Printer Friendly

Invisible pension investments.

B. Other Plan Characteristics

Beyond the great DB/DC divide, some differences in the type and extent of indirect investments were observed earlier among some pension plan sub-varieties. (Compare Figure 5 with Figure 6.) Once first-tier DFE asset holdings are properly attributed to their investor-plans, to what extent do these differences translate into real variation in categorical portfolio composition?

Figure 15 shows the asset allocations of large single-employer cash balance plans for 2008, again separating those plans that do not use DFEs (Group 1, blue columns) from those that report consistent information concerning their DFE investments on Schedules D and H (Group 3, red/pink columns). Comparing Figure 15 with Figure 13A, the similarity between the portfolio compositions of cash balance plans and all DB plans is striking. (143) Cash balance plans comprise only about ten percent of large single-employer DB plans, so the resemblance does not derive from numerical dominance. Although categorized as defined benefit programs, cash balance plans promise a benefit that mimics defined contribution plans and so represent a kind of hybrid arrangement. (144) From that perspective, close alignment between cash balance and traditional DB pension plan portfolio compositions is surprising. From an historical point of view, however, comparable investment behavior becomes understandable. The number of cash balance plans has grown dramatically since 1990, but not by instituting altogether new plans. (145) Instead, a cash balance plan is commonly created by amendment of a preexisting traditional pension plan that has already accumulated substantial investment assets, and the participants of such a converted plan are entitled to receive benefits earned prior to the amendment without diminution. (146) The data appear to support the inference that continuation of an established fund of investment assets and preservation of the plan's prior benefit structure causes the prior investment pattern to endure for an extended period following conversion to the cash balance formula.

A massive exodus from traditional defined benefit pension plans has occurred over the past 25 years. While some employers converted traditional pension plans into cash balance plans (as described in the prior paragraph), many others simply froze their traditional pension plans, amending them to cease further benefit accruals, with the result that thereafter an active employee's services would generate no additional benefits. Often a new defined contribution plan, typically a 401(k) plan, was instituted in conjunction with such a freeze, allowing workers to continue to earn retirement savings. Although its accrued liability does not increase, a frozen DB plan may continue to operate for many years, receiving annual employer funding contributions, accumulating assets, and paying benefits. (147) Does a frozen plan's investment behavior differ from that of an ongoing DB pension program? Figure 16 shows that, at the level of broad categorical asset allocations, the portfolios of frozen and ongoing traditional DB plans are very much alike. (Cash-balance plans have been excluded from the sets of both frozen and ongoing plans.) The differences are small and those that are discernible seem consistent a marginally greater reliance by frozen plans on lower-risk fixed-income investments. U.S. government securities, corporate debt (both preferred and other), interest-bearing cash, and "other investments" (which include state and local government bonds), all contribute somewhat greater proportions of the holdings of frozen than ongoing plans. Figure 16 groups the data differently than the preceding graphs (Figures 13A, 13B, 14A, 14B, and 15) and uses different column colors to highlight that fact. The quantity displayed in each asset category includes both direct and linked first-tier DFE holdings by plans of the designated type, including both plans that do not use DFEs and those that do (i.e., Group 1 and Group 3 plans are combined). Because investments by plans utilizing DFEs are not segregated, the stacked column display is not used, and the DFE numbers shown on the right side of the graph represent only the combination of linked second-tier DFE holdings and residual first-tier DFE investments that were not successfully linked. The most important feature of Figure 16 is that the comparison group does not consist of all ongoing traditional pension plans. Instead, a matching procedure was used to randomly select, for each frozen plan included in the data set, three ongoing plans having an amount of reported gross assets that falls within 2 percent of the reported gross assets of the frozen p1an. (148) As we have already seen, portfolio composition varies (often dramatically) with the level of total plan assets. This matching procedure was developed to neutralize such size effects, providing a comparison of portfolio allocations between plans of different types that is not confounded by differences in the range or distribution of total asset levels between those different plan types. (149) Such an apples-to-apples comparison is particularly important here, because smaller DB plans are much more likely to be frozen, yet any narrow asset range selected to compare frozen and ongoing plans nets very few frozen plans.

Turning to the DC universe, Figure 17A presents the corresponding comparison for all large single-employer 401(k) plans in 2008. To restrict consideration to plans of similar asset size (recall that plans using DFEs tend to have higher total assets than those that do not), Figure 17B presents weighted average portfolio allocations for a subset of high-asset (90th to 95th percentile) single-employer 401(k) plans. With one notable exception, the 401(k) plan portfolios illustrated in Figures 17A and 17B very closely track the results for DC plans generally, shown in Figures 14A and 14B. That's to be expected inasmuch as 401(k) plans account for roughly 75 percent of the DC plan universe. The exception is also quite understandable. Unlike DC plans in general, employer securities account for a much smaller share of 401(k) plan portfolios. The difference is due to the fact that a large share of all employer securities are held by ESOPs, a subtype of DC plan that is included in Figures 14A and 14B but is largely (though not entirely) excluded from the set of 401(k) plans. (150)

Figures 18A and 18B present the corresponding picture for money purchase pension plans (MPPPs), a type of DC plan that has dramatically declined in importance since 2001. (151) The overall pattern is quite similar to the asset allocation of 401(k) plans (Figures 17A and 17B), and hence to DC plans generally (Figures 14A and 14B). One stark difference is the much higher utilization of MTIAs by MPPPs (compare Figures 18A and 17A). The reason seems to lie in the fact that a larger proportion of MPPPs fall into higher asset ranges (recall that MTIA usage is very heavily concentrated among the largest plans, as shown in Figures 4-6) than 401(k) plans. (152) To eliminate the disparity in asset size, Figure 18C compares the set of single-employer money purchase pension plans with a randomly-selected matched sample of 401(k) plans having comparable asset levels. Asset size matching was performed using the procedure described earlier in connection with Figure 16, (153) and as in that figure, the quantity displayed in each asset category includes both direct and linked first-tier DFE holdings by plans of the designated type, including both plans that do not use DFEs and those that do. (The highest-asset money purchase plan was excluded from the data sets compared in Figure 18C, however, because it reports an unusual asset mix--fully 43% as "other receivables"--and its size is so large that inclusion substantially alters the weighted average portfolio of all MPPPs. (154)) The close correspondence between MPPP and 401(k) plan asset allocations, combined with the heavy reliance on registered investment company (mutual fund) shares, suggests that money purchase plans have, like 401(k) plans, largely shifted to responsibility for investment decision-making to participants, who are a choice between a menu of mutual funds, as permitted by ERISA [section] 404(c). (155) And in fact, 69.5 percent of the 1,260 single-employer MPPPs reported plan characteristics codes indicating that the plan provided for total or partial participant direction of investments, as did 96.8 percent of the 3,606 matched 401(k) plans in the comparison group.

The results for large single-employer employee stock ownership (ESOP) plans in 2008 are given by Figures 19A and 19B. An ESOP is defined in part as a plan designed to invest primarily in employer stock, (156) and so it comes as no surprise that for ESOPs that do not use DFEs (Group 1) employer securities is the largest asset category (70% or more) by a wide margin. The shock lies in the asset allocation of plans that use DFEs. To judge by these numbers, many of these plans are not properly classified as ESOPs because they do not invest primarily in employer stock even once indirect investments are properly characterized! No ready explanation for this apparently dramatic misreporting comes to mind.

In unionized workforces, benefits are the subject of collective bargaining and the union often plays an important role in monitoring the implementation of the agreement. Therefore one might expect to find differences in the portfolio composition of pension plans according to whether or not the plans are products of collective bargaining. Multiemployer plans are the best-known example of collectively-bargained pension programs, but those plans cover employees of multiple unrelated employers, have a unique joint governance structure (involving union and employer representatives), are subject to special funding and termination insurance rules, and tend to have very high asset levels. (157) Consequently, if multiemployer plans exhibit different investment allocations than single-employer plans, the difference could be attributable to a number of factors other than union monitoring. To avoid those confounding variables, the authors looked for a unionization effect by sorting the 2008 single-employer pension plan balance sheet data into subsets composed of plans that were or were not collectively-bargained. (158) DB and DC plans were sorted separately, and the subsets (collectively bargained or not) were further subdivided into groups consisting of those plans that did not utilize DFEs (Group 1), and those using DFEs and filing consistent information to permit linking (Group 3).

Figure 20 displays the major categorical asset allocations of collectively-bargained and non-collectively-bargained large single-employer defined benefit plans in 2008. It adopts a clustered stacked column format similar to that used in the preceding charts, but each asset category now contains four columns rather than two. This four-column display permits side-by-side presentation of the portfolio compositions of: (1) non-collectively bargained plans that do not use DFEs (dark blue single columns); (2) linkable non-CBA plans that use DFEs (red/pink stacked columns); (3) collectively-bargained plans that do not use DFEs (light blue single columns); and (4) linkable CBA plans that use DFEs (orange/salmon stacked columns). Also like the preceding charts, the right side of the graph shows the utilization and linking results for the four types of DFEs.

Figure 20 should be compared with Figure 13A, showing the composite results for all large single-employer DB plans regardless of collective bargaining status. So, does unionization of the covered workforce affect DFE utilization or the categorical composition of pension plan portfolios, taking into account linked first-tier DFE investments? The answer for large single-employer DB plans seems to be very little. When DB plans are segregated by collective bargaining status only two conspicuous disparities emerge. First, collectively-bargained plans make substantially greater use of master trust investments than do non-CBA plans (76% versus 51%). Curiously, however, a larger proportion of all non-CBA plans that use DFEs fall into the highest asset ranges than is the case for the CBA plans that use DFEs. Hence higher reliance on MTIA investments by union plans is not attributable (as it was in so many other cases) to higher average asset levels. Second, an exceptionally large proportion of MTIA interests held by collectively-bargained plans could not be successfully linked to the underlying MTIA investments. Even though the amount reported by an investor-plan as the year-end value of its interest in MTIAs on Schedule H matched the total MTIA values reported on Schedule D (that is, we are dealing with potentially linkable Group 3 plans), it is unusually common for these collectively-bargained DB plans to fail to intelligibly identify the particular MTIAs in which they hold an ownership stake. Indeed, there is a much higher rate of unlinked first-tier MTIA investments among DB plans overall than among DC plans (15% versus 3%, as shown in Figure 13A and Figure 14A), and Figure 20 reveals that exceptional failure rate to be almost entirely attributable to collectively-bargained plans. (159) Why collectively-bargained DB plans are far more likely to file defective Schedule D information is a mystery. One wonders whether some sponsors of collectively-bargained DB plans might be blunting union oversight by engaging in deliberate obfuscation. The situation at least looks suspicious.

Figure 21 gives the corresponding breakdown between collectively-bargained and non-CBA large single-employer DC plans in 2008, again using a four-column format and showing the contributions of the direct and indirect holdings of plans that use DFEs by means of stacked columns. Figure 21 should be compared with Figure 14A, showing the composite results for all large single-employer DC plans regardless of collective bargaining status. There is a very close correspondence between Figure 14A and the Figure 21 columns representing asset allocations of plans that are not collectively bargained for the simple reason that collectively-bargained plans are a very small segment of the DC universe. (Note the low "n" reported in Figure 21 for each group of CBA plans.) Comparing the CBA and non-CBA portfolios, the most salient difference is the higher reliance on MTIAs by collectively-bargained plans (45% versus 26% for non-CBA plans using DFEs). Yet this greater utilization of MTIAs by collectively-bargained DC plans is not associated with a high linking failure rate, as was found for collectively-bargained DB plans.

C. Plan-Level Correlations

In order to evaluate individual plan behavior we examine the correlation between direct plan investment in each non-DFE asset and liability category and first-tier DFE values for the same category. Figure 22 presents these correlations together with ninety-five percent confidence intervals for large single-employer defined benefit plans in 2008. Categories with a positive correlation are depicted in boldface type and indicate that as plans hold a higher percentage of their assets (or liabilities) directly in the indicated category, their indirect investment in that category trends upward as well. The categories listed in italics have a statistically significant negative correlation between direct and indirect investment and evidence the opposite pattern. For the categories whose confidence intervals include zero, there is no evidence indicating association between the percentage of assets directly invested in a category and the percentage of assets indirectly invested in a category. (160) Figure 23 reports the correlation results for defined contribution plans.

Strong negative correlations are seen for mutual funds (RICs), and in the case of DC plans, insurance company general accounts. Thus, an increase in a plan's direct holdings of mutual fund shares is associated with investment in DFEs that allocate a lower proportion of their portfolios to mutual funds. It may be relevant that insurance company general account interests often correspond to guaranteed investment contracts (GICs), and under participant-directed DC plans a GIC investment option may be offered in conjunction with an array of mutual fund (RIC) choices. The dearth of other strong negative correlations seems to confirm that DFEs arc not systematically employed to achieve categorically contrarian investment positions (i.e., divergent from a plan's direct holdings). While DFEs could be exploited to obfuscate the broad categorical nature of plan investments, that ulterior usage doesn't appear to be widespread.

D. Characteristics of Plans Filing Inconsistent DFE Information

In 2008, fifty-three percent of large pension plans reported investing in at least one DFE. Such plans are required to report the dollar amount of their DFE interests in two different ways. First, a plan must report the total amount invested in each of the four types of DFE on Schedule H. Second, Schedule D of Form 5500 calls for the plan to identify each specific DFE in which it invested, indicate the type of that DFE (as either MTIA, CCT, PSA, or 103-12 1E), and list the dollar amount invested. The sum of the Schedule H reporting of DFE investment should equal the sum of the DFE investment identified individually on Schedule D. Yet more than thirty-five percent of plans that invest in DFEs report inconsistent numbers on Schedule H and Schedule D. Such inconsistent filings were excluded from the linking protocol--and therefore from the results and analysis presented above--because knowledge of the value of the plan's interest in a DFE is required to impute the correct share of DFE holdings to the plan. (161) Where the numbers on Schedules D and H do not mesh, it is not clear which numbers (if either) should control the imputation. In this section, we in estigate the characteristics of plans that are likely to file inconsistent returns.

The appropriate comparison group for inconsistent filers is the set of plans that invested in DFEs and reported consistent numbers on Schedule H and Schedule D. In 2008, 42,445 large plans reported investing in at least one DFE. Of these, 15,198 filed inconsistent returns. To identify factors associated with bad filings we use a probit regression model setting the outcome variable equal to one if a plan filed an inconsistent return and zero otherwise. The explanatory variables include the various pension plan types and a variety of characteristics that might plausibly influence the quality of the plan's annual return. The results of the model are set forth in Table 2. Coefficients with an asterisk are statistically significant at a 0.05 level.

