Introduction gathered about the reasons for and causes of earnings differences, as well as the corrective action needed. To determine how SSA and IRS ensure that employers consistently report employees' earnings to both agencies, we examined and analyzed records documenting their efforts. We also reviewed SSA and IRS's formal agreement, which stipulates each agency's responsibilities under a system of combined annual wage reporting. To better understand SSA's and IRS's roles and responsibilities on certain legal and procedural questions related to reconciliation activities, we requested the views of the Secretary of Health and Human Services (HHS), through the Commissioner of Social Security (see app. III), and the Secretary of the Treasury, through the Commissioner of Internal Revenue (see app. IV), and discussed the matter with SSA and IRS officials. To determine the actual effects of uncredited wages on beneficiaries and the potential effects on employees (still working), we reviewed SSA and IRS records. These included correspondence, studies, and statistics on employers whose earnings reports (based on IRS's match of IRS and SSA employer reports) may not have been recorded by SSA. In a 1985 study to examine the usefulness of data IRS furnished (indicating that SSA may not have received certain employers' earnings reports), SSA sampled a group of employers for whom SSA had no wage total or a lesser total than IRS. SSA asked the employers to provide some employee W-2s to help SSA track whether it had received and recorded the employees' wages. To determine whether the 1980 and 1981 W-2s had been accurately credited, we checked the W-2s employers provided for 11,151 employees against each individual's Social Security earnings record. For wages that had not been accurately credited, we measured the actual effects on current beneficiaries and, for those still working, we estimated the potential effects on survivors. To measure the actual effects of uncredited earnings, we identified and then selected certain individuals whose earnings had not been credited and who were receiving benefits. At our request, SSA calculated the benefit amounts and monthly and retroactive underpayments. We also estimated a potential effect for individuals who were not yet receiving benefits. Rather than trying to forecast what each individual would earn until retirement, we asked SSA to assume the individual died, leaving a survivor; a benefit payment could thus be calculated based on the actual earnings of each individual. At our request, SSA computed benefits payable as of January 1984, with and without the uncredited earnings. We then calculated the difference in benefits for each individual's survivor. Introduction The measurements of actual and potential effects of uncredited wages are not projectable to all employees with uncredited wages because the universe of affected employees is not known. Rather, the results serve as an indication of the payment effect on employees who were identified as having uncredited wages. To measure these effects for the self-employed, we selected 7,100 selfemployed individuals with uncredited 1979 earnings (we had identified them during a previous review3 ). We then used the same procedure to calculate the difference in their benefits that we had used for employees. Again, the results are not projectable to the universe of all selfemployed, but serve as an indication of the payment effect from the uncredited self-employment earnings. Our work was done from October 1984 to December 1986 in accordance with generally accepted government auditing standards, except that we did not assess the reliability of the IRS system used to record employers' quarterly federal tax returns or the SSA System used to record employees' earnings. 3IRS and SSA Can Improve the Verification and Recording of Data Provided by Self-Employed Taxpayers (GAO/GGD-85-21, May 28, 1985). Chapter 2 IRS and SSA Earnings Differences Total More Than $58 Billion Since 1978, when employers were first required to report employees' earnings annually to SSA and quarterly to IRS, the total earnings amounts recorded by each agency (as reported to them by employers) have differed. SSA, as of March 1987, had recorded about $58.5 billion less than IRS for 1978-84 (data after 1984 are preliminary). These earnings differences, through 1983, involved about 3.5 million earnings reports for about 2.5 million employers. About one-fourth of these employers' earnings reports showed differences between the two agencies for more than 1 year. Chapter 3 illustrates the effects of these differences on some beneficiaries and employees. Chapter 4 discusses the various causes of these differences. SSA Has Consistently Since 1978 employers have been required to report employees' earnings to IRS quarterly (941s) and to SSA annually (W-3s).1 The total of each employer's four quarterly reports to IRS should equal the total earnings that an employer reports annually to SSA. When differences arise in employer reports, it becomes a matter of concern. IRS is usually concerned that the employer reported more earnings to SSA than to IRS; this means that taxes are due on earnings not reported. SSA is usually concerned that the employer reported more earnings to IRS than to SSA; this means that employees' earnings were not credited to their Social Security accounts. Total earnings recorded by each agency and the differences that have existed since 1978 are shown in table 2.1. 'Before 1978, all employer reports with detailed employee data were provided quarterly to IRS, which forwarded the data to SSA. The change to the current format was intended to eliminate the burden of detailed quarterly reporting. As of March 1987, SSA had not resolved the $58.5 billion shortfall in its records for the 1978-84 period. However, the $58.5 billion not recorded by SSA represents only about 0.8 of 1 percent of the $7,412.7 billion that SSA credited from 1978-84. Earnings Differences IRS totals the 941s for each employer and compares this total with the W-3 data SSA recorded for the employer. Differences identified (discussed in ch. 1) indicate employers who may not have reported some or all of their employees' earnings. The number of missing and discrepant employers' earnings reports for 1978-83 are shown in table 2.2. The number of employees potentially affected cannot be determined because the IRS data do not show this number but, rather, the amount by which the SSA and IRS earnings totals differ for specific employers. Some Employers With Reporting Differences in More Than 1 Year Although about 3.5 million employers' earnings reports are missing or discrepant, this number does not represent 3.5 million different employers. There are some employers with reporting differences in more than 1 year. To determine the extent to which reporting differences attributable to some employers occurred repeatedly over the 6-year period, we checked the frequency with which the same EIN was on the SSA-IRS unreconciled lists for 1978-83. The results are shown in table 2.3. About 2.5 million employers accounted for the 3.5 million missing or dis- |