2.1. Development and maintenance of the NFCS State-by-State Survey

The topics covered in the State-by-State Survey are determined by a multidisciplinary team of researchers and policymakers convened by the FINRA Foundation. In past years, researchers have included academics across such fields as economics, consumer sciences and personal finance, psychology, and social work. Policymakers have included representatives from the Consumer Financial Protection Bureau (CFPB), the Securities and Exchange Commission, the Department of Defense, the U.S. Treasury, and the Federal Reserve Board. The questions cover a wide array of financial behaviors and attitudes, many of which align with the four components of financial capability described below. With each wave, questions that are seldom used, for which measurement issues have been identified, or that are no longer pertinent are removed from the questionnaire. These are replaced with items addressing new and emerging issues (e.g., student loan crisis, propagation of fintech, emergence of COVID-19). Whenever possible, new items are obtained or adapted from existing surveys. The questionnaire is designed to take an average of 15 minutes to complete.

The State-by-State Survey has evolved over the years, both in structure and content, but the overarching aims of the study have remained unchanged. At its core, the State-by-State Survey (and other NFCS components) strives to monitor the financial capability of adults living in the United States. While no single, agreed-upon definition of financial capability exists, the State-by-State Survey focuses on the following four components:

1. 1) Making Ends Meet – This component explores respondents’ ability to cover daily expenses and remain financially afloat. It contains items on topics like financial fragility, financial anxiety, medical expenses, and income volatility.

2. 2) Planning Ahead – This component examines savings behaviors for coping with known and unforeseen events. It includes items on topics like investing, retirement planning, and rainy-day funds.

3. 3) Managing Financial Products – This component examines the financial products respondents use and how they use them. It contains items on topics like banking and payments, homeownership, fintech, and different forms of debt (e.g., credit cards, student loans, mortgages).

4. 4) Financial Knowledge and Decision-Making – This component explores respondents’ financial knowledge and self-perceptions of their knowledge, behaviors, and numeracy skills. It includes items on topics such as objective and self-perceived financial knowledge, self-perceived financial capability, and participation in financial education.

The State-by-State Survey also collects a wide variety of demographic information (e.g., gender, income, educational attainment) and attitudinal data (e.g., risk tolerance, financial satisfaction, perceptions of debt) that may fall outside the four pillars of financial capability. In addition to items that can be analyzed individually, the State-by-State Survey contains questions that form multi-item scales. For instance, since 2018, the State-by-State Survey has included the CFPB’s abbreviated five-item financial well-being scale, a set of questions developed to identify consumers’ own perspectives on their financial lives. Created with input from experts and consumers and developed using rigorous psychometric methods, the scale provides a reliable measure of an individual’s financial well-being. More information on the scoring and technical details of the financial well-being scale can be found here: https://www.consumerfinance.gov/data-research/research-reports/financial-well-being-scale/.

2.2. NFCS State-by-State Survey methodology

Respondents for the State-by-State Survey are drawn from established non-probability online research panels using quota sampling. These panels consist of millions of individuals who are offered incentives in exchange for participating in online surveys.Footnote ³ Each uses industry-standard techniques to promote data quality, verifying panel members’ identities and ensuring the validity of panelists’ demographic characteristics. Quotas are set to approximate Census distributions for age by gender, ethnicity, education level, and income within each state based on data from the Census Bureau’s American Community Survey. The State-by-State Survey (and other NFCS components) does not specifically target heads of households or primary decision-makers. For each wave of data, the State-by-State Survey is typically fielded from June through October. Starting with the 2021 wave, Institutional Review Board (IRB) exemptions were obtained from an independent IRB, and all future waves will seek IRB exemptions or approvals, as well. IRB determination letters can be obtained by contacting the FINRA Foundation.

The State-by-State Survey data set includes three weights that are based on data from the American Community Survey and created via raking.

1. 1) National Weights – National weights can be applied to represent the national population in terms of age, gender, ethnicity, education, and Census Division.

2. 2) Regional Weights (i.e., Census Division) – Regional weights can be applied to represent each Census Division in terms of age, gender, ethnicity, education, and state.

3. 3) State Weights – State weights can be applied to represent each of the 50 states and the District of Columbia in terms of age, gender, ethnicity, and education.

To promote the use and interpretability of the State-by-State Survey (and other NFCS components), the data sets include variable and value labels (for SPSS and Stata statistical software), detailed and well-documented codebooks, and methodological documentation. All data sets are available in .csv, .dat (Stata), and .sav (SPSS) formats. The data sets are free to download at https://finrafoundation.org/knowledge-we-gain-share/nfcs/data-and-downloads.

