An Analysis of Overpayments Not Included In the Unemployment Insurance (UI) Government Performance and Results Act (GPRA) Measure for 'Prevention of Overpayments'

Background

As part of its GPRA Strategic and Annual Performance Plans for FY 2002, the Employment and Training Administration (ETA) established a UI goal of “Detection of Overpayments.” After consulting with the UI system through Unemployment Insurance Program Letter 33-02 and other contacts on possible measures, a ratio measure was selected as an indicator for this goal. Its numerator is the amount of overpayments actually established through state integrity operations. Its denominator is an estimate—from the Benefit Accuracy Measurement (BAM) program--of dollars overpaid that are recoverable and that the state is most likely to detect and establish for recovery through existing integrity procedures. The overpayment error rate from which the denominator is determined is termed an “operational overpayment rate” to distinguish it from the broader “Annual Report” rate based on BAM investigations that includes all types of payments that fail to meet state law and policy. The “operational overpayment rate” is the Annual Report rate less all non-recoverable overpayments and four categories of theoretically recoverable overpayments that integrity operations are very unlikely to detect.

All Overpayments vs. Operational Overpayments.

The "operational overpayments" definition is designed to meet the need for a concept of overpayments against which states can evaluate their overpayment detection and recovery activities. It also identifies the more substantive and less “technical” categories of overpayments that occur in state UI operations.

The operational overpayments concept is derived from the standard BAM definition in two steps. First, all overpayments that BAM investigators deem “non-recoverable” are eliminated. States are precluded from recovering overpayments after the fact for various reasons. The most common reasons are finality (too much time has elapsed) or responsibility (many states do not recover overpayments resulting from agency error) or inequity (to recover the overpayment would cause obvious harm to the beneficiary). Subtracting non-recoverable overpayments leaves a combination of overpayments due to fraud--all of which are classified as recoverable--and the recoverable portion of non-fraud overpayments. In the second step, the likely detectability of recoverable overpayments is evaluated. Certain types of overpayments, although theoretically recoverable, are extremely unlikely to be detected after the fact through usual agency integrity procedures. The operational definition excludes four major categories, which BAM terms “causes:” those due to base period wage errors, work search violations, failure to be registered with the Employment Service (ES), and other causes. The following table using CY 2002 data provides a crosswalk between the BAM Annual Report rate and the GPRA operational overpayment rate.

	% of Benefits	$ Overpaid (000)	Percent Distribution
Table 1 Crosswalk: BAM Annual report Overpayment Rate to GPRA "Operational Overpayment" Rate (Calendar Year 2002 data)
BAM Annual Report Rate	9.1%	$3,677,785	100.0
Less: Nonfraud, nonrecoverable	-2.24%	-$904,739	24.6
= Fraud + Nonfraud Recoverable	6.86%	$2,773,046	75.4
Less: Work Search	-1.25%	- $506,475	13.8
Less: Base Period Wage	- 0.40%	- $162,101	4.4
Less: ES Registration	- 0.05%	- $21,243	0.6
Less: Other Causes	- 0.20%	- $82,346	2.2
= GPRA “Operational” Rate	4.96%	$2,000,881	54.4

The table indicates that in 2002, BAM investigators considered approximately 25% of approximately $3.7 billion in UI overpayments “nonrecoverable” for the reasons outlined above. The operational definition eliminates another 21% of overpayments—all of them theoretically recoverable--because of the difficulty of actively detecting them after the fact through systematic procedures such as crossmatches. This process of elimination indicates that only a bit over 50 percent of overpayments are subject to detection, establishment and eventual recovery after the initial error has occurred. These proportions remain relatively constant from year to year.

