2 Methodology and survey design

We partnered with a leading commercial market research firm, Taylor Nelson Sofres (TNS) Global, to develop and administer a survey that reports information on consumers’ financial knowledge related to debt, demographic characteristics, financial experiences, and judgments about their indebtedness.

Our approach to measuring debt literacy has three elements. First, we ask questions to assess understanding of key concepts; i.e., knowledge and skills related to taking on and managing debt, such as an understanding of interest compounding.Footnote ² Debt literacy questions can be solved with simple reasoning and do not require a calculator. Second, we take into consideration preferences regarding fixed monthly payments, often a very expensive way to pay for goods and services, but that many consumers seem to prefer even if it may be cheaper to borrow and repay in one lump sum. Third, we ask participants to judge their financial knowledge, and relate this self-assessment to their performance on the debt literacy questions.

The survey was fielded in November 2007 by the staff of TNS Global. The data were collected via a phone interview from a sample of 1,000 US residents. The survey collected information on a number of self-reported demographic characteristics, such as age, gender, race and ethnicity, marital status, employment, region of residence, family size and type, income, and wealth.Footnote ³

3 Measuring debt literacy

While there are a few national surveys that measure financial knowledge in the United States, such as the HRS, the Rand American Life Panel (ALP), and the Survey of Consumers,Footnote ⁴ few ask questions that focus specifically on borrowing and debt behavior. Our survey included three new questions designed to measure debt literacy. Specifically, respondents were asked questions that assessed their knowledge about the power of interest compounding and the workings of credit card debt and their preference for means of payment, given two options.Footnote ⁵ Tallying respondents’ correct and incorrect responses allowed us to classify individuals according to their respective levels of financial knowledge and to evaluate the link between financial knowledge and borrowing behavior.

The first question, measuring interest compounding, is as follows:

1. (i) 2 years;

2. (ii) Less than 5 years;

3. (iii) 5–10 years;

4. (iv) More than 10 years;

5. (v) Do not know;

6. (vi) Prefer not to answer.

Ignoring interest compounding, borrowing at 20% per year would lead to doubling in 5 years; someone who knew about interest on interest might have selected a number less than 5; someone who knows the ‘rule of 72’ heuristic would know that it would be about 3.6 years (i.e., correct answer (ii) ‘less than 5 years’). Answers (iii) and (iv) reflect a misunderstanding of the concept of interest accrual. Table 1, panel A, reports the responses to this question. Fewer than 36% of respondents answered this question correctly, consistent with the evidence reported in Lusardi and Mitchell (Reference Lusardi and Mitchell2007) that only a small fraction of respondents between the ages of 51 and 56 can correctly perform an interest-compounding calculation when asked to report how the amount in a savings account would grow over a 2-year period at an interest rate of 10%. The larger fraction, 43%, performed only a simple interest rate calculation, without taking into account that interest grows on interest. What we know from psychology and marketing is confirmed here: many people are not numerate and have difficulty grasping percentages (Chen and Rao, Reference Chen and Rao2007; Peters et al., Reference Peters, Hibbard, Slovic and Dieckmann2007).

Table 1. TNS survey debt literacy questions: weighted results

Note: The average score excludes the survey answers ‘Do not know’ and ‘Prefer not to answer’.

The evidence reported in panel A points to two other results. First, a sizable proportion of respondents, close to 20%, reported that they ‘do not know’ the answer to this question. As reported in other papers (Lusardi and Mitchell, Reference Lusardi, Mitchell, Mitchell and Lusardi2011a , Reference Lusardi and Mitchell b , Reference Lusardi and Mitchell c ; van Rooji et al., Reference van Rooij, Lusardi and Alessie2011), ‘do not know’ answers identify respondents with the lowest level of financial knowledge. Second, more than 30% of respondents overestimated, sometimes by a wide margin, the number of years it would take for debt to double when borrowing at a high rate. Overall, while many individuals deal frequently with credit cards and credit card debt, there seems to be limited knowledge of interest compounding.

We find similar evidence from the second literacy question, which asks respondents to calculate how many years it would take to pay off credit card debt when making minimum payments equal to the interest payments on the outstanding debt. Given that one is only paying interest, the principal balance will never decline. The exact wording of the question is as follows:

1. (i) Less than 5 years;

2. (ii) Between 5 and 10 years;

3. (iii) Between 10 and 15 years;

4. (iv) Never, you will continue to be in debt;

5. (v) Do not know;

6. (vi) Prefer not to answer.

Table 1, panel B, shows that only slightly more than 35% of respondents appreciated that making minimum payments equal to the interest payment on the outstanding debt will never eliminate debt. Not surprisingly, responses to the first two questions are highly correlated. More than half (56%) of those who responded correctly to the first question also responded correctly to the second question. ‘Do not know’ responses exhibit an even higher correlation, with 80% of those who respond ‘do not know’ to the first question responding similarly to the second question.

