A system for detecting payment errors in the unemployment insuranceprogram was recently developed by the U.S.
Department of Labor. Thissystem has made it possible to identify the level of both fraud andnonfraud overpayments, as well as underpayments, in the program. Priorto the introduction of this detection system, it was not possible todetermine the extent and nature of payment errors.
Currently, the detection system–known as the random auditsystem–is operating in 46 unemployment insurance jurisdictions. Theremaining jurisdictions will be included in this program or itssuccessor (the UI quality control program) during fiscal year 1985. Atthat time, the audit system will provide a basis for: (1) estimating theextent of payment errors in the nationwide unemployment insuranceprogram; (2) indentifying the primary sources of the payment errors; (3)implementing corrective action, where appropriate; and (4) evaluatingthe effects of such corrective actions (or other programmatic changes)on unemployment insurance payment accuracy. This summary discusses thedesign and methodology of the random audit system and presents findingsfrom the pilot tests conducted in five States–Illinois, Kansas,Lousiana, New Jersey, and Washington–over a 1-year period ending inMarch 1982. Because of the large volume of weekly payments made in theunemployment insurance system, it would be prohibitively expensive(under current law and policy) to verify each claimant’seligibility to receive benefits. Thus, the random audit system relieson a small sample of payments made in each unemployment insurancejurisdiction as the basis for estimating the extent and nature ofpayment errors.
The payments selected for investigation are taken froma specially constructed computer file of weekly statewide unemploymentinsurance payments in each participating jurisdiction. Each week, aprobability sample of cases is selected from the file, and the resultsof verifying benefit eligibility for those cases are used to estimatestatewide payment errors; quarterly estimates are developed for eachunemployment insurance jurisdiction. After a sample has been selected for review, a detailed andconsistent procedure is followed. When cases are selected forinvestigation, it is assumed that claimants have been properly paid, andthis opinion is changed only if documented evidence to the contrary isrpresented. Verification of benefit eligibility includes the followingprocedures: (1) files related to the case are obtained and reviewed; (2)the base period wages upon which the claimant established his or herclaim for benefits are verified (with employers if possible); (3) apersonal interview with the claimant is conducted to verify relevantfacts regarding the individual’s claim for benefits; (4) theclaimant’s reasons for separation from previous employers areverified to determine if any disqualifying circumstances were involved;(5) attempts are made to verify if the claimant was able and availablefor work during the sampled week; (6) if applicable, employers listed bythe claimant as work search contacts during the sampled week arecontacted for verification as to whether the claimant actually appliedfor work; (7) as appropriate, attempts are made to determine if theclaimant refused any offers of “suitable” work that woulddisqualify the individual from receiving benefits; (8) attempts are madeto determine if the claimant accurately reported any earnings or workperformed during the sampled week; and (9) depending on thecircumstances of the case, other individuals may be contacted to verifyany other determinants that could affect the claimant’s eligibilityfor benefits during the sampled week.
On the basis of information acquired during the verificationprocess, the field investigator makes a judgment as to whether theclaimant met eligibility requirements for the benefits received. If anoverpayment is suspected, careful review procedures are followed. First,the investigator interviews the claimant a second time in order toprovide the claimant an opportunity for rebuttal of evidence acquiredduring the investigation. Second, a review is requested from themanager of the local unemployment insurance office in which the claimfor benefits was filed. Third, the case file is reviewed by the Staterandom audit system supervisor and, in some cases, by a Federal reviewteam (representing the national office of the unemployment insuranceservice).
If the State determines that the payment was in error and theclaimant files for an appeal, a representative of the State random auditunit is available to present relevant evidence affecting the case. Inthe event of a reversal of the overpayment determination, the resultsrecorded for the case are modified to reflect this final status of thesampled payment. Verifications of benefit eligibility are conducted by unemploymentinsurance personnel from each participating jurisdiction to ensure thateach sampled case is reviewed in accordance with the respectiveState’s law and policy. Each full-time unemployment insuranceinvestigator assigned to the random audit program normally receives nomore than three cases on a weekly basis. In contrast, a full-time Stateunemployment insurance claims examiner assigned to a local officetypically would process at least 50 times as many cases in a week.