Table 2: Regression Model of Inconsistent Reporting on Schedule
H and Schedule D by Large Single-Employer Plans, 2008

                              Coefficient  Standard  p-value
                                              Error

Intercept                        -0.457 *     0.122    0.001

Total assets                        0.000     0.000    0.340

Total liabilities                   0.000     0.000    0.118

Total DFE investment (D1)           0.000     0.000    0.080

Defined benefit                  -0.315 *     0.117    0.007

401(k)                           -0.341 *     0.055  < 0.001

ESOP                               -0.068     0.080    0.395

Stock bonus                         0.158     0.124    0.203

Profit-sharing                      0.148     0.114    0.192

Money purchase                      0.053     0.118    0.653

Cash balance                        0.048     0.067    0.475

Accountant opinion

Adverse                            -1.084     1.983    0.584

Disclaimer                        -0.15 *     0.028    0.000

Not reported                      1.984 *     0.032  < 0.001

Qualified                           0.064     0.080    0.427

Invested in at least one 000      4.431 *     0.145  < 0.001
DFE

Collective bargaining             -0.24 *     0.031  < 0.001

Frozen plan                      -0.205 *     0.061    0.001

Large amount of "other            0.306 *     0.058  < 0.001
investments"

Data from OPR editor              0.374 *     0.033  < 0.001

Plan maturity                    -0.291 *     0.056  < 0.001

Participant direction

Partial                          -0.140 *     0.067    0.036

Total                            -0.365 *     0.048  < 0.001

Any imputed assets                2.364 *     0.200  < 0.001


Somewhat surprisingly, the amount of money involved does not appear to have much, if any, impact on inconsistent filing. Neither the amount of total assets nor total liabilities is statistically significant. The total dollar amount a plan invests in DFEs is not significant at a 0.05 level, but it is at a more relaxed 0.1 level. Defined benefit plans are less likely to file inconsistent returns than defined contribution plans. Similarly, 401(k) plans are less likely to file inconsistent returns than plans without a cash-or-deferred feature. The rest of the plan types are not significant; holding other variables constant, these plan types experience similar rates of inconsistent filing.

Form 5500 is generally accompanied by an accountant's opinion on the fairness and consistency of the presentation of the plan's financial information, including Schedule H and the schedule of investment assets. The accountant is not required to examine or report on financial information relating to a DFE, nor does the accountant's opinion certify the correctness of the annual return generally. (162) The opinion is characterized on Schedule H as either "unqualified," "qualified," "disclaimer," or "adverse." An unqualified opinion indicates that the accountant has no reservations about the plan's financial statements. (163) This category serves as the baseline in our model. This variable produces some surprising results. An adverse opinion does not significantly raise the probability of an inconsistent return, but the estimate is unreliable due to the fact that the data contain only three instances of adverse opinions. More perplexing is the significant negative result for filings with an accountant's disclaimer, which occurs where the accountant declines to express an opinion because the accountant has not performed an audit sufficient in scope to support a judgment on the fairness of the financial statements. These plans are actually less likely to file inconsistent DFE information than those with which an accountant finds no fault. Conversely, returns filed without an accountant opinion reported at all are more likely to be inconsistent. Finally, there are a substantial number of "qualified" opinions (meaning that the accountant opines that the plan's financial statements are fair in all material respects except for one or more matters described in the opinion), yet these filings are not more likely to be inconsistent than those which are approved without qualification.

The rest of the variables in the model all prove to be statistically significant. First consider factors that increase the probability of an inconsistent return. As explained earlier, a CCT or PSA may file its own Form 5500, in which case it is a DFE, but it is not obligated to do so.164 A plan that invests in a CCT or PSA that does not file as a DFE must allocate its share of the underlying assets and liabilities of the CCT or PSA to the separate asset and liability categories on Schedule H (instead of simply including the net value of the plan's overall interest in the trust or account as the value of an interest in a CCT or PSA). In addition to allocating the underlying assets of the indirect investment vehicle on Schedule H, the plan is also required to file a Schedule D identifying the CCT or PSA, even though it does not file as a DFE. Such a non-DFE CCT or PSA is identified on Schedule D by the name and employer identification number of the CCT or PSA just as if it were a DFE, but the failure to file as a DFE is indicated by reporting a three-digit plan number of "000". (Hence we refer to such a non-DFE indirect investment vehicle as a triple-zero CCT/PSA.) When a plan invests in a triple-zero CCT/PSA in conjunction with at least one true DFE, the probability of inconsistent filing increases. This makes sense due to the confusing mutually exclusive reporting rules applicable to DFE and non-DFE CCTs and PSAs, which is exacerbated by ambiguous terminology used on Schedule H. (165) The other variables that increase the likelihood of being inconsistent all suggest a lack of competency or knowledge on the part of the filing plan. When the dollar amount in the category "Other Investments" is greater than two standard deviations above the mean, or when the calculated sum of all reported asset categories does not equal the reported sum of assets, the probability of an inconsistent filing increases. Finally, when the data in a Form 5500 had to be edited by hand (in the OPR editor) by the private contractor tasked with cleaning the data, this indicates some underlying problem in the Form 5500 itself. Such data problems are correlated with a higher probability of the inconsistencies which are the focus of our investigation. (166)

Frozen plans (under which active employees accrue no additional benefits), more mature plans (meaning plans with a low ratio of active to total participants), and collectively bargained plans all have a lower probability of filing inconsistent returns. (167) Also, plans with total participant direction have the lowest rate of inconsistent returns, followed by plans with partial participant direction. Plans with no participant direction (which is the baseline category) have a higher rate of inconsistent returns than those with either partial or total participant direction.

While the regression output in Table 2 provides information on which variables are statistically significant and whether their effects are positive or negative, it does not immediately translate into intuition about the size of such effects and how they might operate in combination. Consequently, we provide graphical illustrations of how the overall predicted probability of filing an inconsistent return changes at different levels of several variables. For purposes of these graphs, any variables not listed are held at their median value and all point estimates are accompanied by ninety-five percent confidence intervals.

Figure 24A illustrates how the probability of inconsistent filing predicted in the model changes with the level of total plan investment in DFEs at different discrete values of the accountant opinion variable. In all cases the probability of an inconsistent filing rises very slightly as total DFE investment increases. While reports filed with an accountant's disclaimer are less likely to be inconsistent than those accompanied by an unqualified report, this difference is quite small. In addition, for larger dollar amounts of DFE investment the confidence intervals overlap, indicating that the small predictive effect of these two types of accountant opinions tends to disappear entirely as the level of DFE investment goes up. The most striking result is the substantial impact of not reporting an accountant opinion at all. (168) Holding other variables at their median, this variable increases the likelihood of filing an inconsistent return from a range of ten to twenty percent range to around eighty percent.

Next, Figure 24B sets forth the predicted probabilities for the modal plan in the data set together with a variety of comparable estimates in each of which one key variable takes on a different value. The modal plan is a defined contribution 401(k) profit-sharing plan that has an accountant opinion of "Disclaimer", total participant direction, and the other dichotomous variables equal zero. When the continuous variables are set at their median, this modal case has a 0.08 probability of having an inconsistent return. The dotted line in Figure 24B is set at this level to facilitate comparisons with the modal case. Investing in at least one triple-zero CCT/PSA and reporting non-zero imputed assets (which indicates discrepancy in the reported and calculated total assets) both substantially increase the probability of an inconsistent filing, making it much more likely than not. The effect of the other variables is considerably more modest.

Most of the different plan types are not statistically significant in the model. Different plan types are strongly associated with different values of some of the other variables, however. Therefore, merely considering the effect of plan type in isolation (while holding everything else constant) is unrealistic and misleading. Figure 24C presents the predicted probabilities for each different plan type determined by holding the other variables at the median for all plans of that same type. Rather than provide a direct comparison of the isolated effect of plan type, this presents a descriptive account of the estimated probability of filing an inconsistent report within each plan type. The most common plan is one variety of defined contribution plan, namely, a 401(k) profit-sharing plan (i.e., profit-sharing plan with an elective cash-or-deferred contribution feature). Since these three characteristics (DC, profit-sharing, and 401(k)) tend to co-occur, they are presented together. The difference between defined benefit and defined contribution plans is fairly substantial, and is actually the opposite of what might be expected. Table 2 shows that, ceteris paribus, classification as a defined benefit plan has a negative effect. Taking into account the likely values of other variables, however, a typical DB plan has nearly twice the probability of filing an inconsistent return (0.15) as a typical DC plan (0.08). The estimates for other plan types are less precise (because there are fewer data) and overall appear to be somewhat similar.

V. FINANCIAL REPORTING AND DISCLOSURE POLICY

Collective arrangements for the investment of pension funds yield important financial advantages in the form of enhanced diversification and reduced costs via economies of scale. The current system of simplified annual reporting, whereby a plan investing in a collective vehicle that qualifies as a direct filing entity (DFE) need only identify the entity and report the value of its interest therein, obscures pension plan investment allocations. While a fair picture of a plan's overall portfolio composition, both direct and indirect (i.e., through DFEs), can in principle be pieced together by linking each DFE's balance sheet to its investor-plans, that is a complicated and burdensome undertaking (as this study demonstrates). The results reported here suggest that pension plan sponsors have not in any widespread or systematic way utilized DFEs to hide the true nature of plan investments from participants or regulators. Yet they could, and even if the current reporting regime is not abused, reduced visibility blunts monitoring by interested parties. In a digitized world those risks and costs seem unwarranted. DFEs could send data showing each investor-plan's share of the DFE's assets and liabilities classified into the broad categories required by Schedule H, allowing each plan to transfer those amounts into the proper balance sheet lines for combination with the plan's direct investments, rather than reporting indirect investments as an opaque undifferentiated lump (that is, identified only as an interest in an MTIA, CCT, PSA, or 10312 1E). To accommodate plans with different plan years, such reports might need to be transmitted at the end of each month, but such data transmittal has negligible cost in an era of computerized recordkeeping and the internet. Simplified reporting on the plan's Schedule H, although formerly necessitated by cost considerations, is no longer justifiable. A pension plan should be required to report the current value of the plan's allocable portion of the underlying assets and liabilities of each DFE in which it invests, just as current regulations require itemized reporting of a plan's interest in the assets and liabilities of a CCT or PSA that does not file as a DFE. (169) Linking DFE balance sheets to the investor-plans should be mandated.

This simple step, however, is hardly enough. For the many large plans that delegate the investment of a large share of their assets to DFEs, the plan's Schedule H information as currently reported is meaningless. But the truth is that the Schedule H balance sheet categories themselves are so antiquated and undifferentiated that proper allocation of DFE interests among those broad categories (i.e., abolishing simplified reporting as recommended above) is at best palliative. As observed earlier, the single Schedule H category "common stock" lumps together all plan holdings that formally consist of equity interests in incorporated entities, provided that the ownership interest does not entail priority rights to earnings or assets (i.e., is not preferred). Consequently, the reported value of common stock could be broadly diversified or concentrated in one or a few businesses, industries, or sectors; it encompasses ownership of both domestic and foreign enterprises; and it includes stock in companies that are closely held as well as shares subject to broad public trading. Likewise the "real estate" category offers no breakdown between improved and unimproved realty, nor does it provide any information concerning property location or relevant markets. (170) Clearly, there is a disconnect between the breadth and formalism of the Schedule H categories and the substantive goals of disclosure--in most cases a pension plan's balance sheet fails to convey a reliable picture of the risk and return characteristics of its investment portfolio. (171) ERISA gives the Department of Labor authority to prescribe more specific and financially relevant categories for the required statement of plan assets and liabilities, (172) and clearly it should do so.

It might be objected that enhanced Schedule H reporting is unnecessary and duplicative in light of the schedule of investment assets that large plans must submit as a component of their annual reports. (173) That schedule is itemized and requires identification of the "issuer, borrower, or lessor, or similar party to the transaction (including a notation as to whether such party is known to be a party in interest), maturity date, rate of interest, collateral, par or maturity value, cost, and current value." (174) The schedule of investment assets is open to public inspection and available to a plan participant upon written request and payment of the cost of copying. (175) While the schedule of investment assets provides detail that is entirely lacking in the Schedule H balance sheet as currently constituted, it is woefully inadequate as a substitute for a revitalized Schedule H. The schedule of investment assets does not have to be provided to plan participants without charge (unlike the Schedule H contents), (176) Apart from cost, two further deficiencies render the schedule of investment assets inadequate to the task of supplying financially useful information. First, the itemized schedule supplies too much undigested detail to be useful to plan participants; information overload makes it likely that the list would be dismissed out of hand. Second--and most important--the schedule of investment assets is not required to be submitted in a format that supports routine electronic data capture. (177) This format limitation means that government regulators, financial analysts, unions, and other interested parties are not equipped with the information that would be necessary to conduct broad-based statistical analysis of benefit plan portfolios. While the facts of a particular plan's investments are open to inspection, one cannot evaluate whether a plan is an outlier or poses special risks to its participants (or to the PBGC insurance system) unless comparative data are available in a form that allows computer processing.

ERISA demands access to detailed pension plan financial information, but access should not be equated with meaningful ability to assess the information. The policy goals of financial disclosure--to deter misconduct, facilitate enforcement, and give workers information they need to make better (more efficient) personal career and financial planning decisions (178)--depend for their accomplishment on a meaningful ability to assess plan financial data. Those goals are not well served by a system that makes workers and regulators guess whether the structure of a particular plan portfolio poses special risks, or whether any such risk is adequately compensated.

There are important sticking points, of course. Consider the administrative challenge: given the enormous range of characteristics of investment properties, how construct a uniform comprehensive system of digitized investment reporting? Some cases are straightforward. Ticker symbols could be used for publicly traded stocks and corporate bonds. Government bonds could be identified by the issuer (jurisdiction), date of issue, and maturity date. Real estate might be classified as undeveloped, agricultural, industrial, commercial, or residential rental, with an indication of each parcel's location (within the U.S., perhaps using metropolitan statistical areas), and where appropriate the number of units (or acreage) and occupancy rate. Debt or equity interests in privately-held businesses present special difficulties; business name, headquarters address, principal business location (metropolitan statistical area), and principal business activity (perhaps using North American Industry Classification System codes) might all be necessary identifiers. Options would be identified with reference to the underlying property (such as stock or real estate), duration, conditions on exercise, and strike price. Derivatives and synthetic financial products (e.g., mortgage-backed securities and other collateralized debt obligations) pose a conundrum; some mechanism for reporting maximum loss exposure would be desirable, as would an indication of whether the instrument is held for purposes of hedging, speculation, or arbitrage.