The third question seeks to determine not only whether people understand the notion of the time value of money and how skillful they are in comparing payment options, but also their preference for fixed payment, a very common feature of many contracts, which could reflect behavioral biases, such as lack of self-control, as argued by Meier and Sprenger (Reference Meier and Sprenger2010):

1. (i) Option (a);

2. (ii) Option (b);

3. (iii) They are the same;

4. (iv) Do not know;

5. (v) Prefer not to answer.

By paying $100 a month (versus $1,200 at the end of the year), one foregoes interest that could have accrued by having kept those dollars. Consistent with the findings of Stango and Zinman (Reference Stango and Zinman2009) that individuals are systematically biased toward underestimating the interest rate out of a stream of payments, we find that a very small proportion of respondents – close to 7% – selected the ‘lowest cost’ option, (b) (Table 1, panel C). A very high fraction of respondents, 40%, chose option (a) even though its Annual Percentage Rate is about 35% versus the 20% in option (b). About 39% thought that the two payment options were the same, ignoring the time value of money. Overall, these results suggest that individuals may underestimate the interest rate at which they are borrowing or exhibit behavioral biases or may pay for budgeting assistance.Footnote ⁶

To summarize: debt illiteracy is widespread. Only one-third of respondents can correctly answer a question about the power of interest compounding or about the workings of credit cards. Consumers choose expensive methods of payment, either because they do not do any calculations when choosing their methods of payment or because they suffer from behavioral biases, which also result in expensive borrowing. This evidence provides some reason for concern in an economy in which consumers routinely borrow and save using debt-like instruments.

4 Measuring financial experiences

Individuals engage in many different types of borrowing and lending financial transactions. We seek to characterize their behaviors broadly and relate them to levels of debt literacy. We expect to see a negative relationship between financial skills and certain wealth-depleting financial behaviors.Footnote ⁹

Experience measures. The TNS survey allows us to characterize a wide range of borrowing and investing experiences and transaction patterns of respondents. While we cannot measure their intensity or frequency, we can identify the types of transactions in which individuals have engaged.Footnote ¹⁰ Our typology includes four types of transactions: traditional borrowing, alternative financial services borrowing, saving/investing, and credit card use. The parenthetical text below was not part of the survey, but it is provided here to organize this information for the reader.

1. (1) (Experience with traditional borrowing, excluding credit cards.) Have you ever…

   1. a. Taken out a loan for student education

   2. b. Taken out an auto loan

   3. c. Taken out a home equity loan

   4. d. Gotten (or refinanced) a mortgage

2. (2) (Experience with alternative financial services borrowing.) Have you ever…

   1. a. Gotten a short-term ‘payday’ or ‘salary advance’ loan

   2. b. Gotten a ‘refund anticipation loan’ to accelerate the receipt of your taxes

   3. c. Gotten an auto title loan

   4. d. Used a pawn shop

   5. e. Bought goods on a layaway plan or at a rent-to-own store

3. (3) (Experience with saving/investing and payments.) Have you ever…

   1. a. Opened a checking or debit card account

   2. b. Opened a savings account or bought a CD

   3. c. Bought a savings bond or other bonds

   4. d. Invested in mutual funds

   5. e. Invested in individual stocks

4. (4) (Typical transaction mode for credit cards.) In the last 12 months, which of the following describes your use of credit cards?

   1. a. I do not have any credit cards or did not use them

   2. b. In some months, I ran an outstanding balance and paid finance charges

   3. c. In some months, I paid the minimum payment only

   4. d. In some months, I was charged a late charge for late payments

   5. e. In some months, I was charged an over the limit charge for charging more than my credit limit

   6. f. In some months, I used the cards for a cash advance

   7. g. My account was closed down by the credit card company

   8. h. I always paid my credit cards in full

While not exhaustive, this list includes many of the transactions which require a person to consider interest or fees.Footnote ¹¹ Table 2 provides the weighted incidences of the various transaction types for our sample population. Some activities are quite common – 91% of the population has experience with checking accounts, 81% has experience with savings accounts or CDs, and 79% currently has credit cards. Other activities are fairly rare. For example, in our sample only 4.4% had ever gotten a refund anticipation loan, 6.5% had ever had an auto title loan, and 7.8% had ever taken out a payday loan. As for credit cards, some (20%) do not have a card or do not use them. However, a majority of respondents use credit cards and do not pay the balances in full each month.

Table 2. Financial experience measures, total sample

This table reports the mean and standard deviation of the frequencies of the various financial experiences by 1,000 survey respondents. All frequencies are weighted. The survey was conducted in November 2007 by TNS Global.

