Defined Benefit Pensions: Conflicts of Interest Involving High	 
Risk or Terminated Plans Pose Enforcement Challenges (28-JUN-07, 
GAO-07-703).							 
                                                                 
To protect workers' retirement security, the requesters asked GAO
to assess: 1) What is known about conflicts of interest affecting
private sector defined benefit (DB) plans? 2) What procedures	 
does the Pension Benefit Guaranty Corporation (PBGC) have to	 
identify and recover losses attributable to conflicts? 3) What	 
procedures does Employee Benefits Security Administration (EBSA) 
have to detect conflicts among service providers and fiduciaries 
for PBGC-trusteed plans? 4) To what extent do EBSA, PBGC, and the
Securities and Exchange Commission (SEC) coordinate their	 
activities to investigate conflicts? GAO interviewed experts,	 
including agency officials, attorneys, financial industry	 
representatives, and academics, and GAO reviewed PBGC		 
documentation and EBSA enforcement materials. GAO analyzed Labor,
SEC, PBGC, and private sector data, including data on pensions,  
pension consultants, and rates of return data, and conducted	 
statistical and econometric analyses.				 
-------------------------Indexing Terms------------------------- 
REPORTNUM:   GAO-07-703 					        
    ACCNO:   A71627						        
  TITLE:     Defined Benefit Pensions: Conflicts of Interest Involving
High Risk or Terminated Plans Pose Enforcement Challenges	 
     DATE:   06/28/2007 
  SUBJECT:   Conflict of interests				 
	     Consultants					 
	     Employee retirement plans				 
	     Interagency relations				 
	     Losses						 
	     Pensions						 
	     Policy evaluation					 
	     Private sector					 
	     Retirement 					 
	     Retirement benefits				 
	     Retirement income					 
	     Strategic planning 				 
	     Fiduciaries					 

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Product.                                                 **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-07-703

   

     * [1]Results in Brief
     * [2]Background
     * [3]Some Pension Plan Service Providers May Have Conflicts of In

          * [4]Pension Plan Fiduciaries and Money Managers Can Have Conflic
          * [5]Consultants Identified by SEC as Having Significant Ongoing
          * [6]Our Analysis Shows an Association between Inadequate Disclos

     * [7]PBGC's Current Policy and Procedures Are Not Focused on Dete

          * [8]PBGC Policies and Procedures Are Focused on Recovering Plan
          * [9]Although PBGC Has Authority to Recover Losses, Cost Benefit
          * [10]While Most of Its Pension Holders Would Not Likely Benefit,

     * [11]EBSA's Enforcement Strategy Does Not Include Procedures That

          * [12]EBSA Has No Specific Enforcement Strategy for PBGC-Trusteed
          * [13]EBSA Has a New Initiative to Focus on Conflicts of Interest
          * [14]EBSA Officials Cite ERISA as Constraint in Pursuing Conflict

     * [15]Different Authorities and Roles Have Limited Agency Collabor

          * [16]Agency Collaboration on Conflicts of Interest Largely Inform
          * [17]Different Agency Responsibilities Tend to Reinforce Limited

     * [18]Conclusions
     * [19]Matters for Congressional Consideration
     * [20]Recommendations for Executive Action
     * [21]Agency Comments and Our Evaluation

          * [22]GAO Panel Data Sample Constructed from Two Primary Data Sour
          * [23]Standard Econometric Modeling Procedures for Handling Panel
          * [24]Fixed-Effect Model
          * [25]Variables Included in the Model
          * [26]Results: Ordinary Least Squares (OLS) and Random-Effects Mod
          * [27]Results: Fixed-Effects Models
          * [28]Limitations of Our Econometric Model

     * [29]Contact
     * [30]Staff Acknowledgments
     * [31]GAO's Mission
     * [32]Obtaining Copies of GAO Reports and Testimony

          * [33]Order by Mail or Phone

     * [34]To Report Fraud, Waste, and Abuse in Federal Programs
     * [35]Congressional Relations
     * [36]Public Affairs

Report to Congressional Requesters

United States Government Accountability Office

GAO

June 2007

DEFINED BENEFIT PENSIONS

Conflicts of Interest Involving High Risk or Terminated Plans Pose
Enforcement Challenges

GAO-07-703

Contents

Letter 1

Results in Brief 3
Background 6
Some Pension Plan Service Providers May Have Conflicts of Interest, but
Determining Whether Harm Results Is Difficult 10
PBGC's Current Policy and Procedures Are Not Focused on Detecting
Conflicts of Interest among Service Providers 19
EBSA's Enforcement Strategy Does Not Include Procedures That Focus on
Conflicts of Interest Involving PBGC Trusteed Plans or High Risk Plans
Likely to Terminate 24
Different Authorities and Roles Have Limited Agency Collaboration 29
Conclusions 33
Matters for Congressional Consideration 35
Recommendations for Executive Action 35
Agency Comments and Our Evaluation 36
Appendix I Scope and Methodology 39
Appendix II Econometric Analysis of the Effect of Inadequately or
Undisclosed Conflicts of Interest on Pension Plan Rates of Return 43
Appendix III Comments from the Pension Benefit Guaranty Corporation 56
Appendix IV Comments from the Department of Labor 57
Appendix V Comments from the Securities and Exchange Commission 60
Appendix VI GAO Contact and Staff Acknowledgments 61

Tables

Table 1: Pension Plan Sponsors Employing 13 Consultants of Concern to the
SEC Regarding Inadequately Disclosed Conflicts of Interest. 41
Table 2: Selected Descriptive Statistics for Plans included in the
Econometric Model 45
Table 3: Econometric Estimates of the Relationship between Undisclosed
Conflicts of Interest and Plan Returns (OLS and Random Effects) 52
Table 4: Econometric Estimates of the Relationship between Undisclosed
Conflicts of Interest and Plan Returns (Fixed-Effects) 54

Figures

Figure 1: Average Annual Rates of Return Achieved by Plans Terminated in
2005 Sponsored by Publicly Traded Companies Compared to CalPERS, TSP and
the S&P 500 Benchmarks, 1997-2002 18
Figure 2: Identifying Plans for GAO's Pension Plan Sample 44

Abbreviations

CalPERS California Public Employees' Retirement System
CAP Consultant Advisor Program
DB defined benefit
DC defined contribution
DOL Department of Labor
EBSA Employment Benefits Security Administration
ERISA Employee Retirement Income Security Act
GLS generalized least squares
MOU memorandum of understanding
OCIE Office of Compliance Inspection and Examinations
OLS ordinary least squares
PBGC Pension Benefit Guaranty Corporation
SEC Security and Exchange Commission
S&P Standard and Poor's

This is a work of the U.S. government and is not subject to copyright
protection in the United States. It may be reproduced and distributed in
its entirety without further permission from GAO. However, because this
work may contain copyrighted images or other material, permission from the
copyright holder may be necessary if you wish to reproduce this material
separately.

United States Government Accountability Office
Washington, DC 20548

June 28, 2007

The Honorable George Miller
Chairman
Committee on Education and Labor
House of Representatives

The Honorable Edward J. Markey
House of Representatives

The bankruptcies of United Airlines, Bethlehem Steel, and other firms
since 2000 have resulted in the termination of a number of large
underfunded pension plans and their becoming the responsibility of the
Pension Benefit Guaranty Corporation (PBGC), the federal guarantor of
private sector defined benefit (DB) plans. Since then, the number of
pensioners who depend on the agency for their retirement benefits has
almost tripled, and the agency's single employer insurance program has
moved from a surplus of $9.7 billion to an accumulated financial deficit
in 2006 of more than $18 billion. Recent experiences involving the
termination of large DB plans have illustrated the weaknesses in funding
rules.^1 Adding to concern over the health of DB plans have been recent
reports about conflicts of interest among pension consultants - advisers
who often play a major role in guiding plan investments. In June 2005, the
Aircraft Mechanics Fraternal Association, which represents certain
employee groups in terminated United and Northwest Airlines DB plans,
expressed concern that conflicts of interest may have been present in the
DB plans of employees the union represents. The union's concerns were
raised as a result of a May 2005 study by Securities and Exchange
Commission (SEC) staff on conflicts of interest among pension consultants.
The SEC study revealed that many pension consultants have failed to
adequately disclose conflicts of interest in the process of advising
pension plans and their trustees, including DB plans.^2 A conflict of
interest is typically a situation in which someone in a position of trust,
such as a pension plan trustee or investment adviser, has competing
professional or personal interests. Such competing interests can make it
difficult for fiduciaries and others, in general, to fulfill their duties
impartially and could cause them to breach their duty to act solely in the
interest of investors, plan participants, or beneficiaries. Having a
conflict in and of itself does not constitute a breach of fiduciary duty.
However, given the potential of financial harm to plan sponsors and
participants, concerns have been raised about the extent and nature of
these conflicts of interest.

^1The Pension Protection Act of 2006 revamped funding rules for defined
benefit plans-- generally effective in 2008--and makes changes to the PBGC
insurance program. Also see, GAO, Private Pensions: Recent Experiences of
Large Defined Benefit Plans Illustrate Weaknesses in Funding Rules,
GAO-05-294 (Washington, D.C.: May 2005).

In view of the importance of protecting the retirement security of plan
participants and bolstering the financial position of the PBGC, you asked
us to pursue the following questions:

           o What is known about conflicts of interest on the part of service
           providers and plan fiduciaries to single employer, private sector
           DB plans?
           o What policies and procedures does the PBGC have in place to
           identify and recover losses attributable to conflicts of interest
           in plans it trustees?
           o Does the Department of Labor's (Labor) Employee Benefits
           Security Administration (EBSA) have procedures in place to detect
           conflicts of interest among service providers and fiduciaries for
           plans now trusteed by PBGC?
           o To what extent do EBSA, PBGC, and SEC coordinate their
           activities to identify and investigate conflicts of interest?

To determine what is known about the existence of conflicts of interest in
the context of single employer private sector DB plans, we interviewed a
variety of professionals with expert knowledge of the issue, including
agency officials, forensic auditors, accountants, attorneys, financial
industry representatives, and academics. We also analyzed Form 5500 data,
Nelson's Directory of Plan Sponsor data, Nelson's Directory of Pension
Consultants data, SEC examination data, Pensions and Investments
periodicals, and data received from the PBGC associated with terminated DB
plans. In addition, we analyzed Standard and Poors (S&P) rate of return
and asset allocation data for ongoing and terminated plans. To determine
the policies and procedures PBGC has in place on conflicts of interest, we
interviewed PBGC and EBSA officials and reviewed PBGC documentation
related to this issue. To determine the procedures EBSA has in place to
detect and investigate conflicts of interest at service providers or plan
fiduciaries, we reviewed EBSA's enforcement materials, our previously
issued reports on EBSA's enforcement program, and interviewed EBSA
officials. To determine EBSA's, PBGC's, and SEC's coordination efforts, we
interviewed officials at all three agencies and reviewed previously issued
reports that provided related information on this issue. We conducted our
work between February 2006 and May 2007 in accordance with generally
accepted government auditing standards. See appendix I and appendix II for
more information on our scope and methodology.

^2See U.S. Securities and Exchange Commission, Office of Compliance
Inspections and Examinations, Staff Report Concerning Examination of
Select Pension Consultants (Washington, D.C.: May 16, 2005.) The report's
findings were based on a 2002 to 2003 examination of 24 pension
consultants. See [37]http://www.sec.gov/news/speech/spch120505lr.htm
(accessed 2007).

Results in Brief

Although no complete information is available regarding the prevalence of
conflicts of interest, pension plan consultants assisting significant
numbers of pension plan sponsors may have conflicts of interest, as a
result of their affiliations or business arrangements with other firms
that could affect the advice they provide to these sponsors. A May 2005
SEC staff study of pension consultants registered as investment advisers
found that 13 of the 24 consultants reviewed that had provided services to
sponsors of pension plans, including ongoing DB and PBGC trusteed DB
plans, had failed to disclose significant ongoing conflicts of interest to
their pension fund clients. Our analysis of data found that, in 2006,
these 13 consultants had over $4.5 trillion in U.S. assets under
advisement, which included DB, DC, and other types of assets. We also
analyzed a sample of ongoing DB plans associated with the 13 consultants
that, as of year-end 2004, had assets of $183.5 billion for these plans
and average assets of $155.3 million. Additional analysis found that the
DB plans using these 13 consultants had annual returns generally 1.3
percent lower than those that did not. Because many factors can affect
returns, and data and modeling limitations limit the ability to generalize
and interpret the results, this finding, while suggestive, should not be
considered as proof of causality between consultants and lower rates of
return. Lack of data prevented a similar study of PBGC trusteed plans or
high risk plans likely to terminate. Although SEC staff have reported that
some of the consultants examined in its study have since taken some
corrective action, this finding nevertheless illustrates the importance of
detecting the presence of undisclosed conflicts of interest among ongoing
plans, and likely among terminated plans. However, independent experts,
EBSA and PBGC officials all concur that while analyzing rates of return is
a useful first step, determining whether conflicts resulted in financial
harm to individual plans and the magnitude of that harm is often extremely
difficult without a detailed forensic audit.

As a creditor and a trustee of terminated plans, PBGC's policies and
procedures are oriented toward the likely recovery of assets, rather than
specifically focusing on losses associated with conflicts of interest
involving service providers. When assuming responsibility for a terminated
plan that is underfunded, PBGC takes steps to identify improper activities
of the plan fiduciary, such as theft or improper loans of plan assets, but
does not collect and evaluate service providers' records to identify their
conflicts of interest and any associated losses to the plan. PBGC
officials told us that, given the agency's mission, balancing scarce
resources against the likelihood of recovering losses makes pursuing
conflicts of interest cases particularly risky for PBGC. Although PBGC has
broad legal authority to pursue and recover losses attributable to
conflicts of interest, PBGC officials told us that the agency limits its
pursuit of cases to those in which the recovery will likely exceed the
cost of identifying and gathering evidence and bringing a case through the
courts successfully. While monetary recoveries of missing assets by PBGC
may improve the agency's financial position, they generally have little
effect on participant benefits because most participants of PBGC trusteed
plans already receive their full plan benefits. According to a PBGC 2004
report, more than 90 percent of all beneficiaries of PBGC trusteed plans
received their full plan benefit.

EBSA's enforcement program is concerned with conflicts of interest
affecting private-sector pension plans generally and does not have
specific procedures for plans trusteed by PBGC. EBSA officials told us
they do not focus their enforcement efforts on PBGC-trusteed plans and
generally leave the responsibility of identifying potentially harmful
conflicts of interest to PBGC for those plans under PBGC control. EBSA has
recently expanded its enforcement activities to focus more heavily on
conflicts of interest involving service providers through a new initiative
known as the Consultant/Adviser Project (CAP). However, EBSA officials
said there are no explicit procedures in the CAP that focus on service
providers of plans that PBGC deems as likely to terminate or those plans
now under PBGC's control. EBSA officials also noted that existing law
limits their efforts to pursue conflicts involving those service providers
that are not fiduciaries under ERISA or that did not knowingly participate
in a breach by a fiduciary.

The current level of coordination among EBSA, PBGC, and the SEC regarding
conflicts of interest is limited, largely because of the different
authorities and missions of each agency. Exchanges of information are
informal and have occurred both between staff at these agencies'
headquarters and between their local offices. At the national level, for
example, SEC has shared examination reports that it concluded would be
helpful to EBSA and provided access to the non-public exam files related
to its 2005 pension consultants study. Locally, information is generally
exchanged when two or more local agency offices have good working
relations. Differing agency responsibilities tend to reinforce limited
collaboration among these agencies. For example, SEC is primarily
concerned with regulating investment advisers to ensure compliance with
securities laws, while EBSA is tasked with protecting participant
benefits. PBGC, in contrast, provides an insurance program for plans in
the event a plan is terminated without sufficient assets to cover promised
benefits. However, more regularized coordination could improve agency
efforts regarding conflicts of interest. Because of their different
missions, these agencies have not established systematic procedures for
regular sharing and coordinating on conflicts of interest.