Political resistance may present an even greater obstacle than technical feasibility. To provide meaningful ability to assess plan financial information, comparison with representative (not necessarily comprehensive) pension investment data is essential. Individual participants are not going to engage in sophisticated financial analysis of their retirement plans; instead, disclosure policy should aim to empower outside monitors to represent their interests. Such outside monitors can include the Department of Labor, the PBGC (in the case of insured defined benefit plans), and unions representing covered employees. Unfortunately, empowering outside monitors creates the threat of encouraging outside meddlers. The prospect of offering ammunition to plaintiff-side class action litigation firms, disclosing investment strategy to competitors, or dropping clues that might facilitate reverse engineering of hedge fund proprietary trading models, would make employers see red. The outside meddlers concern (the risk that plan financial information will be misused) is longstanding--it can be traced back to the earliest federal foray into financial disclosure. It formed the principal ground for opposition to the Welfare and Pension Plans Disclosure Act in 1958, to its strengthening in 1962, and to its replacement in 1974 with ERISA's more robust disclosure regime. (179) It is a concern that deserves to be studied and taken seriously. Clearly, a disclosure regime that induces pension plan sponsors to switch in and out of investments shortly before and after the required reporting date in an effort to prevent questionable or illegitimate use of particularized plan investment information is worse than useless. That response might hide sensitive information from perceived meddlers, but it would generate large transaction costs and destroy the utility of investment information to outside monitors as well, defeating the purposes of disclosure. Perhaps complete digitized investment data should be made available only to the Department of Labor, the Service, and PBGC, with data sets stripped of all sponsor identifying information released to the public only on a delayed basis. (180)

VI. CONCLUSION

The Department of Labor routinely compiles and publishes aggregate statistics derived from summary financial information filed annually by large private pension plans, and the plan level data are publicly-available, yet that information is all virtually meaningless. The data reveal that a large portion of private pension plan assets--more than sixty percent in the case of defined benefit plans--is held by various indirect investment vehicles (known as direct filing entities) whose portfolios are functionally invisible. Thanks to the failure to systematically link returns filed by direct filing entities with the pension plans that invest through them., the overall pattern of pension plan investments, direct and indirect, is unknown. This article describes a project to carry out such linking (insofar as the quality of the raw data permits) and reports the results.

The linked results confirm that there are large differences in the composition of defined benefit and defined contribution plan investments. Certain plan subtypes also exhibit distinctive asset allocations, such as ESOPs among DC plans. Interestingly, others do not: cash balance plan holdings mirror those of traditional DB pension plans, for example. Once DFE interests are properly categorized and attributed to investor-plans, the overall asset allocation of the investor-plans generally looks quite like the portfolio makeup of plans of the same type and asset level that do not invest through DFEs. When plans that are the product of collective bargaining are compared with those that are not, most holdings appear similar, but collectively-bargained DB plans have a curious propensity to fail to adequately identify DFEs in which they invest. Many plans using DFEs file internally inconsistent returns that preclude successful linking of DFE financial information to the investor-plan (about thirty-five percent in 2008), and a regression analysis reveals several plan characteristics associated with such deficient filings.

This study brings to light a great deal of heretofore inaccessible data concerning private pension plan finances. Yet when these results are evaluated in light of the purposes of pension plan financial disclosure, serious deficiencies remain. Even routine, accurate, and comprehensive matching of DFE financial information with investor-plans does little to inform plan participants or government regulators of the risk and return characteristics of a specific plan, because the asset and liability categories governing electronic data submission are far too broad and formalistic. When "common stock" of all sorts is lumped together, for example, we learn nothing about the marketability of the equity interests that comprise that category, nor is the industry, geographical concentration, or capitalization of the corporations revealed, much less the diversification of the portfolio as a whole. ERISA's text and policies support the regulatory formulation of a far more detailed digital disclosure regime. (181)

I. INTRODUCTION                                       593

II. BACKGROUND                                        599

  A. History and Policy                               599
  B. Data Sources and Limitations                     603
  C. Types of Indirect Investment Vehicles            610
  D. Pension Plan Utilization of Indirect Investment  624
     Vehicles
  E. DFE Asset Holdings                               637

II. METHODOLOGY                                       647

IV. RESULTS                                           653

  A. DB and DC Plan Allocations                       653
    1. Findings of Current Study                      653
    2. Comparison with EBSA Summary Statistics        655
    3. Discussion                                     657
  B. Other Plan Characteristics                       667
  C. Plan-Level Correlations                          686
  D. Characteristics of Plans Filing Inconsistent     689
     DFE Information

V. FINANCIAL REPORTING AND DISCLOSURE POLICY          696

VI. CONCLUSION                                        701


[dagger] This research was supported in part by a grant from the Center for Empirical Research in the Law, Washington University School of Law. Copyright 2013, Peter J. Wiedenbeck, Rachael K. Hinkle & Andrew D. Martin.

(1.) EMP. BENEFITS SEC. ADMIN. (EBSA), U.S. DEP'T OF LABOR, PRIVATE PENSION PLAN BULLETIN: ABSTRACT OF 2010 FORM 5500 ANNUAL REPORTS 3 tbl.A1 (2012), http://www.dol.gov/ebsa/PDE/2010pensionplanbulletin.pdf. The 2010 data are the first to report full recoupment of the 23 percent decline in total private pension plan assets that occurred between 2007 and 2008 ($6.1 trillion in 2007 to $4.7 trillion in 2008), apparently due to the impact of the subprime mortgage crisis and the attendant recession. EBSA, PRIVATE PENSION PLAN BULLETIN HISTORICAL TABLES AND GRAPHS 13 tbl.E11, 14 graph E11g (2012), http://www.dol.gov/ebsa/pdf/historicaltables.pdf.

(2.) In 2010 individual retirement accounts (IRAs) were estimated to hold another $4.8 trillion of retirement savings, PROQUEST, LLC, PROQUEST STATISTICAL ABSTRACT OF THE UNITED STATES: 2013, 777 tb1.1229 (2013). While IRAs were originally designed to make tax-favored retirement savings available to workers who are not covered by an employer-sponsored pension plan (about half of the U.S. labor force), the largest share of IRA assets, 43.2% in 2010, is traceable to tax-free rollovers from private pension plans rather than to individual contributions. Employee Benefit Research Institute (EBRI), Issue Brief No. 375: Individual Account Retirement Plans: An Analysis of the 2010 Survey of Consumer Finances, EBRI.ORG, 21 fig.13a (Sept. 2012), http://www.ebri.org/pdf/briefspdf/EBRI_IB_09_2012_No375_IndvAccts.pdf.

The single largest source of retirement income for most Americans is Social Security. In 2010, 65 percent of elderly households (married couples and nonmarried persons age 65 or older) obtained at least 50 percent of their total income from Social Security. SOCIAL SECURITY ADMINISTRATION (SSA), INCOME OF THE AGED CHARTBOOK, 2010, at 9 (2012), http://www.socialsecurity.gov/policy/docs/chartbooks/income_aged/2010/iac10.pdf.

(3.) Annual reports are mandated by the Employee Retirement Income Security Act of 1974 (ERISA) [section][section] 101(b), 104(a), 29 U.S.C. [section][section] 1021(b), 1024(a) (2012), and are required to contain the financial information set forth in ERISA [section] 103, 29 U.S.C. [section] 1023 (2012). Reports are filed using the Form 5500 Series, "Annual Return/Report of Employee Benefit Plan" and required schedules, instruments which were jointly developed by the Department of Labor. the Internal Revenue Service (Service), and the Pension Benefit Guaranty Corporation (PBGC), to satisfy annual reporting requirements under ERISA Title 1 (administered by Department of Labor), Title IV (the termination insurance program for defined benefit pension plans, administered by the Pension Benefit Guaranty Corporation), and the Internal Revenue Code (Code) (administered by the Service). 29 C.F.R. [section] 2520.103-1 (2012).

(4.) As explained below, the information gap stems from the difficulty of attributing assets held by various indirect investment vehicles, known as direct tiling entities or DFEs, to the pension plans that invest through those vehicles. The Director of Research of Employee Benefits Security Administration (EBSA) told one of the authors: "I caution you that in practice making the links [between pension plans and the DFEs in which they hold interests] can be complicated. We have made some efforts along those lines but that remains an unfinished project here." Email from Joseph Piacentini to Peter Wiedenbeck (July 13, 2010, 1:38 PM). In April 2012, the Department of Labor released its first statistical overview of DFEs. EBSA, FORM 5500 DIRECT FILING ENTITY BLLETIN: ABSTRACT OF 2008 FORM 5500 ANNUAL REPORTS (Mar. 2012), http://www.dol.gov/ebsa/pdf/directfilingentity2008.pdf (ver. 1.0, designated "Preliminary"). The Department of Labor report consists of twelve tables of aggregate statistics, unaccompanied by any explanation of methodology or analysis.

(5.) ERISA [section] 103(b)(2), (3), 29 U.S.C. [section] 1023(b)(2), (3) (2012); 29 C.F.R. [section] 2520.103-1(b) (2012).

(6.) 29 C.F.R. [section] 2520.103-1(e) (2012).

(7.) ERISA [section] 103(b)(4), 29 U.S.C. [section] 1023(b)(4) (2012) (statutory authority to relax reporting requirements for plan assets held in either a common trust fund maintained by a bank or similar institution, or a pooled separate account maintained by an insurance carrier); 29 C.F.R. [section] 2520.103-9 (2012) (regulatory exemption for investor-plans allowing them to dispense with reporting financial information on the underlying assets and liabilities of common trust fund or pooled separate account if the bank or insurance carrier files a Form 5500, including Schedule D and Schedule H, covering the common trust fund or pooled separate account). Accord 29 C.F.R. [section][section] 2520.103-3(c)(2) (current value of investor-plan's proportionate interest in underlying assets and liabilities of common trust fund, along with other financial information, must be reported if the trust does not file Form 5500), 2520.1034(c)(2) (same for investor-plan's interest in insurance company pooled separate account) (2012).

(8.) See ERISA [section] 404(c), 29 U.S.C. [section] 1104(c) (2012); 29 C.F.R. [section] 2550.404c-1 (2012). Mutual fund investments are also common under a special type of defined contribution pension plan that covers employees of charitable organizations or public schools. These so-called "403(b) plans" (exemplified by TIAA-CREF) may only invest in annuity contracts or mutual fund shares. I.R.C. [section][section] 403(b)(7), 851(a).

(9.) See, e.g., INTERNAL REVENUE SERV, & EMP. BENEFITS SEC. ADMIN., FORM 5500, SCHEDULE D, DFE/PARTICIPATING PLAN INFORMATION, Part I (2008), http://www.dol.gov/ebsa/pdf/2008-5500-Schedule-D-mp.pdf ("Information on Interests in MTIAs, CCTs, PSAs, and 103-12 IES"); INTERNAL REVENUE SERV., EMP. BENEFITS SEC. ADMIN. & PENSION BENEFIT GUARANTY CORP., FORM 5500, INSTRUCTIONS, at 11-13, 25 (2008), http://www.dol.gov/ebsa/pdf/2008-5500inst.pdf [hereinafter 2008 INSTRUCTIONS]. 29 C.F.R. [section][section] 2520.103-3(c)(1) (plan must identify DFE and report current value of its investment or units of participation in the common trust fund), -4(c)(1) (same for pooled separate accounts), -12(a) (same for so-called "103-12 investment entities") (2012).

(10.) See, e.g., INTERNAL REVENUE SERV. & EMP. BENEFITS SEC. ADMIN., FORM 5500, SCHEDULE D, DFE/PARTICIPATING PLAN INFORMATION, Part II (2008) ("Information on Participating Plans"), http://www.dol.gov/ebsa/pdf/2008-5500-Schedule-D-mp.pdf; 2008 INSTRUCTIONS, supra note 9, at 11-13, 25.

(11.) See generally, JAMES A. WOOTEN, THE EMPLOYEE RETIREMENT INCOME SECURITY ACT OF 1974: A POLITICAL HISTORY 45-50 (2004). S. REP. No. 85-1440, at 2, 3 (1958), reprinted in U.S. DEP'T OF LABOR, LEGISLATIVE HISTORY OF THE WELFARE AND PENSION PLANS DISCLOSURE ACT OF 1958, AS AMENDED BY PUBLIC LAW 87-420 OF 1962, at 73-74 (1962) [hereinafter WPPDA LEGISLATIVE HISTORY].

(12.) Welfare and Pension Plans Disclosure Act (WPPDA), Pub. L. No. 85-836, [section] 7(b), 72 Stat. 997, 1000 (1958). Certain conflict of interest transactions involving pension plans were required to be listed in detail, including investments in securities or properties of the employer, the union, or plan officials, and fund loans made to such parties. Id. [section] 7(f)(1)(C), (D).

(13.) PUBLIC PAPERS OF THE PRESIDENTS OF THE UNITED STATES: DWIGHT D. EISENHOWER: 1958, at 663 (1959).

(14.) WOOTEN, supra note 11, at 81-83, 98.

(15.) 18 U.S.C. [section] 1027 (2012). The 1962 amendments also made theft or embezzlement from an employee benefit fund a federal crime, and outlawed soliciting or receiving bribes or kickbacks to influence the operation of an employee benefit plan, 18 U.S.C. [section][section] 664, 1954 (2012).

(16.) Welfare and Pension Plans Disclosure Act Amendments of 1962, Pub. L. No. 87-420, [section] 9, 76 Stat. 35, 36.

(17.) S. REP. No. 87-908, at 7, 18 (1961): H.R. REP. NO. 87-998, at 9, 11-12 (1961).

  There were people who wanted the Secretary of Labor to have the power
  to make the report include all types of investments--how much stock
  there was in General Motors or General Electric, or any other
  corporation. We resisted this move. We felt that what was necessary
  here was a general disclosure of the broad category of investments.


(18.) 108 CONG. REC. 1735 (1962) (remarks of Rep. Goodell). Accord, id. at 1736, reprinted in WPPDA LEGISLATIVE HISTORY, supra note 11, at 393-95. (18.) 108 CONG. REC. 1735 (1962) (remarks of Rep. Goodell) ("[T]here shall be no powers given to anyone to control any investment policies in these pension and welfare funds."); id. at 1736.

(19.) Most pension plans are designed to qualify for favorable tax treatment. The regulations provide that a qualified pension, profit-sharing, or stock bonus plan must be a "definite written program or arrangement which is communicated to the employees." Treas. Reg. [section] 1.401-1(a)(2) (as amended in 1976). The definiteness and writing requirements were initially promulgated under section 165 of the Revenue Act of 1938 (which was the source of the anti-diversion rule that now appears as I.R.C. [section] 401(a)(2)), 26 C.F.R. [section] 9.165-1(a) (1939 Supp.), and were apparently intended to ensure that employees would have some enforceable rights under state law. Those criteria prevent the employer's commitment from being construed as illusory, a mere gratuity, or unenforceable under the statute of frauds. Nevertheless, state trust law standards of loyalty and care can be relaxed by inserting exculpatory language in the trust instrument, and employee benefit plans commonly included such language prior to ERISA.

