Statement of
Douglas Holtz-Eakin
Director

Estimating the Costs of the
Pension Benefit Guaranty Corporation

before the
Committee on the Budget
U.S. House of Representatives

June 9, 2005

Chairman Nussle, Mr. Spratt, and Members of the Committee, I am pleased to discuss the ongoing work that the Congressional Budget Office (CBO) is doing at the request of this Committee on budgeting for loans, guarantees, and insurance. Today, I will focus on the economic costs, federal costs, and budgetary treatment of the Pension Benefit Guaranty Corporation's (PBGC's) insurance of defined-benefit pension plans.

At the outset of my statement, however, I would emphasize two important caveats. First, CBO's efforts to estimate the costs of PBGC and to identify alternative, potentially more effective budgetary treatments constitute a work in progress. With further refinements, CBO's estimates and findings are likely to change somewhat. Second, the estimates that I will be reporting today are market measures of the value of financial resources being transferred to or within the defined-benefit pension system under current law. They are not budget cost estimates, nor do the estimates of the effects of changes in policy represent the budget scoring for legislation that would effect those changes.

As economic--rather than budget cost--measures, the estimates provide an opportunity to think broadly about federal policy toward defined-benefit pension plans. Under current policy, the full cost of those pensions is not being shouldered by the plans' sponsors. Rather, because the current rules permit plans to be underfunded and because pension insurance is underpriced for many plans, some of the costs are being borne by the plans' beneficiaries and, potentially, by taxpayers. From a budgetary perspective, the key question is, how much should taxpayers be required to contribute to the defined-benefit pension system?

Under current law, PBGC is liable for insured benefits only to the extent that it has resources from insurance premiums, investment income, the assets of terminated plans, and recoveries from sponsors. However, because PBGC is a federal insurance agency, there is a widespread belief that its obligations have at least an implied federal guarantee that commits the government to use general revenues to honor insured claims.

In pursuing the objective of reducing or eliminating federal costs, policymakers have several general types of approaches available. One group consists largely of regulatory instruments, including raising premiums and adjusting them for risk, tightening the pension funding rules, improving the measurement and reporting of pension liabilities, and attempting to increase the discipline of private sponsors' funding decisions. Higher premiums--in particular, ones linked to PBGC's risk exposure--would offset losses on future claims. More accurate measurement of plans' liabilities would make the existing funding rules and premium schedule more effective.

If beneficiaries understood that they were at risk from plans' underfunding, they would have incentives to press for higher funding or perhaps another form of compensation. Accordingly, increased requirements for plans to publicly and frequently disclose sufficient information about their financial condition could be useful in reducing federal costs. Alternatively, privatizing PBGC so that losses were absorbed by its shareholders or by private reinsurers would also bring the force of market discipline to the task of controlling PBGC's losses.

Policymakers could also use budget instruments to help move toward eliminating federal costs for PBGC. Increasing the transparency of PBGC's own financial condition and performance could be as useful as doing so for the pension plans. For instance, the agency's budget accounts could be reconfigured to recognize the accruing cost to the government from pension insurance.

The Congress may also decide that, for various reasons, subsidizing the defined-benefit pension system is desirable. In that case, policymakers may be willing to accept some level of expected funding through general revenues. The same policy instruments could be used to limit taxpayers' exposure as would be used to eliminate it.

The economic costs of PBGC insurance to taxpayers (if the implicit guarantee is honored) are substantial. In thinking about reducing those costs, however, it is critical to distinguish between costs already incurred and prospective costs. PBGC had accumulated losses of $23.3 billion at fiscal year-end 2004 for single-employer plans that had been terminated or whose termination the agency regarded as probable. "Sunk" costs for plans that have been terminated (in actuality or in effect) cannot be avoided, and policy decisions can determine only who will bear those costs. However, policy changes can reduce prospective costs.

CBO estimates that the economic costs to the public of PBGC insurance for single-employer plans net of premium collections over the next 10 years is $48 billion. That figure describes the estimated net present value of the financial resources that the program will be transferring to sponsors of and participants in defined-benefit pensions. It is also the price that the government would have to pay to private insurers bidding in competitive markets to take on the obligations that PBGC will assume in that period with current premiums and funding rules. Adding sunk costs and prospective costs together results in a total of $71 billion for the upcoming decade, $83 billion for 15 years, and $91 billion for 20 years.