A Cause and Detectability Analysis of Overpayments Excluded from the Operational Definition: CY 2002

The following analysis examines first non-recoverable overpayments, and the recoverable overpayment excluded from the operational overpayments definition because of the difficulty of detecting them after the fact. For each subgroup, the analysis will focus on two dimensions. The first dimension is “cause.” The basic BAM cause codes indicate where the overpayment error occurred in the eligibility determination cycle, e.g., a failure to meet work search requirements or to report earnings or other income for a week claimed. The analysis will use the following standard BAM cause codes:

Base Period Wage (monetary determination) errors
Separation Issues (the claimant was not involuntarily separated from work either before the beginning of or within a UI benefit year)
Able and Available violations: (failure to be able to work or available for work)
Benefit Year Earnings violations (excessive earnings, or unreported earnings, during the week claimed.)
Severance or Vacation Pay violations (many states consider receipt of such payments as “disqualifying income,” the same as earnings)
Work Search violations
Social Security violations (many states consider receipt of such payments as “disqualifying income,” the same as earnings)
Illegal Alien causes
ES Registration (many states require claimants to be actively registered with the Job Service or One-stop system)
Dependents Allowances (errors related to the number of dependents or amount of dependents allowances in those states whose UI benefit amount is related to the number of eligible dependents)
Other Eligibility causes. (items such as refusal of suitable work and other miscellaneous eligibility violations not already identified)
Other Causes (benefits paid during a period of disqualification, redeterminations and appeals reversals, back pay awards, all other causes)

The second dimension is called Prior Agency Action. Prior Agency Action codes indicate the extent of the agency's knowledge, or ability to know, the essential facts needed to make a correct eligibility decision or determine the correct payment amount. All errors receive one of the following five codes:

The issue was not detectable by normal procedures ("not detectable");
The agency was in the process of resolving the issue or had resolved it between the time the BAM sample was drawn and the review completed ("in process");
The agency had identified the issue but took the wrong action ("wrong action");
Despite having sufficient documentation, the agency did not conclude there was an issue ("didn’t resolve"); and
The agency failed to follow procedures that would have detected the issue ("procedures not followed").

In brief, the first code indicates that the agency's normal procedures could not detect the issue; the second code involves an issue that is only an error because of the timing of the BAM review; and the other three codes indicate detectable issues that were either erroneously decided, or not followed through to identification of an issue, or missed entirely because of failure to follow procedures that would have identified the issue the BAM investigator detected.

Non-recoverable Overpayments

In 2002, slightly over half of the overpayments not part of the Operational Overpayment definition, or $900 million out of nearly $1.7 billion (see table 4, below), were excluded because the agency did not consider them recoverable. Recovery is often precluded because the agency was responsible for the overpayment, either by incorrectly determining the claimant was eligible or failing to resolve the issue timely. Table 2 shows these amounts by cause and Prior Agency Action.

The issues that caused approximately half of these nonrecoverable overpayments were detectable at the time the decision to pay benefits was made. This is a significantly higher percentage than for overpayments in general. Of the potentially detectable portion, the most important reason for the overpayment was the state’s failure to resolve an issue ($191 million out of $455 million, or 42%) but failure to follow procedures and taking the wrong action were both important. The relatively high ratio of detectable to all nonrecoverable overpayments indicates that agencies’ efforts to improve their internal controls is an essential step in handling this category of overpayments. These overpayments can only be addressed through prevention, because once they occur, detection and recovery are highly unlikely.

Table 2: Non-Recoverable Overpayments

Causes Undetectable In Process Didn't Resolve Procedures
Not Followed Wrong Action All Actions

Able+Avail. $24,514,368 $0 $8,334,428 $19,447,338 $5,871,883 $58,168,017

BP Wages $73,658,825 $20,916 $14,451,564 $10,909,040 $3,385,092 $102,425,437

BY Earnings $5,198,181 $954,284 $1,468,628 $0 $186,333 $7,807,426

Dependents $3,682,832 $0 $380,684 $1,476,990 $49,562 $5,590,068

ES Reg. $197,930,332 $4,153,773 $57,824,174 $51,958,364 $3,731,049 $315,597,692

Illegal Alien $511,838 $36,289,671 $1,205,806 $0 $38,007,315

Other Causes $16,422,808 $1,419,522 $810,524 $12,037,039 $1,277,037 $31,966,930

Other Eligibility . $7,338,297 $0 $4,928,067 $660,281 $0 $12,926,645

Sep. Issues $82,244,285 $4,548,060 $43,539,577 $7,321,042 $96,766,920 $234,419,884