(20.) ERISA [section][section] 404(a) (general fiduciary duties), 406 (prohibited transactions), 410(a) ("[A]ny provision in an agreement or instrument which purports to relieve a fiduciary from responsibility or liability for any responsibility, obligation, or duty under this part shall be void as against public policy."), 29 U.S.C. [section][section] 1104(a) 1106, 1110(a) (2012).

(21.) ERISA [section][section] 502(a)(2), 409, 29 U.S.C. [section][section] 1132(a)(2), 1109 (2012).

(22.) PETER J. WIEDENBECK, ERISA: PRINCIPLES OF EMPLOYEE BENEFIT LAW 16 (2010).

(23.) H.R. REP. NO. 93-533, at 11 (1973), reprinted in 2 STAFF OF THE SUBCOMM. ON LABOR OF THE S. COMM. ON LABOR AND PUBLIC WELFARE, 94TH CONG., LEGISLATIVE HISTORY OF THE EMPLOYEE RETIREMENT INCOME SECURITY ACT OF 1974, at 2358 (Comm. Print 1976) [hereinafter ERISA LEGISLATIVE HISTORY]. See S. REP. No. 93-127 at 4 (1973) ("Experience ... has demonstrated the inadequacy of the [WPPDA] in ... protecting rights and benefits due to workers. It is weak in its limited disclosure requirements and wholly lacking in substantive fiduciary standards."), reprinted in 1 ERISA LEGISLATIVE HISTORY 587, 590; 2 id. at 3293, 3295 (informal report on substitute version of H.R. 2, to same effect).

(24.) ERISA [section] 103(b)(3)(A), (C), 29 U.S.C. [section] 1023(b)(3)(A), (C) (2012).

(25.) ERISA [section][section] 101(b)(1) (filing), 104(a)(1) (filing deadline and public inspection), 106 (public inspection), 29 U.S.C. [section][section] 1021(b)(1), 1024(a)(1), 1026 (2012).

(26.) ERISA [section][section] 101(a)(2), 104(b)(3), 103(b)(3)(A), (B), 29 U.S.C. [section][section] 1021(a)(2), 1024(b)(3), 1023(b)(3)(A), (B) (2012); 29 C.F.R. [section][section] 104b-10(d) (prescribed form for summary annual report), 104b-1(c) (disclosure via electronic media) (2012).

(27.) 29 C.F.R. [section] 104b-10(d)(3) (2012).

(28.) Id. (penultimate paragraph).

(29.) ERISA [section] 104(b)(2), (4), 29 U.S.C. [section] 1024(b)(2), (4) (2012).

(30.) See 29 C.F.R. [section] 104b-10(d)(3) (2012), which requires that the SAR include notice of the participant's right to additional information, including the schedule of "assets held for investment" and "information concerning any common or collective trusts, pooled separate accounts, master trusts or 103-12 investment entities in which the plan participates" (items 3 and 9 on the list of available annual report information).

(31.) ERISA [section][section] 109(a) (forms), 106(a) (study quote), 29 U.S.C. [section][section] 1029(a), 1026(a) (2012).

(32.) ERISA [section] 109(a), 29 U.S.C. [section] 1029(a) (2012); 29 C.F.R. [section] 2520.103-1 (2012) (prescribed form and required schedules). See ERIS.A [section]-[section] 104(e), 3004(a) (coordination of Labor and Treasury Department rules, practices and forms), 29 U.S.C. [section][section] 1024(e), 1204(a) (2012).

(33.) 29 C.F.R. [section][section] 2520.103-1(b), 2520.104-44(b)(1) (2012) (exemptions for welfare plans under which benefits are paid solely from the general assets of the employer or union maintaining the plan (unfunded plans), welfare plans which provide benefits solely through insurance or a qualified health maintenance organization (insured plans), and welfare plans under which benefits are paid in part from the employer's general assets and partly through insurance (partly unfunded and partly insured plans)). Certain fully-insured pension plans are also exempt. 29 C.F.R. [section] 2520, 104-44 (b) (2).

(34.) E.g., EBSA, PRIVATE PENSION PLAN BULLETIN: ABSTRACT OF 2010 FORM 5500 ANNUAL REPORTS 23-36 tbls.C4-C11 (2012), http://www.dol.gov/ebsa/PDF/2010pensionplanbulletin.pdf

(35.) E.g., INTERNAL REVENUE SERV., EMP, BENEFITS SEC. ADMIN, & PENSION BENEFIT GUARANTY CORP., FORM 5500, SCHEDULE H. FINANCIAL INFORMATION, Line 1(c)(11) (2008), http://www.dol.gov/ebsa/pdf/2008-5500-Schedule-H-mp.pdf [hereinafter SCHEDULE H].

(36.) E.g., INTERNAL REVENUE SERV. & EMP. BENEFITS SEC. ADMIN., FORM 5500, SCHEDULE D. DFE/PARTICIPATING PLAN INFORMATION, Part I (2008), http://www.dol.gov/ebsa/pdf/2008-5500-Schedule-D-mp.pdf. The MTIA also files a schedule D, on which it identifies (Part II) each employee benefit plan that invests in the MTIA.

(37) See Peter J. Wiedenbeck, Rachael K. Hinkle & Andrew D. Martin, Research Protocol, Pension Investment Project, n. 10 (April, 2013), http://invisiblepensioninvestments.wustl.edu/ [hereinafter Research Protocol].

(38.) This total consists of 1652 filings by master trust investment accounts (MTIAs); 3115 common or collective trust funds (CCTs); 2048 insurance company pooled separate accounts (PSAs); and 432 103-12 investment entities (103-12 IEs). Each of those four types of DFEs can be utilized by pension plans. (Differences between them are explained below. See infra notes 61-84 and accompanying text.) Another 105 filings were by group insurance arrangements (GIAs), a type of DFE employed by some insured welfare plans which is not relevant to a study of pension fund investments. There were seventy-three duplicate DFE filings (amended returns) in 2008, leaving a total of 7174 DFEs that were either MTIA, CCT, PSA, or 103-12 IE. Research Protocol, supra note 37, at nn.4, 5. Some of these returns reported zero year-end DFE assets. Others failed to include the Schedule H report of assets and liabilities, making it impossible in those instances to attribute DFE investments to participating plans.

The Department of Labor's count is slightly higher, as it reports 7702 total DFEs in 2008, as follows: 1693 MTIAs; 3448 CCTs; 2128 PSAs; and 433 103-12 IEs. EBSA, supra note 4, tb1.1. The explanation for this discrepancy is not clear because this study derived DFE counts directly from the raw data posted on the Department of Labor's web site. The raw filings, however, include returns by 2201 entities which were not identified as either a plan or a DFE, so perhaps in conducting its study EBSA determined that some of these entities are actually DFEs. No edited or revised DFE data have been posted on the Department of Labor's web site, yet it seems that the EBSA numbers are based on a set of data that is somewhat different than the posted raw data.

(39.) See EBSA. Privacy Impact Assessment: Employee Retirement Income Security Act (ERISA) Filing Acceptance System IEFAST) 2010, U.S. DEP'T OF LABOR. http://www.dol.gov/oasam/ocio/programs/PIA/EBSA/EBSA-EFAST.htm (last visited April 12. 2013).

(40.) The raw, unedited data from all of the Form 5500 and Form 5500-SF (the short form for small plans that meet certain requirements, see 29 C.F.R. [section] 2520.103-1(c)(2) (2012)) filings for each year, including the data reported in the various schedules, are posted on the Department of Labor's web site under the heading, "Form 5500 Data Sets." The raw data include all filings by both pension and welfare plans, whether large (100 or more participants) or small, and also include DFE filings. Consequently, the raw data contain records relating to approximately 800.000 filers. See EBSA, Form 5500 Data Sets, U.S. DEP'T OF LABOR, http://www.dol.gov/ebsa/foia/foia-5500.html (last visited April 12, 2013).

(41.) E.g., Actuarial Research Corporation, User Guide: 2008 Form 5500 Private Pension Plan Research File (Contract D0LI089327412) U.S. DEP'T OF LABOR, 9 (Dec. 2010), http://www.dol.gov/ebsa/pdf/2008-5500-researchfileuserguide.pdf

(42.) EBSA, Pension and Health Plan Bulletins and Form 5550 Data, U.S. DEP'T OF LABOR, http://www.dol.gov/ebsa/publications/form5500dataresearch.html (last visited Mar. 3.2013).

(43.) 29 C.F.R. [section] 2520.104a-2(a) (2012) provides:

  Any annual report (including any accompanying statements or
  schedules) filed with the Secretary under part 1 of title I of the
  Act for any plan year (reporting year, in the case of common or
  collective trusts, pooled separate accounts, and similar non-plan
  entities) beginning on or after January 1. 2009, shall be filed
  electronically in accordance with the instructions applicable to such
  report, and such other guidance as the Secretary may provide.


The ERISA annual reporting and disclosure regulations were revised to mandate electronic filing under EFAST2 in late 2007. Annual Reporting and Disclosure, 72 Fed. Reg. 64,710 (Nov. 16, 2007). For an explanation of the proposed rules, see Annual Reporting and Disclosure, 71 Fed. Reg. 41,392 (proposed July 21, 2006). The corresponding revisions to Forms 5500 and attachments, and changes to the accompanying instructions, are set forth in Revision of Annual Information Return/Reports, 72 Fed. Reg. 64,731 (Nov. 16, 2007).

(44.) EBSA, Frequently Asked Questions: EFAST2 Electronic Filing System, U.S. DEP'T OF LABOR, Q&A-42, -43, http://www.dol.gov/ebsa/faqs/faq-EFAST2.html (last visited Mar. 5, 2013).

(45.) Annual Reporting and Disclosure Requirements, 65 Fed. Reg. 21,068, 21,074 (Apr. 19, 2000). See id at 21,069 (absence of standardized reporting format for common or collective trusts and pooled separate accounts "has made it virtually impossible for the Department to correlate and effectively use the data regarding plan assets held for investment by CCTs and PSAs" while the value of plans assets invested in such entities had grown from $113.9 billion to $280 billion between 1990 and 1996).

(46.) See supra text accompanying notes 33-34. 47 ERISA generally requires diversification of plan investments "so as to minimize the risk of large losses, unless under the circumstances it is clearly prudent not to do so."

(47.) ERISA [section] 404(a)(1)(C), 29 U.S.C. [section] 1104(a)(1)(C) (2012). Many defined contribution plans are excused from that requirement, however. ERISA [section][section] 404(a)(2), 407(d)(3), 29 U.S.C. [section][section] 1 I04(a)(2), 1107(d)(3) (2012).

(48.) See supra note 24 and accompanying text.

(49.) Form 5500 Schedule H, line 4(i) asks whether the plan had assets held for investment during the plan year, and if the answer is yes, a detailed schedule is called for. "The schedules must use the format set forth below or a similar format and the same size paper as the Form 5500." 2008 INSTRUCTIONS, supra note 9, at 35. No standardized presentation format is prescribed (a "similar format" is expressly permitted), the size of the columns is not specified, and some of the descriptive information is quite variable ("Description of investment including maturity date, rate of interest, collateral, par, or maturity value"). Consequently, automated electronic data capture of the detailed schedule of investments is not feasible.

(50.) See INTERNAL REVENUE SERV., EMP. BENEFITS SEC. ADMIN. & PENSION BENEFIT GUARANTY CORP., FROM 5500, INSTRUCTIONS, at 39-40 (2012), http://www.dol.gov/ebsa/pdf/2012-5500inst.pdf. "Any information that cannot be contained on the 5500 series forms and schedules may be submitted as an unstructured attachment on EFAST2" and must be submitted as either a PDF file or an ASCII Text file. EFAST2 Guide for Filers & Service Providers, U.S. DEP'T OF LABOR, [section] 5.2.2 (ver. 5.2, Apr. 27, 2011), http://www.efast.dol.gov/fip/pubs/EFAST2_Guide_Filers_Service_Providers.pdf. Such "unstructured attachments" include the schedule of assets held for investment. See EBSA, supra note 44, Q&A-24a. The serious policy implications of this limitation are discussed in Part V. infra, text accompanying notes 169-180.

(51.) ERISA [section] 3(21)(A)(i), 29 U.S.C. [section] 1002(21)(A)(i) (2012). Uniform federal fiduciary obligations, including both general standards of conduct (derived from the traditional trustee duties of loyalty and care), and a set of objectively-defined prohibited transactions, were a central policy innovation.

  ERISA imposes uniform federal fiduciary obligations to control
  mismanagement and abuse of employee benefit programs. While drawing
  on general principles of trust law, ERISA's fiduciary standards
  include two fundamental departures from prevailing state law. First,
  the statutory definition of fiduciary extends far beyond state law
  trustees, imposing standards of competence and fair dealing on anyone
  who has or exercises any discretionary authority in the
  administration of the plan or the management of its assets, and on
  investment advisors as well. Second, ERISA voids any attempt to relax
  its stringent fiduciary obligations through the inclusion of
  exculpatory clauses in the plan, even though such indulgences are
  common and effective under state law.

  Federal fiduciary standards were designed to work in combination
  with improved disclosure of plan finances and powerful enforcement
  tools to stem misconduct in plan administration. Particularized
  reporting of transactions between the plan and certain related
  parties would give participants and the Department of Labor
  information needed to assert workers' rights, while the federal
  courts, armed with broad remedial powers and supported by
  nationwide service of process, would grant effective relief.
  Moreover, employees would be free to assert their rights without
  fear of employer retaliation by discharge, demotion, or other
  adverse employment action.


WIEDENBECK, supra note 22, at 16 (footnotes omitted).

(52.) ERISA [section] 103(b)(3)(A), 29 U.S.C. [section] 1023(b)(3)(A) (2012).

(53.) The Pension Protection Act of 2006, Pub. L. No. 109-280, [section] 611(f). 120 Stat. 780, 972, added a definition of plan assets as ERISA [section] 3(42), 29 U.S.C. [section] 1002(42) (2012), which essentially reinforces the prior regulatory definition, 29 C.F.R. [section] 2510.3-101 (2012). The principal departure of the new statutory definition from the longstanding Department of Labor rule imposes a limitation on the definition of "benefit plan investor" for purposes of determining significant equity ownership under the twenty-five percent test described below. See infra note 57 and accompanying text. The 1986 plan asset regulation counted equity ownership by all pension or welfare plans, including government plans and church plans that are not subject to ERISA, 29 C.F.R. [section] 2510.3-101(f)(2)(i) (2012); the new statutory definition adopts the twenty-five percent threshold but forbids counting government or church plan investments, ERISA [section] 3(42), 29 U.S.C. [section] 1002(42) (2012).

(54.) 29 C.F.R. [section] 2510.3-101(a)(2) (2012). The Department of Labor's definition of plan assets also applies for purposes of the excise tax on prohibited transactions. 29 C.F.R. [section] 2510.3-101(a)(1) (2012); see I.R.C. [section] 4975.

