept	−0.022	0.122	0.86	0.013	0.107	0.90
Semantic Memory Decrease	0.404	0.300	0.17	0.632**	0.246	0.01
Intercept	0.065	0.134	0.63	0.203*	0.112	0.07
Visuospatial Ability Decrease	0.472**	0.212	0.03	0.024	0.194	0.90
Intercept	0.136	0.143	0.34	−0.060	0.131	0.65
Working Memory Decrease	0.333	0.254	0.19	0.153	0.211	0.47
Intercept	−0.017	0.145	0.68	−0.010	0.120	0.42

This table presents summary statistics and regression results for the effect of cognitive changes within each of the five domains on the components of financial literacy, numeracy and knowledge. Cognition Decrease provides the association of the change in cognitive domain score with the change in the specified component of financial literacy only among the subset of participants who experience a decrease in cognition.

*, **, and *** indicate statistically significant at the 10%, 5%, and 1% level respectively

Table 3 also presents regression results for how decreases in these cognitive domain scores are associated with changes in the components of financial literacy. Numeracy changes are most strongly predicted by a drop in episodic memory; a one unit decrease in episodic memory is associated with a 0.725 decrease in numeracy, which is statistically significant at the 1% level. Numeracy changes are also associated with visuospatial ability (p=.03). Knowledge changes are most strongly predicted by a drop in semantic memory; a one unit decrease in semantic memory is associated with a 0.632 decrease in the participant’s financial knowledge, which is statistically significant at the 5% level.

3.5. Cognition Changes and Confidence

We next examine the effect of changes in global cognition on a variety of confidence measures. First, we examine the effect of a decrease in cognition on general self-confidence. Table 4 shows that a one unit change in cognition is associated with a 0.416 change in self-confidence on a ten-point scale. This weak association in changes is driven by a strong association among the subset of those participants experiencing declining cognition. A one unit decrease in cognition is associated with a 0.968 decrease in self-confidence, which is statistically significant at the 5% level. However, we find a very different result for the effect of a decrease in cognition on one's confidence for managing financial matters. Neither changes in cognition or decreases in cognition are associated with changes in confidence in managing one’s finances. Despite the drop in self-confidence associated with a decrease in cognition, participants who have a decrease in cognition do not reduce their confidence for managing their own finances.

Table 4.

Decrease in Cognition and Confidence

	Change in Self-Confidence			Change in Confidence in Managing Finances
		SE	p-value	Estimate	SE	p-value
Cognition Change	0.416*	0.234	0.08	0.163	0.168	0.33
Intercept	0.131	0.086	0.13	−0.094	0.061	0.13
Cognition Decrease	0.968***	0.350	0.01	0.098	0.267	0.72
Intercept	0.342**	0.147	0.02	−0.092	0.110	0.40
	Change in Confidence in Financial Knowledge
	Estimate	SE	p-value
Cognition Change	1.042**	0.426	0.01
Intercept	0.153	0.157	0.33
Cognition Decrease	1.062*	0.616	0.09
Intercept	0.126	0.259	0.63

This table presents regression results of the effect of cognition changes on three different measures of confidence. Cognition Change and Cognition Decrease are defined as in Table 2.

*, **, and *** indicate statistically significant at the 10%, 5%, and 1% level respectively

Similarly, participants who experience a decrease in cognition do not significantly reduce their confidence in their financial knowledge. Although we find that a one unit change in cognition is associated with a 1.042 change in participants’ confidence in their financial knowledge with statistical significance nearly at the 1% level, this result is not driven primarily by those with a decrease in cognition. In this case those increasing their cognition score are also increasing their confidence in their financial knowledge. Among those participants experiencing a decrease in cognition, there is only weak statistical significance in the association between decreases in cognition in the change in confidence in their financial knowledge with a p-value of 0.09.

In the previous subsection of this paper, we document a statistically significant finding that financial knowledge does drop with decreases in cognitive score; thus, these participants do not appear to recognize fully the detrimental effect of decreased cognition on their financial ability despite their decrease in self-confidence in general.

3.6. Cognitive Changes and Seeking Financial Help

Having shown that decreases in cognition are strongly associated with a decrease in financial literacy but not one's financial confidence, we now examine to what extent those participants who experience a decrease in their cognitive score get help with their financial decision making. Because the dependent variables in this subsection are binary, we alter our regression to the logistic form; otherwise, the explanatory variables are the same. Table 5 presents these logistic regression results. A one unit decrease in cognition results in an increase in the odds that a participant stops making her own financial decisions by e^1.098 − 1 = 203%. This relationship is statistically significant at the 1% level. Similarly, a one unit decrease in measured cognition results in an increase in the odds that both participant and spouse (a household) stop making their own financial decisions by e^1.290 − 1 = 263% Again, this relationship is statistically significant at the 1% level.

