The paper by the Center for Retirement Research at Boston College looked at the different methods of calculating the number of participants in employer-provided pension plans, pointing out the flaws and differences in the five commonly used methods.
The Department of Labor’s Form 5500 series, which requires plan sponsors to submit detailed information about their plan participants, may actually overestimate the number of participants, because the tally might include non-vested employees and 401(k) eligible employees not actively participating in the plan, and a limit to one form per organization means that some participants in both DC and DB plans might be counted twice.
The 5500 series stands apart from the other four methods, in that the number yielded from Form 5500 gives only a number of participants, while the household surveys separate participants along demographic lines.
The first demographic-geared model discussed in the paper is the Survey of Income and Program Participation (SIPP), which does not gauge continuous pension participation and tends to underestimate defined contribution plan participation. The Current Population Survey (CPS) is conducted more frequently than the SIPP but does not provide information on what kind of pension plan a person is in for all years, and tends to underestimate participation rates for all private workers. The Panel Study on Income Dynamics (PSID) by the Survey Research Center at the Institute for Social Research at the University of Michigan only started asking pension participation questions in 1999. The drawback of the Survey of Consumer Finances (SCF), which is considered to keep high-quality data, is that it does not distinguish between public and private employees.
The paper found that regardless of the measurement participation, rates have stayed between 45%-52% during the decade between 1991 to 2003. However, on a short-term basis, the calculations showed greater variations.
For the full paper and comparisons of the data sets go here .