(55.) For the definitions of "equity interest" and "publicly-offered security," see 29 C.F.R. [section] 2510.3-101(b) (2012). Moreover, when it comes to mutual funds, the scope of ERISA's fiduciary obligations have always been limited by statute: "In the case of a plan which invests in any security issued by an investment company registered under the Investment Company Act of 1940, the assets of such plan shall be deemed to include such security but shall not, solely by reason of such investment, be deemed to include any assets of such investment company." ERISA [section] 401(b)(1), 29 U.S.C. [section] 1101(b)(1) (2012).

(56.) 29 C.F.R. [section] 2510.3-101(c)(1) (2012) (emphasis added). Despite the focus on financial services, however, the look-through rule does not apply to certain entities that predominately make venture capital or real estate investments, but only if the entity qualities as a "venture capital operating company" or "real estate operating company". Id. Those categories are defined with reference to whether the entity has the right to substantially participate in the management of the underlying venture capital or real estate investments. 29 C.F.R. [section] 2510.3-101(d)(3), (e) (2012). The look-through rule also does not reach the underlying assets of certain government-guaranteed mortgage pools. 29 C.F.R. [section] 2510.3-101(i) (2012). A special rule provides that if a related group of employee benefit plans owns all of the outstanding equity interests (other than director's qualifying shares) of an entity, then the assets and management of the wholly-owned enterprise are fully subject to ERISA even if it is an operating company and is not engaged in rendering financial services. 29 C.F.R. [section] 2510.3-101(h)(3), (4) (2012).

(57.) 29 C.F.R. [section] 2510.3-101(f) (2012).

(58.) 2008 INSTRUCTIONS, supra note 9, at 29; accord id. at 30 (warning that a plan's interest in the underlying assets of a common or collective trust or pooled separate account that does not file its own Form 5500 "must be allocated and reported in the appropriate categories on a line-by-line basis on Part 1 of the Schedule H"). See also 29 C.F.R. [section] 2520.103-10(b)(1)(ii) (2012) ("Except as provided in the Form 5500 and the instructions thereto, in the case of assets or investment interests of two or more plans maintained in one trust, all entries on the schedule of assets held for investment purposes that relate to the trust shall be completed by including the plan's allocable portion of the trust.").

(59.) 2008 INSTRUCTIONS, supra note 9, at 11-14.

(60.) Tax exemption of a qualified trust requires that it form "part of a stock bonus, pension, or profit-sharing plan of an employer for the exclusive benefit of his employees and their beneficiaries," I.R.C. [section] 401(a), and the use of the singular ("an employer") created doubt about whether multiple employers could pool qualified plan assets in a combined trust without forfeiting favorable tax treatment. At least as early as 1939 the Service announced that "[a] trust forming part of a plan of affiliated corporations for their employees may be exempt if all requirements are otherwise satisfied." Regulation 101, Treas. Reg. [section] 9.165-1(f) (issued under the Revenue Act of 1938 and published in the 1939 Supplement to the Code of Federal Regulations). That language was carried forward in various iterations of the regulations under the Internal Revenue Code of 1939. E.g., Treas. Reg. [section] 39.165-1(b) (1953) (applicable to taxable years beginning after Dec. 31, 1951); id. [section] 39.1-1(b). By its terms, however, this regulation only blessed a trust created under "a plan" (singular) covering the employees of an affiliated group of corporations. Moreover, in determining whether "all requirements are otherwise satisfied" the qualification standards were apparently applied to each member corporation separately because each is a distinct "employer." See I.R.C. 404(a)(3)(B); H.R. REP. No. 1337, at 43, A150-51 (1954) (member of affiliated group lacking current or accumulated earnings and profits cannot make contributions for its employees under common profit-sharing plan of the group, and prior to 1954 profitable group members could not deduct contributions for loss corporation's employees); S. REP. No. 83-1622, at 54-55, A150-51 (1954) (same); Rev. Rul. 69-35, 1969-1 C.B. 117.

By 1944 the Service was permitting unrelated corporations to adopt a single plan and contribute to a common exempt trust, but the qualification requirements of Code section 401(a) and the limits on deductible contributions of section 404(a) were applicable to each participating employer separately. P.S. No. 14 (Aug. 24, 1944), reprinted in GERHARD A. MUNCH, FEDERAL TAXATION OF INSURED PENSIONS, app. 4, at App-103 (1966), restated and superseded by Rev. Rul. 69-230, 1969-1 C.B. 116; see Rev. Rut. 32, [section] 3, 1953-1 C.B. 265 (requests for advance rulings may be submitted by industry-wide or other multiple employer plans); Treas. Reg. [section] 1.401-1(d) (adopted 1956).

With ERISA's enactment in 1974, plans adopted by unaffiliated companies became subject to a few special qualification requirements. A "multiemployer plan" is a plan maintained pursuant to a collective bargaining agreement under which two or more unrelated employers are required to contribute. ERISA [section] 3(37), 29 U.S.C. [section] 1002(37) (2012); I.R.C. [section] 414(f). Some qualification conditions are relaxed for multiemployer plans, I.R.C. [section] 413(b), as are the advance funding and PBGC termination insurance rules for DB plans. ERISA [section][section] 302(a), 304. 4201-4303, 29 U.S.C. [section][section] 1082(a), 1084, 1381-1453 (2012); I.R.C. [section][section] 412(a), 431-32. Where two or more unrelated employers maintain a single plan that is not the product of collective bargaining, the program is now generally called a "multiple employer plan." Mere adoption by distinct employers of a plan having identical terms does not make the program a single plan; a single plan exists if and only if all assets are available to provide benefits to a covered employee of either employer. See 1.R.C. [section] 413(c), Treas. Reg. [section][section] 1.413-2, 1.413-1(a)(2), 1.414 (1)-1(b)(1) (as amended in 1979); see generally I MICHAEL J. CANAN, QUALIFIED RETIREMENT PLANS [section][section] 6:2, 6:11, 6:13 (2012 2013 ed.) (distinguishing collectively bargained plans, multiemployer plans and multiple employer plans). Hence a form plan adopted by several employers ordinarily is not a multiple employer plan; despite uniform terms the arrangement gives rise to independent single employer plans sponsored by each adopting company. Similarly, the Department of Labor takes the position that for purposes of ERISA Title I, including reporting and disclosure rules and fiduciary obligations, a plan and trust adopted by several unrelated employers is not a single multiple employer plan if the adopting employers are not members of a bona fide group or association of employers. ERISA Op. Ltr. 2012-04A (May 25, 2012); see generally U.S. GOV'T ACCOUNTABILITY OFFICE, GA0-12-665, PRIVATE SECTOR PENSIONS: FEDERAL AGENCIES SHOULD COLLECT DATA AND COORDINATE OVERSIGHT OF MULTIPLE EMPLOYER PLANS (2012). See infra notes 63, 69 (concerning master plans and associated trusts).

(61.) 29 C.F.R. [section] 2520.103-1(e) (2012).

(62.) Id. (parenthetical clause): see ERISA [section] 402(a)(1), 29 U.S.C. [section] 1102(a)(1) (2012) (permission for directed trustees).

(63.) For example, the master trust moniker is commonly applied to a collective trust established under a master plan. Master plan means a form plan which "is made available by a sponsor for adoption by employers and for which a single funding medium (for example, a trust or custodial account) is established, as part of the plan, for the joint use of all adopting employers" and which complies with Service procedures for obtaining an advance determination as to its qualification. Rev. Proc. 2011-49, [section] 4.01, 2011-44 I.R.B. 608, 611. Typically, master plans are developed by trade associations, professional organizations, banks, insurance companies, or regulated investment companies, and the sponsor offers the master plan as a low-cost preapprovcd plan that may be utilized by its members or customers. As the employers adopting a master plan are unrelated, a master trust that serves as the collective funding medium for the participating employers is not a "master trust" within the meaning of the Department of Labor's annual report regulations. See Rev. Rul. 71-461, 1971-2 C.B. 227 (where several employers adopted master plan but coverage under one adopting employer's plan became discriminatory, exempt status of master trust not adversely affected if the trustee transfers the funds held under the disqualified plan to an unrelated trust as soon as administratively feasible). Instead, a trust funding vehicle for a master plan is ordinarily classified as a common or collective trust for annual reporting purposes. See infra note 69.

Master plan procedures were first developed to handle the flood of determination letter requests by small employers following the 1963 extension of qualified status to plans covering self-employed individuals (so-called "Keogh" or "H.R. 10" plans covering partners and sole proprietors). Rev. Proc. 63-23, [section] 3.02, 1963-2 C.B. 757. The program was so successful that the Service later extended it to corporate plans, Rev. Proc. 68-45, [section] 2, 1968-2 C.B. 957, 958, and it has been repeatedly expanded, refined, and updated ever since, Rev. Proc. 2011-49, supra. Indeed, a random survey of 1200 sponsors conducted in 2010 found that 86% of 401 (k) plans are some form of pre-approved plan (such as a master, prototype, or volume submitter plan); only 14% were individually-designed plans. EMP. PLANS COMPLIANCE UNIT, INTERNAL REVENUE SERV., SECTION 401 (K) COMPLIANCE CHECK QUESTIONNAIRE INTERIM REPORT 56-57 (2012), http://www.irs.gov/pub/irstege/401k_interim_report.pdf.

(64.) 2008 INSTRUCTIONS, supra note 9, at 11.

(65.) See supra note 63 and accompanying text.

(66.) 29 C.F.R. [section] 2520.103-3(a), (b) (2012).

(67.) 12 C.F.R. [section] 9.18(a) (2012) (authority for national banks to establish collective investment funds for assets held in a fiduciary capacity, referring separately to common trust funds in (a)(1) and group trusts in (a)(2)).

(68.) I.R.C. [section][section] 584 (common trust fund), 581 (definition of bank). Participants in the common trust fund include these amounts on their own returns in combination with similar items of income or loss derived from other sources. Thus, a common trust fund receives a simplified version of conduit or pass-through tax treatment, akin to that accorded regulated investment companies (mutual funds) under I.R.C. [section] 852. I.R.C. [section] 584 originated as section 169 of the Revenue Act of 1936, in response to court decisions that held such commingled investment funds taxable as corporations. S. REP. NO. 74-2156, at 20 (1936).

(69.) Rev. Rul. 2011-1, 2011-2 I.R.B. 256. The "master trust" funding medium under a preapproved master plan (see supra note 63) would ordinarily qualify as a group trust and therefore be classified as a CCT for annual reporting purposes. Such a "master trust" is not an MTIA for reporting purposes because it pools funds accumulated under multiple plans (albeit plans that share common terms) of unrelated employers. See Treas. Reg. [section][section] 1.4132(a)(2) ("mere fact that a plan, or plans, utilizes a common trust fund or otherwise pools plan assets for investment purposes does not, by itself, result in a particular plan being treated as" a multiple employer plan subject to [section] 413(c); master or prototype plan maintained by employers that are not members of commonly controlled group is subject to [section] 413(c) only if it is a single plan), 1.413-1(a)(2), 1.414(l)-1(b)(1) (single plan requires that all plan assets be available to pay benefits to employees covered the plan) (as amended in 1979), Rev. Rul. 2011-1, supra (terms of a group trust required to provide that assets contributed under one participating plan may not be used to benefit employees or beneficiaries under another participating plan).

(70.) The Federal Reserve Board amended its regulations specifying trust powers of national banks in 1955 to permit collective investment of the funds of two or more qualified pension, profit-sharing, or stock bonus plans. 20 Fed. Reg. 3305 (May 14, 1955) (amending 12 C.F.R. [section] 206.10(c) (1955)); see 12 C.F.R. [section] 9.18 (2012) (current version). Shortly thereafter the Service ruled that if participation in such a "group trust" is limited to trusts under qualified pension or profit-sharing plans and specified protections were in place, then the group trust would constitute a qualified trust. Rev. Rul. 56-267, 1956-1 C.B. 206; see also Rev. Proc. 56-12, [section] 2.02(d), 1956-1 C.B. 1029 (procedure to apply for determination of qualified status of "master trust" under a pooled fund arrangement, where "individual trusts under separate plans pooled their funds for investment purposes through a master trust").

(71.) Rev. Rul. 2011-1, 2011-2 I.R.B. 256. But see I.R.S. Gen. Couns. Mem. 39,873 (Apr. 15, 1992) (participation by voluntary employees' beneficiary association will disqualify the group trust). See also H.R. REP. No. 93-1280, at 337-38 (1974) (Conf. Rep.) (ERISA conference committee directs Service to allow IRAs to participate in group trusts).

(72.) 15 U.S.C. [section][section] 77c(a)(2) (registration exemption under the Securities Act of 1933), 78c(a)(12)(A)(iii), (iv) (common trust fund an exempted security under Securities Exchange Act of 1934), 78c(a)(12)(A)(iv) (same for group pension trust), 80a-3(c)(3) (exemption under the Investment Company Act of 1940, provided that interests in the fund are not advertised or offered for sale to the general public), 80a-3(c)(11) (exemption under the Investment Company Act of 1940 for bank-maintained common trust funds consisting solely of assets of qualified plans or certain government or church plans) (2012). See generally 1 TAMAR FRANKEL & ANN TAYLOR SCHWING, THE REGULATION OF MONEY MANAGERS [section] 6.06[B] & [C] (2d ed. 2001).

(73.) Compare 12 C.F.R. [section] 9.18(b)(7) (2012) ("A bank may not advertise or publicize any fund authorized under paragraph (a)(1) of this section, except in connection with the advertisement of the general fiduciary services of the bank.") with 15 U.S.C. [section] 80a-3(c)(3) (2012) (exemption under the Investment Company Act of 1940, provided that interests in the fund are not advertised or offered for sale to the general public). Indeed, an earlier version of the national bank regulation authorizing collective investment of fiduciary funds explained:

  The purpose of this section is to permit the use of Common Trust
  Funds, as defined in section 584 of the Internal Revenue Code, for
  the investment of funds held for true fiduciary purposes; and the
  operation of such Common Trust Funds as investment trusts for other
  than strictly fiduciary purposes is hereby prohibited. No bank
  administering a common trust fund shall issue any document evidencing
  a direct or indirect interest in such common trust fund in any form
  which purports to be negotiable or assignable. The trust investment
  committee of a bank operating a Common Trust Fund shall not permit
  any funds of any MIST to be invested in a Common Trust Fund if it has
  reason to believe that such trust was not created or is not being
  used fur bona fide fiduciary purposes. A bank administering a Common
  Trust Fund shall not, in soliciting business or otherwise, publish or
  make representations which are inconsistent with this paragraph or
  the other provisions of this part and, subject to the applicable
  requirements of the laws of any State. shall not advertise or
  publicize the earnings realized on any Common Trust Fund or the value
  of the assets thereof.