Table 5.

Decrease in Cognition and Seeking Financial Help

	Participant Stopped			Household Stopped
	Estimate	SE	p-value	Estimate	SE	p-value
Cognition Change	−1.123***	0.323	0.00	−1.119***	0.347	0.00
Intercept	−2.113***	0.145	0.00	−2.414***	0.164	0.00
Cognition Decrease	−1.098***	0.406	0.01	−1.290***	0.432	0.00
Intercept	−2.043***	0.203	0.00	−2.462***	0.232	0.00
	Obtained Help			Obtained Help Outside Household
	Estimate	SE	p-value	Estimate	SE	p-value
Cognition Change	−0.658**	0.331	0.05	−0.787**	0.341	0.02
Intercept	−2.065***	0.142	0.00	−2.227***	0.151	0.00
Cognition Decrease	−0.864**	0.427	0.04	−0.878**	0.434	0.04
Intercept	−2.205***	0.217	0.00	−2.284***	0.223	0.00

This table presents logistic regression results of the effect of cognitive changes on participants’ participation in their own financial decisions. Cognition Change and Cognition Decrease are defined as in Table 2.

*, **, and *** indicate statistically significant at the 10%, 5%, and 1% level respectively

Participants who experience a decrease in their cognition are more likely to obtain help with making financial decisions. A one unit decrease in measured cognition results in an increase in the odds that a participant obtained help for her financial decisions by e^0.864 − 1 = 137%. This result is statistically significant at the 5% level. It includes obtaining help from a spouse as well as anyone outside the household. Similarly, a one unit decrease in measured cognition results in an increase in the odds that a participant obtained help for her financial decisions from outside her household by e^0.878 − 1 = 141%, which is statistically significant at the 5% level. Typically, help from outside the household is provided by a son, a daughter, or a professional financial advisor.

Despite the strong association between decreases in cognition and seeking help with financial decisions, there are still many participants who experience significant declines in their cognition who are not getting help. We use each participant's complete history of cognitive scores, including those prior to the start of the Decision Making assessment, to determine the long-term cognitive trajectory of each individual. The number of annual cognition scores for participants in our sample ranges from 2 for the most recent enrollees to 15 for long-time participants. On average participants have 6.6 cognitive scores with median of 7. Thus, we have a long history of cognitive function scores to determine which participants are experiencing a decline in cognition during their time in MAP. For each participant we determine the slope of her cognitive ability by running a simple linear regression of cognition scores on age and a constant. There are 146 participants who have experienced both decreased cognition during the decision making assessment and a statistically significant cognitive decline during their participation in MAP. Of these 146 participants only about half (76) get help with their financial decision making.

4. Conclusion

We utilize the data from the Rush Memory and Aging Project and the Decision Making substudy to identify the detrimental impact of decreases in cognition associated with aging on the financial decision making ability of older Americans. We find that decreases in cognition are associated with decreases in financial literacy. We provide evidence that participants do not recognize this decrease. Despite showing significant drops in their self confidence in general, their confidence in their ability to manage their own finances and their confidence in their financial knowledge do not decrease with drops in measured cognition. Whether it is sought out or unsolicited, participants who experience a decrease in their cognitive score are more likely to obtain help with their financial decisions, though perhaps not as many get assistance as need it and bad advice may be a problem.

The importance of studying financial decision making in the older population has never been greater. Prior to 1980, retirees relied on a combination of employer-sponsored defined benefit pensions and Social Security for monthly income. For these retirees institutions shouldered the responsibility and the risk of investing contributions and managing payouts. Since 1980, many defined benefits plans have been replaced by defined contribution plans, which leave the responsibility of managing investments and withdrawals to the individual retiree. Poterba, Venti, and Wise (2008) document that in 2000, 87% of personal retirement contributions went to individual accounts with the largest proportion of these going to 401(k) accounts. The next generation of retirees will have the responsibility and risk of managing the money in these individual accounts sensibly. As the baby boom generation of Americans begins to retire, there will be an ever larger portion of the population shouldering this great financial responsibility of managing their own retirement wealth.