We are making recommendations to Labor and PBGC that are intended to
improve the detection and oversight of conflicts of interest, and
strengthen EBSA's enforcement ability over non-fiduciaries and recovery of
losses to PBGC-trusteed plans and to improve the collaboration among EBSA,
PBGC, and SEC. We are also asking that Congress consider amending ERISA to
give the Department of Labor greater authority to recover losses from
non-fiduciaries. We provided a draft of this report to Labor, PBGC, and
the SEC for their review and comment.

We obtained comments from the acting Assistant Secretary for the Employee
Benefits Security Administration, the Deputy Director of the Pension
Benefit Guaranty Corporation, and the Director of Compliance Inspections
and Examinations for the Securities and Exchange Commission. Each of the
agencies also provided technical comments, which were incorporated into
the report as appropriate. EBSA, PBGC, and SEC generally agreed with the
findings and conclusions of the report. PBGC noted that although it has no
authority to take action against service providers with conflicts of
interest involving ongoing plans, its recent initiative to enhance its
procedures for identifying and pursuing fiduciary breach and other types
of claims is fully consistent with our recommendation. In its comments,
EBSA agreed to consider our recommendation to expand the focus of its CAP
program to PBGC- identified pension plans that may be trusteed or are high
risk as it reviews the results of its initial efforts under CAP and gains
additional experience through project investigations. EBSA also noted a
number of concerns about our statistical analysis and in particular our
econometric analysis that suggests a negative association between
consultants with undisclosed conflicts of interest and rates of return on
assets. In response to these concerns, we now discuss the limitations of
the analysis more prominently and have added more information on our
statistical analysis and data in appendix II. All three agencies
acknowledge the importance of effective cooperation to facilitate their
respective missions. EBSA's, PBGC's and SEC's comments are reproduced in
appendix III, appendix IV and appendix V.

Background

ERISA is the primary federal law governing the sponsorship and operation
of private sector employee pension plans, including DB plans.^3 Title I of
ERISA gives Labor the primary authority to enforce requirements governing
the conduct of fiduciaries of pension and other employee benefit plans.
EBSA is the Labor agency responsible for administering and enforcing Title
I. ERISA has requirements relating to the standard of conduct of plan
fiduciaries^4 and also prohibits certain transactions^5 between
fiduciaries and parties in interest. Under Title IV of ERISA, PBGC was
established to provide insurance to covered private-sector single-employer
and multiemployer DB plans. The PBGC is not an enforcement agency and
receives no funds from general tax revenues. When a bankrupt plan sponsor
terminates an underfunded pension plan, PBGC assumes trusteeship of the
assets and liabilities of the plan, pays participant and beneficiary
benefits, and acts as a creditor of the bankrupt sponsor in the interest
of the plan's participants and beneficiaries. As plan trustee, PBGC may
file suit to recover missing assets of the plan as well as other assets of
the bankrupt sponsor or to recover losses and debts owed to a plan,
including those resulting from the improper actions of anyone whether or
not they are considered fiduciaries under ERISA.^6

329 U.S.C. SS 1001-1461.

^429 U.S.C. S 1104.

^529 U.S.C. S 1106. Although ERISA generally does not prohibit conflicts
of interest, it establishes a number of prohibited transactions, such as
sales or loans between fiduciaries and parties in interest, as well as any
transaction between a plan and a fiduciary that's in the fiduciary's
interest. ERISA also provides, however, a number of detailed exemptions to
these prohibitions and permits Labor to establish additional ones. 29
U.S.C. S 1108.

Among other things, ERISA provides that private sector employee pension
plans, including DB plans, must have one or more named fiduciaries who
have authority to control or manage the operation and administration of
the plan.^7 ERISA requires fiduciaries to discharge duties solely in the
interest of the participants and beneficiaries with care, skill, prudence,
and diligence. The law states that a person acts as a fiduciary when they
1) exercise any discretionary control or authority over plan management or
any authority or control over plan assets; 2) render investment advice for
a fee or other compensation, direct or indirect, with respect to any
moneys or property of a plan or has any authority or responsibility to do
so;^8 or 3) have any discretionary authority or responsibility in the
administration of a plan.

Pension plans and their fiduciaries often rely on consultants and other
service providers to assist them in plan administration and asset
management, which include selecting money managers and monitoring money
managers' performance and brokerage transactions. Not all of these
consultants and service providers are at all times fiduciaries under
ERISA. ERISA takes a functional approach to fiduciary status. Fiduciaries
that breach their plan duties are personally liable for making up losses
to the plan, restoring any profits made through the use of plan assets,
and face removal as plan fiduciaries.^9 To the extent that a service
provider was not functioning as a fiduciary under ERISA; however, EBSA can
seek recovery against that provider if it knowingly participated in a
breach by a fiduciary. Any such recovery is limited to plan funds the
service provider received (typically in the form of fees paid to it) and
any proceeds derived from those funds to the extent that they remain in
the service provider's possession.

^6For more information on the financial risks facing PBGC, see GAO,
Pension Benefit Guaranty Corporation: Single-Employer Pension Insurance
Program Faces Significant Long-Term Risks, [38]GAO-04-90 (Washington, D.C:
Oct. 29, 2003).

^729 U.S.C. S 1102.

^829 U.S.C. S 1002(21). EBSA officials explained that a consultant or
other adviser fits within this element of the definition only to the
extent that advice was provided (1) as to the purchase or sale of
securities or other property of plan, (2) on a regular basis, (3) pursuant
to a mutual agreement, arrangement, or understanding, (4) as a primary
basis for investment decisions, and (5) based on the particular needs of
the plan.

^929 U.S.C. S 1109.

SEC regulates certain money managers and pension consultants under the
Investment Advisers Act of 1940 (Advisers Act), which requires those firms
meeting certain criteria to register with the commission as investment
advisers.^10 SEC regulates potential conflicts of interests at registered
investment advisers and requires that they disclose information about
affiliations, business interests, and compensation arrangements to their
advisory clients, primarily by providing Part II of SEC's Form ADV or a
brochure containing the same information to clients at the beginning of a
relationship and by offering to provide it annually thereafter.^11
According to SEC, investment advisers have a fiduciary obligation under
the Advisers Act to provide disinterested advice and disclose any material
conflicts of interest to their clients. When an adviser fails to disclose
information regarding material conflicts of interest, clients are unable
to make informed decisions about entering into or continuing the advisory
relationship. Failure to act in accordance with requirements under the
Advisers Act may constitute a violation. According to an SEC official, if
SEC becomes aware of conflicts of interest that are inadequately disclosed
or pose harm to investors, it can require a firm to remedy the
deficiencies or take formal enforcement action against the firm. SEC also
regulates broker-dealers under the Securities Exchange Act of 1934
(Exchange Act), which governs how they may engage in transactions in
securities for their customers and make recommendations to their
customers.^12

The financial services industry and the DB pension system have changed
significantly since the early 1970s. The globalization of financial
markets, as well as technological and international regulatory changes,
has facilitated the development of new financial instruments and the
complexity of investment opportunities.^13 Meanwhile, despite the long
term decline in the number of plans and active participants, DB pension
plans remain a major holder of financial assets. Consequently, the
financial services industry has responded to the growing need for
assistance with managing, investing, transferring, settling, valuing, and
holding pension assets. In 2005, over 81 percent of large
public/government plans utilized a consultant and 42 percent of private
pension plans did so.^14 According to an SEC official, as of October 31,
2005, there were more than 1,800 SEC-registered investment advisers that
indicated on their SEC registration forms that they provide pension
consulting services.^15 The official stated that these firms vary widely
from small one-person operations to large organizations employing
hundreds. Some firms only provide pension consulting, while others may
have started as pension consultants, but then added additional business
operations such as brokerage and money management. It has been reported by
a financial newspaper that in order to remain competitive, some consulting
firms are assuming the fiduciary responsibility of making investment
decisions and have been expanding the range of services they offer.^16

1015 U.S.C. SS 80b-1 et seq. Rules under the Advisers Act permit pension
consultants to plans having an aggregate value of at least $50,000,000 to
register with the Commission (Rule 203A-2(b)).

^11Investment advisers use Form ADV to register with the SEC and state
securities authorities or to amend those registrations and, among other
things, to disclose their conflicts of interest to advisory clients.

^1215 U.S.C. SS 78a et seq.

^13GAO, Financial Regulation: Industry Changes Prompt Need to Reconsider
U.S. Regulatory Structure, [39]GAO-05-61 (Washington, D.C.: Oct. 6, 2004).

This trend toward diversification leads to the potential for conflicts of
interest that could harm pension plans because of competing professional
interests. While conflict of interest is a broad term that can encompass
many specific arrangements, according to an SEC official, conflicts can
occur in the case of money managers, broker-dealers, or pension
consultants when business relationships, particularly those that involve
business among each other, may make them vulnerable to breaching their
fiduciary obligation or duty. Such competing interests can make it
difficult for fiduciaries, in general, to fulfill their duties impartially
and could cause them to breach their duty to act solely in the interest of
investors, plan participants, and beneficiaries. According to Labor
officials, having a conflict in and of itself does not constitute a breach
of fiduciary duty. However, under securities law, acting on and benefiting
from the existence of a conflict without making full and fair disclosure
of all related issues to clients potentially affected by the conflict may
very well constitute a breach. As ERISA fiduciaries, plan trustees also
may face significant conflicts of interest as they may have allegiance
both to the plan and its beneficiaries, as well as to the plan sponsor
that appoints them, and they may receive economic benefits directly or
indirectly from plan service providers.

^14For consultant usage information, see Thomson Nelson, Annual Report of
Pension Fund Consultants 2006 (New York, N.Y.: 2006).

^15For information on SEC's pension consultant examination, see SEC,
Speech by SEC Staff: Conflicts of Interest in Pension Consulting,
(Washington, D.C.: Dec. 5, 2005),
[40]http://www.sec.gov/news/speech/spch1205051r.htm (2007).

^16See, Mark Bruno, "Consultants: Fighting for Revenue", Pensions and
Investments Online. (October 30, 2006).

Some Pension Plan Service Providers May Have Conflicts of Interest, but
Determining Whether Harm Results Is Difficult

Although no complete information is available regarding the prevalence of
conflicts of interests, pension plan consultants assisting significant
numbers of pension plans sponsors may have conflicts of interest as a
result of their affiliations or business arrangements with other firms
that could affect the advice they provide to these plan sponsors. A May
2005 SEC staff study of pension consultants registered as investment
advisers found that more than half (13 out of 24) of the 24 consultants
examined had failed to disclose significant conflicts of interest to their
pension fund clients, including ongoing and PBGC trusteed DB plans.^17 We
determined that, in 2006, these 13 consultants had over $4.5 trillion in
U.S. assets under advisement, including private DB and defined
contribution (DC) plan assets, as well as public pension plan and other
types of assets. We also analyzed a sample of ongoing DB plans associated
with the 13 consultants that, as of year-end 2004, had assets of $183.5
billion for these plans, while average assets were $155.3 million.
Additional analysis found that the DB plans using these 13 consultants had
annual returns that were generally 1.3 percent lower than those that did
not. Because many factors can affect returns, and data and modeling
limitations limit the ability to generalize and interpret the results,
this finding, while suggestive, should not be considered as proof of
causality between consultants and lower rates of return. Although SEC
staff have reported that some of the consultants examined in their study
have since taken corrective action, our analysis illustrates the
importance of detecting the presence of undisclosed conflicts of interest
among ongoing plans, and likely among terminated plans. However,
independent experts, EBSA and PBGC officials all concur that while
analyzing rates of return is a useful first step, determining whether
conflicts resulted in financial harm to individual plans and the magnitude
of that harm is often extremely difficult to detect without a detailed
forensic audit.

^17Most of the consultants examined in the SEC study had disclosures that
SEC staff found inadequate; however, 13 of them were of particular concern
for SEC because the nature of the conflicts of interest information is not
disclosed.

Pension Plan Fiduciaries and Money Managers Can Have Conflicts of Interest,
Although Little Information Is Available on Their Prevalence

According to experts we interviewed, fiduciaries of pension plans often
have an inherent conflict of interest because they are frequently
employees of the plan sponsor. As fiduciaries, they are charged by law to
act solely in the interest of plan participants and beneficiaries, but
they may also have loyalty to the plan sponsor. For example, in 2004,
United Airlines, a plan sponsor, appointed itself fiduciary of its
employee pension plans after all three members of its plan trustee board
resigned during bankruptcy negotiations. A conflict of interest existed
because the newly appointed fiduciaries would have reason to make
decisions that would benefit the plan sponsor instead of the plan
participants. In this instance, the fiduciaries of the United Airlines'
plans faced the obligation to ensure that minimum funding standards
explicitly set in ERISA were satisfied by the plan sponsor. United
Airlines subsequently decided to stop making contributions to the pension
plans it was attempting to terminate. Labor stated that United Airlines'
decision to stop funding its pension plans made clear the need to appoint
an independent fiduciary to represent the interest of workers and retirees
and resolve this conflict of interest. Subsequently, Labor and United
Airlines agreed that United Airlines would appoint an independent
fiduciary.

Plan fiduciaries may also be more prone to conflicts of interest such as
prohibited transactions involving improper loans or more serious actions
such as taking money from the pension plan for personal or business use
when the plan sponsor is financially unstable and may be heading toward
bankruptcy. Experts told us that plan fiduciary conflicts of interest and
other acts such as these are less likely to occur in larger plans since
they often have many professionals to assist with a plan's administration
and management of plan assets. Despite this potential, there is little
information on the extent to which conflicts of interest occur among plan
fiduciaries of DB plans.

Though no formal study has reported statistics quantifying the prevalence
of conflicts of interest among money managers, SEC through its examination
and enforcement efforts has also identified potential conflicts of
interest at money managers that could result in harm to clients, including
pension plans. Money managers may in some cases have incentives to
allocate investment opportunities in a way that could be unfair to certain
advisory clients. For example, an adviser might make more money in fees
based on how it allocates an investment opportunity--such as an initial
public offering--among its clients and steer that opportunity to the
advisors' more lucrative clients. In deciding whether to allocate the
opportunity among its clients, the adviser may have an incentive to
unfairly allocate the investment to a client that pays higher fees. For
example, as between a hedge fund and a pension fund, the adviser could
make more money in fees paid by the hedge fund (for which fees are
generally calculated as a percentage of the fund's overall performance,
which could increase significantly from the investment) than in fees paid
by the pension plan (for which fees are generally calculated as a
percentage of plan assets). Another form of this conflict of interest,
referred to as "cherry picking," occurs when an adviser places a trade
without immediately identifying the client the trade is associated with,
and then allocates the investment after learning of its value. If the
purchase appears valuable based on market conditions, the adviser might
place it in its own portfolio or a more profitable portfolio, but if it
appears less valuable, the advisor might instead place it in one of its
client's portfolios.

Money managers, including those at pension plans, may also face conflicts
of interest because of due to soft dollar payment arrangements. Under soft
dollar arrangements, money managers use part of the brokerage commissions
their clients pay to broker-dealers for executing trades to obtain
research and other services.^18 These arrangements can create a number of
problems. They can create incentives for investment advisers to trade
excessively to obtain more soft dollar services, thereby increasing costs
to pension plan clients or other clients. They can also influence advisers
to place trades with a broker-dealer that provides the adviser with
soft-dollar services rather than another broker-dealer that might provide
best execution.

