Attachment to UIPL 25-93
OPTIONS FOR REVIEWING BENEFIT CHARGING ACCURACY
Issue. Core RQC includes a Program Review of the SESA's system for allocating benefit charges. Should QC review benefit charging accuracy more intensively? If so, should this review cover charge decision accuracy, or charge allocation accuracy, or both? And, how should the review be done, and by whom?
Background. In most States, employers' experience-rated tax (contribution) rates are determined by a formula which includes the amount of UI benefits paid totheir former employees. For these tax rates to be correct, these benefits must be correctly "charged," or "charged back," to each employer's UI account. This first requires determining whether the employer is chargeable for a particular spell of unemployment (for some separation reasons, benefits are "socialized"—charged to the "pool account" instead of to the former employer's account). Secondly, if the claimant had more than one chargeable employer, benefit charges may be allocated among them all in accordance with State law.
Core RQC reviews the SESA's mechanism for allocating benefit charges as part of its review of the Account Maintenance functions. However, partly in response to concerns expressed by the Department's OIG in its 1985 study of experience rating (Audit Report Number 03-3-203-03-315), RQC proposed to include a more extensive review of Benefit Charging (BC) accuracy, subject to pilot testing. The OIG had studied the UI experience-rating of UI taxes in 12 States. Their findings suggested that employers might be systematically responding to SESA notices of potential charge by alleging that the claimant had left employment for reasons which would excuse them from charges. If true, SESAs use faulty data for making charging decisions and allocations; and too many benefits are socialized and UI tax rates are inadequately experience-rated. Accordingly, the RQC design plan envisioned pilot testing a Benefit Charging (BC) module: a more extensive review of total BC accuracy, to include (1) verifying the data on which chargeability decisions are made, (2) reviewing the chargeability decision itself, and (3) reviewing charge allocation accuracy. [A few States have simple charging rules, e.g., charge the last employer, and so not all chargeable employers are base period employers and not all the concerns which stimulated this pilot apply in all States.]
The Benefit Charge Accuracy Pilot.During 1991, six States—Minnesota, New York, North Carolina, Utah, Washington, and West Virginia—conducted a 9-month pilot test of a methodology for assessing BC accuracy in some detail. All but NY assessed charge decision accuracy by having BQC investigators obtain information pertinent to charge decisions from all base period employers in their BQC sample. (This aspect was irrelevant in NY: under NY's charging system every base period employer is automatically chargeable, regardless of the reason for the employee's separation.) The original design envisioned that charge allocation accuracy would be assessed for the same sample of cases using the employer charge statements. Differences in the States' accounting and computer systems and in their accounting system charging cycles meant that not all States were able to do this fully or in the same way.
Results: Charge Decision Accuracy.The basic findings on charge decision accuracy for the 5 States conducting this part of the review are displayed in tables 1, 2, and 3. These tables are taken from the Pilot evaluation, "Unemployment Insurance Revenue Quality Control: Benefit Charging Pilot Project. Final Report" [Abt Associates, January 11, 1993. In brief, the findings are:
Case Error Rate. In the 5 pilot States, the percent of cases with at least one charge decision error ranged from a low of 4.4% to a high of 20%, with three States' error rates close to 12%. (Table 1, A)
Decision Error Rate. The best indicator is the rate of errors per decision (i.e., per base period employer). This indicator shows the range from 2.8% to 12%. The three middle States' rates fell between 7 and 11%. (Table 1, B)
Separating Base Period Employers vs. Non-separating. The study showed that the accuracy of the charge decisions depends critically on whether or not the employer was the "separating" employer for purposes of determining the claimant's eligibility for benefits. SESAs adjudicate separation decisions pertinent to benefit eligibility but do not feel they have resources to follow up on reasons for separation from other BP employers (even though these determine chargeability).
Separating BP Employers: Error rates varied from 0% to 2.3%. All these errors were attributable to SESAs. (Table 1, B)
Nonseparating BP Employers: Error rates varied from 5% to 19% range. (Table 1, B) It is critical to note, however, that the SESA-caused error rate ranged only from 0.9% to 2.5%; the remainder was attributable to employers' failure to respond (timely or at all). (Tables 1, 2) In four of the 5 States, 74 to 89% of these charging errors were attributable to employers' failures to respond to notices of potential charge. (Table 2)
Charges to the Pool Account. Because of employers' failures to respond timely or accurately to notices of potential charge, charges to the pool account were lower than they should have been, not higher (as the OIG feared). The difference was very slight in one State but was as high as 5 percentage points in three others. (Table 3) The pilot collected no data on why employers failed to respond.