Experience segments. We use this information to segment individuals into separate groups. This segmentation is carried out solely on the basis of transaction activity, without referring to demographics or other variables. We used cluster analysis to determine which groups of individuals have had similar financial experiences.Footnote ¹²

Based on the results of the cluster analysis, we reliably identify four main segments defined by common experiences. Table 3 provides the transaction characteristics of the four groups. While we ‘name’ these clusters for the sake of exposition, these names cannot fully characterize the range of behaviors that these groups share. Cluster 1, comprising about 26% of the sample, are people firmly engaged in the traditional financial system. These individuals all have credit cards but do not carry any revolving balances. This is such a distinctive characteristic that, for descriptive purposes, we use the name ‘pay in full’ to identify this cluster. These people have relatively high (but not the highest) levels of experience with mutual funds, stocks, and bonds. With respect to the other clusters, respondents in cluster 1 are most likely to have a mortgage and are fairly likely to have some experience with auto loans and home equity loans. Moreover, they have the lowest levels of alternative financial services usage.

Table 3. Financial experience segments

This table reports the incidences of various financial experiences, conditional on assignment to one of the four experience clusters. The clusters were defined with reference to these experiences and not on the basis of demographic or debt literacy information.

At the other end of the spectrum (cluster 4) is the 30% of our sample that does not use traditional financial services as often as the others. For descriptive purposes, we name them ‘users of alternative financial services’ or ‘AFS users’ in brief. For example, when compared with cluster 1, their usage of alternative financial services is considerably more frequent, using payday loans, tax refund loans, and pawn shops 5, 16, and 9 times more frequently. Most individuals in this cluster (68%) do not have credit cards and are more likely to be ‘unbanked.’ At the same time, the likelihood that they have ever invested in a stock, a bond, or a mutual fund – or held a mortgage – is about one-fifth that of the pay in full group.

In between are two groups that comprise 43% of the sample. Almost all have credit cards and virtually all carry revolving balances most months. They are virtually all ‘banked.’ The smaller subgroup, accounting for about 12% of the sample, is comprised of what we call the ‘borrowers/savers’ (cluster 2). This group has the highest level of experience with savings and investments of any of the four clusters. At the same time, this group has the highest levels of debt exposure too, with the most frequent experience with student loans (46%), home equity loans (54%), auto loans (94%), and virtually the same levels of mortgage loans as the pay in full group (77%). This group seems much more extended than the pay in full group in their credit card behavior.

The final 31% of the sample is what we call the ‘pay fees’ group (cluster 3). Relative to the three other groups, this group has the highest likelihood of paying the minimum amount due on their credit cards (56%), running late fees on their credit cards (17%), incurring over-the-limit fees (11.8%), and using their cards to get cash advances (16.1%). At the same time, they have far less experience than the borrowers/savers or the pay in full group with respect to mutual funds, stocks, or bonds, as well as less experience than these other groups with home equity loans, mortgages, and auto loans.

4.1 Characteristics by experience segment: demographics and debt literacy

Having created these experience segments, it is then natural to determine if there is a relationship between demographics, debt literacy, and these clusters. Are those in the pay in full group financially better off (e.g., in terms of income or wealth), more financially knowledgeable, and/or more secure in their level of indebtedness? Are the AFS users financially worse off, less financially literate, and/or less secure in their level of indebtedness? Finally, who are the fee payers? Table 4 provides descriptive statistics for these four clusters with respect to their demographics (panel A) and debt literacy (panel B). Following this discussion, we report the results of a multinomial logit analysis which examines cluster assignment as a function of all these factors.

Table 4. Characteristics of financial experience segments

This table reports statistics on the demographic and debt literacy variables for the total sample as well as for the four clusters defined in Table 3.

With respect to demographics, the payers in full have the highest income (43% have income over $75,000) and wealth (74% have financial assets in excess of $50,000). They are more likely to be married and to be white than are members of the other three clusters. Borrowers/savers have incomes almost as high as the pay in full group, similar levels of marriage, are the second-oldest group, and tend to be men (62%). In terms of wealth, this group is not quite as wealthy as the payers in full, with only 52% having financial assets above $50,000. The AFS users have the lowest income (53% have income below $30,000) and are most likely to be women (58%) and to be single or separated (47%). Finally, the fee payers look most like the ‘average’ American, with income distributed roughly similarly as in the overall sample, and other demographics (age, gender, marital status, and race) roughly comparable to the entire sample. Both the AFS users and the fee payers have considerably fewer financial assets than do the other two groups, with only 24% and 28%, respectively, having financial assets in excess of $50,000.