^18Some forms of soft dollar payment arrangements are considered legal
under a "safe harbor" provision of the Exchange Act in section 28(e) of
the Securities Exchange Act of 1934 that allows advisers to pay more than
the lowest available commission rate for security transactions in return
for research and brokerage services and not be in breach of their
fiduciary duty. 11 U.S.C S 78bb(e). In order to be protected against a
claim of breach of fiduciary duty under this safe harbor, the adviser must
make a good faith determination that the amount of commission paid is
reasonable in relation to the value of the brokerage and research services
provided by the broker-dealer.

Consultants Identified by SEC as Having Significant Ongoing Conflicts of
Interest Provide Services to Many Pension Plans

No complete information exists about the presence of conflicts of interest
at pension plan service providers. However, a 2005 SEC examination of the
activities of 24 pension consultants from 2002 through 2003 revealed that
13 out of 24 of the firms examined failed to disclose significant ongoing
conflicts of interest.^19 These ongoing conflicts took a number of
different forms. For example, SEC found that 13 pension consultants or
their affiliates were found to have conflicts of interest because they
provided products and services to both pension plan advisory clients and
money managers and mutual funds on an ongoing basis without adequately
disclosing these conflicts. Specifically, the study found that 10 pension
consultants sold money managers analytical software packages, which they
use to help analyze and improve the performance of clients' holdings. This
creates a conflict of interest for the pension consultant that might be
more inclined to recommend to pension plans the money managers that buy
software because those business relationships are profitable for the
consultant. Similarly, 13 pension consultants hosted conferences attended
by pension plan advisory clients, who were typically invited to attend
without charge, and money managers, who were often invited to attend for a
fee. A consultant hosting such a conference has a conflict of interest
because it might be more inclined to recommend to pension plans the money
managers that pay fees to attend conferences as such fees are used to
offset costs incurred in hosting the conference.

SEC staff also found that the majority of pension consultants examined had
business relationships with broker-dealers that raised a number of
concerns about potential harm to pension plans. For example, in certain
directed brokerage arrangements, a pension consultant may convince a
pension plan client to direct their money manager to place plan trades
through a broker-dealer that was affiliated with the consultant as a means
for paying advisor fees a plan owed to its consultant using a portion of
the brokerage commission paid on such trades. These arrangements raised
concerns that plans might not have received the best price for each
trade--or "best execution"--because the directions given to a plan's money
manager by the plan may have restricted the money manager's ability to
select a broker-dealer that was the best able to execute a trade. These
arrangements raised the additional concern that consultants might be
overpaid because the plan did not always know when the fee had been paid
in full because brokerage commissions were being used to pay the fee
rather than checks drawn on the plan's checking account.

^19According to the SEC report, the pension consultants examined
represented a cross section of the pension consultant community and varied
in size (measured in terms of the number and size of their pension plan
clients) and the type of products and services they offered. About half of
the pension consultants examined was among the largest pension consulting
firms, measured in terms of the assets of the plans they advise. The
remainder of the sample consisted of medium and smaller consultants. Since
the consultants were not selected randomly, this sample cannot be
generalized to the population of pension consultants.

Following up on its examinations of 24 pension consultants, in late 2005,
SEC staff subsequently sought to determine what steps these firms had
taken to address the findings from the earlier examinations. According to
SEC staff, in general, most pension consulting firms it had examined had
taken positive steps to reevaluate, revise, and implement changes to their
policies and procedures. Specifically, pension consultants implemented
policies and procedures to insulate their advisory activities from other
activities, including for example, creating separate reporting lines and
firewalls between employees that perform these separate functions, and
considering employee compensation and incentives. In addition, SEC staff
said that most consultants they examined had updated their policies and
procedures to improve their disclosure of material conflicts of interest
to pension plan clients and potential clients. Many pension consultants
the SEC staff examined also reviewed and improved their policies and
procedures to prevent conflicts of interest with respect to brokerage
commissions, gifts, gratuities, entertainment, contributions, and
donations provided to clients or received by money managers. However, SEC
staff noted that while the pension consultants it had examined had
improved their practices, it was not able to conduct examinations of all
1,800 investment advisers that indicated that they provide pension
consulting services.

Our analysis of industry data regarding the 13 pension consultants that
failed to disclose serious conflicts of interest found that these
consultants provided services to a number of pension plans.^20 In
particular, the 13 consultants:

^20The report did not seek to identify the financial harm to pension plans
caused by these conflicts of interest, nor whether there were any
violations of ERISA. U.S. Securities and Exchange Commission's Office of
Compliance Inspections and Examinations, Staff Report Concerning
Examination of Select Pension Consultants, (Washington D.C.: May 16,
2005).

           o in 2006, had over $4.5 trillion in U.S. assets under advisement,
           including private DB and defined contribution (DC) plan assets, as
           well as public pension plan and other types of assets;^21 
           o provided advisory services to 36 percent (9 out of 25) of the
           largest plan sponsors, in terms of claims, currently trusteed by
           PBGC since 2000;^22 
           o provided advisory services to 14 percent (12 out of 86) of the
           plan sponsors that were trusteed by the PBGC in 2005; and
           o provided advisory services to 24 percent (1009 out of 4203) of
           the sponsors of ongoing DB plans between the years 2000 and 2004.

Our Analysis Shows an Association between Inadequate Disclosure of Conflicts and
Lower Rates of Return, Although Proof of Financial Harm Requires a Detailed
Audit

We conducted an analysis using ongoing DB plans that revealed a
statistical association between inadequate disclosure and lower investment
returns for ongoing plans, suggesting the possible adverse financial
effect of such nondisclosure. Specifically, we conducted an econometric
analysis using ongoing DB plans and SEC study data on pension consultants
that either adequately disclosed their conflicts of interest and those who
did not.^23 We found lower annual rates of return for those ongoing plans
associated with consultants that had failed to disclose significant
conflicts of interest, with lower rates generally ranging from a
statistically significant 1.2 to 1.3 percentage points over the 2000 to
2004 period, depending on the different model specifications tested.^24
Since the average return for the ongoing plans that used consultants who
did not have significant disclosure violations was about 4.5 percent, the
model implies that the average returns for ongoing plans that used
consultants who failed to disclosure significant conflicts was 3.2 to 3.3
percent for the period. We did not find significant differences in returns
for those plans that had associations with both types of consultants. As
of year-end 2004, our sample of ongoing plans represented assets of $183.5
billion for these plans, and average assets were $155.3 million. We
conducted our analysis using ongoing plans rather than terminated plans
because the ongoing plans provided the necessary sample size to conduct
our analysis, compared to a much smaller sample of terminated plans.

^21Pensions and Investments periodical's  list of Top 25 consultants
ranked by U.S. institutional, tax exempt assets, 2006. 9 of the 13
consultants made the list of Top 25 consultants.

^22We constructed this analysis so that we looked at plans sponsors rather
than plans. For example, PBGC's 25 largest trusteed sponsors since fiscal
year 2000 had a total of 67 plans and comprised 70 percent of the total
claims against the agency between 1975 and 2006.

^23Our analysis is based on a data set we constructed by matching SEC
consultant data with financial information compiled from the Form 5500
database on 1111 plans over 5 years. Of those, 983 were associated with
the 13 consultants identified by the SEC as having provided services to DB
plans that had serious disclosure problems, while 39 were associated with
11 consultants that either were in compliance or had minor inadequacies
with disclosure and another 89 that were associated with both types of
consultants. A complete discussion of our econometric approach, including
model specification, variables used, data sources, estimation techniques
and limitations is provided in appendix II.

^24These include an ordinary least squares specification with time-fixed
effects and various random effect and fixed effect model specifications.
"Fixed-effects" helps to control for the potentially large number of
unmeasured forces that can explain the difference in plan returns. See
appendix II.

While, the results suggest a negative association between returns and
plans that are associated exclusively with pension consultants that did
not properly disclose significant conflicts of interest, the results
should not be viewed necessarily as evidence of a causal relationship in
light of modeling and data limitations. Although the analysis controlled
for plan size, funding level, performance of asset markets, differences in
plan fiscal years and other key variables, other unknown, omitted factors
could have influenced the results of our analysis. While this result gives
an indication of the potential harm conflicts of interest may cause in the
aggregate, these results cannot be generalized to the population of
pension consultants since the consultants examined by the SEC were not
selected randomly. In addition, while these findings are consistent with
the views of the experts we interviewed concerning the adverse effect that
complex service provider related conflicts of interest can have on pension
plans, we cannot rule out the possibility that some other differences
between the plans could explained the differences in estimated returns.
See appendix II for a fuller discussion of the limitations and caveats.

Although statistical analysis is useful, a detailed audit would be needed
to uncover a conflict of interest in any one plan. Independent experts and
officials stated that though a typical first step to identify harm related
to a conflict of interest is to examine a plan's investment returns,
determining whether any financial harm is caused to an individual pension
plan by a conflict of interest requires a detailed forensic audit. A rate
of return for any single plan is not necessarily a good litmus test for
deciding whether to pursue an investigation. For example, two trusteed
sponsors of plans that had some history with consultants reported as
having business arrangements that could pose conflicts of interest had
very different rates of return for their plans. The U.S. Airways plans,
which were trusteed by PBGC in 2003 and 2005, had a rate of return that
exceeded the average measured the benchmark returns earned by the Standard
and Poors (S&P) 500, CalPERS (a major public plan) and the Thrift Savings
Plan (the defined contribution plan for federal employees). Yet, at some
point from 2000 through 2005, the U.S. Airways plan used the services of
consultants who had business arrangements that are of the form described
in the SEC study that raise concerns regarding conflicts of interest.
These business arrangements included directed brokerage arrangements and
hosting conferences. (See figure 1.)

On the other hand, at some point during that time period United Airlines
used a pension consultant who had been noted for engaging in business
arrangements such as directed brokerage and commission recapture programs
that are similar in form to the type that SEC concluded in their 2005
study posed a conflict of interest. During our analysis period, United
Airlines showed a rate of return somewhat lower than three of the four
benchmarks.^25 (See figure 1). For both cases, and very likely most cases,
a detailed, forensic audit would be necessary to identify any accrued harm
from a conflict of interest. Even then, the magnitude of the harm could be
difficult to determine. Experts told us that determining harm often
involves a resource-intensive audit of a plan's service provider's records
and the investment performance of the plan's assets. To perform such an
audit effectively, experts told us that they would need, at a minimum, 5
years worth of service provider specific documents, including contracts
with the plan sponsor, fees charged, payments and other financial
transactions between service providers and those involving plan
fiduciaries. In addition, experts told us that it would be important to
review the investigative files and complaint records of agencies like the
SEC to determine if there is a history of problems at plans and service
providers.

^25Many factors affect a plan's rate of return. At the individual plan
level, the poor performance of a plan's investments might not reflect the
harm of conflicts of interest but could reflect weakly performing asset
markets, failure to manage these assets in a cost effective manner,
misallocation of the plan assets, or some combination of the three. Lower
returns could also signify more conservative investment strategies as a
result of differences in the age profile of pensioners. As a result, the
plan's rate of return received on its assets must be measured and isolated
from the many other industry and firm specific factors which may have an
impact on an investment's value.

Figure 1: Average Annual Rates of Return Achieved by Plans Terminated in
2005 Sponsored by Publicly Traded Companies Compared to CalPERS, TSP and
the S&P 500 Benchmarks, 1997-2002

Note: Analysis based on calculated average rate of returns for the
publicly traded firms taken over by the PBGC in 2005. We selected a range
of benchmarks beginning with a very conservative benchmark to an all stock
investment portfolio, the S&P 500. The average return for the TSP is based
on a conservative portfolio whose allocation mimics the conservative
Lifecycle Fund 2010, as of April 6, 2007. The TSP 2020 fund, whose
allocation mimics the less conservative Lifecycle Fund 2020, earned less
than the 2010 fund during the period of analysis. As a result, we chose
the 2010 fund as one of our benchmarks. S&P 500 returns are based on the
actual S&P total return index and therefore do not consider the cost
involved with maintaining a portfolio indexed to the S&P 500. The CalPERS
benchmark was selected to provide a more realistic comparison given its
asset mix of bonds and stock for a pension fund than the S&P 500 could.

PBGC's Current Policy and Procedures Are Not Focused on Detecting Conflicts of
Interest among Service Providers

As a creditor and a trustee of a sponsor's terminated plan, PBGC's
policies and procedures are designed to review a plan's assets and
liabilities and recover any shortfall. Agency officials told us that such
audits include identifying missing money and conflicts of interest
involving improper activities by a fiduciary such as improper loans and
other prohibited transactions or those that rise to the level of fraud and
theft of fund assets.^26 However, there is no explicit focus on potential
losses associated with conflicts of interest by service providers since
these losses are likely to be found in service provider records and not in
the plan's financial records. Agency officials told us that they currently
do not collect the service provider's records to the extent needed to
uncover conflicts. Although PBGC has authority to recover losses from a
broad group of service providers and not merely ERISA fiduciaries, agency
officials said it may not be cost effective to do so. Our own analysis
also indicates that while recoveries could have a positive, but likely
small effect on the agency's financial position, they would have little
effect on benefits for the large majority of participants.

PBGC Policies and Procedures Are Focused on Recovering Plan Assets

As the insurer of private sector DB plans, PBGC has a primary
responsibility to provide timely and uninterrupted payment of guaranteed
pension benefits. Given that plans trusteed by the PBGC have insufficient
assets to pay all accrued benefits, the agency seeks to bridge that gap by
reviewing the plan, in part, to help recover assets. Such recoveries
include the difference between the plan's assets and liabilities and
quarterly contributions that employers have failed to make. PBGC uses plan
financial documents and a variety of procedures and processes to identify
and value plan assets and liabilities. The financial documents used also
assist the PBGC in uncovering fiduciary breaches, including conflicts of
interest such as prohibited transactions, improper loans and acts of theft
and fraud. However, these financial documents do not provide financial
information that would assist the PBGC in uncovering conflicts of interest
associated with service providers. Agency officials told us that they
currently do not collect such documents.^27 Experts told us that the
agency would, in fact, need to collect and analyze 5 to 10 years worth of
contracts between service providers and plan sponsors and documents that
reveal fees charged, payments, and other financial transactions in order
to conduct a forensic audit. The agency would also need to collect an
historical list of investment advisers, pension consultants and broker
dealers, the plan's investment strategy, the money managers' selection
process, and the money managers' investment performance gross and net of
fees.

^26PBGC officials explained that, in a majority of instances, missing
money and improper loans by fiduciaries occur at small employers with
small pension plans where the plan fiduciary may also be the owner of the
company sponsoring the plan and has little or no professional assistance
with plan administration or asset management.

According to PBGC officials, the only circumstances in which their agency
would have examined records to detect potentially harmful conflicts by a
service provider would be in the case of a complaint providing specific
allegations of wrongdoing with a plan's assets. Agency officials told us
that they had never received a complaint regarding conflicts of interest
by a service provider. In 2005, they received a letter involving the
underfunded United Airlines plans that had been trusteed by PBGC. The
letter was filed by union representatives of affected participants out of
concern emerging from the findings of the SEC staff report on conflicts of
interest among pension consultants. PBGC officials told us that one reason
they did not conduct a forensic audit of the United Airlines plans under
their control was that the letter did not direct them to specific
violations that they should audit. Agency officials also told us that they
did not find it necessary to conduct a detailed audit of any of the plans
following the SEC study because: 1) the plan's investment performance did
not appear out of line; and 2) after reviewing the fees charged, assets
managed, and the type of disclosures implicated, PBGC concluded that it
was unlikely that the conflicts could have had a material adverse affect
on United's pension plans.^28

^27PBGC does collect some service provider contract and fee information
when a plan fails, but only to assure that the provider is compensated as
appropriate.

