MEASUREMENT / SCORE TOOLKIT

Methodology Overview

The Financial Health Network’s FinHealth ScoreTM is a measurement framework that can be used to assess the financial health of individuals or groups of individuals over time. This page presents an overview of how the FinHealth Score was developed and the theoretical framework upon which it is based.

Financial Health Framework

Definition of Financial Health

The FinHealth Score is based upon the following definition of financial health:

Financial health comes about when your daily financial systems allow you to be resilient and pursue opportunities over time.

The Financial Health Network introduced this definition of financial health to serve as a composite framework that considers the totality of an individual’s financial life. Unlike narrow metrics like credit scores, financial health considers whether individuals are spending, saving, borrowing, and planning in a way that will either contribute to, or detract from, their resilience in the face of unexpected events and ability to thrive in the long term.

Eight Indicators of Financial Health

The Financial Health Network developed eight indicators to help organizations measure financial health. The indicators were informed by existing data from 20 consumer finance studies and consultations with more than 85 financial services providers and industry experts.

For more on how the Financial Health Network developed these eight indicators, please read Eight Ways to Measure Financial Health.

The FinHealth ScoreTM

Overview of the Framework

The FinHealth Score is based upon the theoretical framework presented above. For every individual who responds to eight survey questions that align with the eight indicators of financial health, we can calculate one FinHealth Score and four sub-scores.

FinHealth Scores and sub-scores should be interpreted in the following way:

  • Financial health scores between 0 – 39 are considered Financially Vulnerable. Individuals with scores in this range report healthy outcomes across few, or none, of the eight financial health indicators.
  • Financial health scores between 40 – 79 are considered Financially Coping. Individuals with scores in this range report healthy outcomes across some, but not all, of the eight financial health indicators.
  • Financial health scores between 80 – 100 are considered Financially Healthy. Individuals with scores in this range report healthy outcomes across all eight financial health indicators.

To understand how tier cut-offs were developed, see Identifying Financial Health Tiers.

Developing the Survey Questions

To develop the survey questions underlying the FinHealth Score, the Financial Health Network first conducted extensive industry research to identify questions from existing studies that corresponded with the eight indicators of financial health. We considered survey instruments from FINRA’s National Capability Study, the FDIC’s National Survey of Unbanked and Underbanked Households, and the Federal Reserve’s Survey of Consumer Finances and Report on the Economic Well-Being of U.S. Households.

In 2016, we fielded a nationally representative benchmarking study (n = 400) to test the draft survey questions underlying the FinHealth Score framework. We then analyzed response patterns to identify which questions yielded intuitive results and which questions could be clarified and improved.

In 2017, we hired an external research firm to test the revised survey questions with a select group of low-income and financially struggling consumers. The firm conducted in-depth interviews with consumers in San Francisco and Detroit to ensure that respondents accurately interpreted each of the survey questions.

In 2018, we partnered with the University of Southern California to conduct a second round of cognitive testing with members of the nationally representative Understanding America Study panel. The survey questions were simplified and streamlined into the final versions that now comprise the FinHealth Score.

Constructing the Scoring Logic

To develop the scoring logic for the FinHealth Score, we conducted a comprehensive scan of scoring tools available at that time. We considered the CFPB’s Financial Well-Being Scale, the University of Wisconsin’s Financial Capability Scale, Momentum’s U.K. Financial Wellness Index, and Bain & Company’s Net Promoter Score (NPS). We also considered proprietary models, such as those available from Melius, HelloWallet, NerdWallet, and USAA. Based on this research, as well as consultations with industry experts, we based our FinHealth Score on an 100-point scale. We chose this scale because of its straightforward and intuitive nature.

Weighting the Indicators

For simplicity, we elected to weight all indicator and component response values equally. While we found a lack of evidence to suggest an alternative approach, this decision may be revisited as we continue to research financial health measurement. Any potential methodology revisions could be applied retroactively, ensuring consistent measurement across years.

Determining Response Values

We drew upon internal and external financial health expertise to assign point values to each question response. For example, since the vast majority of Americans pay all of their bills on time, only paying some bills on time should be considered a sign of financial distress. As such, there is a larger gap between the healthiest response in Indicator #2 — “Pay all of our bills on time” (100 points) and the next healthiest response “Pay nearly all of our bills on time” (60 points) — than there would be in a standard summative scale where all response options are weighted using an ordinal scale. The goal of this exercise was to develop a normative score that would a) capture gradations in financial health, particularly at the lower end of the financial health spectrum, and b) serve as an absolute scale, not changing as the population mean might shift, for example.

In 2017, we tested a first draft of this scoring logic in a second nationally representative consumer study (n = 5,000). Based on the results of this study, we adjusted the response values, assessing how different approaches affected the mean, median, and overall distribution of financial health scores across the country.