With respect to debt literacy (panel B), the payers in full and borrowers/savers are both more knowledgeable than the other two segments, scoring a considerably larger fraction of correct answers on the three questions than the latter two groups. The fee payers and AFS users are more likely to admit to not knowing the answers to the debt literacy questions. These patterns are also reflected in measures of self-reported financial knowledge; the fee payers and AFS users judge themselves to be much less knowledgeable than do payers in full and borrowers/savers. We can see this both in the average scores as well as in the distribution of scores. Whereas 48% and 53% of the payers in full and borrowers/savers ranked themselves in the top two scores with respect to their financial knowledge, for fee payers and AFS users, these comparable figures are 15.3% and 23.5%, respectively. In short, from the univariate statistics, the two clusters that pay the highest credit card fees and access the highest cost borrowing methods tend to have lower levels of debt literacy.

Of course, these univariate measures are correlated, and therefore we must consider all of the demographic variables simultaneously by using a multivariate approach to tease out the marginal relationship between debt literacy and behavior. The dependent variable in our analysis is an indicator for the four clusters we have identified in the data, and we use a multinomial logit analysis.

We have four correlated measures of debt literacy: the self-reported measure of financial knowledge and objective measures of debt literacy resulting from the answers to the three questions discussed above. The answers to the latter questions can be more finely characterized. For example, respondents with incorrect answers to the question about interest compounding are divided into two groups: those who underestimated and those who overestimated how quickly debt can double. Moreover, we add a dummy for those who did not know the answer to this question as this is a sizable and also distinct group of respondents (see Lusardi and Mitchell, Reference Lusardi and Mitchell2014, for a detailed discussion of the ‘do not know’ responses). We also include a dummy for those who refused to answer the questions.Footnote ¹³

All incorrect responses to the second literacy question were underestimates of how many years it would take to eliminate credit card debt. We aggregate the responses into those who made large underestimates (answered that it would take less than 5 years and between 5 and 10 years to eliminate credit card debt) versus those who chose a longer yet still incorrect time period (between 10 and 15 years). The incorrect answers to the third question characterize two distinct types of respondents. We keep these two groups separate. We again add dummies for those who did not know the answer or refused to answer.

Among the demographic variables, we include age and age squared and dummies for gender, race, and marital status. We add dummies for larger household sizes, characterizing those with four members and five or more members, and a dummy for those who are not employed; these families may be more vulnerable to shocks. Finally, we add dummies for household income and wealth. To consider how these demographic variables explain the four clusters, we conduct a multinomial logit regression across the four clusters considering the demographic variables and the debt literacy and self-assessed financial knowledge variables.

Table 5 reports the marginal effect of each variable in the multinomial logit regressions across the four clusters. For brevity, we report the estimates of the debt literacy variables only but we comment on the estimates of the demographic variables when appropriate.Footnote ¹⁴ Moreover, rather than reporting the estimates with respect to a reference group, we calculate the marginal effects in comparison to all the other clusters. We first consider the self-reported measure of financial knowledge (Table 5, first set of estimates). Even after accounting for demographics, those who display higher levels of financial knowledge are more likely to locate in cluster 1 (pay in full). Levels of self-assessed financial knowledge above the mean score (score of higher than 4) are associated with higher chances of being among those who pay in full, and the likelihood of being in this group is the greatest for those with high self-assessed knowledge (scores of 6 and 7). Individuals in this cluster are also those with high incomes (income greater than $75,000) and high wealth. African-Americans and Hispanics and those with large families are less likely to be in the pay in full group.

Table 5. Multinomial logit analyses of characteristics of financial experience segments

This table reports the logit estimates belonging to one of four clusters on a set of debt literacy measures and demographic variables (marginal effects are reported). See text for detail.

Definitions of Lit1–Lit4 variables:

Self-assessed financial knowledge: Lit1 = 4, Lit2 = 5, Lit3 = 6, Lit4 = 7. Omitted class: low financial knowledge 1–3.

First measure of debt literacy: Lit1 = underestimate, Lit2 = overestimate, Lit3 = do not know, Lit4 = refuse to answer. Omitted class: Correct.

Second measure of debt literacy: Lit1 = underestimate, Lit2 = overestimate, Lit3 = do not know, Lit4 = refuse to answer. Omitted class: Correct.

Third measure of debt literacy: Lit1 = option a, Lit2 = same, Lit3 = do not know, Lit4 = refuse to answer. Omitted class: Correct.

*p < 0.05; **p < 0.01; ***p < 0.1.

Standard errors in parentheses.

Self-reported financial knowledge is not related to the behavior of those in cluster 2, the borrowers/savers.Footnote ¹⁵ These individuals have relatively high income, and they do not display characteristics that are usually associated with debt problems (e.g., large families, unemployed, or divorced or separated). Income and race are the only variables that characterize those in cluster 2. While borrowers/savers do carry credit card balances and tend to pay finance charges, this behavior seems less likely to be due to lack of knowledge.