^28SEC officials told us that they did not review the consultants for
ERISA violations. Based on PBGC's evaluation work, they also did not
evaluate United Airline's service providers for possible ERISA violations.

Although PBGC Has Authority to Recover Losses, Cost Benefit Considerations Shape
the Agency's Actions

PBGC has the authority, as trustee, to recover losses from any party,
including service providers that are not fiduciaries under ERISA.^29
Specifically, ERISA authorizes PBGC to recover from any entity that has
caused a loss or liability to the plan utilizing any available federal or
state cause of action. However, agency officials and experts explained
that since PBGC is not an enforcement agency, their responsibility is to
bring cases to recover losses to a plan, not to bring cases for ERISA
violations. For example, a kickback arrangement--where an investor
receives a financial benefit for choosing a particular investment--or
other types of self-dealing constitute conflicts of interest that may
violate ERISA's prohibited transaction rules. Although it is often
difficult to determine whether a kickback causes a loss to a plan, under
the prohibited transaction provision, the existence of a violation does
not depend on whether any harm results from the transaction. Hence,
identifying and bringing these types of cases would not necessarily be
something that PBGC would pursue unless the violation caused a loss to the
plan.^30

PBGC has pursued cases against plan fiduciaries in an effort to seek such
recoveries. However, in many instances, according to PBGC officials,
seeking recoveries from the plan fiduciary of a small plan to recover
missing money or improper loans may prove fruitless since the plan
fiduciary may have few assets to place a claim against. In fact, in some
cases, a plan fiduciary's only asset from which to recover may be an
accrued pension benefit.^31 PBGC officials told us that the majority of
their cases of fiduciary breach involve action by a fiduciary that
adversely affected plan assets. The officials and outside experts told us
that the majority of cases against fiduciaries to recover missing money or
involving improper loans or prohibited transactions occur with small plans
rather than large plans since large plans typically have many
professionals involved in the management and administration of the pension
plan.

^2929 U.S.C. S 1342(d)(1)(B)(iv). Under ERISA, PBGC generally has the
later of 6 years after the cause of action arose or 3 years after
trusteeing a plan to initiate a claim to recover the losses. 29 U.S.C.
section S 1303(c)(6).

^30Some kickbacks involve pay-to-play schemes in which a money manager
pays to gain access to pension plan sponsors, such as a pension consultant
who only recommends to pension plans the money manager who pays the
highest fees, while another form of kickback may involve the siphoning off
of investment returns. In this example, kickbacks are siphoned off the
investment returns of a client by the money manager after a marked-up
trade. The siphoned returns may be paid to a plan trustee or fiduciary for
their business, to pension consultants, or others.

^31In such instances, spousal claims may prevent the attachment of or
lessen the claim against a plan fiduciary's pension benefit.

While identifying and pursuing cases against a small plan's fiduciary is
typically not resource intensive, agency officials and experts have told
us that identifying and pursuing harm related to conflicts of interest by
service providers is a resource intensive effort that does not always
result in the ability to quantify associated harm and make a recovery.
Agency officials told us that, given their mission, measuring the
investment of scarce resources against the likelihood of recovering losses
makes pursuing conflicts of interest cases particularly risky for the
PBGC. Further, officials explained that the agency must pursue cases where
the recovery will likely exceed the cost of investing the agency's
resources for identifying and gathering evidence and bringing a case
through the courts successfully. If the agency does not believe that a
recovery will exceed its costs, it would be imprudent for the agency to
pursue that case. Nevertheless, as part of the agency's efforts to
evaluate their exposure to certain risk factors, PBGC has undertaken two
relevant studies, one of which includes an assessment of the risks that
relate to the potential for unidentified claims against outside parties,
which includes conflicts of interest^32

^32According to PBGC officials, the agency has very recently implemented
new screening procedures to identify plans that would justify the
significant expenditure of resources required to conduct an audit of a
plan or service provider for conflicts of interest and other types of
cases.

While Most of Its Pension Holders Would Not Likely Benefit, Additional Monetary
Recoveries Could Potentially Reduce PBGC's Deficit

PBGC recoveries generally have little impact on participants in PBGC
trusteed plans because, as an insurer, the agency may pay benefits up to a
guaranteed limit that is higher than the benefits promised to most
participants of trusteed plans.^33 According to a PBGC 2004 annual report,
more than 90 percent of the participants and beneficiaries of single
employer plans that were trusteed by the agency received their full
promised plan benefits.^34 Officials explained that many plans offer
benefits that often fall under the guaranteed limits. The small percentage
of participants and beneficiaries who currently could be helped by such
monetary recoveries represents those that have lost promised plan benefits
that were not guaranteed by PBGC and were not funded by the plan's
assets.^35

PBGC is required by law to use a portion of its employer liability
recoveries and remaining plan assets to cover the non-guaranteed benefits
of pension holders after guaranteed benefits are funded and allocated.^36
An expert we interviewed explained that a portion of the recovered money
goes toward assisting the agency in covering the guaranteed benefits it
pays out. Agency officials told us that recoveries on claims for employer
liability are distributed as prescribed by law and typically increases a
participant's benefit payment less than $20 per month. It is not clear
whether recoveries related to conflicts of interest would provide
significant additional benefits for participants since, according to
experts we interviewed, recoveries for conflicts are likely to be small
compared to represent a small fraction of a terminated plan's total
underfunding.

^33PBGC pays vested accrued participant benefits as of the date of the
plan's termination according to the provisions of each pension plan,
subject to certain maximum guaranteed limits. Benefits guaranteed under
section 4022 of ERISA (other than those under subsection (c)) are
considered basic benefits. 29 U.S.C. S 1322. Some participants receive a
small portion of non-basic benefits as outlined under section 4022(c) as
well. For plans terminating in 2007, the PBGC maximum monthly annuity for
an employee who retires at age 65 was $4,125 per month or $49,500
annually. The maximum guaranteed limit is dependent, in part, upon the age
at which a pension holder begins receiving benefits and the type of
benefits to be provided. The maximum benefit amount can be lower if
benefits are received at an earlier age or the pension includes benefits
for a survivor. However, if an individual is disabled, the maximum amount
guaranteed is not reduced for age for those who begin receiving benefits
from PBGC before reaching age 65.

^34See GAO See Pension Benefit Guaranty Corporation, Pension Insurance
Data Book PBGC 2004 (Washington, D.C.: Spring 2005) p.25. Also note that
participants of terminated pension plans lose future accrued benefits. See
GAO, [41]Pension Plans: Benefits Lost When Plans Terminate , T-HRD-92-58,
(Washington, D.C.: September 24, 1992).

^35Generally, PBGC guarantees basic monthly benefits that provide income
when participants retire, but does not guarantee basic benefits that
exceed the ERISA specified maximum allowance per year and other non-basic
benefits, such as special supplemental benefits that exceed the amount
payable at normal retirement age, early retirement payments, and lump-sum
death benefits. 29 U.S.C. S 1322. PBGC also pays only a portion of the
benefit increases, the greatest of 20 percent per year or $20 per month,
in effect less than 5 years before plan termination. For those
participants who own more than 10 percent of the business, stricter limits
apply.

^3629 U.S.C. S 1322(c).

Benefit recoveries may still help to reduce PBGC's accumulated deficit and
support the agency's mission. PBGC's financial position declined
dramatically for single-employer pension plans from fiscal year 2000 to
2005, with a four-fold increase in underfunding claims of $25 billion. As
of September 2006, the accumulated deficit for PBGC's single employer
program was $18.1 billion. Though recoveries from conflicts of interest
are likely to be small compared with the agency's accumulated deficit,
agency officials say that pursuing conflicts of interest would be
beneficial as long as the costs do not outweigh the benefits obtained from
the recovery.^37

EBSA's Enforcement Strategy Does Not Include Procedures That Focus on Conflicts
of Interest Involving PBGC Trusteed Plans or High Risk Plans Likely to Terminate

Though EBSA's enforcement program is concerned with conflicts of interest
affecting all private sector pension plans, the agency does not have a
specific focus on plans that are trusteed by PBGC or ongoing high risk
plans that PBGC identifies as most likely to terminate. Among EBSA's
reasons for not having such a focus is the agency's view that the PBGC is
in the best position to detect conflicts of interest at terminated plans
and to refer cases to EBSA. Meanwhile, EBSA has recently expanded its
ERISA enforcement effort by implementing its new Consultant/Adviser
Project (CAP) to focus more heavily on conflicts of interest at all
pension plans. EBSA officials also emphasized, however, that existing law
presents a limitation to their pursuing conflicts on the part of several
types of service providers.

^37According to PBGC officials, these costs include direct costs of
pursuing a case and opportunity costs of diverting PBGC staff from other
mission critical work. Thus, a small net benefit based on direct costs
could easily be outweighed by opportunity costs.

EBSA Has No Specific Enforcement Strategy for PBGC-Trusteed Plans or Plans that
PBGC Deems Likely to Terminate

While EBSA's enforcement program does include a focus on conflicts of
interest affecting all private sector pension plans, agency officials told
us they have no specific procedures for detecting conflicts that may have
involved plans that have been trusteed by PBGC or may be trusteed in the
future.^38 There are several reasons for this, according to EBSA
officials. First, they emphasized the view that it is primarily the
failure of plan sponsors to adequately fund pension plans causing plan
underfunding problems rather than poor investment advice from
self-interested service providers. Second, officials told us that while
they do have the responsibility to enforce fiduciary violations regardless
of whether a plan has terminated, they do not focus their enforcement
efforts on PBGC trusteed plans and generally leave the responsibility of
identifying potentially harmful conflicts of interest to PBGC for the
plans under their control. EBSA officials also said, that while their
agency has subpoena power, PBGC has the necessary authority and access to
the many documents needed to pursue conflicts in the plans it trustees.
However, PBGC officials noted that while this may be true for terminated
plans, it does not have jurisdiction to collect such documents for plans
that have not yet terminated.

Finally, EBSA officials told us they had not had occasion to investigate
any PBGC-trusteed plans for conflicts of interest insofar as PBGC had not
made any investigative referral to EBSA concerning conflicts of interest.
Further, EBSA officials said they had not received any complaints
regarding service providers' conflicts of interests involving a terminated
plan prior to a letter it, along with PBGC, received in 2005 from
representatives of certain United Airlines employees. Agency officials
told us that they responded to concerns raised in that letter by reviewing
the plan performances and the portfolio distributions of United Airlines'
plans and determined that they were in line with those demonstrated in the
industry. In fact, in some discrete years, the performance for the United
Airlines plans exceeded some industry benchmarks. Finally, EBSA used
United's Form 5500 information to review the fees paid to service
providers and found them to be comparable to other plans and reasonable.
Thus, the agency did not believe that a forensic audit for service
provider conflicts at United Airlines' plans was warranted.

^38A good identifier of high risk or future PBGC-trusteed plans is the
PBGC list of probable claims. Historically, over 95 percent of claims
classified as a probable claim on the PBGC's balance sheet do subsequently
terminate in the future. See GAO, Private Pensions: Questions Concerning
the Pension Benefit Guaranty Corporation's Practices Regarding Single
Employer Probable Claims, [42]GAO-05-991R (Washington, D.C.: Sept. 9,
2005).

EBSA Has a New Initiative to Focus on Conflicts of Interest among Service
Providers

Despite the current lack of information about the extent of any harm that
may have occurred as a result of conflicts of interest, EBSA officials
acknowledged that such conflicts are a growing concern for their agency.
In order to address these concerns, the agency has, therefore, undertaken
a new national enforcement project, known as the Consultant/Adviser
Project (CAP), which largely focuses on issues identified in the SEC's
2005 study of pension consultants.^39 In addition, with this project, EBSA
hopes to identify other service providers that may be using or managing
plan assets for personal benefit. Specifically the agency will look for
improper, undisclosed compensation such as kickbacks, pay-to-play
arrangements, and soft dollar arrangements. Further, to acquire more
information about the fees charged by service providers, EBSA has proposed
several revisions to the Form 5500, which plan sponsors are required to
file annually. Among the many changes, the revised form would require
increased disclosure regarding the types and amounts of payments made to
service providers, including amounts paid via third-party arrangements,
both direct and indirect.

Despite these changes, EBSA officials said the CAP will not have specific
procedures focused on examining service providers of high risk and
underfunded plans once they are trusteed by PBGC. In addition, as we
previously reported, challenges remain for pursuing more complicated
conflicts of interest cases impacting plans in the context of EBSA's
overall enforcement program.^40 For example, EBSA uses participant
complaints and other agency referrals as sources of investigative leads
and to detect violations. EBSA also identifies leads through informal
targeting efforts by investigators primarily using data reported by plan
sponsors on their Form 5500 annual returns. While these sources are
important, such methods are generally reactive and may reveal only those
violations that are sufficiently obvious for a plan participant to detect
or those that are disclosed and not those violations that are more
complex. Moreover, complaints have primarily originated from participants
in defined contribution (DC) plans since certain problems (e.g., failure
to credit participants' accounts with deposits) involving DC plans are
often more apparent to participants. Requiring more information on the
Form 5500 could, according to experts, uncover or discourage many abuses
concerning conflicts of interest. However, the Form does not necessarily
offer the agency timely or accurate information, because of the 285 days
allowed for its completion and the possibility that errors may be present
on the Form 5500 using the current paper-based filing system.^41

^39As its investigations under the CAP are completed, EBSA may expand its
scope to focus on additional pension consultants and advisers. Each of
EBSA's 10 regional offices will be involved in identifying and
investigating service provider activities related to conflicts of interest
and prohibited transactions.

^40GAO has recently identified some steps that EBSA could implement that
could enhance its enforcement effectiveness. GAO, Employee Benefits
Security Administration: Enforcement Improvements Made but Additional
Actions Could Further Enhance Pension Plan Oversight, [43]GAO-07-22
(Washington, D.C.: January 2007).

It is also unclear how much time EBSA investigators, given their other
duties, will be able to devote to the complex conflicts of interest cases
similar to those targeted by the CAP. EBSA officials told us that they are
addressing their resource constraints in the CAP by concentrating on a
relatively small number of carefully targeted cases. They said they are
undecided as to whether the agency will expand CAP beyond the cases
identified as a result of the SEC study once these investigations are
completed. We had previously reported that Labor's revised performance
goals for EBSA enforcement may encourage a focus on cases that are more
obvious and easily corrected, such as those involving delinquent employee
contributions to DC plans, rather than on investigations of complex and
emerging violations where the outcome is less certain and may take longer
to attain. We had suggested changes to EBSA's approach to assessing
performance to better promote industry compliance and address emerging
violations although the agency has yet to make substantial changes to its
performance measures.^42

Further, though fiduciaries are considered the first line of defense in
avoiding conflicts of interest, EBSA does not conduct routine compliance
examinations or routinely evaluate plan fiduciaries that are not part of
an ongoing investigation to determine how well they select and monitor
service providers. Agency officials and experts have stated that having a
formal set of procedures and guidelines in place to guide the selection of
service providers as well as a formal investment strategy to guide how
assets are to be invested helps to mitigate conflicts of interest. EBSA
officials said the agency lacks sufficient resources to conduct such
general oversight and, instead, uses outreach programs to educate
fiduciaries on the importance of avoiding conflicts of interest. Other
expert observers, however, commented that EBSA's education program only
addresses abuses that occur as a result of ignorance or unintentional
negligence, not those conflicts that are intentional. The experts
emphasized that it is difficult to detect harmful conflicts of interest
without some form of regularized or routine examinations. Although EBSA,
in concert with the SEC, has issued a "tip list" of questions to help plan
fiduciaries avoid conflicts among service providers, EBSA has no
compliance procedures to determine whether fiduciaries are generally using
this information.^43

^41GAO, Private Pensions: Government Actions Could Improve the Timeliness
and Content of Form 5500 Pension Information, [44]GAO-05-491 (Washington,
D.C.: June 2005).

