This study identifies "causes and states where UI overpayment errors are concentrated." The main part of the study comprises four tables. Two tables present state overpayment error rates that add to the "Annual Report Error Rate," which the Department has published annually since 1988. The tables simply show the rates sorted in two ways -- by states alphabetically and by total error rates. There are two parallel tables presenting the corresponding dollars overpaid.

The data come from the Benefit Accuracy Measurement (BAM) program. The study excludes two states-Colorado and Puerto Rico-because the Department has decided it cannot publish their data for CY 2001. Both states deviated significantly from the prescribed BAM methodology during the year. Omitting them does not materially change the national results; together Colorado and Puerto Rico accounted for less than 2% of total payments in 2001.

As this is written, the Department is beginning to consult with its partners and stakeholders about alternative measures payment accuracy and integrity. This consultation will influence the measure the Department uses to assess UI payment accuracy performance for the Government Performance and Results Act plans. The proposed operational overpayment error rate uses a smaller set of BAM overpayment error categories than the Annual Report rate. It is designed both to reduce the disparities among state overpayment rates and to represent the kinds of errors states can detect, establish and recover through their normal operations and Benefit Payment Control units. It eliminates certain kinds of overpayments that are (a) technical in nature, or (b) that many states do not count as errors under their laws, or (c) that states cannot recover if they are detected after the fact. Tables 3 and 4 display the proposed alternative overpayment rates and the components excluded from the Annual Report measure, and Tables 3a and 4a give the corresponding dollar overpayment amounts.

Definition of BAM Error Causes

Definitions of the types of errors included in the operational error rate and those technical errors that are excluded from it are discussed below.

Operational Errors

• Benefit Year Earnings and Income Errors: These overpayment errors occurred because claimants failed to report, or underreported, earnings or other income during weeks they claimed benefits. In many cases, they returned to a full-time job. This category also includes unreported or underreported severance pay, vacation pay, or Social Security or pension benefits.

  • Overpayment Rates. Eight states have BYE errors under 1%; at the other end of the spectrum, five states are at 5% or above, reaching a high of over 8% for Maryland.
  • Overpayment Amounts. Overall error rates are highly dependent on this error cause. The three states with the highest overall error amount, California, Texas and Illinois, are in the top four for this category, where they are joined by Pennsylvania. This is the only cause for which PA ranks among the top error rate states.

• Separation Errors. These errors involve claimants who quit or were discharged for cause from work and were erroneously determined to be eligible for benefits. These errors could result from a lack of information available at the time the initial claim was filed or inaccurate application of state law and policy.

  • Overpayment Rates. Separation errors are another of the "big three" causes, although one state, Maine, reported none in CY 2001, and about a third of the states reported Separation error rates under 1.0%. On the other hand, five states had rates over 3%, with New Jersey the highest at about 4.5%.
  • Overpayment Amounts. About one quarter of the states had over $10 million in separation overpayments, with New York being the highest at $83 million. BAM estimated that another quarter had less than $1 million.

• Able and Available. These are overpayments due to claimants failing to meet state law and policy requiring them to be able to work and available for work during the weeks they claim benefits.

  • Overpayment Rates. Overall, this is a fairly minor category of errors, 0.56% of benefits for the nation as a whole and totaling about $166 million for the year. However, eight states' error rates for this cause exceeded 1% of benefits; Minnesota's exceeded 3%.
  • Overpayment Amounts. BAM estimated that two states had no dollars overpaid from this cause, but 5 states' amounts exceeded $10 million. The highest was Minnesota, at $19 million.

• Other Eligibility Errors. These errors include instances where claimants refused an offer of suitable work, were self-employed, were found to be illegal aliens, and miscellaneous unspecified eligibility failures.

  • Overpayment Rates. This is another area with a wide variance. Nationwide, these errors accounted for less than 0.4% of benefits, and about 40% of the states had no errors of this type in CY 2001. However, these are major error areas for Wyoming (5.4% of benefits), Maine (6.25%), Virginia (1.4%) and Vermont (1.4%).
  • Overpayment Amounts. For the country as a whole, this category accounted for about $108 million in 2001. California had the highest amount, about $33 million. Maine and Wyoming, the two highest -rate states, had very small dollar error amounts because they are among the smallest states.

Technical/Nonrecoverable

• Base Period Wage Errors. These overpayment errors occur when employers report claimants' earnings during the "base period" incorrectly. These earnings determine weekly benefit amount and total benefits payable.

  • Overpayment Rates. For nearly all states, BP Wage issues are a minor source of overpayments; all but three states had rates of less than 1% in CY 2001. The exceptions are New Jersey, Ohio, and New York. New Jersey's rate exceeds 3%, perhaps due to its complex Alternative Base Period system.
  • Overpayment Amounts. The typical BP error is quite small, but there are many of them, adding up to relatively few dollars. Forty-four states reported less than $10 million in errors due to this cause. The main exceptions are the three high-error rate states, where amounts exceeded $25 million.