Those in cluster 3, the fee payers, are considerably less likely to report high levels of financial knowledge, even after controlling for many demographic traits. These respondents are also more likely to have lower levels of wealth, to be African-American, and to have large families.

Low levels of financial knowledge also characterize AFS users in cluster 4. These respondents are much less likely to report high levels of financial knowledge and more likely to be unemployed or to have lower incomes (income less than $30,000).

We find similar patterns when we use the three measures of debt literacy instead of the self-reported measure of financial knowledge (Table 5, last three sets of estimates). Those who overestimated how long it takes for debt to double may be lulled into borrowing more or not paying on time. Indeed, those who are less likely to be knowledgeable about interest compounding, both because they overestimated the number of years it takes for debt to double or because they did not know the answer to this question, are less likely to belong to the pay in full group and more likely to belong to the AFS users group. Being unable to answer the question about interest compounding also characterizes those who belong to cluster 3, the pay fees segment, who tend to carry balances and pay finance charges and penalty fees. On the other hand, those who did not know the answer to the question about interest compounding are less likely to belong to cluster 2, the borrowers/savers, who are likely to carry balances and not pay on time.

Turning to the second measure of debt literacy, we find that those who make mistakes, both small and large, in answering this question are significantly more likely to belong to the AFS users group. Those who display the lowest level of debt literacy, i.e., responded that they do not know the answer to this question, are also more likely to belong to this group. Conversely, those who made small mistakes or did not know the answer to the question are less likely to belong to the pay in full or borrowers/savers clusters.

Estimates for the third debt literacy question – which may also capture a behavioral bias – show similar findings: those who answered this question ‘incorrectly’ (i.e., chose option (a) or thought the two options were the same or ignored the time value of money) or did not know the answer to the question are much less likely to belong to the pay in full group or the borrower/saver group – and are more likely to belong to the pay fees cluster.Footnote ¹⁶

In summary, for each measure of debt literacy, there is a strong relationship between debt literacy and debt behavior, even after controlling for demographics. The more financially knowledgeable, who grasp basic concepts about debt, are much more likely to pay their credit cards in full, while those who are less debt literate are more likely to pay fees or be AFS users. Our borrowers/savers are rather knowledgeable and have high incomes, yet tend to carry credit card balances and pay finance charges, perhaps because these charges are not particularly consequential for this group. In the next section, we try to address this issue by examining self-reported debt loads. Before we do so, we should mention that, while we find a strong link between debt literacy and debt behavior, we cannot say whether it is a causal one. It is possible that behavior toward debt also has an effect on debt literacy, though one may have to explain why behavior toward debt is responsible for some of the nuances of the debt literacy variables (e.g., overestimating rather than underestimating the effect of compound interest). Also, there could be unobservables driving both debt literacy and debt behavior that we are not able to account for in cross-sectional data. Moreover, debt literacy may be measured with considerable error. These are topics for future work, but we note that in all of the estimates that try to account for these problems (measurement error and/or endogeneity), the effect of financial literacy is always found to be bigger than reported in the simple Ordinary Least Squares (OLS) estimates (see Lusardi and Mitchell, Reference Lusardi and Mitchell2014); thus our estimates may even underestimate the true effect of debt literacy.

5 Overindebtedness

According to intertemporal economic models, consumers borrow to smooth consumption over the life cycle. Variations in debt over time and across individuals would not necessarily indicate that anyone felt ‘over-levered’ or ‘under-levered.’ Yet imperfections in financial markets and shocks might lead individuals to conclude that their debt level was suboptimal. Some may suffer from credit constraints and be unable to borrow as much as they would like. Others may be hit by unexpected negative shocks and carry higher debt loads than they might otherwise prefer. The existing literature has largely failed to consider that some may accumulate too much debt by being unaware of the consequences of their own choices. We consider the latter possibility, looking for links between debt levels and lack of financial knowledge.

In the survey, we sought to understand whether people have difficulties paying off their debt. While we recognize the potential problems with self-reported measures of debt levels, these reports give information about credit constraints and consumers’ interest in additional borrowing. To gauge debt levels, we asked individuals the following question:

1. a. I have too much debt right now and I have or may have difficulty paying it off.

2. b. I have about the right amount of debt right now and I face no problems with it.

3. c. I have too little debt right now, I wish I could get more.

4. d. I just do not know.

In aggregate, in November 2007, before the financial crisis hit the economy, 26.4% of respondents in our representative sample of Americans already said they have or may have difficulty paying off debt (have difficulty with debt). Another group, 11.1%, ‘just did not know’ their debt position (unsure). We focus primarily on these two groups.

