Fraudulent Receipt of Unemployment Insurance Benefits - Characteristics of Those Who Committed Fraud and a Prediction Profile

• NCJ Number: 69547
• Author(s): R D St Louis; P L Burgess; J L Kingston
• Date Published: 1978
• Length: 30 pages
• Corporate Author: Arizona Dept of Economic Security
• Sale Source: National Institute of Justice/
• Language: English
• Country: United States of America

Annotation

Methods used to develop a profile for identifying persons likely to collect fraudulent benefits from Arizona's Unemployment Assurance Program are described, based on data from 5 years of detected fraud cases.

Abstract

The data base of 5,304 claimants who received fraudulent overpayments during fiscal years 1973-77 represented 1.6 percent of all claimants eligible for unemployment insurance benefits during that period. Discriminant analysis was used to estimate two profiles: one based on data available at the time initial benefit eligibility is determined and one that also depends on information generated during the benefit year. The latter approach was more effective in uncovering larger groups and groups with higher fraud rates. Claimants were classified on the basis of sex, age, ethnic group, and occupation into two groups--detected fraud and all other claimants. Statistical methods are detailed, and comparisons between the groups are illustrated in 12 tables. Analysis revealed that 2.4 percent of the total claimants had a fraud rate of 8.9 percent, and that 0.5 percent of the total accounted for 11.5 percent of detected frauds. Rates were higher for men than women and dropped as the age of claimants increased. Fraud rates for white, white with spanish surname, and other ethnic classifications were nearly identical. Among the occupational categories, structural workers had the highest rate of 2.4 percent while the lowest rates were recorded for workers in skilled and semi-skilled trades. Correlations were discovered between the incidence of fraud and levels of previous earnings, weekly benefit amounts, and the length of time that benefits were received. Analysis of information gathered during the benefit year indicated a pattern of sharp increases in fraud as periods of individual unemployment increased. The study demonstrated that discriminant analysis is an extremely promising technique for identifying fraud prone groups, which can then be used to allocate fraud detection resources more effectively by focusing on these individuals. Suggestions for further research included adding more variables to the profile analysis, testing the profiles by applying them to a new group of clients, and including data from other states in a proposed cross-match program. Footnotes are provided.