In terms of the particular instruments that could be used, CBO's calculations suggest the following:

• Premium collections would have to rise fivefold in order to cut net federal costs to zero through increases in premiums alone. For well-funded plans, which do not pay a premium for underfunding, the increase would be relatively modest, but for severely underfunded plans, which do pay an underfunding premium ($9 per $1,000 of underfunding per year), the increase could constitute a large increase in costs.
  
  
• Some proposals that the Administration has made, if enacted, could measurably reduce the economic costs of the system. For example, increasing premiums from $19 to $30 per participant would reduce 10-year net economic costs by $3 billion, while the proposed tighter rules for calculating pension liabilities and the proposed requirements for increased funding by financially distressed sponsors could reduce prospective economic costs significantly.
  
  
• Other policy changes such as reducing the maximum share of a pension plan's assets that could be invested in equities (stocks) to 30 percent from the current unregulated level of about 70 percent would reduce costs by $7 billion over 10 years.
  
  
• Some changes currently being considered could increase prospective costs. For example, making permanent a legislated increase in the discount rate used to calculate the present value of pension liabilities would increase PBGC's net costs by $5.3 billion. Increasing the average time permitted for closing a plan's funding gap by two years would raise net costs by $6 billion.
  
  
• Changing the budgetary treatment of PBGC or changing its ownership by paying a private entity to take it over would not directly affect net costs but could increase the visibility of those costs and contribute to improved monitoring by the Congress.

Estimating the Costs of PBGC

The recent takeover of several airlines' pension plans by the Pension Benefit Guaranty Corporation has focused attention on and raised concerns about this program's costs to the government, taxpayers, the plans' sponsors, and the plans' participants. However, the budgetary and financial information currently available about PBGC is not very informative about the likely costs of the takeovers or the incidence of those costs.

One reason for the absence of such information is that federal pension insurance gives a large number of beneficiaries valuable but highly uncertain claims to future payments. A natural approach to determining the costs of such claims is to find market prices for equivalent uncertain commitments. Although no exact match is currently available in private markets, finance specialists have developed techniques for using the prices of securities that are bought and sold to price contracts that are not traded. In the case of PBGC, the value of defined-benefit pension insurance is equivalent to a type of put option. Specifically, the option held by a pension plan's beneficiaries is to sell, or put, the assets of the plan to PBGC at a price equal to the value of the insured liabilities, contingent on the financial distress of the sponsor.

CBO has used those techniques along with publicly available information to project the three key determinants of PBGC's costs: the probability of a sponsor's bankruptcy, which is necessary before the put can be exercised; the probability distribution of the plan's underfunding (the plan's liabilities minus its assets) at termination, which is the value of the put option when it is exercised (or when the plan is transferred to PBGC); and market risk (the correlation of PBGC's claims with bad economic conditions), which affects the discount rate used to calculate the present value of the option.

The resulting estimated costs are the market value of the financial resources transferred to the defined-benefit pension system by PBGC. The estimates are based on information contained in Securities and Exchange Commission filings by publicly traded sponsors of defined-benefit pension plans. (Data on privately held companies and confidential filings that sponsors of publicly traded companies with significantly underfunded plans submit to PBGC are not available to CBO.) In the data available to CBO, plans' total liabilities amount to about 88 percent of those reported by PBGC. Therefore, CBO has scaled its estimates of PBGC's costs by a factor of 1.14 to adjust them to the size of the defined-benefit pension system.

The estimates are subject to considerable uncertainty for many reasons. CBO's estimates rely on firms' reports that are based on generally accepted accounting principles of pension assets and liabilities, whereas PBGC's figures rely on firms' reports for the Internal Revenue Service and under the Employee Retirement Income Security Act, which indicate a higher initial level of underfunding. Also, CBO's estimates are based on assumptions that simplify the complexities of the defined-benefit pension system. For example, all plans are assumed to fund pensions with the same mix of assets and to exhibit the same jump in liabilities at termination.

