Examining persistent inequities in financial complaint resolution for older Americans and veterans in the United States

Abstract

The Consumer Financial Protection Bureau (CFPB) maintains a centralized complaint intake and routing system, yet whether company response timeliness differs across demographic groups and neighborhoods remains under-investigated. We analyze over 1.27 million consumer financial complaints submitted between 2014 and 2022 to assess whether older adults and service members face disparities in company response timeliness. Logistic regression models of company response timeliness in the CFPB portal show that while the predicted probability of a timely response improved from 97.7% in 2014 to 99.4% in 2022 for the general public, disparities remained. Service members consistently had lower probabilities of timely response, with gaps of 0.3 to 1.8 percentage points. Older adults shifted from a slight advantage in 2014 (+0.5 points) to a disadvantage in 2022 (−0.6 points). Consumers in disadvantaged neighborhoods also had lower predicted probabilities of timely company response, with the largest descriptive disparities observed among older adults.

Introduction

Many US households remain financially vulnerable in the years following the 2008 crisis (1). Older adults face elevated risk from fixed incomes, cognitive decline, and fraud exposure (2, 3), while service members face transition-related hardships and complex lending environments (4, 5). In 2024, older victims reported $4.9 billion in losses to the Federal Bureau of Investigation Internet Crime Complaint Center (6). Even after accounting for underreporting, the Federal Trade Commission estimates fraud costs to older adults may reach $10.1–81.5 billion (7).

When disputes or fraud-related losses occur, consumers may turn to formal complaint channels for redress. A key question is whether these workflows show comparable responsiveness when older adults and service members seek help. The Consumer Financial Protection Bureau (CFPB) runs a centralized process for collecting and forwarding consumer complaints about regulated financial products, and its public Consumer Complaint Database provides a large-scale record of complaints and related workflow indicators. In this system, a public indicator reflects whether the “company” met CFPB response windows in the complaint workflow, rather than the speed or quality of CFPB investigations. Using over 1.27 million complaints from 2014 to 2022, we test whether older adults and service members experience disparities in company response timeliness and whether neighborhood socioeconomic disadvantage, measured by the area deprivation index (ADI), moderates these disparities.

Results

Analyzing 1,273,653 CFPB complaints from 2014 to 2022, we find that predicted probabilities of timely company response rose across all consumer groups, yet service members never matched the general population, and older adults moved from a small advantage in 2014 to a clear disadvantage by 2017 (Figure 1). For the general population, the predicted probability rose from 97.7% (95% CI: 97.6–97.8) in 2014 to 99.4% (95% CI: 99.4–99.5) in 2022, a 1.7-point increase.

Figure 1.

Predicted probability of a timely company response (2014–2022) for the general population, service members, and older adults. Solid lines show annual model-based predictions; shaded areas are 95% CIs.

Service members

Service members had consistently lower predicted probabilities of timely response than the general population. In 2014, their probability was 96.9% (95% CI: 96.4–97.4), 0.8 points below the general population. The gap widened to 1.8 points by 2016, then narrowed and stabilized at ∼0.3–0.6 points from 2017 onward. By 2022, their probability was 99.1% (95% CI: 99.0–99.2) compared with 99.4%, leaving a 0.3-point gap. When comparing early (2014–2017) and later (2018–2022) periods, the increase for service members was similar to the general population and, in many states, larger (Figure 2A). To convey scale, the 2022 service-member gap (99.1 vs. 99.4% “predicted” timely response) corresponds to about 80 more untimely responses among 26,533 service-member complaints than would be expected if they had the general-population timely response rate.

Figure 2.

State-level results, 2014–2022. A) Change in predicted probability of a timely response from 2014–2017 to 2018–2022 by group (percentage points). Higher values indicate larger increases. B) Within-state ADI disparity (low ADI minus high ADI; percentage points). Positive values indicate higher predicted timeliness in advantaged (low ADI) areas; values near zero indicate little or no gap. Model specifications are described in Materials and methods; color scales are shown to the right of each panel.

Older adults

Older adults began the study period with higher probabilities of timely response than the general population. In 2014, their probability was 98.2% (95% CI: 97.9–98.5). This advantage disappeared in 2016, and from 2017 onward, older adults had lower probabilities than the general population in most years. By 2022, their probability was 98.8% (95% CI: 98.6–99.0), 0.6 points below the general population and 0.3 points below service members. Multiplying this 0.6-point gap by the 10,788 older-adult complaints filed in 2022 yields an estimated 65 additional untimely responses relative to the general-population rate. As shown in Figure 2A, while timeliness increased across all states for older adults, the magnitude of the increase was generally smaller than for the other groups.

Geographic disadvantage

Adjusting for ZIP-code-level ADI did not eliminate group disparities, indicating that socioeconomic deprivation alone does not account for the differences. ADI was an independent predictor: each SD increase in ADI rank was associated with a 2.0% decrease in the odds of timely response (odds ratio [OR] = 0.98; 95% CI: 0.96–0.99; P = 0.011) after adjustment for product, year, and state. As shown in Figure 2B, predicted probabilities were lower in high-ADI (disadvantaged) areas than in low-ADI (advantaged) areas for all three groups, with the largest descriptive disparities observed among older adults. For the general population, the weighted difference in predicted timeliness between low- and high-ADI areas was 0.083 percentage points. Multiplying this gap by the total high-ADI complaint volume across included states over 2014–2022 (n = 252,609) yields an estimated 211 fewer timely responses associated with geographic disadvantage.

Discussion

Although company response timeliness improved over time, persistent disparities remained across groups and geographies. Service members faced persistently lower timeliness, and older adults shifted from a slight advantage to a disadvantage over the study period, a finding that held after adjusting for product type, state, and neighborhood deprivation. Linking ADI to complaint-level timeliness revealed an independent geographic gradient, with the largest descriptive gaps among older adults in disadvantaged areas. Product-stratified analyses show that subgroup gaps vary across product categories: some products exhibit near-zero differences while others show larger gaps.

These disparities warrant cautious interpretation. Timely company response is a measure of administrative responsiveness, not an indicator of favorable relief or restitution. Slower responses may reflect case complexity, documentation requirements, or firm-side risk controls (e.g. fraud-related complaints may trigger additional verification). Such verification protocols may be disproportionately concentrated in high-deprivation areas, contributing to observed ADI-related timeliness gaps. Accordingly, we interpret these patterns as descriptive differences in process timeliness, not evidence of discriminatory intent. Regulatory tradeoffs also merit consideration: increased timeliness monitoring could affect credit access or costs for protected populations (8). Such regulatory burden could operate through multiple channels—altered borrower risk profiles following lender adjustment, selective firm retreat from high-risk markets, or increased costs passed to vulnerable consumers—any of which could itself affect timeliness. Our design does not permit us to disentangle firm-side prioritization, case complexity, or market-adjustment mechanisms.

This study has limitations. First, CFPB complaints reflect self-filed behavior not representative of all consumers (9). Filing propensity, expectations of redress, severity thresholds, and product mix may differ by age and service status; as noted above, disparities persist across product categories, though unmeasured self-selection mechanisms may remain (SI Appendix). Second, the public-use files lack a consistent company final-response timestamp across 2014–2022, precluding reliable analysis of response durations; the available date-to-forwarding lag largely reflects CFPB processing rather than company response time. Third, we analyze timeliness rather than outcome quality. Finally, CFPB complaints are primarily directed at regulated financial firms; scam-related issues may nonetheless appear when consumers file against banks or payment services involved (e.g. misattributing imposter scams to their institution), affecting both the composition and complexity of complaints in the database. Despite these limitations, the scale of the dataset and robustness across specifications support persistent gaps in company response timeliness.

Materials and methods

We analyzed CFPB public complaint records (2014–2022) (10, 11). Starting from 1,491,342 records with complete outcome and product data, we excluded those with missing state identifiers (1.0%) or ADI scores (13.6%), yielding 1,273,653 complaints (85.4% of the initial frame). Each record included product type, ZIP code, and two optional submitter-selected demographic tags: “older American” (aged 62+) and “service member” (active duty, reserve, or veteran personnel). Complaints without either tag were coded as general population (1,144,178 records; 89.8% of the final sample). Data also include CFPB-recorded metrics (timeliness flag and dates). Complaints were classified into eight product categories, e.g. mortgages and debt collection.

The primary outcome was the CFPB's “Timely response?” field (Yes/No), which records whether the company responded within CFPB-defined windows (15 calendar days, or up to 60 if marked “in progress”), per CFPB documentation and oversight guidance (11). “Response” refers to the company's update in the complaint workflow, not CFPB's response to the consumer.

ZIP codes were matched to ADI scores (12). We began in 2014, when the CFPB complaint database underwent a documented change in public disclosure (13), and ended in 2022, the latest year in the ADI release. Full-year 2014 records were retained; excluding January–February 2014 (n = 12,715; 1.0%) in a sensitivity analysis yielded unchanged conclusions: the older-adult–general-population gap changed by only −0.016 percentage points and the service-member gap by −0.159 percentage points. ADI values used the most recent release available for each period: 2015 for 2014–2015, 2020 for 2016–2020, and 2022 for 2021–2022 (14).

Key predictors were indicators for older adult and service-member status, coded to allow overlap (i.e. a complaint could carry both tags). We tested for interaction effects between the older American and service member indicators via an interaction term in the primary model (OR = 1.085; 95% CI: 0.925–1.272; P = 0.318, not significant) and reestimated results using a four-group specification; predicted probabilities were directionally consistent across both approaches (general population: 0.987, older only: 0.986, service member only: 0.985, both tags: 0.985). All models included state and product fixed effects, year as a factor, and standardized ADI. Additional robustness checks, including product-stratified analyses, yielded consistent results across specifications. Details are reported in the SI Appendix.

Supplementary Material

Supplementary material is available at PNAS Nexus online.

Competing Interest

The authors declare no competing interests.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Author Contributions

Yidan Sun (Conceptualization, Formal analysis, Investigation, Methodology, Resources, Software, Validation, Visualization, Writing—original draft), Daiqi He (Data curation, Investigation, Methodology, Software, Validation), and Mayank Kejriwal (Conceptualization, Project administration, Supervision, Writing—review & editing)

Data Availability

CFPB complaint data are publicly available at https://www.consumerfinance.gov/data-research/consumer-complaints/. ADI data are from the University of Wisconsin Neighborhood Atlas (https://www.neighborhoodatlas.medicine.wisc.edu/). Analysis code, datasets, and additional results are available at https://github.com/yidans/CFPB-Complaint-Equity-Analysis.