^42GAO, Employee Benefits Security Administration: Enforcement
Improvements Made but Additional Actions Could Further Enhance Pension
Plan Oversight, [45]GAO-07-22 (Washington, D.C.: January 2007).

EBSA Officials Cite ERISA as Constraint in Pursuing Conflicts of Interest

EBSA's ability to recover plan losses related to conflicts of interest by
a service provider is largely limited by the extent to which the service
provider was functioning as a fiduciary under ERISA. Additionally, for
EBSA to take action against an individual or entity, there generally must
be a breach of fiduciary duty. Many service providers carefully structure
their contracts with plans in an attempt to avoid meeting the ERISA
definition of a fiduciary, but whether or not they do depends on the facts
and circumstances in each case. EBSA officials said that many service
providers, such as accountants, auditors, and actuaries are seldom
fiduciaries under ERISA even though they provide important consulting
services to DB plans by evaluating plan assets, calculating required
funding levels, and evaluating financial statements.^44 Experts told us
that broker-dealers are a growing concern, for example, because they have
been expanding their services to include both consulting and investment
services--triggering conflicts of interest questions because offering both
services raises concerns regarding the best execution of trades and
introduces incentives that may not promote practices in the best interest
of plans and participants. Nevertheless, to the extent that a
broker-dealer is not a fiduciary under ERISA, EBSA typically has no
authority to take action against them for not acting solely in the
interest of plans and participants.

^43See U.S. Department of Labor, Employee Benefits Security
Administration, Selecting And Monitoring Pension Consultants - Tips For
Plan Fiduciaries, (Washington, DC.: May 2005).

^44Under ERISA, a person acts as a fiduciary when he or she 1) exercises
any discretionary control or authority over plan management or any
authority or control over plan assets; 2) renders investment advice for a
fee or other compensation, direct or indirect, with respect to any moneys
or property of a plan or has any authority or responsibility to do so; or
3) has any discretionary authority or responsibility in the administration
of a plan. 29 U.S.C. S 1002(21). In addition, according to EBSA officials,
a consultant or other adviser who renders advice for a fee or other
compensation fits within the ERISA definition on that basis only to the
extent that advice was provided (1) as to the purchase or sale of
securities or other property of plan, (2) on a regular basis, (3) pursuant
to a mutual agreement, arrangement, or understanding, (4) as a primary
basis for investment decisions, and (5) based on the particular needs of
the plan.

To the extent that a service provider was functioning as a fiduciary under
ERISA, in addition to recovering any funds taken from the plan and profits
derived from them, EBSA can recover losses to the plan to the same extent
that it can recover them from other fiduciaries. To the extent that a
service provider was not functioning as a fiduciary under ERISA, however,
EBSA cannot recover from them at all unless the service provider knowingly
participated with a fiduciary under ERISA in a fiduciary breach. EBSA
officials said that proving such knowing involvement is often quite
difficult. Even in cases where EBSA can prove that a non-fiduciary
knowingly participated in a fiduciary breach, however, EBSA is limited in
its ability to obtain meaningful recoveries. Specifically, EBSA cannot
recover plan losses but usually only amounts the plan paid to the
non-fiduciary and any profits derived from those payments. Furthermore,
courts have required proof that these amounts remain in the possession of
the non-fiduciary plan in order for them to be recovered.

In addition to such monetary recoveries, EBSA can also obtain injunctions
against fiduciaries initiating or continuing, and non-fiduciaries
knowingly participating with fiduciaries in, activities constituting
fiduciary breaches. Officials explained that, in theory, EBSA does not
have to prove that an activity will cause financial harm to a plan before
obtaining an injunction but as a practical matter it is very difficult to
persuade a judge to grant one without being able to show such harm.

Different Authorities and Roles Have Limited Agency Collaboration

Currently, collaboration on the part of EBSA, PBGC, and SEC regarding
service providers with conflicts of interest is largely informal. At the
national level, SEC and EBSA have communicated about staff examinations
related to the pension consultant study. At the local level, information
about conflicts of interest involving pension plans has been exchanged
between agencies where staffs have developed working relationships. For
example, EBSA officials noted that two of its regional offices have been
collaborating with regional SEC staff on some conflict of interest related
cases. However, exchanges of information generally occur when local
employees of different agency field offices have good working relations
and decide that such contact is helpful. Differences in agency missions
and responsibilities tend to reinforce such informal coordination.

Agency Collaboration on Conflicts of Interest Largely Informal

Collaboration on the part of EBSA, PBGC, and SEC that might facilitate
identifying conflicts of interest is largely informal, particularly with
regard to PBGC trusteed plans. With respect to EBSA and PBGC, there is a
memorandum of understanding (MOU) for sharing quarterly information on the
financial status of plans, but it does not provide for collaborating over
potential conflicts of interest. Moreover, EBSA and PBGC officials told us
that the data currently shared would not likely reveal conflicts of
interest. EBSA officials told us that their weekly discussions with PBGC
representatives are related to financial matters of plans that may be
experiencing financial trouble. With regard to the SEC, there is no formal
agreement with the other two agencies to share information relevant to
conflicts of interest. However, EBSA officials stated that there is some
collaboration between EBSA and SEC both nationally and locally, generally
occurring on an informal basis.

At the national level, SEC has shared examination reports that it
concluded would be helpful to EBSA, including the non-public exam files
related to the pension consultants study. At the local level, information
about conflicts of interest involving pension plans has been exchanged
between agencies where employees developed working relationships. For
example, EBSA officials noted that two of its regional offices had been
collaborating with regional SEC staff on some conflict-of-interest related
cases.^45 However, under securities laws, SEC is subject to
confidentiality restrictions with respect to information it can disclose
to EBSA pertaining to an ongoing investigation, even if the information
pertains to possible violations of ERISA. Likewise, EBSA investigators can
alert SEC to information that is discovered during an ERISA investigation
that might be of interest to SEC. However, unlike EBSA, SEC may not share
documentation associated with its findings unless EBSA submits a written
request for information which, if approved, allows access to any evidence
that SEC has obtained during the course of its investigation. ^46

45In addition to the 2005 SEC study, these two regional projects also
served as an impetus for EBSA's new CAP designed to address issues of
whether plan service providers particularly pension consultants, may have
potential conflicts of interest that could affect the objectivity of the
advice they provide to their pension plan clients.

^46SEC personnel are generally prohibited from disclosing information
obtained as a result of an examination or investigation. The Commission
may provide such information, however, when those to receive it show that
it is needed and provide acceptable assurances of confidentiality. 15
U.S.C. S 78x.

Nevertheless, there is no systematic procedure among the three agencies
that would effectively target or monitor service providers engaged in
conflicts of interest.^47 However, more regularized coordination could
improve agency efforts regarding conflicts of interest. For example,
during investigations and examinations, SEC and EBSA tend to collect
documentation that is specific to their individual enforcement objectives.
Experts told us that creating efficiencies through collaborative and
supportive enforcement practices where both agencies collect and share
information that both agencies would find useful would be a major
improvement in collaboration.

Different Agency Responsibilities Tend to Reinforce Limited Collaboration among
EBSA, SEC, and PBGC

To some extent, differences in each agency's roles and responsibilities
affect the level of collaboration regarding conflicts of interest among
the three agencies. First, EBSA is tasked with enforcing the fiduciary
standards required under Title I of ERISA, which seeks to ensure that
fiduciaries operate their plans in the best interest of plan participants.
Second, SEC enforces securities law and is primarily concerned with
regulating professional entities, such as pension consultants or
investment advisers. Finally, PBGC insures benefits for the beneficiaries
of private- sector DB pension plans.

Federal law and regulation across the agencies are not consistent on the
treatment of conflicts of interest. For example, under securities law, a
conflict of interest that is disclosed may not be a violation, and would
not necessarily prompt investigation by the SEC, although it may prompt
investigation by EBSA. However, Title I of ERISA applies only to those who
have carried out or been associated with fiduciary responsibilities, which
does not always include all types of service providers. In addition,
differences in definitions and terminology create challenges for the
agencies to gather useful information for collaborating on investigations.
For example, all money managers and others that actively manage or invest
pension assets have a fiduciary obligation under the Advisers Act. Money
managers are generally considered fiduciaries under ERISA, though
broker-dealers are considered to be fiduciaries under ERISA only under
certain circumstances. Pension consultants typically have a fiduciary
obligation under the Advisers Act^48 but may not be fiduciaries under
ERISA.

^47GAO has identified the need for greater coordination and made
recommendations for greater and more formal coordination between EBSA and
the SEC to improve enforcement efforts. See GAO, Employee Benefits
Security Administration: Enforcement Improvements Made but Additional
Actions Could Further Enhance Pension Plan Oversight, [46]GAO-07-22
(Washington, D.C.: January 2007) and GAO, Pension Plans: Additional
Transparency and Other Actions Needed in Connection with Proxy Voting,
[47]GAO-04-749 (Washington, D.C.: August 2004).

While PBGC has broad authority to recover losses, it is not an enforcement
agency and therefore is not in the business of investigating conflicts of
interest or other fiduciary violations without the intention of recovering
meaningful losses. To the extent that there would be meaningful losses to
recover, as trustee, PBGC has the ability to collect a range of service
provider documents that might suggest a history of conflicts involving the
pension plans it trustees. EBSA, which does have authority to investigate
and bring conflict cases involving plan fiduciaries to court, does not
have ready access to these documents without a subpoena. Although EBSA
does have broad subpoena powers, the use of these subpoena powers and
enforcing subpoenas can involve significant delays in enforcement and case
resolution.

The lack of formal collaboration between the three agencies also reflects
their differing missions. While the SEC and EBSA both have an enforcement
role, their missions have different orientations. SEC enforces securities
law and is primarily concerned with regulating professional entities, such
as pension consultants or investment advisers, to the extent that all
conflicts of interest are adequately disclosed and plan sponsors can make
informed decisions about whom to hire. In most cases, the SEC is able to
act administratively, in that it can levy fines and suspend registered
advisers without having to use federal courts. Further, since advisers
have a statutorily imposed fiduciary responsibility, investigators do not
have the burden of proving a fiduciary status before taking action.

In contrast, EBSA is tasked with enforcing the fiduciary standards
required under Title I of ERISA, which seeks to ensure that fiduciaries
operate their plans in the best interest of plan participants. In most
cases, EBSA must prove that each violation it pursues was caused by a plan
fiduciary or a party carrying out a fiduciary function. EBSA officials
told us that a consultant or other adviser is a fiduciary investment
adviser only to the extent that advice was provided (1) on the purchase or
sale of securities or other property of the plan, (2) on a regular basis,
(3) pursuant to a mutual agreement, arrangement, or understanding, (4) as
a primary basis for investment decisions, and (5) based on the particular
needs of the plan. While the courts have ruled in some cases that EBSA can
pursue non-fiduciaries that contribute to a fiduciary breach, EBSA
officials stated that the remedies they have available to them under ERISA
are limited when pursuing these entities. Moreover, EBSA officials told us
that there can be situations where a pension consultant may not meet the
conditions necessary to be considered a fiduciary under ERISA; in which
case EBSA generally would not be able to take action against the
consultant.

^48Rules under the Advisers Act require pension consultants to plans
having an aggregate value of at least $50 million to register with the
Commission (Rule 203A-2(b)).

The role of PBGC, in turn, is not to regulate pension plan trustees and
service providers, but to insure benefits for the beneficiaries of
terminated pension plans. Therefore, PBGC's primary goal is to preserve
plan assets to the degree possible in order to pay promised benefits and
keep expenses to a minimum. Accordingly, PBGC generally does not undertake
the cost of litigation without a clear opportunity to recover assets. PBGC
officials stated that the recoveries are typically far smaller than their
claims on assets, as the agency generally recovers at most 10 cents on the
dollar. For plans terminating in fiscal 2005, for example, PBGC reported
$10.8 billion in claims but only $170.7 million in recoveries.^49

Conclusions

The challenge to sound pension sponsorship posed by financial conflicts of
interest is largely a consequence of the changes experienced by financial
markets over the last 30 years. In fact, the pre-ERISA world of 1974 never
anticipated the multiplicity and complexity of financial instruments that
have expanded both investment opportunities and risks for plan
fiduciaries. Index and hedge funds, the growth of complicated financial
derivatives, and access to international financial markets represent only
some of the extraordinary number of choices confronting today's pension
plan fiduciaries.

Of necessity, DB pension plan fiduciaries must utilize the variety of
service providers that have become available to help them assess choices.
While conflicts of interest are not necessarily inherent in the provision
of such financial services, the prevalence and the proliferation of
consulting work and the complexity of business arrangements among
investment advisers, plan consultants, and others have increased the
likelihood. Our analysis of ongoing plans suggests that, in the aggregate,
there may be some cause for concern. While many consultants have taken
remedial action, there are pension plan consultants that advise on a
sizeable portion of U.S. pension assets that did not fully disclose
conflicts of interest in the past. Although not generalizeable to all
consultants and plans, our analysis cautiously suggests an association
with such undisclosed conflicts and plan performance. However, assessing
the extent and magnitude of the problem of conflicts of interest at an
individual plan level, at the outset, may require a coordinated effort
among the regulatory agencies because of the complexities involved and the
significant resources associated with investigative audits. Regardless of
the difficulty of finding a financial trail of damage, to the extent that
financially harmful conflicts of interest exist, they pose a potential
threat to the investment confidence of sponsors and participants. For this
reason, alone, credible and visible enforcement is essential to prevent
such erosion.

^49PBGC publishes detailed statistics regarding its program operations and
benefit protections annually. For more information see, Pension Benefit
Guarantee Corporation, Pension Insurance Data Book, PBGC 2005 (Washington,
D.C.: Summer 2006).

Yet our findings reveal that there is limited regulatory framework for
deterrence in this area, particularly for terminated DB plans and those
likely to terminate. EBSA's recent CAP initiative to target conflicts of
interest among service providers may help, but it does not include any
specific emphasis on service providers of plans either under PBGC's
trusteeship or those considered likely to terminate. In addition, as EBSA
officials have noted, ERISA's definition of fiduciary and associated
remedies and penalties to correct potential breaches of fiduciary
responsibility, conceived to address the pension issues of the 1970s, are
less effective in combating conflicts of interest in the far more complex
world of today. Further, the SEC concluded in its study that many pension
consultants do not consider themselves to be fiduciaries to their clients.
In fact, many pension consultants believe they have taken appropriate
actions to insulate themselves from being considered a fiduciary under
ERISA. As a result, it appears that many consultants believe they do not
have any fiduciary relationships with their advisory clients and ignore or
are not aware of their fiduciary obligations under the Advisers Act.