3.1. NFCS investor survey

Since 2015, the NFCS has included an Investor Survey component. The Investor Survey is a follow-up study to the State-by-State Survey targeting respondents who report owning taxable investment accounts outside of retirement accounts. The survey includes questions about investor-specific issues that are outside the scope of the State-by-State Survey. The data provide an opportunity to better understand the behaviors, attitudes, knowledge, and decision-making processes of investors. While respondents are not told the Investor and State-by-State Surveys are linked, the two data sets can be merged to provide a comprehensive picture of the investing behaviors, financial capability, and demographics of those who complete the Investor Survey. All qualifying individuals in the State-by-State survey are invited to take the Investor survey, yielding approximately 3,000 respondents. The Investor Survey data set includes weights to approximate the age and education distributions of the investing population in the corresponding State-by-State survey. To date, three Investor Surveys have been conducted (2015, 2018, 2021), and the fielding of the 2024 Investor Survey began in July of 2024. For each wave, the Investor Survey typically begins fielding about one month after the start of the State-by-State Survey.

3.2. 2021 NFCS Asian American data set

To supplement the limited sample size of Asian American/Pacific Islander respondents within the 2021 State-by-State Survey data set (n=1,193), the 2021 State-by-State Survey questionnaire was administered to an additional sample of 1,001 U.S. adults who identified their ethnicity as Asian or Pacific Islander. The 2021 NFCS Asian American/Pacific Islander data set combines the two samples of Asian American/Pacific Islander respondents for a total sample of 2,194 respondents. This supplemental data set includes weights that enable researchers to approximate the U.S. population of Asian American/Pacific Islander adults in terms of age, gender, ethnicity, education, and Census Division based on the American Community Survey. More information about this data set is available upon request.

3.3. 2021 NFCS Puerto Rico survey

To shed light on the financial capability of adults living in Puerto Rico, an often-understudied geographic segment of the U.S. population, a modified version of the 2021 State-by-State Survey questionnaire, was fielded to a sample of 1,001 adults living in Puerto Rico. The questionnaire was nearly identical to the State-by-State Survey questionnaire, with a few geography-specific screening questions. The 2021 Puerto Rico Survey questionnaire was offered in English and Spanish. The data set includes weights that researchers can use to approximate Census distributions for Puerto Rico in terms of age, gender, ethnicity, and education based on the American Community Survey. The 2021 Puerto Rico Survey questionnaire was fielded from June to December 2021. More information about the NFCS Puerto Rico Survey data set is available upon request.

3.4. Longitudinal NFCS American Life Panel data

In 2012, the State-by-State Survey questionnaire was administered to 2,075 respondents in RAND’s ALP, a nationally representative probability-based panel consisting of about 6,000 adults ages 18 and older. The same survey respondents who took the survey in 2012 were assessed six years later in 2018 and will again be surveyed in 2024. This cross-lagged panel design provides an opportunity to measure intraindividual changes in financial capability. Further, because ALP panelists are often interviewed online for research purposes, the longitudinal NFCS ALP data can be combined with other existing ALP data, increasing the scope of questions that can be answered. These data sets can be found at https://www.rand.org/education-and-labor/survey-panels/alp.html. The 2012 wave is data set number 284 (NFCS), and the 2018 data set is number 504 (NFCS [W2]). Both are available for free download.

4.1. Financial knowledge over time

With the financial landscape in the United States becoming increasingly complex, having the knowledge and skills to process financial information and make informed decisions is crucial. As such, one pillar of financial capability, “Financial Knowledge and Decision-Making,” has received significant attention over the years: respondents’ objectively measured financial knowledge. More specifically, since 2009, the State-by-State Survey has included a five-item quiz that tests respondents’ knowledge of fundamental financial concepts. These questions include the Big 3 created by Lusardi and Mitchell (Reference Lusardi and Mitchell2011) that assess a respondent’s knowledge of simple interest, inflation, and diversification. To further differentiate respondents’ financial literacy levels and cover other areas of financial knowledge, the State-by-State Survey incorporated two additional items into this battery: one on mortgages and another on bond prices. Together, these five items are referred to as the Big 5 and have been asked in all waves. The Big 5 provides information on the state of financial literacy in the United States and its evolution over the years. The large-scale nature of the State-by-State Survey also affords researchers the ability to explore nuances in financial literacy across geographic, demographic, behavioral, and attitudinal characteristics, an important benefit of the State-by-State Survey given the heterogeneity present in financial literacy (Lusardi and Mitchell Reference Lusardi, Mitchell, Lusardi and Mitchell2011).