Paralleling our analysis in the last section, we first report on the traits of these different groups in univariate terms and then provide a multinomial logit analysis of debt loads. Table 6 shows that relative to those who are comfortable with their level of debt, those experiencing difficulty with debt are younger and have fewer financial assets and lower incomes. Note that they are disproportionately drawn from the pay fees cluster, while almost none are part of the pay in full segment. In terms of debt literacy, they rank themselves the lowest.

Table 6. Characteristics by self-assessed debt levels

This table reports statistics on the demographic, debt literacy, and experience segmentation variables for the total sample as well as for the four groups defined by their self-assessed level of indebtedness.

The ‘unsure,’ the 11% who were unable to judge whether they have too much or too little debt (they answered they just did not know), tend to be disproportionately female (nearly 70%), African-American (18%), and unmarried (60%): the same characteristics displayed by those with low debt literacy. With respect to income, they are disproportionately drawn from the lowest income group (59% with household income under $30,000 per year), and have considerably less wealth than the 60% who categorized their debt load as ‘about right.’ Their debt literacy is considerably weaker than that of respondents who judged their debt to be either about right or have difficulty with debt. Respondents in this group were also more likely to select ‘do not know’ as the answer to the debt literacy questions. This group is disproportionately drawn from the AFS users segment.

We perform a multinomial logit analysis of the three groups mentioned above: those having difficulty with debt, the unsure, and those with the right amount of debt. As predictors for these debt outcomes, we add dummies for the different measures of debt literacy (Table 7). Moreover, we use demographic variables, including age and age squared, and dummies for gender, marital status, race, family size, employment status, and income and wealth.Footnote ¹⁷

Table 7. Multinomial logit analyses of self-assessed debt levels

This table reports the logit estimates of belonging to a specific debt group on a set of debt literacy measures and demographic variables (marginal effects are reported). See text for detail.

Definitions of Lit1–Lit4 variables:

Self-assessed financial knowledge: Lit1 = 4, Lit2 = 5, Lit3 = 6, Lit4 = 7. Omitted class: low financial knowledge 1–3.

First measure of debt literacy: Lit1 = underestimate, Lit2 = overestimate, Lit3 = do not know, Lit4 = refuse to answer. Omitted class: Correct.

Second measure of debt literacy: Lit1 = underestimate, Lit2 = overestimate, Lit3 = do not know, Lit4 = refuse to answer. Omitted class: Correct.

Third measure of debt literacy: Lit1 = option a, Lit2 = same, Lit3 = do not know, Lit4 = refuse to answer. Omitted class: Correct.

*p < 0.01; **p < 0.05; ***p < 0.1.

Standard errors in parentheses.

Self-reported financial knowledge is strongly correlated to self-assessed debt burdens. Those who reported higher levels of financial knowledge are more likely to belong to the group who reported having no difficulty handling their current debt. The effect is not only sizable but it tends to increase with higher scores for self-assessed knowledge. Conversely, those with lower self-reported financial knowledge are much more likely to have reported having difficulty with debt, and again there is a monotonic (negative) relationship between financial knowledge and having too much debt. Although the estimates are less sizable than for those who have or may have difficulty with debt, those who just do not know their current debt position are also much less likely to display high levels of self-assessed financial knowledge (Table 7). Moreover, those who are employed and have higher income and higher wealth are much more likely to have reported that they have the right amount of debt. Finally, women, African-Americans, and those with low income and wealth are less able to judge their debt load.

When we consider the measures of debt literacy, we find similar results. Most importantly, these results are consistent with the multinomial logit for the experience segments. Specifically, those who overestimated the number of years it takes for debt to double (first measure of debt literacy) are also more likely to have reported that they have or may have difficulty paying off debt. On the other hand, those who made mistakes in answering this question or did not know the answer to this question are much less likely to have reported that they have the right amount of debt; they are more likely to belong to the unsure (just do not know) group.

Knowledge about how to eliminate credit card debt (second debt literacy question) is also related to self-assessed levels of debt. Those who display the least knowledge, i.e., claimed not to know the answer to this question, are less likely to have reported having the right amount of debt. Those respondents who were not able to answer the third debt literacy question (installment versus single future payments) are less likely to have reported having the right amount of debt; they are more likely to belong to the unsure group. There seems to be a group of individuals unsure of their financial skills and of their debt loads.