Meanwhile, PBGC's recent decision to conduct an overall risk assessment
and implement new screening procedures acknowledges the need for
improvement in PBGC's reviews for conflicts of interest and other risk
factors. Without procedures to evaluate the effect of conflicts of
interest on high risk and terminated plans, however, potential ERISA
violations related to such conflicts of interest could possibly go
undetected. Moreover, the current levels of collaboration among the three
agencies most involved with DB pension plans -- EBSA, PBGC, and the SEC --
or their service providers, present opportunities that could enhance
enforcement. Because SEC conducts examinations of some registered advisers
it oversees, consistent inquiry by EBSA and PBGC into SEC's inspection
results would be a good first step toward bridging the information gap. In
addition, in the spirit of creating efficiencies with overseeing service
providers doing business with pension plans, EBSA and PBGC may greatly
benefit from SEC's regular exams by giving some thought to what SEC could
collect during its efforts that would be useful to the other two agencies.

Nevertheless, it would be prudent and responsible to carefully weigh the
benefits of any new regulatory approaches against their potential effect
on continued sponsorship of DB plans. Although the percentage of the
private sector labor force covered by a pension plans has remained roughly
constant over the last decade, the number of active DB plan participants
has declined sharply. Nonetheless, given the important role that DB plans
still play in the retirement security of millions of American workers and
their families, it would be prudent to weigh any proposed regulatory
options against the additional administrative costs they may generate on
DB plans.

Matters for Congressional Consideration

Congress may wish to consider amending ERISA to allow EBSA to recover plan
losses against certain types of service providers even if they are not
currently considered fiduciaries under ERISA.

Recommendations for Executive Action

To enhance existing protections of plans and participants, and maintain
participant and sponsor confidence in the private DB pension system, as
part of its current risk assessment efforts, the Director of the PBGC
should:

           o Develop a pilot project to collect the necessary documents on a
           select group of trusteed plans to determine the extent to which
           conflicts of interest may have affected these plans. This pilot
           project should be undertaken with the assistance of EBSA and in
           consultation with the SEC. PBGC and EBSA should provide SEC with
           ideas that would be useful to them on the information SEC could
           gather during its adviser and broker-dealer examinations.

The Secretary of Labor should direct the Assistant Secretary for EBSA:

           o to enhance current enforcement by expanding the scope of the new
           CAP program to include some emphasis on service providers of those
           high risk plans PBGC deems likely to terminate in the future and
           plans PBGC-trusteed.

Building on the existing memorandum of understanding (MOU) between EBSA
and PBGC and a recommendation made in our earlier work, the Assistant
Secretary of EBSA, the Director of the PBGC, and the Chairman of the SEC
should:

           o Enter into an MOU to facilitate information sharing on conflicts
           of interest among service providers that either consult or that
           provide money management services to PBGC-trusteed plans and those
           likely to terminate in the future.

Agency Comments and Our Evaluation

We obtained comments from the acting Assistant Secretary for the Employee
Benefits Security Administration, the Deputy Director of the Pension
Benefit Guaranty Corporation, and the Director of Compliance Inspections
and Examinations for the Securities and Exchange Commission. PBGC's,
EBSA's, and SEC's comments are reproduced in appendix III, appendix IV,
and appendix V. Each of the agencies also provided technical comments,
which were incorporated into the report as appropriate.

EBSA, PBGC and SEC generally agreed with the findings and conclusions of
the report. PBGC noted that although it has no authority to take action
against service providers with conflicts of interest involving ongoing
plans, its recent initiative to enhance its procedures for identifying and
pursuing fiduciary breach and other types of claims is fully consistent
with our recommendations. In comments, EBSA agreed to consider our
recommendation to expand the focus of its CAP program to PBGC- identified
pension plans that may be trusteed or are high risk as it reviews the
results of its initial efforts under CAP and gain additional experience
through project investigations.

All three agencies acknowledge the importance of effective cooperation to
facilitate their respective missions. SEC notes that it looks forward to
further developing its currently cooperative relationship with EBSA and
PBGC through discussion on our recommendations. PBGC pointed to the
existing information sharing arrangement that it has with EBSA and the
Internal Revenue Service that it believes could serve as a useful model to
coordinate with EBSA and SEC, and pledges to work with EBSA and SEC to
more closely coordinate agency action on PBGC-trusteed plans and plans
likely to terminate in the near future. EBSA also notes both the
importance of establishing and maintaining effective working relations
with other agencies to maximize enforcement effectiveness, and has stated
that it is prepared to work with PBGC and SEC to facilitate information
sharing.

EBSA noted a number of concerns about our statistical analysis and in
particular our econometric analysis that suggests a negative association
between consultants with undisclosed conflicts of interest and rates of
return on assets. EBSA expressed important cautions that should be
considered when interpreting our results, including some data limitations
and our use of an estimate for our investment returns variable. We agree
that our econometric study, while suggestive, should not be considered
definitive, or a proof of causality, especially in light of the data and
modeling limitations constraining the analysis. The goal of this analysis
was to shed some light on a critical public policy issue--understanding
the relationship between rates of return and consultants that have been
found to have undisclosed conflicts of interest--given the current state
of econometric techniques and limited real world data. We view our
findings as an indicator of the potential effects that conflicts of
interest can have on returns and as a catalyst for further analysis rather
than evidence of a causal relationship. In response to EBSA's concerns, we
now discuss the limitations of the analysis more prominently and have
added more information on our statistical analysis and our data in
appendix II.

We are sending copies of this report to the Director of the PBGC, the
Secretary of Labor, the Chairman of the SEC, and other interested parties.
We will also make copies available to others on request. In addition, the
report will be available at no charge on the GAO Web site at
[48]http://www.gao.gov .

If you or your staff has any questions concerning this report, please call
me at (202) 512-7215. Key contributors are listed in appendix VI.

Barbara D. Bovbjerg
Director, Education, Workforce, and
Income Security Issues

Appendix I: Scope and Methodology

To conduct our review of the procedures the PBGC, SEC and EBSA have in
place to detect and coordinate on conflicts of interest that may impact
the entities they oversee, we interviewed officials from key agencies as
well as independent outside experts, lawyers and forensic auditors
knowledgeable about conflict of interest issues. We conducted interviews
with officials from SEC's Office of Compliance Inspections and
Examinations (OCIE) and SEC's Division of Enforcement regarding their 2005
staff report concerning examinations of selected pension consultants and
the agency's general enforcement practices. We interviewed various PBGC
officials, including staff attorneys and accountants, on their
departmental procedures policies regarding the pursuit of financial
recoveries regarding underfunded terminated plans, as well as other
related issues. We interviewed various EBSA officials, including those
from EBSA's Office of Enforcement and the Solicitors office. We
interviewed a number of experts on conflict of interest issues, including
lawyers and auditing professionals knowledgeable about conducting audits
related to conflicts of interest, and those with expert knowledge on EBSA
and PBGC's policies and procedures. We also interviewed legal experts on
ERISA and securities laws.

To obtain information about agency procedures, we collected and reviewed
PBGC's operations and policy manual, and reviewed PBGC court cases. We
collected data provided by EBSA's Office of Enforcement pertaining to the
investigation of cases related to conflicts of interest and prohibited
transaction violations and reviewed EBSA's enforcement manual. We reviewed
the relevant section of ERISA and securities laws in consultation with
GAO's legal staff. Finally, we reviewed past GAO work on SEC, PBGC and
EBSA enforcement efforts with respect to conflicts of interest, as well as
agencies' general enforcement efforts and we consulted the teams within
GAO that regularly review SEC, PBGC and EBSA operations.

To conduct our statistical analysis on the 24 pension consultants included
in the SEC study, we obtained details regarding the type of conflicts
found and the disclosure issues involving the 24 pension consultants from
SEC officials. We obtained specific information regarding certain finding
statements made in the SEC's 2005 staff report and reviewed this
information at SEC headquarters under their oversight. To protect the
confidentiality of SEC's exam practice, our data analysis was mostly
conducted at SEC headquarters. Any additional analysis conducted by us at
GAO headquarters which could reveal information that could identify the
consultants reviewed in the SEC's 2005 staff report has been destroyed.

Working with the SEC study data, we compiled a list of clients associated
with the 13 consultants identified by the SEC staff report as having
undisclosed conflicts of interest using the Form 5500 and Thomson Nelson's
database of pension consultants and plan sponsors. ERISA and the IRC
require administrators of pension plans to file annual Form 5500 reports
concerning, among other things, the financial condition and operation of
plans. Form 5500 Reports are shared among Labor's Employee Benefits
Security Administration, IRS, PBGC, and the Social Security
Administration, and each agency uses the Form 5500 to meet its statutory
obligations. Plan year 2004 was the most recent year for which
plan-specific Form 5500 data were available for our review. The Form 5500
data presents a number of data limitations. These limitations have been
well documented in other reports issued by us.^1

We then supplemented our Form 5500 data with information we purchased from
Thomson Nelson (Nelson). We used the Nelson databases and the Thomson
Nelson Annual Report of Pension Fund Consultants 2006 The Nelson database
contains detailed information on various aspects of 20,000 single employer
DB and DC pension plans and on 350 pension consultants and other service
providers who service those plans. With the Form 5500 and Nelson data, we
developed a client list for the 13 pension consultants examined in the SEC
study. We regard our client list to be a partial list since the sources
contained incomplete information and no complete source of information was
available to us. Since creating a complete client list for the 13
consultants identified by the SEC staff report as having undisclosed
conflicts of interest was not possible, we consider our counts of the
clients of these consultants to be conservative.

To determine the relationship between the consultants identified by the
SEC with PBGC trusteed plans and ongoing plans we conducted three matches:

           o 24 SEC identified consultants with PBGC's list of the trusteed
           plans of the 25 largest companies in terms of claims since the
           beginning of fiscal year 2000.^2 We constructed each match so that
           we looked at plan sponsors rather than plans. PBGC's 25 largest
           trusteed companies since the year 2000 had a total of 67 plans.
           o 24 SEC-identified consultants and our client list with PBGC's
           list of plans that underwent PBGC trusteeship in 2005. The total
           number of plans was 118.
           o 24 SEC-identified consultants and our client list with plans
           that were non-terminated and ongoing between 2000 and 2004. The
           total number of plans was 4832.

^1See GAO, Private Pensions: Participants Need Information on the Risks of
Investing in Employer Securities and the Benefits of Diversification,
[49]GAO-02-943 (Washington, D.C.: Sept. 6, 2002); Retirement Income Data:
Improvements Could Better Support Analysis of Future Retirees' Prospects,
[50]GAO-03-337 (Washington, D.C.: Mar. 21, 2003); Private Pensions:
Multiemployer Plans Face Short- and Long-Term Challenges, [51]GAO-04-423
(Washington, D.C.: Mar. 26, 2004); and Private Pensions: Publicly
Available Reports Provide Useful but Limited Information on Plans'
Financial Condition, [52]GAO-04-395 (Washington, D.C.: Mar. 31, 2004).

The compilation and matching of our data sources provide the following
information is illustrated in the table 1:

Table 1: Pension Plan Sponsors Employing 13 Consultants of Concern to the
SEC Regarding Inadequately Disclosed Conflicts of Interest.

                                                                  Sponsors of 
                              25 largest PBGC     Sponsors of ongoing defined 
                           trusteeships since  plans trusteed   benefit plans 
                                         2000 by PBGC in 2005       2000-2004 
Sponsors employing 1 or                  9              12           1,009 
more consultants with                                                      
undisclosed conflicts                                                      
Total number of                         25              86           4,203 
sponsors                                                                   
Percentage                             36%             14%             24% 

Source: GAO Analysis of data from PBGC, SEC, Nelson Information, and Form
5500 filings.

Of the remaining 11 consultants that were of less concern to the SEC:

           o Six were ranked as some of the largest pension consultant firms
           in the U.S. with pension assets under advisement totaling over 1.5
           trillion dollars.^3 
           o One of the 11 had an advisory relationship with one of the 25
           largest PBGC trusteeships at some point during the years they were
           ongoing, although at least one or more of the 13 consultants of
           concern also provided services to this plan sponsor during that
           same period.
           o One of the 11 had an advisory relationship at some point during
           the 5 year period between 2000 through 2004with one of the
           sponsors with plans that terminated in 2005 and was trusteed by
           PBGC, although at least one of the 13 consultants of concern also
           provided services to this plan sponsor during that same period.
           o One or more of the 11 had an advisory relationship with 167 of
           the ongoing defined benefit plan sponsors from 2000-2004, although
           at least one or more of the 13 consultants of concern also
           provided services to 99 of these plan sponsors during that same
           period.

^2These 25 companies' plans with the largest claims against the PBGC
comprised 70 percent of the total claims against the agency, to date.

^3According to Pensions and Investments periodical list of Top 25
consultants ranked by U.S. institutional, tax exempt assets, 2006. 6 of
the 11 consultants made the list of Top 25 consultants.

To match the 13 consultants identified by the SEC with
non-terminated/ongoing DB plans, we analyzed Form 5500 information from
filing years 2000 to 2004 to identify the sponsors of DB plans we
categorized as ongoing. We selected only single employer or multiple
employer DB plans that filed Form 5500 in 2004 and were not on the PBGC
list of plans that terminated in 2005. From these we selected plans whose
filings were not partial year and were not final plan filings.
Additionally, we selected those plans with at least one other full year
filing for the period 2000 to 2003. Additionally, we chose only plans with
information on both beginning and end of year assets reported on Schedule
H. This resulted in a list of 4,203 sponsors of 4,832 plans.

To determine the consultants that worked directly for a plan or indirectly
for a plan through a plans holdings in master trusts accounts and other
such arrangements, we compiled information on service providers reported
on the plans Form 5500 schedule C and also on service providers reported
on the filings of master trust accounts and other such plan holdings. We
compiled a working list of consultants whose service was reported with the
codes 17 (Consulting), 20 (Investment advisory), and 21 (Investment
management). We matched this list to the list of consultants investigated
by the SEC to determine which plans used the services of one or more of
the consultants that were investigated. We augmented this list using
consultant and client list information available from Nelson.

Finally, we conducted an econometric analysis to determine whether there
was a correlation between undisclosed conflicts of interest and rates of
return for the ongoing DB plans identified as employing the services of
one or more of the consultants listed in the SEC study. We included only
the plans that we could link to the 24 consultants either directly on the
basis of plan form 5500 filings or indirectly based on a plans holdings in
master trusts accounts and other such arrangements. For details of this
analysis please see appendix II.

Appendix II: Econometric Analysis of the Effect of Inadequately or
Undisclosed Conflicts of Interest on Pension Plan Rates of Return

The SEC has stated that disclosure helps to mitigate the effects of
conflicts of interest. There is concern that plans that use pension
consultants who have not properly disclosed conflicts of interest may
achieve lower net returns on plan assets either because of higher
administrative costs or due to poor money manager selection, among other
reasons. To investigate the relationship between returns and improperly
disclosed conflicts of interest, GAO compiled a database using SEC data on
pension consultants and the Department of Labor's 5500 data (as well as
some auxiliary data sources to create additional control variables). The
data contains observations on 1111 pension plans over a 5-year period,
2000 to 2004. To analyze the relationship, GAO employed various
multivariate econometric models using the panel data. While, the results
suggest a negative correlation between returns and plans that are
associated exclusively with pension consultants that have not properly
disclosed conflicts of interest, the results should not be viewed
necessarily as evidence of a causal relationship in light of the modeling
and data limitations. This appendix provides additional information on the
construction of GAO's database, the econometric model, additional
descriptive statistics, and the limitations of the analysis.