Newer waves of the State-by-State Survey have additional questions designed to measure financial knowledge. Since 2015, there has been an item that assesses respondents’ understanding of the role of compound interest as it applies to debt. In 2021, an item assessing respondents’ understanding of probability was added, given the important role that understanding probability plays in comprehending risk, a key component of financial decision-making. The Big 3, Big 5, and full financial literacy quiz can be found in Table A.I in the Appendix.

Since the start of the NFCS, financial literacy rates have been low and steadily decreasing. Figure 1 shows two measures of financial literacy from 2009 to 2021: (1) the mean number of correct answers on the Big 5 financial literacy quiz and (2) the percentage of respondents who correctly answered at least four of the Big 5 quiz questions (defined here as having “high” financial literacy). In 2009, the respondents on average answered 3.0 of 5 questions correctly. By 2021, this number had fallen to 2.6. The proportion of respondents displaying high financial literacy has also declined – from 42 percent in 2009 to fewer than one-third (32 percent) of respondents in 2021. The decline is also evident when examining the proportion of respondents correctly responding to all five questions – from 15 percent in 2009 to 12 percent in 2021. While the largest drops occurred for questions on inflation and risk, declines were observed for all questions (Urban and Valdes Reference Urban and Valdes2022).

Figure 1. Respondents’ performance on Big 5 financial literacy quiz, by year.

Low levels of financial literacy are a concern because they are associated with adverse financial behaviors and outcomes (Angrisani et al. Reference Angrisani, Burke, Lusardi and Mottola2023). While declines in financial literacy stem in part from an increase in the number of incorrect responses to these questions, they are also a result of a steady increase in the selection of “don’t know” as a response option, which is often treated as an incorrect response by researchers. Similarly, Angrisani et al. (Reference Angrisani and Kapteyn2024) report a small rise in the percentage of respondents answering “don’t know” to all of the Big 3 questions in the Understanding American Study between 2014 and 2023. In the NFCS data, the rise in “don’t know” responses accounts for 75 percent of the drop in financial knowledge from 2009 to 2021. The rise in “don’t know” responses is particularly evident for certain questions, namely, on interest, a rise from 10 to 15 percent, and mortgage, from 15 to 22 percent. While the underlying reasons behind the declines in financial literacy rates or increases in “don’t know” response rates are unclear, one hypothesis under examination includes the possibility that the increased use of mobile phones for survey completion may lead to a rise in the selection of “don’t know” for cognitively demanding questions, like knowledge-based questions. In any case, the data provide evidence of the vital role of high-quality, accessible financial education (Urban and Valdes Reference Urban and Valdes2022).

While low levels of financial literacy have been widely reported, specific rates of financial knowledge can differ across data sets. These differences are present when comparing responses to the Big 5 questions in the 2012 NFCS State-by-State Survey data set and the 2012 NFCS data collected via RAND’s ALP, a probability-based panel. On average, ALP respondents answered 3.3 questions correctly, compared to 2.9 questions answered correctly by respondents in the 2012 State-by-State Survey. The percentage of respondents who demonstrated high financial literacy (correctly answering four or five questions) was also higher in the ALP, 49 percent in the ALP compared to 39 percent in the NFCS.

While financial literacy levels are lower in the NFCS than in the ALP, the two data sources show similar associations between financial literacy and financial capability measures. Figure 2 shows how three measures of financial capability: emergency savings, planning for retirement, and financial fragility (see Table A.II in the Appendix for these three financial capability questions) vary with financial literacy levels. In both samples, respondents who report having emergency savings or having done some planning for retirement, on average, tend to answer more financial literacy questions correctly than those without emergency savings or those who have not done any planning for retirement, while respondents who report financial fragility tend to score lower than their counterparts who are not financially fragile. For instance, those who reported having emergency savings correctly answered, on average, 3.3 and 3.7 Big 5 questions in the NFCS and ALP, respectively. Those without emergency savings, in turn, answered only 2.6 and 2.8 correctly. The direction and magnitude of the differences between the two samples for the planning for retirement variable and financial fragility variable are also similar.

Figure 2. Associations between the Big 5 and financial capability by sample.