After the massive shift from defined benefit pensions to self-directed defined contribution retirement accounts, economists documented the many heuristics and biases of these new retirement savers (Benartzi and Thaler (2007)). Research also helped to reveal solutions such as automatic enrollment and default investment portfolios (Choi et al. 2004) that have greatly increased retirement savings. As this generation of workers begins to retire, we believe that research on the financial decision making of older Americans will be equally as important in revealing the heuristics, biases, and behaviors of this new generation of retirees. This information is essential to developing the innovations that will help them to maximize their well-being during this last period of their lives when many important and influential financial decisions are made.

Appendix

Cognition Tests

The global cognition score is calculated by converting raw scores on each of the 19 tests listed below to z scores, using the mean and standard deviation from the full cohort at baseline, and then averaging the z scores to produce the composite measure. The composite measure of each cognitive domain is calculated similarly using only the tests in that domain.

Episodic Memory

1. Logical memory (immediate) Story A from the Logical Memory subset of the Wechsler Memory Scale-Revised;

2. Logical memory (delayed) Story A from the Logical Memory subset of the Wechsler Memory Scale-Revised;

3. CERAD Word list recall (immediate)

4. CERAD Word list recall (delayed)

5. CERAD Word list recognition

6. East Boston Story (immediate)

7. East Boston Story (delayed)

Perceptual Speed

1. Oral version of the Symbol Digit Modalities Test

2. Number Comparison

3. 2 indices from a modified version of the Stroop Neuropsychological Screening test

Semantic Memory

1. Verbal fluency from CERAD;

2. 15 item version of the Boston Naming Test

3. 15-item reading test

Visuospatial Ability

1. 15-item version of Judgment of Line Orientation

2. 16-item version of Standard Progressive Matrices

Working Memory

1. Digit Span subtests-forward of the Wechsler Memory Scale-Revised

2. Digit Span subtests-backward of the Wechsler Memory Scale-Revised

3. Digit Ordering

Survey Questions

Numeracy Questions

1. Which of these percentages represents the biggest risk of getting a disease? 1%, 10%, 5%

2. A store is offering 15% off a television that is normally priced at $1000. How much money would you save on the TV during this sale? $15, $150, $1500

3. If a television set is on sale for $899, which is $200 off its normal price, what is the normal price? $699, $1099, $1299

4. If 5 people all have the winning numbers in the lottery and the prize is $2 million, how much will each of them receive? $200,000; $400,000; $600,000

5. If the chance of getting a disease is 10 percent, how many people out of 1,000 would be expected to get the disease? 100, 10, 90, 900

6. In a sale, a shop is selling all items at half price. Before the sale, a sofa costs $300. How much will it cost in the sale? $150, $600, $900

7. Suppose you had $100 in a savings account and the interest rate was 2% per year. After 5 years, how much do you think you would have in the account if you left the money to grow: more than $102, exactly $102, or less than $102?

8. Again, suppose you had $100 in a savings account and the interest rate was 2% per year. After 5 years, how much do you think you would have in the account if you left the money to grow: more than $110, exactly $110, or less than $110?

9. Imagine that the interest rate on your savings account is 1% per year and inflation is 2% per year. After 1 year, will you be able to buy more than, exactly the same as, or less than today with the money in your account?

Financial Knowledge and Confidence Questions for Overconfidence Measure

Note: Each financial knowledge question is followed by the same confidence question below.

How confident are you that you answered that question correctly?

extremely confident, fairly confident, a little confident, not at all confident

1. What do the initials FDIC stand for?

2. What does the FDIC do?

   approves new drugs for clinical use, protects the funds people or depositors place in banks and savings institutions, underwrites mortgages and other loans

3. When interest rates go up, what do bond prices do: go down, go up, or stay the same?

4. True or False. An older person with $100,000 to invest should hold riskier financial investments than a younger person with $100,000 to invest.

5. True or False. Using money in a bank account to pay off credit card debt is usually wise.

6. True or False. To make money in the stock market, you have to buy and sell stocks often.

7. True or False. Stocks and mutual funds generally produce higher average returns above inflation compared to fixed-income investments such as bonds.

Self-confidence Question

Using a 1–10 point rating scale, where 1 indicates that you are not at all confident and 10 indicates that you are completely confident, how would you rate your general level of confidence?

Financial Confidence Question

Please give a number between 1 and 7 indicating the degree to which you agree or disagree with this statement, with 1 being strongly agree and 7 strongly disagree. I am good at managing day to day financial matters such as keeping up with checking accounts, credit cards, payments, and budgeting.

Question about Who Makes Financial Decisions

Currently, who is (are) primarily responsible for making your financial decisions: you, your spouse/partner, your child, or someone else? If someone else, please specify the other person.