EXECUTIVE SUMMARY

The Denial Pilot study was completed In the summer of 1988 following two years of design. Implementation. and analysis. During that time. a total of five states Implemented three denial quality control claim selection and Investigationprograms. This summary provides a brief overview of the major study highlights. A more detailed compilation of study results is presented In the main body of the report.

The major purposes of the pilot study were to develop empirical estimates of erroneous denial determinations and to test denial quality control on a limited basis In anticipation that It might be incorporated Into the current benefit quality control program. The pilot study also focused on testing the feasibility and cost of conducting denial QC investigations. and the estimation of the dollar consequences of erroneous denial determinations. The five states that participated in the study agreed to undertake this difficult Initiative at the same time that they were completing the successful Implementation of the core QC program and meeting the sampling and Investigation requirements for fiscal year 1987. The success of the pilot study Is a direct result of the efforts of investigators and supervisors in each of the states.

The pilot study implemented three methods or options for selecting claimants who had been formally denied unemployment Insurance benefits. The first method retained the core (paid) QC sample and added to it a sample of denied claims that were investigated and managed separately from core QC. One state implemented this option. The second method integrated core (paid) and denied claims into a single population from which a sample-of approved and denied claims were selected and investigated. One state implemented this option. Both of these methods focused on the selection of separate samples for each claimed week and for this reason are considered to be cross-sectional. The third selection method sampled initial claims each week. As denied claims and a sample of paid claims were determined, they were investigated. In addition,   this cohort was tracked throughout the year and each new determination was subjected to QC review.  Three states implemented this option.

The design specified that each state select three types of denied claims: those denied for monetary reasons; those dented for reasons of work separation; and those denied for nonmonetary, nonseparation reasons. The denial investigation focused exclusively on verification of the basis for the denial. All states utilized the same Investigation procedures subject only to variations in state law and procedure.

There are a number of limitations that affect the interpretation of the empirical results produced by the denial pilot study. The most important of these limitations include the following:

• The pilot design only included formal denials--those based on an official review of a filed claim. It did not
  include denials resulting from Informal inquiries made by potential claimants Nor did It Include potential
  claimants who chose not to file a claim.
• The interpretation of the results or effects of each of the claims selection options,is limited because. Options
  1 and 2 were tested in one state each. It is difficult to identify the impact of the selection method separate from
  the impact of state law. procedures and economic conditions.
• There is no accepted standard or benchmark that can be compared to the denial error rates associated with each of
  the options. Thus-it is difficult to judge which option is more or-less effective in detecting erroneous denials.
• In each of the options, a small proportion of denied claims were redetermined to be eligible. However, some
  of these claims were included in the sample but the-information supporting the redetermination was not
  included in the QC investigation. This sampling error had the effect of slightly reducing the estimated impact of redeterminations as a self-correcting mechanism.
• The study did not collect data about the behavior of denied claimants - whether they refiled, when they refiled, and the results of refiling - so that It was not possible to accurately estimate the actual dollar value of erroneous denials. To compensate, an econometric model and computer simulation were developed to approximate the range of behavior that the denied claimant might exhibit during a benefit year.
     
  Some of the major empirical findings for the five pilot test states include the following:
  
• The percent of denials In error varied from a high of 36-3 percent to a low of 7.1 percent depending on the type of denial determination monetary, separation. or nonmonetary nonseparation and the pilot state. The percent of denials in error was somewhat higher in the Option 3 states and lower in the Option 2 state. (See the Summary Table denial error rate at the end of the Executive Summary.)
• Claimant appeals and agency rederminations nations reversed a high proportion of original erroneous denials. Although the reductions produced by these reversals did not substantially reduce the extremely high error rates. in most instances the reduction was substantial. (See Dental Error Rate Corrected for Appeals and Redeterminations in the Summary Table.)   There is a significant self-correcting effect for erroneous denials.
• When the corrected denial error rate is considered in terms of the annual volume of claims. error rites rang, from a high of 5 percent to a low of less than 0.1 percent. (See Corrected Error Rate as Percent of all Determinations In the Summary Table.)
• The results of an econometric simulation to estimate the value of erroneously denied claims for a population of 104.000 claimants produced estimates ranging from $ 3.49M to $7.75M for monetary denial errors, $2.83M to $3M for separation denial errors, and less than $250.000 for nonmonetary nonseparatlons denial errors. These estimates were calibrated using the error rates developed in the five pilot states. The simulation focused on benefit week and not benefit year because of the absence of reliable data. (See Estimated Dollar Value of Errors In the Summary Table.)
• The mean time to complete denial investigations ranged from a high in one state of 101.7 percent of the mean time needed to complete core QC investigations to a low of 37.2 percent of the mean time needed to complete core QC investigations. (See Mean Time to Complete Case of Regular Core Investigation in the Summary Table.)
• Refusal by the claimant to cooperate in the investigation ranged from a high of 27.l percent to a low of. 5.2 percent. The corresponding rate for paid claims is between I and 2 percent. Denial QC must. address this high refusal rate because It threatens the accuracy of information used to check the accuracy of the original denial determination. (See Rate of Refusal to Cooperate in Investigation in the Summary Table.)

The results of the Dental. Pilot study provide important evidence about the relative severity of error in formal denied claims. There is no compelling evidence to suggest that one design option functioned more or less successfully than the other design. Each option encountered operational difficulties but those encountered by the benefit year option appear to have been most severe. These problems included sampling and repeated investigations of claimants who re-entered the U1 'system' during the benefit year.

SUMMARY OF EMPIRICAL RESULTS OF DENIALS QC PILOT TESTING