Financial Well-being and Inclusion of Justice Involved Populations: Evidence from the SHED

Finance and Economics Discussion Series Federal Reserve Board, Washington, D.C. ISSN 1936-2854 (Print) ISSN 2767-3898 (Online) Financial Well-being and Inclusion of Justice Involved Populations: Evidence from the SHED Kabir Dasgupta, Jennifer Fernandez, and Alicia Lloro 2026-024 Please cite this paper as: Dasgupta, Kabir, Jennifer Fernandez, and Alicia Lloro (2026). “Financial Well-being and Inclusion of Justice Involved Populations: Evidence from the SHED,” Finance and EconomicsDiscussionSeries2026-024. Washington: BoardofGovernorsoftheFederalReserve System, https://doi.org/10.17016/FEDS.2026.024. NOTE: Staff working papers in the Finance and Economics Discussion Series (FEDS) are preliminary materials circulated to stimulate discussion and critical comment. The analysis and conclusions set forth are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors. References in publications to the Finance and Economics Discussion Series (other than acknowledgement) should be cleared with the author(s) to protect the tentative character of these papers.

Financial Well-being and Inclusion of Justice Involved Populations: Evidence from the SHED Kabir Dasgupta,† Jennifer Fernandez, and Alicia Lloro May 2026 Abstract This study examines financial challenges faced by justice-involved individuals using 2023-2024 Survey of Household Economics and Decisionmaking data. Individuals with justice system contact experience substantially worse financial outcomes than those without criminal records, with disparities widening by severity of involvement. Compared to individuals with no prior records, those arrested but not convicted are 4 percentage points less likely to report doing at least okay financially, while formerly convicted as well as incarcerated adults are 15 percentage points less likely. Formerly incarcerated individuals are also 21 percentage points less likely to have credit scores above 660 and 13 percentage points less likely to have credit cards. These disparities mirror patterns observed across education levels, where adults with lower educational attainment experience lower financial well-being and inclusion. Our findings document substantial barriers to financial stability among justice-involved populations and may inform policies promoting financial inclusion and improving economic outcomes for this group. JEL classification: K42; G50; I31 Keywords: Justice-involved individuals; financial well-being; financial inclusion; credit score; SHED. † Dasgupta (email: kabir.dasgupta@frb.gov): Senior Economist, Division of Consumer and Community Affairs (DCCA), Federal Reserve Board.  Fernandez (email: jennifer.l.fernandez@frb.gov): Senior Community Development Analyst, Division of Consumer and Community Affairs (DCCA), Federal Reserve Board.  Lloro (email: alicia.lloro@frb.gov): Principal Economist, Division of Consumer and Community Affairs (DCCA), Federal Reserve Board. Note: The results and opinions expressed in this paper reflect the views of the authors and should not be attributed to the Federal Reserve Board or the Federal Reserve System. This draft has been reviewed by Experian and approved for publication. 1

1. Introduction Prior involvement with the criminal justice system, whether through arrest, conviction, or incarceration, can have a profound impact on an individual’s financial well-being. The primary objectives of this study are to identify the various financial challenges facing justice-involved individuals and to explore which financial products and services they can access in managing those challenges. As such, our analysis may inform efforts to foster financial inclusion and improve economic outcomes among the justice-involved population. There are several reasons why justice-involved individuals often experience adverse future economic outcomes (McElhattan, 2024). Existing empirical evidence suggests that people who have interactions with the criminal justice system disproportionately come from socioeconomically disadvantaged backgrounds (Holzer et al. 2007; Jordan et al. 2024). The loss of time due to justice system involvement, especially during incarceration or time served under other forms of custodial sentence, can also substantially erode social and human capital. Such incapacitation effects can reduce future job prospects and economic mobility. As a separate sentencing outcome, monetary sanctions such as legal fees, fines, or restitution orders can further impose substantial debt burden on previously convicted individuals (Harris et al. 2010). Additionally, prior contact with the justice system often subjects individuals to social stigma, which can contribute to significant economic barriers to societal reintegration, particularly when criminal records are easily accessible (Uggen et al., 2014).1 To date, empirical research on the economic effects of justice system involvement has largely focused on employment outcomes (e.g., see Agan and Starr, 2017; Dobbie et al., 2018; Mueller- 1 There have been various federal and state-level initiatives over time such as the “Ban-the-Box” policy and “Clean Slate” initiatives that are intended to restrict or delay employers’ access to individuals’ justice records during hiring processes. However, these interventions have not been found to improve employment prospects for justice-involved individuals (Doleac and Hansen, 2020; Rose 2021; Agan et. al, 2024; Weiss et al, 2025). 2