Using those assumptions, CBO estimates that under current policy, the market price of PBGC insurance going forward for existing plans for 10 years is $48 billion (net of premiums and assets of terminated plans and recoveries). That figure conveys the present value of the commitment to take on PBGC's net obligations for existing single-employer plans for the next 10 years. With the $23.3 billion in accumulated losses reported by PBGC at year-end 2004, the combined total of historical and prospective 10-year costs is about $71 billion.

The $23.3 billion in accumulated losses are sunk costs that cannot be avoided by policy changes now and that will be difficult to recover from surviving sponsors. As a consequence, policymakers have greater latitude in focusing on the second component of costs: claims that are prospective under current policy and, therefore, may be avoided.
 

Measures to Reduce the Federal Costs of PBGC

Two general regulatory approaches may be useful in reducing the future net costs of PBGC insurance. The first is to raise insurance premiums and adjust them for risk. The second is to reduce the level of risk in the defined-benefit pension system.

Raising Premiums

Raising premiums would require sponsors to pay a larger share of costs. To cut federal costs to zero through higher premiums alone would require a fivefold increase in PBGC's receipts from premiums. Those higher premiums might be manageable for well-funded plans, which currently pay only a flat charge of $19 per year per participant for insurance. However, for firms with plans that are significantly underfunded, their current annual premiums also include a charge of $9 per $1,000 of underfunding. A hypothetical firm with 1,000 participants and $50 million in underfunding would pay premiums of $469,000 per year, of which $450,000 is the charge for underfunding. Therefore, for some firms, the increase in premiums could be significant--perhaps to the point of causing them to adjust the form and level of compensation that they offer.

Reducing or Charging for Risk

An alternative to a proportionate increase in premiums for all sponsors would be to make premiums more sensitive to the risk that various plans pose for PBGC. Although the extra charge for underfunding currently provides some adjustment based on risk, increasing the variation in premiums on the basis of risk could reduce the current cross-subsidies from low-risk sponsors and plans to high-risk ones. Some risk-adjusted premiums could also strengthen incentives for sponsors to reduce risk--which could lower the premium rate required to achieve any given level of net costs.

With this approach, premiums would be higher for sponsors that were more likely to encounter financial distress and whose plans would tend to be more deeply underfunded at termination. For example, premiums could vary with the volatility of the market value of a firm and its pension assets, the ratio of the firm's liabilities to its equity (leverage), and the firm's credit rating. The resulting range of premiums could be substantially wider than it is under current policy because risk varies significantly among plans. If, for example, premiums were set so that PBGC's expected net cost for insuring an investment-grade company (which is within the top four broad ratings categories) was the same as that for a lower-rated company, they would need to be about 20 times higher per dollar of liability for the lower-rated company.

Another important correlate of plans' risk that could provide a basis for adjusting premiums is the ratio of a pension plan's assets in equities to its total assets. Sponsors appear to prefer a high proportion of equities because they expect higher average returns on stocks than on bonds. If realized, that risk premium would reduce the cash contributions a sponsor must make to its plan in order to fund the promised pension benefits. Of course, such investments entail the risk that the stock market will do poorly and the plan will become underfunded.

Plans with a high proportion of common stocks, rather than high-quality bonds or other fixed-income securities, exhibit more volatility in the value of their assets than do plans holding more debt securities. Plans with a high share of stocks are thus at greater risk of underfunding when the sponsors encounter financial distress. That increase in risk to PBGC means that fair (full-cost) premiums would be about 16 percent lower for plans with an equity share of 30 percent rather than the average of almost 70 percent currently found in defined-benefit pension plans. Such an adjustment in premiums could create incentives for firms' investment decisions that could lower costs and improve the match between the risk posed and the premiums paid. An alternative to relying on the incentive effects of risk-based premiums to reduce risk would be to limit, through law or regulation, the share of assets that plans could invest in stocks.