GAO Panel Data Sample Constructed from Two Primary Data Sources

To explore the risk areas relating to pension consulting, SEC's Office of
Compliance Inspections and Examinations (OCIE) conducted focused
examinations of 24 pension consultants who were registered investment
advisers, some of whom were considered at high risk for undisclosed
conflicts. These consultants examined ranged in size and by the types of
products and services offered. SEC chose its sample in part based on
geographical dispersion and judgmentally selected the consultants. SEC
found that 13 of these 24 pension consultants failed to disclose
significant conflicts of interest while the remaining 11 were found to
have less significant disclosure issues. Using the Labor's 5500 data, GAO
used the SEC information to identify 983 pension plans associated with
these 13 pension consultants and 39 pension plans associated with the 11
pension consultants found to have less significant disclosure issues. We
were also able to identify 89 plans in the 5500 database that were
associated with both types of pension consultants (see figure 2). Given
the nature of the SEC selection process (it was not selected randomly) and
the small number of pension consultants, the plans included in the
analysis should not be considered as representative of the population of
defined benefit pension plans and the results may not be generalizeable.

Figure 2: Identifying Plans for GAO's Pension Plan Sample

To construct the database used to estimate the econometric model, we
compiled financial information from the 5500 database on these 1111 plans
over 5 years and added additional data on the performance of the S&P 500
over various fiscal year end dates taken from Robert Shiller's Web site
and market performance indicators from Credit Suisse and the Federal
Reserve Board. As a panel data set, data pooled across all plans matched
to the 24 consultants reviewed by SEC over the 2000 to 2004 period, we
were able to account for variances in returns across plans and over a
short period of time and utilizes techniques that enhance the validity of
the parameter estimates. Because some of the plans did not have the
requisite data for each year, the panel is unbalanced. While this requires
minor modifications in the computation of the related statistics, it does
not preclude the estimation of the model. Nevertheless, this panel of 1111
plans was used to empirically evaluate the relationship between returns
and undisclosed conflicts of interest. Table 2 reports some descriptive
statistics on the plans included in the analysis.

Table 2: Selected Descriptive Statistics for Plans included in the
Econometric Model

                               Return  Assets per                  Assets per 
                  Return    (standard        plan                        plan 
               (average,   deviation,   (average,   Sum of assets   (average, 
              2000-2004)   2000-2004)       2004)         in 2004  2000-2004) 
Plans            3.2%        11.5% 186,708,455 183,534,411,177 155,371,479 
associated                                                                 
with Group                                                                 
A                                                                          
Plans            4.5%        12.7% 253,121,820   9,871,750,992 214,063,302 
associated                                                                 
with Group                                                                 
B                                                                          
Plans            4.2%        12.4% 363,821,650  32,380,126,875 317,061,230 
associated                                                                 
with Group                                                                 
A and B                                                                    
All plans        3.4%        11.6% 203,227,983 225,786,289,044 170,426,326 

Source: GAO Analysis.

Note: The groups reflected in the table are as follows: Group A consists
of those consultants that had failed to disclose significant conflicts of
interest, Group B consists of those consultants that failed to disclose
less significant conflicts of interest or that had no conflicts of
interest, and Group A and B consists of members of group A and B.

Standard Econometric Modeling Procedures for Handling Panel Data: Random and
Fixed-Effects Model

Panel data provides potential advantages over pure cross sectional and
pure time series designs as it allows us to factor out the time- and
space-invariant components of the data. As a result, panel data are able
to identify and measure effects that are not detectable in other designs.
There are two commonly accepted approaches to estimating panel data, the
"random-effects" model and the "fixed-effects" model. In a "fixed-
effects" model individual effects are estimated in this case for each plan
to reflect the assumption that special features specific to each plan such
as investment style or skill can be captured best with a different,
time-invariant intercept for each plan. In a "random-effects" model, in
this context, these individual effects are captured through treating the
intercept as a random variable with an unique error term for each plan.
While each model has its advantages and disadvantages, the random effects
model is appropriate when we can plausibly assume that the individual
effects (which are unobserved and unmeasured in the model) are
uncorrelated with the explanatory variables that are measured and included
in the model. If this assumption holds, the random-effects model is
especially attractive if the cross-sectional units (pension plans) are
drawn randomly from a common population or alternatively when the number
of cross-sectional units is large and the time period is small. Otherwise,
the fixed-effects model is preferred, especially as a control for omitted
variables bias.

Using panel data as stated above, basic model takes the form:

                     (1) y[it] = th + X[it]b + Z[i]d + e[it

]where y = the dependent variable (plan returns).

X = a matrix of explanatory variables that varies across time and
individual plans. These variables are control variables that help explain
the variation in returns across plans such as the performance of various
markets over a plan's fiscal year, the size of a plan and its funding
status.

Z = a matrix of variables that vary across individuals plans but for each
individual plan are constant across the 5 years. The variables are
essentially the dummy variables that indicate whether a plan is associated
with the various types of pension consultants outlined above.

th = constant term.

i = 1, 2, . . ., 1111 and represents the individual pension plans in the
panel data.

t = 1, 2, . . ., 5 and represents the number of years (2000 to 2004).

As is the typical case with panel data, we have a large number of
cross-sections (pension plans) and a relatively small number of time
periods. Therefore we specify the composite error structure for the
disturbance term as follows:

                            (2) e[it] = a[i] + y[it

]where a[i] = plan-specific error component which captures the unobserved
heterogeneity across plans (either as a fixed- or random-effect).

E(X[it]y[it]) = 0 (there is no correlation between y[it] and X[it]).

The a[i] is the individual effect which can be treated as either fixed or
random. The fixed- and random-effect models which take account of the
repetition inherent in the data and allow us to use the individual
differences effectively. Correspondingly, if we treat the individual
effect as zero we can estimate the model using the simple ordinary least
squares (OLS) procedure. Essentially, this is a pooled regression model
where we assume the intercept and slope coefficients are constant across
time and space and the normal error term (y[it]) captures differences over
time and individual plans. However, when the true model is a
random-effects model, pooling the observations in this manner using OLS
produces biased estimates that are also not efficient when compared to the
more complex GLS procedure (outlined hereafter). The pooled OLS model is
also susceptible to omitted-variables bias.

The random-effects technique proceeds under the premise that the ignorance
about the unobserved differences in returns across plans is better
captured through the disturbance term rather than the intercept. The
random-effects model basically maintains that the 1,111 pension plans in
the sample are a drawing from a much larger universe of such plans and
that they have a common mean value for plan returns (represented by the
constant term, th) and that the individual differences in the intercept
values of each plan are captured in the error term e[it].^1 Given the
composite nature of the new disturbance term that incorporates the
individual random effect of each plan, the appropriate method for
producing estimates is generalized least squares (GLS).^2 Feasible GLS
derives an estimate of the covariance matrix of the error term and uses
the information (heteroscedasticity from repeated observations of the same
cross-section unit) to estimate the coefficients in the model.

Note that the random-effects model uses the heterogeneity across units to
produce more efficient estimates.^3 However, the drawback to this approach
is that it forces one to make the strong assumption that the unobserved
random-effects are uncorrelated with the explanatory variables in the
model (E(X[it]a[i]) = 0 in addition to the standard assumption
E(X[it]y[it]) = 0). As a result the random effect treatment of the panel
data may also produce estimates that suffer from the inconsistency because
of omitted variables. Therefore, the validity of the results in the case
would depend more heavily on the control variables included in the model
to capture differences across plans, unless the omitted variables
(unobserved heterogeneity across plan) are uncorrelated with the conflict
dummy variable. If this is the case, the random-effect model may produce
more appropriate estimates than the fixed-effects model. In our case, the
unobserved effects, a[i], were found to be unimportant (relative to the
variance of y[i]) as the random-effect estimates approximated the pooled
OLS results. This made the choice between random and pooled OLS a moot
point (see below).

^1The random effects model can be thought of as a regression with a random
constant term. In other words, it is assumed that the intercept is a
random outcome variable that is a function of a mean value plus a random
error.

^2Because a[i] is in the composite error for each time period t, the error
term (e[it] = a[i] + y[it]) is serially correlated across time,
invalidating OLS estimates.

^3The technique uses the additional information (heterogeneity) to achieve
potential efficiency gains, meaning the standard errors on the estimator
can have minimum variance. These efficiency gains come at the risk of
biased estimates when compared to the fixed effects model.

Fixed-Effect Model

When there is heterogeneity that cannot readily be explained, another
analytical approach is to incorporate it into a fixed-effects model. In
the case of the fixed-effects model, a[it] is estimated uniquely for each
plan as a fixed coefficient to be added to the intercept term. In this
way, we take into the account the individuality of each plan (each
cross-sectional unit) by letting the intercept vary by a fixed amount for
each plan. The benefit of the fixed-effects estimator is that it is
consistent in the presence of omitted variables. Because many variables
that impact returns across plans are difficult to measure or could not be
obtained this omission could bias the parameter estimates. With panel data
and a fixed-effect specification it is possible to obtain consistent
estimates of the impact of undisclosed conflicts of interest even when
there are correlated omitted effects. The differences that exist across
plan are essentially pulled out and accounted for explicitly, allowing for
a more valid estimation of the impact of conflicts of interest on plan
returns. Moreover, in many cases the fixed-effects estimates will still
produce consistent estimates even when the random effects model is valid.

While the easiest way to implement the fixed-effects estimator is to
include a dummy variable for each plan, we could not run a fixed-effect
model directly due to the nature of our primary regressor of interest.
Since the fixed-effects are time-invariant and the conflict variable in
our model is a qualitative variable (dummy) that does not change over the
2000 to 2004 period either, the fixed effects model is not able to
identify the impact of the variable. Essentially the variable is collinear
with the fixed effect intercepts. Therefore we used an alternative
procedure to achieve the same effect. To produce the fixed-effects
estimator we used the fixed-effects vector decomposition approach.^4 The
technique estimates the fixed-effects estimator in three stages: the first
stage runs a fixed-effect model without the time invariant variables
(Z[i]). We then decompose the fixed-effects estimator into a portion
explained by the time invariant variables (Z[i]) and an error term. The
final stage re-estimates the first stage with the time invariant and time
variant variables and the error term from the stage two. In the third
stage, estimated by pooled regression, we used robust standard errors
adjusted for the degrees of freedom. In this manner we were able to
approximate the unbiased, consistent estimator in the presence of
time-invariant omitted variables.^5 While some researchers have found that
this procedure has better finite sample properties than the alternative
approaches for estimating the effect of time-invariant variables using
panel data, it should be noted that this is a recently applied econometric
technique.

Variables Included in the Model

The dependent variable in all of our econometric models is plan returns.
Returns were calculated two ways using the Form 5550 data. The first
return measure calculates plan returns relating the change in plan assets
(A[t] - A[t - 1]) over the year, netting out the impact of benefits
payments from the plan (B) and contributions to the plan (C) and also
accounts for net transfers (T) into the plan. The formula can be written:

(3) ROR[1] = [(A[t] - A[t - 1]) + B - C - T] / [(A[t - 1]) +  1/2(C - B +
T)].

As an alternative we slightly amend this calculation to account for
administrative expenses (E) paid by the plan in a different manner. This
alternative formula can be written:

(4) ROR[2] = [(A - A[t - 1]) + B + E - C - T] / [(A[t - 1]) +  1/2(C - B -
E + T)].

The results we report below use this measure of returns but we obtained
similar results using the first estimate of returns.

^4See for example T. Plumper and V. Troeger, "Efficient Estimation of
Time-Invariant and Rarely Changing Variables in Finite Sample Panel
Analysis with Unit Fixed Effects." Political Analysis, Vol. 15, 2007.

^5Because of the need to estimate the model in steps, the estimator is
consistent if the assumption underlying our estimator is correct the
time-invariant variable is uncorrelated with the unobserved unit effects.
Otherwise, the estimates may be inconsistent.

The primary variables of interest are the time invariant variables (Z),
namely a dummy variable (conflict) that equal 1 if the plan is associated
exclusively with pension consultants found to have undisclosed conflicts
of interest and 0 otherwise. In many specifications we also include a
dummy variable (mixed) that equals 1 if the plan is associated with both
types of consultants -- pension consultants found to have undisclosed
conflicts of interest and pension consultants that have no conflicts or
disclosed conflicts properly and 0 otherwise.

Although, the fixed effect model guards against time invariant omitted
variables bias, it is always advisable to explore possible causes of
heterogeneity. We included a number of control variables in attempt to
capture the variation in plan return across plans although time
constraints restricted the variables we could include. Because different
plans may allocate assets differently because of investment style or age
composition of plan participants, some plans may track more conservative
or aggressive benchmarks rather than the overall market. As a result in
addition to a general market indicator, the S&P 500, we also include a
measure of hedge fund performance as well as a fixed income measure. The
broad market measure the performance of the S&P 500 over plan i's fiscal
year for year t.^6 Our measure of hedge fund performance is the Credit
Suisse/Tremont hedge fund index.^7 The fixed-income measure is the Moody's
yield on corporate seasoned Aaa bonds taken from the Federal Reserve
Board. These variables were constructed in a manner that also accounts for
the varied fiscal year end dates across plans. Moreover, since the size
and the funding level of the plan may influence asset allocation and
investment strategy, we included assets at the beginning of the fiscal
year and the degree of under-funding as explanatory variables as well.
Including funding status creates potential simultaneous equations bias
since the funding ratio is most likely dependent on plan returns. Since
lagging the variable resulted in a loss of both a year's data and large
number of observations as well as severe autocorrelation, we included the
contemporaneous funding ratio but did not include the variable all
specifications. The asset variable was substantively and/or statistically
insignificant across multiple specifications and therefore it was not
included in some instances.

^6 [53]http://www.econ.yale.edu/~shiller/data.htm (2007).

^7Credit Suisse/Tremont Hedge Fund Index is compiled by Credit Suisse
Tremont Index LLC. It is an asset-weighted hedge fund index and includes
only funds, as opposed to separate accounts. The Index uses the Credit
Suisse/Tremont database, which track over 4500 funds, and consists only of
funds with a minimum of US$50 million under management, a 12-month track
record, and audited financial statements. It is calculated and rebalanced
on a monthly basis, and shown net of all performance fees and expenses. It
is the exclusive property of Credit Suisse Tremont Index LLC.

We also included time period effects whenever possible. This amounts to
creating a dummy variable for 4 of the 5 time periods covered in the
database. While, this is straightforward in the OLS and fixed-effects
models, two way random effects or random effects with a time period fixed
effect is only possible for balanced panel in the econometric software
used for the modeling procedure. When we included time period fixed
effects in the fixed-effects model some of the explanatory variables
became redundant and added no explanatory power to the models. In our
case, we did not reproduce the random-effects model on the balanced panel
in this appendix, since the only variation across units were the fixed
effects, and the random effects model was equivalent to the pooled OLS
results.

Results: Ordinary Least Squares (OLS) and Random-Effects Models

The simple econometric model (OLS), suggests that plans associated with
undisclosed conflicts of interest achieve returns roughly 1.2 to 1.7
percentage points lower. The results are all significant at the 5 percent
level (table 3). However, this model disregards the space and time
dimensions of the pooled data and is plagued with a number of issues
including omitted variables bias, which can impact the parameter
estimates, as potentially evidenced by the somewhat low Durbin-Watson
statistic. The random-effect model, which assumes that there are
differences between the plans and that these differences are random, did
not produce results distinct from the OLS model. When the unobserved
effects, a[i], are unimportant (relative to the variance of y[it]), the
random-effects estimates will be closer to a pooled OLS model. Our
estimation found that the random-effects were unimportant and there were
no efficiencies to estimating the model via GLS. Nevertheless, the
relationship between undisclosed conflicts and returns estimated by the
OLS and random-effects models remained robust even when additional control
variables were included and, in the case of OLS, when time fixed-effects
were added to the model.^8

8Time period fixed effects were included in the OLS model only since
two-way random effects or mixed random effects and fixed-time effects
cannot be estimated for an unbalanced panel.

