Cost-Benefit Analysis of Expanded Employer Contacts

Background and Summary

This analysis explores the likely costs and benefits (in terms of reduced overpayments) of following up with employers when new or additional initial claims are filed. It uses BAM data on the extent and causes of overpayments for CY 2001 to estimate the benefits of additional employer contacts in preventing overpayments. A 1990 pilot test of conducting BAM investigations by telephone provide data on the time required to make certain kinds of employer contacts, the basis for cost estimates.

The analysis explored benefits and costs of UI agencies interviewing employers via telephone in three different scenarios. The three scenarios, and findings, are:

• Interview all base period and separating employers when the claimant files a new initial claim and is found monetarily eligible to query the employer about all relevant issues (monetary eligibility, separation eligibility, alien status, and severance/vacation pay).
  • Result: Not cost-effective. Benefit/Cost ratio is 0.14 (only 14 cents in overpayments prevented per each additional dollar spent.)
• Interview all separating employers whenever a new initial or additional initial claim has been filed, to query the employer only about separation issues.
  • Result: Not cost-effective. Benefit/Cost ratio is 0.24
• Interview the separating employer to ensure that full information has been obtained before making a decision to pay UI benefits when a separation issue exists. (This step is also a condition of a high-quality nonmonetary determination.)
  • Result: Cost-effective. Benefit/Cost ratio is 1.2.

We thus recommend that states replicate the analysis of the third approach using their own BAM data and figures on contact costs, and change procedures accordingly.

Analysis

All errors BAM finds receive a "Prior Employer Action" code, indicating the completeness of relevant information employers provided to the agency. Table 1 shows that during CY 2001, about 45% of all overpaid weeks, and 54% of dollars overpaid, were not employer issues. In another 8% of the weeks, and 9% of the dollars, the employers provided adequate and timely information. However, for about 48% of the weeks overpaid, and 37% of dollars overpaid, employer response was deficient, either because the agency failed to request it, or because the employer responded late or with inadequate information, or did not respond at all.

Table 2 shows the extent of employer response for overpaid weeks for the standard BAM overpayment cause codes. According to BAM, for some causes of error the employer input or response was either not an issue or was complete (e.g., 94% in the case of work search errors); for others, the opposite was true (e.g., 92% inadequate or missing in the case of Base Period Wage errors). For three error cause categories, responses from the employers from whom the agency regularly requests information were at least potentially useful in preventing errors in over half the cases: Base Period Wages, Separation issues and Separation/Vacation pay.

According to the BAM data summarized above, employers may hold the key to identifying about 48% of the weeks, and 37% of the dollars, overpaid. Additional contact, presumably by telephone, could obtain most or all of such information. The most reasonable possibility for follow up is to contact employers at the time an initial claim is filed to obtain full information regarding the claimant's eligibility to prevent errors. Table 3 takes the information from Table 2 by cause for which employer input could possibly have prevented an error, and indicates whether the cause involved a decision largely made at the initial claim, or continued claim, level, and shows the average overpayment found with each of the type of error.

Table 3 suggests that two alternative approaches can be considered:

• 1. Contact the Separating Employer and All Base Period Employers at the Time a New Initial Claim is Filed. BAM data suggest that the average claimant has about 1.5 base period employers. A contact would explore all eligibility issues; these are most likely to involve Base Period wage, separation, dependents, illegal alien status, and severance/vacation pay issues. About 61% of separation determinations are made at the new initial claim level, so the follow-up at this level would not detect the 39% of separation issues presumed to occur at the additional initial claim level.
  • Table 3, "All Initial Claims" line, suggests that this approach would identify an overpayment in about 5.2% of the cases examined, but because most of them are base period wage errors, the average detected per claim with an error would be small, about $60.
• Contact Separating Employers. Alternatively, states could contact just the separating employers, when both new and additional initial claims are filed. Although such contacts may yield other eligibility information, we assume that it will be primarily information on separation issues and on separation/vacation pay issues. This could be done in either of two ways:
  • 2. Contact all base period and separating employers known when a new initial claim is filed. Although the "All Separations" line of Table 3 suggests that contacting all separating employers would obtain information relevant to an overpayment in 1.3% of the claims examined, other BAM data indicate that employers are not considered responsible for all those overpayments; overall BAM assigns employer responsibility for only 30 percent of those weeks. This implies a productive finding in 0.6% of weeks examined.
  • 3. Contact only employers involved in affirmative nonmon determinations. A more selective approach would be for the agency to ensure that it obtains full information from employers on all separation issues it has detected so that it determines eligibility correctly, and particularly with employers on separation adjudications it affirms. In a typical year, about 21% of all separation determinations involve adjudicable issues. Of these, about 50-55% are denied, and so about 10% of all separation determinations result in affirmative adjudications. BAM data on Prior Agency Action, used in conjunction with the data on Prior Employer Action, indicate that in about 0.3% of all separation determinations involving an overpayment the agency either failed to follow through to determine there was an issue or failure to follow procedures that would have detected an issue. Thus, ensuring that full information is obtained from employers on adjudications the agency was ready to affirm will turn up additional ineligible claims. These 0.3% of all separation determinations involving an overpayment are concentrated among that 10% of contacts that involve affirmative adjudications. Thus, examining only affirmative adjudications should find an otherwise undetected overpayment with about 3% of the employers contacted. Follow-up in this fashion may also detect some severance or vacation pay errors.