Smith and Schnepel, 2021). However, limited empirical evidence exists on broader economic conditions and financial well-being beyond labor market consequences. Our study extends the existing literature by examining various financial indicators that reflect the financial health and inclusion of justice-involved individuals. A recent report by the Consumer Financial Protection Bureau (CFPB, 2022) provides a detailed discussion of the financial difficulties faced by families of justice-involved individuals at different stages of involvement with the criminal justice system. The report notes that justice-involved individuals may have an elevated risk of having higher debts, increased likelihood of credit delinquency, and lower credit scores, all of which can affect people’s ability to obtain credit (including small business loans)2, stable housing, and employment. We use data from the Survey of Household Economics and Decisionmaking (SHED) for our empirical analysis. We focus on three aspects of financial well-being: financial health3, financial inclusion,4 and credit scores. For credit score analysis, we consider SHED’s self-reported assessment of individuals’ credit scores, as well as credit bureau records linked with SHED’s sample. While our study relies on associational evidence rather than causal mechanisms, we present a comprehensive empirical analysis of individuals' financial well-being using a wide array of measures. This provides insight into the broader financial conditions experienced by justiceinvolved individuals, which remain underexplored in previous literature. We further contribute by examining how financial well-being varies with the severity of justice system involvement, from arrest without conviction to incarceration following conviction. 2 Recent empirical evidence shows that justice-involved individuals are more likely to turn to self-employment potentially due to the high barriers to formal employment in the labor market (Finlay et al. 2023). 3 Determined as self-reported measures of overall financial condition, credit scores, and emergency preparedness 4 Measured in terms of banking and credit access. 3

Justice-involved individuals exhibit significantly lower financial well-being and are more likely to be liquidity- or credit-constrained compared to individuals with no criminal records. Moreover, financial disadvantages increase with severity of involvement, with formerly convicted and incarcerated individuals experiencing the most acute challenges. For instance, regression analysis controlling for observable individual characteristics shows that compared to individuals with no criminal records, those arrested but not convicted are 4 percentage points less likely to report living comfortably or doing okay financially (“at least okay”), while those convicted and incarcerated are 15 percentage points less likely to report doing at least okay. These disparities follow qualitatively similar patterns to gaps observed across education levels, where adults with lower educational attainment experience poorer financial outcomes compared to college graduates (those with at least a bachelor’s degree). However, while the educational gaps appear to be relatively larger in magnitude for most outcomes, the differences observed between justice-involved adults and those without any prior records are economically significant even after controlling for respondents’ relevant economic and demographic characteristics including educational attainment. 2. Data and Descriptive Statistics The Survey of Household Economics and Decisionmaking (SHED) supports research focusing on personal finances and economic well-being. The SHED is an annual online survey conducted by the Federal Reserve Board each fall since 2013. The survey focuses on a wide range of topics including self-reported assessments of individuals’ personal finances and household economic conditions, emergency preparedness, savings behavior, housing and living arrangements, access and usage of banking and alternative financial services5 (AFS), etc. The annual survey sample is designed to be representative of the U.S. adult population. 5 Such as pawnshop, payday lender, and/or tax refund advance. 4

The questions on individuals’ prior experiences with the justice system were introduced in the 2023 SHED and were also included in the 2024 SHED.6 Specifically, there are three survey questions that were asked of respondents to indicate whether they have ever been “arrested or taken into custody”; “convicted of criminal offense, but never received a prison sentence”; and “convicted of criminal offense, and received a prison sentence.” Using these survey questions, we classify our sample into four categories– people with no history of being involved with the justice system (our reference base group); people who were previously arrested but never received a conviction; those who were previously convicted but never incarcerated; and people who experienced incarceration. We use SHED 2023-2024 for our empirical analysis. For respondents who participated in both the 2023 and 2024 surveys, we consider the most recent observation to make sure we have only one observation per individual. In total, our analysis is based on a sample of 19,680 individuals. Additionally, to compare the share of individuals with a ‘good’ credit score (i.e. scores above 660) across different groups, we link credit information from Experian to approximately 65 percent of respondents who agreed to have their survey responses anonymously matched with their credit history.7 The resulting credit-merged sample includes around 11,000 individuals, along with a snapshot of their credit history from September 2023-2024. 6 While past SHED surveys included questions on justice system involvement, they were more limited than our current analysis. For instance, the 2019-2021 SHED included information on unpaid legal expenses, fines, fees, and court costs. Watson (2023) uses some of these earlier SHED indicators to analyze economic health of justiceinvolved individuals. 7 We use the VantageScore 4.0 for credit score information in merged credit information from Experian to construct the credit score indicator (Larrimore et al., 2024). 5

In Table 1, we present descriptive statistics of our sample classified by their past involvement with the justice system.8 For comparison with the broader SHED sample of adults, we also report the summary statistics for the credit-linked sample. Compared to people with no prior records, justice-involved individuals, particularly those who have received a prison sentence, disproportionately tend to be male, prime-age, Black, and Hispanic. Justice-involved individuals were also less likely to have a bachelor’s degree, own their home, or be married. The corresponding demographic distributions from the credit-merged sample are similar to those in the overall sample of adult respondents. 8 Since, some of the financial well-being outcomes could be associated with people’s labor market participation, we also provide the descriptive summary of prime-aged adult respondents (aged 25-54) from SHED 2023-2024 in Appendix Table A.1. 6

