Panel Paper:
The Subjective Measurement of Fraud in State Unemployment Programs
*Names in bold indicate Presenter
Biden’s comments follow a long political tradition of emphasizing the rarest and most extreme cases of fraud in public programs to exploit public fear that social programs are overrun with fraud. Politicians undermine public confidence in programs when they call attention to the most egregious fraud cases. They also obscure the primary causes of fraud while oversimplifying the subjectivity of fraud determinations. The most common cause of unemployment insurance fraud is working while claiming benefits, followed by separation issues (claimants misrepresent the reason for job loss), and failing to search for work. State lawmakers may increase the amount of fraud in the system by simply creating more onerous work-search requirements. Furthermore, states define fraud as a “willful misrepresentation of the facts.” What separates a willful misrepresentation from a simple error is often a subjective decision, meaning that the state unemployment insurance agencies may influence the amount of fraud in the system by varying the application of state law. A recent incident in Michigan offers an illustration of the subjective nature of fraud. The state unemployment agency used a new automated IT system to process 53,000 fraud determinations over a two-year period. At the behest of the U.S. Department of Labor, the state agency reviewed these determinations, resulting in a 93 percent reversal rate of cases reviewed thus far.
By manipulating the measurement of fraud in the unemployment insurance system, state lawmakers and administrators may undermine confidence in the program. The present study will determine if political, economic, and social factors relate to differences in the rate of fraud determinations within states over time. I will combine federal administrative data on fraud determinations from 1985 to 2015 for all 50 states with information on party control of state legislatures and governorships; data on state unemployment insurance financing and benefits; and state characteristics related to unionization rates and demographic and industry composition. Linear panel models estimated using first differences indicate that political and economic factors may contribute to changes in reported fraud within states. The findings from this study have important policy implications given the widespread adoption of IT systems for fraud detection and prevention in public programs.