Costs of Contacts

The most relevant information on the time it might take to contact the employers and to perform the two follow-up investigations is the time measurement done during the 1990 BAM Telephone Pilot Study. It yielded the following the average times for obtaining BAM information by telephone:

• Claimant interview, 35 minutes;
• Base Period wages, 24 minutes;
• Separation investigation with base period employers, 5 minutes;
• Other investigations with BP employers, 10 minutes;
• Work Search, 10 minutes;
• Third party contacts, 5 minutes;
• Benefit Year Earnings investigation, less than 5 minutes;
• Locating respondents, 35 minutes.

We used these figures to derive cost estimates for the analysis, as follows. In the pilot, a full investigation of base period wage, separation, and other initial-claim issues with base period and separating employers took about 39 minutes, or 0.65 hours. The simpler separation investigation averaged BAM only 5 minutes in the pilot test. BAM staff averaged 35 minutes to locate all respondents; however, claims staff should have little trouble locating most base period and separating employers if agency records are up to date, and so we have used 8 minutes for all base period and separating employers and 5 minutes for separating employers only. In sum,

• All Base Period and Separating Employers.
  • Contacts with employers: 39 minutes per claim
  • Locating Employers: 8 minutes per claim
  • Total investigation: 47 minutes per claim (0.78 hours)
• Separating Employers Only.
  • Contacts with employers: 5 minutes per claim
  • Locating Employers: 5 minutes per claim
  • Total investigation: 10 minutes per claim (0.17 hours)

The costs of each follow-up use the above times and the national average UI staff wages. The calculation assumes that state staff wage rates per hour in 2001 were $28.93 (based on FY 2001 budget factor of $52,083 per year, 1800 hours per year worked). At this rate, the 47 minutes for a base period and separating employer investigation costs about $22.57 per contact; the 10 minutes to locate and contact only the separating employer cost $4.92.

Return from the Investigations

Table 3 indicates that the combination of monetary, separation, and other overpayments detected in the contact of all base period and separating employers averages about $60 per error. The average overpayment detected through contact with separating employers (both all, and those involved with known affirmative separation decisions) is about $201.

Cost-Benefit Calculation

Table 4 shows the expected results from the three different approaches to enhanced employer contact. In each case, the column labeled "discovered per investigation" is the product of the dollars per error discovered on the average "hit" and the chance of making such a hit. Thus, for example, contacting all base period and separating employers should find an error about 5% of the time. 100 contacts would yield about $300, an average of about $3 per contact.

Conclusion:

This calculation indicates that is not likely to be cost-effective to make additional telephone contacts with all base period and separating employers about all initial claim issues, or with all separating employers about separation issues. At best, these approaches return only about a quarter in prevented overpayments per dollar spent in the effort. Ensuring that employers are fully contacted before any affirmative separation adjudication is rendered, however, appears to be cost-effective. The estimated payoff using national average is about $1.20 for each $1.00 spent in the effort. This analysis suggests that not only does such a step fulfill one of the key requirements of high quality nonmonetary adjudications, but it is also cost-effective. The cost of the time spent making the contact will be less than the amount of overpayments prevented. It would thus be worthwhile for states to make their own calculations of follow-up costs and expected benefits using BAM results.

Agencies could consider more extensive follow-up with all employers involved in separation adjudications to ensure they had full information. This would reduce the cost-benefit ratio by about half (to about 0.6) but it would also prevent some erroneous denials. Preliminary data from the Denied Claim Accuracy program indicate that, in FY 2002, about 7% of denials were erroneous and employers were responsible for about 20% of those errors.

This and every calculation that relies on efforts to improve the extent and quality of issue detection assume that the agency draws the proper conclusion from the additional information. BAM data also show that in CY 2001 about 30% of separation errors occurred because agencies drew the wrong conclusion or failed to detect an issue from adequate information. Providing training to ensure that nonmon staff correctly use available information is a separate matter from obtaining that information, but closely complements it. Agencies also need to examine BAM data on Prior Agency Action to determine the extent that their own nonmon errors occur because of improper handling of adequate information versus the lack of adequate information. In some cases, additional training in resolving issues correctly may yield more cost-effective results than additional efforts to detect issues.