For completeness, we also examine the estimates in which we account for the three dummies characterizing different clusters (the first cluster is the reference group). In this way, we can assess whether financial experiences have a direct effect on the amount of debt that respondents have and whether the effect of debt literacy remains significant after accounting for the behavior characterized by the four clusters. The effect of debt literacy weakens only for the third measure of debt literacy; otherwise, there is still an effect even after accounting for the clusters. Thus, debt literacy is related to debt loads above and beyond the effect it has on financial experiences. Moreover, even after accounting for a large set of demographic characteristics, those who belong to the three segments that do not pay credit cards in full are disproportionately more likely to have difficulty with debt. Similarly, members of clusters 2, 3, and 4 are much less likely to have reported that they have the right amount of debt. Note that not just the fee payers and the AFS users reported having difficulty with debt, but those in cluster 2, who carry some balances and pay some finance charges, also end up with too much debt.Footnote ¹⁸

6 An estimate of the cost of ignorance

In this section, we offer partial estimates of what we call ‘the cost of ignorance,’ or the financial transaction costs incurred by less-informed Americans and the component of these costs that is particularly related to lack of financial knowledge. For the purpose of this calculation, we focus exclusively on credit card debt.

This calculation of expected costs has two components – the likelihood of and the costs of various behaviors. First, we calculate the likelihood of engaging in four credit card behaviors that give rise to explicit fees or finance charges: paying bills late, going over the credit limit, using cash advances, and paying the minimum amount only. These likelihoods come directly from empirical estimates using the data on credit card behavior, debt literacy, and demographics. We compare consumers with higher versus lower financial knowledge, with the least financially savvy in our population defined as those who judge their financial knowledge equal to 4 or lower on our seven-point scale.

The second part of the calculation estimates the costs incurred by the cardholder, conditional on engaging in the particular behavior. For the purpose of this calculation, we use estimated credit card balances across higher versus lower financial knowledge individuals.

As described in more detail in the appendix and Appendix Table A2, these four behaviors give rise to collective fees and charges of $39.3 billion paid by cardholders, most of which are finance charges due to paying only the minimum amount due. The less knowledgeable account for 28.7% of the cardholder population and account for 21% of these charges, because of their higher likelihood of incurring them but a lower balance on their credit cards. Thus, they bear a sizable share of the fees associated with fee-inducing behaviors. Perhaps more importantly, of these four types of charges incurred by less-knowledgeable cardholders, one-third are incremental charges that are empirically linked to low financial knowledge after controlling for many variables, including income, age, family structure, wealth, and other demographic factors. Even in these very conservative calculations, the cost of ignorance is significant.

7 Moving forward

This paper has circulated extensively prior to its publication, and a number of studies have used the questions we designed for this survey and have found very similar results. For example, Gathergood (Reference Gathergood2012) and Disney and Gathergood (Reference Disney and Gathergood2013) used a set of questions very similar to our ‘debt literacy’ measures to study consumer overindebtedness and borrowing in the UK. As in the US, consumers in the UK have low debt literacy and are not familiar with the concepts related to debt. Importantly, these authors also find a strong link between low debt literacy and both overindebtedness and high-cost consumer borrowing. Interestingly, their data show that low debt literacy is also associated with finding financial services complicated or confusing.

Some of the questions designed for this survey were also added to the National Financial Capability Survey (NFCS), in both the 2009 and 2012 waves; the 2015 wave of the NFCS will have not only the set of questions measuring financial experiences and overindebtedness but also a question about debt literacy. De Bassa Scheresberg and Lusardi (Reference de Bassa Scheresberg and Lusardi2013) examine data from the first wave of the NFCS, using the questions on high-cost methods of borrowing that were taken from the TNS survey (payday lons, pawn shops, auto title loans, tax refund loans, and rent-to-own shops). Similar to the findings in our work, they show that those with lower financial literacy are more likely to borrow at high costs. De Bassa Scheresberg (Reference de Bassa Scheresberg2013) examines high-cost methods of borrowing among the young and shows there is a sharp divide in the use of these methods of borrowing among education groups; it is those with low educational attainment who disproportionately use high-cost methods of borrowing. However, he shows that financial literacy also matters; those who have low financial literacy are more likely to borrow at high costs, even after accounting for the effect of education.

In the wake of the financial crisis, academic attention has turned to debt, how individuals manage debt, and the importance of financial skills. Gerardi et al. (Reference Gerardi, Goette and Meier2013) show that numerical ability (assessed with a combination of questions measuring financial literacy and knowledge of math) has contributed substantially to the massive number of defaults on subprime mortgages in the recent financial crisis. According to their estimates, those in the highest numerical ability group have a much lower probability of defaulting on their subprime mortgages than those in the lowest financial numeracy group, and this alone can explain a large part of the chance of defaulting. Similarly, Agarwal and Mazumder (Reference Agarwal and Mazumder2013) examine two suboptimal behaviors related to borrowing, with the first case featuring the use of credit cards for convenience transaction after a balance transfer and the second involving a financial mistake on a home equity loan application. They find that consumers with higher math scores (as measured by the Armed Forces Qualifying Test) are substantially less likely to make a financial mistake. Agarwal et al. (Reference Agarwal, Driscoll, Gabaix and Laibson2009) also focused on financial ‘mistakes’ related to decisions about debt, showing not only that they are consequential in terms of costs associated with them, but also that they are most prevalent among the young and the old, which our data show are the age groups with the lowest levels of debt literacy.

