Vulnerability and fraud: evidence from the COVID-19 pandemic

The COVID-19 pandemic created an unprecedented global shock with widespread health, psychological, and economic impacts. Concurrently, reports of consumer fraud surged in the U.S. and other countries. This study investigates whether and how the spread of COVID-19 is associated with increases in reported fraud and scams, and what factors contribute to this relationship. The authors pose three questions: how rampant fraud became alongside rising COVID-19 cases; which factors (including finance- and non-finance-related) are associated with the spread of scams; and what policy lessons can mitigate fraud, especially finance-related fraud. They contrast two hypotheses: (1) the vulnerable-to-become-victimization hypothesis, grounded in the opportunity model of predatory victimization (greater exposure to offenders, reduced guardianship, increased target attractiveness), predicting more fraud as pandemic stress rises; and (2) the vulnerability-risk-aversion hypothesis from psychology/neuroscience, positing that uncertainty lowers dopamine and increases risk aversion, leading to heightened vigilance and potentially fewer scams. Early descriptive evidence shows fraud incidents rising in parallel with COVID-19 cases from March to June 2020, motivating the empirical analysis.

The study builds on criminological theories of victimization: lifestyle/exposure theory, routine activity theory, and their synthesis in the opportunity model of predatory victimization (Cohen et al., 1981), which links victimization risk to exposure, proximity, guardianship, and target attractiveness. Prior work documents increased online exposure and reduced social guardianship during lockdowns, fostering environments conducive to online fraud (e.g., romance scams, eWhoring). Psychological and neuroeconomic research suggests that uncertainty affects risky decision-making via dopamine modulation and may induce risk-averse behavior, potentially reducing victimization. The paper also connects to consumer fraud literature identifying demographic, psychological, and behavioral correlates of victimization, and to financial literacy research showing that higher literacy improves preparedness for macro shocks and financial decisions. This framework supports two competing hypotheses about the pandemic–fraud nexus.

Data and period: Daily U.S. COVID-19 case data (Our World in Data) and Federal Trade Commission (FTC) fraud/scam complaints from January 4 to July 28, 2020 (208 days). Outcome construction: The main dependent variable is the daily growth rate of scam/fraud cases over a 7-day window: ASCAM_CASES = [ln(SCAM_CASES_t / SCAM_CASES_{t-7})]/7. Main regressor: lagged daily growth rate of COVID-19 cases over a comparable 7-day window using t−1 and t−8 to align reporting patterns: LAGΔCOVID_CASES = [ln(COVID_CASES_{t-1} / COVID_CASES_{t-8})]/7. Empirical model: ASCAM_CASES = α + β1·LAGΔCOVID_CASES + Γ·Controls + μ. Controls include: seven-day average S&P 500 daily return from t−8 to t−1; same-day S&P 500 return; a Seasonal Affective Disorder (SAD) proxy (hours of extra darkness); Government Response Stringency Index (daily); and a non-trading day dummy (weekends/holidays). Descriptive statistics characterize the distributions. Awareness vs vulnerability test: Using Google Trends, the authors compute daily search indices for pandemic stress (queries: pandemic, COVID-19, novel coronavirus) and fraud awareness (queries: fraud, scam, Ponzi scheme) over the same 208 days, then split the sample at the median into high/low pandemic-stress days and high/low fraud-awareness days. They re-estimate the baseline model within subsamples and formally test coefficient differences to infer whether increases are driven by victim vulnerability or reporting awareness. Robustness: Alternative outcome and exposure constructions are tested: (1) day-over-day log change of scams vs lagged day-over-day log change of COVID-19; (2) seven-day moving sums’ log ratios for scams and COVID-19. Cross-sectional analysis of financial literacy: At the state level (50 states + DC), total reported fraud cases from Jan 1–Jul 28, 2020, are scaled by state population and split into finance-related vs non-finance-related categories based on complaint topics. Key regressors include COVID_INFECTED_POPULATION_RATIO (total cases per population), HIGH_FINANCE_LITERACY (top 25 of WalletHub’s financial literacy ranking), and the interaction term. Controls include state-level personal income, education level, poverty rate, and religious population proportion.