12 C.F.R. [section] 206.17(a)(3) (1959).

(74.) 29 C.F.R. [section] 2520.103-4(a), (b) (2012). Insurance company separate account assets are classified as plan assets under ERISA. 29 C.F.R. [section] 2510.3-101(h)(1)(iii) (2012).

(75.) See generally FRANKEL & SCHWING, supra note 72, [section] 5.06[A].

(76.) I.R.C. [section][section] 817(b) (basis of assets in variable contract segregated asset account increased or decreased by any appreciation or diminution in value), (d), (e) (pension plan contracts treated as variable contract), 818(a) (pension plan contract includes contracts with qualified pension, profit-sharing, stock bonus and annuity plans, [section] 403(b) plans, IRAs, governmental plans qualified under [section] 401(a), and eligible deferred compensation plans under [section]457(b)); see I.R.S. Priv, Ltr. Rul. 9230023 (Apr. 29, 1992) (PSA qualifies as variable contract under [section] 817(d)). See also 1.R.C. [section] 817(a) (separate account asset appreciation excluded from increase in reserves, thereby preventing insurer from claiming deduction under [section] 807(b) for an amount not included in gross income); RICHARD S. ANTES ET AL., FEDERAL INCOME TAXATION OF LIFE INSURANCE COMPANIES [section] 17.05[3] (2011).

(77.) See I.R.C. [section][section] 801(d)(1), (e)(1)(B), 818(a) (deductible reserves for pension plan contract set by reference to balance in the policyholder's fund).

(78.) 15 U.S.C. [section][section] 77c(a)(2) (registration exemption under the Securities Act of 1933 for contract issued by insurance company in connection with qualified pension, profit-sharing, stock bonus or annuity plan. as well as certain governmental and church plans), 78c(a)(12)(A)(iv), (a)(12)(C) (exempted security under Securities Exchange Act of 1934), 80a-3(c)(1) (exemption under the Investment Company Act of 1940) (2012). FRANKEL & SCHWING, supra note 72, [section][section] 5.06, 6.06[G]. Observe that the exemptions under the 1933 and 1934 Acts do not apply to contracts issued in connection with a plan covering one or more owners of an unincorporated business (a so-called Keogh or H.R. 10 plan). See H.R. REP. NO. 91-1382, at 44 (1970) (Keogh plans not exempted "because of their fairly complex nature as an equity investment and because of the likelihood that they could be sold to self-employed persons, unsophisticated in the securities field").

(79.) 15 U.S.C. [section][section] 80a-3(c)(11), -2(a)(37), 77c(a)(2)(C) (2012) (exemption under the Investment Company Act of 1940 restricted to insurance company separate accounts the assets of which are "derived solely from" qualified pension, profit-sharing, or annuity plans, or certain governmental plans).

(80.) See 2008 INSTRUCTIONS, supra note 9, at 11 (Form 5500 for MTIA must include "Schedule D, to list CCTs, PSAs, and 103-12 IEs in which the MTIA invested at any time during the MTIA year and to list all plans that participated in the MTIA during its year").

(81.) ERISA [section] 103(b)(3)(G), (b)(4), 29 U.S.C. [section] 1023(b)(3)(G), (b)(4) (2012).

(82.) 29 C.F.R. [section][section] 2520.103-9 (DFE treatment conditioned upon bank or insurance company that maintains the CCT or PSA filing completed Form 5500 and required schedules; administrator of investor-plan must also supply the plan number, and the name and employer identification number of plan sponsor, to the bank or insurer to enable identification of the plan on the return filed for the CCT or PSA (see Schedule D Part II)). -3(c)(1) (CCT), -4(c)(1) (PSA) (2012).

(83.) 29 C.F.R. [section][section] 2520.103-3(c)(2) (CCT), -4(c)(2) (PSA) (2012); see supra note 58 and accompanying text. The sponsoring financial intermediary is required to transmit to the administrator of each investor-plan a certified annual statement of assets kind liabilities of the CCT or PSA and the value of the plan's units of participation, and must also state whether or not the CCT or PSA will file as a DFE. ERISA [section] 103(a)(2), 29 U.S.C. [section] 1023(a)(2) (statutory information-forcing authority) (2012); 29 C.F.R. [section] 2520.103 -5(c)(1)(ii), (iii) (PSA), -5(c)(2)(i), (ii) (CCT) (2012).

(84.) 29 C.F.R. [section] 2520. 103-12(c), (e) (2012). For participating plans to obtain the full benefit of tax-free accumulation, it is important that the investment entity not be subject to the corporate income tax. Consequently, 103-12 IEs are ordinarily organized in a form that qualifies for conduit tax treatment as a partnership (or conceivably, as an S corporation). I.R.C. [section][section] 701, 1363(a); Treas. Reg. [section] 301.7701-3(a) (as amended in 2006).

(85.) 29 C.F.R. [section][section] 2520.103-12(c) (definition of 103-12 Investment Entity), 2510.3-101 (plan asset definition) (2012). See supra notes 54-58 and accompanying text.

(86.) See 29 C.F.R. [section] 2520.103-12(a) (2012); supra text accompanying note 58.

(87.) 2008 INSTRUCTIONS, supra note 9, at 25 (Schedule D Instructions).

(88.) Alternatively, the unidentified investors might be plans that were misidentified on the DFE's Schedule D (incorrect E1NPN supplied), or they might be plans that failed to tile Form 5500 in 2008. Indeed, perhaps these mystery investors are not employee benefits plans at all--maybe the bank or insurance company might simply list the EIN of all entities that invest in the CCT or PSA without regard to whether they are pension or welfare plans.

For the exemption from filing Form 5500 for one-participant plans, see 2008 INSTRUCTIONS, supra note 9, at 3.

(89.) ERISA [section] 401(1)(1), 29 U.S.C. [section] 1101(b)(1) (2012) (mutual fund assets not deemed plan assets); 29 C.F.R. [section] 2510.3-101(a)(2), (h)(1) (2012) (look-through rule inapplicable to security issued by investment company registered under the Investment Company Act of 1940). Insurance company assets are statutorily exempt from the look-through rule only if the plan's policy or contract with the insurance company provides for benefits the amount of which is guaranteed by the insurer. ERISA [section] 401(b)(2), 29 U.S.C. [section] 1101(b)(2) (2012). The scope of this guaranteed benefit policy exception was uncertain until the Supreme Court gave it an expansive reading in John Hancock Mut. Life Ins. Co. v. Harris Trust & Sav. Bank, 510 U.S. 86 (1993). The insurance industry reacted with alarm to the decision in Harris Trust, seeking protection from ERISA fiduciary obligations. Congress responded by providing a safe harbor for policies issued to an employee benefit plan that are supported by the assets of the insurer's general account, but that safe harbor applies only to policies issued before 1999. ERISA [section] 401(c), 29 U.S.C. [section] 1101(c) (2012): 29 C.F.R. [section] 2520.401c-1 (2012). See 29 C.F.R. [section] 2510.3-10 1(h)(2) (2012) (look-through rule does not apply to plan's interest in an insurance company, apparently including an equity interest obtained by virtue of being a policyholder in a mutual insurance company).

(90.) Although it does not file Form 5500, an insurance company that provides funds from its general account for the payment of benefits is obligated to provide information to the plan administrator that is needed to prepare the plan's annual report and schedules, 29 C.F.R. [section] 2520.103-5(b)(1), (c)(1)(i) (2012).

(91.) Generally, a mutual fund (open-end investment company) must redeem its securities within seven days of tender. 15 U.S.C. [section] 80a-22(e) (2012); FRANKEL & SCHWING, supra note 72, [section][section] 5.08[C][2], 26.01.

(92.) See DAN M. MCGILL ET AL., FUNDAMENTALS OF PRIVATE PENSIONS 693 (9th ed. 2010).

(93.) In 2008, the portion of single-employer DB plan assets reported as "Partnership/joint venture interests" increases with the overall level of plan assets, from an average of 0.5% for plans with below-median assets to 3.5% of total assets for the ten percent of plans having the highest total assets. In contrast, single-employer DC plans held hardly any partnership interests in 2008 (less than 0.1% on average), regardless of plan asset size. (Throughout this study any plan reporting no year-end assets is excluded from consideration, so plan asset size categories are determined without reference to the number of such zero-asset plans.)

(94.) ERISA [section] 3(34), 29 U.S.C. [section] 1002(34) (2012); accord I.R.C. [section] 414(i). A defined benefit plan is a pension plan that is not a defined contribution plan. ERISA [section] 3(35), 29 U.S.C. [section] 1002(35) (2012); accord I.R.C. [section] 414(j). Under a traditional DB plan, the employer promises a specified level of benefit payments (typically paid in the form of a life annuity) to provide support commencing on retirement, and the contributions necessary to fund the promised benefits are determined actuarially. The plan sponsor's commitment, therefore, is fixed by reference to contributions (money going in) in the case of a defined contribution plan, and is fixed by reference to benefit distributions (money going out) in the case of a defined benefit plan.

(95.) See WIEDENBECK, supra note 22, at 349-51.

(96.) See infix: notes 144-146, Figure 15, and accompanying text.

(97.) See supra note 40 and accompanying text. The Department of Labor's edited data do not include DFE filings. See EBSA, supra note 42, and accompanying text.

(98.) The first question on Form 5500 calls for identification of the type of filing entity. E.g., INTERNAL REVENUE SERV., EMP. BENEFITS SEC. ADMIN. & PENSION BENEFIT GUARANTY CORP., FORM 5500, ANNUAL RETURN/REPORT OF EMPLOYEE BENEFIT PLAN, Part I, Line A(4) (2008), http://www.dol.gov/ebsa/pdf/2008-5500-mp.pdf. If the DFE box is checked the filer is instructed to enter a one-letter code to specify the type of DFE (M = MTIA; C = CCT; P = PSA; E = 103-12 IE). See 2008 INSTRUCRIONS, supra note 9, at 15.

(99.) For 2008 the total amount reported by all large pension plans (both DB and DC) as the year-end value of interests in MTIAs was $1.195 trillion, while the total net assets reported by MTIAs (reduced by amounts reported as interests held in other MTIAs to avoid double counting) was $1.150 trillion. The small (<4%) apparent excess of the value of plan interests over the total net asset value of MTIAs is presumably attributable to differences in reporting year or reporting errors. In contrast, the Department of Labor's 2008 DFE statistical summary reports the total amount invested by private pension plans in MTIAs as $1.216 trillion, and total MTIA assets of $1.389 trillion. EBSA, supra note 4, tbls.2, 10. The difference apparently stems from variation in the underlying data sets; in conducting its analysis EBSA seems to have corrected or supplemented the raw filings. See supra note 38. To date EBSA has not published its methodology.

(100.) For 2008 the total amount reported by all large pension plans as the year-end value of interests in PSAs was $109 billion, while the total net assets reported by PSAs (reduced by amounts reported as interests held in other PSAs to avoid double counting) was $167 billion. In contrast, the Department of Labor's 2008 DFE statistical summary reports the total amount invested by private pension plans in PSAs was $112 billion, and total PSA assets of $193 billion. EBSA, supra note 4, tbls.2, 10; see supra note 99. Although the Department of Labor labels the $112 billion as "private pension plan" assets invested in PSAs, implying that the number includes investments by both large and small plans, other reported data suggest that $112 billion in fact represents only large plan investments. Compare id. tb1.10 with id. tb1.12 ($112 billion reported as total original large plan PSA interests).

To avoid regulation as an investment company, PSA participation must be limited to qualified retirement plans and governmental plans. See supra note 79. Governmental plans are not subject to ERISA and not required to file annual reports, ERISA [section] [section] 4(b)(1), 3(32), 29 U.S.C. [section][section] 1003(b)(1), 1002(32) (2012), while the simplified financial information reported by small pension plans that is available in digital form (Schedule T) does not break out DFE investments. Therefore, while it seems clear that PSA assets in excess of large plan interests are traceable either to small plans or governmental plans, the data do not allow us to precisely determine their respective contributions. The PSAs in our data set reported a total of 1,065,693 investor "plans" on Schedule D, Part H, of which 830,147 could be confidently identified as pension plans, but only 177,118 of those (21%) were large pension plans. See supra tb1.1, note 88 and accompanying text. This indicates that PSAs are heavily utilized by small pension plans (which is consistent with longstanding experience in the industry), and so it seems likely that the 35% of PSA net assets not accounted for by large plan investments are overwhelmingly attributable to interests owned by small private pension plans.

(101.) See supra notes 67, 84-86 and accompanying text. MTIA participation is limited to plans of a single employer or of a commonly controlled group of employers.

(102.) For 2008 the total amount reported by all large pension plans (both DB and DC) as the year-end value of interests in CCTs was $514 billion, while the total net assets reported by CCTs (reduced by amounts reported as interests held in other CCTs to avoid double counting) was $1.281 trillion. In contrast, the Department of Labor's 2008 DFE statistical summary reports the total amount invested by private pension plans in CCTs was $525 billion, and total CCT assets of $1.796 trillion. EBSA, supra note 4, tbls.2, 10.

For 103-12 IEs the corresponding numbers are $26.1 billion and $211 billion. According to EBSA the total amount invested by private pension plans (both DB and DC) in 103-12 IEs was $27.7 billion, and total 103-12 IE assets of $279 billion. EBSA, supra note 4, tbls.2, 10.

(103.) The aggregate net assets of all small plans in 2008 was approximately $514 billion. See EBSA, PRIVATE PENSION PLAN BULLETIN: ABSTRACT OF 2008 FORM 5500 ANNUAL REPORTS, tbls.A3, D7 (2010). http://www.dol.gov/ebsa/PDF/2008pensionplanbulletin.pdf ($4.577 trillion total net assets all plans, of which $4.063 reported by large plans).

(104.) The 2008 counts are as follows: 1484 MTIAs reported $1.28 trillion in gross total assets; 2877 CCTs reported $2.13 trillion total assets; 1819 PSAs reported $195 billion total assets; and 402 103-12 IEs reported $316 billion total assets. The DFE counts reported in the preceding sentence include only DFEs of each type reporting non-zero assets. See supra note 38. The corresponding numbers, according to the Department of Labor, are as follows: 1693 MTIAs with $1.39 trillion in gross total assets; 3448 CCTs with $1.80 trillion total assets; 2128 PSAs with $193 billion total assets: and 433 103-12 1Es with $279 billion total assets. EBSA, supra note 4, tb1.1.1.

(105.) A third possibility should also be noted: conceivably, small pension plans (those covering fewer than 100 participants) might utilize CCTs much more heavily than MTIAs, just the reverse of the relative importance of these investment vehicles for large plans (Figures 1-2). While this could be a contributing factor, its significance is clearly limited. inasmuch as the total assets of all small private plans in 2008 was only $526 billion, which is considerably less than the difference between reported CCT and MTIA total assets in 2008. See EBSA, supra note 103. tbl.C1.