The current structure of premiums tends to disconnect them from risk because PBGC's costs vary more closely with plans' liabilities rather than their number of participants. The per-participant charge also tends to lower the premium per dollar of insured liabilities for firms with a high proportion of older or high-wage employees compared with firms whose workforce is predominantly younger or lower paid and therefore has few accumulated pension benefits. At the current rate of $19 per participant, those effects may be small, but if rates were raised to be fair on average, the effects on firms' behavior could be significant.

A major source of risk to PBGC is the potentially large gap between the level of pension liabilities reported under the current definitions and funding rules and the economic value of those liabilities at plans' termination. PBGC often reports that plans that appeared to be well-funded prior to termination turn out to be deeply underfunded when they are transferred to the agency. For example, Bethlehem Steel's plan was 84 percent funded on the basis of current reporting requirements but was only 45 percent funded at termination.⁽¹⁾ Underfunding can increase as a sponsor approaches bankruptcy for several reasons, including the discretion that the law allows in calculating the present value of a plan's liabilities and in valuing assets at their purchase price rather than current market value. (Those same funding rules also permit many plans that are effectively underfunded to avoid paying the variable-rate premium of $9 per $1,000 of underfunding.) Changing the definition and measurement of liabilities and tightening the funding rules, especially for sponsors with a greater chance of financial distress, could lessen the risk to PBGC and to the defined-benefit pension system.
 

Increasing the Visibility of PBGC's Costs

The policy changes needed to reduce the costs of pension insurance might be facilitated by increasing the visibility of PBGC's costs through changes in the budgetary treatment of pension insurance or other means. The present budgetary treatment focuses on the cash inflows to PBGC's on-budget account, primarily from premiums, interest income, and transfers from an off-budget trust fund, which holds the assets of plans taken over by PBGC. The inflows are netted against federal outlays for pension benefits in plans run by PBGC's trustees and for administrative expenses. That treatment delays the recognition of insurance claims, often for decades, from when they are realized at a plan's termination to when benefits are paid. As a consequence, and despite large losses, PBGC's budgetary position has contributed to reducing the federal deficit in every year except for fiscal year 2003, when the on-budget account recorded net outlays of $229 million. For fiscal year 2004, net budget outlays for PBGC were once again negative, representing a net cash inflow of $247 million. Such budgetary treatment is not designed to indicate or suited to describing the expected risk and magnitude of losses in the pension insurance system.

The financial statements issued by PBGC include losses on plans that have been terminated and those whose takeover the agency can foresee. In addition, PBGC publishes financial projections based on its Pension Insurance Modeling System, which indicate that the midpoint of the agency's distribution of accumulated deficits in 10 years is about $30 billion. Although both of those indicators of PBGC's financial status provide useful information to policymakers and are good starting points for further analysis, the first focuses primarily on losses that have occurred, including losses on probable terminations (the $23.3 billion cited earlier); and the latter excludes the cost of market risk.

Information on the present market value of future transfers to the defined-benefit pension system net of future premiums might be provided to the Congress through a supplementary reporting system or through changes in budget presentation. The first approach would offer the advantage of avoiding the need for changes to the budget, which are difficult to make piecemeal; the second, the advantage of citing budget numbers, which are more frequently used for policy decisions than supplementary information is.

Budgetary treatments of pension insurance that would better indicate full costs should be the following:

• Timely. According to a recommendation of the President's Commission on Budget Concepts, the budget should reflect outlays when the government incurs the obligation to pay.⁽²⁾ In the case of PBGC, that point suggests that costs should include losses on pension plans when they are terminated.
  
  
• Based on Market Value. In general, the budget uses market prices to measure the value of inputs consumed by various federal programs. For consistency, market prices should be used in estimating insurance costs. For PBGC, the market price of risk is significant because the events that precipitate a transfer of pension liabilities to PBGC, including low investment returns, high rates of financial distress, and low interest rates, occur when the market value of all assets is down.
  
  
• Prospective. The costs relevant to budgeting are those to which the government is committing in the budget period. Although sunk costs need to be recorded and paid, it is those costs that are being incurred in the budget period that are the focus of decisions. Of course, the extent to which the government is committing to pay under current law is restricted to the resources available to PBGC from premiums, assets of terminated plans, and recoveries from sponsors.