Though the Big 5 is not present in many other data sets, limiting the ability for direct comparisons, as Figure 3 shows, there is a fair amount of variation in responses to the Big 3 questions across six different samples. Four of the samples were obtained from probability-based panels, one was primarily collected via phone with some in-person interviewing, and the last sample studied was the State-by-State 2021 NFCS. Data for the six samples were collected between 2020 and 2023. Using range and standard deviation as measures of sample-to-sample variation, we identified that the Diversification question varies most across samples (range = 42%–66% correct, SD = 11%), followed by the Inflation question (range = 69%–84% correct, SD = 10%). The least variability occurs for the Interest question, with a standard deviation of 7 percent. The NFCS has the lowest mean level of correct answers across the three questions (55 percent) followed by the Probability 1 sample (59 percent). Respondents from the other four sources show higher levels of financial literacy, ranging from 69 percent to 75 percent.

Figure 3. Big 3 financial literacy performance by sample.

Sample estimates may vary for a number of reasons. One notable factor is that the NFCS is a non-probability-based sample while the others are probability-based, and research shows that probability-based samples tend to provide more accurate population estimates (Mercer et al. Reference Mercer, Kennedy and Keeter2024). Other factors include the timing of survey administration, which could affect responses through history effects (Cook and Campbell Reference Cook and Campbell1979) both within and between samples. The mode of administration could also play a role – for example, surveys administered online could result in different responses than surveys administered over the phone or in person (AAPOR 2024). Survey context can also affect how subjects respond to survey questions (Beals et al. Reference Beals, Carr, Mottola, Deevy and Carstensen2017; Tourangeau et al. Reference Tourangeau, Rips and Rasinski2000). That is, survey questions that come before the Big 3 could, arguably, affect how subjects answer the Big 3. Finally, differences in survey length and question placement (i.e., early in the survey or late in the survey) could play a role via increased dropout rates and lower levels of respondent engagement (Jeong et al. Reference Jeong, Aggarwal, Robinson, Kumar, Spearot and Park2023; Revilla and Höhne Reference Revilla and Höhne2020). Of course, this is not an exhaustive list of factors that could affect the different results shown in Figure 3, and while a robust empirical analysis of what drives variation in sample estimates for the Big 3 is beyond the scope of this paper, the field could benefit from such an analysis.

Despite differences in the estimated financial literacy rates, multiple studies have found financial literacy varies by demographic and behavioral characteristics. On average, those without a college degree and with lower income tend to have lower financial literacy (Lusardi Reference Lusardi2019). Behaviorally, financial literacy is correlated with activities that are indicative of financial capabilities – such as having emergency savings and saving for retirement (Lin et al. Reference Lin, Bumcrot, Mottola, Valdes, Ganem, Kieffer, Lusardi and Walsh2022). There is also evidence that financial literacy is predictive of downstream financial behaviors. For example, using longitudinal NFCS data from the ALP, Angrisani et al. (Reference Angrisani, Burke, Lusardi and Mottola2023) found that financial literacy was predictive of future financial outcomes. Specifically, respondents with higher levels of financial literacy in 2012 reported that they were more able to handle a financial shock, more likely to plan for retirement, and were more satisfied with their financial condition in 2018. These findings were statistically significant even after controlling for a wide range of variables. However, using longitudinal data from the Health and Retirement Study (HRS), Schmeiser and Seligman (Reference Schmeiser and Seligman2013) found that the Big 3 questions were not significantly related to changes in wealth over time, though the HRS sample only includes those aged 50 and over.

4.2. Other measures of financial capability over time

Unlike changes in financial knowledge, trend data from the State-by-State Survey suggests improvements in other aspects of U.S. adults’ financial capability from 2009 to 2021 (Lin et al. Reference Lin, Bumcrot, Mottola, Valdes, Ganem, Kieffer, Lusardi and Walsh2022). Even considering the economic disruptions of the COVID-19 pandemic in 2020, we find steady increases in the proportion of respondents reporting no difficulty covering monthly expenses or bills (36 percent in 2009 to 54 percent in 2021), a twofold increase in those indicating satisfaction with their personal financial situation (from 16 percent in 2009 to 33 percent in 2021), a rise in those paying their credit cards in full (from 41 percent in 2009 to 59 percent in 2021), and a steady increase in the proportion of respondents who have set aside three months’ worth of emergency funds (35 percent in 2009 to 53 percent in 2021). These improvements in financial capability may reflect the broader economic recovery in the United States and, for the 2018 to 2021 time period, government support during the pandemic. The first wave of the State-by-State Survey, collected in 2009, occurred in the wake of the Great Recession, a particularly challenging economic time in the United States. Increases in financial capability from 2009 to 2021 align well with the economic recovery that transpired post-recession. However, while these findings are encouraging overall, there are wide and persistent disparities across populations. Younger respondents, those with lower educational attainment, and Black/African American and Hispanic/Latino respondents show higher levels of vulnerability across various measures of financial capability.