Table 3: Econometric Estimates of the Relationship between Undisclosed
Conflicts of Interest and Plan Returns (OLS and Random Effects)

                    OLS/Random OLS/Random                                     
                     Effects I Effects II  OLS III   OLS IV    OLS V   OLS VI 
Independent variable                                                       
Conflict           -0.012^a    -0.014a  -0.014a  -0.013a  -0.014a  -0.017b 
Mixed                                                               -0.006 
SP500               0.600^a    0.676^a -0.090^a    0.050    0.053    0.050 
Hedge                         -0.575^a          -0.252^a -0.263^a -0.253^a 
Bond yield                    -0.023^a           0.062^a  0.062^a 0.0612^a 
Assets                          -0.000             0.000                   
Funding ratio                                              -0.000          
Constant            0.062^a    0.252^a  0.043^a -0.341^a -0.336^a -0.330^a 
Time period              No         No      Yes      Yes      Yes      Yes 
fixed-effects                                                              
R-square             0.5276     0.5584   0.6388   0.6430   0.6411   0.6423 
Durbin-Watson          1.86       1.95     1.61     1.61     1.61     1.61 
Sample size           4,170      4,170    4,170    4,170    4,170    4,170 

Source: GAO analysis.

adenotes significance at the .01 level.

bdenotes significance at the .05 level.

Notes: Conflict indicates a plan that is associated with one of the
pension consultants identified has having undisclosed conflicts of
interest. Mixed indicates a plan that is associated with a pension
consultant with undisclosed conflicts but also a pension consult found to
be free of conflict or having disclosed them properly.

Bond Yield, SP500, and Hedge are: Moody's average yield on corporate Aaa
bonds, the change in the S&P 500, and the change in the Credit
Suisse/Tremont hedge fund index return over a plan's fiscal year,
respectively. Assets denote plan assets at the beginning of the year
(squared). The funding ratio is the ratio of assets to liabilities.

Results: Fixed-Effects Models

The fixed-effect model, which helps guard against omitted variable bias,
supports the results from the pooled OLS model. The R-square from the
fixed-effect regression suggest that the models explain roughly 75 percent
of the variation in plan returns. Again, the results are highly
significant as the probability of an erroneous statistical conclusion in
most models is substantially lower than what is commonly accepted as
significant in hypothesis testing (5 percent or a p-value of .05). There
is one exception to be noted, when the dummy variable is included for
those plans associated with both conflicted and non-conflicted pension
consultants, the significance of the conflict variable falls to the 10
percent level (p-value is roughly 6 percent). This implies the probability
of concluding a negative relationship when none is present has increased
to about 6 percent. Moreover, when we drop those observations associated
with both types of pension consultants, the conflicted variable was again
significant only at the 10 percent level (p-value on roughly 6.7 percent).
However, when we used the return calculation expressed in equation (3) the
conflict dummy remains significant at the 5 percent level (p-value of
roughly 3%) even when an independent dummy variable is included for the
plans associated with both conflicted and non-conflicted pension
consultants. It should be noted that, against the one-sided alternative,
returns are lower for conflicted plans (H1: Conflict<0) and the results
remain highly significant at the 5 percent level or lower. Otherwise, the
conflict variable is robust to the inclusion of any of the additional
control variables discussed above and the two measures of returns. We
generally find that an exclusive association with one of the pension
consultants identified has having conflicts of interest is associated with
a lower return by about 1.2 to 1.3 percentage points, and higher in one
case.

Table 4: Econometric Estimates of the Relationship between Undisclosed
Conflicts of Interest and Plan Returns (Fixed-Effects)

                                                             Model V          
                                                           (no mixed          
                       Model I Model II Model III Model IV     plans Model VI 
Independent                                                                
variable                                                                   
Conflict           -0.013^a  -0.012a   -0.025a  -0.013c   -0.013c  -0.013a 
Mixed                                            -0.001                    
SP500               0.103^a    0.051   0.053^a    0.051     0.036  0.054^c 
Hedge                       -0.115^b  -0.125^c -0.115^b    -0.063 -0.129^b 
Bond Yield                   0.108^a   0.106^a  0.108^a   0.113^a  0.107^a 
Assets                                -0.000^a                             
Funding ratio                                                       -0.000 
Constant            0.042^a -0.648^a  -0.594^a   -0.647    -0.687   -0.637 
Time period             Yes      Yes       Yes      Yes       Yes      Yes 
fixed-effects                                                              
R-square              0.742    0.746     0.747    0.746     0.753    0.745 
Durbin-Watson          1.94     1.94      1.93     1.94      1.91     1.94 
Sample Size           4,170    4,170     4,170    4,170     3,385    4,170 

Source: GAO Analysis.

aDenotes significance at the .01 level.

bDenotes significance at the .05 level.

cDenotes significance at the .10 level. Standard errors (not reported)
were adjusted for heteroscedasticity using White's procedure.

Notes: Conflict indicates a plan that is associated with one of the
pension consultants identified as having undisclosed conflicts of
interest. Mixed indicates a plan that is associated with a pension
consultant with undisclosed conflicts, but also a pension consult found to
be free of conflict or having disclosed them properly.

Bond Yield, SP500, and Hedge are: Moody's average yield on corporate Aaa
bonds, the change in the S&P 500, and the change in the Credit
Suisse/Tremont hedge fund index return over plan i's fiscal year,
respectively. Assets denote plan assets at the beginning of the year
(squared). The funding ratio is the ratio of assets to liabilities.

Limitations of Our Econometric Model

Like many statistical analyses, the results should be interpreted with
care. Although the panel data provides many advantages and can produce
more valid and efficient estimates, drawing causal inferences is still
difficult. Even with control variables and the fixed-effects models there
are a number of threats to the validity of our results. First, although
the fixed-effects estimator is robust to the omission of any relevant
time-invariant variables, if there are time-varying differences that have
been omitted the result could be biased. Although the analysis controlled
for plan size, funding level, the performance of asset markets and other
key variables, other unknown, omitted factors could still influence the
results of our analysis or account for the differences in estimated
returns. There may be additional biases resulting from the vector
decomposition procedure used to obtain the fixed-effect estimates. Second,
the existence of statistical relationship is not in and of itself, enough
to assert causality. Fixed-effects, while strengthening the validity of
model's parameters, do not completely solve the problem of drawing causal
inferences. Third, the use of the 5500 data could lead to measurement
error in the dependent variable (plan returns). We assume that any errors
are random and therefore do not impact the validity of the parameter
estimates. Similarly, although we were careful in identifying and
reviewing the plans associated with the two types of pension consultants
any error, random or non-random, would impact the parameter estimates.
Moreover, we used a potentially unrepresentative sample of pension
consultants to identify the pension plans included in our investigation
that therefore limits the ability to generalize the results. A few pension
consultants that had significant conflicts of interest that impacted their
activity could very well drive the observed negative relationship.
Further, the imbalance between the large number of plans associated
exclusively with conflicted consultants and the small number of those that
were not raise additional statistical issues and limits the ability to
generalize the results. Lastly, given the short time period analyzed, it
could be possible that some plans' return were abnormally low due to their
investment strategies, and would have higher returns had the time period
analyzed been lengthened.

Appendix III: Comments from the Pension Benefit Guaranty Corporation

Appendix IV: Comments from the Department of Labor

Appendix V: Comments from the Securities and Exchange Commission

Appendix VI: GAO Contact and Staff Acknowledgments

Contact

Barbara Bovbjerg (202)512-7215

Staff Acknowledgments

In addition to the above, Charles A. Jeszeck, Kimberley M. Granger, Joseph
Applebaum, Susan Bernstein, Megan Birney, Richard Burkard, Julie DeVault,
Lawrance Evans Jr., Randall Fasnacht, Cody Goebel, Gene Kuehneman, Michael
Morris, Walter Vance, and Craig Winslow made important contributions to
this report.

(130554)

GAO's Mission

The Government Accountability Office, the audit, evaluation and
investigative arm of Congress, exists to support Congress in meeting its
constitutional responsibilities and to help improve the performance and
accountability of the federal government for the American people. GAO
examines the use of public funds; evaluates federal programs and policies;
and provides analyses, recommendations, and other assistance to help
Congress make informed oversight, policy, and funding decisions. GAO's
commitment to good government is reflected in its core values of
accountability, integrity, and reliability.

Obtaining Copies of GAO Reports and Testimony

The fastest and easiest way to obtain copies of GAO documents at no cost
is through GAO's Web site ( [54]www.gao.gov ). Each weekday, GAO posts
newly released reports, testimony, and correspondence on its Web site. To
have GAO e-mail you a list of newly posted products every afternoon, go to
[55]www.gao.gov and select "Subscribe to Updates."

Order by Mail or Phone

The first copy of each printed report is free. Additional copies are $2
each. A check or money order should be made out to the Superintendent of
Documents. GAO also accepts VISA and Mastercard. Orders for 100 or more
copies mailed to a single address are discounted 25 percent. Orders should
be sent to:

U.S. Government Accountability Office 441 G Street NW, Room LM Washington,
D.C. 20548

To order by Phone: Voice: (202) 512-6000
TDD: (202) 512-2537
Fax: (202) 512-6061

To Report Fraud, Waste, and Abuse in Federal Programs

Contact:

Web site: [56]www.gao.gov/fraudnet/fraudnet.htm
E-mail: [57][email protected]
Automated answering system: (800) 424-5454 or (202) 512-7470

Congressional Relations

Gloria Jarmon, Managing Director, [58][email protected] (202) 512-4400 U.S.
Government Accountability Office, 441 G Street NW, Room 7125 Washington,
D.C. 20548

Public Affairs

Paul Anderson, Managing Director, [59][email protected] (202) 512-4800
U.S. Government Accountability Office, 441 G Street NW, Room 7149
Washington, D.C. 20548

[60]www.gao.gov/cgi-bin/getrpt?GAO-07-703 .

To view the full product, including the scope
and methodology, click on the link above.

For more information, contact Barbara Bovbjerg at (202) 512-7215 or
[email protected].

Highlights of [61]GAO-07-703 , a report to congressional requesters

June 2007

DEFINED BENEFIT PENSIONS

Conflicts of Interest Involving High Risk or Terminated Plans Pose
Enforcement Challenges

To protect workers' retirement security, the requesters asked GAO to
assess: 1) What is known about conflicts of interest affecting private
sector defined benefit (DB) plans? 2) What procedures does PBGC have to
identify and recover losses attributable to conflicts? 3) What procedures
does Employee Benefits Security Administration (EBSA) have to detect
conflicts among service providers and fiduciaries for PBGC-trusteed plans?
4) To what extent do EBSA, PBGC, and SEC coordinate their activities to
investigate conflicts?

GAO interviewed experts, including agency officials, attorneys, financial
industry representatives, and academics, and GAO reviewed PBGC
documentation and EBSA enforcement materials. GAO analyzed Labor, SEC,
PBGC, and private sector data, including data on pensions, pension
consultants, and rates of return data, and conducted statistical and
econometric analyses.

[62]What GAO Recommends

GAO recommends that PBGC assess the risks from conflicts of interest; that
EBSA expand enforcement to include a focus on PBGC-identified plans; and
that each agency share data on conflicts. Congress should consider
amending ERISA to expand Labor's authority to recover losses against
non-fiduciaries. Each agency generally concurred with the report, although
EBSA expressed some methodological concerns.

A conflict of interest typically exists when someone in a position of
trust, such as a pension consultant, has competing professional or
personal interests. Though data are limited on the prevalence of conflicts
involving plan fiduciaries and consultants, a 2005 SEC staff report
examining 24 registered pension consultants identified 13 that failed to
disclose significant conflicts. GAO's analysis found that, in 2006, these
13 consultants had over $4.5 trillion in U.S. assets under advisement. GAO
also analyzed a sample of ongoing DB plans associated with the 13
consultants that, as of year-end 2004, had total assets of $183.5 billion
and average assets of $155.3 million. Additional sample analysis showed
that the DB plans using these 13 consultants had annual returns generally
1.3 percent lower than those that did not. Because many factors can affect
returns, and data as well as modeling limitations limit the ability to
generalize and interpret the results, this finding should not be
considered as proof of causality between consultants and lower rates of
return, although it suggests the importance of detecting the presence of
conflicts among pension plans. Whether specific financial harm was caused
by a conflict of interest is difficult to determine without a detailed
audit.

As a creditor and a trustee of terminated plans, PBGC's policies and
procedures are oriented toward the likely recovery of assets, rather than
explicitly focusing on losses associated with conflicts of interest
involving service providers. Although PBGC has broad legal authority to
recover losses attributable to conflicts of interest, PBGC officials told
us that the agency limits its pursuit of cases to those in which the
recovery will likely exceed the cost of bringing a case to court
successfully. While monetary recoveries by PBGC may improve the agency's
financial position, they generally have little effect on participant
benefits because most affected participants already receive their full
benefits promised by their plans. According to PBGC, more than 90 percent
of all beneficiaries of PBGC trusteed plans received their full promised
plan benefit.

While EBSA's enforcement program is concerned with conflicts of interest
affecting all private pension plans, it does not have specific procedures
for plans trusteed or likely to be trusteed by PBGC. EBSA has recently
initiated the Consultant/Advisor Project (CAP) to focus on conflicts among
service providers, though it includes no specific focus on high risk or
terminated plans. Moreover, existing law limits EBSA's efforts to pursue
conflicts and redress for financial harm when certain service providers
are either not fiduciaries under ERISA or did not knowingly act in concert
with a fiduciary.

Coordination among EBSA, PBGC, and the SEC on conflicts of interest is
primarily informal, in part because of agencies' different
responsibilities. The agencies' investigative activities for conflicts of
interest tend to operate independently. Differences in agency missions
pose challenges to the three agencies' developing a coordinated focus to
pursue conflicts of interest affecting individual pension plans.

References

Visible links
  37. http://www.sec.gov/news/speech/spch120505lr.htm
  38. http://www.gao.gov/cgi-bin/getrpt?GAO-04-90
  39. http://www.gao.gov/cgi-bin/getrpt?GAO-05-61
  40. http://www.sec.gov/news/speech/spch1205051r.htm
  41. http://searching.gao.gov/cs.html?charset=iso-8859-1&url=http%3A//archive.gao.gov/t2pbat6/147637.pdf&qt=terminate&col=&n=13&la=en
  42. http://www.gao.gov/cgi-bin/getrpt?GAO-05-991R
  43. http://www.gao.gov/cgi-bin/getrpt?GAO-07-22
  44. http://www.gao.gov/cgi-bin/getrpt?GAO-05-491
  45. http://www.gao.gov/cgi-bin/getrpt?GAO-07-22
  46. http://www.gao.gov/cgi-bin/getrpt?GAO-07-22
  47. http://www.gao.gov/cgi-bin/getrpt?GAO-04-749
  48. http://www.gao.gov/
  49. http://www.gao.gov/cgi-bin/getrpt?GAO-02-943
  50. http://www.gao.gov/cgi-bin/getrpt?GAO-03-337
  51. http://www.gao.gov/cgi-bin/getrpt?GAO-04-423
  52. http://www.gao.gov/cgi-bin/getrpt?GAO-04-395
  53. http://www.econ.yale.edu/~shiller/data.htm
  54. http://www.gao.gov/
  55. http://www.gao.gov/
  56. http://www.gao.gov/fraudnet/fraudnet.htm
  57. mailto:[email protected]
  58. mailto:[email protected]
  59. mailto:[email protected]
  60. http://www.gao.gov/cgi-bin/getrpt?GAO-07-703
  61. http://www.gao.gov/cgi-bin/getrpt?GAO-07-703
*** End of document. ***