(106.) See supra text accompanying notes 66-73.

(107.) The data presented in Figure 10 tracks the DFE balance sheet report contained in the Department of Labor's 2008 DFE statistical summary. EBSA, supra note 4, tb1.2. One important difference is that the Department of Labor numbers do not report DFE ownership interests in lower-tier DFEs, instead allocating those lower-tier DFE interests to other asset categories according to the nature of the lower-tier DFE's investments. As a result of this difference, the size ranking of the non-DFE asset categories reflected in Figure 10 corresponds to the ranking reported in the EBSA balance sheet, but the actual proportions (portfolio shares) differ.

(108.) See supra note 80 and accompanying text. The Department of Labor's DFE statistical summary reports that in 2008 no PSAs or 103-12 IEs invested in an MTIA, while ninety-five MTIAs invested in another MTIA. It also reports ten CCTs holding interests in an MTIA. EBSA, supra note 4, tb1.9.

(109.) Accord EBSA, supra note 4. tbl.10

(110.) Id.

(111.) Id. tb1.2.

(112.) A statutory exemption from ERISA's prohibited transaction rules authorizes a bank that is a plan fiduciary to invest all or part of the plan's assets in deposits in the fiduciary bank itself, provided that the deposits bear a reasonable rate of interest and that such investment is expressly authorized by a provision of the plan or by an independent fiduciary who has the power to direct the bank with respect to such investment. ERISA [section] 408(b)(4)(B), 29 U.S.C. [section] 1108 (b)(4)(B)(2012): 29 C.F.R. [section] 2550.408b-4 (2012).

It is also possible that the pressure on banks to maintain adequate capital reserves in the face of declining asset values caused by the financial crisis might have induced CCT trustees to allocate an unusually large share of CCT assets to deposits in the sponsoring bank in 2008. The objective of this study is to link direct and indirect pension plan investments in 2008, so no comprehensive analysis of prior-year financial data has been undertaken. Nevertheless, reports of CCT asset holdings in 2007 were extracted from the Department of Labor's raw data files for comparison with the 2008 CCT portfolio allocation described above. The combined CCT balance sheet data (Schedule H) show that interest-bearing cash accounted for 7.1% of aggregate CCT investments in 2007, which is substantially less than the 11.3% average portfolio share reported in 2008. This finding suggests that the onset of the 2008 credit crunch may have contributed to the higher CCT utilization of interest-bearing accounts relative to other DFEs. Indeed, the two largest changes in CCT portfolio allocations between 2007 and 2008 were a 4.2 percentage point decrease in common stock holdings and a 4.2 percentage point increase in interest-bearing cash.

(113.) EBSA, supra note 4, tb1.2.

(114.) Id.

(115.) See 2008 INSTRUCTIONS, supra note 9, at 30 (instructions for Schedule H, line c) (15), "other investments").

(116.) Beth J. Dickstein & Robert A. Ferencz, Qualified Plans--Investments, BNA TAX MGMT. PORTFOLIO No. 377, at A-26 (2007).

(117.) Andrew W. Needham & Christian Brause, Hedge Funds, BNA TAX MGMT. PORTFOLIO No. 736, at A-1 to A-7 (2007); STAFF OF THE SEC. & EXCH. COMM'N. IMPLICATIONS OF THE GROWTH OF HEDGE FUNDS 33-36 (2003), http://www.sec.gov/news/studies/hedgefunds0903.pdf [hereinafter SEC STAFF HEDGE FUND REPORT]. Private equity funds that are owned 25 percent or more by benefit plan investors (so that the plan asset look-though rule applies) may also qualify as 103-12 IEs. Dickstein & Ferencz, supra note 116. Private equity funds are turn-around specialists that buy a controlling interest in underperforming companies and enhance their value through expert management and strategic realignment. Hence a private equity fund is heavily invested in common stock which it typically holds for several years. The 21% of 103-12 IE assets consisting of common stock may be attributable, to some extent, to private equity funds.

(118.) Of course, partial ownership of an unincorporated operating company would also appear in the joint venture asset category. The instructions for Schedule H. line 1(c)(5), for reporting investments in "Partnership/joint venture interests" state:

  Include the value of the plan's participation in a partnership
  or joint venture if the underlying assets of the partnership or
  joint venture are not considered to be plan assets under 29 CFR
  2510.3-101. Do not include the value of a interest in a
  partnership or joint venture that is a 103-12 IE. Include the
  value of a 103-12 IE in 1c(12).


2008 INSTRUCTIONS, supra note 9, at 30.

(119.) I.R.C. [section] [section] 514, 511(a); Elliot Knitwear Profit Sharing Plan v. Commissioner, 614 F.2d 347 (3d Cir. 1980) (income generated by securities purchased on margin subject to tax on unrelated debt-financed income). Indeed, entry of an amount on Schedule H, line Ii, "Acquisition indebtedness," is virtually a concession that the tax on unrelated debt-financed income applies. See 2008 INSTRUCTIONS, supra note 9, at 30-31 (indicating that the acquisition indebtedness liability category does not pertain to real property and applies as provided in Code section 514(c)).

If certain conditions are satisfied, a qualified plan may incur debt to acquire real property without triggering the tax on unrelated business taxable income (UBTI). I.R.C. [section] 514(c)(9). Therefore the prospect of tax exposure does not discourage leveraged real estate investments by qualified plans, although the exception is subject to special rules where a plan is a partner in a partnership that holds real estate. I.R.C. [section] 514(c)(9)(B)(vi), (c)(9)(C). (e)(9)(E). Another debt-financed income exception provides that participation in securities lending programs will not trigger UBTI. I.R.C. [section] 514(c)(8).

(120.) See SEC STAFF HEDGE FUND REPORT, supra note 117, at 37-38.

(121.) If the fund is a partnership for tax purposes its borrowing is attributed to the partners, giving rise to UBTI exposure for tax-exempt investors. The feeder corporation "blocker" strategy prevents this result because shareholders (unlike partners) are not attributed a share of the underlying debt. That is, the hedge fund's leveraged returns belong to its corporate partner, and do not pass through to the corporation's shareholders. Where the feeder corporation is organized offshore (is a foreign corporation), the blocking can be accomplished without incurring corporate income tax. Needham & Brause, supra note 117, at A-46 to A-47: Vadim Mahmoudov, Rafael Kariyev & Daniel Backenroth, Playing With Blocky: Testing a Fund's Blocker Allocations, 133 TAX NOTES 993 (Nov. 21, 2011).

(122.) These averages are referred to as minima because they are computed without reference to the large number of unidentified DFE investors, some (perhaps many) of which might be pension plans. See supra note 88. The data reported in the Department of Labor's DFE statistical summary apparently do not permit computation of the average number of pension plans per DFE. Table 1 of the Department of Labor summary reports the "number of invested private pension plans" for each DFE type, but where a plan invests in several DFEs of the same type--for example, where a high asset plan invest in five MTIAs--it seems to be counted only once. See EBSA, supra note 4, tb1.1, n.1 (multiple counting of plan if it invests "in more than one type of DFE"). Another table displays the distribution of DFEs by type among specified ranges of the number of private pension plan investors. Id tb1.3. That distribution seems consistent with the ordinal ranking of the average number of private pension plans per DFE type reported here (i.e., MTIA < 103-12 IE < CCT << PSA), but it does not allow computation of precise averages.

(123.) Research Protocol, supra note 37. The web appendix also reports the complete results of this study, including summary spreadsheets showing the aggregate portfolio composition of various categories of large single-employer pension plans and multi-color versions of the figures presented in this article.

(124.) Actuarial Research Corporation, supra note 41, at 8 10.

(125.) Id. at 4.

(126.) See id. at 23-31 (reported source for all original data used in Pension Research Files is either Form 5500, Schedule H, or Schedule I; no data taken from Schedule D).

(127.) See Research Protocol, supra note 37, at Step II.

(128.) Plans that did not invest in a DFE at any time during the year were identified by two conditions: (1) no non-zero entry appeared in any of the Schedule H DFE asset categories; and (2) either no Schedule D was filed, or the plan tiled a Schedule D reporting interests only in CCTs or PSAs that were identified as not being DFEs.

(129.) Specifically, a plan is put into Group 3 if the reported Schedule H amount for the EOY value of the plan's interest in a given type of DFE is not more than $10 different from the sum of the amounts reported on Schedule D as the plan's investment in DFEs of that type, and that condition is met for all four types of DFEs. CCTs and PSAs identified on Schedule D as not filing their own Form 5500 for the year (non-DFE CCTs and PSAs) are excluded from the sum used for comparison with the reported Schedule H value of interests in CCTs or PSAs. (In accordance with the regulations, the Form 5500 instructions require such non-DFE CCTs and PSAs to be identified on Schedule D, but their assets and liabilities must be allocated among the appropriate specific balance sheet categories on Schedule H, not reported as a unitary interest in a CCT or PSA. A non-DFE CCT or PSA is identified by the sponsor's name and EIN, while a plan number of 000 is used to indicate that the CCT or PSA in question did not file as a DFE.) Any plan that is not in Group 1 (because it invested in a DFE) and does not satisfy the reporting consistency tests for Group 3 is classified in Group 2.

(130.) See infra Part IV.D and Table 2.

(131.) 2008 INSTRUCTIONS, supra note 9, at 25.

(132.) 2008 INSTRUCTIONS, supra note 9, at 11 provides: "Form 5500 filed for the DFE, including all required schedules and attachments, must report information for the DFE year (not to exceed 12 months in length) that ends with or within the participating plan's year."

(133.) Actuarial Research Corporation, supra note 41, at 3 (80% of pension plans file on a calendar year basis). The authors' tally from 2008 DFE Form 5500 filings shows that 87% of DFEs of all sorts filed a return for a plan year beginning on January 1. The proportion is 93% in the case of MTIAs, which are associated with higher-asset plans (see supra Figures 3&4).

(134.) See supra Figure 10.

(135.) To pursue the example in Figure 12, our protocol attributes to Plan A1 A1% of the $U of MTIA direct investments, preserving their proper characterization, plus A1% of the $V invested in CCT1, but this latter amount is reported as an undifferentiated interest in a CCT. Proper multilevel attribution would attribute to Plan A1 A1% of the SU MTIA direct investments, plus [V/(W + X)] of SW CCT1 assets (other than its interest in CCT2), plus A1% of [V/(W + X)] of $X CCT2 assets, and the attributions from CCT1 and CCT2 would consist of a share of each underlying investment held by these DFEs.

(136.) Many DC plan returns (10,795 in 2008) report $0 in all Schedule H asset categories except participant loans, but also report as their total assets a number that is larger than the reported amount of participant loans. This unidentified excess was coded as another asset category, "imputed assets." This situation appears almost exclusively among plans that do not use DFEs (Group 1 plans) having low levels of total assets. On average, imputed assets made up fully 25% of the portfolio of Group 1 DC plans with less than $1.33 million in gross assets (roughly the bottom 20% of DC plans ranked by assets size), but only 0.5% for plans with gross assets greater than $14.69 million (about the top 20% of DC plans).

We hypothesize that such reports correspond to plans that invest all assets in mutual fund shares. If so, Schedule H should report only one entry in the asset categories, namely, an interest in registered investment companies (RICs, ordinarily mutual funds), and the same number should be reported as the plan's total assets, provided that the plan does not make loans to participants. If it does allow participant loans, then the technically correct reporting would be to show entries in two asset categories, both for interests in RICs and loans to participants. (The loan asset amount might well also show up in the "other liability" category if the plan borrows from a bank to raise the cash to make loans to participants.) Where participant loans are allowed, total gross assets should equal the sum of the RIC holdings and participant loans. However, because ERISA provides that the underlying assets of a RIC are not plan assets (in contrast to many other types of indirect investments), inexpert trustees who do no more than pass contribution dollars along to one or more mutual fund companies might not view themselves as holding any plan assets other than claims for repayment of plan loans.

(137.) The range is from $32.91 million to $72.69 million in reported gross assets, constituting the 90th to 95th percentile of single-employer DC plans classified in either Group 1 or Group 3 when ranked by asset size.

(138.) EBSA, supra note 4.

(139.) Id. tb1.11.

(140.) See infra text accompanying notes 169-180.

(141.) Indirect liabilities dominate direct liabilities for Group 3 DC plans as well, but in this case the ratio of total liabilities (direct and linked) to total gross assets (direct and linked) among DC plans that use DFEs (Group 3) is actually less than the ratio of liabilities to gross assets for DC plans that do not use DFEs (Group 1). This lower overall debt ratio appears to be attributable to the very high level of liabilities characteristic of leveraged ESOPs, which typically do not use DFEs (See infra Figure 1 9A and text accompanying note 156). If one focuses exclusively on the large majority of DC plans that are 401(k) plans, the liability pattern is similar to the findings for DB plans. Indirect liabilities dominate direct liabilities for Group 3 401(k) plans and the ratio of total liabilities (direct and linked) to total gross assets (direct and linked) among 401(k) plans that use DFEs (Group 3) is greater than the ratio of liabilities to gross assets for 401(k) plans that do not use DFEs (Group 1).

(142.) Results not displayed here. The complete results of this study, including summary spreadsheets showing the aggregate portfolio composition of various categories of large single-employer pension plans, are posted as a web appendix at http://invisiblepensioninvestments.wustl.edu/.

(143.) This finding is consistent with the similarity, noted earlier, in the utilization of indirect investment vehicles by cash balance and traditional DB plans. See supra Figure 7 and text accompanying note 96.

(144.) Specifically, a cash balance plan promises a benefit equal to the balance of a hypothetical account which is credited annually with a specified percentage of pay and notional interest tied to the accumulated balance of the account. Such pay credits and interest credits make cash balance plans functionally equivalent to money purchase pension plans, which are defined contribution plans. See Kyle N. Brown, Specialized Qualified Plans--Cash Balance, Target, Age-Weighted and Hybrids, BNA TAX MGMT. PORTFOLIO 352-3d, at A-37 to A-66(12) (2007); Cooper v. IBM Pers. Pension Plan, 457 F.3d 636 (7th Cir. 2006). Under a cash balance plan, however, the employer is legally obligated to pay the full promised benefit (balance of the hypothetical account) regardless of the actual investment performance of the pension fund assets, and that allocation of risk causes the arrangement to be classified as a defined benefit rather than a defined contribution plan. ERISA [section] 3(34), (35). 29 U.S.C. [section] 1002(34), (35) (20l2); I.R.C. [section] 414(i), (j).

(145.) See generally TOWERS WATSON, PENSIONS IN TRANSITION: RETIREMENT PLAN CHANGES AND EMPLOYER MOTIVATIONS 4-5, 8-9 (2012).