Table 1: Individual-level characteristics of people with prior contact with the justice system – SHED 2023-2024 All adults Credit merged sample Arrested, Arrested, No Convicted, Convicted, No Convicted, Convicted, Characteristics not not record w/o prison w/ prison record w/o prison w/ prison convicted convicted (1) (2) (3) (4) (5) (6) (7) (8) Male 0.44 0.65 0.68 0.78 0.45 0.65 0.68 0.77 White 0.63 0.59 0.62 0.42 0.66 0.63 0.63 0.45 Black 0.11 0.14 0.12 0.23 0.10 0.15 0.13 0.26 Hispanic 0.17 0.20 0.17 0.29 0.16 0.17 0.17 0.23 Age 18-24 0.10 0.03 0.05 0.03 0.07 0.02 0.04 0.02 Age 25-54 0.48 0.53 0.62 0.65 0.50 0.54 0.63 0.66 Age 55-64 0.16 0.20 0.16 0.20 0.15 0.19 0.16 0.22 Age 65 and above 0.24 0.22 0.16 0.11 0.26 0.24 0.16 0.09 Married 0.56 0.49 0.44 0.31 0.58 0.51 0.44 0.36 College degree (BA or more) 0.41 0.23 0.19 0.12 0.44 0.27 0.18 0.13 At least one parent has a college degree 0.32 0.24 0.20 0.17 0.34 0.25 0.20 0.17 Owns home 0.66 0.63 0.52 0.44 0.69 0.65 0.54 0.46 Household: No. of children (<18 years) 0.55 0.54 0.63 0.58 0.55 0.53 0.67 0.60 Household: No. of adults 2.24 2.14 2.22 2.12 2.21 2.08 2.22 2.17 Sample size 17078 1293 912 397 9346 799 564 245 Notes: Authors’ calculations using SHED 2023-2024. The above table presents survey-based shares by individual-level demographic, education, and household characteristics across each group defined by people’s prior contact with the justice system. The descriptive information in columns (1)-(4) is based on all adult respondents aged 18 and above who participated in the SHED 2023 and 2024 surveys. The descriptive information in columns (5)-(8) are based on the credit-merged sample. For individuals who were surveyed in both the surveys, we consider the latest observation. The shares are calculated using the SHED’s person-level weights. We present the summary statistics of prime-aged adults (25-54 years) in Appendix Table A.1. 3. Empirical Findings: Financial Health and Inclusion 3.1 Unadjusted differences in financial outcomes In Table 2, we focus on a range of indicators of financial health and financial inclusion. In total, we consider 12 binary indicators based on responses provided to relevant SHED questions. We provide a detailed description of how the indicators were constructed utilizing various SHED instruments in Appendix Table A.2. Looking at the unadjusted raw differences for the outcomes, justice-involved individuals have poorer perceived financial health and cash liquidity. For instance, while three-fourths of individuals with no criminal justice records managing financially okay or living comfortably, the share progressively goes down by the severity of the judicial outcomes experienced by each group of justice-involved people. For previously incarcerated individuals, the share is less than 7

half (44%). Additionally, while half of the respondents with no prior records can manage an emergency expense exceeding $2,000 using only their savings, that share drops to 17 percent for previously incarcerated people. Consistent with that finding, the share of individuals who skipped medical treatment because of cost is substantially higher for justice-involved people compared to people with no prior contact with the justice system. Regarding measures of financial inclusion, the share of justice involved individuals with a bank account (checking, savings, or money market) or with at least one credit card is lower than the corresponding shares among people with no justice records. Justice-involved individuals, particularly previously incarcerated individuals, are more likely to borrow credit from AFS, pay an overdraft fee, and use a check-cashing service at a non-bank institution. Additionally, while justice-involved individuals are more likely to apply for credit, they are also less likely to exhibit a high level of confidence in having their credit application approved, suggesting that justiceinvolved individuals are more likely to be credit-constrained compared to individuals without prior justice system involvement. While our analysis does not establish causality, the indicators measure outcomes following individuals' contact with the justice system.9 However, a key data limitation is that we are unable measure the time since last justice system interaction, an important predictor of recidivism (Bushway 2024). This information could also enable examination of whether economic challenges persist or diminish over time for justice-involved individuals. 9 The SHED sample is representative of noninstitutionalized, civilian population age 18 and older. 8