• Work Search. These overpayments are due to claimants' failures to comply with the state work search requirements. These requirements vary greatly among states.
  • Overpayment Rates. When benefit accuracy measurement began, work search was the principal source of estimated overpayments. Many states have since eliminated or vitiated work search requirements, and it now ranks third among causes. BAM estimated no official overpayments for 15 states. Of the remainder, however, some had exceedingly high rates: the Virginia's rate was about 11-½ % and Maryland's was about 8%. Seven other states' rates were over 4%. Half of Virginia's aggregate overpayment rate was due to work search errors.
  • Overpayment Amounts. BAM estimates that a quarter of the states had overpayments exceeding $10 million due to work search errors, headed by Texas and Illinois at about $60 million each.
• Job Service Registration. Failure to be actively registered with the Job Service or One-Stop in accordance with state law and policy.
  • Overpayment Rates. State laws differ widely with respect to requiring active registration with the Job Service. About half of the states had no Job Service registration errors in CY 2001. On the other hand, nine had rates exceeding 2%, and Mississippi's error rate for this cause was nearly 15%.
  • Overpayment Amounts. Total overpayments in four states-Washington, Texas, Michigan, and Mississippi-exceeded $25 million for this cause. Washington's amount was nearly $50 million.
• All Other Causes. This includes errors related to dependents' allowances; back pay awards; redeterminations and appeals reversals; and receipt of benefits during a period of disqualification even though a stop-pay order was in effect.
  • Overpayment Rates. The pattern for this error category is similar to the Other Eligibility cause. BAM reports no errors for this cause in about half the states. Only four states have error rates of 1.0% or more: Nebraska, Illinois, Iowa, and Texas. Illinois is the highest at 1.8%.
  • Overpayment Amounts. Total overpayments for this cause were estimated at $110 million in CY 2001. Three states--Illinois, Pennsylvania, and Texas--account for about 60%.

Findings by Error Cause: National Summary

The following table summarizes the national totals. As has been the case for the last several years, benefit year earnings violations is the leading cause, followed by separation errors and work search errors. Together those three causes account for two thirds of all overpayments estimated by BAM.

Note: Figures Exclude Colorado and Puerto Rico

Findings by Error Cause: Findings Including State Detail

Total Error Rates. Table 1 presents aggregate overpayment rates by state; Table 1a contains the data on dollars overpaid. In Tables 2 and 2a, the results are sorted from high to low. Four states reported rates of 3% or less; five states had rates of 18% or above, with the highest being Virginia at 22%. Estimated amounts overpaid range from less than $700,000 in North Dakota (which combines a low error rate with small size), to the vicinity of $200 million or more in California, Illinois, and Texas. California is in the high-error amount group because its size (its error rate is 5.6%). Texas and Illinois both have double-digit error rates.

An Analysis of the Proposed Operational Overpayment Rate

The proposed alternative accuracy measure focuses on the categories of overpayment errors that are most detectable and recoverable by state operations and BPC units, and can thus be most directly related to overpayments established and recovered by BPC units. These are overpayments which state law deems to be recoverable, and comprise the categories due to benefit year earnings and income, Separation, Able and Available, and the Other Eligibility category which includes payments made to claimants who refused an offer of suitable work, were self-employed, or who were illegal aliens. It excludes overpayments under all cause categories that the state could not recover, as well as the following categories of recoverable violations: Work Search, ES Registration, Base Period Earnings, and minor miscellaneous violations including dependents' allowance violations. Work Search and ES registration violations are excluded because many state do not make them eligibility conditions. Also, such violations are very difficult for state BPC units to detect even if they are recoverable. BPC units rarely detect base period wage errors, about 40% of which are not recoverable nationwide. Only a handful of states provide dependents' allowances. Errors in the Other Eligibility category again are relatively unlikely to be detected by BPC methods.

Table 3 shows that this measure produces an aggregate error rate of about 4.7%, and state rates range from about 1% in Oklahoma, Georgia, and North Dakota to about10% in Maryland and Delaware. Benefit year earnings and income and separation violations represent four fifths of these overpayments. The alternative measure yields about $1.4 billion in overpayments, versus nearly $2.45 billion for the usual Annual Report measure (Table 3a).

Tables 4 and 4a are the complement to Tables 3 and 3a. Excluded overpayments amounted to 3.5% of benefits paid in CY 2001. Three error causes account for three quarters of the overpayments in this table: work search (35%), ES Registration (20%) and Base Period Wage violations (19%). Nearly half (48%) of the excluded overpayments were non-recoverable.