(146.) ERISA [section] 204(g), 29 U.S.C. [section] 1054(g) (2012); I.R.C. [section] 411(d)(6). For cash balance conversions after June 29, 2005, benefits accrued prior to amendment under the traditional pension formula must be preserved and in addition benefits earned by post-amendment service under the new cash balance formula must serve to increase total benefits, not be credited against the pre-amendment entitlement (i.e., so-called "wear away" transition rules are treated as prohibited age discrimination). ERISA [section][section] 204(b)(5)(B)(ii), (iii), 203(0(3), 29 U.S.C. [section][section] 1054(b)(5)(B)(ii), (iii), 1053(f)(3) (2012); I.R.C. [section] 411(b)(5)(B)(ii)-(iii), (a)(13). See STAFF OF JOINT COMM. ON TAXATION, 109TH CONG., GENERAL EXPLANATION OF TAX LEGISLATION ENACTED IN THE 109TH CONGRESS 329, 460-68 (Comm. Print 2007); Brown, supra note 144. at A-57 to A-60.

(147.) Ordinarily, frozen plans continue operations because they have not accumulated sufficient assets to pay all accrued benefits. Under current law, a financially healthy sponsor cannot voluntarily terminate an underfunded plan. See ERISA [section] 4041(a)(1), (b), 29 U.S.C. [section] 1341(a)(1), (b) (2012).

(148.) Where the same ongoing plan was randomly selected as a match for more than one frozen plan it is included in the comparison set of ongoing plans only once. Hence the number of matched ongoing plans is less than three times the number of frozen plans. Also, where only one or two ongoing plans report total gross assets falling within a two-percent collar of the asset level of a particular frozen plan, the frozen plan and the available matches were retained in the data set, but this was the case for only one frozen single-employer plan. Frozen plans were excluded from the comparison if no ongoing plans satisfied the asset-level screen, which ruled out thirty-eight plans.

(149.) The authors investigated using a matching protocol that would screen for comparability of both gross and net asset levels, but the simpler gross asset test was adopted because gross and net asset levels were found to track very closely for the overwhelming majority of plans.

(150.) ESOPs can include an elective contribution element, and such a cash-or-deferred component would cause the plan to be classified as both an ESOP and a 401(k) plan. I.R.C. [section][section] 401(k)(2), 4975(e)(7). That combination, however, is relatively rare. According to the Department of Labor, in 2008 there were 7048 ESOP filings, of which 1374 had a 401(k) feature and 5672 did not. By way of comparison, in 2008 there were 510,209 401(k) plans that were not ESOPs. EBSA, supra note 103, tbl.D16 at 62.

(151.) See supra Figure 8 and note 95.

(152.) The Department of Labor reports average assets per large (i.e., 100 or more participants) MPPP of $41.6 million in 2008; the corresponding number for large 401(k) plans is $29.7 million. See EBSA, supra note 103, author's computation from Tables Al(a), Dl (2192 large MPPPs; total gross assets $91.194 million), D4, and D9 (64,263 large 401(k) plans with $1,910,099 million total assets).

(153.) See supra text accompanying notes 148-149.

(154.) The excluded plan, with $4.28 billion in reported gross assets in 2008, is the American Airlines, Inc. Pilot Retirement Benefit Program Variable Income Plan.

(155.) ERISA [section] 404(c), 29 U.S.C. [section] 1104(c) (2012).

(156.) ERISA [section] 407(d)(6), (5), 29 U.S.C. [section] 1107(d)(6), (5) (2012): I.R.C. [section][section] 4975(e)(7), (8), 409(l): Treas. Reg. [section] 54.4975-11 (as amended in 2012).

(157.) See 29 U.S.C. [section] 186(c)(5) (2012) (governance structure): ERISA [section][section] 3(37) (definition of multiemployer plan). 302(a)(2)(C), 304 (minimum funding standards). 4001(a)(3) (definition for purposes of PBGC insurance program), 4006 (premium rates), 4022A (benefits guaranteed), 4201 4303 (withdrawal liability), 29 U.S.C. [section][section] 1002(37), 1082(a)(2)(C), 1084, 1301(a)(3), 1306, 1322a, 1381-1453 (2012). In 2008, 2939 multiemployer plans reported total assets of $471.5 billion, or on average $160 million per plan: fewer than one percent of single employer plans in 2008 reported assets comparable to or greater than this amount. EBSA, supra note 103, tbls.A6, B2.

(158.) The sort was based upon responses to Form 5500, Part I, question C. which instructs: "If the plan is collectively bargained, check here." According to the Department of Labor tallies, in 2008 there were 46,926 single-employer plans, of which 3399 were collectively bargained and 43,526 were not. Of the 667,156 single-employer DC plans in 2008, only 7627 were collectively bargained. EBSA, supra note 103, tbl.A6.

(159.) But see infra text accompanying note 167.

(160.) Categories listed in square brackets contained insufficient data to calculate a correlation.

(161.) In the methodology explanation these unlinkable plans are referred to as Group 2 plans. See supra text accompanying notes 128-129. Group 2 plans are identified in the web appendix to this article, together with the portfolio information each reported on Schedule H. http://invisiblepensioninvestments.wustl.edu/.

(162.) ERISA [section] 103(a)(1)(B), (a)(3) (examination and opinion by independent qualified public accountant required), 29 U.S.C. [section] 1023(a)(1)(B), (a)(3) (2012); 29 C.F.R. [section][section] 2520.103-1(b)(5) (rule for large pension plans), 2520.103-8 (limit on scope of examination and report for assets held by bank or insurance carrier, including DFEs that are MTIA. CCT or PSA), 2520.103-12(d) (limit on examination and report concerning 103-12 IE) (2012).

(163.) An unqualified opinion reports the accountant's conclusion "that the plan's financial statements present fairly, in all material respects, the financial status of the plan as of the end of the period audited and the changes in its financial status for the period under audit in conformity with generally accepted accounting principles" or another comprehensive accounting system such as the cash basis. 2008 INSTRUCTIONS, supra note 9, at 33 (Sch. H, line 3(a)(1)).

(164.) See supra notes 81-83 and accompanying text.

(165.) The Schedule H balance sheet categories call for reports of the "Value of interest in common/collective trusts" and "Value of interest in pooled separate accounts" (lines 1(c)(9) and (10), respectively) without explicitly indicating that entries should be made only for CCTs and PSAs that tile as DFEs. SCHEDULE H. supra note 35. That DFE limitation is set forth in the instructions to Form 5500 and Schedule H, but is not made clear on the Schedule itself. See 2008 INSTRUCTIONS, supra note 9.

(166.) The private contractor does not clean data from Schedule H or Schedule D. just the Form 5500 itself.

(167.) The result for collectively bargained plans is somewhat surprising in light of the Finding that collectively bargained DB plans that utilize DFEs and file consistent returns frequently fail to properly identify the DFEs in which they invest on Schedule D. See supra Figure 20 and text accompanying note 159.

(168.) The Schedule H instructions warn that "If the required accountant's report is not attached to Form 5500, the filing is subject to rejection as incomplete and penalties may be assessed." 2008 INSTRUCTIONS, supra note 9, at 33.

(169.) See supra notes 82-83 and accompanying text.

(170.) See supra text accompanying notes 46-49.

(171.) In a few instances Schedule H requires itemized reporting of investment positions that have distinctive and highly relevant financial characteristics. Most notably, interests in employer securities must be separately reported (distinct from the generic categories of common stock. preferred stock, and corporate debt), and U.S. government securities are also broken out. It should also be observed that the rise of participant-directed 401(k) plans has probably greatly improved the investment information available to DC plan participants because of the mutual fund information distributed under the rules of ERISA [section] 404(c), 29 .U.S.C. [section] 1104(c) (2012). 29 C.F.R. [section][section] 2550.404c-1(b)(2)(i)(B)(2), 2550.404a-5 (2012).

(172.) The statute demands that the annual report include a statement of assets and liabilities "aggregated by categories and valued at their current value," ERISA [section] 103(b)(3)(A), 29 U.S.C. [section] 1023(b)(3)(A) (2012), but ERISA does not specify the categories. That task was left to the implementing regulation, 29 C.F.R. [section] 2520.103-1(b)(1) (2012), which requires large plans to report the financial information called for by Form 5500 Schedule H and the instructions thereto. Because ERISA grants the Secretary of Labor express authority to prescribe forms and to promulgate regulations to carry out ERISA title 1, only administrative action is needed to compel more functional and informative balance sheet reporting. ERISA [section][section] 109(a) (authority to prescribe forms), 505 (regulations), 29 U.S.C. [section][section] 1029(a), 1135 (2012).

(173.) See generally 29 C.F.R. [section][section] 2520.103-1(b)(1), -10, -11 (2012). These regulations implement ERISA [section] 103(b)(3)(C), 29 U.S.C. [section] 1023(b)(3)(C) (2012).

(174.) ERISA [section] 103(b)(3)(C), 29 U.S.C. [section] 1023(b)(3)(C) (2012); 29 C.F.R. [section][section] 2520.10310 (2012).

(175.) See supra notes 28-30 and accompanying text.

(176.) Compare items 3 and 9 listed under "Your Rights to Additional Information" in 29 C.F.R. [section] 2520.104b-10(d)(3) (2012), which provide that a pension plan's schedule of investment assets and information relating to DFEs in which the plan participates may be obtained upon request and payment of copying costs, with the penultimate paragraph of the same subsection, which provides that the Schedule H balance sheet information can be obtained without charge. 29 C.F.R. [section] 2520.104b-10(d)(3) (2012).

(177.) See supra note 49 and accompanying text.

(178.) See WIEDENBECK, supra note 22, at 14-16, 57-58.

(179.) In 1958 business leaders and Republican legislators attempted to limit coverage of the WPPDA to jointly-managed Tall-Hartley plans, exempting single-employer defined benefit pension plans from mandatory disclosure. They contended that fiduciary abuses had been uncovered only in the operation of multiemployer plans. Because the company sponsoring a single-employer defined benefit plan (then referred to as "level-of-benefits plans") bears the risk of underfunding, the employer has a strong incentive to monitor and properly manage the fund, rendering disclosure to participants unnecessary. Moreover, they asserted that disclosure of funding levels and investment information would be harmful, because the information would be misused, particularly by unions seeking wage and benefit increases, and that the resulting bargaining pressure would discourage adequate funding. "Where there has been no testimony of abuse and where it is conceded by almost everyone that this type of plan is not susceptible to abuse, it seems to me we have no right to legislate away confidential information which in the hands of unions would create a whipsaw at the bargain table between management and labor." 104 CONG. REC. 16,439 (1958) (remarks of Rep. Bosch), reprinted in WPPDA LEGISLATIVE HISTORY, supra note 11, at 370. Accord H.R. REP. NO. 85-2283, at 23-25 (1958) (minority views assert that a "great deal of harm could result from requiring disclosure of minute details of finance by a level-of-benefits plan. Such a requirement might shift the bargaining emphasis away from benefits and [facilitate] a whipsawing technique by unions."), reprinted in WPPDA LECHSLATIVE HISTORY, supra note 11, at 376-78. The specter of social investing was also raised: "Disclosure of the volume and distribution of investments of level-of-benefits plans will also inevitably culminate in those parties not responsible for providing the benefits to demand a voice in determining the type of investments to be made." Id.

These arguments for exempting single-employer defined benefit plans from financial disclosure were renewed and persistently promoted in connection with congressional consideration of the 1962 amendments to WPPDA. To Amend the Welfare and Pension Plans Disclosure Act: Hearing on S. 1994 Bcybre the Subcomm. on Labor of the S. Comm. on Labor and Public Welfare, 87th Cong. 18 (1961) (prepared statement of Secretary of Labor Arthur Goldberg); id. at 29-34 (extended colloquy between Sen. Goldwater and Department of Labor officials): id. at 44-47 (prepared statement of Sen. Allott) (including concerns over divulging investment practices to competitors and pressure for social investing): see supra note 17.

Proposals for more detailed disclosure of investment information also drew fire during the early development of the pension reform proposals that evolved into ERISA. See PRESIDENT'S COMM. ON CORN. PENSION FUNDS & OTHER PRIVATE RET. & WELFARE PROGRAMS. PUBLIC POLICY AND PRIVATE PENSION PROGRAMS 77-79 (1965) (Cabinet committee recommends detailed disclosure); WOOTEN, supra note 11 at 123-24, 127-28 (objections to more stringent disclosure).

(180.) Even stripped of sponsor identifying information, a complete plan-level investment report might be easily associated with the corresponding simplified Schedule H balance sheet, which of course is not anonymous. But the Schedule 11 report (founded, as it is, on largely dysfunctional categories) would become unnecessary under a regime of standardized comprehensive digital reporting. Granting participants a right of access to the complete financial data pertaining to their own plans seems appropriate, but once released to participants such sponsor-identified data might be disseminated to others, and over time many sponsors in the publicly released data set might be identified. To counteract that erosion of anonymity, the arbitrary identifier assigned to each plan in the publicly released data set might be changed annually, but that measure would hinder plan-specific longitudinal studies by scholars and other private-sector analysts. Suffice it to say that devising an effective mechanism to keep sponsor identifying information out of the hands of outside meddlers is no simple task.

(181.) The research protocol and the complete results of this study, including summary spreadsheets showing the aggregate portfolio composition of various categories of large single-employer pension plans and multi-color versions of the figures, are available as a web appendix.

Peter J. Wiedenbeck, Rachael K. Hinkle & Andrew D. Martin *

* Peter J. Wiedenbeck is the Joseph H. Zumbalen Professor of the Law of Property, Washington University in St. Louis. Rachael K. Hinkle, J.D., Ph.D., is an Assistant Professor of Political Science at the University of Maryland, Baltimore County. Andrew D. Martin is Vice Dean of the School of Law, Charles Nagel Chair of Constitutional Law and Political Science at Washington University, and Director of the Center for Empirical Research in the Law (CERL). The authors are grateful for helpful comments on earlier drafts of this paper provided by Marion Crain, Peter Joy, Pauline Kim, Ronald Levin, Russell Osgood, and other faculty colleagues, as well as participants in the conference on Employee Benefits in an Era of Retrenchment, held at Washington University School of Law. March 29, 2012. Correspondence concerning this project should be directed by e-mail to Peter Wiedenbeck, at pwiedenbeck@law.wust.edu.
COPYRIGHT 2013 Virginia Tax Review
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:IV. Results B. Other Plan Characteristics through VI. Conclusion, with footnotes, p. 667-702
Author:Wiedenbeck, Peter J.; Hinkle, Rachael K.; Martin, Andrew D.
Publication:Virginia Tax Review
Date:Mar 22, 2013
Words:21245
Previous Article:Invisible pension investments.
Next Article:Taxing market discount on distressed debt.
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters