Background on PBGC Modeling
PBGC has for two decades designed and used stochastic models to project possible scenarios for the pension system and possible exposure for PBGC. Work on the Single-Employer Pension Insurance Modeling System (SE-PIMS) began in 1992; it became operational in 1998. The initial specifications for the multi-employer model (ME-PIMS) were written in 2007, and it became operational in 2009.
PBGC initially developed SE PIMS to project general trends in PBGC's financial condition. Its principal purpose is to generate PBGC's Exposure Report, which provides an actuarial evaluation of PBGC's expected operations and the status of its trust funds and is a component of PBGC's Annual Report. Section 4008 of the Employee Retirement Income Security Act (ERISA) requires PBGC to include in its Annual Report "a detailed statement of the actuarial assumptions and methods used in making such evaluation." and to summarize "the specific simulation parameters, specific initial values, temporal parameters, and policy parameters" embedded in PIMS each year.
Within the actuarial and economic community, SE-PIMS is generally well-regarded. It has been the subject of seminars and scholarly articles and is given deference within the actuarial community1. In October 2010, PBGC completed an agreement under which the Society of Actuaries now has and uses SE-PIMS for its own analyses. The Society has published and presented several such analyses. PBGC provides no financial support to the Society and its use of PIMS is entirely at its discretion.
In 2010, we undertook a similar independent review of the ME-PIMS model. However, the actuarial community has less interest in projections of the multi-employer system, so we contracted with Buck Consultants, an actuarial consulting firm with experience in multi-employer plans. Buck has just delivered its recommendations, and PBGC is in the process of determining which of them to implement, and on what schedule it will do so. The report has positive things to say about the basic structure of PIMS (e.g. "It is certainly worth noting that we found many important areas where we believe the current system works just fine and for which we had no significant suggestions") and is focused on improving what is already there to meet the new reality in the ME world.
Outside the actuarial community, PIMS is not widely understood. Concerns have been raised by industry and Congressional staff about transparency, reliability, and about whether its results were being properly presented. Some of this concern probably arises from the fact that PIMS projects both considerable shortcomings in pension funding and long-term shortfalls in PBGC's ability to pay benefits.
The Pension Protection Act of 2006 included a provision requiring substantially expanded discussions in PBGC's Annual Exposure Reports of the methods and assumptions used in PIMS; these reports have been made. Also, PBGC has held PIMS teaching sessions and provided written materials to Hill staff. There have been and are ongoing efforts to be transparent and provide as much information as possible. However, Congressional staffs have shown little interest in these efforts or their conclusions.
Last year, our OIG found lapses in quality control policies and processes in both the inputs to, and outputs from, PIMS models. OIG found errors in transcribing the outputs into PBGC's exposure reports. The OIG did not review the structure or operation of the PIMS models themselves, only the lack of procedures or care in reviewing the inputs and outputs. Nonetheless, the OIG report was seized upon as evidence of the inadequacy of the PIMS models, rather than of the PBGC and contractor staffs that operated it.
In MAP-21, Congress, without consultation or apparent knowledge of existing third-party reviews, required an annual review of the PIMS models by "a capable agency or organization that is independent of the Corporation." The bill mentioned SSA as a possible reviewer, but there is no legislative history. Consultation with congressional staff revealed differences in views between majority and minority staffs.
Prior external reviews of PIMS
Before PIMS first became operational in 1998, it underwent extensive reviews by groups that included two distinguished academics working under a year-long contract with PBGC: a Nobel Prize winning economist and his colleagues, two leading actuaries in the field of stochastic modeling, and a nationally recognized expert on bankruptcy. PIMS also was reviewed at a two-day conference at the Wharton School by several members of the Wharton faculty and by an array2 of distinguished economists, actuaries, and investment bankers from throughout the country.
PIMS was the subject of an award-winning 50-page paper3 in May 2002 in the Journal of Risk and Insurance that was written by the model's two primary architects, one of whom remains a key member of PRAD's PIMS team. The paper won the Robert C. Witt Award, which is given "for the article published the previous calendar year in The Journal of Risk and Insurance, judged to be the best by an independent committee of experts."4
In 2003, PIMS was reviewed by a major investment bank, which was tasked by PBGC with quantifying the value of the PBGC put. Initially, the bank's analytic team planned to use its own models to perform this analysis. However, after reviewing PIMS and determining that PIMS produced results consistent with their models, the lead consultant decided to use PIMS for the project rather than their own models.
In October 2010, PBGC gave SE-PIMS to the Society of Actuaries under an MOA that I signed. As part of a year-long vetting of PIMS to assure that it met the standards of the actuarial profession. Joe Silvestri, the SOA actuary, performed the review.
The SOA has used the SE-PIMS model in several major reports and in briefings to congressional staff. In an October 6, 2011 press release5, the Society of Actuaries announced the release of a report, "The Rising Tide of Pension Contributions Post-2008: How Much and When?" and shared with legislators and staff the findings and implications of the SOA report in an October 11 Capitol Hill Briefing.
The press release states that "the report used data from the regulatory filings of the private sector defined benefit pension system and the Pension Insurance Modeling System, originally developed for the Pension Benefit Guaranty Corp., to analyze the private, single-employer defined benefit system. By simulating the demographic and economic experience of those plans over more than 10 years, the Society of Actuaries projects the funding requirements of the system for each year."
The SOA has completed a second PIMS-based report on minimum funding relief. They presented their report6 to Hill staff at a briefing on June 14, and released it to the public on June 19.
1 See Appendix A for a detailed description of professional and academic reviews of SE-PIMS and ME-PIMS.
2 See also a GAO report that refers to the PIMS technical review panel: p. 170, footnote 4 of http://www.gao.gov/assets/160/155990.pdf.
3 http://onlinelibrary.wiley.com/doi/10.1111/1539-6975.00012/abstract.
4 http://onlinelibrary.wiley.com/journal/10.1111/(ISSN)1539-6975/homepage/the_robert_c._witt_award.htm.
5 "Private Pension System Stable Despite Economy and Increase in Required Contributions About the Report," press release from Society of Actuaries, October 2011.
6 American Academy of Actuaries Calls New Report on Pension Funding “Valuable Contribution” of Expertise.