Improving unemployment benefit payment accuracy and integrity

Official letter (https://www.dol.gov/sites/dolgov/files/ETA/advisories/UIPL/2003/UIPL23-03.html):

EMPLOYMENT AND TRAINING ADMINISTRATION ADVISORY SYSTEM
U.S. Department of Labor
Washington, D.C. 20210
CLASSIFICATION: UI
CORRESPONDENCE SYMBOL: OWS/DPM
DATE: March 31, 2003
UNEMPLOYMENT INSURANCE PROGRAM LETTER NO. 23-03
TO: ALL STATE WORKFORCE AGENCIES
FROM: CHERYL ATKINSON /s/, Administrator, Office of Workforce Security
SUBJECT: Unemployment Insurance (UI) Benefit Payment Accuracy and Integrity
RESCISSIONS: None
EXPIRATION DATE: March 31, 2004

1. Purpose. To provide states with data analyses that suggest effective ways to improve detection, prevention and recovery of UI benefit overpayments.

2. References. Unemployment Insurance Program Letter (UIPL) 33-02, Development of An Unemployment Insurance Payment Accuracy/Integrity Measure (July 31, 2002); UIPL 15-03, Government Performance and Results Act (GPRA) Fiscal Year (FY) Unemployment Insurance (UI) Program Goals (January 22, 2003).

3. Background.
   Since 1987, when the Department of Labor (Department) instituted a Benefits Quality Control program (now called Benefit Accuracy Measurement, BAM), states have developed estimates of benefit overpayments and underpayments by thoroughly examining samples of payments each week to determine whether the initial decisions related to benefit qualifying and eligibility were correct and whether all eligibility requirements for the week in question were met.
   In developing a GPRA goal for FY 2003 related to improving benefit payment integrity, the Department examined the various reasons for overpayments that make up the BAM rate. It developed an "operational" overpayment rate that reflects an estimate of the types of overpayments that states can be expected to detect and recover through their program operations. About one quarter of recoverable overpayments found through BAM investigations cannot be detected through normal program operations, work-search violations being the principal example. However, a significant portion of recoverable overpayments, such as those involving claiming benefits while working, are almost all detectable. The preliminary operational overpayment rate for CY 2002 is 4.9%.
   The GPRA goal that the Department adopted following consultations with state partners uses the operational overpayment data compared with the amount of overpayments actually established for recovery. The goal for FY 2003 is for states to detect and establish for collection at least 59% of the estimated total detectable and recoverable overpayments.

4. Displays and Analyses of Erroneous Payments. The data analyses provided here are intended to help states develop and adopt procedures that will increase the level of overpayments that are prevented or detected. These include analytical displays of error rates and overpayment amounts by reason (cause) for the various states; and benefit-cost analyses of various methods of detection and recovery. Highlights from some of the analyses are discussed briefly below. A short synopsis of each analysis is attached and the complete analyses are available on the ETA/OWS Web site, https://ows.doleta.gov and by e-mail from Burman Skrable at [email protected].
   Causes of Overpayments. Analyses of BAM data show that three "causes" of overpayments accounted for about two thirds of all overpaid benefits. These causes are (a) benefits claimed while working (31% in CY 2001); (b) voluntary quits and discharges that should result in disqualifications but do not (21%); and (c) failure to meet work search requirements (15%).

• Benefits Claimed While Working. Many if not all of these overpayments can be detected by crossmatch with state new hire data and wage record data. Use of new hire data can also prevent some overpayments by early detection. Data from states using new hire data suggest that they prevented about $55 million in CY 2002.

• Voluntary Quits and Discharges. Payment errors resulting from incorrect "separation" eligibility decisions at the time a new or additional claim is filed tend to be large because many weeks of overpaid benefits can result from that erroneous decision and each involves the full weekly benefit amount. The agencies' greatest obstacle to making the right decision, affecting about three fifths of error cases, is their inability to detect a potential eligibility issue, often because they lack employer information about the reason for separation. In addition, agencies allow about a third of all separation overpayment errors by mishandling information or failing to follow procedures.