Results: Charge Allocation Accuracy. All six pilot States reviewed whether benefits for the sampled claims were allocated accurately among charged employers or to the pool account. (This portion of the review assumed that the SESA's initial decision to charge an employer was correct.) As shown in Table 4, the error rate was very low between 0 and 3% for four of the six States. State S-2 only issues charge statements once a year, on the same date for all employers, to employers whose former employees filed claims with benefit years ending during the preceding year. Thus it was able to assess charge allocations in only 74 of its 675 cases. In those it found 14 errors an error rate of 18.9%. State S-6, with a manual allocation system, had allocation errors in over 19% of its cases.
Clean vs. Complex Cases. The study assumed that if a case involved such factors as monetary redeterminations, overpayments or underpayments, supplemental checks, benefit offsets, etc. it might be more prone to charge allocation errors. Therefore, various such complicating factors were noted for each case. Their incidence in the pilot States is shown in Chart 1. Table 5 compares the rates of allocation errors for cases without such complicating factors ("clean" cases) and those with at least one complicating element ("complex cases"). Such complications contribute significantly to allocation errors:No State made more than 3% errors on "clean" cases. The incidence of errors was at least twice, and as much as 13 times, as high in the complex cases.
Associated Costs. To minimize the burden on participating States, the pilot did not track costs in detail. States were, however, asked to estimate the approximate effort required. In return for participating in the pilot, SESAs were given relief from investigating 100 QC cases (one staff year). All States but one reported that this one position freed up sufficient resources to investigate BC accuracy. (That State concluded that its procedures, which went beyond what DOL and Abt asked, probably caused the extra costs.)
Options for Including Benefit Charge Accuracy in QC
Benefit Charge Decision Accuracy
Option I: Include Charge Decision Accuracy in QC PRO:The pilot indicated that overall many charge decisions are made inaccurately at least 6.7% in 4 of the 5 pilot States.
SESAs have no mechanism for reviewing the total accuracy of these decisions since BC "falls into the cracks" between tax and benefit operations.
Most SESAs no longer adjudicate separation decisions which have no bearing on claimant eligibility. When nonseparating BP employers fail to respond to charge notices, SESA staff lack complete information on reasons for separation, and make charge decision errors. It could be argued that even though this is employers' failure to act in their own interest, SESAs still have responsibilities to their employer customers and should at least measure the seriousness of the problem to decide whether some remedy is needed.
The review is not costly if added on to BQC investigations.
Charge errors cause errors in the experience rating system.
The error rate attributable to SESA procedures is miniscule approximately 2 percent at the most.
The great majority of errors is attributable to employers' failures to act in their own interest by responding to notices of potential charge and inform the SESA why the claimant was separated from work.
In the States examined, the benefit charge errors actually increased the degree of experience rating. If this did not actually increase trust fund revenues,it at least lowered the percent of charges that were socialized.
Core RQC does review the chargeability decision for the claimant whose case is selected for the Acceptance Sample review. (This review, however, is based on the information available in SESA records and thus can only identify SESA-caused errors.)
PRO: See CONS, above.
CON: See PROS, above.
Charge Allocation Accuracy
Option 1: Stick with Core RQC Review as designed, at least until some experience is gathered.The Core RQC review involves drawing acceptance samples of 60 employer charge notices. If that notice lists a claimant for whom the employer received a credit, that claim is to be selected for followup, to ensure a complex case; if there is none, the median claimant is investigated. The review ascertains, based on the information in SESA records, whether the charge for that one claimant is properly allocated to the employer's account or to the pool account, or that the credit is correct.
PRO:This review is already being implemented and is intended to assess allocation accuracy. It is extremely low-cost.
Pilot results suggested that States fall into either of two groups: extremely good or extremely inaccurate. The existing Program Review should be able to discriminate among them.
Whether a more extensive review is needed can be determined by comparing theresults of the Core RQC Program Review with the results of the Benefit Charge pilot for the 6 pilot States.
The existing Core RQC acceptance sample (AS) review may miss some potential errors because it is limited to the review of the accuracy of the allocated charge for only one claimant per employer selected.
This approach will include a very high ratio of claims with only one BP employer; for these, "allocation" should be moot.
This approach can ensure that cases with 2+ chargeable employers are selected, among whom allocation errors are relevant.
The review is not highly time-consuming.
This methodology is more thorough because it examines all chargeable employers involved in a given case, not just one.
Half of the States had charge allocation accuracy rates which were above our accuracy standard. Allocation accuracy may not be a serious enough problem to go beyond the existing design.
The existing Core RQC program review may suffice to detect insufficient allocation programs, and is already being implemented.
Implementing this will entail some programming costs to assemble the new sampling frame.
This would force a review of allocations to all employers involved in a UI payment.
The cases most error-prone as shown by the pilot would be explicitly reviewed.
This would involve reprogramming; may be difficult and expensive in the short run.
For any given sample size, the investigative effort will be greater than required now for Core RQC.
This effort may not be necessary. The current design may afford a sufficiently thorough review.