Other papers have pointed to the relationship between self-control problems and debt, which could help us interpret the evidence from one of our three debt literacy questions. For example, Meier and Sprenger (Reference Meier and Sprenger2010) show that present-biased individuals are more likely to have credit card debt and to have significantly higher amounts of credit card debt. Similarly, McCarthy (Reference McCarthy2011) shows that self-control is an important predictor of financial distress (keeping up with bills and credit commitments).

Given the extent to which debt has become a persistent feature of household balance sheets at every stage of life, including close to retirement (Lusardi and Mitchell, Reference Lusardi and Mitchell2013) or at the start the economic life, due to the growing size of student loans or other loans, we think it is more important than ever before to pay attention to how individuals manage indebtedness.

With consumer credit becoming so widely available and accessed among households, one may argue about the possibility of societal or peer pressure regarding debt. For example, are people who perceive themselves poorer than their social circles more likely to borrow to keep up with peers, and can these types of borrowing contribute to potential financial distress? These are the questions that a recent paper investigates (Georgarakos et al., Reference Georgarakos, Haliassos and Pacini2014). The study shows that such pressure exists; those who perceive others as having higher average incomes consume more (trying to emulate the peer spending) and also borrow more, potentially increasing their financial distress in the future. Thus, more than ever, debt behavior and over-indebtedness are important topics to study.

8 Implications and conclusions

With this work, we focus attention on an important component of financial literacy – debt literacy. Moreover, we consider individuals’ rich set of financial experiences, rather than simply focusing on one behavior. We also take into account individuals’ assessments of their own debt levels. Finally, we design a project that blends scholarly research with market research. Our conclusions suggest a complex set of interactions among debt literacy, financial experiences, demographics, and debt loads.

Low levels of debt literacy are the norm, and understanding of the basic mechanics of debt is especially limited among the elderly, women, certain minorities, and people with lower incomes and wealth. Particularly intriguing is the notion that certain respondent groups, like the elderly, think they know considerably more than they actually do. This disparity may help explain the incidence of financial frauds perpetrated against the elderly. Moreover, women – both young and old – exhibit substantially lower debt literacy than men.

Second, people have rich sets of financial experiences. Our work collapses these experiences into four segments and shows that the segments are closely linked with both demographics and debt literacy. It is troubling that the people whose financial transaction patterns are characterized by high-cost borrowing are those who come from vulnerable demographic groups and – even after controlling for these factors – are less debt literate. People who make financial choices that incur high fees and charges (e.g., only paying the minimum balance on credit cards, incurring late or over-the-limit fees, using alternative financial service credit such as payday loans, tax refund loans, or pawnshops) are those with a weaker understanding of the implications of debt. Finally, in November 2007, over a quarter of Americans felt overburdened with respect to their debt loads and another 11% were unable to assess their debt position. Thus, even before the onset of the financial crisis, more than 40% of families had issues with their debt position. Moreover, those facing difficulty paying off debt were drawn from certain demographic groups, had common financial experiences characterized by costly borrowing, and tended to have lower levels of debt literacy.

Our empirical results suggest a sizeable cost of financial ignorance as well. Using credit cards as an example, we find that the less financially knowledgeable pay a sizeable fraction of fees and finance charges. Our empirical analysis suggests that about a third of the fees and charges paid by low financial knowledge individuals are related to lack of knowledge, even after controlling for observable differences in income, wealth, family status, and other factors.

We think there are a number of implications from our findings. If poor financial decisions partly result from lack of financial knowledge, then in certain circumstances, one may be able to design mechanisms to compensate for it. These solutions might be embodied in auto enrollment options, such as those studied by Choi et al. (Reference Choi, Laibson, Madrian and Metrick2003, Reference Choi, Laibson, Madrian, Metrick and Wise2004a ) and Choi et al. (Reference Choi, Laibson and Madrian2004b ), among others. However, once one recognizes the wide range of financial choices that consumers potentially face, it becomes harder to conceive that poor financial decisions can be overcome in this fashion. For example, someone who needs additional funds will have to search for and compare alternatives ranging from extending their borrowing on their credit cards to taking out a home equity loan to overdrafting a bank account to taking out a payday loan to borrowing from a friend or going to a pawn shop. As much as we could try to circumscribe the choices, individuals will need to make active decisions. Our work suggests that financial literacy is related to the choices that people make, with less knowledgeable people making more costly decisions – even after controlling for a host of other factors. We interpret this to mean that additional research on financial literacy – and education to enhance financial literacy – remains an important priority, especially to understand the debt-related choices that consumers make.