The current budgetary treatment of PBGC recognizes the inflow of premium collections during the budget period but not the value of claims arising under the insurance. It thus falls short of having the attributes outlined above. CBO is currently exploring budgetary alternatives that might attain those qualities. One possibility would be to estimate the net prospective economic costs of PBGC over a specified period and to treat those values as the budget baseline costs of the program. Future year budgets could recognize the changes in the value of the insurance due to changes in law, regulation, or variables such as insured liabilities or interest rates. In the language of credit reform, those changes in costs might be treated either as reestimates (the result of unexpected economic changes) or modifications (the result of policy changes).

Another possibility would be to structure the accounts to recognize as budget costs the unpaid fair-value premiums for PBGC insurance. That is, estimates of the annual premiums required to cut the net budget costs of insurance to zero could be compared with the premiums expected to be paid by sponsors, and the difference could be shown as the budget costs of PBGC.

A more extreme approach would be to transfer PBGC to private owners. That step would probably accelerate the recognition of past losses in the budget because the current deficit would have to be covered, presumably by Congressional appropriations, before a private entity would be willing to assume the program's obligations. In addition, a private owner might require either an annual or lump-sum payment from the government to continue to operate the insurance program under current funding rules and premiums. Because PBGC insurance is mandatory for defined-benefit pension plans, the government would probably remain involved in regulating the terms of the insurance--which raises the question of the amount of risk and responsibility the government effectively could transfer to private owners. Nevertheless, the risk to the government would most likely be less than it is under current policy.
 

The Administration's Proposals

The Bush Administration has proposed several changes in the defined-benefit pension system intended to reduce its financial shortfall and increase transparency.⁽³⁾ Generally, the Administration would raise premiums and permit further risk-adjustment of them; change the measure of plans' liabilities and funding requirements; and increase public disclosures of plans' funding status. Plans' sponsors would also be permitted to fund the liabilities at higher levels during good economic conditions (without loss of tax benefits) as a buffer against underfunding during bad economic conditions and to use a higher discount rate to calculate plans' liabilities. Most of those changes are consistent with the objective of reducing the federal costs of pension insurance. More specifically, the major provisions being proposed would do the following:

• Raise the fixed premium per participant from $19 to $30 per year and index the premium to future wage growth. CBO estimates that this change would reduce the prospective 10-year economic costs of PBGC insurance by $3 billion.⁽⁴⁾
  
  
• Authorize PBGC's directors (the Secretaries of Labor, Treasury, and Commerce) to adjust the variable-rate portion of the premium so that PBGC's income would cover expected losses. The change would require more than a sixfold increase in the premiums paid by plans' sponsors.
  
  
• Require that plans' liabilities reflect the effects of early retirements, lump-sum distributions, and increased longevity. The proposal would also require sponsors with credit ratings below investment-grade to calculate pension liabilities by assuming that employees retire at the earliest opportunity, thereby increasing estimated liabilities. Such sponsors would also be required to fund completely any increases in the plans' benefits. Although it is difficult to estimate the effect of the tighter rules for calculating liabilities, they are potentially the largest source of savings among the Administration's proposals.
  
  
• For the purpose of discounting in calculating pension liabilities, funding requirements, and premiums, mandate the use of a three-month average of interest rates on corporate bonds whose duration matches the scheduled payments to beneficiaries. The proposal would make permanent the change from a Treasury rate to a corporate rate for discounting pension liabilities. It would permit plans' sponsors to avoid making up the additional underfunding that resulted from the legislated increase in discount rates for 2004 and 2005. According to CBO's estimates, this proposal would increase PBGC's costs by $5 billion over 10 years.

The Administration's proposals incorporate many of the policy options discussed here to reduce PBGC's risk exposure and to improve the transparency of the system. However, they also omit several options that are relatively important for reducing risk exposure and cross-subsidies between sponsors. First, premiums would continue to be unrelated to the risk of how pension assets are invested. Second, no new limitations would be placed on sponsors' investment policies. Third, the proposals retain a fixed charge per worker, rather than establishing charges per dollar of coverage, which would perpetuate a transfer from plans with younger, lower-paid workers to those with a higher proportion of older workers, higher-paid workers, and retirees.