• Work Search Requirements. Failure to meet work search requirements cannot be detected cost-effectively; however, periodic reviews of work search efforts may increase compliance.
  Detectable and Recoverable Overpayments. There are two dimensions of overpayment detection. The first dimension is whether an eligibility issue can be detected when payment decisions are being made—the time when an overpayment could be prevented. BAM determined that in FY 2001 states could not have prevented about $1.7 billion, or 73%, of the estimated $2.3 billion dollars overpaid because the eligibility issues could not be detected when payment decisions were being made. For example, information necessary to detect when individuals inappropriately continue to claim UI benefits after returning to work is typically not available until much later.
  The second dimension is detection of actual overpayments, after the fact. Current BPC procedures are likely to be able to detect slightly over half of all overpayments. For example, the overpayments made to individuals who continue to claim UI benefits after returning to work, almost impossible to detect when payment decisions are being made, are almost all detectable after the fact using various crossmatches. Comparing the actual detected (and established) and recovered amounts with the estimated "detectable and recoverable" overpayment amount—$1.24 billion in 2001—is the most reasonable basis for judging the completeness of state BPC efforts. Certain overpayments, such as work search violations, are very difficult and expensive to detect, both at the time the decision is made to pay the week claimed, and later, after the payment has been made. BAM estimates that in 2001 there were between $550 and $600 million of such overpayments that state laws would have allowed the state to recover, but their current BPC methods were very unlikely to detect. Work search violations and base period wage errors accounted for two thirds of these recoverable-but-hard-to-detect overpayments.
  Cost-Benefit Analyses. Cost-benefit calculations of different approaches for preventing or recovering overpayments were made using BAM and BPC data. One approach for preventing erroneous payments that appears to be cost-effective is to interview separating employers via telephone to ensure that the agency has complete information before making a nonmonetary determination to pay UI benefits. The analysis showed that interviewing separating employers prior to making a decision to pay benefits is cost effective: for every dollar spent this way the agency could prevent about $1.20 in overpayments. This type of follow-up with employers not only helps to prevent potential overpayments, but also fulfills one of the key requirements for quality nonmonetary determinations.
  A recovery approach that appears to be very cost-effective for some states is to increase their current BPC efforts to detect and establish overpayments, e.g., by making more intensive use of New Hire and wage-benefit crossmatches. Nationally, over four dollars could be recovered per additional dollar expended for BPC, even after allowing for some drop-off in establishment and recovery productivity. This is a cost-effective approach in states where overpayment amounts established for recovery are less than 80% of the amount estimated by BAM to be recoverable and which seem to be most detectable using current BPC methods. Using this measure of overpayment management, 41 states showed values of less than 80% for FY 2001. (This overpayment management measure is a new measure under the Government Performance and Results Act. Nationally, the target for FY 2003 is to establish for recovery 59% of the overpayment amount BAM estimates is detectable and recoverable. See UIPLs 33-02 and 15-03.)
  The analyses suggest that it is not cost-effective to expand BPC efforts to include additional steps to identify, establish and recover the types of overpayments that are not usually identified through BPC operations, e.g., work search verifications.

5. Recommended State Actions. The subject analyses are intended to provide ideas for changes in processes and their likely payoff. We recommend that states use them as models for their own analyses. In certain state environments, some of the process changes may prove to be highly cost-effective and should be given serious consideration for implementation.

6. Action Required. State Workforce Administrators are requested to:

• Provide this information to all interested parties;

• Replicate benefit-cost calculations using BAM data and data on how long claimant and employer contacts would take, to see whether the additional contacts referenced in papers cited above might be cost-beneficial in their state; and

• Focus on remedying errors in benefit payment operations where BAM indicates that sufficient data were available to make a correct decision or procedures were not used that could have detected ineligibility issues.

7. Inquiries. Please refer any questions to the appropriate Regional Office.

8. Attachment.
   Synopses of Erroneous Payments and Integrity Analyses

Official attachment (https://www.dol.gov/sites/dolgov/files/ETA/advisories/UIPL/2003/UIPL23-03_Attach.html):

Attachment
Attachment
Synopses of Erroneous Payments and Integrity Analyses
The complete analyses are available on the OWS Web site, https://ows.doleta.gov or by e-mail by contacting Burman Skrable at [email protected].

• The Distribution of UI Overpayment Rates and Dollars by State, CY 2001. This analysis arrays BAM's estimate of the percentages of dollars overpaid and estimated dollar overpayment amounts for all states by the point in the eligibility determination process where the error occurred. The analysis also makes a distinction between errors that are recoverable and those that are not recoverable due to provisions in state law. BAM estimated that total overpayments for CY 2001 were 8.2% of dollars paid, a total of about $2.5 billion for all causes. The three major causes were receipt of earnings and other disqualifying income while claiming UI benefits (31%); voluntary separation from work (21%; three fifths of these issues were not detectable by the state agency); and failure to meet work search requirements (15%). Eliminating technical and non-recoverable overpayments produces an operational overpayment rate of about 4.7%. This operational overpayment rate was discussed UIPL 33-02. The analysis covers all states except Colorado and Puerto Rico (which did not complete a sufficient number of BAM cases to produce reliable estimates) and the Virgin Islands, which has no BAM program.

• Detectable and Recoverable UI Overpayments in FY 2001. The BAM data used in this analysis indicates that about 85% of the instances—73% of dollars—of overpayments occur because UI agencies cannot detect eligibility issues at the time payment decisions are made. The primary reason states could not detect the issues was because pertinent information--such as whether a claimant who returned to work continued to file for UI benefits--was not available when the payment decision was made. Further, BAM estimates that 21% of dollars overpaid cannot be recovered after they are made because of state finality rules, or because the agency is responsible for the error. This amounted to nearly $500 million in FY 2001. Of the remaining $1.8 billion, state BPC operations are most likely to detect about 69%. This most detectable portion of recoverable overpayments amounted to about 54% of all overpayments—in FY 2001, about $1.235 billion out of $2.285 billion overpaid in total. The analysis concludes that this $1.235 billion of recoverable and most detectable overpayments is the most reasonable basis for judging the completeness of state BPC establishment efforts.

• Cost Benefit Analysis of Reducing UI Payment Errors. This paper combines data from BAM, Denied Claim Accuracy and administrative reports to identify where errors occur in the payment process and how these errors relate to the extent of information is available to agencies when they make payment decisions. The paper's primary focus is on the analysis of overpayment errors. It concludes that these errors occur primarily because agencies lack enough information to know that a claim has an eligibility issue. This is particularly the case with errors made at the continuing eligibility level, which account for about 70% of all dollars overpaid. However, BAM data indicate that overpayments due to voluntary separations have the highest average weekly overpayment—$200 nationally—because these errors involve the whole week's payment amount. Thus, efforts to prevent or recover separation errors should be given priority. (The average continuing eligibility error is about $135, smaller than separation errors because many involve partials and part-total benefits. The higher incidence of such errors accounts for their predominance among total overpayments. The average monetary error is $33.) The paper develops a methodological framework for analyses to explore the benefits and costs of different approaches to attempting to prevent overpayments and of attempting to detect and recover them after the fact. This framework has been the basis for National Office analyses and could also be the basis for state analyses.

• Preliminary Cost Benefit Analysis of Identifying Overpayments BPC Misses. This analysis explores the costs and benefits of agencies taking additional steps to identify, establish and collect the types of overpayments that they do not usually identify through BPC operations. It explored three alternative telephone contact approaches, all carried out at the 10th week of the claim: contacting base period and separating employers to obtain eligibility information; contacting the claimant to obtain eligibility information; and (based on a prior claimant contact) following up with work search employers to verify work search contacts. Benefits were computed using BAM data on the incidence of error and overpayment per case to determine likely payoff in recoveries. Costs were based on data from a 1990 telephone claims pilot that indicated the time to make and complete telephone contacts. None of the approaches appeared to be cost-effective; benefit-to-cost ratios ranged from about 0.01 to 0.06—in other words, at best, spending a dollar on this kind of follow-up would yield about 6 cents in recoveries.

• Cost-Benefit Analysis of Expanded Employer Contacts. This analysis complements the preliminary cost-benefit analysis (above) by exploring three scenarios for interviewing employers via telephone before the first payment is made, i.e., what might be the value of obtaining additional information in time to prevent overpayments instead of only in time to detect and recover overpayments. (The analysis did not explore potential effects on timeliness of nonmonetary determinations or first payments.) Three scenarios were explored: (1) interviewing all base period and separating employers to obtain monetary, separation, and related information; (2) interviewing all separating employers at the time a claimant filed either a new or additional claim to verify separation eligibility; and (3) interviewing separating employers to ensure complete information before a decision to pay benefits is rendered on a separation issue. As with the previous study, BAM data for FY 2001 plus information from the telephone pilot study were used to compute costs and benefits. The results showed that only (3), interviewing separating employers prior to making a decision to pay benefits after a separation issue has been raised, is likely to be cost effective: the ratio was over 1.20. This analysis suggests that not only does ensuring contact with employers fulfill one of the key requirements for a quality nonmonetary determination, but it is also a cost-effective step in preventing potential overpayments. The analysis also noted that about 30% of separation overpayments occurred because the agency failed to detect an issue or draw the right conclusion even though it had adequate information. Therefore, it would be valuable for states to examine the extent to which overpayments occur because they are mishandling information that would identify eligibility issues or are drawing the wrong conclusion on separation issues and other issues from adequate information.

• Cost-Benefit Analysis of Expanding Benefit Payment Control Activities. This analysis explored the costs and benefits of expanding the scale of BPC activities in 41 states. In these states, the overpayment amounts established in FY 2000 and FY 2001 averaged less than 80% of the amount BAM estimated was most detectable and recoverable by BPC. BPC cost data for FY 2001 were obtained from the Resource Justification Model. The basic analysis assumed that:
  states could increase their ratios of overpayments established to the BAM estimate of what was most detectable and recoverable by BPC by one half of the difference between where they are and 80% (this seemed to represent a reasonable and attainable level of effort for most states); and

• the productivity in establishing and recovering additional overpayments was the same as it had been in FY 2000 and 2001.
  The analysis also examined whether establishment and recovery productivity might decline if effort increased and concluded that declines might be on the order of 5 to 10 percent. It concludes that nationally, over four dollars could be recovered per additional dollar expended for BPC, even after allowing for a 10% drop-off in establishment and recovery productivity.