Table 2 – Measures of financial health and financial inclusion of individuals classified by prior experiences of being involved in the justice system Arrested, not Convicted, w/o Convicted, w/ No record convicted prison prison (1) (2) (3) (4) Financial health – Well-being and stability Financially doing okay 0.75 0.66 0.54 0.44 Emergency expense >=$2000 0.50 0.43 0.34 0.17 Skipped medical expenses 0.26 0.35 0.43 0.58 Credit score ‘good’ or ‘excellent’ (self-reported)  0.83 0.71 0.59 0.46 Credit score >660; (credit-merge)  0.81 0.69 0.58 0.39 Financial inclusion – Banking and credit access Has a bank account 0.94 0.93 0.89 0.81 Has a credit card 0.84 0.77 0.67 0.54 Used alternative financial services (AFS) 0.05 0.10 0.15 0.28 Non-bank check cashing 0.06 0.09 0.17 0.27 Overdraft fee 0.11 0.16 0.25 0.27 Credit application confidence  0.68 0.58 0.47 0.34 Applied for any credit 0.34 0.39 0.43 0.44 Sample size (range) 9346-17078 799-1293 564-912 245-397 Notes: Authors’ calculations using SHED 2023-2024. The descriptive statistics are calculated using survey weights. Since the sample  sizes may vary by the outcome evaluated, we provide the range of sample sizes used to derive the shares reported above. : Estimates reported for these indicators are based on smaller samples than full samples.  : Estimates reported are based on the credit-merged sample. Among individuals who applied for credit in the year preceding the survey, justice-involved adults were more likely to experience one of three outcomes: outright rejection, partial approval for less than the requested amount, or self-selection out of the application process due to anticipated rejection (Table 3). Additionally, among those who did not apply for any credit in the year preceding the survey, the proportion who desired credit but refrained from applying was significantly higher for individuals with prior contact with the justice system. This share was the highest among the individuals who were previously incarcerated (32%) and the lowest for people with no prior records (9%). Finally, justice-involved individuals have lower credit scores than people without any past criminal records. Based on the credit-merged sample, the share of people who have a credit score above 660 was significantly greater for people who never came in contact with the justice system 9

(80%) than those who were incarcerated as an outcome of their conviction (36%).10 These findings are consistent with the self-reported evaluations of respondents’ credit scores in the annual SHED 2023-2024. Table 3 – Outcomes experienced by those who applied for credit and reasons cited by those who did not apply for credit No Arrested, not Convicted, Convicted, record convicted w/o prison w/ prison (1) (2) (3) (4) Those who did not apply Desired credit but did not apply 0.09 0.19 0.23 0.32 Sample size 11416 804 530 222 Those who applied Turned down for credit 0.22 0.32 0.47 0.70 Approved but not given the full amount applied 0.16 0.21 0.29 0.31 Put off applying to fear of rejection or partial approval 0.18 0.26 0.35 0.50 Sample size 5662 489 382 175 Notes: Authors’ calculations using SHED 2023-2024. The descriptive statistics are calculated using survey weights. 3.2 Regression analysis The raw differences across the groups considered in Table 2 hold across most indicators when we incorporate a regression framework to adjust for differences in various observable characteristics across different groups of justice-involved individuals. We present our regressionbased findings in Table 4. The covariate-adjusted regressions include controls for demographic attributes such as age, sex, race, and ethnicity, educational attainment, marital status, household size and state-specific differences that could potentially capture variations in state-specific criminal justice policies. 10 We use the Vantage credit score information in the merged credit data from Experian to construct the credit score indicator. 10

Table 4 – Regression-adjusted differences in the financial well-being outcomes experienced by justice-involved individuals (1) (2) (3) (4) (5) (6) (7) Panel A – Financial health Financially Emergency Skipped Good/ Credit score > - doing okay expenses medical Excellent 660 >= $2000 credit score (Credit merge) Arrested, not convicted -0.037*** -0.013 0.068*** -0.065*** -0.063*** (0.014) (0.013) (0.014) (0.014) (0.015) Convicted, w/o prison sentence -0.118*** -0.075*** 0.127*** -0.142*** -0.103*** (0.017) (0.016) (0.017) (0.017) (0.018) Convicted, w/ prison sentence -0.154*** -0.178*** 0.240*** -0.183*** -0.205*** (0.026) (0.028) (0.027) (0.029) (0.029) Observations 19,668 19,665 19,679 14,854 10,948 Sample mean 0.734 0.510 0.271 0.822 0.803 Panel B – Financial inclusion Has bank Has credit Used AFS Non-bank Overdraft fee Credit Applied account card check application for credit cashing confidence Arrested, not convicted 0.000 -0.030*** 0.036*** 0.021** 0.040*** -0.051*** 0.058*** (0.007) (0.011) (0.008) (0.008) (0.012) (0.014) (0.016) Convicted, w/o prison sentence -0.019** -0.079*** 0.067*** 0.083*** 0.110*** -0.119*** 0.102*** (0.009) (0.014) (0.011) (0.012) (0.016) (0.017) (0.019) Convicted, w/ prison sentence -0.045*** -0.130*** 0.133*** 0.138*** 0.099*** -0.164*** 0.104*** (0.014) (0.021) (0.019) (0.021) (0.023) (0.028) (0.028) Observations 19,609 19,679 19,484 19,632 18,674 18,769 19,679 Sample mean 0.949 0.846 0.051 0.063 0.110 0.685 0.341 Notes: Authors’ calculations using SHED 2023-2024. All outcome variables considered in the above regression analysis are binary indicators. The credit score information used in column (5), panel A is VantageScore 4.0. The main explanatory variable is a categorical indicator denoting each group considered in our analysis. The sample of adults with no criminal record is the reference group (omitted category). The above regression estimates represent marginal effects obtained from Probit regressions. Robust standard errors correcting for heteroskedasticity are estimated and reported in parentheses. In each regression, we control for individual-level characteristics including indicators of sex, age, race, ethnicity, marital status, educational attainment, and household size (including number of children aged under 18 and adults in the respondents’ households), state fixed effects, and survey year fixed effects. *** p<0.01, ** p<0.05, * p<0.1. In Table 4, we report the marginal effects derived from non-linear (Probit) regressions. According to the results reported in Table 4 Panel A, justice-involved individuals are more likely to experience lower financial well-being, emergency preparedness, and credit scores compared to adults with no prior records (reference group). They are also more likely to skip medical expenses due to lack of affordability. In most cases, differences from the reference group increase with the severity of prior justice system involvement, with the largest disparities observed for individuals with prior convictions, particularly those who experienced incarceration. For instance, while justice-involved individuals with an arrest record but no 11