Limitations of the random audit system Several limitations of the random audit system and its data shouldbe noted. For example, the random audit system and its data should benoted. For example, the random audit system tends to produce”low-side” estimates of the payment errors that characterizeState unemployment insurance programs. This tendency appears to resultfrom the following: First, unemployment insurance benefits are paid withat least a 1-week lag, so that “ex post facto” efforts arerequired to determine if benefits have been paid in accordance with theState’s employment security law and policies; the longer theseinvestigations are delayed, the more difficult it is for claimants andothers to accurately recall relevant facts, making it more difficult todocument payment errors. Second, the provisions of each State’semployment security laws and policies limit the extent to which aclaimant’s activities may be investigated to determine if a paymenterror occurred. Third, because of the very long time lags usuallyinvolved in detecting instances of unreported earnings in unemploymentinsurance-covered employment through a “postaudit,” thisprocedure is not utilized as part of the standard random auditinvestigation, resulting in some understatement of overpayments thatactually occur.
Fourth, unreported earnings in the “casheconomy” are extremely difficult to detect, even if”postaudit” procedures are utilitized. Fifth, sampledpayments are considered correct ulness documented evidence to thecontrary is made available; given the complexities of the employmentsecurity laws and policies that specify the eligibilitycriteria–especially those related to the “availability forwork” and “active-search-for-work” requirements–it islikely that overpayment errors are somewhat understated simply becauseunrefutable documentation could not be obtained. The nature of thepayment errors that cannot be detected by the random audit system issuch that many would be established as fraud overpayments if they weredetected; hence, the estimates provided by the random audit system offraud overpayments are very likely to be more understated than is thecase for all overpayments. The principal findings of the random audit system pilot tests aresummarized below.
These results are indicative of the types ofinformation currently being produced on a quarterly basis in the 46unemployment insurance jurisdictions in which the random audit system iscurrently operating, but it should be noted that a variety of other dataelements also are collected in this system. Table 1 shows the estimated percentages of weeks paid statewidewith either an overpayment or an underpayment of any amount. The totalpercentage of weeks paid with such errors ranged from 12.2 percent inLouisiana to 52.
1 percent in New Jersey; the findings also indicate thatoverpayment errors tended to be much more common than underpaymenterrors in the five pilot test States. Underpayments, as a proportion ofall dollars paid, were estimated to be 1 percent or less in each State,indicating the insignificance of underpayments. In sharp contrast, the rates of unemployment insurance overpaymentsin the five States ranged from 7.3 percent in Louisiana to 24.3 percentin New Jersey; overall, double-digit overpayment rates were estimatedfor 3 of the 5 States. A comparison of the percentage of dollarsoverpaid with the percentage of weeks overpaid indicates that paymenterrors of small dollar amounts were relatively frequent in these States.In Washington, for example, 20 percent of the weeks paid but only 9.
3percent of the dollars paid were estimated to be overpaid. Similarly,in New Jersey, 38.2 percent of the weeks paid but only 24.3 percent ofthe dollars paid were estimated to be overpaid.
The principal cause ofthese relatively frequent overpayments involving small dollar amountswas errors in the reporting or recording of base period earnings. Because of the historical interest in and concern about fraud inthe unemployment insurance program, a separate measure of fraudulent payments is provided by the random audit system; estimates for the fivepilot test States indicate that “officially established”fraudulent payments constituted only a small portion of the totaldollars paid in each State; fraud rates ranged from 0.2 percent inKansas to 2.7 percent in Louisiana.
As noted earlier, however, theabsence of postaudits to detect unreported earnings in coveredemployment and the difficulty of detecting unreported earnings in the”cash economy” tend to understate the “true”magnitude of the fraud problem in the unemployment insurance program. The random audit system also produces information on both the”types” and causes” of payment errors in the unemploymentinsurance program. Types of payment errors are classified on the basisof whether the error was the “responsibility” of theunemployment insurance claimant, covered employers, the Stateunemployment insurance agency, or a combination of the three. Causes ofpayment errors are classified on the basis of which aspects ofemployment security law or policy were violated, including: errors inthe reporting or recording of earnings during the sampled week for whichthe payment was made; errors in the reporting or recording of baseperiod earnings; violations of “continuing” eligibilitycriteria (refusals of suitable work, nonavailability for work, inactive job search); disqualifying reasons for separation from previousemployers; and other factors. In the current system, statisticalinformation is provided for specific causes of unemployment insurancepayment errors only if such causes account for at least 1 percent ofquarterly unemployment insurance payments. THE NATIONAL RANDOM AUDIT is a major step forward in controllingpayment errors in the unemployment insurance program. This is anessential program because it provides statistically reliable estimatesof payment error rates for entire unemployment insurance jurisdictions.This permits not only identification of payment errors, but also themeans through which the fundamental problems can be diagnosed andsolved.
Furthermore, the capability of the system to provide timelyevidence on such payment errors facilitates evaluation of the effects ofthe various types of corrective actions that may be undertaken inindividual unemployment insurance jurisdictions. The compilation ofthis systemwide data base should prove to be a valuable research tool.