Severance & Vacation Pay $1,311,572 $0 $1,897,089 $0 $1,936,182 $5,144,843

Soc. Security $6,337,340 $57,369 $11,530,553 $0 $1,850,875 $19,776,137

Work Search $31,072,655 $934,088 $9,779,029 $25,894,777 $5,227,781 $72,908,330

All Causes $449,711,495 $12,599,850 $191,233,988 $130,910,677 $120,282,714 $904,738,724

Table 2 illustrates that two causes account for three fifths of the overpayments: failures to register with the Employment Service or One-stop, and separation errors. The data also indicate that these errors have considerably different root causes. Over sixty percent of ES registration errors cannot be detected by the agency, with failures to resolve issues or to follow procedures that would detect an issue accounting for most of the remainder. The separation issue problem is different: only about one third of these issues are undetectable, and over 40% occur because agencies make the wrong decision and another fifth because they fail to pursue issues.

The other major cause of nonrecoverable overpayments is Base Period Wage issues. About 70% of these tend to be undetectable by the agency at the time it is making the monetary determination.

Recoverable Overpayments

The other category of overpayments, accounting for about $772 million or 46% of the overpayments excluded from the operational measure, comprises only four causes of overpayments. Table 3 shows the dollar amounts using the same categories as used for nonrecoverable overpayments.

Table 3: Recoverable Overpayments

Causes Undetectable In Process Didn't Resolve Procedures
Not Followed Wrong Action All Actions

BP Wages $114,589,022 $679,951 $22,181,766 $19,411,367 $5,238,856 $162,100,962

ES Reg. $19,585,654 $0 $1,657,614 $0 $0 $21,243,268

Other Causes $23,874,777 $7,669,815 $31,566,228 $7,965,621 $11,269,513 $82,345,954

Work Search $499,284,468 $2,796,345 $2,561,044 $601,498 $1,231,287 $506,474,642

All Causes $657,333,921 $11,146,111 $57,966,652 $27,978,486 $17,739,656 $772,164,826

This category is dominated by work search violations, which account for two thirds of these recoverable overpayments. A comparison of tables 2 and 3 indicates that the overwhelming percentage (88%) of work search violations results in potentially recoverable overpayments. The other major category is base period wage violations, one fifth of the excluded recoverable overpayments. Comparing tables 2 and 3 also shows that very few ES registration overpayments are recoverable, only about 6 percent.

On the whole the overpayments in this category are very difficult to detect at the time the payment is being made. These data indicate that about 85% are undetectable. Three of the causes fit the largely-undetectable pattern, ranging from 70% for base period wage issues to nearly 99% for work search violations. The “other causes” category is the exception to the general rule. Less than 30% of these overpayments are considered undetectable; the principal reason these errors occur is failure of the agency to pursue and resolve potential issues (39%).

Conclusion

Table 4 aggregates the two types of excluded overpayments. It indicates that in 2002, states made nearly $1.7 billion in overpayments that they were unlikely to be able to establish for recovery. About half was due to the fact that the overpayments were not recoverable under their laws, the other half because their normal BPC detection activities are unlikely to identify them. Overall, BAM estimates that front-line staff could not have detected about two thirds of these overpayments in the normal course of business, so with existing administrative procedures (e.g., remote claims taking systems seem to make detection of issues more difficult) these overpayments could neither have been prevented nor recovered. Work search violations accounted for nearly half of the undetectable overpayments, followed by ES registration (20%) and Base Period Wage violations (17%).

However, about one third—an estimated $570 million in 2002—were more controllable. Agencies were somewhat aware, or should have been aware, of the pertinent issues that underlay these “non-GPRA” or “non-operational” overpayments. They were in the process of dealing with about 4% of these when BAM drew and investigated the case. About 19 of 20 were not in the process of correction, and:

The largest reason for making the error (about 44% of the controllable category) involved the state’s failure to use information it had in hand to determine that this information definitely pointed to an eligibility issue. This seemed particularly to involve ES registration, base period wages, separation issues, and illegal alien issues.
About 28% involved taking the wrong action on adequate information; this was a particular problem where separation errors are concerned.
Failures to thoroughly apply their own procedures--a quarter of the “controllable” errors—were especially implicated in ES registration errors, base period wage errors and work search errors.

Causes	Undetectable	In Process	Didn't Resolve	Procedures Not Followed	Wrong Action	All Actions
	Table 4: All Non-GPRA Overpayments
Able+Avail.	$24,514,368	$0	$8,334,428	$19,447,338	$5,871,883	$58,168,017
BP Wages	$188,247,847	$700,867	$36,633,330	$30,320,407	$8,623,948	$264,526,399
BY Earnings	$5,198,181	$954,284	$1,468,628	$0	$186,333	$7,807,426
Dependents	$3,682,832	$0	$380,684	$1,476,990	$49,562	$5,590,068
ES Reg.	$217,515,986	$4,153,773	$59,481,788	$51,958,364	$3,731,049	$336,840,960
Illegal Alien	$0	$511,838	$36,289,671	$1,205,806	$0	$38,007,315
Other Causes	$40,297,585	$9,089,337	$32,376,752	$20,002,660	$12,546,550	$114,312,884
Other Eligibility	$7,338,297	$0	$4,928,067	$660,281	$0	$12,926,645
Sep. Issues	$82,244,285	$4,548,060	$43,539,577	$7,321,042	$96,766,920	$234,419,884
Severance & Vacation Pay	$1,311,572	$0	$1,897,089	$0	$1,936,182	$5,144,843
Soc. Security	$6,337,340	$57,369	$11,530,553	$0	$1,850,875	$19,776,137
Work Search	$530,357,123	$3,730,433	$12,340,073	$26,496,275	$6,459,068	$579,382,972
All Causes	$1,107,045,416	$23,745,961	$249,200,640	$158,889,163	$138,022,370	$1,676,903,550
All Causes	66.0%	1.4%	14.9%	9.5%	8.2%	100.0%

The analysis above shows that in CY 2002 agencies had at least some control over approximately $550 million in non-recoverable overpayments or overpayments their procedures are not likely to detect after a payment is made, and were not in the process of correcting them. In theory, perfect internal controls could have prevented all $550 million. This suggests that improving adherence to established procedures, ensuring follow-through to detect issues, and improving understanding and application of state law and policy are called for. A closer analysis of the error categories indicates that not all these overpayments warrant equal attention, however, and furthermore that the most significant areas are being addressed:

Errors involving failure to register with the ES (21% of the $550 million) are in a unique technical category. Many states do not require claimants to register; and in practice because the ES is responsible for less than 5% of UI claimants’ return to work, these overpayments seem to have little or no relation to reemployment outcomes.
Separation and illegal alien errors (together, a third of the $550 million) are very important, on the other hand, because one error can affect multiple weeks of benefit eligibility and usually involves the whole weekly benefit amount. The federal partner works very hard with states through training and technical assistance to improve the quality of separation adjudication processes. With funding assistance from the Department, states are implementing data interchange with the Social Security Administration (SSA) to detect fraudulent use of social security numbers.
Each monetary error (these were 14% of the $550 million) also affects the entire benefit stream. However, the average monetary overpayment is only about one sixth of the weekly benefit amount. Monetary overpayments and underpayments roughly balance one another.
All other errors—work search, able and available, and other issues--comprise the remaining fifth of these controllable overpayments. They seem to deserve much less attention because in most cases each error relates to only one week of eligibility. Thus the roughly $100 million in overpayments they summed to in 2002 represents the effect of a great many disparate overpayment events.

In looking at the “controllable” overpayment problems identified above, one cannot overlook some of the contributing factors states faced in 2002. These included the considerable staff turnover states have experienced due to waves of retirements sweeping much of government, the high workloads from the Temporary Extended Unemployment Compensation program, and to various budget difficulties. The most significant parts of the underlying problem—the detection and proper handling of separation issues and illegal alien errors—are already high focus areas for the Department and the states. Efforts designed to improve the quality of nonmonetary determinations and to ensure full use of the SSA data interchange will continue and should show fruit in succeeding years. As state workforces stabilize, they will also show improvement in their detection and handling of monetary and miscellaneous overpayment issues.