Comparing Proposals for Social Security Reform

Studies | Social Security

No. 227
Wednesday, September 01, 1999
by Liqun Liu and Andrew J. Rettenmaier

Assumptions Behind the Forecasts of the Reform Proposals

Figure II - Social Security Revenues and Expenses

Using a computer simulation model developed by the Private Enterprise Research Center (PERC) at Texas A&M University and the National Center for Policy Analysis (NCPA),13 this study analyzes four reform proposals: (1) the Gramm-Domenici plan, (2) the Archer-Shaw plan, (3) the bipartisan plan and (4) the alternative developed by the NCPA and PERC.

All of the proposals analyzed here have in common: contributions to personal retirement accounts, annuities funded by the private accounts that partially or fully replace government-paid Social Security benefits and the promise that the current payroll tax rate will not be increased (although the bipartisan proposal increases the maximum income subject to taxation and, as a result, increases tax revenues). The plans differ on other provisions, such as how money is put into the private accounts and how the funds are invested. However, the simulations below use a common set of assumptions to make a comparison of the reform proposals more feasible.

Program Changes. Our analysis considers only the Old-Age and Survivors Insurance (OASI) part of the Social Security system, the program designed specifically to provide old age pensions. The OASI and Disability Insurance programs have separate trust funds and dedicated tax sources, so they are easily separated. The current payroll tax of 15.3 percent (employer and employee combined) funds OASI (10.7 percent), Disability Insurance (1.7 percent) and Medicare (2.9 percent). OASDI expenditures currently equal 12.4 percent of payroll.

"It is reasonable to assume a 5.3 percent real rate of return after administrative costs."

Rate of Return on Personal Retirement Accounts (PRAs). We assume that the private accounts will be invested in a portfolio made up of 60 percent stocks and 40 percent corporate bonds, with an annual real rate of return of 5.5 percent. This rate is quite reasonable given the historic rates of return on portfolios held over long periods. Before corporations pay federal corporate income taxes and state and local taxes, the real rate of return is 8.5 percent.14 Further we assume that of the 5.5 percent after-tax rate of return, 0.2 percent (20 basis points) goes to administrative costs for managing the private accounts, leaving a net annual rate of return of 5.3 percent. We assume that the annuities based on the private accounts will also pay a net annual real rate of return of 5.3 percent.

Balanced Budget Restriction. We assume that all program changes must be paid for. If taxes are not to be increased or benefits reduced (as is the case in three of the four plans analyzed), additional funds must come from one of two sources: (1) the federal budget surplus and (2) taxes collected as a result of the increased economic activity made possible by the private accounts. For the purpose of this analysis, the budget surplus is the unified budget surplus minus Medicare taxes and expenditures.15 Additional revenue will become available to the extent that the investment of the private accounts increases the nation's capital stock. This additional capital will produce taxable income that otherwise would not exist. We assume that the federal government can recapture 85 percent of these new revenues.16 Of the 8.5 percent rate of return earned by new capital, federal and state and local taxes take 3 percent.

Table I - Percent of Social Security Benefits That Can Be Replaced by Private Retirement Accounts

Promises Made to the Baby Boomers. Most reform proposals are focused on securing retirement benefits for young people entering the labor market today. Yet well before today's young workers reach the retirement age we will face the crisis of funding retirement benefits for the baby boom generation. As part of our general balanced budget constraint, we require each plan not only to fund deposits made to PRAs but also to meet all future obligations to each retiree cohort.

Real Accounting. The Social Security Administration treats the nonnegotiable bonds in the OASI Trust Fund as real assets that can be used to pay benefits as they come due. Yet every asset of the Trust Fund is a liability of the Treasury. So the sum of the assets and liabilities equals zero. Similarly, any interest earned on Trust Fund bonds is simply a promise by one government agency to pay another government agency and is not a source of real funds. By contrast, our model relies on real accounting. Every expenditure must be paid for by a real funding source - not by IOUs the government writes to itself.17

"Under the current system, expenses exceed revenues beginning in 2014."

Retirement Age and Benefits. For three of the proposals, we assume that the retirement age, promised Social Security benefits and the tax on Social Security benefits continue to be determined by current law. The exception is the bipartisan proposal, which makes specific changes in these parameters.

Forecasts of Taxes and Expenditures. The economic forecasts of future income - on which taxes are paid and benefits are calculated - were developed by the Private Enterprise Research Center based on historical economic data. The life expectancy assumptions are the intermediate assumptions of the Census Bureau. Figure II shows the current projected OASI revenues and expenses between 1999 and 2075. (All values in the analysis are in constant 1999 dollars.) As noted above, the OASI payroll tax rate is 10.7 percent of earnings up to a taxable maximum of $72,600. In future years the tax rate is scheduled to drop to 10.6 percent, and the taxable maximum will rise at the same rate as average earnings. As Figure II shows, expenses begin to exceed revenues in 2014 and do so in all subsequent years.18

"The net effect of eliminating the earnings penalty will be small."

The Social Security Earnings Penalty. All the proposals except the Gramm-Domenici plan repeal the earnings penalty,19 under which retirees ages 62 through 69 are limited to a modest amount of earned income and penalized if they earn more. Specifically, in 1999 seniors ages 62 through 64 will lose one dollar in benefits for every two dollars earned above $9,600 and those ages 65 through 69 will lose one dollar in benefits for every three dollars earned above $15,500.20 Both Social Security Administration and private sector estimates agree that the net effect of eliminating the earnings test will be small, so we do not explicitly model it in the simulations.21 However, the earnings test does have an effect on seniors' work effort, and eliminating the penalty entirely for the age 65 to 69 group would increase the aggregate labor output of affected seniors by an estimated 5.3 percent.22

Required Contribution Rate. Is it possible to fully fund young workers' Social Security benefits with PRA deposits? As Table I shows:23

  • With a 2.0 percent contribution rate, about half of an average individual's estimated benefits can be paid by a PRA-funded annuity.
  • With a 2.5 percent contribution rate, about 62 percent can be paid from the annuity.
  • Fully funding Social Security benefits requires a 4.2 percent contribution rate.

Most workers are not average, however. Owing to Social Security's progressive benefit structure, benefits equal a higher percent of preretirement income for low-income workers than for those with higher incomes. Our model identifies lifetime income with the level of education. The benefit paid to those with more years of schooling replaces a smaller percent of preretirement earnings. Those with less schooling have a higher replacement rate. As a result, a percentage point of wages deposited in a PRA replaces more of promised Social Security benefits, the higher the education level. For example,

  • At a contribution rate of 2.5 percent, a male worker who enters the labor force after high school in 1999 can expect his PRA to purchase an annuity equal to $8,575 at the time of his retirement in 2048. This annuity will replace only 64 percent of his promised Social Security benefits.
  • By contrast, a male worker with a graduate school degree can expect his PRA annuity to be $18,335 - about 79 percent of his promised Social Security benefit.

"Higher-earning individuals can replace more of their benefits with annuities than can those with lower lifetime earnings."

For women, the story is the same: higher-earning women, those with higher levels of education, can replace more of their expected Social Security benefits with their annuities than can those with lower lifetime earnings. Moreover, the magnitude of these differences remains at higher contribution levels as well. For example:

  • With a 2.5 percent contribution rate, the amount of Social Security benefits PRAs can replace ranges from 52 to 79 percent for men.
  • For women, the range is 33 to 67 percent.

As Table I shows, a contribution rate of 4.2 percent over the work life of an average worker is sufficient to replace Social Security completely. But here again, most people are not average:

  • With a 4.2 percent contribution rate, PRAs can replace from 87 percent to 132 percent of expected Social Security benefits for men.
  • The replacement rate ranges between 55 percent and 112 percent of the benefits for women.

"Social Security redistributes income."

Thus, even with a contribution rate that ultimately produces an average annuity equal to the average benefit, tax revenues will continue to be required to bring some workers' annuities up to the level of benefits they would have received under Social Security. The alternative is to construct a higher contribution rate at lower levels of earnings and a lower rate at higher levels of earnings. A consequence of such action is that once the program is fully phased in, lower-income individuals would be contributing a greater percentage of their earnings to their PRAs than would those with higher incomes.

Again, these problems arise because of Social Security's requirement that lower-income retirees have a higher replacement rate than higher-income retirees. In effect, Social Security redistributes income among members of each cohort. The question is when to do the redistribution - during the years in the labor force or during retirement? These alternatives are explored below.

Read Article as PDF