WASHINGTON, DC-Differences in men's and women's earnings patterns and work histories would have a large impact on retirement accumulations under a Social Security program that required all workers to have individual accounts, according to new research by the nonpartisan Employee Benefit Research Institute (EBRI). Such differences are avoided when the accounts are made voluntary, such as the voluntary accounts proposed by both GOP presidential candidate Gov. George W. Bush within Social Security and Democratic presidential candidate Vice President Gore outside the Social Security program.

Previous analyses of the mandatory individual account approach have assumed continuous work histories for both male and female workers and a specific earnings pattern for workers. But a new analysis published in the October issue of EBRI Notes uses more realistic earnings patterns of workers, taking into account that earnings change over time and that men and women have different work histories. For instance, the EBRI report looks at how women's retirement accumulations would be affected by temporary absences from the work force.

If realistic earnings histories are used, the individual account retirement account balances of average-earning women would fall by almost 25 percent and the account balances of average-earning men would increase by approximately 15 percent, compared with the economywide average wage used in Social Security estimates.

The EBRI analysis is based on data from the U.S. Census Bureau's Current Population Survey and the Social Security OASDI trustees' report, and uses the Institute's Social Security simulation model (SSASIM) to compare individual account balance accumulations of workers with different employment backgrounds.

The EBRI analysis does not specifically model either the Bush or Gore proposals-since both are voluntary-nor does it compare an individual account system with the current Social Security program.

Factors must be considered before switching

EBRI President and CEO Dallas Salisbury noted that any individual account that is part of or complements Social Security deserves careful scrutiny. Because it would likely affect different workers in different ways, analysis such as this allows predictions of how many workers, and which worker groups, would choose to participate in a voluntary individual account. "If a voluntary individual account system is going to be pursued or implemented, the impact of different earnings patterns and work histories on the decision to participate, and on resulting retirement balances are critically important factors that cannot be ignored," Salisbury said.

Salisbury noted that in the annual Social Security trustees' report, estimated benefits for workers assume that the workers earn the same percentage of the economywide average wage throughout their entire working careers, without distinguishing between genders. Therefore, the workers' relative earnings, compared with the average wage, are constant under Social Security assumptions. But research data, such as those in the U.S. Census Bureau's Current Population Survey, indicate that workers' average earnings at different ages are not constant and that average earnings by gender are not equal.

Some of the EBRI report's key findings:

* Based on data from the Census Bureau's Current Population Survey, the average earnings for both male and female workers relative to the economywide average wage rise from age 16 through the late 40s for males, and the early 40s for females, before declining. In addition, female average earnings are always below both the economywide average wage and the male average wage, while average earnings for males are above it during the middle-age years and below it at the beginning and the end of their working careers.

* For the 1962 birth cohort, who would have only a portion of their working lives covered by the proposed individual accounts, a worker who earns the economywide average wage throughout his or her career would accumulate $41,073 (in 1999 dollars) at a retirement age of 67, assuming historical equity returns. However, when the more realistic earnings patterns for men and women are used, females would accumulate $35,101 (14.5 percent less than the worker with the economywide average wage) and males would accumulate $62,192 (51.4 percent more).

* For the 1982 birth cohort, who would have a full career under the proposed individual accounts, the economywide average wage earner would accumulate an account balance of $140,557 at retirement. In contrast, assuming more realistic earnings patterns, the account balance for females would be $104,825 (25.4 percent less than the economywide average wage earner), while males would accumulate $163,533 (16 percent more).

* Time out of the work force-especially at a young age-would significantly reduce women's individual account accumulations. The account balances for women who are assumed to work every year starting at age 22 and continuing through age 66 differ from those based on realistic female lifetime earnings patterns. For example, when an average-earning female worker is out of the work force for five years (between ages 26-30, 30-34, or 35-39), she will accumulate approximately $16,00 to $18,000 (or 15.6 percent to 17.9 percent) less, on average (based on historical equity returns), than if she had not taken any time out of the work force. The younger the worker with similar earnings pattern is when she is out of the work force, the smaller the account balance will be at retirement. In addition to not making contributions during this period, this worker would also miss out on the compounding of the account, which is a significant portion of the final account balance.

Concerning methodology, the EBRI report calculates the individual account accumulation at retirement using the SSASIM model. The intermediate assumptions from the 1999 OASDI trustees' report are used, and it is assumed that workers contribute 2 percent of covered earnings to the account starting in the year 2002 and continue to do so during their remaining working lives. Two nominal equity return rates, 10.49 percent and 8.12 percent, are used, and administrative costs are assumed to be at 50 basis points annually. The model is run stochastically, meaning that the economic and demographic variables will vary around the trustees' assumptions, and that equity return assumptions will vary as well, producing the specified assumptions on average over the entire simulation period of the 5,000 different plausible scenarios.

# # #