conviction were 4 percentage points less likely to either live comfortably or do okay financially (column 1) and 6 percentage points less likely to have a credit score above 660 (column 5) when compared to people with no records, these differences rise to roughly 15 percentage points (overall financial well-being) and 21 percentage points (above 660 credit score) for those who experienced incarceration. In line with the raw differences in Table 2, the marginal effects in Panel B indicate that justiceinvolved individuals also exhibit lower rates of financial inclusion–they are less likely to have bank accounts or credit cards and more likely to use alternative financial services such as payday or pawnshop loans and to pay overdraft fees. Despite having greater credit needs, justiceinvolved individuals express lower confidence in credit application approval. Compared to respondents with no prior records, those who were convicted and incarcerated are 10 percentage points more likely to have applied for credit, but 16 percentage points less likely to express confidence in approval, while those arrested but never convicted are 6 percentage points more likely to have applied but 5 percentage points less confident in approval. It is worth noting that controlling for individual-level observable characteristics accounts for substantial portions of the raw or baseline financial well-being gaps across different groups of justice-involved individuals. We report our baseline regression estimates in Appendix Table A.4, which exclude the individual-level covariates and state fixed effects considered in Table 4 analysis. To elaborate, Table A.4 shows that compared to people with no prior contact with the justice system, previously convicted individuals who experienced incarceration were 30 percentage points less likely to either live comfortably or do okay financially and 42 percentage points less likely to likely to have a credit score above 660–almost twice the corresponding differences seen in the saturated regression models in Table 4. 12

For further insights, we perform a supplemental analysis to estimate differences in financial wellbeing and inclusion by educational attainment, with adults holding at least a bachelor’s degree serving as the reference group. The marginal effects highlighting differences in the indicators of financial well-being and inclusion across education levels are reported in Appendix Table A.5 and provide important context for interpreting the magnitude of disparities associated with justice system involvement shown in Table 4. Notably, the financial difficulties associated with justice system involvement are qualitatively consistent with the gaps observed across education levels.11 However, the differences across educational levels for most outcomes appear to be larger in magnitude than the comparable differences observed across different types of justiceinvolved individuals. Finally, as further shown in Appendix Table A.6, our key findings remain largely consistent when we restrict our regression analysis to prime-aged adults. 4. Conclusion Justice-involved individuals, particularly the previously incarcerated, experience lower overall financial well-being, worse credit outcomes, and lower rates of bank account ownership. Several mechanisms likely drive these adverse outcomes. Many justice-involved individuals come from economically disadvantaged backgrounds, suggesting pre-existing conditions may partially explain their long-term financial outcomes. However, even after controlling for age, education, and other observable characteristics, we find statistically and economically significant gaps in financial well-being and inclusion between justice-involved individuals and those without any prior records. While not causal, our regression results–taken together with existing literature on employment outcomes–offer suggestive evidence that justice system contact leads to adverse 11 We perform this analysis based on adults with no prior records, in order to remove possible underlying effects of prior contact with the justice system. 13

financial outcomes through multiple channels, including skill erosion, weakened social connections, heightened risk of social stigma, and persistent financial obligations such as courtimposed fines and fees. Overall, given the high prevalence of justice system interactions in the US population (Umez and Pirius, 2018), our findings have important implications for household financial stability. In particular, the findings highlight the relevance of financial inclusion and financial health needs of justice-involved populations, particularly for individuals seeking to reintegrate into society. The analysis suggests promising avenues for future research to examine whether interventions such as credit building products, second chance loans, and improvements in bank account access can help reduce financial well-being gaps between justice-involved individuals and those without prior justice system records. These findings can also inform future policy considerations for identifying strategies that ensure better social reintegration of justice-involved individuals. 14

References: Jordan, Andrew, Ezra Karger, and Derek Neal. Early Predictors of Racial Disparities in Criminal Justice Involvement. No. w32428. National Bureau of Economic Research, 2024. Holzer, Harry J., Diane Whitmore Schanzenbach, Greg J. Duncan, and Jens Ludwig. "The economic costs of poverty in the United States: Subsequent effects of children growing up poor." Washington, DC: Center for American Progress (2007). Harris, Alexes, Heather Evans, and Katherine Beckett. "Drawing blood from stones: Legal debt and social inequality in the contemporary United States." American journal of sociology 115, no. 6 (2010): 1753-1799. McElhattan, David. "Criminal background check laws and labor market inequality in the United States." Criminology & Public Policy 23, no. 2 (2024): 391-429. Uggen, Christopher, Mike Vuolo, Sarah Lageson, Ebony Ruhland, and Hilary K. Whitham. “The edge of stigma: An experimental audit of the effects of low‐level criminal records on employment.” Criminology 52, no. 4 (2014): 627-654. Weiss, Deborah M., William Dawson, Ronald McKinley, and Lee Webster. "Do ban-the-box policies increase the hiring of applicants with criminal records?." Plos one 20, no. 4 (2025): e0320736. Doleac, Jennifer L., and Benjamin Hansen. "The unintended consequences of “ban the box”: Statistical discrimination and employment outcomes when criminal histories are hidden." Journal of Labor Economics 38, no. 2 (2020): 321-374. Rose, Evan K. "Does banning the box help ex-offenders get jobs? Evaluating the effects of a prominent example." Journal of Labor Economics 39, no. 1 (2021): 79-113. Agan, Amanda Y., Andrew Garin, Dmitri K. Koustas, Alexandre Mas, and Crystal Yang. Can you erase the mark of a criminal record? labor market impacts of criminal record remediation. No. w32394. National Bureau of Economic Research, 2024. Agan, Amanda, and Sonja Starr. "The effect of criminal records on access to employment." American Economic Review 107, no. 5 (2017): 560-564. Dobbie, Will, Jacob Goldin, and Crystal S. Yang. "The effects of pre-trial detention on conviction, future crime, and employment: Evidence from randomly assigned judges." American Economic Review 108, no. 2 (2018): 201-240. Mueller-Smith, Michael, and Kevin T. Schnepel. "Diversion in the criminal justice system." The Review of Economic Studies 88, no. 2 (2021): 883-936. Financial, Consumer, and Protection Bureau. "Justice-Involved Individuals and the Consumer Financial Marketplace." (2022). 15

Finlay, Keith, Michael Mueller‐Smith, and Brittany Street. "Criminal Justice Involvement, Self‐ Employment, and Barriers in Recent Public Policy." Journal of Policy Analysis and Management 42, no. 1 (2023): 11-34. Larrimore, Jeff, Alicia Lloro, Zofsha Merchant, and Anna Tranfaglia. "“The Only Way I Could Afford It”: Who Uses BNLP and Why." FEDS Notes 2024-12 (2024): 20-4. Umez, Chidi, and Rebecca Pirius. "Barriers to work: Improving employment in licensed occupations for individuals with criminal records." In National Conference on State Legislatures. https://documents. ncsl. org/wwwncsl/Criminal-Justice/Barriers-to-Work-People-with-Criminal- Records. pdf. 2018. Watson, Spencer. Rep. Economic Wellbeing of U.S. Adults with Experiences with Incarceration & Unpaid Legal Costs. First Step Alliance, November 26, 2023. https://www.firststepalliance.org/post/first-step-alliance-report-economic-health-of-justiceinvolved-individuals. Bushway, Shawn D. "Resetting the Record: The Facts on Hiring People with Criminal Histories." RAND Research Brief, RBA2968-1 (2024). 16

APPENDIX Table A.1: Individual-level characteristics of prime-age adults (25-54) with prior contact with the justice system – SHED 2023-2024 Arrested, not Convicted, Convicted, Characteristics No record convicted w/o prison w/ prison Male 0.44 0.59 0.65 0.76 White 0.57 0.54 0.60 0.38 Black 0.12 0.14 0.12 0.23 Hispanic 0.20 0.26 0.18 0.31 Married 0.57 0.43 0.41 0.28 College degree (BA or more) 0.48 0.25 0.19 0.11 At least one parent has a college degree 0.38 0.28 0.21 0.18 Own home 0.60 0.51 0.44 0.42 Household: No. of children (<18 years) 0.90 0.88 0.86 0.75 Household: No. of adults 2.22 2.19 2.25 2.26 Sample size 7435 600 522 230 Notes: Authors’ calculations using SHED 2023-2024. The above table presents survey-based shares by individual-level demographic, education, and household characteristics across each group defined by people’s prior contact with the justice system. The descriptive information is based on all adult respondents aged 25-54 who participated in the SHED 2023 and 2024 surveys. For individuals who were surveyed in both the surveys, we consider the latest observation. The shares are calculated using the SHED’s person-level weights. 17

Table A.2: Construction of binary indicators of financial health and inclusion using survey questions from the SHED 2023-2024 Indicator Survey question Coding – Response options Financially doing okay Overall, which one of the following best describes 1: “Living comfortably” or how well you are managing financially these “Doing okay” days? 0: “Just getting by” or “Finding it difficult to get by” Emergency expense >=$2000 Based on your current financial situation, what is 1: $2000 or more the largest emergency expense that you could 0: for all response options under handle right now using only your savings? $2000 Skipped medical expenses Based on a series of five questions related 1: if respondent chose “Yes” to to “Prescription medicine”; “Seeing a any of the five questions. doctor or specialist”; “Mental health care 0: Otherwise. or counseling”; “Dental care”; and “Follow-up care”. During the past 12 months, was there a time when you needed each of the following, but went without because you couldn’t afford it? Credit score ‘good’ or ‘excellent’ Where do you think your credit score falls? 1: Good/ Excellent 0: Fair/Poor/Very poor Has bank account Do you (and/or your spouse or partner) 1: Yes currently have a checking, savings or money 0: No market account? Has credit card Do you currently have at least one credit card? 1: Yes Please do not include debit cards or prepaid 0: No cards. Used alternative financial services Based on a series of three questions: 1: if respondent selected “yes” “In the past 12 months, did you (and/ or your to any of the three questions. spouse or partner) take out: 0: Otherwise - take out a payday loan or payday advance - take out a pawn shop loan or an auto title loan - obtain a tax refund advance Cashed check at non-bank Cash a check at a place other than a bank - In the 1: Yes organizations past 12 months, did you (and/or your spouse or 0: No partner) Overdraft fee Pay an overdraft fee on a bank account - In the 1: Yes past 12 months, did you (and/or your spouse or 0: No partner) Credit application confidence If you were to apply for a credit card today, how 1: Very confident confident are you that your application would be 0: Somewhat or not confident approved? Applied for any credit In the past 12 months have you applied for any 1: Yes credit (such as a credit card, higher credit card 0: No limit, mortgage, refinance, student loan, personal loan, or other loan)? Note: The above table provides a detailed description of the survey questions utilized to create the binary indicators of financial health, financial inclusion, and credit applications that are reported in Table 2. 18

Table A.3: Measures of financial health and financial inclusion of prime-aged adults, classified by prior experiences of being involved in the justice system Arrested, not Convicted, Convicted, No record convicted w/o prison w/ prison (1) (2) (3) (4) Financial health – Wellbeing and stability Financially doing okay 0.70 0.60 0.46 0.40 Emergency expense >=$2000 0.46 0.33 0.26 0.13 Skipped medical expenses 0.31 0.42 0.50 0.63 Credit score ‘good’ or ‘excellent’ (self-reported)  0.78 0.63 0.50 0.46 Credit score >660; credit-merge)  0.76 0.59 0.50 0.34 Financial inclusion – Banking and credit access Has bank account 0.93 0.91 0.87 0.79 Has credit card 0.82 0.71 0.62 0.51 Used alternative financial services 0.07 0.16 0.18 0.33 Non-bank check cashing 0.06 0.12 0.21 0.28 Overdraft fee 0.14 0.22 0.28 0.31 Credit application confidence  0.64 0.47 0.41 0.29 Applied for any credit 0.38 0.43 0.46 0.48 Sample size (range) 4128-7435 379-600 323-522 145-230 Notes: Authors’ calculations using SHED 2023-2024. The descriptive statistics are based on respondents aged 25- 54 and are calculated using survey weights. Since the sample sizes may vary by the outcome evaluated, we provide  the range of sample sizes used to derive the shares reported above. : Estimates reported for these indicators are based on samples smaller than full samples.  : Estimates reported are based on the credit-merged sample. 19

Table A.4: Baseline differences in the financial well-being outcomes experienced by justice-involved prime-aged adults without controlling for individual-level characteristics (1) (2) (3) (4) (5) (6) (7) Panel A – Financial health Financially Emergency Skipped Good/ Credit score - doing okay expenses medical Excellent > 660 >= $2000 credit score (Credit merge) Arrested, not convicted -0.088*** -0.067*** 0.087*** -0.115*** -0.115*** (0.015) (0.016) (0.015) (0.017) (0.019) Convicted, w/o prison sentence -0.204*** -0.163*** 0.172*** -0.241*** -0.227*** (0.019) (0.017) (0.019) (0.022) (0.024) Convicted, w/ prison sentence -0.303*** -0.332*** 0.317*** -0.371*** -0.421*** (0.028) (0.021) (0.028) (0.033) (0.034) Observations 19,680 19,680 19,680 14,855 10,954 Sample mean 0.734 0.509 0.271 0.822 0.802 Panel B – Financial inclusion Has bank Has credit Used AFS Non-bank Overdraft fee Credit Applied account card check application for credit cashing confidence Arrested, not convicted -0.015* -0.068*** 0.055*** 0.034*** 0.055*** -0.104*** 0.051*** (0.009) (0.014) (0.010) (0.010) (0.012) (0.016) (0.016) Convicted, w/o prison sentence -0.051*** -0.167*** 0.101*** 0.114*** 0.144*** -0.210*** 0.097*** (0.012) (0.018) (0.014) (0.015) (0.018) (0.019) (0.019) Convicted, w/ prison sentence -0.133*** -0.303*** 0.237*** 0.212*** 0.162*** -0.343*** 0.102*** (0.024) (0.028) (0.026) (0.025) (0.027) (0.027) (0.028) Observations 19,680 19,680 19,680 19,680 18,683 18,774 19,680 Sample mean 0.949 0.846 0.050 0.063 0.110 0.685 0.341 Notes: Authors’ calculations using SHED 2023-2024. All outcome variables considered in the above regression analysis are binary indicators. The main explanatory variable is a categorical indicator denoting each group considered in our analysis. The sample of adults with no criminal record is the reference group (omitted category). The above regression estimates represent marginal effects obtained from Probit regressions. Robust standard errors correcting for heteroskedasticity are estimated and reported in parentheses. In each regression, we only control for survey year fixed effects. *** p<0.01, ** p<0.05, * p<0.1. 20

Table A.5: Regression-adjusted differences in the financial well-being outcomes by education levels (1) (2) (3) (4) (5) (6) (7) Panel A – Financial health Financially Emergency Skipped Good/ Credit score - doing okay expenses medical Excellent > 660 >= $2000 credit score (Credit merge) Some college -0.150*** -0.225*** 0.112*** -0.124*** -0.151*** (0.008) (0.009) (0.008) (0.007) (0.009) High school graduate -0.213*** -0.336*** 0.110*** -0.210*** -0.225*** (0.009) (0.010) (0.009) (0.010) (0.012) Less than high school -0.298*** -0.442*** 0.177*** -0.323*** -0.345*** (0.019) (0.019) (0.018) (0.022) (0.025) Observations 17,069 17,066 17,078 12,927 9325 Sample mean 0.755 0.528 0.252 0.845 0.830 Panel B – Financial inclusion Has bank Has credit Used AFS Non-bank Overdraft fee Credit Applied account card check application for credit cashing confidence Some college -0.024*** -0.080*** 0.032*** 0.034*** 0.066*** -0.162*** -0.001 (0.003) (0.006) (0.004) (0.004) (0.006) (0.009) (0.010) High school graduate -0.073*** -0.191*** 0.043*** 0.038*** 0.063*** -0.238*** -0.076*** (0.005) (0.008) (0.005) (0.005) (0.007) (0.010) (0.010) Less than high school -0.170*** -0.381*** 0.079*** 0.084*** 0.096*** -0.382*** -0.121*** (0.013) (0.018) (0.010) (0.011) (0.016) (0.021) (0.017) Observations 16,774 17,078 16,771 17,000 16,293 16,292 17,078 Sample mean 0.954 0.864 0.040 0.053 0.099 0.707 0.332 Notes: Authors’ calculations using SHED 2023-2024. All outcome variables considered in the above regression analysis are binary indicators. The main explanatory variable is a categorical indicator denoting different levels of educational attainment ranging from those with no high school education to college graduates. The sample of adults with at least bachelor’s degree is the reference group (omitted category). The above regression estimates represent marginal effects obtained from Probit regressions. Robust standard errors correcting for heteroskedasticity are estimated and reported in parentheses. In each regression, we control for individual-level characteristics including indicators of sex, age, race, ethnicity, marital status, educational attainment, and household size (including number of children aged under 18 and adults in the respondents’ households), state fixed effects, and survey year fixed effects. *** p<0.01, ** p<0.05, * p<0.1. 21

Table A.6: Regression-adjusted differences in the financial well-being outcomes experienced by justice-involved prime-aged adults (25-54 years) (1) (2) (3) (4) (5) (6) (7) Panel A – Financial health Financially Emergency Skipped Good/ Credit score - doing okay expenses medical Excellent > 660 >= $2000 credit score (Credit merge) Arrested, not convicted -0.024 -0.026 0.075*** -0.070*** -0.088*** (0.020) (0.020) (0.022) (0.021) (0.024) Convicted, w/o prison sentence -0.146*** -0.076*** 0.149*** -0.162*** -0.124*** (0.023) (0.021) (0.024) (0.025) (0.025) Convicted, w/ prison sentence -0.148*** -0.155*** 0.248*** -0.126*** -0.211*** (0.034) (0.039) (0.035) (0.036) (0.039) Observations 8,782 8,780 8,784 6,431 4,973 Sample mean 0.668 0.439 0.337 0.759 0.731 Panel B – Financial inclusion Has bank Has credit Used AFS Non-bank Overdraft fee Credit Applied account card check application for credit cashing confidence Arrested, not convicted 0.006 -0.028* 0.052*** 0.030** 0.046** -0.066*** 0.062*** (0.011) (0.016) (0.014) (0.012) (0.018) (0.021) (0.023) Convicted, w/o prison sentence -0.023* -0.084*** 0.086*** 0.112*** 0.113*** -0.120*** 0.093*** (0.013) (0.018) (0.017) (0.018) (0.022) (0.022) (0.025) Convicted, w/ prison sentence -0.039** -0.111*** 0.166*** 0.129*** 0.109*** -0.150*** 0.122*** (0.019) (0.027) (0.029) (0.026) (0.034) (0.037) (0.037) Observations 8,700 8,782 8,659 8,689 8,141 8,365 8,782 Sample mean 0.928 0.807 0.079 0.0770 0.153 0.614 0.394 Notes: Authors’ calculations using SHED 2023-2024. All outcome variables considered in the above regression analysis are binary indicators. The main explanatory variable is a categorical indicator denoting each group considered in our analysis. The sample of adults with no criminal record is the reference group (omitted category). The above regression estimates represent marginal effects obtained from Probit regressions. Robust standard errors correcting for heteroskedasticity are estimated and reported in parentheses. In each regression, we control for individual-level characteristics including indicators of sex, age, race, ethnicity, marital status, educational attainment, and household size (including number of children aged under 18 and adults in the respondents’ households), state fixed effects, and survey year fixed effects. *** p<0.01, ** p<0.05, * p<0.1. 22