• Baseline association: LAGΔCOVID_CASES is positively associated with fraud growth (coef = 0.1285, p = 0.001). Economically, a 10% increase in confirmed COVID-19 cases corresponds to a 1.285% daily increase in fraud cases; given the average daily increase in fraud cases is ~2%, this implies a 64.25% relative jump. - Controls: Seven-day average market returns are negatively related to fraud growth (coef = −6.9210, p = 0.002), implying a 1% average daily market drop over seven days associates with a 6.92% increase in fraud cases. SAD is positive (0.1681, p = 0.001), and Government Stringency Index is positive (0.0071, p < 0.001). Same-day returns and non-trading-day dummy are not significant. - Vulnerability vs awareness: Pandemic stress split: The effect of COVID-19 growth on fraud growth is stronger on high-stress days (coef = 1.2582, p < 0.001) than on low-stress days (coef = 0.1454, p = 0.001); difference = 1.1129, χ² = 43.97, p < 0.001. Fraud awareness split: Significant effect only on low-fraud-awareness days (coef = 0.1984, p = 0.001); on high-fraud-awareness days the effect is not significant (coef = −0.0086, p = 0.798). Difference = −0.2070, χ² = 17.89, p < 0.001. These patterns indicate increased victim vulnerability, not reporting awareness, drives the rise in fraud. - Robustness checks: Results remain significant using alternative definitions: (a) day-over-day logs: coef = 0.3912, p = 0.005; (b) 7-day moving sums: coef = 0.0428, p = 0.002. - Financial literacy analysis (state level): Finance-related scams per population increase with COVID-19 infection ratio (coef = 0.0098, p = 0.015), but the interaction with HIGH_FINANCE_LITERACY is negative and significant (−0.0090, p = 0.064), implying higher financial literacy attenuates finance-related fraud increases. For non-finance-related scams, the interaction is not significant (−0.0012, p = 0.304). Personal income is positively associated with finance-related scams; religious population share is negatively associated with both finance and non-finance-related scams.

The findings support the vulnerable-to-become-victimization hypothesis: as COVID-19 spreads, individuals’ exposure to fraudsters increases (greater online engagement), guardianship decreases (isolation, reduced social support), and target attractiveness rises (heightened psychological and financial strain). The stronger relationship on high pandemic-stress days and the absence of an effect on high fraud-awareness days indicate that increased vulnerability, rather than awareness or reporting propensity, explains the surge in fraud. Control results reinforce this interpretation: tighter government restrictions and lower market performance coincide with conditions that amplify vulnerability and reduce guardianship. The financial literacy analysis reveals that higher literacy mitigates finance-related victimization, aligning with literature on literacy improving decision-making under shocks. These results emphasize the need to integrate fraud prevention into public responses during crises, focusing on enhancing cognitive functioning, social support/guardianship, and financial education, with particular attention to groups such as older adults who face disproportionate risks.

Fraud and scams escalated alongside the spread of COVID-19 in the U.S., with empirical evidence favoring the opportunity model of predatory victimization. The increase is driven by pandemic-induced vulnerability rather than by increased awareness or reporting. Financial literacy appears protective against finance-related fraud during the pandemic. Policy implications include: strengthening cybercrime enforcement during crises; implementing targeted fraud prevention that bolsters cognitive functioning and reduces isolation; expanding financial literacy programs at community and state levels; and enhancing guardianship mechanisms, especially for older adults. Future research should leverage more granular pre- and intra-pandemic data, incorporate victim demographics, and examine local policy impacts to deepen understanding of fraud dynamics under systemic shocks.

• Lack of daily pre-pandemic fraud/scam data limits direct before–after comparisons (e.g., Jan–Jul 2019 vs Jan–Jul 2020). - Inability to identify victim and offense characteristics at the individual level (e.g., age, gender, education, self-control) constrains profiling analyses. - Absence of daily state- or county-level fraud/scam data precludes assessing local policy and regulatory impacts. - No demographic profile of victims in the available data prevents formal tests focused on particularly vulnerable groups, such as older adults.