APPENDIX A

ANALYTIC METHOD AND DETAILED RESULTS

This appendix provides additional details about the comparisons of federal and private-sector employee benefits prepared by the Congressional Budget Office (CBO) in consultation with Watson Wyatt & Company of Bethesda, Maryland. The amounts compared represent the present value of benefits earned in a year by five hypothetical employees. The dollar values cover major benefits for rank-and-file employees. CBO selected analytic methods and assumptions so that the comparisons would show federal benefits to their best relative advantage. The estimates may therefore be viewed as offering an upper bound on the generosity of the federal benefits included in the comparisons. Appendix B shows results under a different set of assumptions.

The Watson Wyatt & Company Database

The dollar value of private-sector employee benefits used in the comparisons reflects the practices of 800 firms in a database maintained by Watson Wyatt for 1996. Those firms employ nearly 12 million workers. They are also large: about 90 percent of them employ more than 1,000 workers (see Table A-1). Those large firms account for nearly the entire workforce covered by the database. By contrast, census data for 1994 show that only about half of all employees work for firms of 500 or more employees. Also, firms in the database are disproportionately engaged in manufacturing--about 34 percent compared with 6 percent of all U.S. firms.
 

The values presented in this analysis should not be interpreted as representative of the private sector as a whole. The large firms in the database are more likely to offer benefit packages than are other, smaller private firms. For example, all 800 firms offer some retirement program, but data from the Employee Benefits Research Institute show that, nationwide, 40 percent of all employees have no employer-sponsored retirement plan. Because large firms predominate in the database, the results reported apply best to jobs for which such firms would be more likely to serve as a source of competition in recruitment. Those jobs would include many professional and administrative jobs as well as some higher-level technical positions. Such jobs make up a growing share of the federal workforce--up from 58 percent in 1985 to 70 percent in 1995.

Comparing Federal Retirement with Private-Sector Practices

The retirement values compared are the present values of benefits earned for the current year of employment. Watson Wyatt & Company computed separate values for the Civil Service Retirement System (CSRS) and each part of the three-part Federal Employees Retirement System (FERS). Values computed for FERS assume that employees have been covered by the program for their entire career, even though the length of the careers assumed for some of the hypothetical employees exceeds the length of time the program has been in existence. For the private sector, values representing the average for the database are shown for two types of retirement programs--defined benefit and defined contribution plans. (Averages for each type reflect the fact that some employers have no plan of that type.) Separate values are also shown for Social Security. As described below, the method of calculating the values used in comparisons varies by type of retirement plan.

Estimating Values for Defined Benefit Pensions. The values calculated in the comparisons for defined benefit plans represent the present value of future benefits divided by the expected length of service (the projected unit credit normal costs). Generally, the values are the amount the employer would have to put aside in a year in order to have enough on hand at a hypothetical employee's retirement to pay the benefits earned in that year.⁽¹⁾

Making such computations requires assumptions about expected salary growth, age at retirement, mortality, separation rates, interest rates, and, if the benefit is indexed, inflation. Watson Wyatt & Company used the same set of assumptions in calculating values for all pension plans (see Table A-2). Assumptions reflect federal practices based on data from actuaries at the Office of Personnel Management. Regarding mortality, the analysis assumes a fixed schedule of deaths through age 110. That schedule reflects mortality rates in the 1983 Group Annuity Mortality Table for male lives with a three-year age setback (that is, the mortality rate used approximates that for both males and females).
 

Estimating Values for Social Security. The amounts used in comparisons for Social Security in the private sector and in FERS, consistent with the approach used in assigning values to defined benefit programs, represent the present value of future benefits earned in a year. The level of those future benefits reflects provisions of current law, which could change. Calculations assume that employers pay for half of the benefits under Social Security, reflecting current law under which employees and employers each pay 6.2 percent of pay up to the taxable wage cap.

Estimating Values for Defined Contribution Pensions. The dollar values compared for defined contribution plans are simply the employer contribution that the employee earns during the year. The employer contribution is calculated as the amount the employer would match for a given level of employee contribution, plus any automatic contributions. All employees are assumed to make contributions according to the same fixed schedule that varies contributions by income. Thus, differences in values among plans reflect differences in the generosity of employer contributions and not differences in the level of actual employee participation in various plans. The assumed employee contribution for the hypothetical employee with a salary of $25,000 is 3.25 percent of pay; for the two employees with an annual salary of $45,000 and the one with a salary of $50,000, 5.5 percent of pay; and for the employee at $75,000, 6.1 percent of pay. For plans that capped contributions below the levels just described, calculations assumed the lower contributions.

Comparing the Government's Employee Health Insurance Program with Private-Sector Practices

The comparisons value health insurance as the employee's estimated medical costs covered by insurance in the year, minus any contributions the employee makes. The method for calculating the values involves two steps. First, Watson Wyatt & Company estimates a package of medical costs that each hypothetical employee could be expected to incur in a year. It then applies the provisions of each insurance plan against those medical costs to determine the portion each plan would cover. Private-sector values for each hypothetical employee represent the average medical costs covered for all firms in the database. The amount for federal employees is the weighted average medical costs covered by four typical plans: the Government Employees Hospital Association's standard benefit plan, the Kaiser Foundation's standard health plan for the mid-Atlantic region, Blue Cross and Blue Shield's standard benefit plan, and Mail Handlers' high-option benefit plan. Together, those four plans cover about half of the federal civilian workforce (see Table A-3).
 

Watson Wyatt estimates the package of medical costs for each hypothetical employee using an extensive database that tracks medical expenses and the use of medical services by age, sex, type of insurer, employment status, and family status. (Most major medical services are represented. They fall into the categories of medical care, alcohol and drug treatment, physician office visits, physician and other charges for surgery, physician and other charges for obstetrics, treatment for nervous and mental conditions, prescription drugs, lab and X-ray services, home health care, hospice care, nursing and other extended care, and emergency room care.) The database contains information on average costs for specific services and user groups and on the units of services purchased, which allows the calculation of copayments and cost sharing. Data on medical expenses come from private firms and insurers. Information on dental expenses was provided by the Michigan chapter of the American Dental Association.

Watson Wyatt & Company estimated a different package of medical costs for each hypothetical employee. Each package represents the costs that claims experience in the database indicates is typical for employees with characteristics that match those of the hypothetical employees. For purposes of developing the packages of medical expenses, each employee is assumed to be a composite of a single person and a person with dependents. Forty-two percent of the costs in a package are those for a single person, 22 percent are for an employee and one dependent, and 35 percent are for a family. Calculations assume that 1 percent of all employees would elect no coverage.

Once the package of medical costs has been developed for each hypothetical employee, Watson Wyatt applies the provisions of each insurance plan against those costs to determine the portion that would be covered by insurance. All major provisions are included in the analysis, including deductibles and payment percentages, employee copayments, and out-of-pocket limits. In applying the provisions, Watson Wyatt holds medical costs for each employee constant, except when estimating values for HMOs. In that case, some variation is adopted in the packages of costs to reflect the unique claims experience of HMOs; HMOs generally emphasize preventive care and early intervention and tend to have lower rates of hospitalization than other plans. Watson Wyatt adjusts its database to approximate employees' experience in HMOs using information collected from selected HMOs, academic studies, and other sources.

For preferred-provider plans, the estimates assume that 70 percent of the services occur within the network. The expenses under a point-of-service plan assume that 90 percent of the services are provided within the network.

By holding the package of medical costs fairly constant in determining the amounts of health insurance benefits, the comparison understates the value of the Federal Employees Health Benefits (FEHB) program. Because federal employees can choose from among a variety of insurance plans in FEHB, they have a greater opportunity than most private-sector employees to select insurance that offers the most generous coverage for their particular needs. That would reflect itself in variations in the typical claims experience under each participating plan, variations not reflected in the method used here except through the distinction between HMOs and non-HMOs. Thus, the federal benefit may be more valuable to federal employees than indicated in the results presented.

Comparing the Government's Retiree Health Insurance Program with Private-Sector Practices

The dollar values estimated are the amounts needed to fund the expected future medical benefits of retirees over each employee's career. Estimated future medical costs for private-sector firms are based on the experience of selected Watson Wyatt & Company clients. Plan provisions were applied against those expected costs to determine the portion covered by insurance, taking into consideration eligibility requirements, caps on coverage, and other factors. The calculations used to determine the amounts needed to fund those benefits incorporate the same methods and assumptions used to compute unit credit amounts under defined benefit retirement plans.

The dollar values for the federal government are based on benefits provided under the Government Employees Hospital Association insurance plan. The amounts for the private sector reflect the benefits offered by 12 firms selected to represent the different levels of benefits offered by firms in the database. All of the firms selected offer benefits for retirees both before and after age 65. Given the small number of firms and the judgmental nature of the sample, the results may not be representative of the database as a whole.

Two key assumptions in estimating values relate to expected inflation in the medical sector and the method insurance plans use to coordinate benefits with Medicare after age 65. For inflation, the calculations assume an initial rate of 2.6 percent, increasing to an ultimate annual rate of 5.7 percent by 2003. For coordinating benefits with Medicare, plans follow four basic methods, described below in general terms.

• Under carve-out, the most common method used in the private sector, plans determine what they would have paid in the absence of Medicare and then deduct any amount paid by Medicare. The retiree pays expenses that are not covered by the employers' insurance plan.
  
  
• Coordination of benefits, the method used by the federal government, offers the most generous benefit. Under that approach, plans pay amounts not covered by Medicare but no more than the amount they would have paid in the absence of Medicare. Many retirees enjoy a level of benefit superior to that received while employed.
  
  
• Exclusion is a method that often provides a level of benefit that falls between the two just described. Under this approach, the plans apply copayments and deductibles to any amount not covered by Medicare.
  
  
• Under the wrap-around method, plans cover any deductibles and copayments required by Medicare. They may also offer employees coverage for services that Medicare does not cover--prescription drugs, for example.

Each of the major forms of coordination is represented among the 12 private firms used in the comparisons. Results are weighted to reflect the prevalence of the different forms of coordination in the database. For 1996, 51 percent of firms offering retiree health benefits used carve-out, 14 percent used coordination of benefits, 9 percent used exclusion, and 26 percent offered wrap-around benefits, or supplements to Medicare. Results are also weighted to reflect the prevalence of retiree health programs in the database.

Comparing Federal and Private-Sector Sick Leave and Disability Programs

Benefits for each hypothetical employee are the present value of payments employees receive from employers each year as part of the basic sick and disability benefit programs. For each hypothetical employee, those payments take into consideration the benefits available under employer plans for absences of different durations and the probability that those absences will occur. The probabilities and durations of absences are based on data from the Society of Actuaries.

Private-sector values are averages for the database. Long-term disability provisions differ under CSRS and FERS, and separate values are computed for each. Benefits under both are funded in part by employee retirement contributions. For purposes of this comparison, however, estimates assume that federal disability benefits are fully funded by employers. All employee contributions to federal pension plans are assumed to fund retirement benefits. Thus, the disability benefit reported here is somewhat overstated. That irregularity, however, is offset by a corresponding understatement of employer funding for retirement, so that total values for all benefits combined are not affected.

1. The projected unit credit normal cost for an employee would, all else being equal, increase with length of service because of the greater probability of reaching retirement age, among other reasons. Under an alternative measure used by the government and others, the normal cost is instead calculated as a level amount.