Identifying Financial Health Tiers

Once we finalized the scoring logic, we employed a number of analytical techniques to identify financial health tiers:

  1. We conducted a cluster analysis to determine where segment cut-offs naturally occurred within the data set. Using the eight scored indicator questions as inputs, we conducted a k-means cluster analysis. Three clusters were determined to be the optimal number based on a screeplot.
  2. To ensure that this empirical analysis yielded conceptually consistent results, we also developed “archetypes” based on the U.S. Financial Diaries to evaluate whether the hypothetical response pattern for each archetype generated scores that were logically aligned with our expert diagnosis of the archetype’s financial health.
  3. We then assessed all possible scores that might arise from different combinations of responses to ensure that the hypothesized tier cut-offs of 80 and 40 would not yield any counterintuitive scoring results. (For example, if a person were to select five “vulnerable” responses, but still receive a healthy score.) While the archetypes exercise described above allowed us to assess intuitive answer combinations, this analysis allowed us to rule out counter-intuitive scoring combinations.
  4. As a final step, we coded responses in the 2017 dataset to the eight indicator questions as healthy, coping, or vulnerable based on the response scoring. We then plotted the number of “healthy,” “coping,” and “vulnerable” responses per respondent to examine the patterns between the coded responses and overall financial health score. This exercise showed us the average number of “healthy” responses (in most cases the top one or two response options) started to exceed the average number of “coping” responses (the middle responses) and “vulnerable” responses (the bottom one or two responses) at a score of 80. Likewise, at a score of 40, the average number of “healthy” and “coping” responses started to exceed the average number of “vulnerable” responses.

Considered together, these analyses allowed us to determine that the Healthy/Coping tier cut-off was located at a score of 80, and the Coping/Vulnerable tier cut-off was located at a score of 40. We performed a similar set of analytical exercises to identify tier cut-offs for Spend, Save, Borrow, and Plan sub-scores and determined that the same cut-offs of 40 and 80 applied to the sub-scores as well.

However, it should be noted that there are likely few material differences between an individual with a score of 80 and a score of 79, or between individuals with scores of 40 and 39. Each of these individuals is likely to be on the cusp of financial health and could benefit from additional guidance and access to high-quality financial products and services.

Testing Alternate Scoring Logic

In 2019, we tested several alternative scoring methodologies in order to assess the rigor and consistency of our FinHealth Score. Using the U.S. Financial Health Pulse dataset, we compared the different methodologies’ distributions, descriptive statistics, and correlations. This analysis was designed to reveal if there were anomalies, biases, or large differences in how the FinHealth Score measures financial health relative to alternative approaches. This work builds on the validation analytics conducted in the U.S. Financial Health Pulse: 2018 Baseline Survey, where we found the FinHealth Score™ to be highly correlated with the CFPB’s Financial Well-Being Scale (see pages 55-58 in Appendix of the report).

We explored the following alternative methodological approaches for constructing the score:

 

 

Summative Score

 

We first used a typical methodology for creating simple scales, where the categorical responses to each question are added together to create a simple summation. The resulting scale is used as the output measure. Whereas the original FinHealth Score methodology involved assigning a specific point value to a given response, the summative score methodology assumes that the scale for all response options is consistent. For example, for indicator question #2, the FinHealth Score methodology scores the responses with values 100, 60, 40, 20, 0, such that the scale between responses is not consistent. For the summative score methodology, the values used for those responses would be consistent: 5, 4, 3, 2, 1.

For this process, the 8 indicator variables were re-coded as consistent likert scales, so that a larger summative score would be indicative of greater financial health. Each indicator variable was then normalized so that it had a mean of zero and standard deviation of 1, and then summed to create the scale. Finally, this measure was rescaled to range between 0 and 100 to allow for comparison to the FinHealth Score. This process yielded a score with a mean of 65.1 and standard deviation of 21.0. (see the table below for more summary statistics and a comparison across the scoring methodologies)

 

Initial Principal Components Analysis

 

Using the 2018 Pulse dataset, we conducted a principal components analysis (PCA) on the eight indicator questions. PCA is a dimension reduction procedure where the information from the input variables can be broken down into the underlying traits or components that best explain the input data. Our analysis revealed only one underlying component, which supported our design and use of the 8 indicator questions — they all contributed to the measurement of the underlying latent trait of financial health.

We rescaled the PCA-derived score to range between 0 and 100, allowing for comparison to the FinHealth Score. The rescaled component had a mean score of 65.3 and a standard deviation of 21.3. (See the table below for more summary statistics and a comparison across the scoring methodologies.)

 

 

Comparisons Between Score Methodologies

Alongside the summary statistics and high correlations between the alternate scales, the distributions of the different scoring methodologies were very similar, indicating that the FinHealth Score does not substantively differ from the alternative scoring approaches.

We plotted the distributions of the rescaled scores to determine if there were differences not addressed in the summary statistics. While the FinHealth Score has slightly greater density toward the extremes of the score range relative to the Summative score and PCA score, the distributions were very similar overall.

We also compared the score distributions of the different methodologies across responses to other questions in the Pulse dataset in order to assess the consistency of the methodologies across variables that are correlated with financial health. We found the methodologies to be consistently correlated with the key variables: for example, with respondents’ satisfaction with their financial situation (see example below). 

Conclusions from Comparisons of FinHealth Score™ to Alternate Methodologies

The original FinHealth Score methodology yielded scores and results consistent with other scoring methodologies, further suggesting that the FinHealth Score is an accurate and reliable measure of financial health. Given these findings, we elected not to change the FinHealth Score methodology, allowing us to preserve the simplicity and continuity of the measurement tool.

Looking Ahead

However, the FinHealth Score is meant to be a starting point; a proof point that financial health can be measured and ultimately improved. In the coming years, the Financial Health Network will continue to work with industry experts, academics, and network members to test and validate the measurement tool. Through the U.S. Financial Health Pulse, we are also exploring how to augment this tool with financial data. Please direct any questions or feedback about the FinHealth Score to finhealth@finhealthnetwork.org.

 

Last Updated: November 2019

Get the Toolkit

Complete the form below to start measuring and improving the financial lives of your customers and employees.